1 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4 with lower case letter.
6 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
8 * config/i386/i386.md (bswaphi2): New expander.
9 (*bswaphi2_movbe): New insn pattern.
10 (bswaphi -> rorhi pepehole2): New peephole pattern.
12 2017-11-20 Jeff Law <law@redhat.com>
14 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
15 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
16 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
17 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
18 gimple-ssa-evrp-analyze.h.
20 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
22 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
24 2017-11-20 Bin Cheng <bin.cheng@arm.com>
26 * tree-predcom.c: Add general comment on Store-Store chains.
27 (split_data_refs_to_components): Postpone clearing eliminate_store_p
29 (get_chain_last_ref_at): Rename into...
30 (get_chain_last_write_at): ...this.
31 (get_chain_last_write_before_load): New function.
32 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
33 CT_STORE_STORE when write reference is added.
34 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
35 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
36 (initialize_root_vars_store_elim_1): Ditto.
37 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
38 definition is created.
39 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
40 chain by replacing it with dominant stored value.
42 2017-11-20 Bin Cheng <bin.cheng@arm.com>
44 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
46 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
48 * vr-values.c (extract_range_from_binary_expr): Use a full range
51 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
53 * config/arm/arm.md (R4_REGNUM): Define constant.
54 (nonsecure_call_internal): Remove r4 clobber.
55 (nonsecure_call_value_internal): Likewise.
56 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
57 clobber and resequence match_operands.
58 (nonsecure_call_value_reg_thumb1_v5): Likewise.
59 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
60 (nonsecure_call_value_reg_thumb2): Likewise.
62 2017-11-20 Jakub Jelinek <jakub@redhat.com>
64 PR tree-optimization/78821
65 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
66 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
67 to the address of the base rather than the base itself.
68 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
69 (find_bswap_or_nop_finalize): New function.
70 (find_bswap_or_nop): Use it.
71 (bswap_replace): Return a tree rather than bool, change first
72 argument from gimple * to gimple_stmt_iterator, allow inserting
73 into an empty sequence, allow ins_stmt to be NULL - then emit
74 all stmts into gsi. Fix up MEM_REF address gimplification.
75 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
76 (struct store_immediate_info): Add N and INS_STMT non-static
78 (store_immediate_info::store_immediate_info): Initialize them
79 from newly added ctor args.
80 (merged_store_group::apply_stores): Formatting fixes. Sort by
82 (stmts_may_clobber_ref_p): For stores call also
83 refs_anti_dependent_p.
84 (gather_bswap_load_refs): New function.
85 (imm_store_chain_info::try_coalesce_bswap): New method.
86 (imm_store_chain_info::coalesce_immediate_stores): Use it.
87 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
88 (imm_store_chain_info::output_merged_store): Fail if number of
89 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
90 and NOP_EXPR rhs_code.
91 (pass_store_merging::process_store): Compute n and ins_stmt, if
92 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
93 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
95 (pass_store_merging::execute): Calculate dominators.
97 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
98 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
99 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
100 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
101 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
102 class pass_optimize_bswap, bswap_replace,
103 pass_optimize_bswap::execute): Moved to ...
104 * gimple-ssa-store-merging.c: ... this file.
105 Include optabs-tree.h.
106 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
107 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
108 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
109 anonymous namespace, remove static keywords.
110 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
111 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
113 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
116 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
118 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
120 * vec.h (debug_helper): New function.
121 (DEFINE_DEBUG_VEC): New macro.
122 * hash-set.h (debug_helper): New function.
123 (DEFINE_DEBUG_HASH_SET): New macro.
124 * cfg.c (debug_slim (edge)): New function.
125 Call DEFINE_DEBUG_VEC for edges.
126 Call DEFINE_DEBUG_HASH_SET for edges.
127 * cfghooks.c (debug_slim (basic_block)): New function.
128 Call DEFINE_DEBUG_VEC for basic blocks.
129 Call DEFINE_DEBUG_HASH_SET for basic blocks.
130 * print-tree.c (debug_slim): New function to handle trees.
131 Call DEFINE_DEBUG_VEC for trees.
132 Call DEFINE_DEBUG_HASH_SET for trees.
133 (debug (vec<tree, va_gc>) &): Remove.
134 (debug (<vec<tree, va_gc>) *): Remove.
135 * print-rtl.c (debug_slim): New function to handle const_rtx.
136 Call DEFINE_DEBUG_VEC for rtx_def.
137 Call DEFINE_DEBUG_VEC for rtx_insn.
138 Call DEFINE_DEBUG_HASH_SET for rtx_def.
139 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
140 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
141 (debug (vec<rtx_insn *> *ptr): Remove.
142 (debug_insn_vector): Remove.
143 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
145 2017-11-20 Tom de Vries <tom@codesourcery.com>
147 PR rtl-optimization/82020
148 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
149 IF_THEN_ELSE condition.
151 2017-11-19 Jeff Law <law@redhat.com>
153 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
154 of degenerates resulting from ignoring an edge.
156 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
159 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
161 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
164 * profile-count.c (profile_count::to_sreal_scale): Fix return value
165 for uninitialied counts.
167 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
170 * tree-inline.c (estimate_num_insns): Set to 1 at least.
172 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
175 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
177 2017-11-19 Tom de Vries <tom@codesourcery.com>
179 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
182 2017-11-19 Tom de Vries <tom@codesourcery.com>
184 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
187 2017-11-19 Tom de Vries <tom@codesourcery.com>
189 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
192 2017-11-19 Tom de Vries <tom@codesourcery.com>
194 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
196 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
198 2017-11-19 Tom de Vries <tom@codesourcery.com>
200 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
202 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
203 semicolon after MCORE_EXPORT_NAME call.
205 2017-11-19 Tom de Vries <tom@codesourcery.com>
208 * vmsdbgout.c (vmsdbgout_early_finish): New function.
209 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
211 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
213 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
214 (cgraph_edge::clone): Cleanup updating of profile.
215 * ipa-cp.c (update_profiling_info): Likewise.
216 * ipa-inline-transform.c (inline_transform): Likewise.
217 * ipa-inline.c (inline_small_functions): Add missing space to dump.
218 * ipa-split.c (execute_split_functions): Do not split when function
220 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
221 * profile-count.c (profile_count::dump): Add global0.
222 (profile_count::to_cgraph_frequency): Do not ICE when entry is
224 (profile_count::to_sreal_scale): Likewise.
225 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
226 (profile_count::combine_with_ipa_count): New function.
227 * profile-count.h (profile_guessed_global0adjusted): New.
228 (profile_count::adjusted_zero): New.
229 (profile_count::global0adjusted): New.
230 (profile_count::combine_with_ipa_count): New.
231 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
232 correct profile of return block of split functions.
233 (copy_cfg_body): Remove unused profile_count.
234 (copy_body): Likewise.
235 (expand_call_inline): Update.
236 (tree_function_versioning): Update.
238 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
240 * hash-set.h (hash_set::empty): New.
241 * tree-ssa-threadbackward.h: Delete.
242 * tree-ssa-threadbackward.c (class thread_jumps): New.
243 Move max_threaded_paths into class.
244 (fsm_find_thread_path): Remove arguments that are now in class.
245 (profitable_jump_thread_path): Rename to...
246 (thread_jumps::profitable_jump_thread_path): ...this.
247 (convert_and_register_jump_thread_path): Rename to...
248 (thread_jumps::convert_and_register_current_path): ...this.
249 (check_subpath_and_update_thread_path): Rename to...
250 (thread_jumps::check_subpath_and_update_thread_path): ...this.
251 (register_jump_thread_path_if_profitable): Rename to...
252 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
253 (handle_phi): Rename to...
254 (thread_jumps::handle_phi): ...this.
255 (handle_assignment): Rename to...
256 (thread_jumps::handle_assignment): ...this.
257 (fsm_find_control_statement_thread_paths): Rename to...
258 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
259 (find_jump_threads_backwards): Rename to...
260 (thread_jumps::find_jump_threads_backwards): ...this.
261 Initialize path local data.
262 (pass_thread_jumps::execute): Call find_jump_threads_backwards
263 from within thread_jumps class.
264 (pass_early_thread_jumps::execute): Same.
266 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
268 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
270 2017-11-17 Steve Ellcey <sellcey@cavium.com>
272 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
273 to canonical location.
275 2017-11-17 Steve Ellcey <sellcey@cavium.com>
278 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
280 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
282 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
284 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
285 rcrt1.o%s/grcrt1.o%s for -static-pie.
287 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
289 * i386.c (ix86_multiplication_cost, ix86_division_cost,
290 ix86_shift_rotate_cost): Break out from ...
291 (ix86_rtx_costs): ... here.
292 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
295 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
297 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
298 when entry block was promoted unlikely.
299 (estimate_bb_frequencies): Increase frequency scale.
300 * profile-count.h (profile_count): Export precision info.
302 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
304 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
305 disturbin profile of entry block.
307 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
309 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
312 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
314 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
315 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
317 2017-11-17 Jeff Law <law@redhat.com>
319 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
320 from evrp_dom_walker class. Various methods moved into new class.
321 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
322 (evrp_range_analyzer::enter): New method.
323 (evrp_range_analyzer::leave): New method.
324 (evrp_dom_walker): Remove delegators no longer needed by this class.
325 Replace vr_values data member with evrp_range_analyzer
327 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
328 method extracted from evrp_dom_walker::before_dom_children.
329 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
330 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
332 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
333 Add private copy constructor and move assignment operators.
334 Privatize methods and class data where trivially possible.
335 (evrp_dom_walker::cleanup): New function, extracted from
336 execute_early_vrp. Simplify access to class data.
338 * vr-values.h (get_output_for_vrp): Prototype.
339 * vr-values.c (get_output_for_vrp): New function extracted from
340 vrp_visit_assignment_or_call and extract_range_from_stmt.
341 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
343 2017-11-17 Luis Machado <luis.machado@linaro.org>
345 * config/aarch64/aarch64.c
346 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
347 (qdf24xx_tunings) <autoprefetcher_model>: Set to
348 tune_params::AUTOPREFETCHER_WEAK.
350 2017-11-17 Tamar Christina <tamar.christina@arm.com>
353 * config/arm/arm.c (arm_valid_target_attribute_rec):
354 Parse "arch=" and "+<ext>".
355 (arm_valid_target_attribute_tree): Re-init global options.
356 (arm_option_override): Make non-static.
357 (arm_options_perform_arch_sanity_checks): Make errors fatal.
358 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
359 (__ARM_FEATURE_CRC32): Support undef.
360 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
361 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
363 2017-11-17 David Malcolm <dmalcolm@redhat.com>
365 * gdbinit.in (break-on-diagnostic): New command.
367 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
369 * config/i386/i386.c (ix86_expand_epilogue): Change simple
370 return to indirect jump for EH return if control-flow
371 protection is enabled. Change explicit 'false' argument in
372 pro_epilogue_adjust_stack with a value of
374 * config/i386/i386.md (simple_return_indirect_internal):
375 Remove SImode restriction to support 64-bit.
377 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
379 * combine.c (added_notes_insn): New.
380 (try_combine): Handle added_notes_insn like added_links_insn.
381 Rewrite return value code.
382 (distribute_notes): Set added_notes_insn to the earliest insn we added
385 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
387 PR rtl-optimization/82621
388 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
389 dest of one of those SETs is unused.
391 2017-11-17 Richard Biener <rguenther@suse.de>
394 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
395 * tree-pretty-print.c (dump_generic_node): Handle
396 annot_expr_parallel_kind.
397 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
398 * gimplify.c (gimple_boolify): Likewise.
400 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
402 * config.gcc (extra_headers): Add cet.h for x86 targets.
403 * config/i386/cet.h: New file.
404 * doc/install.texi: Add --enable-cet/--disable-cet.
406 2017-11-17 Richard Biener <rguenther@suse.de>
408 PR tree-optimization/83017
409 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
410 (gen_parallel_loop): Properly count iterations.
411 (parallelize_loops): Handle loop->can_be_parallel independent
412 of flag_loop_parallelize_all. Make static profitability test match
414 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
415 * invoke.texi (parloops-min-per-thread): Document.
417 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
419 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
420 upstreaming review comments.
422 2017-11-17 Tamar Christina <tamar.christina@arm.com>
424 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
425 with fast unaligned access.
426 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
428 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
430 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
431 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
433 2017-11-17 Richard Biener <rguenther@suse.de>
435 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
436 folding of references.
438 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
441 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
442 of replacing call to strncmp with corresponding call to strcmp when
445 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
447 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
448 option prefer-avx256 for skylake-avx512 configuration.
449 * config/i386/i386.c (ix86_option_override_internal): Ditto.
450 (get_builtin_code_for_version): Ditto.
452 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
453 Monk Chiang <sh.chiang04@gmail.com>
455 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
456 (FIXED_REGISTERS): Reserve more register numbers.
457 (CALL_USED_REGISTERS): Likewise.
458 (REG_ALLOC_ORDER): Likewise.
459 (REG_CLASS_CONTENTS): Likewise.
460 (REGISTER_NAMES): Likewise.
462 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
463 Kito Cheng <kito.cheng@gmail.com>
465 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
467 * config/nds32/iterators.md: Add vector mode iterators and attributes.
469 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
471 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
472 parameter list for vec_splats.
474 2017-11-16 Joseph Myers <joseph@codesourcery.com>
476 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
478 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
480 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
483 * doc/install.texi: Document incompatibility of Perl >=5.6.26
484 with the required version of automake 1.11.6.
486 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
488 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
490 (DU_C2_3_power9): Correct reservation combinations.
491 (FP_DIV_power9, VEC_DIV_power9): New.
492 (power9-alu): Split out rotate/shift...
493 (power9-rot): ...to here, correct dispatch resource.
494 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
496 (power9-fp): Correct latency.
497 (power9-sdiv): Add div/sqrt resource.
498 (power9-ddiv): Correct latency, add div/sqrt resource.
499 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
500 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
502 (power9-qpdiv, power9-qpmul): Adjust resource usage.
504 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
506 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
507 switch statement mapping KF built-ins to TF built-ins if we don't
508 have the proper ISA 3.0 assembler support.
510 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
512 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
513 (lower_emutls_function_body): Do not compute it.
515 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
517 * ipa-split.c (split_bb_info): Turn time to sreal.
518 (split_point): Likewise.
519 (dump_split_point): Likewise.
520 (fine_split_points): Likewise.
521 (execute_split_functions): Only zero split_bbs; turn time to sreals.
523 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
525 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
528 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
530 * predict.c (combine_predictions_for_bb): Preserve zero predicted
532 (expensive_function_p): Remove useless assert.
533 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
535 2017-11-16 Martin Sebor <msebor@redhat.com>
537 PR tree-optimization/82588
538 PR tree-optimization/82583
539 * tree-vrp.c (check_array_ref): Handle flexible array members,
540 string literals, and inner indices.
541 (search_for_addr_array): Add detail to diagnostics.
543 2017-11-16 Nathan Sidwell <nathan@acm.org>
547 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
548 (SET_DECL_ASSEMBLER_NAME): Forward to
549 overwrite_decl_assembler_name.
550 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
551 (overwrite_decl_assembler_name): Declare.
552 * tree.c (overwrite_decl_assembler_name): New.
553 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
554 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
555 (LANG_HOOKS_INITIALIZER): Add it.
556 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
557 * langhooks.c (lhd_set_decl_assembler_name): Use
558 SET_DECL_ASSEMBLER_NAME.
559 (lhd_overwrite_decl_assembler_name): Default implementation.
561 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
562 Jackson Woodruff <jackson.woodruff@arm.com>
564 PR tree-optimization/71026
565 * match.pd: Canonicalize constant multiplies in division.
567 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
569 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
570 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
571 Remove OPT_fomit_frame_pointer entry.
572 * common/config/arc/arc-common.c: Likewise.
573 * common/config/arm/arm-common.c: Likewise.
574 * common/config/avr/avr-common.c: Likewise.
575 * common/config/c6x/c6x-common.c: Likewise.
576 * common/config/cr16/cr16-common.c: Likewise.
577 * common/config/cris/cris-common.c: Likewise.
578 * common/config/epiphany/epiphany-common.c: Likewise.
579 * common/config/fr30/fr30-common.c: Likewise.
580 * common/config/frv/frv-common.c: Likewise.
581 * common/config/ia64/ia64-common.c: Likewise.
582 * common/config/iq2000/iq2000-common.c: Likewise.
583 * common/config/lm32/lm32-common.c: Likewise.
584 * common/config/m32r/m32r-common.c: Likewise.
585 * common/config/mcore/mcore-common.c: Likewise.
586 * common/config/microblaze/microblaze-common.c: Likewise.
587 * common/config/mips/mips-common.c: Likewise.
588 * common/config/mmix/mmix-common.c: Likewise.
589 * common/config/mn10300/mn10300-common.c: Likewise.
590 * common/config/nios2/nios2-common.c: Likewise.
591 * common/config/pa/pa-common.c: Likewise.
592 * common/config/pdp11/pdp11-common.c: Likewise.
593 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
594 * common/config/riscv/riscv-common.c: Likewise.
595 * common/config/rs6000/rs6000-common.c: Likewise.
596 * common/config/rx/rx-common.c: Likewise.
597 * common/config/s390/s390-common.c: Likewise.
598 * common/config/sh/sh-common.c: Likewise.
599 * common/config/sparc/sparc-common.c: Likewise.
600 * common/config/tilegx/tilegx-common.c: Likewise.
601 * common/config/tilepro/tilepro-common.c: Likewise.
602 * common/config/v850/v850-common.c: Likewise.
603 * common/config/visium/visium-common.c: Likewise.
604 * common/config/xstormy16/xstormy16-common.c: Likewise.
605 * common/config/xtensa/xtensa-common.c: Likewise.
606 * invoke.texi (-fomit-frame-pointer): Update documentation.
608 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
610 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
612 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
614 * cfg.c (scale_bbs_frequencies_int,
615 cale_bbs_frequencies_gcov_type): Remove.
616 * cfg.h (scale_bbs_frequencies_int,
617 cale_bbs_frequencies_gcov_type): Remove.
619 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
621 * tree-ssa-loop-manip.c
622 (scale_dominated_blocks_in_loop): Update to profile counts.
623 (tree_transform_and_unroll_loop): Likewise.
625 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
627 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
628 scale_bbs_frequencies_int.
630 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
632 * final.c (compute_alignments): Use counts rather than frequencies.
634 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
636 * cfgloopanal.c: Include sreal.h
637 (average_num_loop_insns): Use counts and sreal for accounting.
639 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
641 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
644 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
645 Kito Cheng <kito.cheng@gmail.com>
647 * config/nds32/constraints.md: Provide more constraints.
648 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
649 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
650 support constraints usage.
652 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
654 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
656 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
657 Kito Cheng <kito.cheng@gmail.com>
659 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
660 * config/nds32/nds32.opt: Refine the layout.
661 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
662 TARGET_EXT_STRING): Support new options.
663 * config/nds32/nds32.h: Likewise.
664 * config/nds32/nds32.md: Likewise.
665 * config/nds32/nds32-predicates.c: Likewise.
666 * config/nds32/constraints.md: Likewise.
667 * common/config/nds32/nds32-common.c: Likewise.
669 2017-11-16 Julia Koval <julia.koval@intel.com>
672 * config/i386/gfniintrin.h: Add sse check.
673 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
675 2017-11-16 Julia Koval <julia.koval@intel.com>
677 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
678 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
679 (ix86_handle_option): Handle -mavx512vbmi2.
680 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
681 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
682 * config/i386/i386-c.c (__AVX512VBMI2__): New.
683 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
684 (ix86_valid_target_attribute_inner_p): Ditto.
685 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
686 * config/i386/i386.opt (mavx512vbmi2): New option.
687 * doc/invoke.texi: Add new option.
689 [2017-11-16 Julia Koval <julia.koval@intel.com>
691 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
692 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
693 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
694 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
695 _mm512_gf2p8mul_epi8): New intrinsics.
696 * config/i386/i386-builtin-types.def
697 (V64QI_FTYPE_V64QI_V64QI): New type.
698 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
699 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
700 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
701 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
702 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
703 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
705 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
707 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
708 explicitly as a stream of bytes.
710 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
712 * config/rs6000/altivec.h (vec_xst_be): New #define.
713 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
714 and externalize from *altivec_vperm_<mode>_internal.
715 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
717 (XL_BE_V8HI): Likewise.
718 (XL_BE_V4SI): Likewise.
719 (XL_BE_V4SI): Likewise.
720 (XL_BE_V2DI): Likewise.
721 (XL_BE_V4SF): Likewise.
722 (XL_BE_V2DF): Likewise.
723 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
724 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
725 all array entries with these keys: VSX_BUILTIN_VEC_XL,
726 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
727 VSX_BUILTIN_VEC_XST_BE.
728 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
729 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
731 (altivec_init_builtins): Replace conditional calls to def_builtin
732 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
733 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
734 with unconditional calls. Remove calls to def_builtin for
735 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
736 __builtin_vec_xst_be.
737 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
738 to define_expand, and add alternate RTL generation for P8.
739 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
741 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
742 add alternate RTL generation for P8.
743 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
744 vsx_ld_elemrev_v16qi.
745 (vsx_st_elemrev_v8hi): Convert define_insn
746 to define_expand, and add alternate RTL generation for P8.
747 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
749 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
750 add alternate RTL generation for P8.
751 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
752 vsx_st_elemrev_v16qi.
754 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
757 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
758 TARGET_AVX512ER check.
759 (ix86_option_override_internal): Set MASK_VZEROUPPER if
760 neither -mzeroupper nor -mno-zeroupper is used and
761 TARGET_EMIT_VZEROUPPER is set.
762 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
763 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
765 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
767 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
768 folding of vector compares.
769 (fold_build_vec_cmp): New helper function.
770 (fold_compare_helper): New helper function.
771 (builtin_function_type): Add compare builtins to the list of functions
772 having unsigned arguments. Cosmetic updates to comment indentation.
773 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
774 the not+eq combination.
776 2017-11-15 Bin Cheng <bin.cheng@arm.com>
778 PR tree-optimization/82726
779 PR tree-optimization/70754
780 * tree-predcom.c (order_drefs_by_pos): New function.
781 (combine_chains): Move code setting has_max_use_after to...
782 (try_combine_chains): ...here. New parameter. Sort combined chains
783 according to position information.
784 (tree_predictive_commoning_loop): Update call to above function.
785 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
787 2017-11-15 Bin Cheng <bin.cheng@arm.com>
789 PR tree-optimization/82726
791 2017-01-23 Bin Cheng <bin.cheng@arm.com>
793 PR tree-optimization/70754
794 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
795 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
796 combined stmt before it if not NULL.
797 (combine_chains): Process refs reversely and compute dominance point
801 2017-02-23 Bin Cheng <bin.cheng@arm.com>
803 PR tree-optimization/79663
804 * tree-predcom.c (combine_chains): Process refs in reverse order
805 only for ZERO length chains, and add explaining comment.
807 2017-11-15 Tamar Christina <tamar.christina@arm.com>
809 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
810 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
811 * doc/invoke.texi (ARM Options): Add armv8.3-a.
813 2017-11-15 Tamar Christina <tamar.christina@arm.com>
815 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
817 2017-11-15 Martin Liska <mliska@suse.cz>
819 * tree-cfg.c (pass_warn_function_return::execute):
820 Compare warn_return_type for greater than zero.
822 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
826 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
827 to return true on Xeon and not on Xeon Phi.
828 (ix86_check_avx256_register): Changed to ...
829 (ix86_check_avx_upper_register): ... this. Add extra check for
830 VALID_AVX512F_REG_OR_XI_MODE.
831 (ix86_avx_u128_mode_needed): Changed
832 ix86_check_avx256_register to ix86_check_avx_upper_register.
833 (ix86_check_avx256_stores): Changed to ...
834 (ix86_check_avx_upper_stores): ... this. Changed
835 ix86_check_avx256_register to ix86_check_avx_upper_register.
836 (ix86_avx_u128_mode_after): Changed
837 avx_reg256_found to avx_upper_reg_found. Changed
838 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
839 (ix86_avx_u128_mode_entry): Changed
840 ix86_check_avx256_register to ix86_check_avx_upper_register.
841 (ix86_avx_u128_mode_exit): Ditto.
842 * config/i386/i386.h: (host_detect_local_cpu): New define.
844 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
846 * config/arm/xgene1.md (xgene1): Split into automatons
847 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
848 (xgene1_f_load): Adjust reservations and/or types.
849 (xgene1_f_store): Likewise.
850 (xgene1_load_pair): Likewise.
851 (xgene1_store_pair): Likewise.
852 (xgene1_fp_load1): Likewise.
853 (xgene1_load1): Likewise.
854 (xgene1_store1): Likewise.
855 (xgene1_move): Likewise.
856 (xgene1_alu): Likewise.
857 (xgene1_simd): Likewise.
858 (xgene1_bfm): Likewise.
859 (xgene1_neon_load1): Likewise.
860 (xgene1_neon_store1): Likewise.
861 (xgene1_neon_logic): Likewise.
862 (xgene1_neon_st1): Likewise.
863 (xgene1_neon_ld1r): Likewise.
864 (xgene1_alu_cond): Added.
865 (xgene1_shift_reg): Likwise.
866 (xgene1_bfx): Likewise.
867 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
869 2017-11-15 Jakub Jelinek <jakub@redhat.com>
872 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
874 (can_widen_mult_without_libcall): New function.
875 (expand_mul_overflow): If only checking unsigned mul overflow,
876 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
877 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
878 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
879 (expand_DIVMOD): Formatting fix.
880 * expmed.h (expand_mult): Add NO_LIBCALL argument.
881 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
882 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
884 PR tree-optimization/82977
885 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
886 constructed temporary to strlen_to_stridx.put.
888 2017-11-15 Martin Liska <mliska@suse.cz>
890 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
891 * configure: Regenerate.
893 2017-11-15 Martin Liska <mliska@suse.cz>
896 * config/sh/sh-mem.cc: Use proper probability for
899 2017-11-14 Jeff Law <law@redhat.com>
901 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
902 the red zone for stack_clash_protection_final_dynamic_probe targets
903 when the total dynamic stack size is zero bytes.
905 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
906 blocks is post order.
908 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
910 * dumpfile.h (TDF_COMPARE_DEBUG): New.
911 * final.c (rest_of_clean_state): Set it for the
912 -fcompare-debug dump.
913 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
914 class when TDF_COMPARE_DEBUG is set.
916 * dwarf2out.c (gen_producer_string): Discard
919 2017-11-15 Joseph Myers <joseph@codesourcery.com>
922 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
923 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
924 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
925 (__TGMATH_REAL_2_3): Remove macros.
926 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
927 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
929 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
931 (remquo): Define using __TGMATH_REAL_3.
933 2017-11-14 Jeff Law <law@redhat.com>
935 * vr-values.c: New file with contents extracted from tree-vrp.c.
936 * Makefile.in (OBJS): Add vr-values.o
937 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
938 (set_value_range, set_and_canonicalize_value_range): Likewise.
939 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
940 (value_range_constant_singleton, symbolic_range_p): Likewise.
941 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
942 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
943 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
944 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
945 (find_case_label_range, find_case_label_index): Likewise.
946 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
947 (range_int_cst_singleton_p, value_inside_range): Likewise.
948 (get_single_symbol): Likewise.
949 (switch_update): Move structure definition here.
950 (to_remove_edges, to_update_switch_stmts): Provide externs.
951 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
952 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
953 (vrp_val_is_min, set_value_range): Likewise.
954 (set_and_canonicalize_value_range, copy_value_range): Likewise.
955 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
956 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
957 (range_is_nonnull, range_int_cst_p): Likewwise.
958 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
959 (get_single_symbol, operand_less_p): Likewise
960 (compare_values_warnv, compare_values): Likewise.
961 (value_inside_range, value_range_constant_singleton): Likewise.
962 (zero_nonzero_bitgs_from_vr): Likewise.
963 (extract_range_from_binary_expr_1): Likewise.
964 (overflow_comparison_p): Likewise.
965 (to_remove_edges, to_update_switch_stmts): Likewise.
966 (find_case_label-index, find_case_label_range): Likewise.
967 (switch_update, set_value_range_to_nonnegative): Remove.
968 (set_value_range_to_truthvalue): Likewise.
969 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
970 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
971 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
972 (find_case_label_ranges, test_for_singularity): Likewise.
973 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
974 (x_vr_values): Move to its remaining use site.
976 2017-11-10 Jeff Law <law@redhat.com>
978 * vr-values.h (VR_INITIALIZER): Move #define here.
979 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
980 * Makefile.in (OBJS): Add tree-evrp.o
981 * tree-vrp.h (assert_info): Move structure definition here.
982 (set_value_range_to_varying): Prototype.
983 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
984 (infer_value_range, register_edge_assert_for): Likewise.
985 (stmt_interesting_for_vrp): Likewise.
986 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
987 (set_value_range_to_varying): No longer static.
988 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
989 (infer_value_range, register_edge_assert_for): Likewise.
991 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
993 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
994 generate the XXBRD instruction.
996 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
997 (rs6000_builtin_type_compatible): Treat _Float128 and long double
998 as being compatible if -mabi=ieeelongdouble.
999 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1000 to setup float128 built-ins with hardware support.
1001 (BU_FLOAT128_HW_2): Likewise.
1002 (BU_FLOAT128_HW_3): Likewise.
1003 (BU_FLOAT128_HW_VSX_1): Likewise.
1004 (BU_FLOAT128_HW_VSX_2): Likewise.
1005 (scalar_extract_expq): Change float128 built-in functions to
1006 accommodate having both KFmode and TFmode functions. Use the
1007 KFmode variant as the default.
1008 (scalar_extract_sigq): Likewise.
1009 (scalar_test_neg_qp): Likewise.
1010 (scalar_insert_exp_q): Likewise.
1011 (scalar_insert_exp_qp): Likewise.
1012 (scalar_test_data_class_qp): Likewise.
1013 (sqrtf128_round_to_odd): Delete processing the round to odd
1014 built-in functions as special built-in functions, and define them
1015 as float128 built-ins. Use the KFmode variant as the default.
1016 (truncf128_round_to_odd): Likewise.
1017 (addf128_round_to_odd): Likewise.
1018 (subf128_round_to_odd): Likewise.
1019 (mulf128_round_to_odd): Likewise.
1020 (divf128_round_to_odd): Likewise.
1021 (fmaf128_round_to_odd): Likewise.
1022 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1023 support for KFmode and TFmode xststdcqp calls.
1024 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1025 point, switch the built-in handlers for the get/set float128
1026 exponent, get float128 mantissa, float128 test built-ins, and the
1027 float128 round to odd built-in functions. Eliminate creating the
1028 float128 round to odd built-in functions as special built-ins.
1029 (rs6000_init_builtins): Eliminate special creation of the float128
1030 round to odd built-in functions.
1031 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1032 function insns to support both TFmode and KFmode variants.
1033 (xsxsigqp_<mode>): Likewise.
1034 (xsiexpqpf_<mode>): Likewise.
1035 (xsiexpqp_<mode>): Likewise.
1036 (xststdcqp_<mode>): Likewise.
1037 (xststdcnegqp_<mode>): Likewise.
1038 (xststdcqp_<mode>): Likewise.
1040 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1042 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1043 unused path_in_freq_ptr parameter.
1044 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1046 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1048 * ipa-inline.c (edge_badness): Dump sreal frequency.
1049 (compute_inlined_call_time): Match natural implementaiton ...
1050 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1051 forgotten division by CGRAPH_FREQ_BASE.
1053 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1055 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1056 Solaris 11. Update comment.
1057 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1059 * config/sol2.h (STARTFILE_SPEC): Likewise.
1060 * configure: Regenerate.
1062 2017-11-14 Carl Love <cel@us.ibm.com>
1064 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1065 le_ and be_ prefixes to swap* variables. Remove
1066 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1069 2017-11-14 Jason Merrill <jason@redhat.com>
1071 Support GTY((cache)) on hash_map.
1072 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1073 (ggc_cache_remove): Override it instead of ggc_mx.
1074 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1075 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1076 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1077 (simple_cache_map_traits): Override maybe_mx.
1078 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1079 (hash_map): Friend gt_cleare_cache.
1080 (gt_cleare_cache): New.
1081 * tree.h (tree_cache_traits): New hash_map traits class.
1082 (tree_cache_map): New typedef.
1084 2017-11-14 Richard Biener <rguenther@suse.de>
1086 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1087 paramter and handling.
1088 (cleanup_control_flow_bb): Likewise.
1089 (cleanup_control_flow_pre): New helper performing a DFS walk
1090 to call cleanup_control_flow_bb in PRE order.
1091 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1092 via cleanup_control_flow_pre.
1094 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1096 * config/aarch64/aarch64-simd.md
1097 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1098 (*aarch64_simd_bsl<mode>_alt): Likewise.
1099 (aarch64_simd_bsldi_internal): New.
1100 (aarch64_simd_bsldi_alt): Likewise.
1102 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1104 * tracer.c (better_p): Do not compare frequencies.
1105 * reg-stack.c (better_edge): Likewise.
1106 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1109 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1111 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1112 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1113 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1114 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1115 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1116 * predict.c (maybe_hot_frequency_p): Inline to ...
1117 (maybe_hot_count_p): ... here; rewrite to counts.
1118 (counts_to_freqs): Rename to ...
1119 (update_max_bb_count): ... this one.
1120 (expensive_function_p): Use counts.
1121 (estimate_bb_frequencies): Update.
1122 (rebuild_frequencies): Update.
1123 * predict.h (counts_to_freqs): Rename to ...
1124 (update_max_bb_count): ... this one.
1125 * profile.c (compute_branch_probabilities): Add debug info
1126 * tree-inline.c (expand_call_inline): Update debug info.
1127 (optimize_inline_calls): Use update_max_bb_count..
1128 (tree_function_versioning): Use update_max_bb_count..
1129 * value-prof.c (gimple_value_profile_transformations):
1130 Do not use update_max_bb_count.
1132 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1134 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1135 always use frequencies.
1137 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1139 * bb-reorder.c: Remove frequencies from comments.
1140 (better_edge_p): Use profile counts.
1141 (find_traces): Dump profile counts.
1142 (rotate_loop): Use profile counts.
1143 (find_traces_1_round): Likewise.
1144 (connect_better_edge_p): Use counts instead of probabilities for
1146 (copy_bb_p): Drop early check for non-0 frequency.
1147 (sanitize_hot_paths): Update comments.
1149 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1151 * ipa-split.c (struct split_point): Add count.
1152 (consider_split): Do not compute incoming frequency; compute incoming
1153 count and store it to split_point.
1154 (split_function): Set count of the call to split part correctly.
1156 2017-11-13 Carl Love <cel@us.ibm.com>
1158 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1160 2017-11-13 Tom Tromey <tom@tromey.com>
1162 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1164 2017-11-13 Carl Love <cel@us.ibm.com>
1166 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1167 Add support for builtins:
1168 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1169 vector {un,}signed {char,int,short},
1170 vector {un,}signed {char,int,short}) arguments.
1171 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1172 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1173 Add BU_P9V_AV_2 expansions for the builtins.
1174 * config/rs6000/altivec.h (vec_first_match_index,
1175 vec_first_mismatch_index, vec_first_match_or_eos_index,
1176 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1177 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1178 new extern declaration.
1179 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1180 * config/rs6000/vsx.md (first_match_index_<mode>,
1181 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1182 first_mismatch_or_eos_index_<mode>): Add define expand.
1183 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1184 * doc/extend.texi: Update the built-in documenation file for the new
1187 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1189 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1190 into the min/max operations for _Float<N> and _Float<N>X types.
1192 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1195 * dwarf2out.c (do_eh_frame): New static variable.
1196 (dwarf2out_begin_prologue): Set it.
1197 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1199 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1201 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1203 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1205 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1208 * debug.h (dwarf2out_do_eh_frame): Declare.
1209 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1210 (dwarf2out_do_frame): Use it.
1211 (dwarf2out_do_cfi_asm): Likewise.
1212 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1213 (dwarf2out_assembly_start): Likewise.
1214 (dwarf2out_begin_prologue): Fix comment.
1215 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1216 if the target needs either debug or unwind DWARF2 info.
1217 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1218 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1219 -ftrapping-math, -ftrapv and -fwrapv.
1221 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1223 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1224 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1225 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1226 (estimate_edge_size_and_time): Likewise.
1227 (ipa_merge_fn_summary_after_inlining): Likewise.
1228 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1229 (compute_uninlined_call_time): Use sreal_frequency.
1230 (compute_inlined_call_time): Likewise.
1231 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1232 * profile-count.c: Include sreal.h.
1233 (profile_count::to_sreal_scale): New.
1234 * profile-count.h: Forward declare sreal.
1235 (profile_count::to_sreal_scale): Declare.
1237 2017-11-13 Nathan Sidwell <nathan@acm.org>
1239 * diagnostic.c (maybe_line_and_column): New.
1240 (diagnostic_get_location_text): Use it.
1241 (diagnostic_report_current_module): Likewise.
1242 (test_diagnostic_get_location_text): Add tests.
1244 2017-11-13 Luis Machado <luis.machado@linaro.org>
1246 * doc/md.texi (Specifying processor pipeline description): Fix
1247 incorrect latency for the div instruction example.
1249 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1251 PR tree-optimization/78821
1252 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1253 that bit_not_p is the same.
1254 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1255 (split_group): Count precisely bit_not_p bits in each statement.
1256 (invert_op): New function.
1257 (imm_store_chain_info::output_merged_store): Use invert_op to
1258 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1259 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1261 2017-11-13 Martin Liska <mliska@suse.cz>
1263 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1264 (struct source_info): Likewise.
1265 (add_branch_counts): Likewise.
1266 (add_line_counts): Likewise.
1267 (function_summary): Likewise.
1268 (output_intermediate_line): Likewise.
1269 (generate_results): Likewise.
1271 2017-11-13 Martin Liska <mliska@suse.cz>
1273 * gcov.c (struct block_info): Remove typedef for block_t.
1274 (struct line_info): Likewise.
1275 (line_info::has_block): Likewise.
1276 (EXIT_BLOCK): Likewise.
1277 (unblock): Likewise.
1278 (circuit): Likewise.
1279 (get_cycles_count): Likewise.
1280 (process_file): Likewise.
1281 (read_graph_file): Likewise.
1282 (solve_flow_graph): Likewise.
1283 (find_exception_blocks): Likewise.
1284 (add_line_counts): Likewise.
1285 (accumulate_line_info): Likewise.
1286 (output_line_details): Likewise.
1288 2017-11-13 Martin Liska <mliska@suse.cz>
1290 * gcov.c (struct arc_info): Remove typedef for arc_t.
1291 (struct line_info): Likewise.
1292 (add_branch_counts): Likewise.
1293 (output_branch_count): Likewise.
1294 (function_info::~function_info): Likewise.
1295 (circuit): Likewise.
1296 (output_intermediate_line): Likewise.
1297 (read_graph_file): Likewise.
1298 (solve_flow_graph): Likewise.
1299 (find_exception_blocks): Likewise.
1300 (add_line_counts): Likewise.
1301 (accumulate_line_info): Likewise.
1302 (output_line_details): Likewise.
1303 (output_function_details): Likewise.
1305 2017-11-13 Martin Liska <mliska@suse.cz>
1307 * gcov.c (struct function_info): Remove typedef for function_t.
1308 (struct source_info): Likewise.
1309 (source_info::get_functions_at_location): Likewise.
1310 (solve_flow_graph): Likewise.
1311 (find_exception_blocks): Likewise.
1312 (add_line_counts): Likewise.
1313 (output_intermediate_file): Likewise.
1314 (process_file): Likewise.
1315 (generate_results): Likewise.
1316 (release_structures): Likewise.
1317 (read_graph_file): Likewise.
1318 (read_count_file): Likewise.
1319 (accumulate_line_counts): Likewise.
1320 (output_lines): Likewise.
1322 2017-11-13 Martin Liska <mliska@suse.cz>
1324 * gcov.c (function_info::function_info): Remove num_counts
1325 and add vector<gcov_type>.
1326 (function_info::~function_info): Use the vector.
1327 (process_file): Likewise.
1328 (read_graph_file): Likewise.
1329 (read_count_file): Likewise.
1330 (solve_flow_graph): Likewise.
1332 2017-11-13 Martin Liska <mliska@suse.cz>
1334 * gcov.c (function_info::is_artificial): New function.
1335 (process_file): Erase all artificial early.
1336 (generate_results): Skip as all artificial are already
1339 2017-11-13 Martin Liska <mliska@suse.cz>
1341 * gcov.c (read_graph_file): Store to global vector of functions.
1342 (read_count_file): Iterate the vector.
1343 (process_file): Likewise.
1344 (generate_results): Likewise.
1345 (release_structures): Likewise.
1347 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1349 PR tree-optimization/82954
1350 * gimple-ssa-store-merging.c
1351 (imm_store_chain_info::coalesce_immediate_stores): If
1352 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1354 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1356 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1357 Upddate call to ENDIAN_LANE_N.
1358 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1359 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1360 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1361 and use aarch64_endian_lane_rtx.
1362 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1364 2017-11-12 Tom de Vries <tom@codesourcery.com>
1366 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1368 2017-11-12 Tom de Vries <tom@codesourcery.com>
1370 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1372 2017-11-12 Tom de Vries <tom@codesourcery.com>
1374 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1376 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1377 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1378 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1380 2017-11-11 Martin Sebor <msebor@redhat.com>
1383 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1386 * prefic.c (translate_name): Replace strncpy with memcpy to
1387 avoid -Wstringop-truncation.
1389 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1391 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1393 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1395 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1397 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1399 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1402 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1404 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1405 enable generating XXBRH if the value is in a vector register.
1406 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1407 value is in a vector register.
1408 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1409 register to register bswap64's instead of doing the GPR sequence
1410 used on previous machines.
1411 (bswapdi2_xxbrd): New insn.
1412 (bswapdi2_reg): Disallow on ISA 3.0.
1413 (register to register bswap64 splitter): Do not split the insn on
1414 ISA 3.0 systems that use XXBRD.
1416 2017-11-10 Martin Sebor <msebor@redhat.com>
1419 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1421 (find_subframework_file): Same.
1423 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1425 * auto-profile.c (afdo_indirect_call): Drop frequency.
1426 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1427 (cgraph_node::create_edge): Drop frequency argument.
1428 (cgraph_node::create_indirect_edge): Drop frequency argument.
1429 (cgraph_edge::make_speculative): Drop frequency arguments.
1430 (cgraph_edge::resolve_speculation): Do not update frequencies
1431 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1432 (cgraph_node::dump): Check consistency in IPA mode.
1433 (cgraph_edge::maybe_hot_p): Use IPA counter.
1434 (cgraph_edge::verify_count_and_frequency): Rename to ...
1435 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1436 (cgraph_node::verify_node): Update.
1437 * cgraph.h (struct cgraph_edge): Drop frequency.
1438 (cgraph_edge::frequency): New function.
1439 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1441 (cgraph_edge::rebuild_edges): Likewise.
1442 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1443 (duplicate_thunk_for_node): Do not pass frequency.
1444 (cgraph_node::create_clone): Scale only counts.
1445 (cgraph_node::create_virtual_clone): Do not pass frequency.
1446 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1447 (cgraph_node::create_version_clone): Do not pass frequency.
1448 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1449 (cgraph_node::expand_thunk): Do not pass frequency.
1450 (cgraph_node::create_wrapper): Do not pass frequency.
1451 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1453 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1454 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1455 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1456 (gather_caller_stats): Use frequency function.
1457 (ipcp_cloning_candidate_p): Use frequency function.
1458 (ipcp_propagate_stage): Use frequency function.
1459 (get_info_about_necessary_edges): Use frequency function.
1460 (update_profiling_info): Update only IPA profile.
1461 (update_specialized_profile): Use frequency functoin.
1462 (perhaps_add_new_callers): Update only IPA profile.
1463 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1464 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1465 (dump_ipa_call_summary): Use frequency function.
1466 (estimate_edge_size_and_time): Use frequency function.
1467 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1468 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1469 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1470 (update_noncloned_counts): ... ths one; scale counts only.
1471 (clone_inlined_nodes): Do not scale frequency.
1472 (inline_call): Do not pass frequency.
1473 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1474 (compute_inlined_call_time): Use IPA profile.
1475 (want_inline_small_function_p): Use IPA profile.
1476 (want_inline_self_recursive_call_p): Use IPA profile.
1477 (edge_badness): Use IPA profile.
1478 (lookup_recursive_calls): Use IPA profile.
1479 (recursive_inlining): Do not pass frequency.
1480 (resolve_noninline_speculation): Do not update frequency.
1481 (inline_small_functions): Collect max of IPA profile.
1482 (dump_overall_stats): Dump IPA porfile.
1483 (dump_inline_stats): Dump IPA porfile.
1484 (ipa_inline): Collect IPA stats.
1485 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1486 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1487 (ipa_propagate_frequency): Use frequency function.
1488 (ipa_profile): Cleanup.
1489 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1490 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1491 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1492 (input_node): Do not stream frequency.
1493 (input_edge): Do not stream frequency.
1494 (merge_profile_summaries): Scale only IPA profiles.
1495 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1496 * predict.c (drop_profile): Do not recompute frequency.
1497 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1498 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1499 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1500 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1501 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1502 * tree-inline.c (copy_bb): Do not scale frequency.
1503 (expand_call_inline): Do not scale frequency.
1504 (tree_function_versioning): Do not scale frequency.
1505 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1507 2017-11-10 Julia Koval <julia.koval@intel.com>
1509 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1510 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1511 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1512 (_mm256_maskz_gf2p8affine_epi64_epi8)
1513 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1514 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1515 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1516 (__builtin_ia32_vgf2p8affineqb_v32qi)
1517 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1518 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1520 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1524 (arm_option_override): Refactor.
1525 (arm_option_reconfigure_globals): New.
1526 (arm_options_perform_arch_sanity_checks): New.
1527 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1529 (arm_options_perform_arch_sanity_checks): Likewise
1531 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1533 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1534 (power9-qpmul): New.
1535 * rs6000/rs6000.md ("type" attr): Add qmul.
1536 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1537 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1538 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1540 2017-11-10 Martin Sebor <msebor@redhat.com>
1543 * builtins.c (compute_objsize): Handle arrays that
1544 compute_builtin_object_size likes to fail for. Make extern.
1545 * builtins.h (compute_objsize): Declare.
1546 (check_strncpy_sizes): New function.
1547 (expand_builtin_strncpy): Call check_strncpy_sizes.
1548 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1549 -Wstringop-truncation.
1550 (gimple_fold_builtin_strncat): Same.
1551 * gimple.c (gimple_build_call_from_tree): Set call location.
1552 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1553 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1554 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1555 (handle_builtin_strlen): Use strlen_to_stridx.
1556 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1558 Use strlen_to_stridx.
1559 (pass_strlen::execute): Release strlen_to_stridx.
1560 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1561 (-Wstringop-truncation): Document new option.
1563 2017-11-10 Martin Liska <mliska@suse.cz>
1565 PR gcov-profile/82702
1566 * gcov.c (main): Handle intermediate files in a different
1568 (get_gcov_intermediate_filename): New function.
1569 (output_gcov_file): Remove support of intermediate files.
1570 (generate_results): Allocate intermediate file.
1571 (release_structures): Clean-up properly fn_end.
1572 (output_intermediate_file): Start iterating with line 1.
1574 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1576 PR tree-optimization/82929
1577 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1578 ops_swapped_p non-static data member.
1579 (store_immediate_info::store_immediate_info): Clear it.
1580 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1581 ops set ops_swapped_p.
1582 (count_multiple_uses): Handle ops_swapped_p.
1584 2017-11-10 Martin Liska <mliska@suse.cz>
1586 * coverage.c (coverage_init): Stream information about
1587 support of has_unexecuted_blocks.
1588 * doc/gcov.texi: Document that.
1589 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1590 * gcov.c (read_graph_file): Likewise.
1591 (output_line_beginning): Fix a small issue with
1594 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1596 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1597 reference of trivial component.
1599 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1602 * gimple-ssa-store-merging.c
1603 (pass_store_merging::terminate_all_aliasing_chains): For
1604 gimple_store_p stmts also call refs_output_dependent_p.
1606 PR rtl-optimization/82913
1607 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1610 2017-11-09 Jeff Law <law@redhat.com>
1612 * vr-values.h: New file with vr_values class.
1613 * tree-vrp.c: Include vr-values.h
1614 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1615 data objects into the vr_values class.
1616 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1617 (get_value_range): Make it a member function within vr_values class.
1618 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1619 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1620 (op_with_constant_singleton_value_range): Likewise.
1621 (extract_range_for_var_from_comparison_expr): Likewise.
1622 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1623 (extract_range_from_binary_expr): Likewise.
1624 (extract_range_from_unary_expr): Likewise.
1625 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1626 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1627 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1628 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1629 (compare_name_with_value, compare_names): Likewise.
1630 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1631 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1632 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1633 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1634 (extract_range_from_phi_node): Likewise.
1635 (simplify_truth_ops_using_ranges): Likewise.
1636 (simplify_div_or_mod_using_ranges): Likewise.
1637 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1638 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1639 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1640 (simplify_float_conversion_using_ranges): Likewise.
1641 (simplify_internal_call_using_ranges): Likewise.
1642 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1643 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1644 poitner via x_vr_values for calls into gimple folder.
1645 (vrp_initialize_lattice): Make this the vr_values ctor.
1646 (vrp_free_lattice): Make this the vr_values dtor.
1647 (set_vr_value): New function.
1648 (class vrp_prop): Add vr_values data member. Add various member
1649 functions as well as member functions that delegate to vr_values.
1650 (check_array_ref): Make a member function within vrp_prop class.
1651 (search_for_addr_array, vrp_initialize): Likewise.
1652 (vrp_finalize): Likewise. Revamp to avoid direct access to
1653 vr_value, values_propagated, etc.
1654 (check_array_bounds): Extract vrp_prop class instance pointer from
1655 walk info structure. Use it to call member functions.
1656 (check_all_array_refs): Make a member function within vrp_prop class.
1657 Smuggle class instance pointer via walk info structure.
1658 (x_vr_values): New local static.
1659 (vrp_valueize): Use x_vr_values to get class instance.
1660 (vr_valueize_1): Likewise.
1661 (class vrp_folder): Add vr_values data member. Add various member
1662 functions as well as member functions that delegate to vr_values.
1663 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1664 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1665 class instance from vr_values. Use it to call member functions.
1666 (vrp_dom_walker): Add vr_values data member.
1667 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1668 instance via x_vr_values.
1669 (identify_jump_threads): Accept vr_values as argument. Store
1670 it into the walker structure.
1671 (evrp_dom_walker): Add vr_values class data member. Add various
1673 (evrp_dom_walker::try_find_new_range): Use vr_values data
1674 member to access the memory allocator.
1675 (evrp_dom_walker::before_dom_children): Store vr_values class
1676 instance into the vrp_folder class.
1677 (evrp_dom_walker::push_value_range): Rework to avoid direct
1678 access to num_vr_values and vr_value.
1679 (evrp_dom_walker::pop_value_range): Likewise.
1680 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1681 Use vr_values to get to dump_all_value_ranges member function.
1682 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1683 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1684 Pass vr_values class instance down to identify_jump_threads.
1685 Remove call to vrp_free_lattice.
1686 (debug_all_value_ranges): Remove.
1688 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1689 (vrp_folder): Likewise.
1691 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1692 Get it from the existing bitmap instead.
1693 (vrp_intersect_ranges_1): Likewise.
1695 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1697 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1699 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1700 set bit_not_p to it.
1701 (imm_store_chain_info::coalesce_immediate_stores): Break group
1702 if bit_not_p is different.
1703 (count_multiple_uses, split_group,
1704 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1705 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1706 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1707 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1708 to store_immediate_info ctor.
1710 2017-11-09 Jim Wilson <jimw@sifive.com>
1712 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1713 (scan_prog_file): Likewise.
1715 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1717 * bb-reorder.c (max_entry_frequency): Remove.
1718 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1719 connect_traces, push_to_next_round_p): Remove prototypes.
1720 (find_traces_1_round): Use counts only.
1721 (push_to_next_round_p): Likewise.
1722 (find_traces): Likewise.
1723 (rotate_loop): Likewise.
1724 (find_traces_1_round): Likewise.
1725 (connect_traces): Likewise.
1726 (edge_order): Likewise.
1728 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1730 * config/arm/arm.c (output_return_instruction): Add comments to
1731 indicate requirement for cmse_nonsecure_entry return to account
1732 for the size of clearing instruction output here.
1733 (thumb_exit): Likewise.
1734 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1735 return in hardfloat mode.
1737 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1739 * config/rs6000/rs6000.c (machine_function): Add a bool,
1740 "toc_is_wrapped_separately".
1741 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1742 if it wasn't explicitly set or unset, we are optimizing for speed, and
1743 doing separate shrink-wrapping.
1744 (rs6000_get_separate_components): Enable the TOC component if
1745 saving the TOC register in the prologue.
1746 (rs6000_components_for_bb): Handle the TOC component.
1747 (rs6000_emit_prologue_components): Store the TOC register where needed.
1748 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1749 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1751 2017-11-09 Martin Jambor <mjambor@suse.cz>
1753 * ipa-param-manipulation.c: New file.
1754 * ipa-param-manipulation.h: Likewise.
1755 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1756 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1757 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1758 (ipa_parm_adjustment): Likewise.
1759 (ipa_parm_adjustment_vec): Likewise.
1760 (ipa_get_vector_of_formal_parms): Moved declaration to
1761 ipa-param-manipulation.h.
1762 (ipa_get_vector_of_formal_parm_types): Likewise.
1763 (ipa_modify_formal_parameters): Likewise.
1764 (ipa_modify_call_arguments): Likewise.
1765 (ipa_combine_adjustments): Likewise.
1766 (ipa_dump_param_adjustments): Likewise.
1767 (ipa_modify_expr): Likewise.
1768 (ipa_get_adjustment_candidate): Likewise.
1769 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1770 ipa-param-manipulation.c.
1771 (ipa_get_vector_of_formal_parm_types): Likewise.
1772 (ipa_modify_formal_parameters): Likewise.
1773 (ipa_modify_call_arguments): Likewise.
1774 (ipa_modify_expr): Likewise.
1775 (get_ssa_base_param): Likewise.
1776 (ipa_get_adjustment_candidate): Likewise.
1777 (index_in_adjustments_multiple_times_p): Likewise.
1778 (ipa_combine_adjustments): Likewise.
1779 (ipa_dump_param_adjustments): Likewise.
1780 * tree-sra.c: Also include ipa-param-manipulation.h
1781 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1784 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1785 Alan Hayward <alan.hayward@arm.com>
1786 David Sherwood <david.sherwood@arm.com>
1788 * doc/sourcebuild.texi (vect_masked_store): Document.
1790 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1796 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1797 Alan Hayward <alan.hayward@arm.com>
1798 David Sherwood <david.sherwood@arm.com>
1800 * doc/sourcebuild.texi (vect_variable_length): Document.
1802 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1803 Alan Hayward <alan.hayward@arm.com>
1804 David Sherwood <david.sherwood@arm.com>
1806 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1808 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1809 Alan Hayward <alan.hayward@arm.com>
1810 David Sherwood <david.sherwood@arm.com>
1812 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1814 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1815 Alan Hayward <alan.hayward@arm.com>
1816 David Sherwood <david.sherwood@arm.com>
1818 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1819 previously undocumented selectors.
1820 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1822 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1824 * doc/rtl.texi (const_vector): Say that elements can be
1825 const_wide_ints too.
1826 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1827 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1828 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1829 * optabs.c (expand_vector_broadcast): Likewise.
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 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1836 scaled index even if the unscaled address was invalid.
1837 Don't increase the complexity of using a scale in that case.
1839 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1840 Alan Hayward <alan.hayward@arm.com>
1841 David Sherwood <david.sherwood@arm.com>
1843 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1844 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1845 UNITS_PER_WORD bytes.
1846 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1847 based on REGMODE_NATURAL_SIZE of the inner mode.
1848 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1849 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1850 * expmed.c (lowpart_bit_field_p): Divide the value up into
1851 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1852 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1853 whether something is likely to occupy more than one register.
1855 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1858 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1859 profile_count::adjust_for_ipa_scaling.
1860 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1861 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1863 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1865 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1867 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1868 (split_group): Add total_orig and total_new arguments, estimate the
1869 number of statements related to the store group without store merging
1870 and with store merging.
1871 (imm_store_chain_info::output_merged_store): Adjust split_group
1872 callers, punt if estimated number of statements with store merging
1873 is not smaller than estimated number of statements without it.
1875 (handled_load): Remove has_single_use checks.
1876 (pass_store_merging::process_store): Likewise.
1878 2017-11-09 Richard Biener <rguenther@suse.de>
1880 PR tree-optimization/82902
1881 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1883 2017-11-09 Martin Liska <mliska@suse.cz>
1886 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1889 2017-11-09 Martin Liska <mliska@suse.cz>
1891 PR tree-optimization/82669
1892 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1894 2017-11-09 Martin Liska <mliska@suse.cz>
1896 PR gcov-profile/48463
1897 * coverage.c (coverage_begin_function): Output also end locus
1898 of a function and information whether the function is
1900 * gcov-dump.c (tag_function): Parse and print the information.
1901 * gcov.c (INCLUDE_MAP): Add include.
1902 (INCLUDE_SET): Likewise.
1903 (struct line_info): Move earlier in the source file because
1904 of vector<line_info> in function_info structure.
1905 (line_info::line_info): Likewise.
1906 (line_info::has_block): Likewise.
1907 (struct source_info): Add new member index.
1908 (source_info::get_functions_at_location): New function.
1909 (function_info::group_line_p): New function.
1910 (output_intermediate_line): New function.
1911 (output_intermediate_file): Use the mentioned function.
1912 (struct function_start): New.
1913 (struct function_start_pair_hash): Likewise.
1914 (process_file): Add code that identifies group functions.
1915 Assign lines either to global or function scope.
1916 (generate_results): Skip artificial functions.
1917 (find_source): Assign index for each source file.
1918 (read_graph_file): Read new flag artificial and end_line.
1919 (add_line_counts): Assign it either to global of function scope.
1920 (accumulate_line_counts): Isolate core of the function to
1921 accumulate_line_info and call it for both function and global
1923 (accumulate_line_info): New function.
1924 (output_line_beginning): Fix GNU coding style.
1925 (print_source_line): New function.
1926 (output_line_details): Likewise.
1927 (output_function_details): Likewise.
1928 (output_lines): Iterate both source (global) scope and function
1930 (struct function_line_start_cmp): New class.
1931 * doc/gcov.texi: Reflect changes in documentation.
1933 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1936 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1937 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1938 and similarly for not instead of neg.
1940 2017-11-08 Andi Kleen <ak@linux.intel.com>
1942 * config/i386/i386.opt: Add -mforce-indirect-call.
1943 * config/i386/predicates.md: Check for flag_force_indirect_call.
1944 * doc/invoke.texi: Document -mforce-indirect-call
1946 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1948 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1950 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1951 config/riscv/riscv.c (predict.h): New include.
1952 (riscv_slow_unaligned_access_p): No longer static.
1953 (riscv_block_move_straight): Add require.
1954 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1956 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1959 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1960 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1961 nonimmediate_operand predicate for operand 1 instead of
1964 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1966 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1968 * config/aarch64/constraints.md (Uml): New constraint.
1969 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1972 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1974 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1975 of two vec_duplicates into a vec_concat.
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 vec_concat.
1981 * config/aarch64/constraints.md (Utq): New constraint.
1982 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1985 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1987 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1988 Simplify vec_merge of vec_duplicate and const_vector.
1989 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1991 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1992 mode iterator. Update predicate on operand 1 to
1993 handle non-const_vec constants. Delete constraints.
1994 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1996 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1998 PR tree-optimization/78821
1999 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2001 (store_operand_info::store_operand_info): Initialize it to false.
2002 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2003 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2004 store in the group, and if chain_info is non-NULL, to ignore altogether
2006 (compatible_load_p): Fail if bit_not_p does not match.
2007 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2008 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2009 (pass_store_merging::process_store): Adjust
2010 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2011 call terminate_all_aliasing_chains newly when adding a store into
2012 a chain with non-NULL chain_info.
2014 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2016 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2018 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2020 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2022 (aarch64_layout_frame): Initialise emit_frame_chain.
2023 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2024 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2026 2017-11-08 Martin Liska <mliska@suse.cz>
2028 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2029 of gimple_call_internal_p.
2031 2017-11-07 Tom de Vries <tom@codesourcery.com>
2033 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2035 2017-11-07 Tom de Vries <tom@codesourcery.com>
2037 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2040 2017-11-08 Martin Liska <mliska@suse.cz>
2043 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2045 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2047 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2049 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2052 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2053 (*cmp<mode>_ccz_1): New insn with $k alternative.
2056 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2057 RTX_COMM_COMPARE as commutative as well.
2058 (ix86_binary_operator_ok): Formatting fix.
2059 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2060 *<code><mode>3<mask_name><round_saeonly_name>,
2061 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2062 *<s>mul<mode>3_highpart<mask_name>,
2063 *vec_widen_umult_even_v16si<mask_name>,
2064 *vec_widen_umult_even_v8si<mask_name>,
2065 *vec_widen_umult_even_v4si<mask_name>,
2066 *vec_widen_smult_even_v16si<mask_name>,
2067 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2068 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2069 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2070 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2071 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2072 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2073 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2074 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2075 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2076 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2077 ix86_binary_operator_ok. Formatting fixes.
2078 (*<plusminus_insn><mode>3<mask_name><round_name>,
2079 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2082 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2084 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2085 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2086 eq and ne if TARGET_ISEL.
2087 (cmp): New code_iterator.
2088 (UNS, UNSU_, UNSIK): New code_attrs.
2089 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2090 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2092 ("eq<mode>3"): ... this.
2093 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2095 ("ne<mode>3"): ... this.
2097 2017-11-07 Julia Koval <julia.koval@intel.com>
2100 * common/config/i386/i386-common.c
2101 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2102 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2103 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2104 * config/i386/i386.opt: Ditto.
2105 * config/i386/i386.c (ix86_target_string): Ditto.
2106 (ix86_option_override_internal): Ditto.
2107 (ix86_init_mpx_builtins): Move MPX to args2.
2108 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2109 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2110 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2111 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2112 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2113 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2114 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2116 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2119 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2120 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2121 (zero-extendsidi peephole2): Remove peephole.
2123 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2126 * stor-layout.c: Include attribs.h.
2127 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2128 explicit lookup of "aligned" attribute.
2130 2017-11-07 Andrew Waterman <andrew@sifive.com>
2132 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2133 (riscv_expand_block_move): Likewise.
2134 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2136 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2137 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2138 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2139 (riscv_adjust_block_mem): Likewise.
2140 (riscv_block_move_loop): Likewise.
2141 (riscv_expand_block_move): Likewise.
2142 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2144 2017-11-07 Michael Clark <michaeljclark@mac.com>
2146 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2147 (MUSL_DYNAMIC_LINKER): Likewise.
2149 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2151 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2152 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2155 2017-11-07 Richard Biener <rguenther@suse.de>
2157 * match.pd: Fix build.
2159 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2160 Jackson Woodruff <jackson.woodruff@arm.com>
2162 PR tree-optimization/71026
2163 * match.pd: Canonicalize negate in division.
2165 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2168 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2170 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2172 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2173 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2175 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2177 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2178 non-scalar integral types.
2179 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2180 (-(A-B), -(~A)): New transformations.
2182 2017-11-07 Tom de Vries <tom@codesourcery.com>
2184 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2185 semicolon after "do {} while (0)".
2186 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2187 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2188 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2189 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2190 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2192 2017-11-07 Tom de Vries <tom@codesourcery.com>
2194 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2195 after "do {} while (0)".
2196 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2197 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2198 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2199 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2200 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2202 2017-11-07 Tom de Vries <tom@codesourcery.com>
2205 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2207 (arm_rtx_costs_internal): Add missing semicolon after
2208 HANDLE_NARROW_SHIFT_ARITH call.
2210 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2213 disable isel if it was not set explicitly.
2215 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2217 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
2218 (add_type_attribute) likewise.
2220 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2222 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2223 of struct ix86_frame.
2224 (ix86_initial_elimination_offset): Likewise.
2225 (ix86_expand_split_stack_prologue): Likewise.
2227 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2229 * tree-vrp.h (enum value_range_type): Update stale comment.
2231 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2232 Alan Hayward <alan.hayward@arm.com>
2233 David Sherwood <david.sherwood@arm.com>
2235 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2236 (aarch64_expand_vec_perm_const): Take the number of units too.
2237 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2238 (aarch64_expand_vec_perm_const): Likewise.
2239 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2240 (vec_perm<mode>): Update accordingly.
2242 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2243 Alan Hayward <alan.hayward@arm.com>
2244 David Sherwood <david.sherwood@arm.com>
2246 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2247 Take the number of units too.
2248 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2249 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2250 but check for a vector mode before rather than after the call.
2251 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2252 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2253 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2254 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2255 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2256 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2257 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2258 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2259 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2260 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2261 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2262 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2263 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2264 (aarch64_sqdmull2_n<mode>): Update accordingly.
2266 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2267 Alan Hayward <alan.hayward@arm.com>
2268 David Sherwood <david.sherwood@arm.com>
2270 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2271 the number of units too.
2272 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2273 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2274 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2275 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2276 (vec_store_lanesxi<mode>): Update accordingly.
2278 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2279 Alan Hayward <alan.hayward@arm.com>
2280 David Sherwood <david.sherwood@arm.com>
2282 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2283 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2284 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2285 of units rather than the mode.
2286 * config/aarch64/iterators.md (nunits): New mode attribute.
2287 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2288 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2289 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2290 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2291 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2292 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2293 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2294 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2295 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2296 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2297 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2298 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2299 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2300 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2301 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2302 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2303 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2304 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2305 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2306 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2307 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2308 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2309 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2310 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2311 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2312 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2313 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2314 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2315 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2316 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2317 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2318 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2319 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2320 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2321 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2322 (aarch64_simd_vec_setv2di): Likewise.
2324 2017-11-06 Carl Love <cel@us.ibm.com>
2326 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2328 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2329 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2331 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2333 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2334 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2335 Add power 8 macro expansions.
2336 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2337 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2338 power 8 instructions. (VSX_XXBR): Add iterator.
2340 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2342 * config/arm/arm.md (predicable_short_it): Change default to "no",
2343 improve documentation, remove uses that are identical to the default.
2344 (enabled_for_depr_it): Rename to enabled_for_short_it.
2345 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2346 Remove default uses.
2347 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2348 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2349 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2350 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2352 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2355 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2356 float128 helper macros, which are no longer used after deleting
2357 the old 'q' built-in functions, and moving the round to odd
2358 built-in functions to being special built-in functions.
2359 (BU_FLOAT128_2): Likewise.
2360 (BU_FLOAT128_1_HW): Likewise.
2361 (BU_FLOAT128_2_HW): Likewise.
2362 (BU_FLOAT128_3_HW): Likewise.
2363 (FABSQ): Delete old 'q' built-in functions.
2364 (COPYSIGNQ): Likewise.
2365 (SQRTF128_ODD): Move round to odd built-in functions to be
2366 special built-in functions, so that we can handle
2367 -mabi=ieeelongdouble.
2368 (TRUNCF128_ODD): Likewise.
2369 (ADDF128_ODD): Likewise.
2370 (SUBF128_ODD): Likewise.
2371 (MULF128_ODD): Likewise.
2372 (DIVF128_ODD): Likewise.
2373 (FMAF128_ODD): Likewise.
2374 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2375 built-in names to 'f128'.
2376 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2377 old 'q' built-in functions, as the machine independent code for
2378 'f128' built-in functions handles this.
2379 (rs6000_expand_builtin): Add expansion for float128 round to odd
2380 functions, keying off on -mabi=ieeelongdouble of whether to use
2381 the KFmode or TFmode variant.
2382 (rs6000_init_builtins): Initialize the _Float128 round to odd
2384 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2385 _Float128 'q' built-in functions are now mapped into the new
2386 'f128' built-in functions.
2388 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2390 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2391 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2393 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2396 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2397 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2398 * toplev.c: Include "ipa-fnsummary.h".
2399 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2401 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2403 PR tree-optimization/82838
2404 * gimple-ssa-store-merging.c
2405 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2406 on a separate gimple_seq which is then appended to seq.
2408 2017-11-06 Jeff Law <law@redhat.com>
2411 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2412 (get_probe_interval): New functions.
2413 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2414 (ix86_adjust_stack_and_probe): Likewise.
2415 (output_adjust_stack_and_probe): Likewise.
2416 (ix86_emit_probe_stack_range): Likewise.
2417 (ix86_expand_prologue): Likewise.
2419 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2421 PR tree-optimization/82816
2422 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2423 if the modes of the two types are the same.
2424 (convert_plusminus_to_widen): Likewise.
2426 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2428 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2430 (usadv16qi): New define_expand.
2431 (usadv8hi): New define_expand.
2433 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2436 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2438 (inline_transform): Likewise.
2439 * predict.c (counts_to_freqs): Remove useless conditional.
2440 * profile-count.h (profile_count::apply_scale): Move sanity check.
2441 * tree-inline.c (copy_bb): Always scale.
2442 (copy_cfg_body): Likewise.
2444 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2447 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2449 (*cmp_ite0): Add enabled_for_depr_it attribute.
2450 (*cmp_ite1): Likewise.
2452 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2454 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2457 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2459 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2460 return the value by pointer.
2461 (extract_range_from_multiplicative_op_1): Update accordingly.
2462 Return as soon as an operation fails.
2464 2017-11-05 Tom de Vries <tom@codesourcery.com>
2467 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2468 (DEF_SANITIZER_BUILTIN): ... here.
2469 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2470 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2472 2017-11-05 Tom de Vries <tom@codesourcery.com>
2475 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2477 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2478 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2479 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2481 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2482 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2483 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2485 2017-11-05 Tom de Vries <tom@codesourcery.com>
2488 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2491 2017-11-04 Michael Clark <michaeljclark@mac.com>
2493 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2494 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2496 (*addsi3_extended): Likewise.
2497 (*addsi3_extended2): Likewise.
2498 (<optab>si3): Likewise.
2499 (<optab>di3): Likewise.
2500 (<optab><mode>3): Likewise.
2501 (<*optabe>si3_internal): Likewise.
2502 (zero_extendqi<SUPERQI:mode>2): Likewise.
2503 (*add<mode>hi3): Likewise.
2504 (*xor<mode>hi3): Likewise.
2505 (<optab>di3): Likewise.
2506 (*<optab>si3_extend): Likewise.
2507 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2508 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2509 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2511 2017-11-04 Andrew Waterman <andrew@sifive.com>
2513 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2514 TARGET_STRICT_ALIGN based upon -mtune argument.
2516 2017-11-04 Andrew Waterman <andrew@sifive.com>
2518 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2520 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2522 config/i386/i386.c (choose_basereg): Use optional scratch
2523 register and add assertion.
2524 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2525 needed, and don't allocate stack.
2526 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2527 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2528 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2530 2017-11-03 Jeff Law <law@redhat.com>
2532 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2533 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2534 to probe at the start of a noreturn function.
2536 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2538 PR tree-optimization/78821
2539 * gimple-ssa-store-merging.c: Update the file comment.
2540 (MAX_STORE_ALIAS_CHECKS): Define.
2541 (struct store_operand_info): New type.
2542 (store_operand_info::store_operand_info): New constructor.
2543 (struct store_immediate_info): Add rhs_code and ops data members.
2544 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2545 and op1r arguments to the ctor, initialize corresponding data members.
2546 (struct merged_store_group): Add load_align_base and load_align
2548 (merged_store_group::merged_store_group): Initialize them.
2549 (merged_store_group::do_merge): Update them.
2550 (merged_store_group::apply_stores): Pick the constant for
2551 encode_tree_to_bitpos from one of the two operands, or skip
2552 encode_tree_to_bitpos if neither operand is a constant.
2553 (class pass_store_merging): Add process_store method decl. Remove
2554 bool argument from terminate_all_aliasing_chains method decl.
2555 (pass_store_merging::terminate_all_aliasing_chains): Remove
2556 var_offset_p argument and corresponding handling.
2557 (stmts_may_clobber_ref_p): New function.
2558 (compatible_load_p): New function.
2559 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2560 if there is overlap and rhs_code is not INTEGER_CST. For
2561 non-overlapping stores terminate group if rhs is not mergeable.
2562 (get_alias_type_for_stmts): Change first argument from
2563 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2564 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2565 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2567 (get_location_for_stmts): Change first argument from
2568 auto_vec<gimple *> & to vec<gimple *> &.
2569 (struct split_store): Remove orig_stmts data member, add orig_stores.
2570 (split_store::split_store): Create orig_stores rather than orig_stmts.
2571 (find_constituent_stmts): Renamed to ...
2572 (find_constituent_stores): ... this. Change second argument from
2573 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2574 to info structures rather than the statements.
2575 (split_group): Rename ALLOW_UNALIGNED argument to
2576 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2577 it. Adjust find_constituent_stores caller.
2578 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2579 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2580 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2581 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2582 (mem_valid_for_store_merging): New function.
2583 (handled_load): New function.
2584 (pass_store_merging::process_store): New method.
2585 (pass_store_merging::execute): Use process_store method. Adjust
2586 terminate_all_aliasing_chains caller.
2588 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2590 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2591 Return true for more constants, symbols and label references.
2592 (aarch64_valid_floating_const): Remove unused function.
2594 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2597 * config/aarch64/aarch64.c (aarch64_layout_frame):
2598 Undo forcing of LR at bottom of frame.
2600 2017-11-03 Jeff Law <law@redhat.com>
2603 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2604 for int_registers_saved.
2606 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2607 extracted from tree-ssa-dom.c.
2608 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2609 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2610 (record_equivalences_from_incoming_edge): Add additional argument
2611 to single_pred_edge_ignoring_loop_edges call.
2612 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2613 (uncprop_dom_walker::before_dom_children): Add additional argument
2614 to single_pred_edge_ignoring_loop_edges call.
2615 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2616 single_pred_edge_ignoring_loop_edges rather than open coding.
2617 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2619 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2621 * match.pd (-(-A)): Rewrite.
2623 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2625 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2626 (rs6000_emit_int_cmove): New declaration.
2627 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2628 (rs6000_emit_sISEL): Delete.
2629 (rs6000_emit_int_cmove): Make non-static.
2630 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2631 instead of rs6000_emit_sISEL.
2633 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2635 * asan.c (create_cond_insert_point): Maintain profile.
2636 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2638 * basic-block.h (struct basic_block_def): Remove frequency.
2639 (EDGE_FREQUENCY): Use to_frequency
2640 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2642 (find_traces): Update to use to_frequency.
2643 (find_traces_1_round): Likewise; use only IPA counts.
2644 (bb_to_key): Likewise.
2645 (connect_traces): Use IPA counts only.
2646 (copy_bb_p): Update to use to_frequency.
2647 (fix_up_crossing_landing_pad): Likewise.
2648 (sanitize_hot_paths): Likewise.
2649 * bt-load.c (basic_block_freq): Likewise.
2650 * cfg.c (init_flow): Set count_max to uninitialized.
2651 (check_bb_profile): Remove frequencies; check counts.
2652 (dump_bb_info): Do not dump frequencies.
2653 (update_bb_profile_for_threading): Update counts only.
2654 (scale_bbs_frequencies_int): Likewise.
2655 (MAX_SAFE_MULTIPLIER): Remove.
2656 (scale_bbs_frequencies_gcov_type): Update counts only.
2657 (scale_bbs_frequencies_profile_count): Update counts only.
2658 (scale_bbs_frequencies): Update counts only.
2659 * cfg.h (struct control_flow_graph): Add count-max.
2660 (update_bb_profile_for_threading): Update prototype.
2661 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2662 (find_many_sub_basic_blocks): Likewise.
2663 * cfgcleanup.c (try_forward_edges): Likewise.
2664 (try_crossjump_to_edge): Likewise.
2665 * cfgexpand.c (expand_gimple_cond): Likewise.
2666 (expand_gimple_tailcall): Likewise.
2667 (construct_init_block): Likewise.
2668 (construct_exit_block): Likewise.
2669 * cfghooks.c (verify_flow_info): Check consistency of counts.
2670 (dump_bb_for_graph): Do not dump frequencies.
2671 (split_block_1): Do not update frequencies.
2672 (split_edge): Do not update frequencies.
2673 (make_forwarder_block): Do not update frequencies.
2674 (duplicate_block): Do not update frequencies.
2675 (account_profile_record): Do not update frequencies.
2676 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2677 for global heuristics.
2678 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2679 (expected_loop_iterations_unbounded): Use counts only.
2680 * cfgloopmanip.c (scale_loop_profile): Simplify.
2681 (create_empty_loop_on_edge): Simplify
2683 (duplicate_loop_to_header_edge): Simplify
2684 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2685 (update_br_prob_note): Take care of removing note when profile
2687 (relink_block_chain): Do not dump frequency.
2688 (rtl_account_profile_record): Use to_frequency.
2689 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2690 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2691 (cgraph_update_edges_for_call_stmt_node): Likewise.
2692 (cgraph_edge::verify_count_and_frequency): Update.
2693 (cgraph_node::verify_node): Temporarily disable frequency verification.
2694 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2695 to_cgraph_frequency.
2696 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2697 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2699 (cgraph_node::expand_thunk): Update profile.
2700 * except.c (dw2_build_landing_pads): Do not update frequency.
2701 * final.c (compute_alignments): Use to_frequency.
2702 (dump_basic_block_info): Do not dump frequency.
2703 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2704 (dump_gimple_bb_header): Do not dump frequency.
2705 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2707 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2708 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2709 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2710 (init_before_recovery): Do not update frequency.
2711 (sched_create_recovery_edges): Do not update frequency.
2712 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2713 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2714 (ipa_cp_c_finalize): Set max_count to uninitialized.
2715 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2716 (param_change_prob): Use counts.
2717 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2719 * ipa-split.c (consider_split): Use to_frequency.
2720 (split_function): Use to_frequency.
2721 * ira-build.c (loop_compare_func): Likewise.
2722 (mark_loops_for_removal): Likewise.
2723 (mark_all_loops_for_removal): Likewise.
2724 * loop-doloop.c (doloop_modify): Do not update frequency.
2725 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2727 * lto-streamer-in.c (input_function): Update count_max.
2728 * omp-expand.c (expand_omp_taskreg): Update count_max.
2729 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2730 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2731 (maybe_hot_count_p): Use ipa counts only.
2732 (maybe_hot_bb_p): Simplify.
2733 (maybe_hot_edge_p): Simplify.
2734 (probably_never_executed): Do not take frequency argument.
2735 (probably_never_executed_bb_p): Do not pass frequency.
2736 (probably_never_executed_edge_p): Likewise.
2737 (combine_predictions_for_bb): Check that profile is nonzero.
2738 (propagate_freq): Do not set frequency.
2739 (drop_profile): Simplify.
2740 (counts_to_freqs): Simplify.
2741 (expensive_function_p): Use to_frequency.
2742 (propagate_unlikely_bbs_forward): Simplify.
2743 (determine_unlikely_bbs): Simplify.
2744 (estimate_bb_frequencies): Add hack to silence graphite issues.
2745 (compute_function_frequency): Use ipa counts.
2746 (pass_profile::execute): Update.
2747 (rebuild_frequencies): Use counts only.
2748 (force_edge_cold): Use counts only.
2749 * profile-count.c (profile_count::dump): Dump new count types.
2750 (profile_count::differs_from_p): Check compatiblity.
2751 (profile_count::to_frequency): New function.
2752 (profile_count::to_cgraph_frequency): New function.
2753 * profile-count.h (struct function): Declare.
2754 (enum profile_quality): Add profile_guessed_local and
2755 profile_guessed_global0.
2756 (class profile_proability): Decrease number of bits to 29;
2757 update from_reg_br_prob_note and to_reg_br_prob_note.
2758 (class profile_count: Update comment; decrease number of bits
2759 to 61. Check compatibility.
2760 (profile_count::compatible_p): New private member function.
2761 (profile_count::ipa_p): New member function.
2762 (profile_count::operator<): Handle global zero correctly.
2763 (profile_count::operator>): Handle global zero correctly.
2764 (profile_count::operator<=): Handle global zero correctly.
2765 (profile_count::operator>=): Handle global zero correctly.
2766 (profile_count::nonzero_p): New member function.
2767 (profile_count::force_nonzero): New member function.
2768 (profile_count::max): New member function.
2769 (profile_count::apply_scale): Handle IPA scalling.
2770 (profile_count::guessed_local): New member function.
2771 (profile_count::global0): New member function.
2772 (profile_count::ipa): New member function.
2773 (profile_count::to_frequency): Declare.
2774 (profile_count::to_cgraph_frequency): Declare.
2775 * profile.c (OVERLAP_BASE): Delete.
2776 (compute_frequency_overlap): Delete.
2777 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2778 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2779 * sched-ebb.c (rank): Use counts only.
2780 * shrink-wrap.c (handle_simple_exit): Use counts only.
2781 (try_shrink_wrapping): Use counts only.
2782 (place_prologue_for_one_component): Use counts only.
2783 * tracer.c (find_best_predecessor): Use to_frequency.
2784 (find_trace): Use to_frequency.
2785 (tail_duplicate): Use to_frequency.
2786 * trans-mem.c (expand_transaction): Do not update frequency.
2787 * tree-call-cdce.c: Do not update frequency.
2788 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2789 (gimple_merge_blocks): Likewise.
2790 (gimple_split_edge): Likewise.
2791 (gimple_duplicate_sese_region): Likewise.
2792 (gimple_duplicate_sese_tail): Likewise.
2793 (move_sese_region_to_fn): Likewise.
2794 (gimple_account_profile_record): Likewise.
2795 (insert_cond_bb): Likewise.
2796 * tree-complex.c (expand_complex_div_wide): Likewise.
2797 * tree-eh.c (lower_resx): Update profile.
2798 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2800 (initialize_cfun): Do not initialize frequencies
2801 (freqs_to_counts): Delete.
2802 (copy_cfg_body): Ignore count parameter.
2803 (copy_body): Update.
2804 (expand_call_inline): Update count_max.
2805 (optimize_inline_calls): Update count_max.
2806 (tree_function_versioning): Update count_max.
2807 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2808 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2810 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2811 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2812 (try_peel_loop): Likewise.
2813 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2815 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2816 (tree_transform_and_unroll_loop): Do not use frequencies
2817 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2818 Use reliable prediction only.
2819 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2820 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2821 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2822 probability scaling.
2823 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2824 not update frequency
2825 (any_remaining_duplicated_blocks): Likewise.
2826 (update_profile): Likewise.
2827 (estimated_freqs_path): Delete.
2828 (freqs_to_counts_path): Delete.
2829 (clear_counts_path): Delete.
2830 (ssa_fix_duplicate_block_edges): Likewise.
2831 (duplicate_thread_path): Likewise.
2832 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2833 * tree-tailcall.c (decrease_profile): Do not update frequency.
2834 (eliminate_tail_call): Likewise.
2835 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2836 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2837 (optimize_mask_stores): Likewise.
2838 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2839 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2840 (ubsan_expand_ptr_ifn): Update profile.
2841 * value-prof.c (gimple_ic): Simplify.
2842 * value-prof.h (gimple_ic): Update prototype.
2843 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2844 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2846 (want_inline_self_recursive_call_p): Likewise.
2847 (resolve_noninline_speculation): Only cummulate defined counts.
2848 (inline_small_functions): Use nonzero_p.
2849 (ipa_inline): Do not access freed node.
2851 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2853 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2854 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2856 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2858 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2859 non-legitimate address.
2861 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2863 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2864 (*lt0_<mode>di, *lt0_<mode>si): New.
2866 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2868 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2869 TARGET_PAIRED_FLOAT.
2871 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2872 Jim Wilson <jim.wilson@linaro.org>
2874 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2875 * config/aarch64/aarch64-tune.md: Regenerated.
2876 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2877 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2879 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2881 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2882 (arc_expand_prologue): Restore blink for millicode.
2883 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2885 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2888 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2889 gen_vec_duplicate after forcing the scalar into a register.
2891 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2893 * combine (try_combine): Print the insns input to try_combine to the
2896 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2899 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2900 Remove second argument from aarch64_process_target_attr call.
2901 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2903 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2905 (aarch64_handle_attr_arch): Remove second argument.
2906 (aarch64_handle_attr_cpu): Ditto.
2907 (aarch64_handle_attr_tune): Ditto.
2908 (aarch64_handle_attr_isa_flags): Ditto.
2909 (aarch64_process_one_target_attr): Ditto.
2910 (aarch64_process_target_attr): Ditto.
2911 (aarch64_option_valid_attribute_p): Remove second argument.
2912 on aarch64_process_target_attr call.
2914 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2916 * diagnostic.c: Include "selftest-diagnostic.h".
2917 (selftest::assert_location_text): New function.
2918 (selftest::test_diagnostic_get_location_text): New function.
2919 (selftest::diagnostic_c_tests): Call it.
2921 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2923 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2924 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2925 (class selftest::test_diagnostic_context): Move to...
2926 * selftest-diagnostic.c: New file.
2927 * selftest-diagnostic.h: New file.
2929 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2931 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2932 offset range for FT32B.
2933 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2934 * config/ft32/ft32.md: Add TARGET_NOPM.
2935 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2936 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2938 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2940 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2942 2017-11-02 Jeff Law <law@redhat.com>
2944 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2945 virtual keyword on FINAL OVERRIDE members.
2947 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2948 virtuals together. Add virtual destructor.
2949 (substitute_and_fold_engine): Similarly.
2951 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2953 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2955 2017-11-02 Richard Biener <rguenther@suse.de>
2957 PR tree-optimization/82795
2958 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2960 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2962 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2964 (gcc_GAS_CHECK_FEATURE): Remove.
2965 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2966 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2967 * configure: Regenerate.
2969 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2971 * config/arc/arc.c (hwloop_optimize): Account for empty
2974 2017-11-02 Richard Biener <rguenther@suse.de>
2977 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2978 Truncate ARRAY_REF index and element size.
2980 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2982 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2984 2017-11-01 Jeff Law <law@redhat.com>
2986 * tree-ssa-ccp.c (ccp_folder): New class derived from
2987 substitute_and_fold_engine.
2988 (ccp_folder::get_value): New member function.
2989 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2990 (ccp_fold_stmt): Remove prototype.
2991 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2992 * tree-ssa-copy.c (copy_folder): New class derived from
2993 substitute_and_fold_engine.
2994 (copy_folder::get_value): Renamed from get_value.
2995 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2996 * tree-vrp.c (vrp_folder): New class derived from
2997 substitute_and_fold_engine.
2998 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2999 (vrp_folder::get_value): New member function.
3000 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3001 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3002 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3003 provide a class interface to folder/substitute routines.
3004 (ssa_prop_fold_stmt_fn): Remove typedef.
3005 (ssa_prop_get_value_fn): Likewise.
3006 (subsitute_and_fold): Remove prototype.
3007 (replace_uses_in): Likewise.
3008 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3009 Renamed from replace_uses_in. Call the virtual member function
3010 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3011 (substitute_and_fold_dom_walker): Remove initialization of
3012 data member entries for calbacks. Add substitute_and_fold_engine
3013 member and initialize it.
3014 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3015 member functions for get_value, replace_phi_args_in c
3016 replace_uses_in, and fold_stmt calls.
3017 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3018 substitute_and_fold. Remove assert. Update ctor call.
3020 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3021 (ssa_prop_visit_phi_fn): Likewise.
3022 (class ssa_propagation_engine): New class to provide an interface
3024 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3026 (ssa_prop_visit_phi): Likewise.
3027 (ssa_propagation_engine::simulate_stmt): Moved into class.
3028 Call visit_phi/visit_stmt from the class rather than via
3029 file scoped static variables.
3030 (ssa_propagation_engine::simulate_block): Moved into class.
3031 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3032 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3033 set file scoped statics for the visit_stmt/visit_phi callbacks.
3034 * tree-complex.c (complex_propagate): New class derived from
3035 ssa_propagation_engine.
3036 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3037 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3038 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3040 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3041 ssa_propagation_engine.
3042 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3043 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3044 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3045 * tree-ssa-copy.c (copy_prop): New class derived from
3046 ssa_propagation_engine.
3047 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3048 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3049 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3050 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3051 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3052 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3053 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3055 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3057 PR rtl-optimization/82778
3058 PR rtl-optimization/82597
3059 * compare-elim.c (struct comparison): Add in_a_setter field.
3060 (find_comparison_dom_walker::before_dom_children): Remove killed
3061 bitmap and df_simulate_find_defs call, instead walk the defs.
3062 Compute last_setter and initialize in_a_setter. Merge definitions
3063 with first initialization for a few variables.
3064 (try_validate_parallel): Use insn_invalid_p instead of
3065 recog_memoized. Return insn rather than just the pattern.
3066 (try_merge_compare): Fix up comment. Don't uselessly test if
3067 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3069 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3072 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3073 Alan Hayward <alan.hayward@arm.com>
3074 David Sherwood <david.sherwood@arm.com>
3076 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3077 aarch64_hard_regno_nregs to get the number of registers
3080 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3081 Alan Hayward <alan.hayward@arm.com>
3082 David Sherwood <david.sherwood@arm.com>
3084 * config/aarch64/constraints.md (Upl): Rename to...
3086 * config/aarch64/aarch64.md
3087 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3090 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3091 Alan Hayward <alan.hayward@arm.com>
3092 David Sherwood <david.sherwood@arm.com>
3094 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3095 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3098 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3099 Alan Hayward <alan.hayward@arm.com>
3100 David Sherwood <david.sherwood@arm.com>
3102 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3103 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3104 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3106 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3107 of a single element.
3108 * config/aarch64/iterators.md: Add a comment above the permute
3109 unspecs to say that they are generated directly by
3110 aarch64_expand_vec_perm_const.
3111 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3113 2017-11-01 Nathan Sidwell <nathan@acm.org>
3115 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3117 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3119 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3120 and medany code models, and describe what they do.
3122 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3124 Revert accidental duplicate:
3126 * combine.c (can_change_dest_mode): Reject changes in
3127 REGMODE_NATURAL_SIZE.
3129 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3131 PR rtl-optimization/64682
3132 PR rtl-optimization/69567
3133 PR rtl-optimization/69737
3134 PR rtl-optimization/82683
3135 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3136 register mentioned in the note, drop the note, unless it came from I3,
3137 in which case it should go to I3 again.
3139 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3141 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3142 and return false if not.
3143 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3145 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3147 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3148 range is known to be empty.
3150 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3151 Alan Hayward <alan.hayward@arm.com>
3152 David Sherwood <david.sherwood@arm.com>
3154 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3155 and CONST_VECTOR_NUNITS instead of computing the number of units from
3156 the byte sizes of the vector and element.
3157 (simplify_binary_operation_1): Likewise.
3158 (simplify_const_binary_operation): Likewise.
3159 (simplify_ternary_operation): Likewise.
3161 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3162 Alan Hayward <alan.hayward@arm.com>
3163 David Sherwood <david.sherwood@arm.com>
3165 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3166 (int_mem_offset): ...this new function.
3167 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3168 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3169 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3172 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3173 Alan Hayward <alan.hayward@arm.com>
3174 David Sherwood <david.sherwood@arm.com>
3176 * lower-subreg.c (interesting_mode_p): New function.
3177 (compute_costs, find_decomposable_subregs, decompose_register)
3178 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3179 (resolve_clobber, dump_choices): Use it.
3181 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3182 Alan Hayward <alan.hayward@arm.com>
3183 David Sherwood <david.sherwood@arm.com>
3185 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
3192 truncations. Check the precision rather than the size.
3194 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3195 Alan Hayward <alan.hayward@arm.com>
3196 David Sherwood <david.sherwood@arm.com>
3198 * machmode.h (is_narrower_int_mode): New function
3199 * optabs.c (expand_float, expand_fix): Use it.
3200 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
3207 * ira-color.c (update_costs_from_allocno): Use it.
3209 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3210 Alan Hayward <alan.hayward@arm.com>
3211 David Sherwood <david.sherwood@arm.com>
3213 * optabs-query.h (convert_optab_p): New function, split out from...
3214 (convert_optab_handler): ...here.
3215 (widening_optab_handler): Delete.
3216 (find_widening_optab_handler): Remove permit_non_widening parameter.
3217 (find_widening_optab_handler_and_mode): Likewise. Provide an
3218 override that operates on mode class wrappers.
3219 * optabs-query.c (widening_optab_handler): Delete.
3220 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3221 parameter. Assert that the two modes are the same class and that
3222 the "from" mode is narrower than the "to" mode. Use
3223 convert_optab_handler instead of widening_optab_handler.
3224 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3225 instead of widening_optab_handler.
3226 * expr.c (expand_expr_real_2): Update calls to
3227 find_widening_optab_handler.
3228 * optabs.c (expand_widen_pattern_expr): Likewise.
3229 (expand_binop_directly): Take the insn_code as a parameter.
3230 (expand_binop): Only call find_widening_optab_handler for
3231 conversion optabs; use optab_handler otherwise. Update calls
3232 to find_widening_optab_handler and expand_binop_directly.
3233 Use convert_optab_handler instead of widening_optab_handler.
3234 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3235 find_widening_optab_handler and use scalar_mode rather than
3237 (convert_plusminus_to_widen): Likewise.
3239 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3240 Alan Hayward <alan.hayward@arm.com>
3241 David Sherwood <david.sherwood@arm.com>
3243 * machmode.h (fixed_size_mode): New class.
3244 * rtl.h (get_pool_mode): Return fixed_size_mode.
3245 * gengtype.c (main): Add fixed_size_mode.
3246 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3247 (get_raw_arg_mode): Likewise.
3248 * doc/tm.texi: Regenerate.
3249 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3250 * targhooks.c (default_get_reg_raw_mode): Likewise.
3251 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3252 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3253 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3254 (msp430_get_raw_result_mode): Likewise.
3255 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3256 * dbxout.c (dbxout_parms): Require fixed-size modes.
3257 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3258 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3259 * omp-low.c (lower_oacc_reductions): Likewise.
3260 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3261 (simplify_subreg): Update accordingly.
3262 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3263 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3264 that aren't fixed-size.
3265 (get_pool_mode): Return a fixed_size_mode.
3266 (output_constant_pool_2): Take a fixed_size_mode.
3268 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3269 Alan Hayward <alan.hayward@arm.com>
3270 David Sherwood <david.sherwood@arm.com>
3272 * doc/rtl.texi (vec_series): Document.
3273 (const): Say that the operand can be a vec_series.
3274 * rtl.def (VEC_SERIES): New rtx code.
3275 * rtl.h (const_vec_series_p_1): Declare.
3276 (const_vec_series_p): New function.
3277 * emit-rtl.h (gen_const_vec_series): Declare.
3278 (gen_vec_series): Likewise.
3279 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3280 (gen_vec_series): Likewise.
3281 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3282 * simplify-rtx.c (simplify_unary_operation): Handle negations
3284 (simplify_binary_operation_series): New function.
3285 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3286 (test_vector_ops_series): New function.
3287 (test_vector_ops): Call it.
3288 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3289 gen_const_vec_series.
3290 (altivec_lvsr): Likewise.
3291 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3293 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3294 Alan Hayward <alan.hayward@arm.com>
3295 David Sherwood <david.sherwood@arm.com>
3297 * doc/rtl.texi (const): Update description of address constants.
3298 Say that vector constants are allowed too.
3299 * common.md (E, F): Use CONSTANT_P instead of checking for
3301 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3302 checking for CONST_VECTOR.
3303 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3305 * expr.c (expand_expr_real_2): Check for vector modes instead
3306 of checking for CONST_VECTOR.
3307 * rtl.h (const_vec_p): New function.
3308 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3309 (unwrap_const_vec_duplicate): Handle them here too.
3311 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3312 David Malcolm <dmalcolm@redhat.com>
3313 Alan Hayward <alan.hayward@arm.com>
3314 David Sherwood <david.sherwood@arm.com>
3316 * rtl.h (vec_duplicate_p): New function.
3317 * selftest-rtl.c (assert_rtx_eq_at): New function.
3318 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3319 (assert_rtx_eq_at): Declare.
3320 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3321 * selftest-run-tests.c (selftest::run_tests): Call it.
3322 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3323 (simplify_unary_operation_1): Recursively handle vector duplicates.
3324 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3326 (simplify_subreg): Handle subregs of vector duplicates.
3327 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3328 (selftest::simplify_rtx_c_tests): New functions.
3330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3331 Alan Hayward <alan.hayward@arm.com>
3332 David Sherwood <david.sherwood@arm.com>
3334 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3335 (gen_vec_duplicate): Likewise.
3336 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3338 (gen_const_vector): ...here.
3339 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3340 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3341 whose elements are all equal.
3342 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3343 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3344 (simplify_relational_operation): Likewise.
3345 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3347 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3348 (aarch64_expand_vector_init): Likewise.
3349 * config/arm/arm.c (neon_vdup_constant): Likewise.
3350 (neon_expand_vector_init): Likewise.
3351 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3352 (arm_block_set_unaligned_vect): Likewise.
3353 (arm_block_set_aligned_vect): Likewise.
3354 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3355 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3356 (expand_vec_perm_even_odd_pack): Likewise.
3357 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3358 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3359 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3360 gen_const_vec_duplicate.
3361 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3362 * config/mips/mips.c (mips_gen_const_int_vector): Use
3363 gen_const_vec_duplicate.
3364 (mips_expand_vector_init): Use CONST0_RTX.
3365 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3366 (define_split): Use gen_const_vec_duplicate.
3367 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3368 (define_split): Use gen_const_vec_duplicate.
3369 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3370 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3371 * config/spu/spu.c (spu_const): Likewise.
3373 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3374 Alan Hayward <alan.hayward@arm.com>
3375 David Sherwood <david.sherwood@arm.com>
3377 * combine.c (can_change_dest_mode): Reject changes in
3378 REGMODE_NATURAL_SIZE.
3380 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3382 * configure.ac (--enable-libssp): New.
3383 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3384 trying to determine target-specific default. Adjust indentation.
3385 * configure: Regenerated.
3386 * doc/install.texi (Configuration): Expand --disable-libssp
3389 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3391 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3394 2017-10-31 Martin Jambor <mjambor@suse.cz>
3397 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3399 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3401 * auto-profile.c (autofdo_source_profile::read): Use
3402 UNKNOWN_LOCATION rather than 0.
3403 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3404 (warning_at): ...this overload.
3405 (warning_at_rich_loc_n): Rename to...
3406 (warning_n): ...this overload.
3407 (error_at_rich_loc): Rename to...
3408 (error_at): ...this overload.
3409 (pedwarn_at_rich_loc): Rename to...
3410 (pedwarn): ...this overload.
3411 (permerror_at_rich_loc): Rename to...
3412 (permerror): ...this overload.
3413 (inform_at_rich_loc): Rename to...
3414 (inform): ...this overload.
3415 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3416 (diagnostic_n_impl_richloc): Rename to...
3417 (diagnostic_n_impl): ...this rich_location *-based decl.
3418 (inform_at_rich_loc): Rename to...
3419 (inform): ...this, and add an assertion.
3420 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3421 (warning_at_rich_loc): Rename to...
3422 (warning_at): ...this, and add an assertion.
3423 (warning_at_rich_loc_n): Rename to...
3424 (warning_n): ...this, and add an assertion.
3425 (warning_n): Update location_t-based implementation for removal of
3426 location_t-based diagnostic_n_impl.
3427 (pedwarn_at_rich_loc): Rename to...
3428 (pedwarn): ...this, and add an assertion.
3429 (permerror_at_rich_loc): Rename to...
3430 (permerror): ...this, and add an assertion.
3431 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3432 (error_at_rich_loc): Rename to...
3433 (error_at): ...this, and add an assertion.
3434 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3435 (driver::do_spec_on_infiles): Likewise.
3436 * substring-locations.c (format_warning_va): Update for renaming
3437 of inform_at_rich_loc.
3439 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3441 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3442 _Float<N>X built-in functions so that the variant without the
3443 "__builtin_" prefix is only enabled for the GNU C and Objective C
3444 languages when they are in non-strict ANSI/ISO mode.
3445 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3446 * target.def (floatn_builtin_p): Add a target hook to control
3447 whether _Float<N> and _Float<N>X built-in functions without the
3448 "__builtin_" prefix are enabled, and return true for C and
3449 Objective C in the default hook. Include langhooks.h in
3451 * targhooks.h (default_floatn_builtin_p): Likewise.
3452 * targhooks.c (default_floatn_builtin_p): Likewise.
3453 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3454 floatn_builtin_p target hook.
3455 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3457 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3458 Eric Botcazou <ebotcazou@adacore.com>
3460 PR rtl-optimization/81803
3461 * lra-constraints.c (curr_insn_transform): Also reload the whole
3462 register for a strict subreg no wider than a word if this is for
3463 a WORD_REGISTER_OPERATIONS target.
3465 2017-10-31 Jason Merrill <jason@redhat.com>
3467 * gdbinit.in: Skip over inlines from timevar.h.
3469 2017-10-31 Martin Liska <mliska@suse.cz>
3471 * doc/gcov.texi: Document new option.
3472 * gcov.c (print_usage): Likewise print it.
3473 (process_args): Support the argument.
3474 (format_count): New function.
3475 (format_gcov): Use the function.
3477 2017-10-31 Martin Liska <mliska@suse.cz>
3479 * gcov.c (struct name_map): do not use typedef.
3480 Define operator== and operator<.
3481 (name_search): Remove.
3482 (name_sort): Remove.
3483 (main): Do not allocate names.
3484 (process_file): Add vertical space.
3485 (generate_results): Use std::find.
3486 (release_structures): Do not release memory.
3487 (find_source): Use std::find.
3489 2017-10-31 Martin Liska <mliska@suse.cz>
3491 * gcov.c (struct line_info): Remove it's typedef.
3492 (line_info::line_info): Add proper ctor.
3493 (line_info::has_block): Do not use a typedef.
3494 (struct source_info): Do not use typedef.
3495 (circuit): Likewise.
3496 (get_cycles_count): Likewise.
3497 (output_intermediate_file): Iterate via vector iterator.
3498 (add_line_counts): Use std::vector methods.
3499 (accumulate_line_counts): Likewise.
3500 (output_lines): Likewise.
3502 2017-10-31 Martin Liska <mliska@suse.cz>
3504 * gcov.c (struct source_info): Remove typedef.
3505 (source_info::source_info): Add proper ctor.
3506 (accumulate_line_counts): Use struct, not it's typedef.
3507 (output_gcov_file): Likewise.
3508 (output_lines): Likewise.
3509 (main): Do not allocate an array.
3510 (output_intermediate_file): Use size of vector container.
3511 (process_file): Resize the vector.
3512 (generate_results): Do not preallocate, use newly added vector
3514 (release_structures): Do not release sources.
3515 (find_source): Use vector methods.
3516 (add_line_counts): Do not use typedef.
3518 2017-10-31 Martin Liska <mliska@suse.cz>
3520 * doc/gcov.texi: Document that.
3521 * gcov.c (add_line_counts): Mark lines with a non-executed
3523 (output_line_beginning): Handle such lines.
3524 (output_lines): Pass new argument.
3525 (output_intermediate_file): Print it in intermediate format.
3527 2017-10-31 Martin Liska <mliska@suse.cz>
3529 * color-macros.h: New file.
3530 * diagnostic-color.c: Factor out color related to macros to
3532 * doc/gcov.texi: Document -k option.
3533 * gcov.c (INCLUDE_STRING): Include string.h.
3534 (print_usage): Add -k option.
3535 (process_args): Parse it.
3536 (pad_count_string): New function.
3537 (output_line_beginning): Likewise.
3538 (DEFAULT_LINE_START): New macro.
3539 (output_lines): Support color output.
3541 2017-10-31 Martin Liska <mliska@suse.cz>
3543 PR gcov-profile/82633
3544 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3545 their interaction with GCOV infrastructure.
3546 * configure.ac: Add -fkeep-{inline,static}-functions to
3548 * configure: Regenerate.
3550 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3553 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3555 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3558 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3559 into a register if it does not fit into an immediate offset field.
3561 2017-10-31 Olivier Hainque <hainque@adacore.com>
3563 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3565 2017-10-31 Julia Koval <julia.koval@intel.com>
3567 * config.gcc: Add gfniintrin.h.
3568 * config/i386/gfniintrin.h: New.
3569 * config/i386/i386-builtin-types.def
3570 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3571 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3572 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3573 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3574 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3575 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3576 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3577 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3578 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3579 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3580 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3581 * config/i386/immintrin.h: Include gfniintrin.h.
3582 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3584 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3586 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3588 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3590 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3591 (arm_ashldi3_1bit): Remove pattern.
3592 (ashrdi3): Remove shift by 1 expansion.
3593 (arm_ashrdi3_1bit): Remove pattern.
3594 (lshrdi3): Remove shift by 1 expansion.
3595 (arm_lshrdi3_1bit): Remove pattern.
3596 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3597 cost of ashldi3 by 1.
3598 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3599 (<shift>di3_neon): Likewise.
3601 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3603 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3604 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3605 and (*aarch64_simd_mov<VQ:mode>).
3606 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3607 pattern alternative.
3608 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3609 attributes to match pattern alternative.
3611 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3613 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3614 * config/rs6000/emmintrin.h: New file.
3615 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3617 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3619 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3620 * (movdi_vfp_cortexa8): Remove pattern.
3622 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3624 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3626 (Specific, alpha*-dec-osf5.1): Remove.
3627 (Specific, mips-sgi-irix5): Remove.
3628 (Specific, mips-sgi-irix6): Remove.
3630 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3633 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3634 arguments to clear_bit_region_be.
3636 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3638 * gimplify.c: Include memmodel.h.
3640 2017-10-30 Martin Jambor <mjambor@suse.cz>
3642 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3643 condition whether loop should be executed at all.
3645 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3647 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3648 gimple folding of vec_madd() intrinsics.
3649 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3650 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3651 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3653 2017-10-30 Richard Biener <rguenther@suse.de>
3655 PR tree-optimization/82762
3657 2017-10-23 Richard Biener <rguenther@suse.de>
3659 PR tree-optimization/82129
3661 2017-08-01 Richard Biener <rguenther@suse.de>
3663 PR tree-optimization/81181
3664 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3665 (compute_antic): ... end of iteration here.
3667 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3669 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3670 -std=iso9899:2017 and -std=gnu17.
3671 * doc/standards.texi (C Language): Document C17 support.
3672 * doc/cpp.texi (Overview): Mention -std=c17.
3673 (Standard Predefined Macros): Document C11 and C17 values of
3674 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3675 * doc/extend.texi (Inline): Do not list individual options for
3676 standards newer than C99.
3677 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3679 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3682 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3684 * asan.c (asan_finish_file): Align asan globals array by shadow
3687 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3690 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3691 (struct store_immediate_info): Add bitregion_start and bitregion_end
3693 (store_immediate_info::store_immediate_info): Add brs and bre
3694 arguments and initialize bitregion_{start,end} from those.
3695 (struct merged_store_group): Add bitregion_start, bitregion_end,
3696 align_base and mask fields. Drop unnecessary struct keyword from
3697 struct store_immediate_info. Add do_merge method.
3698 (clear_bit_region_be): Use memset instead of loop storing zeros.
3699 (merged_store_group::do_merge): New method.
3700 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3701 stores as long as the surrounding bitregions have no gaps.
3702 (merged_store_group::merge_overlapping): Use do_merge.
3703 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3704 is byte aligned, rather than requiring that start and width are
3705 byte aligned. Drop unnecessary struct keyword from
3706 struct store_immediate_info. Allocate and populate also mask array.
3707 Make start of the arrays relative to bitregion_start rather than
3708 start and size them according to bitregion_{end,start} difference.
3709 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3710 struct store_immediate_info.
3711 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3712 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3713 struct keyword from struct store_immediate_info.
3714 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3715 between stores as long as the surrounding bitregions have no gaps.
3717 (struct split_store): Add orig non-static data member.
3718 (split_store::split_store): Initialize orig to false.
3719 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3720 if there is exactly a single original stmt. Change stmts argument
3721 to pointer from reference, if NULL, don't push anything to it. Add
3722 first argument, use it to optimize skipping over orig stmts that
3723 are known to be before bitpos already. Simplify.
3724 (split_group): Return unsigned int count how many stores are or
3725 would be needed rather than a bool. Add allow_unaligned argument.
3726 Change split_stores argument from reference to pointer, if NULL,
3727 only do a dry run computing how many stores would be produced.
3728 Rewritten algorithm to use both alignment and misalign if
3729 !allow_unaligned and handle bitfield stores with gaps.
3730 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3731 from bitregion_start instead of start. Compute allow_unaligned
3732 here, if true, do 2 split_group dry runs to compute which one
3733 produces fewer stores and prefer aligned if equal. Punt if
3734 new count is bigger or equal than original before emitting any
3735 statements, rather than during that. Remove no longer needed
3736 new_ssa_names tracking. Replace num_stmts with
3737 split_stores.length (). Use 32-bit stack allocated entries
3738 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3739 if possible. Handle bitfields with gaps.
3740 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3741 Compute bitregion_{start,end} for the stores and construct
3742 store_immediate_info with that. Formatting fixes.
3744 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3747 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3748 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3750 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3752 * gimplify.c: Include tm_p.h.
3754 * common.opt (gcoff): Re-add as ignored option.
3755 (gcoff1, gcoff2, gcoff3): Likewise.
3757 * Makefile.in (OBJS): Delete sdbout.o.
3758 (GTFILES): Delete $(srcdir)/sdbout.c.
3759 * debug.h: Delete sdb_debug_hooks.
3760 * final.c: Delete sdbout.h include.
3761 (final_scan_insn): Delete SDB_DEBUG check.
3762 (rest_of_clean_state): Likewise.
3763 * output.h: Delete sdb_begin_function_line.
3766 * toplev.c: Delete sdbout.h include.
3767 (process_options): Delete SDB_DEBUG check.
3768 * tree-core.h (tree_type_common): Delete pointer field of
3770 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3771 TYPE_SYMTAB_POINTER.
3772 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3773 (TYPE_SYMTAB_IS_POINTER): Delete.
3774 (TYPE_SYMTAB_IS_DIE): Renumber.
3775 * xcoffout.c: Refer to former sdbout.c file.
3776 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3778 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3779 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3780 (Debugging Options): Delete -gcoff.
3781 (-gstabs): Delete SDB reference.
3783 (-gcoff@var{level}): Delete.
3784 * doc/passes.texi (Debugging information output): Delete SDB and
3785 sdbout.c references.
3786 * doc/tm.texi: Regenerate.
3787 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3788 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3790 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3791 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3792 (SDB_DEBUGGING_INFO): Delete.
3793 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3794 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3795 * target.def (output_source_filename): Delete COFF reference.
3797 * common.opt (gcoff): Delete.
3798 (gxcoff+): Update Negative chain.
3799 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3801 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3802 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3803 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3805 (expand_function_start): Change sdb reference to past tense.
3806 (expand_function_end): Change sdb reference to past tense.
3807 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3808 * opts.c (debug_type_names): Delete coff entry.
3809 (common_handle_option): Delete OPT_gcoff case.
3810 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3812 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3813 * config/cris/cris.h: Delete SDB reference in comment.
3814 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3815 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3816 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3817 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3819 (ix86_expand_prologue): Likewise.
3820 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3821 * config/ia64/ia64.h: Likewise.
3822 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3823 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3825 * config/mmix/mmix.h: Likewise.
3826 * config/nds32/nds32.c: Likewise.
3827 * config/stormy/storym16.h: Likewise.
3828 * config/visium/visium.h: Likewise.
3829 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3831 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3833 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3834 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3835 FRAME_POINTER_REGNUM point at high end of local var area.
3837 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3839 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3840 Move comment around. Do not reset best_edge for a copiable
3841 destination if the copy would cause a partition change.
3842 (better_edge_p): Remove redundant check.
3844 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3846 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3848 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3850 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3851 for math functions that have _Float<N> and _Float<N>X variants.
3852 (mathfn_built_in_2): Add support for math functions that have
3853 _Float<N> and _Float<N>X variants.
3854 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3855 (expand_builtin_mathfn_ternary): Add support for fma with
3856 _Float<N> and _Float<N>X variants.
3857 (expand_builtin): Likewise.
3858 (fold_builtin_3): Likewise.
3859 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3860 create math function _Float<N> and _Float<N>X variants as external
3862 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3863 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3864 the __builtin_ prefix and if not strict ansi, without the prefix.
3865 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3866 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3867 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3868 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3869 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3870 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3871 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3872 function signatures for fma _Float<N> and _Float<N>X variants.
3873 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3874 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3875 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3876 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3877 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3878 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3879 * gencfn-macros.c (print_case_cfn): Add support for math functions
3880 that have _Float<N> and _Float<N>X variants.
3881 (print_define_operator_list): Likewise.
3882 (fltfn_suffixes): Likewise.
3884 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3885 for math functions that have _Float<N> and _Float<N>X variants.
3886 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3887 and _Float<N>X variants.
3888 (COPYSIGN): Likewise.
3891 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3892 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3894 (integer_valued_read_call_p): Likewise.
3895 * fold-const-call.c (fold_const_call_ss): Likewise.
3896 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3897 _Float<N> and _Float<N>X variants.
3898 (fold_const_call_ssss): Add support for fma _Float<N> and
3899 _Float<N>X variants.
3900 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3901 support for copysign and fma _Float<N> and _Float<N>X variants.
3902 (backprop::process_builtin_call_use): Likewise.
3903 * tree-call-cdce.c (can_test_argument_range); Add support for
3904 sqrt _Float<N> and _Float<N>X variants.
3905 (edom_only_function): Likewise.
3906 (get_no_error_domain): Likewise.
3907 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3908 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3909 copysign _Float<N> and _Float<N>X variants.
3910 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3911 handled by machine independent code.
3912 (FMAF128): Likewise.
3913 * doc/cpp.texi (Common Predefined Macros): Document defining
3914 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3915 fma _Float<N> and _Float<N>X variants.
3917 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3920 * config/i386/i386-modes.def (CCFPU): Remove definition.
3921 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3922 (ix86_cc_modes_compatible): Ditto.
3923 (ix86_expand_carry_flag_compare): Ditto.
3924 (ix86_expand_int_movcc): Ditto.
3925 (ix86_expand_int_addcc): Ditto.
3926 (ix86_reverse_condition): Ditto.
3927 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3928 Return true/false for unordered/ordered fp comparisons.
3929 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3930 (ix86_prepare_fp_compare_args): Update for rename.
3931 (ix86_expand_fp_compare): Update for rename. Generate unordered
3932 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3933 (ix86_expand_sse_compare_and_jump): Ditto.
3934 * config/i386/predicates.md (fcmov_comparison_operator):
3935 Remove CCFPU mode handling.
3936 (ix86_comparison_operator): Ditto.
3937 (ix86_carry_flag_operator): Ditto.
3938 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3939 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3940 (*cmpu<mode>_cc_i387): Ditto.
3941 (FPCMP): Remove mode iterator.
3942 (unord): Remove mode attribute.
3943 (unord_subst): New define_subst transformation
3944 (unord): New define_subst attribute.
3946 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3947 (*cmpi<unord>xf_i387): Ditto.
3948 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3949 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3950 using unord_subst transformation.
3951 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3952 (round_saeonly): Also handle CCFP mode.
3953 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3954 Remove UNSPEC_SAHF unspec handling.
3956 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3958 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3960 2017-10-27 Jeff Law <law@redhat.com>
3962 * gimple-ssa-sprintf.c: Include domwalk.h.
3963 (class sprintf_dom_walker): New class, derived from dom_walker.
3964 (sprintf_dom_walker::before_dom_children): New function.
3965 (struct call_info): Moved into sprintf_dom_walker class
3966 (compute_formath_length, handle_gimple_call): Likewise.
3967 (sprintf_length::execute): Call the dominator walker rather
3968 than walking the statements.
3970 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3971 gimple statement locations.
3972 (check_array_bounds): Corresponding changes. Get the statement's
3973 location directly from wi->stmt.
3975 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3978 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3980 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3982 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3983 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3985 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3988 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3989 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3990 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3991 maybe_get_pool_constant.
3992 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3995 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3997 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3998 2.26 caveat. Update gas and gld versions.
3999 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4002 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4004 * cgraph.h (set_malloc_flag): Declare.
4005 * cgraph.c (set_malloc_flag_1): New function.
4006 (set_malloc_flag): Likewise.
4007 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4008 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4010 (read_ipa_call_summary): Add support for reading is_return_callee.
4011 (write_ipa_call_summary): Stream is_return_callee.
4012 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4013 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4014 ipa-prop.h, ipa-fnsummary.h.
4015 (pure_const_names): Change to static.
4016 (malloc_state_e): Define.
4017 (malloc_state_names): Define.
4018 (funct_state_d): Add field malloc_state.
4019 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4020 (check_retval_uses): New function.
4021 (malloc_candidate_p): Likewise.
4022 (analyze_function): Add support for malloc attribute.
4023 (pure_const_write_summary): Stream malloc_state.
4024 (pure_const_read_summary): Add support for reading malloc_state.
4025 (dump_malloc_lattice): New function.
4026 (propagate_malloc): New function.
4027 (warn_function_malloc): New function.
4028 (ipa_pure_const::execute): Call propagate_malloc and
4029 ipa_free_fn_summary.
4030 (pass_local_pure_const::execute): Add support for malloc attribute.
4031 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4032 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4034 2017-10-27 Martin Liska <mliska@suse.cz>
4036 PR gcov-profile/82457
4037 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4038 for fork-like functions to be properly instrumented.
4040 2017-10-27 Richard Biener <rguenther@suse.de>
4043 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4044 info when we redirected EH.
4046 2017-10-26 Michael Collison <michael.collison@arm.com>
4048 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4050 (<optab>_trunchf<GPI:mode>2: New pattern.
4051 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4052 * config/aarch64/iterators.md (wv): New mode attribute.
4053 (vf, VF): New mode attributes.
4054 (vgp, VGP): New mode attributes.
4055 (s): Update attribute with SImode and DImode prefixes.
4057 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4059 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4060 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4061 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4062 (nios2_option_overide): Initialize it. Don't allow R0-relative
4063 addressing with PIC.
4064 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4065 (nios2_symbolic_constant_p): Likewise.
4066 (nios2_legitimate_address_p): Likewise.
4067 (nios2_r0rel_section_name_p): New.
4068 (nios2_symbol_ref_in_r0rel_data_p): New.
4069 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4070 (r0rel_constant_p): New.
4071 (nios2_print_operand_address): Handle r0rel_constant_p.
4072 (nios2_cdx_narrow_form_p): Likewise.
4073 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4074 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4075 (Nios II Options): Document -mr0rel-sec.
4077 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4079 * config/nios2/nios2.c: Include xregex.h.
4080 (nios2_gprel_sec_regex): New.
4081 (nios2_option_overide): Initialize it. Don't allow GP-relative
4082 addressing with PIC.
4083 (nios2_small_section_name_p): Check for regex match.
4084 * config/nios2/nios2.opt (mgprel-sec=): New option.
4085 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4086 (Nios II Options): Document -mgprel-sec.
4088 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4090 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4092 2017-10-26 Tom de Vries <tom@codesourcery.com>
4094 PR tree-optimization/82707
4095 * gimple.c (gimple_copy): Fix unsharing of
4096 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4098 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4100 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4101 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4102 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4103 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4104 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4105 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4106 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4107 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4108 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4109 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4110 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4111 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4112 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4113 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4114 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4115 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4116 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4118 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4120 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4122 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4123 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4124 warning to rs6000.c. Remove the Undocumented flag, since it has
4126 (-mabi=ibmlongdouble): Likewise.
4127 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4128 already set, set the default format for long double.
4129 (rs6000_debug_reg_global): Print whether long double is IBM or
4131 (rs6000_option_override_internal): Rework setting long double
4132 format. Only warn if the user is changing the long double default
4133 and they did not use -Wno-psabi.
4134 * doc/invoke.texi (PowerPC options): Update the documentation for
4135 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4137 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4138 Alan Hayward <alan.hayward@arm.com>
4139 David Sherwood <david.sherwood@arm.com>
4141 * rtl.h (wider_subreg_mode): New function.
4142 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4143 rather than an unsigned int *.
4144 * ira-color.c (regno_max_ref_width): Replace with...
4145 (regno_max_ref_mode): ...this new variable.
4146 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4147 Use wider_subreg_mode.
4148 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4149 rather than an unsigned int *.
4150 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4151 (process_alt_operands): Likewise.
4152 (invariant_p): Likewise.
4153 * lra-spills.c (assign_mem_slot): Likewise.
4154 (add_pseudo_to_slot): Likewise.
4155 * lra.c (collect_non_operand_hard_regs): Likewise.
4156 (add_regs_to_insn_regno_info): Likewise.
4157 * reload1.c (regno_max_ref_width): Replace with...
4158 (regno_max_ref_mode): ...this new variable.
4159 (reload): Update accordingly. Update call to
4160 ira_sort_regnos_for_alter_reg.
4161 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4162 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4163 (scan_paradoxical_subregs): Likewise.
4165 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4167 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4168 (aarch64_frame): Add emit_frame_chain boolean.
4169 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4170 Move eh_return case to aarch64_layout_frame.
4171 (aarch64_layout_frame): Initialize emit_frame_chain.
4172 (aarch64_expand_prologue): Use emit_frame_chain.
4174 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4176 * config/aarch64/aarch64.c (aarch64_layout_frame):
4177 Ensure LR is always stored at the bottom of the callee-saves.
4178 Remove rarely used frame layout which saves callee-saves at top of
4179 frame, so the store of LR can be used as a valid probe in all cases.
4181 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4183 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4184 Improve unaligned TImode/TFmode base/offset split.
4186 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4187 Alan Hayward <alan.hayward@arm.com>
4188 David Sherwood <david.sherwood@arm.com>
4190 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4191 * combine.c (find_single_use_1): Likewise.
4192 (expand_field_assignment): Likewise.
4193 (move_deaths): Likewise.
4194 * lra-constraints.c (simplify_operand_subreg): Likewise.
4195 (curr_insn_transform): Likewise.
4196 * lra.c (collect_non_operand_hard_regs): Likewise.
4197 (add_regs_to_insn_regno_info): Likewise.
4198 * rtlanal.c (reg_referenced_p): Likewise.
4199 (covers_regno_no_parallel_p): Likewise.
4201 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4203 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4204 Don't print any bits outside the precision of the value.
4205 * wide-int.cc (test_printing): Add some new tests.
4207 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4209 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4210 supports -xbrace_comment option.
4211 * configure: Regenerate.
4212 * config.in: Regenerate.
4213 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4214 (ASM_CPU_SPEC): Use it.
4216 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4218 * target.def (static_rtx_alignment): New hook.
4219 * targhooks.h (default_static_rtx_alignment): Declare.
4220 * targhooks.c (default_static_rtx_alignment): New function.
4221 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4222 * doc/tm.texi: Regenerate.
4223 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4224 instead of targetm.constant_alignment. Remove call to
4226 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4227 (cris_preferred_mininum_alignment): New function, split out from...
4228 (cris_constant_alignment): ...here.
4229 (cris_static_rtx_alignment): New function.
4230 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4232 (ix86_constant_alignment): ...here.
4233 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4234 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4235 (mmix_static_rtx_alignment): New function.
4236 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4237 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4239 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4242 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4243 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4245 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4247 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4248 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4249 mark as a sign-extending load.
4250 (local_pic_load_u): Define.
4252 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4255 * fold-const.c (operand_equal_for_comparison_p): Also return true
4256 if ARG0 is a simple variant of ARG1 with narrower precision.
4257 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4259 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4261 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4263 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4264 scatter_static, scatter_per_elt.
4265 * x86-tune-costs.h: Add new cost entries.
4267 2017-10-25 Richard Biener <rguenther@suse.de>
4269 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4270 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4271 class pass_fre): Move to ...
4272 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4273 class pass_fre): ... here and adjust for statistics.
4275 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4278 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4279 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4280 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4281 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4284 2017-10-25 Richard Biener <rguenther@suse.de>
4286 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4287 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4288 eliminate_push_avail, eliminate_insert): Move inside...
4289 (class eliminate_dom_walker): ... this class in preparation
4291 (fini_eliminate): Remove by merging with ...
4292 (eliminate): ... this function. Adjust for class changes.
4293 (pass_pre::execute): Remove fini_eliminate call.
4294 (pass_fre::execute): Likewise.
4296 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4299 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4300 (VPERMI2, VPERMI2I): New mode iterators.
4301 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4302 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4304 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4305 mode iterator. Remove 3 old define_insn patterns.
4306 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4307 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4308 VPERMI2 mode iterator, remove the other two expanders.
4309 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4310 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4311 instructions, remove the other two patterns.
4312 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4313 mode iterator, remove the other two patterns.
4314 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4315 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4316 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4317 and adjust argument order accordingly.
4318 (ix86_expand_vec_perm): Adjust caller.
4319 (expand_vec_perm_1): Likewise.
4320 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4321 (expand_vec_perm_vpermt2_vpshub2): ... this.
4322 (ix86_expand_vec_perm_const_1): Adjust caller.
4323 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4326 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4327 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4328 (vec_shl_<mode>): Remove unused expander.
4329 (avx512bw_<shift_insn><mode>3): New define_insn.
4330 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4331 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4333 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4336 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4339 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4341 PR rtl-optimization/82396
4342 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4343 (autopref_multipass_init): Simplify initialization.
4344 (autopref_rank_data): Simplify sort order.
4345 * gcc/sched-int.h (autopref_multipass_data_): Remove
4346 multi_mem_insn_p, min_offset and max_offset.
4348 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4351 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4352 Check special value of flag_omit_frame_pointer.
4353 (aarch64_can_eliminate): Likewise.
4354 (aarch64_override_options_after_change_1): Simplify handling of
4355 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4357 2017-10-24 Richard Biener <rguenther@suse.de>
4359 PR tree-optimization/82697
4360 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4361 zero for conditional load and unconditional store.
4363 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4365 * doc/install.texi: Document bootstrap-cet.
4367 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4370 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4371 ENDBR instruction at function entrance if function is only
4374 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4377 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4378 patterns to better describe from which operation the CF is computed.
4379 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4380 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4381 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4382 is 0, use _0 suffixed expanders instead of emitting a comparison
4385 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4387 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4388 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4390 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4393 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4394 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4395 * loop-iv.c (iv_get_reaching_def): Likewise.
4396 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4397 variable is promoted and the partition contains undefined values.
4399 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4401 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4403 (nios2_address_cost): Define.
4404 (nios2_legitimize_address): Recognize (exp + constant) directly.
4405 (TARGET_ADDRESS_COST): Define.
4407 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4409 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4410 (nios2_symbolic_memory_operand_p): Declare.
4411 (nios2_split_large_constant): Declare.
4412 (nios2_split_symbolic_memory_operand): Declare.
4413 * config/nios2/nios2.c: Adjust includes.
4414 (nios2_symbolic_constant_allowed): New.
4415 (nios2_symbolic_constant_p): New.
4416 (nios2_plus_symbolic_constant_p): New.
4417 (nios2_valid_addr_expr_p): Recognize addresses involving
4419 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4420 (nios2_symbolic_memory_operand_p): New.
4421 (nios2_large_constant_p): New.
4422 (nios2_split_large_constant): New.
4423 (nios2_split_plus_large_constant): New.
4424 (nios2_split_symbolic_memory_operand): New.
4425 (nios2_legitimize_address): Code refactoring. Handle addresses
4426 involving symbolic constants.
4427 (nios2_emit_move_sequence): Likewise.
4428 (nios2_print_operand): Improve error output.
4429 (nios2_print_operand_address): Handle LO_SUM.
4430 (nios2_cdx_narrow_form_p): Likewise.
4431 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4432 operands involving symbolic constants.
4433 (movhi_internal, movsi_internal): Likewise.
4434 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4435 (extendhisi2, extendqi<mode>2): Likewise.
4437 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4439 * tree-pass.h (PROP_rtl_split_insns): Define.
4440 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4442 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4444 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4446 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4449 * target.def (const_not_ok_for_debug_p): Default to
4450 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4451 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4452 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4453 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4454 which targetm.const_not_ok_for_debug_p returned true.
4455 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4457 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4459 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4460 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4461 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4463 (ix86_const_not_ok_for_debug_p): New function.
4464 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4465 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4467 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4470 * system.h: Conditionally include "unique-ptr.h" if
4471 INCLUDE_UNIQUE_PTR is defined.
4472 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4473 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4475 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4477 * config/rl78/rl78.md: New define_expand "subdi3".
4479 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4482 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4483 DF_REF_INSN if DF_REF_INSN_INFO is false.
4485 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4487 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4488 xmm_move instead of sse_move.
4489 (sse_store_index): New function.
4490 (ix86_register_move_cost): Be more sensible about mismatch stall;
4491 model AVX moves correctly; make difference between sse->integer and
4493 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4494 moves; make difference between SSE and AVX.
4495 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4496 and zmm_move. Increase size of sse load and store tables;
4497 add unaligned load and store tables; add ssemmx_to_integer.
4498 * x86-tune-costs.h: Update all entries according to real
4499 move latencies from Agner Fog's manual and chip documentation.
4501 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4504 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4505 * config/i386/constraints.md (Wf): New constraint.
4506 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4507 (cmp<dwi>_doubleword): Removed.
4508 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4509 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4510 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4511 expand with cmp<dwi>_doubleword. For LTU and GEU use
4512 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4514 * common.opt (gcolumn-info): Enable by default.
4515 * doc/invoke.texi (gcolumn-info): Document new default.
4517 2017-10-23 Richard Biener <rguenther@suse.de>
4519 PR tree-optimization/82672
4520 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4521 Fold the stmt if we propagated into it.
4523 2017-10-23 Richard Biener <rguenther@suse.de>
4525 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4526 (bitmap_remove_expr_from_set): ... this. All callers call this
4527 for non-constant values.
4528 (bitmap_set_subtract): Rename to...
4529 (bitmap_set_subtract_expressions): ... this. Adjust and
4531 (bitmap_set_contains_value): Remove superfluous check.
4532 (bitmap_set_replace_value): Inline into single caller ...
4533 (bitmap_value_replace_in_set): ... here and simplify.
4534 (dependent_clean): Merge into ...
4535 (clean): ... this using an overload. Adjust.
4536 (prune_clobbered_mems): Adjust.
4537 (compute_antic_aux): Likewise.
4538 (compute_partial_antic_aux): Likewise.
4540 2017-10-23 Richard Biener <rguenther@suse.de>
4542 PR tree-optimization/82129
4544 2017-08-01 Richard Biener <rguenther@suse.de>
4546 PR tree-optimization/81181
4547 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4548 (compute_antic): ... end of iteration here.
4550 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4552 * target.def (starting_frame_offset): New hook.
4553 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4554 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4555 * doc/tm.texi.in: Regenerate.
4556 * hooks.h (hook_hwi_void_0): Declare.
4557 * hooks.c (hook_hwi_void_0): New function.
4558 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4559 STARTING_FRAME_OFFSET.
4560 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4561 * reload1.c (reload): Likewise.
4562 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4563 instead of STARTING_FRAME_OFFSET.
4564 * function.c (try_fit_stack_local): Likewise.
4565 (assign_stack_local_1): Likewise
4566 (instantiate_virtual_regs): Likewise.
4567 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4568 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4569 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4570 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4571 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4572 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4573 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4574 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4575 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4576 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4577 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4578 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4579 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4580 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4581 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4582 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4583 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4584 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4585 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4586 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4587 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4588 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4589 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4590 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4591 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4592 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4593 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4594 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4595 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4596 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4597 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4598 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4599 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4600 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4601 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4602 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4603 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4604 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4605 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4606 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4607 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4608 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4609 return a HOST_WIDE_INT.
4610 (avr_builtin_setjmp_frame_value): Use it instead of
4611 STARTING_FRAME_OFFSET.
4612 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4613 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4614 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4616 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4617 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4618 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4619 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4620 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4621 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4622 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4623 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4624 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4625 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4626 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4627 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4629 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4630 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4631 * config/mips/mips.c (mips_compute_frame_info): Refer to
4632 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4633 (mips_starting_frame_offset): New function.
4634 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4635 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4636 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4637 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4638 and return a HOST_WIDE_INT.
4639 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4640 (mmix_initial_elimination_offset): Refer to
4641 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4642 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4643 * config/pa/pa.c (pa_starting_frame_offset): New function.
4644 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4645 (pa_expand_prologue): Likewise.
4646 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4647 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4648 !FRAME_GROWS_DOWNWARD handling to...
4649 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4650 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4651 !FRAME_GROWS_DOWNWARD handling to...
4652 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4653 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4654 !FRAME_GROWS_DOWNWARD handling to...
4655 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4656 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4658 (rs6000_starting_frame_offset): New function.
4659 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4660 !FRAME_GROWS_DOWNWARD handling to...
4661 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4662 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4663 !FRAME_GROWS_DOWNWARD handling to...
4664 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4665 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4666 !FRAME_GROWS_DOWNWARD handling to...
4667 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4668 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4669 (rs6000_starting_frame_offset): New function.
4670 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4671 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4672 * config/vax/vax.c (vax_starting_frame_offset): New function.
4673 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4674 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4675 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4676 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4677 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4678 * system.h (STARTING_FRAME_OFFSET): Poison.
4680 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4682 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4683 SCALAR_TYPE_MODE instead of TYPE_MODE.
4685 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4686 Alan Hayward <alan.hayward@arm.com>
4687 David Sherwood <david.sherwood@arm.com>
4689 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4691 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4692 Alan Hayward <alan.hayward@arm.com>
4693 David Sherwood <david.sherwood@arm.com>
4695 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4697 2017-10-23 Richard Biener <rguenther@suse.de>
4699 PR tree-optimization/82129
4700 * tree-ssa-pre.c (bitmap_set_and): Remove.
4701 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4702 canonicalizing expressions in the set to those with lowest
4703 ID rather than taking that from the first edge.
4705 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4707 * combine.c (rtx_equal_for_field_assignment_p): Use
4708 byte_lowpart_offset.
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 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4715 to a SUBREG_PROMOTED_VAR.
4717 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4718 Alan Hayward <alan.hayward@arm.com>
4719 David Sherwood <david.sherwood@arm.com>
4721 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4722 (expand_debug_source_expr): Likewise.
4723 * combine.c (combine_simplify_rtx): Likewise.
4724 * cse.c (fold_rtx): Likewise.
4725 * optabs.c (expand_float): Likewise.
4726 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4727 (simplify_binary_operation_1): Likewise.
4729 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4730 Alan Hayward <alan.hayward@arm.com>
4731 David Sherwood <david.sherwood@arm.com>
4733 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4734 (record_promoted_value): Likewise.
4735 * expr.c (expand_expr_real_2): Likewise.
4736 * ree.c (update_reg_equal_equiv_notes): Likewise.
4737 (combine_set_extension): Likewise.
4738 * rtlanal.c (low_bitmask_len): Likewise.
4739 * simplify-rtx.c (neg_const_int): Likewise.
4740 (simplify_binary_operation_1): Likewise.
4742 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4743 Alan Hayward <alan.hayward@arm.com>
4744 David Sherwood <david.sherwood@arm.com>
4746 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4747 * regcprop.c (maybe_mode_change): Likewise.
4748 * reload1.c (alter_reg): Likewise.
4750 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4752 * inchash.h (inchash::hash::add_wide_int): New function.
4753 * lto-streamer-out.c (hash_tree): Use it.
4755 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4757 * inchash.h (inchash::hash::add_wide_int): Rename to...
4758 (inchash::hash::add_hwi): ...this.
4759 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4760 (polymorphic_call_target_hasher::hash): Likewise.
4761 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4762 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4763 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4764 * lto-streamer-out.c (hash_tree): Likewise.
4765 * optc-save-gen.awk: Likewise.
4766 * tree.c (add_expr): Likewise.
4768 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4771 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4772 for ordered inequality comparisons even with TARGET_IEEE_FP.
4774 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4777 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4778 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4779 Expand with cmp<dwi>_doubleword.
4781 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4783 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4784 List CET intrinsics.
4785 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4786 specific to -fcf-protection option.
4788 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4790 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4791 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4792 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4793 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4794 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4795 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4796 (extra_objs): Add cet.o for Linux/x86 targets.
4797 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4798 * config/i386/cet.c: New file.
4799 * config/i386/cetintrin.h: Likewise.
4800 * config/i386/t-cet: Likewise.
4801 * config/i386/cpuid.h (bit_SHSTK): New.
4802 (bit_IBT): Likewise.
4803 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4804 pass IBT and SHSTK bits.
4805 * config/i386/i386-builtin-types.def
4806 (VOID_FTYPE_UNSIGNED_PVOID): New.
4807 (VOID_FTYPE_UINT64_PVOID): Likewise.
4808 * config/i386/i386-builtin.def: Add CET intrinsics.
4809 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4810 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4811 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4812 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4814 * config/i386/i386.c (rest_of_insert_endbranch): New.
4815 (pass_data_insert_endbranch): Likewise.
4816 (pass_insert_endbranch): Likewise.
4817 (make_pass_insert_endbranch): Likewise.
4818 (ix86_notrack_prefixed_insn_p): Likewise.
4819 (ix86_target_string): Add -mibt, -mshstk flags.
4820 (ix86_option_override_internal): Add flag_cf_protection
4822 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4823 (ix86_print_operand): Add 'notrack' prefix output.
4824 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4825 (ix86_expand_builtin): Expand CET intrinsics.
4826 (x86_output_mi_thunk): Add 'endbranch' instruction.
4827 * config/i386/i386.h (TARGET_IBT): New.
4828 (TARGET_IBT_P): Likewise.
4829 (TARGET_SHSTK): Likewise.
4830 (TARGET_SHSTK_P): Likewise.
4831 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4832 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4833 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4834 (builtin_setjmp_setup): New pattern.
4835 (builtin_longjmp): Likewise.
4836 (rdssp<mode>): Likewise.
4837 (incssp<mode>): Likewise.
4838 (saveprevssp): Likewise.
4839 (rstorssp): Likewise.
4840 (wrss<mode>): Likewise.
4841 (wruss<mode>): Likewise.
4842 (setssbsy): Likewise.
4843 (clrssbsy): Likewise.
4844 (nop_endbr): Likewise.
4845 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4847 * config/i386/immintrin.h: Include <cetintrin.h>.
4848 * config/i386/linux-common.h
4849 (file_end_indicate_exec_stack_and_cet): New prototype.
4850 (TARGET_ASM_FILE_END): New.
4852 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4854 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4855 latencies instead of having separate table; make difference between
4856 integer and float costs.
4857 * i386.h (processor_costs): Remove scalar_stmt_cost,
4858 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4859 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4861 * x86-tune-costs.h: Remove entries which has been removed in
4862 procesor_costs from all tables; make cond_taken_branch_cost
4863 and cond_not_taken_branch_cost COST_N_INSNS based.
4865 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4867 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4869 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4871 * config/i386/i386.md (isa): Remove fma_avx512f.
4872 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4873 <avx512>_fmadd_<mode>_mask3<round_name>,
4874 <avx512>_fmsub_<mode>_mask<round_name>,
4875 <avx512>_fmsub_<mode>_mask3<round_name>,
4876 <avx512>_fnmadd_<mode>_mask<round_name>,
4877 <avx512>_fnmadd_<mode>_mask3<round_name>,
4878 <avx512>_fnmsub_<mode>_mask<round_name>,
4879 <avx512>_fnmsub_<mode>_mask3<round_name>,
4880 <avx512>_fmaddsub_<mode>_mask<round_name>,
4881 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4882 <avx512>_fmsubadd_<mode>_mask<round_name>,
4883 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4884 (*vec_widen_umult_even_v16si<mask_name>,
4885 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4886 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4888 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4890 * extend.texi: Add 'nocf_check' documentation.
4891 * gimple.texi: Add second parameter to
4892 gimple_build_call_from_tree.
4893 * invoke.texi: Add -fcf-protection documentation.
4894 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4896 2017-10-20 Richard Biener <rguenther@suse.de>
4898 PR tree-optimization/82473
4899 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4900 the largest input type.
4902 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4904 * c-attribs.c (handle_nocf_check_attribute): New function.
4905 (c_common_attribute_table): Add 'nocf_check' handling.
4906 * gimple-parser.c: Add second argument NULL to
4907 gimple_build_call_from_tree.
4908 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4909 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4911 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4912 * common.opt: Add fcf-protection flag.
4913 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4914 * flag-types.h: Add enum cf_protection_level.
4915 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4916 Add 'nocf_check' attribute propagation to gimple call.
4917 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4918 (gimple_build_call_from_tree): Update prototype.
4919 (gimple_call_nocf_check_p): New function.
4920 (gimple_call_set_nocf_check): Likewise.
4921 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4922 * ipa-icf.c: Add nocf_check attribute in statement hash.
4923 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4924 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4925 * toplev.c (process_options): Add flag_cf_protection handling.
4927 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4929 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4931 2017-10-20 Richard Biener <rguenther@suse.de>
4933 PR tree-optimization/82603
4934 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4935 remove false predicated stores.
4937 2017-10-20 Richard Biener <rguenther@suse.de>
4939 * graphite-isl-ast-to-gimple.c
4940 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4941 Remove return value and simplify, dump copied stmt after lhs
4943 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4944 Reduce dump verbosity.
4945 (gsi_insert_earliest): Likewise.
4946 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4947 * graphite.c (print_global_statistics): Adjust dumping.
4948 (print_graphite_scop_statistics): Likewise.
4949 (print_graphite_statistics): Do not dump loops here.
4950 (graphite_transform_loops): But here.
4952 2017-10-20 Nicolas Roche <roche@adacore.com>
4954 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4955 * configure: Regenerate.
4957 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4960 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4961 functions when optimizing replace GIMPLE_RETURN stmts with
4962 calls to __builtin_unreachable ().
4965 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4966 for -fsanitize=thread link of executables.
4967 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4968 link of executables.
4971 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4973 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4974 define_insns for logical vector shifts to use VI248_AVX512BW
4975 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4976 condition, useless isa and prefix attributes. Change the first
4977 2 of these define_insns to ...
4978 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4979 define_insn for avx512vl.
4980 (<shift_insn><mode>3): ... and this, new define_insn without
4981 masking for non-avx512vl.
4984 * config/i386/sse.md (*andnot<mode>3,
4985 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4986 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4987 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4988 not applied use empty suffix even for TARGET_AVX512VL.
4989 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4990 is applied, supply evex,evex or evex,evex,evex instead of just
4993 2017-10-20 Julia Koval <julia.koval@intel.com>
4995 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4996 (OPTION_MASK_ISA_GFNI_UNSET): New.
4997 (ix86_handle_option): Handle OPT_mgfni.
4998 * config/i386/cpuid.h (bit_GFNI): New.
4999 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5000 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5001 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5002 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5003 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5004 * config/i386/i386.opt: Add mgfni.
5006 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5008 * config/msp430/msp430.c (msp430_option_override): Disable
5009 -fdelete-null-pointer-checks.
5010 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5012 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5014 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5015 of x87 and SSE instructions.
5017 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5019 * asan.c (create_cond_insert_point): Do not update edge count.
5020 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5021 (afdo_propagate_circuit): Likewise.
5022 (afdo_calculate_branch_prob): Likewise.
5023 (afdo_annotate_cfg): Likewise.
5024 * basic-block.h (struct edge_def): Remove count.
5025 (edge_def::count): New accessor.
5026 * bb-reorder.c (rotate_loop): Update.
5027 (find_traces_1_round): Update.
5028 (connect_traces): Update.
5029 (sanitize_hot_paths): Update.
5030 * cfg.c (unchecked_make_edge): Update.
5031 (make_single_succ_edge): Update.
5032 (check_bb_profile): Update.
5033 (dump_edge_info): Update.
5034 (update_bb_profile_for_threading): Update.
5035 (scale_bbs_frequencies_int): Update.
5036 (scale_bbs_frequencies_gcov_type): Update.
5037 (scale_bbs_frequencies_profile_count): Update.
5038 (scale_bbs_frequencies): Update.
5039 * cfganal.c (connect_infinite_loops_to_exit): Update.
5040 * cfgbuild.c (compute_outgoing_frequencies): Update.
5041 (find_many_sub_basic_blocks): Update.
5042 * cfgcleanup.c (try_forward_edges): Update.
5043 (try_crossjump_to_edge): Update
5044 * cfgexpand.c (expand_gimple_cond): Update
5045 (expand_gimple_tailcall): Update
5046 (construct_exit_block): Update
5047 * cfghooks.c (verify_flow_info): Update
5048 (redirect_edge_succ_nodup): Update
5049 (split_edge): Update
5050 (make_forwarder_block): Update
5051 (duplicate_block): Update
5052 (account_profile_record): Update
5053 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5054 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5055 * cfgloopmanip.c (scale_loop_profile): Update.
5057 (lv_adjust_loop_entry_edge): Update.
5058 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5059 (force_nonfallthru_and_redirect): Update.
5060 (purge_dead_edges): Update.
5061 (rtl_flow_call_edges_add): Update.
5062 * cgraphunit.c (init_lowered_empty_function): Update.
5063 (cgraph_node::expand_thunk): Update.
5064 * gimple-pretty-print.c (dump_probability): Update.
5065 (dump_edge_probability): Update.
5066 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5067 * haifa-sched.c (sched_create_recovery_edges): Update.
5068 * hsa-gen.c (convert_switch_statements): Update.
5069 * ifcvt.c (dead_or_predicable): Update.
5070 * ipa-inline-transform.c (inline_transform): Update.
5071 * ipa-split.c (split_function): Update.
5072 * ipa-utils.c (ipa_merge_profiles): Update.
5073 * loop-doloop.c (add_test): Update.
5074 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5075 * lto-streamer-in.c (input_cfg): Update.
5076 (input_function): Update.
5077 * lto-streamer-out.c (output_cfg): Update.
5078 * modulo-sched.c (sms_schedule): Update.
5079 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5080 * predict.c (maybe_hot_edge_p): Update.
5081 (unlikely_executed_edge_p): Update.
5082 (probably_never_executed_edge_p): Update.
5083 (dump_prediction): Update.
5084 (drop_profile): Update.
5085 (propagate_unlikely_bbs_forward): Update.
5086 (determine_unlikely_bbs): Update.
5087 (force_edge_cold): Update.
5088 * profile.c (compute_branch_probabilities): Update.
5089 * reg-stack.c (better_edge): Update.
5090 * shrink-wrap.c (handle_simple_exit): Update.
5091 * tracer.c (better_p): Update.
5092 * trans-mem.c (expand_transaction): Update.
5093 (split_bb_make_tm_edge): Update.
5094 * tree-call-cdce.c: Update.
5095 * tree-cfg.c (gimple_find_sub_bbs): Update.
5096 (gimple_split_edge): Update.
5097 (gimple_duplicate_sese_region): Update.
5098 (gimple_duplicate_sese_tail): Update.
5099 (gimple_flow_call_edges_add): Update.
5100 (insert_cond_bb): Update.
5101 (execute_fixup_cfg): Update.
5102 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5103 * tree-complex.c (expand_complex_div_wide): Update.
5104 * tree-eh.c (lower_resx): Update.
5105 (unsplit_eh): Update.
5106 (cleanup_empty_eh_move_lp): Update.
5107 * tree-inline.c (copy_edges_for_bb): Update.
5108 (freqs_to_counts): Update.
5109 (copy_cfg_body): Update.
5110 * tree-ssa-dce.c (remove_dead_stmt): Update.
5111 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5112 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5113 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5114 (unloop_loops): Update.
5115 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5116 * tree-ssa-loop-split.c (connect_loops): Update.
5117 (split_loop): Update.
5118 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5119 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5120 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5121 * tree-ssa-reassoc.c (branch_fixup): Update.
5122 * tree-ssa-tail-merge.c (replace_block_by): Update.
5123 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5124 (compute_path_counts): Update.
5125 (update_profile): Update.
5126 (recompute_probabilities): Update.
5127 (update_joiner_offpath_counts): Update.
5128 (estimated_freqs_path): Update.
5129 (freqs_to_counts_path): Update.
5130 (clear_counts_path): Update.
5131 (ssa_fix_duplicate_block_edges): Update.
5132 (duplicate_thread_path): Update.
5133 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5134 (case_bit_test_cmp): Update.
5135 (collect_switch_conv_info): Update.
5136 (gen_inbound_check): Update.
5137 (do_jump_if_equal): Update.
5138 (emit_cmp_and_jump_insns): Update.
5139 * tree-tailcall.c (decrease_profile): Update.
5140 (eliminate_tail_call): Update.
5141 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5142 (vect_do_peeling): Update.
5143 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5144 * ubsan.c (ubsan_expand_null_ifn): Update.
5145 (ubsan_expand_ptr_ifn): Update.
5146 * value-prof.c (gimple_divmod_fixed_value): Update.
5147 (gimple_mod_pow2): Update.
5148 (gimple_mod_subtract): Update.
5149 (gimple_ic): Update.
5150 (gimple_stringop_fixed_value): Update.
5152 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5155 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5157 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5159 PR rtl-optimization/82395
5160 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5161 based on non_spilled_static_chain_regno_p.
5163 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5165 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5166 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5167 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5169 2017-10-19 Martin Sebor <msebor@redhat.com>
5171 PR tree-optimization/82596
5172 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5174 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5176 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5177 (get_mem_refs_of_builtin_call): Likewise.
5178 * builtins.c (expand_builtin_apply): Adjust call to
5179 allocate_dynamic_stack_space.
5180 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5181 the third argument to allocate_dynamic_stack_space, otherwise -1.
5182 (expand_builtin): Deal with all alloca variants.
5183 (is_inexpensive_builtin): Likewise.
5184 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5185 * calls.c (special_function_p): Deal with all alloca variants.
5186 (initialize_argument_information): Adjust call to
5187 allocate_dynamic_stack_space.
5188 (expand_call): Likewise.
5189 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5190 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5191 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5192 use it for the stack usage computation.
5193 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5194 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5195 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5196 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5197 (in_loop_p): Remove first argument and useless check.
5198 (pass_walloca::execute): Remove useless test and adjust call to above.
5199 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5200 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5201 (gimplify_call_expr): Deal with all alloca variants.
5202 * hsa-gen.c (gen_hsa_alloca): Likewise.
5203 (gen_hsa_insns_for_call): Likewise.
5204 * ipa-pure-const.c (special_builtin_state): Likewise.
5205 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5206 * tree-object-size.c (alloc_object_size): Likewise.
5207 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5208 (call_may_clobber_ref_p_1): Likewise.
5209 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5210 (ccp_fold_stmt): Likewise.
5211 (optimize_stack_restore): Likewise.
5212 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5213 (mark_all_reaching_defs_necessary_1): Likewise.
5214 (propagate_necessity): Likewise.
5215 (eliminate_unnecessary_stmts): Likewise.
5216 * tree.c (build_common_builtin_nodes): Build
5217 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5218 (build_alloca_call_expr): New function.
5219 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5220 (CASE_BUILT_IN_ALLOCA): Likewise.
5221 (build_alloca_call_expr): Declare.
5222 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5224 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5227 * dwarf2out.c (new_die_raw): New static inline function.
5228 (new_die): Use it to create the DIE.
5229 (add_AT_external_die_ref): Likewise.
5230 (clone_die): Likewise.
5231 (clone_as_declaration): Likewise.
5232 (dwarf2out_vms_debug_main_pointer): Likewise.
5233 (base_type_die): Likewise. Remove early return for corner cases.
5234 Do not call add_pubtype on the DIE here.
5235 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5236 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5237 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5238 native order exists for base types, attach the DIE manually and call
5239 add_pubtype on it. Do not equate a reverse order DIE to the type.
5241 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5243 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5244 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5246 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5248 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5250 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5251 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5252 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5253 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5254 * opts.c (sanitizer_opts): Add builtin.
5255 * ubsan.c (instrument_builtin): New function.
5256 (pass_ubsan::execute): Call it.
5257 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5258 * doc/invoke.texi: Document -fsanitize=builtin.
5260 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5261 builtins, store max (log2 (align), 0) into uchar field instead of
5262 align into uptr field.
5263 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5264 store uchar 0 field instead of uptr 0 field.
5265 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5266 instead of passing one address of struct with 2 locations pass
5267 two addresses of structs with 1 location each.
5268 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5269 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5270 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5271 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5272 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5273 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5274 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5275 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5277 2017-10-19 Martin Liska <mliska@suse.cz>
5280 * file-find.c (remove_prefix): Remove.
5281 * file-find.h (remove_prefix): Likewise.
5282 * gcc-ar.c: Remove smartness of lookup.
5284 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5286 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5287 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5288 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5290 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5293 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5294 (setcc + and to xor + setcc): New peephole2.
5296 2017-10-19 Tom de Vries <tom@codesourcery.com>
5298 * doc/sourcebuild.texi (Test Directives, Variants of
5299 dg-require-support): Add dg-require-stack-size.
5301 2017-10-19 Martin Liska <mliska@suse.cz>
5304 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5305 that have a large alignment.
5306 (gimplify_target_expr): Likewise.
5308 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5310 PR rtl-optimization/82602
5311 * ira.c (rtx_moveable_p): Return false for volatile asm.
5313 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5316 * config/i386/i386-modes.def (CCGZ): New CC mode.
5317 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5318 * config/i386/predicates.md (ix86_comparison_operator):
5320 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5321 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5322 with double-word subtraction.
5323 (put_condition_code): Handle CCGZmode.
5325 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5327 * wide-int.cc (debug (const wide_int &)): New.
5328 (debug (const wide_int *)): New.
5329 (debug (const widest_int &)): New.
5330 (debug (const widest_int *)): New.
5332 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5335 * lra-constraints.c (curr_insn_transform): Use non-input operand
5336 instead of output one for matched reload.
5338 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5340 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5341 (tree-ssa-loop-ivopts.h): New header file.
5342 (struct builtin_info): New fields.
5343 (classify_builtin_1): Compute and record base and offset parts for
5344 memset builtin partition by calling strip_offset.
5345 (offset_cmp, fuse_memset_builtins): New functions.
5346 (finalize_partitions): Fuse adjacent memset partitions by calling
5348 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5349 Expose the interface.
5350 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5352 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5354 PR tree-optimization/82574
5355 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5356 that data reference must be executed exactly once per iteration
5357 against the outermost loop in nest.
5358 (classify_partition): Update call to above function.
5360 2017-10-18 Richard Biener <rguenther@suse.de>
5362 PR tree-optimization/82591
5363 * graphite.c (graphite_transform_loops): Move code gen message
5365 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5366 Here. Handle scop_to_isl_ast failing.
5367 (scop_to_isl_ast): Limit the number of ISL operations.
5369 2017-10-18 Richard Biener <rguenther@suse.de>
5371 * graphite-isl-ast-to-gimple.c
5372 (translate_isl_ast_to_gimple::set_rename): Simplify.
5373 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5374 (graphite_copy_stmts_from_block): ... here.
5375 (copy_bb_and_scalar_dependences): Simplify.
5376 (add_parameters_to_ivs_params): Canonicalize.
5377 (generate_entry_out_of_ssa_copies): Simplify.
5378 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5379 by passing in ISL dimension.
5380 (parameter_index_in_region_1): Rename to ...
5381 (parameter_index_in_region): ... this.
5382 (extract_affine): Adjust assert, pass down parameter index.
5383 (add_param_constraints): Use range-info when available.
5384 (build_scop_context): Adjust.
5385 * sese.c (new_sese_info): Adjust.
5386 (free_sese_info): Likewise.
5387 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5388 Remove unused typedefs.
5389 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5391 2017-10-18 Martin Liska <mliska@suse.cz>
5393 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5395 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5397 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5398 (s390_sched_init): Do not reset s390_sched_state if we entered the
5399 current basic block via a fallthru edge and all others are unlikely.
5401 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5403 * config/s390/s390.c (NUM_SIDES): New variable.
5404 (LONGRUNNING_THRESHOLD): New variable.
5405 (LATENCY_FACTOR): New variable.
5406 (s390_sched_score): Decrease score for long-running instructions on
5408 (s390_sched_variable_issue): Perform bookkeeping for long-running
5411 2017-10-18 Richard Biener <rguenther@suse.de>
5413 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5414 Simplify with removal of the parameter rename map.
5415 (set_rename): Likewise.
5416 (should_copy_to_new_region): Likewise.
5417 (graphite_copy_stmts_from_block): Likewise.
5418 (copy_bb_and_scalar_dependences): Remove initialization of
5419 unused copied_bb_map.
5421 (copy_internal_parameters): Likewise.
5422 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5423 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5424 Use INTEGRAL_TYPE_P.
5425 (parameter_index_in_region_1): Rename to ...
5426 (assign_parameter_index_in_region): ... this. Assert we have
5427 a parameter we handle.
5428 (scan_tree_for_params): Adjust.
5429 * sese.h (parameter_rename_map_t): Remove.
5430 (struct sese_info_t): Remove unused parameter_rename_map and
5431 copied_bb_map members.
5432 * sese.c (new_sese_info): Adjust.
5433 (free_sese_info): Likewise.
5435 2017-10-18 Martin Liska <mliska@suse.cz>
5438 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5439 on an abnormal edge.
5441 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5443 * doc/invoke.texi (ffunction-sections and fdata-sections):
5446 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5448 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5449 the use statement can throw internally.
5451 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5453 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5454 any RTX present on the RHS of a SET.
5455 * compare-elim.c (try_eliminate_compare): Restore comment.
5457 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5459 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5460 may be also called on tcc_type nodes.
5461 * langhooks.c (lhd_tree_size): Likewise.
5463 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5465 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5466 format_warning_at_substring.
5467 (maybe_warn): Convert source_range * param to a location_t. Pass
5468 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5469 (format_directive): Remove code to extract source_ranges and
5470 source_range * in favor of just a location_t.
5471 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5473 * substring-locations.c (format_warning_va): Convert
5474 source_range * param to a location_t.
5475 (format_warning_at_substring): Likewise.
5476 * substring-locations.h (format_warning_va): Likewise.
5477 (format_warning_at_substring): Likewise.
5479 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5481 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5483 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5484 and scatter/gather ops.
5486 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5487 vec_gather_load and vec_scatter_store.
5488 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5489 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5490 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5491 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5492 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5493 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5495 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5497 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5498 Detect FTST instruction and handle its register pops. Only pop
5499 second operand if can_pop_second_op is true.
5500 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5501 set can_pop_second_op to false in the compare_for_stack_reg call.
5503 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5504 output_fp_compare for stack register operands.
5505 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5506 instructions here. Do not emit stack register pops here. Assert
5507 that FCOMPP pops next to top stack register. Rewrite function.
5509 2017-10-17 Nathan Sidwell <nathan@acm.org>
5512 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5513 use DECL_ASSEMBLER_NAME_RAW.
5516 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5519 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5520 Wilco Dijkstra <wilco.dijkstra@arm.com>
5522 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5523 converstion to Pmode from the buf_addr.
5525 2017-10-17 Richard Biener <rguenther@suse.de>
5527 * graphite-dependences.c (scop_get_reads_and_writes): Change
5528 output parameters to references.
5530 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5532 PR 71026/tree-optimization
5533 * fold-const.c (distribute_real_division): Removed.
5534 (fold_binary_loc): Remove calls to distribute_real_divison.
5536 2017-10-17 Richard Biener <rguenther@suse.de>
5538 * graphite-scop-detection.c
5539 (scop_detection::stmt_has_simple_data_refs_p): Always use
5540 the full nest as region.
5541 (try_generate_gimple_bb): Likewise.
5542 * sese.c (scalar_evolution_in_region): Simplify now that
5543 SCEV can handle instantiation in regions.
5544 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5545 in the non-loop part of a function if requested.
5547 2017-10-17 Richard Biener <rguenther@suse.de>
5549 PR tree-optimization/82563
5550 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5552 (graphite_regenerate_ast_isl): Call it.
5553 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5555 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5557 PR tree-optimization/82549
5558 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5559 Formatting fixes. Instead of calling make_bit_field_ref with negative
5562 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5564 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5565 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5566 _mm_maskz_reduce_ss): New.
5567 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5568 __builtin_ia32_reducess_mask): Ditto..
5569 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5570 * config/i386/sse.md (reduces<mode>): Renamed to ...
5571 (reduces<mode><mask_scalar_name>): ... this.
5572 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5574 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5575 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5577 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5579 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5580 * selftest-run-tests.c (selftest::run_tests): Call
5581 selftest::unique_ptr_tests_cc_tests.
5582 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5583 * unique-ptr-tests.cc: New file.
5585 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5588 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5590 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5591 (make_hard_regno_born, make_hard_regno_dead): Update
5592 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5594 2017-10-16 Jeff Law <law@redhat.com>
5596 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5598 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5600 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5602 2017-10-16 Olivier Hainque <hainque@adacore.com>
5604 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5605 with_cpu if we were configured for an e500v2 target cpu name.
5607 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5609 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5610 * doc/invoke.texi: Document +nodsp as a valid extension for
5613 2017-10-16 Martin Liska <mliska@suse.cz>
5615 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5616 (test_set_range): Likewise.
5617 (test_range_functions): Rename to ...
5618 (test_bit_in_range): ... this.
5619 (sbitmap_c_tests): Add new test.
5621 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5623 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5625 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5626 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5628 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5630 * config/aarch64/aarch64-builtins.c
5631 (aarch64_types_quadopu_lane_qualifiers): New.
5632 (TYPES_QUADOPU_LANE): New.
5633 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5634 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5635 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5636 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5637 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5638 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5639 (Vdottype, DOTPROD): New.
5640 (sur): Add SDOT and UDOT.
5642 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5644 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5645 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5646 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5648 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5649 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5650 Enable TARGET_DOTPROD.
5651 (cortex-a75.cortex-a55): Likewise.
5652 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5654 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5656 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5657 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5659 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5661 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5662 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5663 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5664 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5665 * config/arm/types.md (neon_dot, neon_dot_q): New.
5666 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5668 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5670 * config/arm/arm.h (TARGET_DOTPROD): New.
5671 * config/arm/arm.c (arm_arch_dotprod): New.
5672 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5673 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5674 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5675 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5676 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5677 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5678 * doc/invoke.texi (armv8.2-a): Document dotprod
5680 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5682 * i386.c (ix86_vec_cost): New function.
5683 (ix86_rtx_costs): Handle vector operations better.
5684 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5685 * x86-tune-costs.h: Add new costs to all tables.
5687 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5689 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5691 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5692 divsd, sqrtss and sqrtsd
5693 * x86-tune-costs.h: Add new entries to all costs.
5694 (znver1_cost): Fix to match real instruction latencies.
5696 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5697 Michael Collison <michael.collison@arm.com>
5699 * compare-elim.c: Include emit-rtl.h.
5700 (can_merge_compare_into_arith): New function.
5701 (try_validate_parallel): Likewise.
5702 (try_merge_compare): Likewise.
5703 (try_eliminate_compare): Call the above when no previous clobber
5705 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5708 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5712 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5713 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5717 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5718 to be any operand_equal_p operands. For & (B - 1) require
5719 B to be power of 2. Recognize
5720 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5722 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5725 * optabs.c (expand_memory_blockage): Fix call of
5726 targetm.have_memory_blockage.
5728 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5731 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5732 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5733 objects to extra_objs instead of overwriting it.
5735 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5737 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5738 Use any_fp_register_operand as operand[3] predicate. Simplify
5739 equality test for operands[2] and operands[4] memory location.
5740 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5741 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5742 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5743 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5744 any_fp_register_operand as operand[1] predicate. Simplify
5745 equality test for operands[0] and operands[3] memory location.
5746 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5747 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5748 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5750 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5752 * target-insns.def: Add memory_blockage.
5753 * optabs.c (expand_memory_blockage): New function.
5754 (expand_asm_memory_barrier): Rename ...
5755 (expand_asm_memory_blockage): ... to this.
5756 (expand_mem_thread_fence): Call expand_memory_blockage
5757 instead of expand_asm_memory_barrier.
5758 (expand_mem_singnal_fence): Ditto.
5759 (expand_atomic_load): Ditto.
5760 (expand_atomic_store): Ditto.
5761 * doc/md.texi (Standard Pattern Names For Generation):
5762 Document memory_blockage instruction pattern.
5764 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5766 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5767 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5768 * config/rl78/rl78.md: New define_expand "adddi3".
5770 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5772 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5775 2017-10-13 Jeff Law <law@redhat.com>
5777 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5779 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5782 * internal-fn.c (expand_mul_overflow): If both operands have
5783 the same highpart of -1 or 0 and the topmost bit of lowpart
5784 is different, overflow is if res <= 0 rather than res < 0.
5786 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5788 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5789 TARGET_P9_VECTOR code for unaligned_load case.
5791 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5793 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5795 2017-10-13 Nathan Sidwell <nathan@acm.org>
5797 * tree-core.h (tree_contains_struct): Make bool.
5798 * tree.c (tree_contains_struct): Likewise.
5799 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5800 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5801 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5802 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5803 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5805 2017-10-13 Richard Biener <rguenther@suse.de>
5807 * graphite-isl-ast-to-gimple.c
5808 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5809 parameters and dominance check.
5810 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5811 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5812 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5813 Do not update SSA form here or do intermediate IL verification.
5814 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5815 (graphite_initialize): Remove check on the number of loops in
5816 the function and inline into graphite_transform_loops.
5817 (graphite_finalize): Inline into graphite_transform_loops.
5818 (graphite_transform_loops): Perform SSA update and IL verification
5820 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5822 2017-10-13 Richard Biener <rguenther@suse.de>
5824 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5825 graphite_expression_type_precision): Avoid global constructor
5827 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5828 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5829 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5830 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5831 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5832 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5834 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5837 * config/i386/i386.h (ix86_red_zone_size): New.
5838 * config/i386/i386.md (push peephole2s): Replace
5839 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
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 * combine.c (can_change_dest_mode): Reject changes in
5846 REGMODE_NATURAL_SIZE.
5848 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5849 Alan Hayward <alan.hayward@arm.com>
5850 David Sherwood <david.sherwood@arm.com>
5852 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5853 (expand_debug_source_expr): Likewise.
5854 * combine.c (combine_simplify_rtx): Likewise.
5855 * cse.c (fold_rtx): Likewise.
5856 * fwprop.c (canonicalize_address): Likewise.
5857 * targhooks.c (default_shift_truncation_mask): Likewise.
5859 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5860 Alan Hayward <alan.hayward@arm.com>
5861 David Sherwood <david.sherwood@arm.com>
5863 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5864 (widened_mode): Likewise.
5865 (expand_unop): Likewise.
5866 * ree.c (transform_ifelse): Likewise.
5867 (merge_def_and_ext): Likewise.
5868 (combine_reaching_defs): Likewise.
5869 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5871 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5872 Alan Hayward <alan.hayward@arm.com>
5873 David Sherwood <david.sherwood@arm.com>
5875 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5876 * combine.c (gen_lowpart_for_combine): Likewise.
5877 * dwarf2out.c (rtl_for_decl_location): Likewise.
5878 * final.c (alter_subreg): Likewise.
5879 * rtlhooks.c (gen_lowpart_general): Likewise.
5880 (gen_lowpart_if_possible): Likewise.
5882 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5883 Alan Hayward <alan.hayward@arm.com>
5884 David Sherwood <david.sherwood@arm.com>
5886 * calls.c (expand_call): Use subreg_lowpart_offset.
5887 * cse.c (cse_insn): Likewise.
5888 * regcprop.c (copy_value): Likewise.
5889 (copyprop_hardreg_forward_1): Likewise.
5891 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5894 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5895 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5896 =Q constraints to +Q and into insn condition add check
5897 that operands[0] and operands[1] are equal.
5898 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5899 =Q constraints to +Q and into insn condition add check
5900 that operands[0] is equal to either operands[1] or operands[2].
5903 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5904 instead of handling MINUS_EXPR twice (once for each argument),
5905 canonicalize operand order and handle just once, use rtype where
5906 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5909 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5910 any values of __C while still being pattern recognizable as a simple
5913 2017-10-13 Richard Biener <rguenther@suse.de>
5915 PR tree-optimization/82451
5917 2017-10-02 Richard Biener <rguenther@suse.de>
5919 PR tree-optimization/82355
5920 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5921 a mapping for the enclosing loop but avoid generating one for
5923 (copy_bb_and_scalar_dependences): Remove premature codegen
5924 error on PHIs in blocks duplicated into multiple places.
5925 * graphite-scop-detection.c
5926 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5927 in the region use it as loop and nest to analyze the DR in.
5928 (try_generate_gimple_bb): Likewise.
5929 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5930 (add_loop_constraints): For blocks in a loop not in the region
5931 create a dimension with a single iteration.
5932 * sese.h (gbb_loop_at_index): Remove assert.
5934 * cfgloop.c (loop_preheader_edge): For the loop tree root
5935 return the single successor of the entry block.
5936 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5937 Reset the SCEV hashtable and niters.
5938 * graphite-scop-detection.c
5939 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5940 assert that we only have POLYNOMIAL_CHREC that vary in loops
5941 contained in the region.
5942 (scop_detection::graphite_can_represent_expr): Adjust.
5943 (scop_detection::stmt_has_simple_data_refs_p): For loops
5944 not in the region set loop to NULL. The nest is now the
5945 entry edge to the region.
5946 (try_generate_gimple_bb): Likewise.
5947 * sese.c (scalar_evolution_in_region): Adjust for
5948 instantiate_scev change.
5949 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5950 Make nest parameter the edge into the region.
5951 (create_data_ref): Likewise.
5952 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5953 entry edge into a region and adjust instantiate_scev calls.
5954 (create_data_ref): Likewise.
5955 (graphite_find_data_references_in_stmt): Likewise.
5956 (find_data_references_in_stmt): Pass the loop preheader edge
5957 from the nest argument.
5958 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5959 parameter the edge into the region.
5960 (instantiate_parameters): Use the loop preheader edge as entry.
5961 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5963 (get_instantiated_value_entry): Make instantiate_below parameter
5964 the edge into the region.
5965 (instantiate_scev_name): Likewise. Adjust dominance checks,
5966 when we cannot use loop-based instantiation instantiate by
5967 walking use-def chains.
5968 (instantiate_scev_poly): Adjust.
5969 (instantiate_scev_binary): Likewise.
5970 (instantiate_scev_convert): Likewise.
5971 (instantiate_scev_not): Likewise.
5972 (instantiate_array_ref): Remove.
5973 (instantiate_scev_3): Likewise.
5974 (instantiate_scev_2): Likewise.
5975 (instantiate_scev_1): Likewise.
5976 (instantiate_scev_r): Do not blindly handle N-operand trees.
5977 Do not instantiate array-refs. Handle all constants and invariants.
5978 (instantiate_scev): Make instantiate_below parameter
5979 the edge into the region.
5980 (resolve_mixers): Use the loop preheader edge for the region
5981 parameter to instantiate_scev_r.
5982 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5984 2017-10-13 Richard Biener <rguenther@suse.de>
5986 PR tree-optimization/82525
5987 * graphite-isl-ast-to-gimple.c
5988 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5990 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5991 Fail code generation when we cannot represent the isl integer.
5992 (binary_op_to_tree): Elide modulo operations that are no-ops
5993 in the type we code generate. Remove now superfluous code
5996 2017-10-13 Richard Biener <rguenther@suse.de>
5998 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5999 (scop_detection::harmful_loop_in_region): Remove premature
6000 IV type restriction.
6001 (scop_detection::graphite_can_represent_scev): We can handle
6002 pointer IVs just fine.
6004 2017-10-13 Alan Modra <amodra@gmail.com>
6006 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6007 "Clobbers and Scratch Registers". Add paragraph on
6008 alternative to clobbers for scratch registers and OpenBLAS
6011 2017-10-13 Alan Modra <amodra@gmail.com>
6013 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6014 example of a memory input for a string of known length. Move
6015 commentary out of table. Add a number of new examples
6016 covering array memory inputs.
6018 2017-10-12 Martin Liska <mliska@suse.cz>
6020 PR tree-optimization/82493
6021 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6022 (test_range_functions): New function.
6023 (sbitmap_c_tests): Likewise.
6024 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6025 * selftest.h (sbitmap_c_tests): New function.
6027 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6029 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6031 * config/rs6000/amo.h: Fix spacing issue.
6033 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6036 * config/i386/i386.md (*ashl<mode>3_mask_1,
6037 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6038 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6041 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6043 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6044 (profile_probability): Set max_probability
6045 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6047 * profile-count.c (profile_probability::differs_from_p): Do not
6048 rely on max_probaiblity == 10000
6050 2017-10-12 Jeff Law <law@redhat.com>
6052 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6055 2017-10-12 Martin Sebor <msebor@redhat.com>
6059 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6060 (handle_alias_pairs): Call it.
6061 * common.opt (-Wattribute-alias): New option.
6062 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6063 * doc/invoke.texi (-Wattribute-alias): Document.
6065 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6068 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6070 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6072 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6073 (make_hard_regno_born, make_hard_regno_dead): Update
6074 bb_killed_pseudos and bb_gen_pseudos.
6076 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6078 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6080 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6082 * config/alpha/alpha.c (alpha_split_conditional_move):
6083 Use std::swap instead of manually swapping.
6084 (alpha_stdarg_optimize_hook): Ditto.
6085 (alpha_canonicalize_comparison): Ditto.
6087 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6089 * tree-loop-distribution.c (struct builtin_info): New struct.
6090 (struct partition): Refactor fields into struct builtin_info.
6091 (partition_free): Free struct builtin_info.
6092 (build_size_arg_loc, build_addr_arg_loc): Delete.
6093 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6094 information from struct builtin_info.
6095 (find_single_drs): New function refactored from classify_partition.
6096 Also moved builtin validity checks to this function.
6097 (compute_access_range, alloc_builtin): New functions.
6098 (classify_builtin_st, classify_builtin_ldst): New functions.
6099 (classify_partition): Refactor code into functions find_single_drs,
6100 classify_builtin_st and classify_builtin_ldst.
6101 (distribute_loop): Don't do runtime alias check when distributing
6103 (find_seed_stmts_for_distribution): New function.
6104 (pass_loop_distribution::execute): Refactor code finding seed
6105 stmts into above function. Support distribution for the innermost
6106 two-level loop nest. Adjust dump information.
6108 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6110 * tree-loop-distribution.c: Adjust the general comment.
6111 (NUM_PARTITION_THRESHOLD): New macro.
6112 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6113 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6114 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6115 in call to build_partition_graph.
6116 (finalize_partitions): New parameter. Make loop distribution more
6117 conservative by fusing more partitions.
6118 (distribute_loop): Don't do runtime alias check in case of loop nest
6120 (find_seed_stmts_for_distribution): New function.
6121 (prepare_perfect_loop_nest): New function.
6122 (pass_loop_distribution::execute): Refactor code finding seed stmts
6123 and loop nest into above functions. Support loop nest distribution.
6124 Adjust dump information accordingly.
6126 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6128 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6129 and set PTYPE_SEQUENTIAL for merged partition.
6131 2017-10-12 Richard Biener <rguenther@suse.de>
6133 PR tree-optimization/69728
6135 2017-09-19 Richard Biener <rguenther@suse.de>
6137 PR tree-optimization/69728
6138 * graphite-sese-to-poly.c (schedule_error): New global.
6139 (add_loop_schedule): Handle empty domain by failing the
6141 (build_original_schedule): Handle schedule_error.
6143 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6144 domain by returning an unchanged schedule.
6146 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6148 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6149 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6151 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6153 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6156 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6159 * expr.c (store_field): Don't optimize away bitsize == 0 store
6160 from CALL_EXPR with addressable return type.
6162 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6164 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6165 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6166 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6167 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6168 TARGET_ISEL instead of TARGET_ISEL<sel>.
6170 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6172 * config/rs6000/rs6000.c
6173 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6175 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6177 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6179 (reg_or_cint_operand): Fix comment.
6180 (reg_or_zero_operand): New predicate.
6181 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6182 * config/rs6000/rs6000.c (output_isel): Delete.
6183 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6184 instead of reg_or_cint_operand. Output instruction directly (not via
6186 (isel_unsigned_<mode>): Ditto.
6187 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6188 gpc_reg_operand. Add an instruction alternative for this. Output
6189 instruction directly.
6190 (*isel_reversed_unsigned_<mode>): Ditto.
6192 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6194 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6195 (TARGET_CANONICALIZE_COMPARISON): Define.
6197 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6200 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6201 Check whether the dest is REG before adding REG_EQUIV note.
6203 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6206 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6208 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6209 (make_hard_regno_born, make_hard_regno_dead): Update
6210 bb_killed_pseudos and bb_gen_pseudos.
6212 2017-10-11 Nathan Sidwell <nathan@acm.org>
6214 * incpath.h (enum incpath_kind): Name enum, prefix values.
6215 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6216 * incpath.c (heads, tails): Use INC_MAX.
6217 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6218 (merge_include_chains, split_quote_chain,
6219 register_include_chains): Update incpath_kind names.
6220 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6221 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6223 (add_framework_path, darwin_register_objc_includes): Likewise.
6224 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6226 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6228 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6229 Do not use float_operator operator predicate.
6230 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6231 * config/i386/predicates.md (float_operator): Remove predicate.
6233 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6235 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6236 (*jccxf_i387): Ditto.
6237 (*jcc<mode>_i387): Ditto.
6238 (*jccu<mode>_i387): Ditto.
6239 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6240 (*jcc_*_i387 splitters): Remove.
6241 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6242 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6243 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6246 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6248 * profile-count.h (slow_safe_scale_64bit): New function.
6249 (safe_scale_64bit): New inline.
6250 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6251 * profile-count.c: Include wide-int.h
6252 (slow_safe_scale_64bit): New.
6254 2017-10-11 Nathan Sidwell <nathan@acm.org>
6256 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6257 HAS_DECL_ASSEMBLER_NAME_P.
6258 * gimple-expr.c (gimple_decl_printable_name: Check
6259 HAS_DECL_ASSEMBLER_NAME_P too.
6260 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6261 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6262 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6263 * passes.c (rest_of_decl_compilation): Check
6264 HAS_DECL_ASSEMBLER_NAME_P too.
6265 * recog.c (verify_changes): Likewise.
6266 * tree-pretty-print.c (dump_decl_name): Likewise.
6267 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6269 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6270 (SET_DECL_ASSEMBLER_NAME): Use it.
6271 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6272 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6273 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6275 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6277 * config.gcc (i386, x86_64): Add extra objects.
6278 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6279 (ix86_min_insn_size): Declare.
6280 (ix86_issue_rate): Declare.
6281 (ix86_adjust_cost): Declare.
6282 (ia32_multipass_dfa_lookahead): Declare.
6283 (ix86_macro_fusion_p): Declare.
6284 (ix86_macro_fusion_pair_p): Declare.
6285 (ix86_bd_has_dispatch): Declare.
6286 (ix86_bd_do_dispatch): Declare.
6287 (ix86_core2i7_init_hooks): Declare.
6288 (ix86_atom_sched_reorder): Declare.
6289 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6290 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6291 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6292 (rip_relative_addr_p): Rename to ...
6293 (ix86_rip_relative_addr_p): ... this one; export.
6294 (memory_address_length): Update.
6295 (ix86_issue_rate): Move to x86-tune-sched.c.
6296 (ix86_flags_dependent): Move to x86-tune-sched.c.
6297 (ix86_agi_dependent): Move to x86-tune-sched.c.
6298 (exact_dependency_1): Move to x86-tune-sched.c.
6299 (exact_store_load_dependency): Move to x86-tune-sched.c.
6300 (ix86_adjust_cost): Move to x86-tune-sched.c.
6301 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6302 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6303 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6304 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6305 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6306 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6307 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6308 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6309 (min_insn_size): Rename to ...
6310 (ix86_min_insn_size): ... this one; export.
6311 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6312 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6313 (core2i7_first_cycle_multipass_backtrack): Move to
6314 x86-tune-sched-core.c.
6315 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6316 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6317 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6318 (ix86_avoid_jump_mispredicts): Update.
6319 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6320 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6321 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6322 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6323 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6324 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6325 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6326 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6327 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6328 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6329 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6330 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6331 (BIG): Move to ix86-tune-sched-bd.c.
6332 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6333 (enum insn_path): Move to ix86-tune-sched-bd.c.
6334 (get_mem_group): Move to ix86-tune-sched-bd.c.
6335 (is_cmp): Move to ix86-tune-sched-bd.c.
6336 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6337 (is_branch): Move to ix86-tune-sched-bd.c.
6338 (is_prefetch): Move to ix86-tune-sched-bd.c.
6339 (init_window): Move to ix86-tune-sched-bd.c.
6340 (allocate_window): Move to ix86-tune-sched-bd.c.
6341 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6342 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6343 (process_end_window): Move to ix86-tune-sched-bd.c.
6344 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6345 (find_constant): Move to ix86-tune-sched-bd.c.
6346 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6347 (has_immediate): Move to ix86-tune-sched-bd.c.
6348 (get_insn_path): Move to ix86-tune-sched-bd.c.
6349 (get_insn_group): Move to ix86-tune-sched-bd.c.
6350 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6351 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6352 (add_insn_window): Move to ix86-tune-sched-bd.c.
6353 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6354 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6355 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6356 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6357 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6358 (do_dispatch): Move to ix86-tune-sched-bd.c.
6359 (has_dispatch): Move to ix86-tune-sched-bd.c.
6360 * config/i386/t-i386: Add new object files.
6361 * config/i386/x86-tune-costs.h: New file.
6362 * config/i386/x86-tune-sched-atom.c: New file.
6363 * config/i386/x86-tune-sched-bd.c: New file.
6364 * config/i386/x86-tune-sched-core.c: New file.
6365 * config/i386/x86-tune-sched.c: New file.
6367 2017-10-11 Liu Hao <lh_mouse@126.com>
6369 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6370 the generic version below instead.
6371 (should_colorize): Recognize Windows consoles as terminals
6373 * pretty-print.c [__MINGW32__] (write_all): New function.
6374 [__MINGW32__] (find_esc_head): Likewise.
6375 [__MINGW32__] (find_esc_terminator): Likewise.
6376 [__MINGW32__] (eat_esc_sequence): Likewise.
6377 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6379 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6382 2017-10-11 Richard Biener <rguenther@suse.de>
6384 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6385 Properly call analyze_scalar_evolution with the loop of the stmt.
6387 2017-10-11 Richard Biener <rguenther@suse.de>
6389 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6390 * tree-core.h (tree_base): Add chrec_var union member.
6391 * tree.h (CHREC_VAR): Remove.
6392 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6393 * tree-chrec.h (build_polynomial_chrec): Adjust.
6394 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6395 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6397 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6399 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6400 * match.pd: ... here.
6401 ((T) X == (T) Y): Relax condition.
6403 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6405 PR tree-optimization/82472
6406 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6408 (break_alias_scc_partitions): Update postorder number.
6410 2017-10-11 Martin Liska <mliska@suse.cz>
6413 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6415 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6417 2017-10-11 Martin Liska <mliska@suse.cz>
6419 * print-rtl.c (print_insn): Move declaration of idbuf
6420 to same scope as name.
6422 2017-10-11 Martin Liska <mliska@suse.cz>
6427 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6430 2017-10-11 Martin Liska <mliska@suse.cz>
6433 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6436 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6438 * config.gcc (default_gnu_indirect_function): Default to yes for
6439 arm*-*-linux* with glibc.
6441 2017-10-11 Richard Biener <rguenther@suse.de>
6443 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6444 default-defs and types we do not want to analyze.
6445 (interpret_loop_phi): Replace unreachable code with an assert.
6446 (compute_scalar_evolution_in_loop): Remove and inline ...
6447 (analyze_scalar_evolution_1): ... here, replacing condition with
6448 what makes the intent clearer. Remove handling of cases
6449 get_scalar_evolution now handles.
6451 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6453 PR rtl-optimization/81434
6454 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6455 comment for main loop. In sched_group_found if, also add checks for
6456 pass and min_cost_group.
6458 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6460 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6461 (rs6000_insn_cost): New function.
6462 * config/rs6000/rs6000.md (cost): New attribute.
6464 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6465 H.J. Lu <hongjiu.lu@intel.com>
6469 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6470 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6471 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6472 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6473 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6474 OPTION_MASK_ISA_MMX and some other bit set require both
6475 mmx and the ISAs without the mmx bit.
6476 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6477 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6478 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6479 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6480 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6481 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6482 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6483 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6484 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6485 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6486 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6487 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6488 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6489 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6490 Add OPTION_MASK_ISA_MMX.
6492 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6494 * config.gcc (armv7*-*-freebsd*): New target.
6495 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6497 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6499 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6500 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6503 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6505 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6506 a template parameter.
6507 (WIDE_INT_REF_FOR): Update accordingly.
6508 * tree.h (wi::int_traits <const_tree>): Delete.
6509 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6510 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6511 (wi::tree_to_wide_ref): New typedef.
6512 (wi::to_wide): New function.
6513 * calls.c (get_size_range): Use wi::to_wide when operating on
6515 * cgraph.c (cgraph_node::create_thunk): Likewise.
6516 * config/i386/i386.c (ix86_data_alignment): Likewise.
6517 (ix86_local_alignment): Likewise.
6518 * dbxout.c (stabstr_O): Likewise.
6519 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6520 * expr.c (const_vector_from_tree): Likewise.
6521 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6522 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6523 (fold_negate_expr_1, int_const_binop_1, const_binop)
6524 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6525 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6526 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6527 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6528 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6529 (fold_not_const, round_up_loc): Likewise.
6530 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6531 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6532 (alloca_call_type): Likewise.
6533 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6534 * godump.c (go_output_typedef): Likewise.
6535 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6536 * internal-fn.c (get_min_precision): Likewise.
6537 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6538 * ipa-polymorphic-call.c
6539 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6540 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6541 (ipa_modify_call_arguments): Likewise.
6542 * match.pd: Likewise.
6543 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6544 * print-tree.c (print_node_brief, print_node): Likewise.
6545 * stmt.c (expand_case): Likewise.
6546 * stor-layout.c (layout_type): Likewise.
6547 * tree-affine.c (tree_to_aff_combination): Likewise.
6548 * tree-cfg.c (group_case_labels_stmt): Likewise.
6549 * tree-data-ref.c (dr_analyze_indices): Likewise.
6550 (prune_runtime_alias_test_list): Likewise.
6551 * tree-dump.c (dequeue_and_dump): Likewise.
6552 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6553 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6554 * tree-pretty-print.c (dump_generic_node): Likewise.
6555 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6556 (simple_iv_with_niters): Likewise.
6557 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6558 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6559 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6560 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6561 (refine_value_range_using_guard, number_of_iterations_ne_max)
6562 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6563 (get_cst_init_from_scev, record_nonwrapping_iv)
6564 (scev_var_range_cant_overflow): Likewise.
6565 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6566 * tree-ssa-pre.c (compute_avail): Likewise.
6567 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6568 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6569 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6570 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6571 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6572 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6573 (dump_case_nodes, try_switch_expansion): Likewise.
6574 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6575 (vect_do_peeling): Likewise.
6576 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6577 * tree-vect-stmts.c (vectorizable_load): Likewise.
6578 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6579 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6580 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6581 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6582 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6583 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6584 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6585 (evrp_dom_walker::before_dom_children): Likewise.
6586 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6587 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6588 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6589 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6590 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6591 (get_range_pos_neg): Likewise.
6592 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6593 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6594 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6595 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6596 * config/avr/avr.c (avr_fold_builtin): Likewise.
6597 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6598 * config/msp430/msp430.c (msp430_attr): Likewise.
6599 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6600 * config/powerpcspe/powerpcspe-c.c
6601 (altivec_resolve_overloaded_builtin): Likewise.
6602 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6603 (rs6000_expand_ternop_builtin): Likewise.
6604 * config/rs6000/rs6000-c.c
6605 (altivec_resolve_overloaded_builtin): Likewise.
6606 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6607 (rs6000_expand_ternop_builtin): Likewise.
6608 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6610 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6612 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6613 when copying loop nest with only one inner loop.
6615 2017-10-10 Richard Biener <rguenther@suse.de>
6617 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6618 blocks if SCEV is active.
6619 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6621 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6622 (scev_initialize): Assert we are not yet initialized.
6624 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6626 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6627 inner loop's exit stmt by making it always exit the loop, otherwise
6628 we would generate an infinite empty loop.
6630 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6632 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6633 renaming variables in new preheader if it's deleted.
6635 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6637 * tree-loop-distribution.c (struct partition): Remove unused field
6638 loops of the structure.
6639 (partition_alloc, partition_free): Ditto.
6640 (build_rdg_partition_for_vertex): Ditto.
6642 2017-10-09 Jeff Law <law@redhat.com>
6644 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6645 return type to match prototype and documentation.
6647 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6649 * config/rs6000/rs6000.c (processor_costs): Move to ...
6650 * config/rs6000/rs6000.h: ... here.
6651 (rs6000_cost): Declare.
6653 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6655 * except.c (setjmp_fn): New global variable.
6656 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6657 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6658 if DONT_USE_BUILTIN_SETJMP is defined.
6660 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6662 * target.def (insn_cost): New hook.
6663 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6664 * doc/tm.texi: Regenerate.
6665 * rtlanal.c (insn_cost): Use the new hook.
6667 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6669 * combine.c (combine_validate_cost): Compute the new insn_cost,
6670 not just pattern_cost.
6671 (try_combine): Adjust comment.
6673 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6675 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6677 * combine.c (uid_insn_cost): Adjust comment.
6678 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6680 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6681 * dse.c (find_shift_sequence): Ditto.
6682 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6683 (bb_valid_for_noce_process_p): Use pattern_cost.
6684 * rtl.h (insn_rtx_cost): Delete.
6685 (pattern_cost): New prototype.
6686 (insn_cost): New prototype.
6687 * rtlanal.c (insn_rtx_cost): Rename to...
6688 (pattern_cost): ... this.
6691 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6693 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6694 (*jcc<mode>_0_r_i387): Ditto.
6695 (*jccxf_r_i387): Ditto.
6696 (*jcc<mode>_r_i387): Ditto.
6697 (*jccu<mode>_r_i387): Ditto.
6698 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6699 (*jcc): Rename from *jcc_1.
6701 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6703 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6704 deferred rescans after the lvx/stvx recombination pre-pass.
6706 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6708 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6709 memory operation instruction support.
6710 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6711 (rs6000-ibm-aix[789]*): Likewise.
6712 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6713 Document new functions.
6715 2017-10-09 Richard Biener <rguenther@suse.de>
6717 PR tree-optimization/82397
6718 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6719 equality only for semantically equal trees.
6721 2017-10-09 Richard Biener <rguenther@suse.de>
6723 PR tree-optimization/82449
6724 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6725 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6726 allow constant addresses.
6727 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6730 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6732 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6735 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6738 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6741 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6744 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6746 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6749 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6750 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6752 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6754 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6755 (WI_BINARY_PREDICATE_RESULT): Likewise.
6756 (wi::binary_traits::operator_result): New type.
6757 (wi::binary_traits::predicate_result): Likewise.
6758 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6759 (generic_wide_int::operator==, generic_wide_int::operator!=)
6760 (generic_wide_int::operator&, generic_wide_int::and_not)
6761 (generic_wide_int::operator|, generic_wide_int::or_not)
6762 (generic_wide_int::operator^, generic_wide_int::operator+
6763 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6765 (operator~, unary operator-, operator==, operator!=, operator&)
6766 (operator|, operator^, operator+, binary operator-, operator*): New
6768 * expr.c (get_inner_reference): Use wi::bit_and_not.
6769 * fold-const.c (fold_binary_loc): Likewise.
6770 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6771 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6772 (bit_value_binop): Likewise.
6773 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6774 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6775 (extract_range_from_binary_expr_1): Likewise.
6776 (masked_increment): Likewise.
6777 (simplify_bit_ops_using_ranges): Likewise.
6779 2017-10-09 Martin Jambor <mjambor@suse.cz>
6782 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6783 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6784 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6786 (hsa_fixup_mov_insn_type): New function.
6787 (hsa_op_with_type::get_in_type): Use it.
6788 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6789 immediates in an assert.
6790 (hsa_op_with_type::extend_int_to_32bit): New method.
6791 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6792 types. Convert to dest type if necessary.
6793 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6794 (reg_for_gimple_ssa): Pass false as min32int to
6795 hsa_type_for_scalar_tree_type.
6796 (gen_hsa_addr): Fixup type when creating addresable temporary.
6797 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6798 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6799 necessary. Call hsa_fixup_mov_insn_type.
6800 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6801 extend operands and convert to dest type if necessary.
6802 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6803 to dest type if necessary.
6804 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6805 if conversion nt necessary and size matches.
6806 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6807 to dest type if necessary.
6808 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6809 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6811 (gen_hsa_clrsb): Likewise.
6812 (gen_hsa_ffs): Likewise.
6813 (gen_hsa_divmod): Extend operands and convert to dest type if
6815 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6817 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6819 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6820 Remove empty default arguments. Use a brace block as output
6822 (conditional return): Ditto.
6824 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6825 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6827 (group_ending_nop): Ditto.
6828 (doloop_end): Ditto.
6829 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6830 (splitters for those): Ditto.
6832 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6834 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6835 a conditional jump (and the compare for it) so that pc_rtx is the
6837 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6838 for the deleted and renamed ctr<mode>_internal[234] patterns.
6839 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6840 Delete second conditional return pattern.
6841 (ctr<mode>_internal2): Delete this second bdnz pattern.
6842 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6843 (ctr<mode>_internal4): Delete this second bdz pattern.
6845 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6847 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6848 (always_initialized_rtx_for_ssa_name_p): New predicate.
6849 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6850 (finish_out_of_ssa): Free new field of SA.
6851 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6852 * tree-ssa-coalesce.c: Include tree-ssa.h.
6853 (get_parm_default_def_partitions): Remove extern keyword.
6854 (get_undefined_value_partitions): New function.
6855 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6856 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6858 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6860 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6862 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6864 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6866 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6867 for targets that preffer 128bit.
6869 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6871 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6873 2017-10-08 Olivier Hainque <hainque@adacore.com>
6875 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6876 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6878 (thumb_set_return_address): Likewise.
6880 2017-10-08 Olivier Hainque <hainque@adacore.com>
6882 * common/config/arm/arm-common.c (arm_except_unwind_info):
6883 Handle DWARF2_UNWIND_INFO.
6885 2017-10-07 Michael Collison <michael.collison@arm.com>
6887 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6890 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6892 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6893 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6894 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6895 defined, force the creation of a new block for a dispatch label.
6897 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6899 * invoke.texi (Wsuggest-attribute=cold): Document.
6900 * common.opt (Wsuggest-attribute=cold): New
6901 * ipa-pure-const.c (warn_function_cold): New function.
6902 * predict.c (compute_function_frequency): Use it.
6903 * predict.h (warn_function_cold): Declare.
6905 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6907 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6910 2017-10-06 Martin Liska <mliska@suse.cz>
6912 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6913 keyword for member functions.
6914 (struct sanopt_tree_couple): New struct.
6915 (struct sanopt_tree_couple_hash): New function.
6916 (struct sanopt_ctx): Add new hash_map.
6917 (has_dominating_ubsan_ptr_check): New function.
6918 (record_ubsan_ptr_check_stmt): Likewise.
6919 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6920 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6921 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6923 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6926 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6927 aarch64_simd_valid_immediate on CONST_VECTORs.
6928 (aarch64_reg_or_bic_imm): Likewise.
6930 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6932 PR rtl-optimization/82396
6933 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6935 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6937 * graphite-dependences.c (scop_get_reads): Move code to...
6938 (scop_get_must_writes): Move code to...
6939 (scop_get_may_writes): Move code to...
6940 (scop_get_reads_and_writes): ... here.
6941 (scop_get_dependences): Call scop_get_reads_and_writes.
6943 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6945 PR tree-optimization/82434
6946 * fold-const.h (can_native_encode_type_p,
6947 can_native_encode_string_p): Remove.
6948 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6949 don't encode anything, just return what would be otherwise returned.
6950 (native_encode_fixed, native_encode_complex, native_encode_vector):
6952 (native_encode_string): Likewise. Inline by hand
6953 can_native_encode_string_p.
6954 (can_native_encode_type_p): Remove.
6955 (can_native_encode_string_p): Remove.
6956 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6957 STRING_CSTs using can_native_encode_string_p, test all
6958 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6959 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6960 argument from native_encode_expr.
6961 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6962 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6965 2017-10-06 Richard Biener <rguenther@suse.de>
6967 PR tree-optimization/82397
6968 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6969 operand_equal_p but rely on data_ref_compare_tree for detecting
6971 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6972 to match up with dr_group_sort_cmp.
6974 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6977 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6979 * config/s390/s390-builtin-types.def: Regenerate.
6981 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6984 * config/s390/s390-builtin-types.def: Regenerate.
6985 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6986 Change flag from B_VXE to B_VX.
6987 (s390_vec_min_dbl): Remove B_VXE flag.
6989 2017-10-06 Richard Biener <rguenther@suse.de>
6991 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6992 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6993 translate_isl_ast_to_gimple::is_valid_rename,
6994 translate_isl_ast_to_gimple::get_rename,
6995 translate_isl_ast_to_gimple::get_def_bb_for_const,
6996 translate_isl_ast_to_gimple::get_new_name,
6997 translate_isl_ast_to_gimple::collect_all_ssa_names,
6998 translate_isl_ast_to_gimple::copy_loop_phi_args,
6999 translate_isl_ast_to_gimple::collect_all_ssa_names,
7000 translate_isl_ast_to_gimple::copy_loop_phi_args,
7001 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7002 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7003 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7004 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7005 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7006 translate_isl_ast_to_gimple::copy_cond_phi_args,
7007 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7008 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7009 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7010 translate_isl_ast_to_gimple::rename_uses,
7011 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7012 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7013 (set_rename_for_each_def): Likewise.
7014 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7015 here. Handle rewriting SCEV analyzable uses here.
7016 (copy_bb_and_scalar_dependences): Generate code for PHI
7018 (graphite_regenerate_ast_isl): Adjust.
7019 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7020 (add_write, add_read): New functions.
7021 (build_cross_bb_scalars_def): Use it and simplify.
7022 (build_cross_bb_scalars_use): Likewise.
7023 (graphite_find_cross_bb_scalar_vars): Inline into...
7024 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7025 simulating out-of-SSA. Compute liveout and add dependencies.
7026 (build_scops): Force an empty entry block.
7027 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7029 (sese_build_liveouts): Declare.
7030 (sese_trivially_empty_bb_p): Likewise.
7031 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7032 compute liveout and debug_liveout.
7033 (sese_bad_liveouts_use): Remove.
7034 (sese_reset_debug_liveouts_bb): Likewise.
7035 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7036 (sese_build_liveouts): Build liveout and debug_liveout and store
7038 (new_sese_info): Adjust.
7039 (free_sese_info): Likewise.
7040 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7041 do not build liveout here.
7042 (move_sese_in_condition): Adjust region entry.
7043 (scev_analyzable_p): Match up with chrec_apply requirements.
7044 (sese_trivially_empty_bb_p): New.
7045 * tree-into-ssa.c (get_reaching_def): Properly support generating
7046 default-defs for incremental rewrite of anonymous names.
7048 2017-10-06 Richard Biener <rguenther@suse.de>
7050 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7051 precision do not perform modulo reduction.
7053 2017-10-06 Richard Biener <rguenther@suse.de>
7055 PR tree-optimization/82436
7056 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7057 conservatively choose the vectorization factor when checking
7058 whether we can perform the required load permutation.
7059 (vect_transform_slp_perm_load): Assert when we may not fail.
7061 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7063 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7064 message for incompatible -msdata=* and -mcall-* options.
7066 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7068 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7069 rate for post-reload scheduling.
7071 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7073 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7075 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7077 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7078 to improve monte carlo in scimark.
7080 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7082 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7083 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7084 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7085 pentium4_cost, nocona_cost): Set reassociation width to 1.
7086 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7087 width to 2 for fp operations and 1 otherwise.
7088 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7090 (atom_cost): Set reassociation width to 2.
7091 (slm_cost, generic_cost): Set fp reassociation width
7092 to 2 and 1 otherwise.
7093 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7094 (core_cost): Set fp reassociation width to 4 and vector to 2.
7095 (ix86_reassociation_width): Rewrite using cost table; special case
7096 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7097 and TARGET_AVX128_OPTIMAL.
7098 * config/i386/i386.h (processor_costs): Add
7099 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7100 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7101 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7102 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7103 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7104 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7106 2017-10-05 Nathan Sidwell <nathan@acm.org>
7108 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7110 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7112 * config/arm/arm.c (arm_test_fpu_data): New.
7113 (arm_run_selftests): Call arm_test_fpu_data.
7115 2017-10-04 Nathan Sidwell <nathan@acm.org>
7117 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7119 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7121 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7123 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7124 check type for aarch64_simd_valid_immediate.
7125 (aarch64_output_simd_mov_immediate): Update prototype.
7126 (aarch64_simd_valid_immediate): Update prototype.
7127 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7128 support for ORR-immediate.
7129 (and<mode>3): modified pattern to add support for BIC-immediate.
7130 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7131 now checks for valid immediate for BIC and ORR based on new enum
7133 (aarch64_output_simd_mov_immediate): Function now used to output
7134 BIC/ORR imm as well based on new enum argument.
7135 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7137 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7138 (aarch64_reg_or_bic_imm): Likewise.
7140 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7142 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7143 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7145 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7149 PR rtl-optimization/82396
7150 * haifa-sched.c (autopref_multipass_init): Simplify
7152 (autopref_rank_data): Simplify sort order.
7153 * sched-int.h (autopref_multipass_data_): Remove
7154 multi_mem_insn_p, min_offset and max_offset.
7156 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7158 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7160 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7162 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7165 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7167 * doc/sourcebuild.texi: Document vect_long_mult.
7169 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7171 PR tree-optimization/82413
7172 * fold-const.c (build_range_check): Use widest_int when comparing
7173 the maximum ETYPE value with HIGH.
7175 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7177 PR rtl-optimization/82396
7178 * haifa-sched.c (autopref_multipass_init): Simplify
7180 (autopref_rank_data): Simplify sort order.
7181 * sched-int.h (autopref_multipass_data_): Remove
7182 multi_mem_insn_p, min_offset and max_offset.
7184 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7186 PR tree-optimization/82381
7187 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7188 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7191 PR tree-optimization/82374
7192 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7193 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7194 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7195 current_function_decl to the new decl.
7197 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7199 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7200 helper macro for IEEE float128 hardware built-in functions.
7201 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7203 (TRUNCF128_ODD): Likewise.
7204 (ADDF128_ODD): Likewise.
7205 (SUBF128_ODD): Likewise.
7206 (MULF128_ODD): Likewise.
7207 (DIVF128_ODD): Likewise.
7208 (FMAF128_ODD): Likewise.
7209 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7210 UNSPEC_TRUNC_ROUND_TO_ODD.
7211 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7212 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7213 floating point round to odd instructions.
7214 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7215 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7216 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7217 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7218 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7219 (trunc<mode>sf2_hw): Change the truncate with round to odd
7220 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7221 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7222 to odd hardware instructions.
7223 (sub<mode>3_odd): Likewise.
7224 (mul<mode>3_odd): Likewise.
7225 (div<mode>3_odd): Likewise.
7226 (sqrt<mode>2_odd): Likewise.
7227 (fma<mode>4_odd): Likewise.
7228 (fms<mode>4_odd): Likewise.
7229 (nfma<mode>4_odd): Likewise.
7230 (nfms<mode>4_odd): Likewise.
7231 (trunc<mode>df2_odd): Change the truncate with round to odd
7232 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7234 * doc/extend.texi (PowerPC built-in functions): Update documentation
7235 for existing IEEE float128-bit built-in functions. Add built-in
7236 functions that generate the IEEE 128-bit floating point round to
7239 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7241 PR rtl-optimization/77729
7242 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7243 to (X&(C1&~C2))|C2 transformations.
7245 2017-10-03 Martin Jambor <mjambor@suse.cz>
7247 PR tree-optimization/82363
7248 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7249 mismatch, mark lacc written regardless of racc.
7251 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7253 PR tree-optimization/82381
7254 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7255 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7256 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7257 fallthrough into reassoc_stmt_dominates_stmt_p.
7260 * combine.c (combine_instructions): Don't combine in unreachable
7263 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7266 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7267 function to not use the have_cpu variable. Do not set cpu_index,
7268 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7270 (rs6000_valid_attribute_p): Remove duplicate initializations of
7271 old_optimize and func_optimize.
7272 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7273 (rs6000_activate_target_options): Make global.
7274 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7277 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7279 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7280 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7281 Set *poffset to 0 and *psize and *pmax_size to -1 if
7282 *poffset + *psize overflows in HOST_WIDE_INT.
7284 PR tree-optimization/82387
7285 PR tree-optimization/82388
7286 PR tree-optimization/82389
7287 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7288 instead of live_bytes non-NULL.
7290 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7293 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7296 2017-10-02 Richard Biener <rguenther@suse.de>
7298 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7299 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7300 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7302 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7304 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7305 requested precision matches the type's.
7306 * calls.c (alloc_max_size): Calculate the new candidate size as
7307 a widest_int and use wi::to_widest when comparing it with the
7308 current candidate size.
7309 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7310 zero rather than integer_zero_node.
7311 * match.pd: Check for a no-op conversion before using wi::add
7312 rather than after. Use tree_to_uhwi when summing small shift
7313 counts into an unsigned int.
7315 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7316 Alan Hayward <alan.hayward@arm.com>
7317 David Sherwood <david.sherwood@arm.com>
7320 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7321 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7322 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7323 POINTER_AND_FP_REGS.
7325 2017-10-02 Richard Biener <rguenther@suse.de>
7327 PR tree-optimization/82355
7328 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7329 a mapping for the enclosing loop but avoid generating one for
7331 (copy_bb_and_scalar_dependences): Remove premature codegen
7332 error on PHIs in blocks duplicated into multiple places.
7333 * graphite-scop-detection.c
7334 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7335 in the region use it as loop and nest to analyze the DR in.
7336 (try_generate_gimple_bb): Likewise.
7337 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7338 (add_loop_constraints): For blocks in a loop not in the region
7339 create a dimension with a single iteration.
7340 * sese.h (gbb_loop_at_index): Remove assert.
7342 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7344 * omp-expand.c (adjust_context_scope): New function.
7345 (expand_parallel_call): Call adjust_context_scope.
7347 2017-10-01 Jeff Law <law@redhat.com>
7349 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7350 dom_opt_dom_walker class with direct access to private members.
7351 Add comments. Call test_for_singularity.
7352 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7353 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7354 m_dummy_cond anymore.
7355 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7357 (pass_dominator:execute): Build the dummy_cond here and pass it
7358 to the dom_opt_dom_walker ctor.
7359 (test_for_singularity): New function.
7361 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7362 Maya Rashish <coypu@sdf.org>
7364 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7365 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7366 (alpha*-*-netbsd*) Use nbsd_tm_file.
7367 (arm*-*-netbsdelf*) Likewise.
7368 (i[34567]86-*-netbsdelf*) Likewise.
7369 (x86_64-*-netbsd*) Likewise.
7370 (mips*-*-netbsd*) Likewise.
7371 (powerpc-*-netbsd*) Likewise.
7372 (sh*-*-netbsd*) Likewise.
7373 (sparc-*-netbsdelf*) Likewise.
7374 (sparc64-*-netbsd*) Likewise.
7375 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7377 (vax-*-netbsdelf*) Likewise.
7378 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7379 (UINT_FAST8_TYPE) Likewise.
7380 (INT_FAST16_TYPE) Check CHAR_FAST16.
7381 (UINT_FAST16_TYPE) Likewise.
7383 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7386 * config/i386/i386.md
7387 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7388 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7389 *divmodsi4_zext_2): New define_insn_and_split.
7390 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7391 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7392 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7393 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7394 New define_insn_and_split.
7395 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7396 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7397 operands[1] having DImode when mode is SImode.
7399 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7400 always SImode for DIV and MOD in REG_EQUAL notes.
7402 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7405 * match.pd: Fix handling of NaNs in pattern.
7407 2017-09-29 Jeff Law <law@redhat.com>
7409 * sbitmap.c (bitmap_bit_in_range_p): New function.
7410 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7411 * tree-ssa-dse.c (live_bytes_read): New function.
7412 (dse_classify_store): Ignore reads of dead bytes.
7414 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7415 typos and whitespace errors.
7416 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7417 * config/s390/s390.c (s390_emit_prologue): Likewise.
7419 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7422 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7423 with a symbol for LRA.
7425 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7427 PR rtl-optimization/82338
7428 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7430 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7432 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7433 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7434 (qsort_chk): Declare.
7435 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7436 (qsort_chk): New function.
7438 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7440 PR tree-optimization/82337
7441 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7442 phi definition if the PHI result appears in an abnormal PHI.
7443 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7444 basis appears in an abnormal PHI.
7446 2017-09-29 Richard Biener <rguenther@suse.de>
7448 * graphite-isl-ast-to-gimple.c
7449 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7450 (binary_op_to_tree): Use it.
7451 (get_rename_from_scev): Likewise.
7452 (copy_loop_phi_nodes): Likewise.
7453 (copy_bb_and_scalar_dependences): Likewise.
7454 (translate_pending_phi_nodes): Likewise.
7456 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7459 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7460 for movabsq $(i32 << shift), r64.
7462 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7464 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7465 index when encoding %esp as %rsp to avoid 0x67 prefix.
7467 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7469 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7470 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7472 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7474 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7475 Extensions with more than 16 double VFP registers.
7476 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7477 to_clear_mask and all code related to it. Replace the remaining
7478 entry by a sbitmap and adapt code accordingly.
7480 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7482 * brig-builtins.def: Change pure attributes to const.
7484 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7486 * config.gcc (default_gnu_indirect_function): Default to yes for
7487 sparc*-*-linux* with glibc.
7489 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7491 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7492 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7493 when creating .init_array and .fini_array sections with priority
7496 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7499 * config/aarch64/aarch64.c
7500 (aarch64_builtin_support_vector_misalignment): Always return false
7501 when misalignment is unknown.
7503 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7505 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7506 this function to return false if the definition used by the swap
7507 instruction is artificial, or if the memory address from which the
7508 constant value is loaded is not represented by a base address held
7509 in a register or if the base address register is a frame or stack
7510 pointer. Additionally, return false if the base address of the
7511 loaded constant is a SYMBOL_REF but is not considered to be a
7513 (replace_swapped_load_constant): New function.
7514 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7515 loaded constant vector with a load of a swapped constant vector.
7517 2017-09-27 Carl Love <cel@us.ibm.com>
7519 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7520 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7521 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7524 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7526 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7527 first, always call autopref_rank_data otherwise.
7529 2017-09-27 Richard Biener <rguenther@suse.de>
7531 * graphite-scop-detection.c (find_scop_parameters): Move
7532 loop bound handling ...
7533 (gather_bbs::before_dom_children): ... here, avoiding the need
7534 to build scop_info->loop_nest.
7535 (record_loop_in_sese): Remove.
7536 * sese.h (sese_info_t::loop_nest): Remove.
7537 * sese.c (new_sese_info): Do not allocate loop_nest.
7538 (free_sese_info): Do not free loop_nest.
7540 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7543 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7544 lhs from calls if the lhs has addressable type.
7546 2017-09-27 Richard Biener <rguenther@suse.de>
7548 * graphite.h (scop::max_alias_set): New member.
7549 * graphite-scop-detection.c: Remove references to non-existing
7550 --param in comments.
7551 (build_alias_sets): Record the maximum alias set used for drs.
7552 (build_scops): Support zero as unlimited for
7553 --param graphite-max-arrays-per-scop.
7554 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7556 (build_poly_sr_1): ... here. Compute alias set based on the
7557 maximum alias set used for drs rather than
7558 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7560 2017-09-27 Richard Biener <rguenther@suse.de>
7562 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7563 --param loop-block-tile-size=0 to disable tiling.
7565 2017-09-27 Richard Biener <rguenther@suse.de>
7567 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7568 (graphite-max-nb-scop-params): Document special value zero.
7569 * domwalk.h (dom_walker::STOP): New symbolical constant.
7570 (dom_walker::dom_walker): Add optional parameter for bb to
7572 (dom_walker::~dom_walker): Declare.
7573 (dom_walker::before_dom_children): Document STOP return value.
7574 (dom_walker::m_user_bb_to_rpo): New member.
7575 (dom_walker::m_bb_to_rpo): Likewise.
7576 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7577 mapping here if not provided by the user.
7578 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7579 provided by the user.
7580 (dom_walker::STOP): Define.
7581 (dom_walker::walk): Do not compute bb to RPO mapping here.
7582 Support STOP return value from before_dom_children to stop
7584 * graphite-optimize-isl.c (optimize_isl): If the schedule
7585 is the same still generate code if -fgraphite-identity
7586 or -floop-parallelize-all are given.
7587 * graphite-scop-detection.c: Include cfganal.h.
7588 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7590 (gather_bbs::before_dom_children): Return STOP for BBs
7592 (build_scops): Compute bb to RPO mapping and pass it to
7593 the domwalk. Treat --param graphite-max-nb-scop-params=0
7594 as not limiting the number of params.
7595 * graphite.c (graphite_initialize): Remove limit on the
7596 number of basic-blocks in a function.
7597 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7598 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7599 default value of 10.
7601 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7603 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7604 Adjust code to eliminate needing to do the shift right 32-bits
7605 operation after XSCVDPSPN.
7607 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7609 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7610 ((X / Y) != 0 -> X >= Y): Likewise.
7612 2017-09-26 Carl Love <cel@us.ibm.com>
7614 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7615 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7616 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7617 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7618 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7619 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7620 definitions and overloading.
7621 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7622 statement for P9V_BUILTIN_XST_LEN_R.
7623 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7624 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7625 define_expand and define_insn for the instructions and builtins.
7626 * doc/extend.texi: Update the built-in documentation file for the new
7628 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7629 define_insn for the instructions
7631 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7634 * gcc/config/netbsd-protos.h: New file.
7635 * gcc/config/netbsd.c: New file.
7636 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7637 * gcc/config/t-netbsd: New file.
7638 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7639 (tmake_file) Add t-netbsd.
7640 (extra_objs) Add netbsd.o.
7642 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7646 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7648 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7650 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7651 sign extension from a vector register to a GPR by doing a 32-bit
7652 direct move and then an EXTSW.
7653 (extendsi<mode>2 splitter): Likewise.
7654 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7655 right or vector extract after doing XSCVDPSPN. Use
7656 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7658 (movdi_from_sf_zero_ext): Likewise.
7659 (reload_gpr_from_vsxsf): Likewise.
7660 (p8_mfvsrd_4_disf): Delete, no longer used.
7661 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7662 then needing to move the SFmode to a GPR to use the XSCVDPSP
7663 instruction instead of FRSP and XSCVDPSPN.
7664 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7665 it is adjacent to the other XSCVSPDP insns.
7666 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7667 SFmode to be in traditional Altivec registers.
7668 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7669 (vsx_xscvspdpn): Likewise.
7670 (vsx_xscvspdpn_scalar): Likewise.
7672 2017-09-26 Martin Jambor <mjambor@suse.cz>
7674 * tree-sra.c (compare_access_positions): Put integral types first,
7675 stabilize sorting of integral types, remove conditions putting
7676 non-full-precision integers last.
7677 (sort_and_splice_var_accesses): Disable scalarization if a
7678 non-integert would be represented by a non-full-precision integer.
7680 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7682 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7683 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7684 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7685 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7686 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7687 conditionals inside the function instead of around it. Call
7688 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7689 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7691 2017-09-26 Richard Biener <rguenther@suse.de>
7693 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7695 (scop_detection::build_scop_breadth): ... this. Removed.
7696 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7697 (scop_detection::harmful_stmt_in_bb): Likewise.
7698 (scop_detection::graphite_can_represent_stmt): Likewise.
7699 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7700 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7701 (scop_detection::can_represent_loop_1): ... this. Removed.
7702 (scop_detection::harmful_loop_in_region): Simplify after inlining
7703 the above and remove more quadraticness.
7704 (build_scops): Adjust.
7705 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7708 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7711 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7712 REGNO (base) == SP_REG if base is a REG.
7715 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7716 if it is different SSA_NAME.
7717 (optimize_range_tests_cmp_bitwise): New function.
7718 (optimize_range_tests): Call it.
7720 2017-09-26 Richard Biener <rguenther@suse.de>
7722 PR tree-optimization/82321
7723 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7724 for the def being inside the loop.
7726 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7728 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7730 * config/s390/s390-builtins.def: Fix constraint on op4.
7732 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7734 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7735 independent expanders.
7736 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7737 ("vec_ordered", "vec_unordered"): New expanders.
7739 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7741 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7744 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7746 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7747 vec_unpacks_lo_v16qi.
7748 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7750 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7752 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7753 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7754 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7756 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7758 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7760 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7762 ("*vec_slb<mode>"): New insn pattern.
7763 ("vec_shr_<mode>"): New expander.
7764 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7765 and force the shift count operand to V16QImode.
7766 ("vec_srb<mode>"): Set shift count mode to V16QI.
7768 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7770 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7771 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7772 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7774 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7777 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7779 2017-09-26 Richard Biener <rguenther@suse.de>
7781 PR tree-optimization/82320
7782 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7785 2017-09-25 Jeff Law <law@redhat.com>
7787 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7788 prototype for new argument.
7789 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7790 mostly extracted from rs6000_emit_allocate_stack.
7791 (rs6000_emit_probe_stack_range_stack_clash): New function.
7792 (rs6000_emit_allocate_stack): Call
7793 rs6000_emit_probe_stack_range_stack_clash as needed.
7794 (rs6000_emit_probe_stack_range): Add additional argument
7795 to call to gen_probe_stack_range{si,di}.
7796 (output_probe_stack_range): New.
7797 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7798 (output_probe_stack_range_stack_clash): New.
7799 (rs6000_emit_prologue): Emit notes into dump file as requested.
7800 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7801 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7802 Add additional operand and pass it to output_probe_stack_range.
7804 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7806 PR tree-optimization/82163
7807 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7808 (checking_verify_loop_closed_ssa): New parameter.
7809 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7810 (check_loop_closed_ssa_stmt): Delete.
7811 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7812 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7813 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7816 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7818 * brig-builtins.def: Treat HSAIL barrier builtins as
7819 setjmp/longjump style functions.
7821 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7823 * target.def (constant_alignment): New hook.
7824 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7825 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7826 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7827 * doc/tm.texi: Regenerate.
7828 * targhooks.h (default_constant_alignment): Declare.
7829 (constant_alignment_word_strings): Likewise.
7830 * targhooks.c (default_constant_alignment): New function.
7831 (constant_alignment_word_strings): Likewise.
7832 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7833 instead of CONSTANT_ALIGNMENT.
7834 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7835 (force_const_mem): Likewise.
7836 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7837 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7838 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7839 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7840 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7842 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7843 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7844 constant_alignment_word_strings.
7845 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7846 (CONSTANT_ALIGNMENT): Likewise.
7847 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7848 (arm_constant_alignment): New function.
7849 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7850 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7851 constant_alignment_word_strings.
7852 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7853 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7854 constant_alignment_word_strings.
7855 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7856 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7857 (cris_constant_alignment): New function.
7858 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7859 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7860 (epiphany_constant_alignment): New function.
7861 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7862 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7863 constant_alignment_word_strings.
7864 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7865 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7866 constant_alignment_word_strings.
7867 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7868 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7869 constant_alignment_word_strings.
7870 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7871 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7872 * config/i386/i386.c (ix86_constant_alignment): Make static.
7873 Use the same interface as the target hook.
7874 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7875 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7876 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7877 constant_alignment_word_strings.
7878 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7879 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7880 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7881 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7882 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7883 constant_alignment_word_strings.
7884 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7885 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7886 constant_alignment_word_strings.
7887 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7888 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7889 constant_alignment_word_strings.
7890 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7891 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7893 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7894 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7895 * config/mips/mips.c (mips_constant_alignment): New function.
7896 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7897 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7898 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7899 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7900 (mmix_constant_alignment): Make static. Use the same interface
7902 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7903 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7904 constant_alignment_word_strings.
7905 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7906 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7907 constant_alignment_word_strings.
7908 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7909 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7910 constant_alignment_word_strings.
7911 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7912 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7913 (rs6000_constant_alignment): New function.
7914 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7915 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7916 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7917 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7918 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7919 (rs6000_constant_alignment): New function.
7920 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7921 * config/s390/s390.c (s390_constant_alignment): New function.
7922 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7923 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7924 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7925 constant_alignment_word_strings.
7926 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7927 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7928 (sparc_constant_alignment): New function.
7929 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7930 * config/spu/spu.c (spu_constant_alignment): New function.
7931 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7932 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7933 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7934 constant_alignment_word_strings.
7935 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7936 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7937 constant_alignment_word_strings.
7938 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7939 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7940 constant_alignment_word_strings.
7941 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7942 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7943 (visium_constant_alignment): New function.
7944 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7945 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7946 (xtensa_constant_alignment): New function.
7947 * system.h (CONSTANT_ALIGNMENT): Poison.
7949 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7951 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7952 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7953 (rs6000_builtin_valid_without_lhs): New helper function.
7954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7955 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7957 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7959 * target.h (vec_perm_indices): Use unsigned short rather than
7961 (auto_vec_perm_indices): Likewise.
7962 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7963 Use unsigned int rather than unsigned char.
7964 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7966 2017-09-25 Richard Biener <rguenther@suse.de>
7968 * cfgloop.h (sort_sibling_loops): Declare.
7969 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7970 (sort_sibling_loops): New function sorting the sibling loop list
7972 * graphite.c (graphite_transform_loops): Sort sibling loops.
7974 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7976 * target.def (vec_perm_const_ok): Change sel parameter to
7978 * optabs-query.c (can_vec_perm_p): Update accordingly.
7979 * doc/tm.texi: Regenerate.
7980 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7981 auto_vec_perm_indices and remove separate nelt field.
7982 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7983 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7984 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7985 (aarch64_expand_vec_perm_const): Update accordingly.
7986 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7987 to vec_perm_indices.
7988 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7989 auto_vec_perm_indices and remove separate nelt field.
7990 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7991 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7992 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7994 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7995 to vec_perm_indices.
7996 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7997 sel to vec_perm_indices.
7998 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7999 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8000 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8002 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8005 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8008 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8009 on the FUNCTION_DECL function context if it has a DIE that is a
8012 2017-09-25 Richard Biener <rguenther@suse.de>
8014 PR tree-optimization/82285
8015 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8018 2017-09-25 Tom de Vries <tom@codesourcery.com>
8022 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8025 2017-09-25 Richard Biener <rguenther@suse.de>
8027 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8028 ISL errors other than isl_error_quota happen. Dump if the
8029 schedule is the same.
8030 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8031 errors instead of aborting inside ISL.
8033 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8036 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8037 of libgcc_eh for m64.
8038 * config/i386/darwin64.h: Likewise.
8040 2017-09-25 Richard Biener <rguenther@suse.de>
8043 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8044 attribute for incomplete types nor twice for complete ones.
8046 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8049 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8050 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8053 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8056 * config/i386/i386.opt (mprefer-avx256): Use
8057 ix86_target_flags variable.
8058 * config/i386/i386.c (ix86_target_string): Move
8059 -mprefer-avx256 to flag2_opts.
8061 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8064 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8065 and x != -1 | y != -1 into (x & y) != -1.
8067 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8069 * config.gcc: Add new case statement to set
8070 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8071 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8072 s390x-*-linux* case statements. Added aarch64 to the list of
8073 supported architectures.
8075 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8077 PR tree-optimization/82289
8078 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8079 STMT_VINFO_RELEVANT_P.
8081 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8082 Alan Hayward <alan.hayward@arm.com>
8083 David Sherwood <david.sherwood@arm.com>
8085 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8086 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8087 (extract_range_from_binary_expr_1): Don't call
8088 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8090 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8091 Alan Hayward <alan.hayward@arm.com>
8092 David Sherwood <david.sherwood@arm.com>
8094 * target.def (preferred_vector_alignment): New hook.
8095 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8097 * doc/tm.texi: Regenerate.
8098 * targhooks.h (default_preferred_vector_alignment): Declare.
8099 * targhooks.c (default_preferred_vector_alignment): New function.
8100 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8102 (DR_TARGET_ALIGNMENT): New macro.
8103 (aligned_access_p): Update commentary.
8104 (vect_known_alignment_in_bytes): New function.
8105 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8107 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8108 Calculate the misalignment based on the target alignment rather than
8110 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8111 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8112 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8113 the target alignment, rather than masking the element misalignment
8114 with the number of elements in a vector. Also use the target
8115 alignment when calculating the maximum number of peels.
8116 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8117 instead of TYPE_ALIGN_UNIT.
8118 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8119 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8120 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8121 (vect_create_data_ref_ptr): Likewise.
8122 (vect_setup_realignment): Realign by ANDing with
8123 -DR_TARGET_MISALIGNMENT.
8124 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8125 the number of peels based on DR_TARGET_ALIGNMENT.
8126 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8127 with the guaranteed alignment boundary when deciding whether
8129 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8130 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8131 (ensure_base_align): Remove stmt_info parameter. Get the
8132 target base alignment from DR_TARGET_ALIGNMENT.
8133 (vectorizable_store): Update call accordingly. Interpret
8134 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8136 (vectorizable_load): Likewise.
8138 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8139 Alan Hayward <alan.hayward@arm.com>
8140 David Sherwood <david.sherwood@arm.com>
8142 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8143 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8144 (vect_enhance_data_refs_alignment): Likewise.
8146 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8148 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8149 error. Only quit immediately if parsing is complete.
8150 (BEGIN): Initialize fatal_err and parse_done.
8151 (begin fpu, end fpu): Check number of arguments.
8152 (begin arch, end arch): Likewise.
8153 (begin cpu, end cpu): Likewise.
8154 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8155 (optalias): Likewise.
8157 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8159 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8160 * config/arm/arm-isa.h: Delete. Move definitions to ...
8161 * arm-cpus.in: ... here. Use new feature and fgroup values.
8162 * config/arm/arm.c (arm_option_override): Use lower case for feature
8164 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8165 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8166 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8168 (print_isa_bits_for): New function.
8169 (gen_isa): New function.
8170 (gen_comm_data): Use print_isa_bits_for.
8171 (define feature): New keyword.
8172 (define fgroup): New keyword.
8173 * config/arm/t-arm (TM_H): Remove.
8174 (GTM_H): Add arm-isa.h.
8175 (arm-isa.h): Add rule to generate file.
8176 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8177 case for feature bit names.
8179 2017-09-22 Richard Biener <rguenther@suse.de>
8181 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8183 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8184 print of no dependency loops ...
8185 * graphite.c (graphite_transform_loops): ... here.
8186 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8188 (same_close_phi_node, remove_duplicate_close_phi,
8189 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8190 (canonicalize_loop_closed_ssa): ... here and simplify.
8191 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8192 (optimize_isl): Use dump_printf_loc to tell when we stopped
8193 optimizing because of an ISL timeout.
8195 2017-09-22 Richard Biener <rguenther@suse.de>
8197 PR tree-optimization/82291
8198 * tree-if-conv.c (predicate_mem_writes): Make sure to
8199 remove writes in blocks predicated with false.
8201 2017-09-22 Richard Biener <rguenther@suse.de>
8203 * sese.c: Include cfganal.h.
8204 (if_region_set_false_region): Remove.
8205 (create_if_region_on_edge): Likewise.
8206 (move_sese_in_condition): Re-implement without destroying
8209 2017-09-22 Richard Biener <rguenther@suse.de>
8211 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8212 Verify both BBs contain loop PHI nodes before dispatching to
8214 (graphite_regenerate_ast_isl): Do not recompute dominators,
8215 do not verify three times. Restructure for clarity.
8216 * graphite-scop-detection.c (same_close_phi_node,
8217 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8218 defined_in_loop_p, canonicalize_loop_closed_ssa,
8219 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8220 checking and SSA rewrite, move to ...
8221 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8222 (graphite_initialize): Do not pass in ctx, do not reset the
8223 SCEV cache, compute only dominators.
8224 (graphite_transform_loops): Allocate ISL ctx after
8225 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8226 Maintain post-dominators only around build_scops.
8227 * sese.c (if_region_set_false_region): Make static. Free
8228 and recompute dominators.
8229 (move_sese_in_condition): Assert we don't get called with
8230 post-dominators computed.
8231 * sese.h (if_region_set_false_region): Remove.
8233 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8235 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8236 mode attribute for TARGET_AVX512VL.
8238 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8240 * config/i386/i386.opt (mprefer-avx256): New option.
8241 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8243 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8244 for TARGET_PREFER_AVX256.
8245 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8247 2017-09-21 Jeff Law <law@redhat.com>
8249 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8250 Fix dump output if the only stack space is for pushed registers.
8252 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8254 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8257 2017-09-21 Martin Sebor <msebor@redhat.com>
8260 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8261 code (in C++) or code that triggers warnings.
8263 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8265 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8267 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8269 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8270 * sched-rgn.c: Ditto.
8271 * sel-sched-ir.c: Ditto.
8273 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8275 * toplev.h (set_random_seed): Adjust return type.
8276 * toplev.c (init_local_tick): Move eager initialization of random_seed
8277 to get_random_seed. Adjust comment.
8278 (init_random_seed): Inline to get_random_seed, delete.
8279 (get_random_seed): Initialize random_seed lazily.
8280 (set_random_seed): Do not return previous value.
8281 (print_switch_value): Do not call get_random_seed.
8283 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8285 * cgraph.c (delete_function_version): New, broken out from...
8286 (cgraph_node::delete_function_version): ...here. Rename to
8287 cgraph_node::delete_function_version_by_decl. Update all uses.
8288 (cgraph_node::remove): Call delete_function_version.
8290 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8293 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8294 VAR_DECLs to which addressable non-volatile parameters are mapped
8295 and for id->retvar after the return value assignment. Clear
8296 id->retval and id->retbnd after inlining.
8298 2017-09-21 Richard Biener <rguenther@suse.de>
8300 PR tree-optimization/82276
8301 PR tree-optimization/82244
8302 * tree-vrp.c (build_assert_expr_for): Set
8303 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8305 (remove_range_assertions): Revert earlier change.
8307 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8310 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8312 2017-09-21 Richard Biener <rguenther@suse.de>
8314 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8315 Restore valid IL after code generation errors.
8316 * graphite.c (graphite_transform_loops): Diagnose code
8317 generation issues as MSG_MISSED_OPTIMIZATION and continue
8318 with processing SCOPs.
8320 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8321 Alan Hayward <alan.hayward@arm.com>
8322 David Sherwood <david.sherwood@arm.com>
8324 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8325 rather than choosing between plus_constant and gen_rtx_<CODE>.
8326 * expr.c (emit_push_insn): Likewise.
8327 (expand_expr_real_2): Likewise.
8329 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8330 Alan Hayward <alan.hayward@arm.com>
8331 David Sherwood <david.sherwood@arm.com>
8333 * loop-unroll.c (split_iv): Call copy_rtx on the step.
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.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8340 calling tree_to_uhwi.
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-ssa-ccp.c (get_value_for_expr): Use a positive test for
8347 INTEGER_CST rather than a negative test for ADDR_EXPR.
8349 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8350 Alan Hayward <alan.hayward@arm.com>
8351 David Sherwood <david.sherwood@arm.com>
8353 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8354 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8356 2017-09-21 Richard Biener <rguenther@suse.de>
8358 PR tree-optimization/71351
8359 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8360 graphite_create_new_loop_guard): Remove, fold remaining parts
8362 (translate_isl_ast_node_for): ... here and simplify.
8364 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8367 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8368 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8369 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8370 alternative always use QI mode, for -Os imov (=R,R) alternative
8371 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8374 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8375 Jeff Law <law@redhat.com>
8377 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8378 (allocate_stack_space): New function, partially extracted from
8380 (s390_emit_prologue): Track offset to most recent stack probe.
8381 Code to allocate space moved into allocate_stack_space.
8382 Dump actions when no stack is allocated.
8383 (s390_prologue_plus_offset): New function.
8384 (s390_emit_stack_probe): Likewise.
8386 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8388 * common.opt (Wa, Wl, Wp, g, gz=): Add
8390 (gno-column-info): Remove.
8391 (gcolumn-info): Drop RejectNegative.
8393 (gno-record-gcc-switches): Remove.
8394 (grecord-gcc-switches): Drop RejectNegative.
8395 (gno-split-dwarf): Remove.
8396 (gsplit-dwarf): Drop RejectNegative.
8397 (gno-strict-dwarf): Remove.
8398 (gstrict-dwarf): Drop RejectNegative.
8399 * config/darwin.opt (gfull, gused): Add RejectNegative.
8400 * dwarf2out.c (gen_producer_string): Drop
8401 gno-record-gcc-switches handler.
8402 * optc-gen.awk: Add g to prefixes with negative forms.
8403 * opts-common.c (remapping_prefix_p): New.
8404 (find_opt): Check it.
8405 (generate_canonical_option): Test g prefix.
8406 (option_map): Add -gno- mapping.
8407 (add_misspelling_candidates): Check remapping_prefix_p.
8409 2017-09-20 Jeff Law <law@redhat.com>
8411 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8412 thinko in stack clash protection support.
8414 * explow.c (compute_stack_clash_protection_loop_data): Use
8415 CONST_INT_P instead of explicit test. Verify object is a
8416 CONST_INT_P before looking at INTVAL.
8417 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8418 instead of explicit test.
8420 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8423 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8424 address instead of to r1 and r11.
8426 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8428 * config.gcc: Support "knm".
8429 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8430 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8432 * config/i386/i386.c (m_KNM): Define.
8433 (processor_target_table): Add "knm".
8435 (ix86_option_override_internal): Add "knm".
8436 (ix86_issue_rate): Add PROCESSOR_KNM.
8437 (ix86_adjust_cost): Ditto.
8438 (ia32_multipass_dfa_lookahead): Ditto.
8439 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8440 (fold_builtin_cpu): Add M_INTEL_KNM.
8441 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8442 (processor_type): Add PROCESSOR_KNM.
8443 * config/i386/x86-tune.def: Add m_KNM.
8444 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8446 2017-09-20 Richard Biener <rguenther@suse.de>
8448 PR tree-optimization/80213
8449 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8450 are allowed in empty BBs as well.
8451 (canonicalize_loop_closed_ssa): Also look for other complex
8453 (scop_detection::get_sese): Include the loop-closed PHI block
8455 (scop_detection::merge_sese): Remove code adding extra blocks.
8456 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8457 (build_scops): Assert the final returned scop is invalid.
8459 2017-09-20 Richard Biener <rguenther@suse.de>
8461 PR tree-optimization/82264
8462 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8464 (vn_phi_lookup): Likewise.
8465 (vn_phi_insert): Likewise.
8467 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8469 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8470 that fits into uhwi or shwi, add DW_AT_const_value regardless
8471 of early_dwarf without going through RTL, using add_AT_unsigned
8474 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8475 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8476 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8477 (DEBUG_MACRO_SECTION): Likewise.
8478 (DEBUG_LTO_MACRO_SECTION): Likewise.
8479 (DEBUG_STR_DWO_SECTION): Likewise.
8480 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8481 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8482 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8483 (DEBUG_LTO_LINE_STR_SECTION): Define.
8484 (init_sections_and_labels): Initialize debug_line_str_section
8485 variable. Initialize debug_loc_section for -gdwarf-5 to
8486 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8488 2017-09-20 Richard Biener <rguenther@suse.de>
8490 * graphite-sese-to-poly.c (extract_affine): Properly handle
8491 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8493 2017-09-20 Richard Biener <rguenther@suse.de>
8495 PR tree-optimization/81373
8496 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8497 Force SESE live-out defs to be handled even if they are
8500 2017-09-19 Jeff Law <law@redhat.com>
8502 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8503 nothing for stack adjustments with REG_STACK_CHECK.
8504 * sched-deps.c (parse_add_or_inc): Reject insns with
8505 REG_STACK_CHECK from dependency breaking.
8506 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8507 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8508 * reg-notes.def (STACK_CHECK): New note.
8510 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8511 (ix86_expand_prologue): Dump stack clash info as needed.
8512 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8514 * function.c (dump_stack_clash_frame_info): New function.
8515 * function.h (dump_stack_clash_frame_info): Prototype.
8516 (enum stack_clash_probes): New enum.
8518 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8519 flag_stack_clash_protection.
8520 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8521 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8522 (arm_frame_pointer_required): Likewise.
8523 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8524 (ia64_expand_prologue): Likewise.
8525 * config/mips/mips.c (mips_expand_prologue): Likewise.
8526 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8527 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8528 (sparc_flat_expand_prologue): Likewise.
8529 * config/spu/spu.c (spu_expand_prologue): Likewise.
8531 * explow.c: Include "params.h".
8532 (anti_adjust_stack_and_probe_stack_clash): New function.
8533 (get_stack_check_protect): Likewise.
8534 (compute_stack_clash_protection_loop_data): Likewise.
8535 (emit_stack_clash_protection_loop_start): Likewise.
8536 (emit_stack_clash_protection_loop_end): Likewise.
8537 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8538 Use anti_adjust_stack_and_probe_stack_clash.
8539 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8540 (emit_stack_clash_protection_loop_start): Likewise.
8541 (emit_stack_clash_protection_loop_end): Likewise.
8542 * rtl.h (get_stack_check_protect): Prototype.
8543 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8544 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8545 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8547 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8549 * doc/tm.texi: Rebuilt.
8550 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8551 get_stack_check_protect.
8552 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8553 * config/arm/arm.c (arm_expand_prologue): Likewise.
8554 (arm_frame_pointer_required): Likewise.
8555 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8556 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8557 * config/mips/mips.c (mips_expand_prologue): Likewise.
8558 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8559 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8560 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8561 (sparc_flat_expand_prologue): Likewise.
8563 * common.opt (-fstack-clash-protection): New option.
8564 * flag-types.h (enum stack_check_type): Note difference between
8565 -fstack-check= and -fstack-clash-protection.
8566 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8567 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8568 * toplev.c (process_options): Issue warnings/errors for cases
8569 not handled with -fstack-clash-protection.
8570 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8571 (-fstack-check): Note additional problem with -fstack-check=generic.
8572 Note that -fstack-check is primarily for Ada and refer users
8573 to -fstack-clash-protection for stack-clash-protection.
8574 Document new params for stack clash protection.
8576 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8578 * config/i386/i386.c (ix86_split_long_move): Do not handle
8579 address used for LEA in a special way.
8581 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8583 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8585 2017-09-19 Martin Sebor <msebor@redhat.com>
8588 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8589 of incompatible types.
8591 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8593 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8594 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8595 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8596 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8598 2017-09-19 Richard Biener <rguenther@suse.de>
8600 PR tree-optimization/82244
8601 * tree-vrp.c (remove_range_assertions): Do not propagate
8602 a constant to abnormals but replace the assert with a copy.
8604 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8606 PR rtl-optimization/57878
8607 PR rtl-optimization/68988
8608 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8609 avoidance test involving non_reload_pseudos. Move frequency test
8610 below the general fragmentation avoidance test.
8612 2017-09-19 Richard Biener <rguenther@suse.de>
8614 PR tree-optimization/69728
8615 * graphite-sese-to-poly.c (schedule_error): New global.
8616 (add_loop_schedule): Handle empty domain by failing the
8618 (build_original_schedule): Handle schedule_error.
8620 2017-09-19 Richard Biener <rguenther@suse.de>
8622 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8623 Do not iterate to sibling loops but only to siblings of inner
8626 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8629 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8630 that the registers are different.
8632 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8634 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8635 to processor_model and "amdfam17h" to arch_names_table.
8636 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8638 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8641 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8643 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8644 Alan Hayward <alan.hayward@arm.com>
8645 David Sherwood <david.sherwood@arm.com>
8647 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8649 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8651 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8652 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8653 vect_schedule_slp_instance.
8654 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8655 Update call to vect_slp_analyze_node_operations. Simplify return
8657 (vect_slp_analyze_bb_1): Update call accordingly.
8658 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8659 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8660 (vect_schedule_slp): Update call accordingly.
8662 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8663 Alan Hayward <alan.hayward@arm.com>
8664 David Sherwood <david.sherwood@arm.com>
8666 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8667 with types that aren't in fact scalar.
8669 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8671 * tree-vect-slp.c (vect_record_max_nunits): New function,
8673 (vect_build_slp_tree_1): ...here.
8674 (vect_build_slp_tree_2): Call it for phis too.
8676 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8678 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8679 to vect_get_vec_def_for_operand when getting the mask operand.
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 type of
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-loop.c (vectorizable_live_operation): Fix element size
8693 calculation for vector booleans.
8695 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8696 Alan Hayward <alan.hayward@arm.com>
8697 David Sherwood <david.sherwood@arm.com>
8699 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8701 (vect_transform_stmt): ...here.
8702 (vect_analyze_stmt): Use it instead of calling
8703 vectorizable_live_operation directly.
8705 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8707 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8708 non-SIMT targets in acc vector loops.
8710 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8712 * configure.ac: Add arc and check if assembler supports gdwarf2.
8713 * configure: Regenerate.
8715 2017-09-18 Richard Biener <rguenther@suse.de>
8717 PR tree-optimization/82220
8718 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8719 epilogue niters from the min_profitable_iters compute.
8721 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8724 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8725 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8726 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8728 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8731 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8732 switching to a new text section.
8734 2017-09-18 Richard Biener <rguenther@suse.de>
8736 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8738 (build_alias_set): Reject aliases with no access function.
8740 2017-09-18 Richard Biener <rguenther@suse.de>
8742 PR tree-optimization/79622
8743 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8745 (build_cross_bb_scalars_use): Likewise.
8747 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8749 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8751 2017-09-18 Alan Modra <amodra@gmail.com>
8754 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8755 stack_pointer_rtx for count 0. Update comments. Break up
8756 large rtl expression.
8758 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8760 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8761 Increase to 20 bytes.
8762 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8763 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8764 or avx version of the stub.
8766 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8769 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8770 compute the minimum stack alignment. Also update preferred stack
8771 boundary for leaf functions.
8773 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8775 PR tree-optimization/82228
8776 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8779 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8781 * common/config/nds32/nds32-common.c
8782 (nds32_option_optimization_table): Refine formatting.
8783 (nds32_option_optimization_table): Use -fsched-pressure and
8784 -fomit-frame-pointer for specific optimization level.
8786 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8788 * config/nds32/nds32.c: Refine formatting and comments.
8789 * config/nds32/nds32.h: Likewise.
8790 * config/nds32/nds32.md: Likewise.
8791 * config/nds32/nds32-cost.c: Likewise.
8792 * config/nds32/nds32-isr.c: Likewise.
8793 * config/nds32/nds32-md-auxiliary.c: Likewise.
8794 * config/nds32/nds32-multiple.md: Likewise.
8795 * config/nds32/nds32-predicates.c: Likewise.
8797 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8798 Jakub Jelinek <jakub@redhat.com>
8800 Add support for -std=c++2a.
8801 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8803 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8805 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8808 * doc/extend.texi (Common Function Attributes): Add
8809 references to ARM, AArch64, and S/390 specific attributes.
8810 (Function Specific Option Pragmas): Add AArch64 and S/390
8811 to list of back ends that support the target pragma.
8813 2017-09-15 Nathan Sidwell <nathan@acm.org>
8815 * doc/standards.texi: Fix C++17 description. Update URLs for
8818 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8820 * common.opt (Wcast-align=strict): New warning option.
8821 * doc/invoke.texi: Document -Wcast-align=strict.
8823 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8825 * cgraph.h (cgraph_thunk_info): Add comments.
8826 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8827 assert for VIRTUAL_* arguments stricter.
8829 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8831 PR tree-optimization/71026
8832 * match.pd: Move RDIV patterns from fold-const.c
8833 * fold-const.c (distribute_real_division): Removed.
8834 (fold_binary_loc): Remove calls to distribute_real_divison.
8836 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8838 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8839 c++1z and gnu++1z as deprecated. Change other references to
8840 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8841 Change -Wc++1z-compat to -Wc++17-compat.
8842 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8843 * dwarf2out.c (highest_c_language): Handle C++17.
8844 (gen_compile_unit_die): Likewise.
8846 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8848 PR rtl-optimization/82192
8849 * combine.c (make_extraction): Don't look through non-paradoxical
8850 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8853 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8854 Alan Hayward <alan.hayward@arm.com>
8855 David Sherwood <david.sherwood@arm.com>
8857 * target.def (function_arg_offset): New hook.
8858 * targhooks.h (default_function_arg_offset): Declare.
8859 * targhooks.c (default_function_arg_offset): New function.
8860 * function.c (locate_and_pad_parm): Use
8861 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8862 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8863 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8864 * doc/tm.texi: Regenerate.
8865 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8866 * config/spu/spu.c (spu_function_arg_offset): New function.
8867 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8868 * system.h (FUNCTION_ARG_OFFSET): Poison.
8870 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8871 Alan Hayard <alan.hayward@arm.com>
8872 David Sherwood <david.sherwood@arm.com>
8874 * target.def (truly_noop_truncation): New hook.
8875 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8876 than TRULY_NOOP_TRUNCATION.
8877 * hooks.h (hook_bool_uint_uint_true): Declare.
8878 * hooks.c (hook_bool_uint_uint_true): New function.
8879 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8880 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8881 * doc/tm.texi: Regenerate.
8882 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8883 rather than TRULY_NOOP_TRUNCATION in comments.
8884 (simplify_comparison): Likewise.
8885 (record_truncated_value): Likewise.
8886 * expmed.c (extract_bit_field_1): Likewise.
8887 (extract_split_bit_field): Likewise.
8888 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8889 instead of TRULY_NOOP_TRUNCATION.
8890 * function.c (assign_parm_setup_block): Likewise.
8891 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8892 * rtlhooks.c: Include target.h.
8893 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8894 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8895 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8896 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8897 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8898 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8899 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8900 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8901 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8902 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8903 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8904 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8905 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8906 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8907 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8908 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8909 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8910 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8911 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8912 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8913 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8914 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8915 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8916 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8917 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8918 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8919 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8920 rather than TRULY_NOOP_TRUNCATION in comments.
8921 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8922 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8923 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8924 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8925 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8926 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8927 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8928 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8929 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8930 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8931 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8932 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8933 rather than TRULY_NOOP_TRUNCATION in comments.
8934 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8935 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8936 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8937 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8938 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8939 TRULY_NOOP_TRUNCATION condition.
8940 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8941 (TRULY_NOOP_TRUNCATION): Delete.
8942 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8943 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8944 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8945 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8946 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8947 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8948 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8949 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8950 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8951 rather than TRULY_NOOP_TRUNCATION in comments.
8952 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8953 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8954 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8955 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8956 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8957 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8959 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8962 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8963 (*cmp_ior): Likewise.
8964 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8965 (*ior_scc_scc_cmp): Likewise.
8966 (*and_scc_scc): Likewise.
8967 (*and_scc_scc_cmp): Likewise.
8969 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8970 Alan Hayard <alan.hayward@arm.com>
8971 David Sherwood <david.sherwood@arm.com>
8973 * target.def (can_change_mode_class): New hook.
8974 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8975 (hard_regno_nregs): Likewise.
8976 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8977 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8978 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8979 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8980 (LOAD_EXTEND_OP): Update accordingly.
8981 * doc/tm.texi: Regenerate.
8982 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8983 CANNOT_CHANGE_MODE_CLASS.
8984 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8985 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8986 * combine.c (simplify_set): Update accordingly.
8987 * emit-rtl.c (validate_subreg): Likewise.
8988 * recog.c (general_operand): Likewise.
8989 * regcprop.c (mode_change_ok): Likewise.
8990 * reload1.c (choose_reload_regs): Likewise.
8991 (inherit_piecemeal_p): Likewise.
8992 * rtlanal.c (simplify_subreg_regno): Likewise.
8993 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8994 instead of CANNOT_CHANGE_MODE_CLASS.
8995 (reload_cse_simplify_operands): Likewise.
8996 * reload.c (push_reload): Use targetm.can_change_mode_class
8997 instead of CANNOT_CHANGE_MODE_CLASS.
8998 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8999 REG_CANNOT_CHANGE_MODE_P.
9000 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9001 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9002 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9003 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9004 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9005 (arm_can_change_mode_class): New function.
9006 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9007 than CANNOT_CHANGE_MODE_CLASS in comments.
9008 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9009 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9010 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9011 (ix86_can_change_mode_class): ...this new function, inverting the
9012 sense of the return value.
9013 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9014 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9015 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9016 (ia64_can_change_mode_class): New function.
9017 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9018 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9019 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9020 (m32c_can_change_mode_class): ...this new function, inverting the
9021 sense of the return value.
9022 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9023 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9024 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9025 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9026 (mips_can_change_mode_class): ...this new function, inverting the
9027 sense of the return value.
9028 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9029 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9030 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9031 (msp430_can_change_mode_class): New function.
9032 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9033 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9034 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9035 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9036 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9037 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9038 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9039 (pa_cannot_change_mode_class): Replace with...
9040 (pa_can_change_mode_class): ...this new function, inverting the
9041 sense of the return value.
9042 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9043 than CANNOT_CHANGE_MODE_CLASS in comments.
9044 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9045 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9046 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9047 (pdp11_cannot_change_mode_class): Replace with...
9048 (pdp11_can_change_mode_class): ...this new function, inverting the
9049 sense of the return value.
9050 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9051 * config/powerpcspe/powerpcspe-protos.h
9052 (rs6000_cannot_change_mode_class_ptr): Delete.
9053 * config/powerpcspe/powerpcspe.c
9054 (rs6000_cannot_change_mode_class_ptr): Delete.
9055 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9056 (rs6000_option_override_internal): Assign to
9057 targetm.can_change_mode_class instead of
9058 rs6000_cannot_change_mode_class_ptr.
9059 (rs6000_cannot_change_mode_class): Replace with...
9060 (rs6000_can_change_mode_class): ...this new function, inverting the
9061 sense of the return value.
9062 (rs6000_debug_cannot_change_mode_class): Replace with...
9063 (rs6000_debug_can_change_mode_class): ...this new function.
9064 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9065 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9066 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9067 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9068 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9070 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9071 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9072 (rs6000_option_override_internal): Assign to
9073 targetm.can_change_mode_class instead of
9074 rs6000_cannot_change_mode_class_ptr.
9075 (rs6000_cannot_change_mode_class): Replace with...
9076 (rs6000_can_change_mode_class): ...this new function, inverting the
9077 sense of the return value.
9078 (rs6000_debug_cannot_change_mode_class): Replace with...
9079 (rs6000_debug_can_change_mode_class): ...this new function.
9080 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9081 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9082 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9083 (s390_can_change_mode_class): ...this new function, inverting the
9084 sense of the return value.
9085 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9086 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9087 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9088 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9089 (sh_cannot_change_mode_class): Replace with...
9090 (sh_can_change_mode_class): ...this new function, inverting the
9091 sense of the return value.
9092 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9093 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9094 (sparc_can_change_mode_class): New function.
9095 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9096 * config/spu/spu.c (spu_can_change_mode_class): New function.
9097 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9098 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9099 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9100 (visium_can_change_mode_class): New function.
9101 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9103 2017-09-15 Richard Biener <rguenther@suse.de>
9105 PR tree-optimization/82217
9106 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9107 but not undefined case.
9109 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9112 * postreload.c (reload_cse_simplify_operands): Skip
9113 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9115 2017-09-15 Richard Biener <rguenther@suse.de>
9117 PR tree-optimization/68823
9118 * graphite-scop-detection.c (build_alias_set): If we have a
9119 possible dependence check whether we can handle them by just
9120 looking at the DRs DR_ACCESS_FNs.
9121 (build_scops): If build_alias_set fails, fail the SCOP.
9123 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9125 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9126 to support float128 built-in functions that require the ISA 3.0
9128 (BU_FLOAT128_3_HW): Likewise.
9129 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9131 (FMAF128): Likewise.
9133 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9134 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9135 floating point instructions.
9136 (rs6000_invalid_builtin): Likewise.
9137 (rs6000_builtin_mask_names): Likewise.
9138 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9139 (RS6000_BTM_FLOAT128_HW): Likewise.
9140 (RS6000_BTM_COMMON): Likewise.
9141 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9143 * doc/extend.texi (RS/6000 built-in functions): Document the
9144 IEEE 128-bit floating point square root and fused multiply-add
9147 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9149 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9150 reg (r2) isn't in the set of registers defined in the prologue.
9152 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9153 Alan Hayward <alan.hayward@arm.com>
9154 David Sherwood <david.sherwood@arm.com>
9156 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9157 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9158 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9159 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9160 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9162 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9163 max_vectorization_factor.
9164 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9166 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9167 Alan Hayward <alan.hayward@arm.com>
9168 David Sherwood <david.sherwood@arm.com>
9170 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9171 (vect_worthwhile_without_simd_p): Declare.
9172 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9173 (vectorizable_reduction): Use it.
9174 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9175 (vectorizable_operation): Likewise.
9177 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9178 Alan Hayward <alan.hayward@arm.com>
9179 David Sherwood <david.sherwood@arm.com>
9181 * tree-vectorizer.h (vect_get_num_copies): New function.
9182 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9183 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9184 (vectorizable_induction): Likewise.
9185 (vectorizable_live_operation): Likewise.
9186 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9187 (vectorizable_bswap): Likewise.
9188 (vectorizable_call): Likewise.
9189 (vectorizable_conversion): Likewise.
9190 (vectorizable_assignment): Likewise.
9191 (vectorizable_shift): Likewise.
9192 (vectorizable_operation): Likewise.
9193 (vectorizable_store): Likewise.
9194 (vectorizable_load): Likewise.
9195 (vectorizable_condition): Likewise.
9196 (vectorizable_comparison): Likewise.
9197 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
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 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9204 of vect_init_vector.
9206 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9207 Alan Hayward <alan.hayward@arm.com>
9208 David Sherwood <david.sherwood@arm.com>
9210 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9211 an inline wrapper that provides a location.
9212 (gimple_build_vector): Likewise.
9213 * gimple-fold.c (gimple_build_vector_from_val): New function.
9214 (gimple_build_vector): Likewise.
9215 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9216 functions to build the initial value. Always return a gimple value.
9217 (get_initial_defs_for_reduction): Likewise. Only compute
9219 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9220 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9221 (vectorizable_induction): Use gimple_build_vector rather than
9224 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9225 Alan Hayward <alan.hayward@arm.com>
9226 David Sherwood <david.sherwood@arm.com>
9228 * target.h (vec_perm_indices): New typedef.
9229 (auto_vec_perm_indices): Likewise.
9230 * optabs-query.h: Include target.h
9231 (can_vec_perm_p): Take a vec_perm_indices *.
9232 * optabs-query.c (can_vec_perm_p): Likewise.
9233 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9234 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9235 * tree-vect-generic.c (lower_vec_perm): Likewise.
9236 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9237 (vect_grouped_load_supported): Likewise.
9238 (vect_shift_permute_load_chain): Likewise.
9239 (vect_permute_store_chain): Use auto_vec_perm_indices.
9240 (vect_permute_load_chain): Likewise.
9241 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9242 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9243 Update uses of can_vec_perm_p.
9244 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9245 mode with a number of elements. Take a vec_perm_indices *.
9246 (vect_create_epilog_for_reduction): Update accordingly.
9247 Use auto_vec_perm_indices.
9248 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9249 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9250 (vect_transform_slp_perm_load): Likewise.
9251 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9252 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9253 (vect_gen_perm_mask_checked): Likewise.
9254 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9255 (vect_gen_perm_mask_checked): Likewise.
9256 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9257 (vectorizable_store): Likewise.
9258 (vectorizable_load): Likewise.
9259 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9260 (vectorizable_bswap): Likewise.
9262 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9263 Alan Hayward <alan.hayward@arm.com>
9264 David Sherwood <david.sherwood@arm.com>
9266 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9267 * tree.c (build_vector): Likewise.
9268 (build_vector_from_ctor): Update accordingly.
9269 (build_vector_from_val): Likewise.
9270 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9271 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9272 * tree-vect-generic.c (add_rshift): Likewise.
9273 (expand_vector_divmod): Likewise.
9274 (optimize_vector_constructor): Likewise.
9275 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9276 (vect_transform_slp_perm_load): Likewise.
9277 (vect_schedule_slp_instance): Likewise.
9278 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9279 (vectorizable_call): Likewise.
9280 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9281 * expmed.c (make_tree): Likewise.
9282 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9283 a vector passed to build_vector.
9284 (fold_convert_const): Likewise.
9285 (exact_inverse): Likewise.
9286 (fold_ternary_loc): Likewise.
9287 (fold_relational_const): Likewise.
9288 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9289 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9290 (const_unop): Likewise. Store the reduction accumulator in a
9291 variable rather than an array.
9292 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9293 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9294 the new vector, rather than constructing it after the input arrays.
9295 (native_interpret_vector): Use auto_vec<tree> when building
9296 a vector passed to build_vector. Add elements in order.
9297 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9298 auto_vec<tree> when building a vector passed to build_vector.
9299 (vect_create_epilog_for_reduction): Likewise.
9300 (vectorizable_induction): Likewise.
9301 (get_initial_def_for_reduction): Likewise. Fix indentation of
9303 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9305 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9306 passed to build_vector.
9308 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9309 Alan Hayward <alan.hayward@arm.com>
9310 David Sherwood <david.sherwood@arm.com>
9312 * tree-core.h (tree_base::u): Add an "nelts" field.
9313 (tree_vector): Use VECTOR_CST_NELTS as the length.
9314 * tree.c (tree_size): Likewise.
9315 (make_vector): Initialize VECTOR_CST_NELTS.
9316 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9317 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9318 TYPE_VECTOR_SUBPARTS.
9319 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9320 as the number of units, setting it from VECTOR_CST_NELTS.
9321 (const_vector_from_tree): Likewise.
9322 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9323 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9324 (fold_negate_expr_1): Likewise.
9325 (fold_convert_const): Likewise.
9326 (const_binop): Likewise. Differentiate the number of output and
9328 (const_unop): Likewise.
9329 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9330 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9331 in cases that did the opposite.
9333 2017-09-14 Richard Biener <rguenther@suse.de>
9335 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9338 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9340 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9342 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9345 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9346 if and where to split a bb, except for splitting before debug insn
9347 sequences followed by non-label real insn. Delete debug insns
9348 in between basic blocks.
9350 * combine.c (make_compound_operation_int): Formatting fixes.
9352 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9353 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9354 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9355 * config/sol2.h (LINK_EH_SPEC): Likewise.
9356 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9357 * config/s390/linux.h (LINK_SPEC): Likewise.
9358 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9359 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9360 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9361 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9362 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9363 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9364 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9365 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9367 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9369 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9371 (ENDFILE_LINUX_SPEC): Likewise.
9372 (LINK_EH_SPEC): Likewise.
9373 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9374 (LINK_OS_LINUX_SPEC32): Likewise.
9375 (LINK_OS_LINUX_SPEC64): Likewise.
9376 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9377 (LINK_OS_LINUX_SPEC): Likewise.
9379 2017-09-13 Martin Liska <mliska@suse.cz>
9382 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9383 CASE_HIGH is NULL_TREE.
9385 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9386 Alan Hayward <alan.hayward@arm.com>
9387 David Sherwood <david.sherwood@arm.com>
9389 * target.def (secondary_memory_needed): New hook.
9390 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9391 instead of SECONDARY_MEMORY_NEEDED.
9392 (secondary_memory_needed_mode): Likewise.
9393 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9394 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9395 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9396 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9397 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9398 * doc/tm.texi: Regenerate.
9399 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9400 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9401 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9402 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9403 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9404 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9405 mode argument first and change the reg_class arguments to reg_class_t.
9406 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9407 Make static. Update the call to inline_secondary_memory_needed.
9408 (ix86_register_move_cost): Update the call to
9409 inline_secondary_memory_needed.
9410 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9411 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9413 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9414 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9416 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9417 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9418 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9419 and match hook interface. Add comment from mips.h.
9420 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9421 * config/mmix/mmix.md (truncdfsf2): Refer to
9422 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9424 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9425 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9426 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9427 (pa_secondary_memory_needed): New function.
9428 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9429 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9430 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9431 (pdp11_secondary_memory_needed): Make static and match hook interface.
9432 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9433 * config/powerpcspe/powerpcspe-protos.h
9434 (rs6000_secondary_memory_needed_ptr): Delete.
9435 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9437 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9438 (rs6000_option_override_internal): Assign to
9439 targetm.secondary_memory_needed rather than
9440 rs6000_secondary_memory_needed_ptr.
9441 (rs6000_secondary_memory_needed): Match hook interface.
9442 (rs6000_debug_secondary_memory_needed): Likewise.
9443 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9444 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9445 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9446 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9447 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9448 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9450 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9451 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9452 (rs6000_option_override_internal): Assign to
9453 targetm.secondary_memory_needed rather than
9454 rs6000_secondary_memory_needed_ptr.
9455 (rs6000_secondary_memory_needed): Match hook interface.
9456 (rs6000_debug_secondary_memory_needed): Likewise.
9457 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9458 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9459 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9460 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9461 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9462 (sparc_secondary_memory_needed): New function.
9463 * lra-constraints.c (check_and_process_move): Refer to
9464 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9466 (curr_insn_transform): Likewise.
9467 (process_alt_operands): Use targetm.secondary_memory_needed
9468 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9469 (check_secondary_memory_needed_p): Likewise.
9470 (choose_split_class): Likewise.
9471 * reload.c: Unconditionally include code that was previously
9472 conditional on SECONDARY_MEMORY_NEEDED.
9473 (push_secondary_reload): Use targetm.secondary_memory_needed
9474 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9475 (push_reload): Likewise.
9476 * reload1.c: Unconditionally include code that was previously
9477 conditional on SECONDARY_MEMORY_NEEDED.
9478 (choose_reload_regs): Use targetm.secondary_memory_needed
9479 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9480 (gen_reload): Likewise.
9481 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9483 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9484 Alan Hayward <alan.hayward@arm.com>
9485 David Sherwood <david.sherwood@arm.com>
9487 * target.def (secondary_memory_needed_mode): New hook:
9488 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9489 * targhooks.h (default_secondary_memory_needed_mode): New function.
9490 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9491 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9492 * doc/tm.texi: Regenerate.
9493 * lra-constraints.c (check_and_process_move): Use
9494 targetm.secondary_memory_needed_mode instead of
9495 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9496 (curr_insn_transform): Likewise.
9497 * reload.c (get_secondary_mem): Likewise.
9498 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9499 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9501 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9502 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9503 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9504 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9505 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9507 * config/powerpcspe/powerpcspe-protos.h
9508 (rs6000_secondary_memory_needed_mode): Delete.
9509 * config/powerpcspe/powerpcspe.c
9510 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9511 (rs6000_secondary_memory_needed_mode): Make static.
9512 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9513 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9515 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9517 (rs6000_secondary_memory_needed_mode): Make static.
9518 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9519 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9520 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9521 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9522 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9524 (sparc_secondary_memory_needed_mode): New function.
9525 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9527 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9529 * config/aarch64/constraints.md (Umq): New constraint.
9530 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9532 (mov<mode>): Update condition.
9534 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9536 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9537 when bitposition is the same.
9539 2017-09-13 Richard Biener <rguenther@suse.de>
9541 * dwarf2out.c (output_die_symbol): Remove.
9542 (output_die): Do not output a DIEs symbol.
9544 2017-09-13 Richard Biener <rguenther@suse.de>
9547 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9548 default-def to avoid breaking iterator update with the weird
9549 interaction with cgraph_update_edges_for_call_stmt_node.
9551 2017-09-13 Richard Biener <rguenther@suse.de>
9553 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9554 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9555 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9556 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9558 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9560 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9561 Disable pc relative literal load irrespective of
9562 TARGET_FIX_ERR_A53_84341 for default.
9564 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9566 * config/sparc/sparc.c (output_return): Output the source location of
9567 the insn in the delay slot, if any.
9568 (output_sibcall): Likewise.
9570 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9573 * common.opt (-static-pie): New alias.
9574 (shared): Negate static-pie.
9575 (-no-pie): Update help text.
9577 (static-pie): New option.
9578 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9579 -static-pie support.
9580 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9581 (LINK_EH_SPEC): Likewise.
9582 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9583 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9584 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9585 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9586 (init_gcc_specs): Likewise.
9587 (init_spec): Likewise.
9588 (display_help): Update help message for -pie.
9589 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9592 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9594 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9595 (movdi_aarch64): Likewise.
9596 (movti_aarch64): Likewise.
9598 2017-09-12 Simon Wright <simon@pushface.org>
9601 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9602 calculation of the minor version, always output as 0.
9604 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9607 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9608 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9609 on it early, rather than manual conversion late. For
9610 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9611 instead of performing manual conversion.
9613 2017-09-12 Carl Love <cel@us.ibm.com>
9615 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9616 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9618 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9619 VMULOSW): Add definitions.
9620 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9621 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9622 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9624 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9626 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9628 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9630 (movti_aarch64): Likewise.
9631 (movdf_aarch64): Likewise.
9632 (movtf_aarch64): Likewise.
9633 (load_pairdi): Likewise.
9634 (store_pairdi): Likewise.
9635 (load_pairdf): Likewise.
9636 (store_pairdf): Likewise.
9637 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9638 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9639 (ldr_got_small_<mode>): Likewise.
9640 (ldr_got_small_28k_<mode>): Likewise.
9641 (ldr_got_tiny): Likewise.
9642 * config/aarch64/iterators.md (ldst_sz): New.
9643 (ldpstp_sz): Likewise.
9644 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9646 (thunderx_load): Split load_8 to load_16.
9647 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9649 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9650 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9651 (xgene1_store_pair): Split store_8 to store_16.
9652 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9653 (falkor_st_0_st_sd): Split store_8 to store_16.
9655 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9657 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9658 and store1/2/3/4 to store_4/8/12/16.
9659 * config/aarch64/aarch64.md: Update for rename.
9660 * config/arm/arm.md: Likewise.: Likewise.
9661 * config/arm/arm.c: Likewise.
9662 * config/arm/thumb1.md: Likewise.
9663 * config/arm/thumb2.md: Likewise.
9664 * config/arm/vfp.md: Likewise.
9665 * config/arm/arm-generic.md: Likewise.
9666 * config/arm/arm1020e.md: Likewise.
9667 * config/arm/arm1026ejs.md: Likewise.
9668 * config/arm/arm1136jfs.md: Likewise.
9669 * config/arm/arm926ejs.md: Likewise.
9670 * config/arm/cortex-a15.md: Likewise.
9671 * config/arm/cortex-a17.md: Likewise.
9672 * config/arm/cortex-a5.md: Likewise.
9673 * config/arm/cortex-a53.md: Likewise.
9674 * config/arm/cortex-a57.md: Likewise.
9675 * config/arm/cortex-a7.md: Likewise.
9676 * config/arm/cortex-a8.md: Likewise.
9677 * config/arm/cortex-a9.md: Likewise.
9678 * config/arm/cortex-m4.md: Likewise.
9679 * config/arm/cortex-m7.md: Likewise.
9680 * config/arm/cortex-r4.md: Likewise.
9681 * config/arm/exynos-m1.md: Likewise.
9682 * config/arm/fa526.md: Likewise.
9683 * config/arm/fa606te.md: Likewise.
9684 * config/arm/fa626te.md: Likewise.
9685 * config/arm/fa726te.md: Likewise.
9686 * config/arm/fmp626.md: Likewise.
9687 * config/arm/iwmmxt.md: Likewise.
9688 * config/arm/ldmstm.md: Likewise.
9689 * config/arm/marvell-pj4.md: Likewise.
9690 * config/arm/xgene1.md: Likewise.
9691 * config/aarch64/thunderx.md: Likewise.
9692 * config/aarch64/thunderx2t99.md: Likewise.
9693 * config/aarch64/falkor.md: Likewise.
9695 2017-09-12 Martin Liska <mliska@suse.cz>
9697 * attribs.c (private_lookup_attribute): New function.
9698 * attribs.h (private_lookup_attribute): Declared here.
9699 (lookup_attribute): Called from this place.
9701 2017-09-12 Richard Biener <rguenther@suse.de>
9703 PR tree-optimization/82157
9704 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9705 stmts with side-effects.
9707 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9708 Alan Hayward <alan.hayward@arm.com>
9709 David Sherwood <david.sherwood@arm.com>
9711 * target.def (hard_regno_nregs): New hook.
9712 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9713 * targhooks.h (default_hard_regno_nregs): Declare.
9714 * targhooks.c (default_hard_regno_nregs): New function.
9715 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9716 (TARGET_HARD_REGNO_NREGS): ...this hook.
9717 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9718 (CLASS_MAX_NREGS): Likewise.
9719 * doc/tm.texi: Regenerate.
9720 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9721 instead of HARD_REGNO_NREGS.
9722 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9723 HARD_REGNO_NREGS in the comment.
9724 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9725 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9726 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9727 Return an unsigned int.
9728 (TARGET_HARD_REGNO_NREGS): Redefine.
9729 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9730 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9731 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9732 (arc_hard_regno_nregs): New function.
9733 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9734 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9735 (arm_hard_regno_nregs): New function.
9736 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9737 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9738 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9739 (TARGET_HARD_REGNO_NREGS): Redefine.
9740 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9741 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9742 (HARD_REGNO_NREGS): Delete.
9743 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9744 (cr16_hard_regno_nregs): New function.
9745 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9746 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9747 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9748 (cris_hard_regno_nregs): New function.
9749 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9750 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9751 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9752 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9753 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9754 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9755 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9756 (frv_hard_regno_nregs): Make static. Take and return an
9758 (frv_class_max_nregs): Remove outdated copy of documentation.
9759 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9760 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9761 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9762 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9763 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9764 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9765 (TARGET_HARD_REGNO_NREGS): Redefine.
9766 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9767 (CLASS_MAX_NREGS): Update comment.
9768 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9769 (ia64_hard_regno_nregs): New function.
9770 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9771 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9772 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9773 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9774 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9776 (m32c_hard_regno_nregs): Likewise. Make static.
9777 (TARGET_HARD_REGNO_NREGS): Redefine.
9778 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9779 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9780 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9781 (m68k_hard_regno_nregs): New function.
9782 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9783 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9784 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9785 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9786 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9787 Take and return an unsigned int.
9788 (TARGET_HARD_REGNO_NREGS): Redefine.
9789 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9790 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9791 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9792 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9793 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9794 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9795 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9796 (msp430_hard_regno_nregs): Make static. Take and return an
9798 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9799 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9800 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9801 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9802 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9803 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9804 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9805 (TARGET_HARD_REGNO_NREGS): Redefine.
9806 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9807 (PA_HARD_REGNO_NREGS): ...this.
9808 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9809 (PA_HARD_REGNO_NREGS): ...this.
9810 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9811 (pa_hard_regno_nregs): New function.
9812 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9813 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9814 (pdp11_hard_regno_nregs): New function.
9815 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9816 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9817 (rs6000_hard_regno_nregs_hook): New function.
9818 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9819 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9820 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9821 Take and return an unsigned int. Move earlier in file.
9822 (TARGET_HARD_REGNO_NREGS): Redefine.
9823 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9824 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9825 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9826 (rl78_hard_regno_nregs): Make static. Take and return an
9828 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9829 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9830 (rs6000_hard_regno_nregs_hook): New function.
9831 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9832 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9833 (TARGET_HARD_REGNO_NREGS): Redefine.
9834 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9835 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9836 instead of HARD_REGNO_NREGS.
9837 (s390_hard_regno_nregs): New function.
9838 (s390_hard_regno_mode_ok): Add comment from s390.h.
9839 (TARGET_HARD_REGNO_NREGS): Redefine.
9840 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9841 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9842 (sh_hard_regno_nregs): New function.
9843 (sh_pass_in_reg_p): Use it.
9844 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9845 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9846 (sparc_hard_regno_nregs): New function.
9847 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9848 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9849 (spu_function_arg_advance): Use it, supplying a valid register number.
9850 (TARGET_HARD_REGNO_NREGS): Redefine.
9851 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9852 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9853 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9854 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9855 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9856 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9857 (CLASS_MAX_NREGS): Remove copy of old documentation.
9858 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9859 (visium_hard_regno_nregs): New function.
9860 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9861 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9862 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9863 xtensa_hard_regno_nregs): New function.
9864 * system.h (HARD_REGNO_NREGS): Poison.
9866 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9868 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9869 hard_regno_nregs instead of HARD_REGNO_NREGS.
9870 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9871 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9872 (c6x_expand_epilogue): Likewise.
9873 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9874 (frv_read_iacc_argument): Likewise.
9875 * config/sh/sh.c: Include regs.h.
9876 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9877 (regs_used): Likewise.
9878 (output_stack_adjust): Likewise.
9879 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9880 * expmed.c: Include regs.h.
9881 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9882 * ree.c: Include regs.h.
9883 (combine_reaching_defs): Use hard_regno_nregs instead of
9885 (add_removable_extension): Likewise.
9887 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9889 * regs.h (hard_regno_nregs): Turn into a function.
9890 (end_hard_regno): Update accordingly.
9891 * caller-save.c (setup_save_areas): Likewise.
9892 (save_call_clobbered_regs): Likewise.
9893 (replace_reg_with_saved_mem): Likewise.
9894 (insert_restore): Likewise.
9895 (insert_save): Likewise.
9896 * combine.c (can_change_dest_mode): Likewise.
9897 (move_deaths): Likewise.
9898 (distribute_notes): Likewise.
9899 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9900 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9901 (rs6000_split_multireg_move): Likewise.
9902 (rs6000_register_move_cost): Likewise.
9903 (rs6000_memory_move_cost): Likewise.
9904 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9905 (rs6000_split_multireg_move): Likewise.
9906 (rs6000_register_move_cost): Likewise.
9907 (rs6000_memory_move_cost): Likewise.
9908 * cselib.c (cselib_reset_table): Likewise.
9909 (cselib_lookup_1): Likewise.
9910 * emit-rtl.c (set_mode_and_regno): Likewise.
9911 * function.c (aggregate_value_p): Likewise.
9912 * ira-color.c (setup_profitable_hard_regs): Likewise.
9913 (check_hard_reg_p): Likewise.
9914 (calculate_saved_nregs): Likewise.
9915 (assign_hard_reg): Likewise.
9916 (improve_allocation): Likewise.
9917 (calculate_spill_cost): Likewise.
9918 * ira-emit.c (modify_move_list): Likewise.
9919 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9920 (ira_hard_reg_in_set_p): Likewise.
9921 * ira.c (setup_reg_mode_hard_regset): Likewise.
9922 (clarify_prohibited_class_mode_regs): Likewise.
9923 (check_allocation): Likewise.
9924 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9925 (lra_setup_reg_renumber): Likewise.
9926 (setup_try_hard_regno_pseudos): Likewise.
9927 (spill_for): Likewise.
9928 (assign_hard_regno): Likewise.
9929 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9930 * lra-constraints.c (in_class_p): Likewise.
9931 (lra_constraint_offset): Likewise.
9932 (simplify_operand_subreg): Likewise.
9933 (lra_constraints): Likewise.
9934 (split_reg): Likewise.
9935 (split_if_necessary): Likewise.
9936 (invariant_p): Likewise.
9937 (inherit_in_ebb): Likewise.
9938 * lra-lives.c (process_bb_lives): Likewise.
9939 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9940 (get_hard_regs): Likewise.
9941 (do_remat): Likewise.
9942 * lra-spills.c (assign_spill_hard_regs): Likewise.
9943 * mode-switching.c (create_pre_exit): Likewise.
9944 * postreload.c (reload_combine_recognize_pattern): Likewise.
9945 * recog.c (peep2_find_free_register): Likewise.
9946 * regcprop.c (kill_value_regno): Likewise.
9947 (set_value_regno): Likewise.
9948 (copy_value): Likewise.
9949 (maybe_mode_change): Likewise.
9950 (find_oldest_value_reg): Likewise.
9951 (copyprop_hardreg_forward_1): Likewise.
9952 * regrename.c (check_new_reg_p): Likewise.
9953 (regrename_do_replace): Likewise.
9954 * reload.c (push_reload): Likewise.
9955 (combine_reloads): Likewise.
9956 (find_dummy_reload): Likewise.
9957 (operands_match_p): Likewise.
9958 (find_reloads): Likewise.
9959 (find_equiv_reg): Likewise.
9960 (reload_adjust_reg_for_mode): Likewise.
9961 * reload1.c (count_pseudo): Likewise.
9962 (count_spilled_pseudo): Likewise.
9963 (find_reg): Likewise.
9964 (clear_reload_reg_in_use): Likewise.
9965 (free_for_value_p): Likewise.
9966 (allocate_reload_reg): Likewise.
9967 (choose_reload_regs): Likewise.
9968 (reload_adjust_reg_for_temp): Likewise.
9969 (emit_reload_insns): Likewise.
9970 (delete_output_reload): Likewise.
9971 * rtlanal.c (subreg_get_info): Likewise.
9972 * sched-deps.c (sched_analyze_reg): Likewise.
9973 * sel-sched.c (init_regs_for_mode): Likewise.
9974 (mark_unavailable_hard_regs): Likewise.
9975 (choose_best_reg_1): Likewise.
9976 (verify_target_availability): Likewise.
9977 * valtrack.c (dead_debug_insert_temp): Likewise.
9978 * var-tracking.c (track_loc_p): Likewise.
9979 (emit_note_insn_var_location): Likewise.
9980 * varasm.c (make_decl_rtl): Likewise.
9981 * reginfo.c (choose_hard_reg_mode): Likewise.
9982 (init_reg_modes_target): Refer directly to
9983 this_target_regs->x_hard_regno_nregs.
9985 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9987 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9988 instead of hard_regno_nregs.
9990 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9992 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9993 end_hard_regno instead of hard_regno_nregs.
9994 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9995 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9996 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9997 * ira-color.c (improve_allocation): Likewise.
9998 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9999 * lra-lives.c (mark_regno_live): Likewise.
10000 (mark_regno_dead): Likewise.
10001 * lra-remat.c (operand_to_remat): Likewise.
10002 * lra.c (collect_non_operand_hard_regs): Likewise.
10003 * postreload.c (reload_combine_note_store): Likewise.
10004 (move2add_valid_value_p): Likewise.
10005 * reload.c (regno_clobbered_p): Likewise.
10007 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10009 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10011 * config/v850/v850.c (v850_reorg): Likewise.
10012 * reload.c (refers_to_regno_for_reload_p): Likewise.
10013 (find_equiv_reg): Likewise.
10014 * reload1.c (reload_reg_reaches_end_p): Likewise.
10016 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10018 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10020 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10021 * config/arm/arm.c (output_move_neon): Likewise.
10022 (arm_attr_length_move_neon): Likewise.
10023 (neon_split_vcombine): Likewise.
10024 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10025 (c6x_mark_reg_written): Likewise.
10026 (c6x_dwarf_register_span): Likewise.
10027 * config/i386/i386.c (ix86_save_reg): Likewise.
10028 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10029 (rws_access_reg): Likewise.
10030 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10031 * mode-switching.c (create_pre_exit): Likewise.
10032 * ree.c (combine_reaching_defs): Likewise.
10033 (add_removable_extension): Likewise.
10034 * regcprop.c (find_oldest_value_reg): Likewise.
10035 (copyprop_hardreg_forward_1): Likewise.
10036 * reload.c (reload_inner_reg_of_subreg): Likewise.
10037 (push_reload): Likewise.
10038 (combine_reloads): Likewise.
10039 (find_dummy_reload): Likewise.
10040 (reload_adjust_reg_for_mode): Likewise.
10041 * reload1.c (find_reload_regs): Likewise.
10042 (forget_old_reloads_1): Likewise.
10043 (reload_reg_free_for_value_p): Likewise.
10044 (reload_adjust_reg_for_temp): Likewise.
10045 (emit_reload_insns): Likewise.
10046 (delete_output_reload): Likewise.
10047 * sel-sched.c (choose_best_reg_1): Likewise.
10048 (choose_best_pseudo_reg): Likewise.
10050 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10051 Alan Hayward <alan.hayward@arm.com>
10052 David Sherwood <david.sherwood@arm.com>
10054 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10055 * target.def (slow_unaligned_access): New hook.
10056 * targhooks.h (default_slow_unaligned_access): Declare.
10057 * targhooks.c (default_slow_unaligned_access): New function.
10058 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10059 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10060 * doc/tm.texi: Regenerate.
10061 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10062 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10063 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10065 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10066 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10068 (rs6000_slow_unaligned_access): New function.
10069 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10070 (expand_block_compare): Likewise.
10071 (expand_strn_compare): Likewise.
10072 (rs6000_rtx_costs): Likewise.
10073 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10074 (riscv_slow_unaligned_access): Likewise.
10075 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10076 (riscv_slow_unaligned_access_p): ...this and make static.
10077 (riscv_option_override): Update accordingly.
10078 (riscv_slow_unaligned_access): New function.
10079 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10080 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10081 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10082 (rs6000_slow_unaligned_access): New function.
10083 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10084 (rs6000_rtx_costs): Likewise.
10085 * config/rs6000/rs6000-string.c (expand_block_compare)
10086 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10087 of SLOW_UNALIGNED_ACCESS.
10088 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10089 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10090 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10091 of SLOW_UNALIGNED_ACCESS.
10092 * expmed.c (simple_mem_bitfield_p): Likewise.
10093 * expr.c (alignment_for_piecewise_move): Likewise.
10094 (emit_group_load_1): Likewise.
10095 (emit_group_store): Likewise.
10096 (copy_blkmode_from_reg): Likewise.
10097 (emit_push_insn): Likewise.
10098 (expand_assignment): Likewise.
10099 (store_field): Likewise.
10100 (expand_expr_real_1): Likewise.
10101 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10102 * lra-constraints.c (simplify_operand_subreg): Likewise.
10103 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10104 * gimple-ssa-store-merging.c: Likewise in block comment at start
10106 * tree-ssa-strlen.c: Include target.h.
10107 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10108 of SLOW_UNALIGNED_ACCESS.
10109 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10111 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10113 PR rtl-optimization/82185
10114 * expmed.c (emit_store_flag_int): Only test tem if it has been
10117 2017-09-12 Richard Biener <rguenther@suse.de>
10119 PR middle-end/82149
10120 * match.pd ((FTYPE) N CMP CST): Fix typo.
10122 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10124 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10126 (mips_near_type_p): Add 'short_call' attribute as a synonym
10128 * doc/extend.texi (short_call): Document new function attribute.
10130 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10133 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10134 assertion check that in the condition.
10135 (get_atomic_generic_size): Likewise. Before testing if parameter
10136 has pointer type, if it has array type, call for C++
10137 default_conversion to perform array-to-pointer conversion.
10139 2017-09-12 Richard Biener <rguenther@suse.de>
10141 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10142 for operations we cannot scalarize.
10144 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10146 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10147 vectors heap vectors. Clean up comments.
10148 Make visited_bbs a reference.
10149 (profitable_jump_thread_path): Make GC
10150 vectors heap vectors. Clean up comments.
10152 (convert_and_register_jump_thread_path): Make GC vectors heap
10154 (check_subpath_and_update_thread_path): Same. Clean up comments.
10155 Make visited_bbs a reference.
10156 (handle_phi): Abstract common code to to
10157 register_jump_thread_path_if_profitable.
10158 Rename VAR_BB to DEF_BB.
10160 Make GC vectors heap vectors.
10161 Make visited_bbs a reference.
10162 (handle_assignment): Same.
10163 (register_jump_thread_path_if_profitable): New.
10164 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10166 Make GC vectors heap vectors. Clean up comments.
10167 Make visited_bbs a reference.
10168 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10169 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10172 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10175 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10176 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10178 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10180 Revert r251800 and r251799.
10182 2017-09-11 Martin Jambor <mjambor@suse.cz>
10185 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10186 arguments in advance.
10187 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10188 use it to find predecessor edges.
10189 (naive_outof_ssa): Collect vector of predecessors.
10191 2017-09-08 Jason Merrill <jason@redhat.com>
10193 PR c++/70029 - ICE with ref-qualifier and -flto
10194 * langhooks.h (struct lang_hooks_for_types): Add
10195 copy_lang_qualifiers.
10196 * attribs.c (build_type_attribute_qual_variant): Use it.
10197 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10199 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10200 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10202 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10205 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10206 (*mulsi3_sp64): New instruction.
10207 (mulsi3): New expander.
10209 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10211 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10213 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10215 * sancov.c: Include memmodel.h.
10217 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10220 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10223 2017-09-07 Carl Love <cel@us.ibm.com>
10225 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10226 the sldi instruction.
10228 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10230 * sancov.c: Include tm_p.h.
10232 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10235 * output.h (switch_to_other_text_partition): New declaration.
10236 * varasm.c (switch_to_other_text_partition): New function.
10237 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10238 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10239 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10240 to the other text partition before emitting LCL label and switch back
10241 after emitting the word after it.
10243 2017-09-07 Richard Biener <rguenther@suse.de>
10245 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10246 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10247 critical edges here, after loop init.
10248 (pass_data_pre): Remove PROP_no_crit_edges flags.
10249 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10250 for valueization of call args to avoid leaking VN_TOP.
10251 (visit_use): Assert we do not visit default defs.
10252 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10253 Use error_mark_node to more easily detect leaking VN_TOP.
10254 All default-defs are varying, not VN_TOP. Mark them visited.
10255 (run_scc_vn): Make code match comment.
10257 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10259 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10260 OPTION_MASK_FLOAT128_KEYWORD.
10261 (POWERPC_MASKS): Likewise.
10262 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10263 support for the -mfloat128-type option, and make -mfloat128
10264 default on PowerPC Linux systems. Define or undefine
10265 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10266 Define __float128 to be __ieee128 if IEEE 128-bit support is
10267 enabled, or undefine it.
10268 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10269 Delete defining __FLOAT128_TYPE__.
10270 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10271 -mfloat128-type option and make -mfloat128 default on PowerPC
10273 (TARGET_FLOAT128_TYPE): Likewise.
10274 (-mfloat128-type): Likewise.
10275 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10276 Delete the -mfloat128-type option and make -mfloat128 default on
10277 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10278 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10279 128-bit floating point is enabled. Change tests from using
10280 -mfloat128-type to -mfloat128.
10281 (rs6000_mangle_type): Use the correct mangling for the __float128
10282 type even if normal long double is restricted to 64-bits.
10283 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10285 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10286 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10287 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10289 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10290 documentation for -mfloat128.
10292 2017-09-06 Olivier Hainque <hainque@adacore.com>
10294 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10296 2017-09-06 Wish Wu <wishwu007@gmail.com>
10297 Jakub Jelinek <jakub@redhat.com>
10299 * asan.c (initialize_sanitizer_builtins): Add
10300 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10301 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10302 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10303 BT_FN_VOID_UINT64_PTR variables.
10304 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10305 (BT_FN_VOID_UINT16_UINT16): Likewise.
10306 (BT_FN_VOID_UINT32_UINT32): Likewise.
10307 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10308 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10309 (BT_FN_VOID_UINT64_PTR): Likewise.
10310 * common.opt (flag_sanitize_coverage): New variable.
10311 (fsanitize-coverage=trace-pc): Remove.
10312 (fsanitize-coverage=): Add.
10313 * flag-types.h (enum sanitize_coverage_code): New enum.
10314 * fold-const.c (fold_range_test): Disable non-short-circuit
10315 optimization if flag_sanitize_coverage.
10316 (fold_truth_andor): Likewise.
10317 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10318 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10319 (coverage_sanitizer_opts): New array.
10320 (get_closest_sanitizer_option): Add OPTS argument, handle also
10321 OPT_fsanitize_coverage_.
10322 (parse_sanitizer_options): Adjusted to also handle
10323 OPT_fsanitize_coverage_.
10324 (common_handle_option): Add OPT_fsanitize_coverage_.
10325 * sancov.c (instrument_comparison, instrument_switch): New function.
10326 (sancov_pass): Add trace-cmp support.
10327 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10328 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10329 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10330 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10331 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10332 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10333 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10334 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10335 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10336 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10338 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10340 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10341 error. Only quit immediately if parsing is complete.
10342 (BEGIN): Initialize fatal_err and parse_done.
10343 (begin fpu, end fpu): Check number of arguments.
10344 (begin arch, end arch): Likewise.
10345 (begin cpu, end cpu): Likewise.
10346 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10347 (optalias): Likewise.
10349 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10351 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10352 * config/arm/arm-isa.h: Delete. Move definitions to ...
10353 * arm-cpus.in: ... here. Use new feature and fgroup values.
10354 * config/arm/arm.c (arm_option_override): Use lower case for feature
10356 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10357 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10358 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10360 (print_isa_bits_for): New function.
10361 (gen_isa): New function.
10362 (gen_comm_data): Use print_isa_bits_for.
10363 (define feature): New keyword.
10364 (define fgroup): New keyword.
10365 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10366 (arm-isa.h): Add rule to generate file.
10367 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10368 case for feature bit names.
10370 2017-09-06 Richard Biener <rguenther@suse.de>
10372 * tree-ssa-pre.c (NECESSARY): Remove.
10373 (create_expression_by_pieces): Do not touch pass-local flags.
10374 (insert_into_preds_of_block): Likewise.
10375 (do_pre_regular_insertion): Likewise.
10376 (eliminate_insert): Likewise.
10377 (eliminate_dom_walker::before_dom_children): Likewise.
10378 (fini_eliminate): Do not look at inserted_exprs.
10379 (mark_operand_necessary): Remove.
10380 (remove_dead_inserted_code): Replace with simple work-list
10381 algorithm based on inserted_exprs and SSA uses.
10382 (pass_pre::execute): Re-order fini_eliminate and
10383 remove_dead_inserted_code.
10385 2017-09-06 Olivier Hainque <hainque@adacore.com>
10387 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10388 for VxWorks 7. Adjust surrounding comments.
10390 2017-09-06 Richard Biener <rguenther@suse.de>
10392 * gimple-ssa-strength-reduction.c
10393 (find_candidates_dom_walker::before_dom_children): Also allow
10396 2017-09-06 Richard Biener <rguenther@suse.de>
10398 PR tree-optimization/82108
10399 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10400 for gap in the non-permutation SLP case.
10402 2017-09-06 Martin Jambor <mjambor@suse.cz>
10404 PR tree-optimization/82078
10405 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10406 add_access_to_work_queue...
10407 (build_accesses_from_assign): ...here.
10408 (propagate_all_subaccesses): Make sure racc is the group
10409 representative, if there is one.
10411 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10413 PR middle-end/82095
10414 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10417 2017-09-06 Richard Biener <rguenther@suse.de>
10419 * gimple-ssa-strength-reduction.c
10420 (find_candidates_dom_walker::before_doom_children): Use a
10421 type and not a mode check.
10423 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10426 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10428 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10430 2017-09-05 Jeff Law <law@redhat.com>
10432 PR tree-optimization/64910
10433 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10434 cases where we have 3 or more operands.
10436 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10438 PR middle-end/81768
10439 * omp-low.c (lower_omp_for): Recompute tree invariant if
10440 gimple_omp_for_initial/final is ADDR_EXPR.
10442 PR middle-end/81768
10443 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10444 into gimple val before gimplification fo the COND_EXPR.
10446 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10448 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10449 REGION_COPY argument.
10450 (thread_through_all_blocks): Remove unused argument to
10451 duplicate_thread_path.
10453 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10458 Take a scalar_mode rather than a machine_mode.
10459 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10460 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10461 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10462 (aarch64_gen_adjusted_ldpstp): Likewise.
10463 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10465 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10466 Alan Hayward <alan.hayward@arm.com>
10467 David Sherwood <david.sherwood@arm.com>
10469 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10470 Take a scalar_int_mode instead of a machine_mode.
10471 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10472 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10473 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10474 (aarch64_simd_attr_length_rglist): Delete.
10475 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10476 a scalar_int_mode instead of a machine_mode.
10477 (aarch64_add_offset): Likewise.
10478 (aarch64_internal_mov_immediate): Likewise
10479 (aarch64_add_constant_internal): Likewise.
10480 (aarch64_add_constant): Likewise.
10481 (aarch64_movw_imm): Likewise.
10482 (aarch64_rtx_arith_op_extract_p): Likewise.
10483 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10484 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10485 Remove assert that the mode isn't a vector.
10486 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10487 (aarch64_expand_mov_immediate): Update calls after above changes.
10488 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10489 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10490 (aarch64_move_imm): Likewise.
10491 (aarch64_can_const_movi_rtx_p): Likewise.
10492 (aarch64_strip_extend): Likewise.
10493 (aarch64_extr_rtx_p): Likewise.
10494 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10495 a CONST_INT when the mode parameter is VOIDmode.
10496 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10498 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10500 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10501 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10502 (bitwise_type_for_mode): Update accordingly.
10504 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10506 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10507 * stor-layout.c (mode_for_size_tree): Likewise.
10508 (mode_for_array): Update accordingly.
10509 (layout_decl): Likewise.
10510 (compute_record_mode): Likewise. Only set the mode once.
10512 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10514 * target.def (get_mask_mode): Change return type to opt_mode.
10516 * doc/tm.texi: Regenerate.
10517 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10518 * targhooks.c (default_get_mask_mode): Likewise.
10519 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10520 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10521 targetm.get_mask_mode.
10522 * tree.c (build_truth_vector_type): Likewise.
10524 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10526 * machmode.h (mode_for_vector): Return an opt_mode.
10527 * stor-layout.c (mode_for_vector): Likewise.
10528 (mode_for_int_vector): Update accordingly.
10529 (layout_type): Likewise.
10530 * config/i386/i386.c (emit_memmov): Likewise.
10531 (ix86_expand_set_or_movmem): Likewise.
10532 (ix86_expand_vector_init): Likewise.
10533 (ix86_get_mask_mode): Likewise.
10534 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10536 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10537 * expmed.c (extract_bit_field_1): Likewise.
10538 * expr.c (expand_expr_real_2): Likewise.
10539 * optabs-query.c (can_vec_perm_p): Likewise.
10540 (can_vec_mask_load_store_p): Likewise.
10541 * optabs.c (expand_vec_perm): Likewise.
10542 * targhooks.c (default_get_mask_mode): Likewise.
10543 * tree-vect-stmts.c (vectorizable_store): Likewise.
10544 (vectorizable_load): Likewise.
10545 (get_vectype_for_scalar_type_and_size): Likewise.
10547 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10549 * machmode.h (mode_for_int_vector): New function.
10550 * stor-layout.c (mode_for_int_vector): Likewise.
10551 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10552 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10553 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10554 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10555 (s390_expand_vcond): Likewise.
10557 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10559 * machmode.h (opt_machine_mode): New type.
10560 (opt_mode<T>): Allow construction from anything that can be
10562 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10563 (mode_for_size): Return an opt_machine_mode.
10564 * stor-layout.c (mode_for_size): Likewise.
10565 (mode_for_size_tree): Update call accordingly.
10566 (bitwise_mode_for_mode): Likewise.
10567 (make_fract_type): Likewise.
10568 (make_accum_type): Likewise.
10569 * caller-save.c (replace_reg_with_saved_mem): Update call
10571 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10572 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10573 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10574 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10575 * expmed.c (extract_bit_field_1): Likewise.
10576 * reload.c (get_secondary_mem): Likewise.
10577 * varasm.c (assemble_integer): Likewise.
10578 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10581 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10583 * machmode.h (decimal_float_mode_for_size): New function.
10584 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10585 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10586 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10587 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10588 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10589 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10591 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10593 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10594 (get_builtin_sync_mode): Likewise.
10595 (expand_ifn_atomic_compare_exchange): Likewise.
10596 (expand_builtin_atomic_clear): Likewise.
10597 (expand_builtin_atomic_test_and_set): Likewise.
10598 (fold_builtin_atomic_always_lock_free): Likewise.
10599 * calls.c (compute_argument_addresses): Likewise.
10600 (emit_library_call_value_1): Likewise.
10601 (store_one_arg): Likewise.
10602 * combine.c (combine_instructions): Likewise.
10603 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10604 * config/arm/arm.c (arm_function_value): Likewise.
10605 (aapcs_allocate_return_reg): Likewise.
10606 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10607 * config/i386/i386.c (construct_container): Likewise.
10608 (ix86_gimplify_va_arg): Likewise.
10609 (ix86_expand_sse_cmp): Likewise.
10610 (emit_memmov): Likewise.
10611 (emit_memset): Likewise.
10612 (expand_small_movmem_or_setmem): Likewise.
10613 (ix86_expand_pextr): Likewise.
10614 (ix86_expand_pinsr): Likewise.
10615 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10616 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10618 * config/mips/mips.c (mips_function_value_1) Likewise.
10619 (mips_block_move_straight): Likewise.
10620 (mips_expand_ins_as_unaligned_store): Likewise.
10621 * config/powerpcspe/powerpcspe.c
10622 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10623 (rs6000_darwin64_record_arg_flush): Likewise.
10624 * config/rs6000/rs6000.c
10625 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10626 (rs6000_darwin64_record_arg_flush): Likewise.
10627 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10628 (sparc_function_value_1): Likewise.
10629 * config/spu/spu.c (adjust_operand): Likewise.
10630 (spu_emit_branch_or_set): Likewise.
10631 (arith_immediate_p): Likewise.
10632 * emit-rtl.c (gen_lowpart_common): Likewise.
10633 * expr.c (expand_expr_real_1): Likewise.
10634 * function.c (assign_parm_setup_block): Likewise.
10635 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10636 * reload1.c (alter_reg): Likewise.
10637 * stor-layout.c (mode_for_vector): Likewise.
10638 (layout_type): Likewise.
10640 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10642 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10643 (spu_convert_move): Likewise.
10644 * lower-subreg.c (resolve_simple_move): Likewise.
10646 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10649 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10650 define_insn to a define_expand.
10651 (altivec_vsum2sws_direct): New define_insn.
10652 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10654 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10656 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10659 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10664 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10665 realign stack if stack alignment needed is less than incoming
10668 2017-09-05 Marek Polacek <polacek@redhat.com>
10671 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10674 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10676 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10678 2017-09-05 Richard Biener <rguenther@suse.de>
10680 PR tree-optimization/82084
10681 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10683 2017-09-05 Richard Biener <rguenther@suse.de>
10685 PR tree-optimization/82102
10686 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10688 2017-09-05 Martin Liska <mliska@suse.cz>
10690 PR tree-optimization/82032
10691 * tree-cfg.c (generate_range_test): New function.
10692 * tree-cfg.h (generate_range_test): Declared here.
10693 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10694 (cleanup_control_expr_graph): Use it.
10695 * tree-switch-conversion.c (try_switch_expansion): Remove
10697 (emit_case_nodes): Use generate_range_test.
10699 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10702 * config/i386/i386.md (*<btsc><mode>_mask): Add
10703 TARGET_USE_BT to insn constraint.
10704 (*btr<mode>_mask): Ditto.
10706 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10708 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10709 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10711 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10714 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10715 TARGET_NEON and TARGET_IWMMXT.
10716 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10717 TARGET_NEON and TARGET_IWMMXT.
10718 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10720 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10722 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10723 (ix86_rewrite_tls_address): Ditto.
10724 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10725 (ix86_rewrite_tls_address_1): Ditto.
10726 (ix86_rewrite_tls_address): Ditto.
10727 * config/i386/predicates.md (tls_address_pattern): New predicate.
10728 * config/i386/i386.md (TLS address splitter): New splitter.
10730 2017-09-04 Richard Biener <rguenther@suse.de>
10732 PR tree-optimization/82084
10733 * fold-const.h (can_native_encode_string_p): Declare.
10734 * fold-const.c (can_native_encode_string_p): Factor out from ...
10735 (native_encode_string): ... here.
10736 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10737 vectorizing stores from constants we later cannot handle.
10739 2017-09-04 Marek Polacek <polacek@redhat.com>
10742 * doc/invoke.texi: Update -Wtautological-compare documentation.
10744 2017-09-04 Jeff Law <law@redhat.com>
10746 PR tree-optimization/64910
10747 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10748 swap the first and last operand if the last is a constant.
10750 2017-09-04 Marek Polacek <polacek@redhat.com>
10753 * convert.c (do_narrow): When sanitizing signed integer overflows,
10754 bail out for signed types.
10755 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10757 2017-09-04 Richard Biener <rguenther@suse.de>
10759 PR tree-optimization/82060
10760 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10761 Move devirtualization after stmt folding and before EH/AB/noreturn
10762 cleanup to get the stmt refs canonicalized. Use a bool instead
10763 of gimple_modified_p since that doesn't work for NOPs. Schedule
10764 NOPs generated by folding for removal.
10766 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10767 Alan Hayward <alan.hayward@arm.com>
10768 David Sherwood <david.sherwood@arm.com>
10770 * coretypes.h (pad_direction): New enum.
10771 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10772 (FUNCTION_ARG_PADDING): Likewise.
10773 * target.def (function_arg_padding): New hook.
10774 * targhooks.h (default_function_arg_padding): Declare.
10775 * targhooks.c (default_function_arg_padding): New function.
10776 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10777 (TARGET_FUNCTION_ARG_PADDING): ...this.
10778 * doc/tm.texi: Regenerate.
10779 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10780 instead of direction.
10781 (compute_argument_addresses): Likewise.
10782 (load_register_parameters): Likewise.
10783 (emit_library_call_value_1): Likewise.
10784 (store_one_arg): Use targetm.calls.function_arg_padding instead
10785 of FUNCTION_ARG_PADDING.
10786 (must_pass_in_stack_var_size_or_pad): Likewise.
10787 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10788 (emit_group_store): Likewise.
10789 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10790 instead of FUNCTION_ARG_PADDING.
10791 (emit_push_insn): Likewise, and propagate enum change throughout
10793 * function.h (direction): Delete.
10794 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10796 * function.c (assign_parm_find_stack_rtl): Likewise.
10797 (assign_parm_setup_block_p): Likewise.
10798 (assign_parm_setup_block): Likewise.
10799 (gimplify_parameters): Likewise.
10800 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10801 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10803 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10804 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10805 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10806 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10807 (aarch64_function_arg_padding): ...this new function.
10808 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10809 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10810 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10811 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10812 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10813 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10814 (arm_pad_arg_upward): Replace with...
10815 (arm_function_arg_padding): ...this new function.
10816 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10818 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10819 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10820 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10821 (ia64_hpux_function_arg_padding): Replace with...
10822 (ia64_function_arg_padding): ...this new function. Use pad_direction
10823 instead of direction. Check for TARGET_HPUX.
10824 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10825 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10826 (iq2000_function_arg_padding): New function.
10827 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10828 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10829 (mips_function_arg_padding): ...this new function.
10830 (mips_pad_reg_upward): Update accordingly.
10831 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10832 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10833 targetm.calls.function_arg_padding.
10834 (FUNCTION_ARG_PADDING): Delete.
10835 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10836 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10837 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10838 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10839 (nios2_block_reg_padding): Return pad_direction instead of direction.
10840 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10841 instead of direction.
10842 (nios2_function_arg_padding): Likewise. Make static.
10843 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10844 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10845 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10846 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10847 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10848 (pa_function_arg_padding): Make static. Return pad_direction instead
10850 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10851 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10852 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10853 instead of direction. Use targetm.calls.function_arg_padding.
10854 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10855 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10856 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10857 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10858 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10860 (function_arg_padding): Rename to...
10861 (rs6000_function_arg_padding): ...this. Make static. Return
10862 pad_direction instead of direction.
10863 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10864 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10865 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10866 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10867 instead of direction. Use targetm.calls.function_arg_padding.
10868 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10869 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10870 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10871 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10872 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10873 (function_arg_padding): Rename to...
10874 (rs6000_function_arg_padding): ...this. Make static. Return
10875 pad_direction instead of direction.
10876 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10877 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10878 * config/s390/s390.c (s390_function_arg_padding): New function.
10879 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10880 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10881 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10882 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10883 (function_arg_padding): Rename to...
10884 (sparc_function_arg_padding): ...this. Make static. Return
10885 pad_direction instead of direction.
10886 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10887 * config/spu/spu.c (spu_function_arg_padding): New function.
10888 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10889 * system.h (FUNCTION_ARG_PADDING): Poison.
10891 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10892 Alan Hayward <alan.hayward@arm.com>
10893 David Sherwood <david.sherwood@arm.com>
10895 * target.def (modes_tieable_p): New hook.
10896 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10897 (TARGET_MODES_TIEABLE_P): ...this.
10898 * doc/tm.texi.in: Regenerate.
10899 * hooks.h (hook_bool_mode_mode_true): Declare.
10900 * hooks.c (hook_bool_mode_mode_true): New function.
10901 * combine.c (subst): Use targetm.modes_tieable_p instead of
10903 * dse.c (find_shift_sequence): Likewise.
10904 * expmed.c (extract_low_bits): Likewise.
10905 * lower-subreg.c: Include target.h.
10906 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10908 * rtlanal.c (rtx_cost): Likewise.
10909 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10910 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10911 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10912 (TARGET_MODES_TIEABLE_P): Redefine.
10913 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10914 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10915 (TARGET_MODES_TIEABLE_P): Redefine.
10916 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10917 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10918 (arc_modes_tieable_p): New function.
10919 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10920 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10921 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10922 (arm_modes_tieable_p): Make static.
10923 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10924 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10925 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10926 (TARGET_MODES_TIEABLE_P): Redefine.
10927 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10928 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10929 (TARGET_MODES_TIEABLE_P): Redefine.
10930 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10931 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10932 (cr16_modes_tieable_p): New function.
10933 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10934 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10935 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10936 (TRULY_NOOP_TRUNCATION): Update comment.
10937 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10938 (TRULY_NOOP_TRUNCATION): Update comment.
10939 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10940 (frv_modes_tieable_p): New function.
10941 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10942 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10943 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10944 (TARGET_MODES_TIEABLE_P): Redefine.
10945 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10946 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10947 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10948 (TARGET_MODES_TIEABLE_P): Redefine.
10949 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10950 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10951 (ia64_modes_tieable_p): New function.
10952 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10953 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10954 (iq2000_modes_tieable_p): New function.
10955 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10956 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10957 (lm32_modes_tieable_p): New function.
10958 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10959 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10960 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10961 (TARGET_MODES_TIEABLE_P): Redefine.
10962 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10963 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10964 (m32r_modes_tieable_p): New function.
10965 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10966 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10967 (m68k_modes_tieable_p): New function.
10968 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10969 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10970 (mcore_modes_tieable_p): New function.
10971 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10972 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10974 (TARGET_MODES_TIEABLE_P): Redefine.
10975 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10976 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10977 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10978 (TARGET_MODES_TIEABLE_P): Redefine.
10979 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10980 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10981 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10982 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10983 (mn10300_modes_tieable_p): ...this and make static.
10984 (TARGET_MODES_TIEABLE_P): Redefine.
10985 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10986 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10987 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10988 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10989 (msp430_modes_tieable_p): Make static.
10990 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10991 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10992 (TARGET_MODES_TIEABLE_P): Redefine.
10993 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10994 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10995 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10996 (TARGET_MODES_TIEABLE_P): Redefine.
10997 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10998 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10999 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11000 (TARGET_MODES_TIEABLE_P): Redefine.
11001 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11002 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11003 (pdp11_modes_tieable_p): New function.
11004 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11005 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11006 (rs6000_modes_tieable_p): New function.
11007 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11008 * config/powerpcspe/powerpcspe.md: Update comment.
11009 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11010 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11011 (TARGET_MODES_TIEABLE_P): Redefine.
11012 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11013 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11014 (rl78_modes_tieable_p): New function.
11015 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11016 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11017 (rs6000_modes_tieable_p): New function.
11018 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11019 * config/rs6000/rs6000.md: Update comment.
11020 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11021 * config/rx/rx.c (rx_modes_tieable_p): New function.
11022 (TARGET_MODES_TIEABLE_P): Redefine.
11023 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11024 * config/s390/s390.c (s390_modes_tieable_p): New function.
11025 (TARGET_MODES_TIEABLE_P): Redefine.
11026 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11027 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11028 (sh_modes_tieable_p): New function.
11029 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11030 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11031 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11032 (sparc_modes_tieable_p): Make static.
11033 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11034 * config/spu/spu.c (spu_modes_tieable_p): New function.
11035 (TARGET_MODES_TIEABLE_P): Redefine.
11036 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11037 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11038 (TARGET_MODES_TIEABLE_P): Redefine.
11039 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11040 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11041 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11042 * config/v850/v850.c (v850_modes_tieable_p): New function.
11043 (TARGET_MODES_TIEABLE_P): Redefine.
11044 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11045 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11046 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11047 (visium_modes_tieable_p): New function.
11048 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11049 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11050 (xtensa_modes_tieable_p): New function.
11051 * system.h (MODES_TIEABLE_P): Poison.
11053 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11054 Alan Hayward <alan.hayward@arm.com>
11055 David Sherwood <david.sherwood@arm.com>
11057 * target.def (hard_regno_mode_ok): New hook.
11058 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11059 (TARGET_HARD_REGNO_MODE_OK): ...this.
11060 * doc/tm.texi.in: Regenerate.
11061 * hooks.h (hook_bool_uint_mode_true): Declare.
11062 * hooks.c (hook_bool_uint_mode_true): New function.
11063 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11064 HARD_REGNO_MODE_OK.
11065 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11066 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11067 instead of HARD_REGNO_MODE_OK.
11068 * caller-save.c: Include target.h.
11069 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11070 HARD_REGNO_MODE_OK.
11071 * combine.c (can_combine_p): Likewise.
11072 (combinable_i3pat): Likewise.
11073 (can_change_dest_mode): Likewise.
11074 * expr.c (init_expr_target): Likewise.
11075 (convert_move): Likewise.
11076 (convert_modes): Likewise.
11077 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11078 (setup_prohibited_mode_move_regs): Likewise.
11079 * ira.h (target_ira): Likewise.
11080 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11081 * lra-constraints.c (process_alt_operands): Likewise.
11082 (split_reg): Likewise.
11083 * recog.c (peep2_find_free_register): Likewise.
11084 * ree.c (combine_reaching_defs): Likewise.
11085 * regcprop.c (maybe_mode_change): Likewise.
11086 * reginfo.c (init_reg_sets_1): Likewise.
11087 (choose_hard_reg_mode): Likewise.
11088 (simplifiable_subregs): Likewise.
11089 * regrename.c (check_new_reg_p): Likewise.
11090 * reload.c (find_valid_class): Likewise.
11091 (find_valid_class_1): Likewise.
11092 (reload_inner_reg_of_subreg): Likewise.
11093 (push_reload): Likewise.
11094 (combine_reloads): Likewise.
11095 (find_dummy_reload): Likewise.
11096 (find_reloads): Likewise.
11097 * reload1.c (find_reg): Likewise.
11098 (set_reload_reg): Likewise.
11099 (allocate_reload_reg): Likewise.
11100 (choose_reload_regs): Likewise.
11101 (reload_adjust_reg_for_temp): Likewise.
11102 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11103 (simplify_subreg_regno): Likewise.
11104 * sel-sched.c (init_regs_for_mode): Likewise.
11105 * varasm.c (make_decl_rtl): Likewise.
11106 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11107 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11108 HARD_REGNO_MODE_OK.
11109 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11110 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11111 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11112 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11113 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11114 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11115 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11116 (arc_mode_class): Delete.
11117 (HARD_REGNO_MODE_OK): Delete.
11118 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11119 (arc_hard_regno_mode_ok): Rename old array to...
11120 (arc_hard_regno_mode_ok_modes): ...this.
11121 (arc_conditional_register_usage): Update accordingly.
11122 (arc_mode_class): Make static.
11123 (arc_hard_regno_mode_ok): New function.
11124 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11125 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11126 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11127 (arm_hard_regno_mode_ok): Make static.
11128 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11129 HARD_REGNO_MODE_OK.
11130 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11131 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11132 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11135 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11136 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11137 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11138 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11140 * config/bfin/predicates.md (valid_reg_operand): Use
11141 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11142 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11143 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11144 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11145 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11146 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11147 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11148 (cr16_hard_regno_mode_ok): Make static and return a bool.
11149 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11150 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11151 (cris_hard_regno_mode_ok): New function.
11152 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11153 (epiphany_mode_class): Delete.
11154 (HARD_REGNO_MODE_OK): Delete.
11155 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11156 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11157 (hard_regno_mode_ok): Rename to...
11158 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11159 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11160 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11161 HARD_REGNO_MODE_OK.
11162 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11163 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11164 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11165 (frv_hard_regno_mode_ok): Make static and return a bool.
11166 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11167 HARD_REGNO_MODE_OK.
11168 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11169 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11170 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11171 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11174 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11175 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11176 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11178 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11179 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11180 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11181 (ia64_hard_regno_mode_ok): New function.
11182 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11183 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11184 (iq2000_hard_regno_mode_ok): New function.
11185 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11186 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11187 (lm32_hard_regno_mode_ok): New function.
11188 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11189 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11190 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11191 instead of HARD_REGNO_MODE_OK.
11192 (m32c_hard_regno_ok): Rename to...
11193 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11194 (m32c_cannot_change_mode_class): Update accordingly.
11195 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11196 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11197 (m32r_mode_class): Delete.
11198 (HARD_REGNO_MODE_OK): Delete.
11199 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11200 (m32r_hard_regno_mode_ok): Rename to...
11201 (m32r_hard_regno_modes): ...this.
11202 (m32r_mode_class): Make static.
11203 (m32r_hard_regno_mode_ok): New function.
11204 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11205 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11206 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11207 (m68k_hard_regno_mode_ok): Make static.
11208 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11209 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11210 (mcore_hard_regno_mode_ok): New function.
11211 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11212 (HARD_REGNO_MODE_OK): Delete.
11213 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11215 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11216 (microblaze_hard_regno_mode_ok): New function.
11217 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11218 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11219 (mips_hard_regno_mode_ok): Delete.
11220 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11221 (mips_hard_regno_mode_ok_p): ...this and make static.
11222 (mips_hard_regno_mode_ok_p): Rename to...
11223 (mips_hard_regno_mode_ok_uncached): ...this.
11224 (mips_hard_regno_mode_ok): New function.
11225 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11226 of HARD_REGNO_MODE_OK.
11227 (mips_option_override): Update after above name changes.
11228 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11229 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11230 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11231 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11232 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11233 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11234 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11235 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11236 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11237 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11238 (msp430_hard_regno_mode_ok): Make static and return a bool.
11239 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11240 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11241 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11243 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11244 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11245 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11246 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11247 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11248 (PA_HARD_REGNO_MODE_OK): ...this
11249 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11250 (PA_HARD_REGNO_MODE_OK): ...this.
11251 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11252 (pa_hard_regno_mode_ok): New function.
11253 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11254 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11255 (pdp11_hard_regno_mode_ok): New function.
11256 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11257 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11259 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11261 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11262 (rs6000_hard_regno_mode_ok): Rename to...
11263 (rs6000_hard_regno_mode_ok_uncached): ...this.
11264 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11265 (rs6000_hard_regno_mode_ok): New function.
11266 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11267 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11268 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11269 (riscv_hard_regno_mode_ok): ...this and make static.
11270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11271 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11272 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11273 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11274 (rl78_hard_regno_mode_ok): Make static and return bool.
11275 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11276 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11278 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11279 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11280 (rs6000_hard_regno_mode_ok): Rename to...
11281 (rs6000_hard_regno_mode_ok_uncached): ...this.
11282 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11283 (rs6000_hard_regno_mode_ok): New function.
11284 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11285 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11286 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11287 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11288 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11289 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11290 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11291 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11292 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11293 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11294 (sh_hard_regno_mode_ok): Make static.
11295 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11296 instead of HARD_REGNO_MODE_OK.
11297 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11298 (sparc_mode_class): Delete.
11299 (HARD_REGNO_MODE_OK): Delete.
11300 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11301 (hard_regno_mode_classes): Make static.
11302 (sparc_mode_class): Likewise.
11303 (sparc_hard_regno_mode_ok): New function.
11304 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11305 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11306 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11308 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11309 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11310 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11311 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11312 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11313 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11314 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11315 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11316 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11317 (visium_hard_regno_mode_ok): New function.
11318 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11319 instead of HARD_REGNO_MODE_OK.
11320 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11321 (HARD_REGNO_MODE_OK): Delete.
11322 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11323 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11324 (xtensa_option_override): Update accordingly.
11325 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11326 (xtensa_hard_regno_mode_ok): New function.
11327 * system.h (HARD_REGNO_MODE_OK): Poison.
11329 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11330 Alan Hayward <alan.hayward@arm.com>
11331 David Sherwood <david.sherwood@arm.com>
11333 * target.def (hard_regno_call_part_clobbered): New hook.
11334 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11335 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11336 * doc/tm.texi: Regenerate.
11337 * hooks.h (hook_bool_uint_mode_false): Declare.
11338 * hooks.c (hook_bool_uint_mode_false): New function.
11339 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11340 * cselib.c (cselib_process_insn): Use
11341 targetm.hard_regno_call_part_clobbered instead of
11342 HARD_REGNO_CALL_PART_CLOBBERED.
11343 * ira-conflicts.c (ira_build_conflicts): Likewise.
11344 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11345 * lra-constraints.c (need_for_call_save_p): Likewise.
11346 * lra-lives.c: Include target.h.
11347 (check_pseudos_live_through_calls): Use
11348 targetm.hard_regno_call_part_clobbered instead of
11349 HARD_REGNO_CALL_PART_CLOBBERED.
11350 * regcprop.c: Include target.h.
11351 (copyprop_hardreg_forward_1): Use
11352 targetm.hard_regno_call_part_clobbered instead of
11353 HARD_REGNO_CALL_PART_CLOBBERED.
11354 * reginfo.c (choose_hard_reg_mode): Likewise.
11355 * regrename.c (check_new_reg_p): Likewise.
11356 * reload.c (find_equiv_reg): Likewise.
11357 * reload1.c (emit_reload_insns): Likewise.
11358 * sched-deps.c (deps_analyze_insn): Likewise.
11359 * sel-sched.c (init_regs_for_mode): Likewise.
11360 (mark_unavailable_hard_regs): Likewise.
11361 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11362 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11363 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11365 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11366 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11367 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11369 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11371 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11372 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11373 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11375 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11376 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11377 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11379 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11380 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11382 * config/powerpcspe/powerpcspe.c
11383 (rs6000_hard_regno_call_part_clobbered): New function.
11384 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11385 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11386 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11388 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11389 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11390 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11392 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11393 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11394 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11396 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11397 Alan Hayward <alan.hayward@arm.com>
11398 David Sherwood <david.sherwood@arm.com>
11400 * rtl.h (subreg_memory_offset): Declare.
11401 * emit-rtl.c (subreg_memory_offset): New function.
11402 * expmed.c (store_bit_field_1): Use it.
11403 * expr.c (undefined_operand_subword_p): Likewise.
11404 * simplify-rtx.c (simplify_subreg): Likewise.
11406 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11408 PR rtl-optimization/57448
11411 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11412 using atomic_load pattern.
11413 (expand_atomic_store): Likewise.
11415 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11418 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11419 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11420 handling. Use replace_call_with_value with NULL instead of
11421 gsi_replace, unlink_stmt_vdef and release_defs.
11423 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11426 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11428 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11431 * rtl.h (emit_library_call_value_1): Declare.
11432 (emit_library_call): Replace declaration with a series of overloads.
11433 Remove the parameter count argument.
11434 (emit_library_call_value): Likewise.
11435 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11436 with an "rtx_mode_t *".
11437 (emit_library_call_value): Delete.
11438 (emit_library_call): Likewise.
11439 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11440 (asan_emit_allocas_unpoison): Likewise.
11441 * builtins.c (expand_builtin_powi): Likewise.
11442 (expand_asan_emit_allocas_unpoison): Likewise.
11443 * cfgexpand.c (expand_main_function): Likewise.
11444 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11445 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11446 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11447 * config/arm/arm.c (arm_trampoline_init): Likewise.
11448 (arm_call_tls_get_addr): Likewise.
11449 (arm_expand_divmod_libfunc): Likewise.
11450 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11451 (smulsi3_highpart): Likewise.
11452 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11453 (c6x_expand_compare): Likewise.
11454 (c6x_expand_movmem): Likewise.
11455 * config/frv/frv.c (frv_trampoline_init): Likewise.
11456 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11457 (ix86_expand_divmod_libfunc): Likewise.
11458 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11459 (ia64_expand_compare): Likewise.
11460 (ia64_profile_hook): Likewise.
11461 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11462 (nonlocal_goto): Likewise.
11463 (restore_stack_nonlocal): Likewise.
11464 * config/m32r/m32r.c (block_move_call): Likewise.
11465 (m32r_trampoline_init): Likewise.
11466 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11467 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11468 (m68k_call_m68k_read_tp): Likewise.
11469 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11470 (microblaze_expand_divide): Likewise.
11471 * config/mips/mips.h (mips_args): Likewise.
11472 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11473 (MIPS_ICACHE_SYNC): Likewise.
11474 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11475 (nios2_trampoline_init): Likewise.
11476 * config/pa/pa.c (hppa_tls_call): Likewise.
11477 (pa_trampoline_init): Likewise.
11478 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11479 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11480 (expand_strn_compare): Likewise.
11481 (rs6000_generate_compare): Likewise.
11482 (rs6000_expand_float128_convert): Likewise.
11483 (output_profile_hook): Likewise.
11484 (rs6000_trampoline_init): Likewise.
11485 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11486 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11487 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11488 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11489 (rs6000_generate_compare): Likewise.
11490 (rs6000_expand_float128_convert): Likewise.
11491 (output_profile_hook): Likewise.
11492 (rs6000_trampoline_init): Likewise.
11493 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11494 * config/sh/sh.c (sh_trampoline_init): Likewise.
11495 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11496 (sparc_emit_float_lib_cmp): Likewise.
11497 (sparc32_initialize_trampoline): Likewise.
11498 (sparc64_initialize_trampoline): Likewise.
11499 (sparc_profile_hook): Likewise.
11500 * config/spu/spu.c (ea_load_store): Likewise.
11501 * config/spu/spu.md (floatunssidf2): Likewise.
11502 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11503 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11504 * config/visium/visium.c (expand_block_move_4): Likewise.
11505 (expand_block_move_2): Likewise.
11506 (expand_block_move_1): Likewise.
11507 (expand_block_set_4): Likewise.
11508 (expand_block_set_2): Likewise.
11509 (expand_block_set_1): Likewise.
11510 (visium_trampoline_init): Likewise.
11511 (visium_profile_hook): Likewise.
11512 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11513 (xtensa_setup_frame_addresses): Likewise.
11514 (xtensa_trampoline_init): Likewise.
11515 * except.c (sjlj_emit_function_enter): Likewise.
11516 (sjlj_emit_function_exit): Likewise.
11517 * explow.c (allocate_dynamic_stack_space): Likewise.
11518 (probe_stack_range): Likewise.
11519 * expr.c (convert_mode_scalar): Likewise.
11520 * optabs.c (expand_binop): Likewise.
11521 (expand_twoval_binop_libfunc): Likewise.
11522 (expand_unop): Likewise.
11523 (prepare_cmp_insn): Likewise.
11524 (prepare_float_lib_cmp): Likewise.
11525 (expand_float): Likewise.
11526 (expand_fix): Likewise.
11527 (expand_fixed_convert): Likewise.
11528 (maybe_emit_sync_lock_test_and_set): Likewise.
11529 (expand_atomic_compare_and_swap): Likewise.
11530 (expand_mem_thread_fence): Likewise.
11531 (expand_atomic_fetch_op): Likewise.
11533 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11535 * doc/generic.texi (OpenACC): Adjust URL.
11536 * doc/invoke.texi (C Dialect Options): Ditto.
11538 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11540 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11541 predicate for operand 1. Add (m,<S>) constraint.
11542 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11543 Prevent memory operand 1 with register operand 2.
11545 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11547 PR rtl-optimization/82024
11548 * combine.c (try_combine): If the combination result is a PARALLEL,
11549 and we only need to retain the SET in there that would be placed
11550 at I2, check that we can place that at I3 instead, before doing so.
11552 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11555 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11557 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11558 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11561 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11562 Jeff Law <law@redhat.com>
11564 * varasm.c (bss_initializer_p): Do not put constants into .bss
11565 (categorize_decl_for_section): Handle bss_initializer_p returning
11566 false when DECL_INITIAL is NULL.
11568 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11571 * config/s390/s390.c (s390_can_inline_p): New function.
11573 2017-09-01 Jeff Law <law@redhat.com>
11575 PR tree-optimization/82052
11576 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11577 Always initialize the returned slot after a hash table miss
11578 when INSERT is true.
11580 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11582 * config/s390/s390.md (mem_signal_fence): Remove.
11583 * doc/md.texi (mem_signal_fence): Remove.
11584 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11586 * target-insns.def (mem_signal_fence): Remove.
11588 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11591 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11594 * asan.c (create_odr_indicator): Strip name encoding from assembler
11595 name before appending it after __odr_asan_.
11597 2017-09-01 Martin Liska <mliska@suse.cz>
11599 PR tree-optimization/82059
11600 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11601 frequency only when an edge is redirected.
11603 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11605 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11606 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11607 (arc_conditional_register_usage): Remove ARC600 lp_count
11609 (arc_file_start): Emit Tag_ARC_CPU_variation.
11610 (arc_can_use_doloop_p): New conditions to use ZOLs.
11611 (hwloop_fail): New function.
11612 (hwloop_optimize): Likewise.
11613 (hwloop_pattern_reg): Likewise.
11614 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11615 (arc_reorg_loops): New function, calls reorg_loops.
11616 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11617 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11619 (arc_loop_hazard): Remove function, functionality moved into
11621 (arc_hazard): Remove arc_loop_hazard call.
11622 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11623 into hwloop_optimize.
11624 (arc_label_align): Remove ZOL handling.
11625 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11626 * config/arc/arc.md (doloop_begin): Remove pattern.
11627 (doloop_begin_i): Likewise.
11628 (doloop_end_i): Likewise.
11629 (doloop_fallback): Likewise.
11630 (doloop_fallback_m): Likewise.
11631 (doloop_end): Reimplement expand.
11632 (arc_lp): New pattern for LP instruction.
11633 (loop_end): New pattern.
11634 (loop_fail): Likewise.
11635 (decrement_and_branch_until_zero): Likewise.
11636 * config/arc/arc.opt (mlpc-width): New option.
11637 * doc/invoke.texi (mlpc-width): Document option.
11639 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11641 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11642 (arc_ccfsm_advance): Fix checking for delay slots.
11643 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11645 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11647 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11649 (movhi_insn): Update store instruction constraint which are saving
11650 6-bit short immediates.
11651 (movsi_insn): Consider also short scaled load operations.
11652 (zero_extendhisi2_i): Use Usd constraint instead of T.
11653 (extendhisi2_i): Add q constraint.
11654 (arc_clzsi2): Add type and length attributes.
11655 (arc_ctzsi2): Likewise.
11656 * config/arc/constraints.md (Usc): Update constraint, the
11657 assembler can parse two relocations for a single instruction.
11659 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11661 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11662 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11664 2017-08-31 Olivier Hainque <hainque@adacore.com>
11666 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11667 match as powerpc-wrs-vxworks*.
11669 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11671 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11672 register constraint for by-element operand.
11673 (aarch64_mls_elt_merge<mode>): Likewise.
11675 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11677 * config/arc/arc.c (arc_can_follow_jump): Check for short
11680 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11682 * config.gcc: Use g.opt for arc.
11683 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11684 functionality moved to ...
11685 (legitimate_scaled_address_p): New function, ...here.
11686 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11687 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11688 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11690 (arc_override_options): Handle G option.
11691 (arc_output_pic_addr_const): Correct function definition.
11692 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11693 (arc_decl_anon_ns_mem_p): Delete.
11694 (arc_in_small_data_p): Overhaul this function to take into
11695 consideration the value given via G option.
11696 (arc_rewrite_small_data_1): Renamed and corrected old
11697 arc_rewrite_small_data function.
11698 (arc_rewrite_small_data): New function.
11699 (small_data_pattern): Don't use pic_offset_table_rtx.
11700 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11701 * config/arc/simdext.md (movmisalignv2hi): Use
11702 prepare_move_operands function.
11704 (movmisalign*): Likewise.
11705 * doc/invoke.texi (ARC options): Document -G option.
11707 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11709 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11711 * config/arc/arc.c (arc_print_operand): Output scalled address for
11712 sdata whenever is possible.
11713 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11714 load/stores are available.
11715 (compact_sda_memory_operand): Check for the alignment required by
11716 code density instructions.
11717 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11719 * config/arc/constraints.md (Usd): Update constraint.
11720 (Us0): New constraint.
11721 (Usc): Update constraint.
11723 2017-08-31 Richard Biener <rguenther@suse.de>
11725 PR middle-end/82054
11726 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11727 function only once.
11729 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11731 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11732 Resize type_signature.
11734 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11735 Alan Hayward <alan.hayward@arm.com>
11736 David Sherwood <david.sherwood@arm.com>
11738 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11739 subregs whose inner modes can be stored in GPRs.
11740 (aarch64_classify_index): Likewise.
11742 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11743 Alan Hayward <alan.hayward@arm.com>
11744 David Sherwood <david.sherwood@arm.com>
11746 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11747 (V_INT_EQUIV): ...this.
11748 (v_cmp_result): Rename to...
11749 (v_int_equiv): ...this.
11750 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11751 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11752 (copysign<mode>3): Likewise.
11753 (aarch64_simd_bsl<mode>_internal): Likewise.
11754 (aarch64_simd_bsl<mode>): Likewise.
11755 (vec_cmp<mode><mode>): Likewise.
11756 (vcond<mode><mode>): Likewise.
11757 (vcond<v_cmp_mixed><mode>): Likewise.
11758 (vcondu<mode><v_cmp_mixed>): Likewise.
11759 (aarch64_cm<optab><mode>): Likewise.
11760 (aarch64_cmtst<mode>): Likewise.
11761 (aarch64_fac<optab><mode>): Likewise.
11762 (vec_perm_const<mode>): Likewise.
11763 (vcond_mask_<mode><v_cmp_result>): Rename to...
11764 (vcond_mask_<mode><v_int_equiv>): ...this.
11765 (vec_cmp<mode><v_cmp_result>): Rename to...
11766 (vec_cmp<mode><v_int_equiv>): ...this.
11768 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11769 Alan Hayward <alan.hayward@arm.com>
11770 David Sherwood <david.sherwood@arm.com>
11772 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11774 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11775 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11776 (UNSPEC_LD4_DREG): New unspecs.
11777 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11778 (aarch64_ld2<mode>_dreg_be): Replace with...
11779 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11781 (aarch64_ld3<mode>_dreg_le)
11782 (aarch64_ld3<mode>_dreg_be): Replace with...
11783 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11785 (aarch64_ld4<mode>_dreg_le)
11786 (aarch64_ld4<mode>_dreg_be): Replace with...
11787 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11790 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11792 PR tree-optimization/81987
11793 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11794 insert an initializer in a location not dominated by the stride
11797 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11799 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11800 on the entire header of the finally block in the fallthru case.
11802 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11804 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11806 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11808 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11809 rs6000_emit_move_from_cr and call renamed function.
11810 (rs6000_emit_prologue): Call renamed functions.
11811 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11812 movesi_from_cr, remove volatile CRs.
11814 2017-08-30 Jon Beniston <jon@beniston.com>
11815 Richard Biener <rguenther@suse.de>
11817 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11818 of VECTOR_MODE_P check.
11819 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11820 element vector types.
11822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11824 * df.h (df_read_modify_subreg_p): Remove in favor of...
11825 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11826 const_rtx instead of an rtx.
11827 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11828 * df-problems.c (df_word_lr_mark_ref): Likewise.
11829 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11830 (mark_pseudo_reg_dead): Likewise.
11831 (mark_ref_dead): Likewise.
11832 * reginfo.c (init_subregs_of_mode): Likewise.
11833 * sched-deps.c (sched_analyze_1): Likewise.
11834 * df-scan.c (df_def_record_1): Likewise.
11835 (df_uses_record): Likewise.
11836 (df_read_modify_subreg_p): Remove in favor of...
11837 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11838 const_rtx instead of an rtx.
11840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11841 Alan Hayward <alan.hayward@arm.com>
11842 David Sherwood <david.sherwood@arm.com>
11844 * rtl.h (partial_subreg_p): New function.
11845 * caller-save.c (save_call_clobbered_regs): Use it.
11846 * calls.c (expand_call): Likewise.
11847 * combine.c (combinable_i3pat): Likewise.
11848 (simplify_set): Likewise.
11849 (make_extraction): Likewise.
11850 (make_compound_operation_int): Likewise.
11851 (gen_lowpart_or_truncate): Likewise.
11852 (force_to_mode): Likewise.
11853 (make_field_assignment): Likewise.
11854 (reg_truncated_to_mode): Likewise.
11855 (record_truncated_value): Likewise.
11856 (move_deaths): Likewise.
11857 * cse.c (record_jump_cond): Likewise.
11858 (cse_insn): Likewise.
11859 * cselib.c (cselib_lookup_1): Likewise.
11860 * expmed.c (extract_bit_field_using_extv): Likewise.
11861 * function.c (assign_parm_setup_reg): Likewise.
11862 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11863 * ira-build.c (create_insn_allocnos): Likewise.
11864 * lra-coalesce.c (merge_pseudos): Likewise.
11865 * lra-constraints.c (match_reload): Likewise.
11866 (simplify_operand_subreg): Likewise.
11867 (curr_insn_transform): Likewise.
11868 * lra-lives.c (process_bb_lives): Likewise.
11869 * lra.c (new_insn_reg): Likewise.
11870 (lra_substitute_pseudo): Likewise.
11871 * regcprop.c (mode_change_ok): Likewise.
11872 (maybe_mode_change): Likewise.
11873 (copyprop_hardreg_forward_1): Likewise.
11874 * reload.c (push_reload): Likewise.
11875 (find_reloads): Likewise.
11876 (find_reloads_subreg_address): Likewise.
11877 * reload1.c (alter_reg): Likewise.
11878 (eliminate_regs_1): Likewise.
11879 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11881 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11883 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11884 back to if statements, including unpack.
11886 2017-08-30 Martin Liska <mliska@suse.cz>
11888 PR inline-asm/82001
11889 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11891 (func_checker::compare_asm_inputs_outputs): ... this function.
11892 (func_checker::compare_gimple_asm): Use the function to compare
11893 also ASM constrains.
11894 * ipa-icf-gimple.h: Rename the function.
11896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11897 Alan Hayward <alan.hayward@arm.com>
11898 David Sherwood <david.sherwood@arm.com>
11900 * coretypes.h (complex_mode): New type.
11901 * gdbhooks.py (build_pretty_printer): Handle it.
11902 * machmode.h (complex_mode): New class.
11903 (complex_mode::includes_p): New function.
11904 (is_complex_int_mode): Likewise.
11905 (is_complex_float_mode): Likewise.
11906 * genmodes.c (get_mode_class): Handle complex mode classes.
11907 * function.c (expand_function_end): Use is_complex_int_mode.
11909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11910 Alan Hayward <alan.hayward@arm.com>
11911 David Sherwood <david.sherwood@arm.com>
11913 * coretypes.h (scalar_mode_pod): New typedef.
11914 * gdbhooks.py (build_pretty_printer): Handle it.
11915 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11916 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11917 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11918 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11919 as_a <scalar_mode>.
11921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11922 Alan Hayward <alan.hayward@arm.com>
11923 David Sherwood <david.sherwood@arm.com>
11925 * machmode.h (mode_for_vector): Take a scalar_mode instead
11927 * stor-layout.c (mode_for_vector): Likewise.
11928 * explow.c (promote_mode): Use as_a <scalar_mode>.
11929 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11932 Alan Hayward <alan.hayward@arm.com>
11933 David Sherwood <david.sherwood@arm.com>
11935 * target.def (preferred_simd_mode): Take a scalar_mode
11936 instead of a machine_mode.
11937 * targhooks.h (default_preferred_simd_mode): Likewise.
11938 * targhooks.c (default_preferred_simd_mode): Likewise.
11939 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11940 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11941 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11942 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11943 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11944 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11945 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11946 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11948 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11949 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11950 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11951 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11952 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11953 * doc/tm.texi: Regenerate.
11954 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11958 Alan Hayward <alan.hayward@arm.com>
11959 David Sherwood <david.sherwood@arm.com>
11961 * target.def (scalar_mode_supported_p): Take a scalar_mode
11962 instead of a machine_mode.
11963 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11964 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11965 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11966 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11967 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11968 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11969 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11970 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11971 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11972 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11973 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11974 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11975 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11976 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11978 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11979 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11980 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11981 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11982 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11984 * doc/tm.texi: Regenerate.
11986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11987 Alan Hayward <alan.hayward@arm.com>
11988 David Sherwood <david.sherwood@arm.com>
11990 * coretypes.h (opt_scalar_mode): New typedef.
11991 * gdbhooks.py (build_pretty_printers): Handle it.
11992 * machmode.h (mode_iterator::get_2xwider): Add overload for
11994 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11996 * expr.c (convert_mode_scalar): Likewise.
11997 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11998 * optabs.c (expand_float): Likewise.
11999 (expand_fix): Likewise.
12000 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12003 Alan Hayward <alan.hayward@arm.com>
12004 David Sherwood <david.sherwood@arm.com>
12006 * optabs.c (expand_float): Explicitly check for scalars before
12007 using a branching expansion.
12008 (expand_fix): Likewise.
12010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12011 Alan Hayward <alan.hayward@arm.com>
12012 David Sherwood <david.sherwood@arm.com>
12014 * expr.c (convert_mode): Split scalar handling out into...
12015 (convert_mode_scalar): ...this new function. Treat the modes
12018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12019 Alan Hayward <alan.hayward@arm.com>
12020 David Sherwood <david.sherwood@arm.com>
12022 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12024 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12027 Alan Hayward <alan.hayward@arm.com>
12028 David Sherwood <david.sherwood@arm.com>
12030 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12031 rather than a machine_mode.
12032 (fixed_from_string): Likewise.
12033 (fixed_convert): Likewise.
12034 (fixed_convert_from_int): Likewise.
12035 (fixed_convert_from_real): Likewise.
12036 (real_convert_from_fixed): Likewise.
12037 * fixed-value.c (fixed_from_double_int): Likewise.
12038 (fixed_from_string): Likewise.
12039 (fixed_convert): Likewise.
12040 (fixed_convert_from_int): Likewise.
12041 (fixed_convert_from_real): Likewise.
12042 (real_convert_from_fixed): Likewise.
12043 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12046 Alan Hayward <alan.hayward@arm.com>
12047 David Sherwood <david.sherwood@arm.com>
12049 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12050 of separate mode class checks. Do not allow vector modes here.
12051 (immed_wide_int_const): Use as_a <scalar_mode>.
12052 * explow.c (trunc_int_for_mode): Likewise.
12053 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12054 (wi::shwi): Likewise.
12055 (wi::min_value): Likewise.
12056 (wi::max_value): Likewise.
12057 * dwarf2out.c (loc_descriptor): Likewise.
12058 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12059 for CONST_WIDE_INT.
12061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12062 Alan Hayward <alan.hayward@arm.com>
12063 David Sherwood <david.sherwood@arm.com>
12065 * tree.h (SCALAR_TYPE_MODE): New macro.
12066 * expr.c (expand_expr_addr_expr_1): Use it.
12067 (expand_expr_real_2): Likewise.
12068 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12069 (fold_convert_const_fixed_from_int): Likewise.
12070 (fold_convert_const_fixed_from_real): Likewise.
12071 (native_encode_fixed): Likewise
12072 (native_encode_complex): Likewise
12073 (native_encode_vector): Likewise.
12074 (native_interpret_fixed): Likewise.
12075 (native_interpret_real): Likewise.
12076 (native_interpret_complex): Likewise.
12077 (native_interpret_vector): Likewise.
12078 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12079 (simd_clone_adjust_argument_types): Likewise.
12080 (simd_clone_init_simd_arrays): Likewise.
12081 (simd_clone_adjust): Likewise.
12082 * stor-layout.c (layout_type): Likewise.
12083 * tree.c (build_minus_one_cst): Likewise.
12084 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12085 * tree-inline.c (estimate_move_cost): Likewise.
12086 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12087 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12088 (vectorizable_reduction): Likewise.
12089 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12090 (vect_recog_mixed_size_cond_pattern): Likewise.
12091 (check_bool_pattern): Likewise.
12092 (adjust_bool_pattern): Likewise.
12093 (search_type_for_mask_1): Likewise.
12094 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12095 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12096 (vectorizable_load): Likewise.
12097 (vectorizable_store): Likewise.
12098 * ubsan.c (ubsan_encode_value): Likewise.
12099 * varasm.c (output_constant): Likewise.
12101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12102 Alan Hayward <alan.hayward@arm.com>
12103 David Sherwood <david.sherwood@arm.com>
12105 * coretypes.h (scalar_mode): New class.
12106 * machmode.h (scalar_mode): Likewise.
12107 (scalar_mode::includes_p): New function.
12108 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12109 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12110 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12111 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12112 * expmed.c (store_bit_field_1): Likewise.
12113 (extract_bit_field_1): Likewise.
12114 * expr.c (write_complex_part): Likewise.
12115 (read_complex_part): Likewise.
12116 (emit_move_complex_push): Likewise.
12117 (expand_expr_real_2): Likewise.
12118 * function.c (assign_parm_setup_reg): Likewise.
12119 (assign_parms_unsplit_complex): Likewise.
12120 * optabs.c (expand_binop): Likewise.
12121 * rtlanal.c (subreg_get_info): Likewise.
12122 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12123 * varasm.c (output_constant_pool_2): Likewise.
12125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12126 Alan Hayward <alan.hayward@arm.com>
12127 David Sherwood <david.sherwood@arm.com>
12129 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12131 (expmed_mult_highpart_optab): Likewise.
12132 (expmed_mult_highpart): Likewise.
12134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12135 Alan Hayward <alan.hayward@arm.com>
12136 David Sherwood <david.sherwood@arm.com>
12138 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12139 instead of a machine_mode.
12140 (builtin_memset_read_str): Likewise.
12141 * builtins.c (c_readstr): Likewise.
12142 (builtin_memcpy_read_str): Likewise.
12143 (builtin_strncpy_read_str): Likewise.
12144 (builtin_memset_read_str): Likewise.
12145 (builtin_memset_gen_str): Likewise.
12146 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12147 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12148 instead of a machine_mode.
12149 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12151 (make_extraction): Likewise.
12152 (try_widen_shift_mode): Take and return scalar_int_modes instead
12154 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12155 a scalar_int_mode instead of a machine_mode.
12156 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12157 (avr_addr_space_pointer_mode): Likewise.
12158 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12159 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12160 (msp430_unwind_word_mode): Likewise.
12161 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12162 (spu_addr_space_pointer_mode): Likewise.
12163 (spu_addr_space_address_mode): Likewise.
12164 (spu_libgcc_cmp_return_mode): Likewise.
12165 (spu_libgcc_shift_count_mode): Likewise.
12166 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12167 (rl78_addr_space_pointer_mode): Likewise.
12168 (fl78_unwind_word_mode): Likewise.
12169 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12171 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12172 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12173 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12174 (mips_valid_pointer_mode): Likewise.
12175 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12176 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12177 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12179 (ft32_addr_space_address_mode): Likewise.
12180 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12181 scalar_int_mode instead of a machine_mode.
12182 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12184 (m32c_addr_space_address_mode): Likewise.
12185 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12186 (rs6000_eh_return_filter_mode): Likewise.
12187 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12188 (rs6000_eh_return_filter_mode): Likewise.
12189 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12190 (s390_libgcc_shift_count_mode): Likewise.
12191 (s390_unwind_word_mode): Likewise.
12192 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12194 * target.def (mode_rep_extended): Likewise.
12195 (valid_pointer_mode): Likewise.
12196 (addr_space.valid_pointer_mode): Likewise.
12197 (eh_return_filter_mode): Return a scalar_int_mode rather than
12199 (libgcc_cmp_return_mode): Likewise.
12200 (libgcc_shift_count_mode): Likewise.
12201 (unwind_word_mode): Likewise.
12202 (addr_space.pointer_mode): Likewise.
12203 (addr_space.address_mode): Likewise.
12204 * doc/tm.texi: Regenerate.
12205 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12207 (do_jump): Use scalar_int_mode for local variables.
12208 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12209 rather than a machine_mode.
12210 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12211 (scompare_loc_descriptor_wide): Likewise.
12212 (scompare_loc_descriptor_narrow): Likewise.
12213 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12215 * except.c (sjlj_emit_dispatch_table): Likewise.
12216 (expand_builtin_eh_copy_values): Likewise.
12217 * explow.c (convert_memory_address_addr_space_1): Likewise.
12218 Take a scalar_int_mode rather than a machine_mode.
12219 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12220 than a machine_mode.
12221 (memory_address_addr_space): Use scalar_int_mode for local variables.
12222 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12223 rather than a machine_mode.
12224 * expmed.c (mask_rtx): Likewise.
12225 (init_expmed_one_conv): Likewise.
12226 (expand_mult_highpart_adjust): Likewise.
12227 (extract_high_half): Likewise.
12228 (expmed_mult_highpart_optab): Likewise.
12229 (expmed_mult_highpart): Likewise.
12230 (expand_smod_pow2): Likewise.
12231 (expand_sdiv_pow2): Likewise.
12232 (emit_store_flag_int): Likewise.
12233 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12235 (extract_low_bits): Likewise.
12236 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12238 * expr.c (pieces_addr::adjust): Likewise.
12239 (can_store_by_pieces): Likewise.
12240 (store_by_pieces): Likewise.
12241 (clear_by_pieces_1): Likewise.
12242 (expand_expr_addr_expr_1): Likewise.
12243 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12244 (expand_expr_real_1): Likewise.
12245 (try_casesi): Likewise.
12246 * final.c (shorten_branches): Likewise.
12247 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12248 type of "mode" to machine_mode.
12249 * internal-fn.c (expand_arith_overflow_result_store): Take a
12250 scalar_int_mode rather than a machine_mode.
12251 (expand_mul_overflow): Use scalar_int_mode for local variables.
12252 * loop-doloop.c (doloop_modify): Likewise.
12253 (doloop_optimize): Likewise.
12254 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12255 than a machine_mode.
12256 (expand_doubleword_shift_condmove): Likewise.
12257 (expand_doubleword_shift): Likewise.
12258 (expand_doubleword_clz): Likewise.
12259 (expand_doubleword_popcount): Likewise.
12260 (expand_doubleword_parity): Likewise.
12261 (expand_absneg_bit): Use scalar_int_mode for local variables.
12262 (prepare_float_lib_cmp): Likewise.
12263 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12264 rather than a machine_mode.
12265 (convert_memory_address_addr_space): Likewise.
12266 (get_mode_bounds): Likewise.
12267 (get_address_mode): Return a scalar_int_mode rather than a
12269 * rtlanal.c (get_address_mode): Likewise.
12270 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12271 than a machine_mode.
12272 * targhooks.c (default_mode_rep_extended): Likewise.
12273 (default_valid_pointer_mode): Likewise.
12274 (default_addr_space_valid_pointer_mode): Likewise.
12275 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12276 than a machine_mode.
12277 (default_libgcc_cmp_return_mode): Likewise.
12278 (default_libgcc_shift_count_mode): Likewise.
12279 (default_unwind_word_mode): Likewise.
12280 (default_addr_space_pointer_mode): Likewise.
12281 (default_addr_space_address_mode): Likewise.
12282 * targhooks.h (default_eh_return_filter_mode): Likewise.
12283 (default_libgcc_cmp_return_mode): Likewise.
12284 (default_libgcc_shift_count_mode): Likewise.
12285 (default_unwind_word_mode): Likewise.
12286 (default_addr_space_pointer_mode): Likewise.
12287 (default_addr_space_address_mode): Likewise.
12288 (default_mode_rep_extended): Take a scalar_int_mode rather than
12290 (default_valid_pointer_mode): Likewise.
12291 (default_addr_space_valid_pointer_mode): Likewise.
12292 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12294 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12295 rather than a machine_mode.
12296 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12297 for local variables.
12298 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12299 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12300 than a machine_mode.
12302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12303 Alan Hayward <alan.hayward@arm.com>
12304 David Sherwood <david.sherwood@arm.com>
12306 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12307 the mode argument to scalar_int_mode.
12308 (do_jump_by_parts_zero_rtx): Likewise.
12309 (do_jump_by_parts_equality_rtx): Likewise.
12310 (do_jump_by_parts_greater): Take a mode argument.
12311 (do_jump_by_parts_equality): Likewise.
12312 (do_jump_1): Update calls accordingly.
12314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12315 Alan Hayward <alan.hayward@arm.com>
12316 David Sherwood <david.sherwood@arm.com>
12318 * is-a.h (safe_dyn_cast): New function.
12319 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12320 (jump_table_for_label): Likewise.
12321 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12322 instead of an rtx_insn *.
12323 (shorten_branches): Use dyn_cast instead of LABEL_P and
12324 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12325 rtx_jump_table_data::get_data_mode.
12326 (final_scan_insn): Likewise.
12328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12329 Alan Hayward <alan.hayward@arm.com>
12330 David Sherwood <david.sherwood@arm.com>
12332 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12333 trying to combine a full-register integer set with a subreg
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_addr_expr): Add a new_tmode local variable
12341 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12348 when the two are known to be equal.
12350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12351 Alan Hayward <alan.hayward@arm.com>
12352 David Sherwood <david.sherwood@arm.com>
12354 * simplify-rtx.c (simplify_const_unary_operation): Use
12355 is_a <scalar_int_mode> instead of checking for a nonzero
12356 precision. Forcibly convert op_mode to a scalar_int_mode
12357 in that case. More clearly differentiate the operand and
12358 result modes and use the former when deciding what the value
12359 of a count-bits operation should be. Use is_int_mode instead
12360 of checking for a MODE_INT. Remove redundant check for whether
12361 this mode has a zero precision.
12363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12364 Alan Hayward <alan.hayward@arm.com>
12365 David Sherwood <david.sherwood@arm.com>
12367 * optabs.c (widen_leading): Change the type of the mode argument
12368 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12369 (widen_bswap): Likewise.
12370 (expand_parity): Likewise.
12371 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12372 (expand_ffs): Likewise.
12373 (epand_unop): Check for scalar integer modes before calling the
12376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12377 Alan Hayward <alan.hayward@arm.com>
12378 David Sherwood <david.sherwood@arm.com>
12380 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12382 (expand_expr_real_1): Update call accordingly.
12384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12385 Alan Hayward <alan.hayward@arm.com>
12386 David Sherwood <david.sherwood@arm.com>
12388 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12389 value_mode arguments. Use scalar_int_mode internally.
12390 (store_bit_field_1): Rename the new integer mode from imode
12391 to op0_mode and use it instead of GET_MODE (op0). Update calls
12392 to store_split_bit_field, store_bit_field_using_insv and
12393 store_fixed_bit_field.
12394 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12395 Use scalar_int_mode internally. Use a bit count rather than a mode
12396 when calculating the largest bit size for get_best_mode.
12397 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12398 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12399 Remove assertion that OP0 has a scalar integer mode.
12400 (store_split_bit_field): Add op0_mode and value_mode arguments.
12401 Update calls to extract_fixed_bit_field.
12402 (extract_bit_field_using_extv): Add an op0_mode argument.
12403 Use scalar_int_mode internally.
12404 (extract_bit_field_1): Rename the new integer mode from imode to
12405 op0_mode and use it instead of GET_MODE (op0). Update calls to
12406 extract_split_bit_field, extract_bit_field_using_extv and
12407 extract_fixed_bit_field.
12408 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12409 to extract_split_bit_field and extract_fixed_bit_field_1.
12410 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12411 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12412 on the target mode.
12413 (extract_split_bit_field): Add an op0_mode argument. Update call
12414 to extract_fixed_bit_field.
12416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12417 Alan Hayward <alan.hayward@arm.com>
12418 David Sherwood <david.sherwood@arm.com>
12420 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12421 * explow.c (hard_function_value): Likewise.
12422 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12423 convert_to_mode call outside the loop.
12424 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12425 for the mode iterator. Require the mode specified by max_pieces
12427 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12429 (copy_blkmode_to_reg): Likewise.
12430 (set_storage_via_setmem): Likewise.
12431 * optabs.c (prepare_cmp_insn): Likewise.
12432 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12433 * stor-layout.c (finish_bitfield_representative): Likewise.
12435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12436 Alan Hayward <alan.hayward@arm.com>
12437 David Sherwood <david.sherwood@arm.com>
12439 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12440 * expr.c (convert_move): Use them.
12441 (convert_modes): Likewise.
12442 (store_expr_with_bounds): Likewise.
12444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12445 Alan Hayward <alan.hayward@arm.com>
12446 David Sherwood <david.sherwood@arm.com>
12448 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12449 parameter for the mode of "x". Remove the "known_x", "known_mode"
12450 and "known_ret" arguments. Change the type of the mode argument
12451 to scalar_int_mode.
12452 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12453 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12454 (reg_num_sign_bit_copies_for_combine): Likewise.
12455 * rtlanal.c (nonzero_bits1): Likewise.
12456 (num_sign_bit_copies1): Likewise.
12457 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12458 (reg_num_sign_bit_copies_general): Likewise.
12459 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12460 (reg_nonzero_bits_general): Likewise.
12462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12463 Alan Hayward <alan.hayward@arm.com>
12464 David Sherwood <david.sherwood@arm.com>
12466 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12467 than in subroutines. Return 1 for non-integer modes.
12468 (cached_num_sign_bit_copies): Change the type of the mode parameter
12469 to scalar_int_mode.
12470 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12471 classes. Handle CONST_INT_P first and then check whether X also
12472 has a scalar integer mode. Check the same thing for inner registers
12473 of a SUBREG and for values that are being extended or truncated.
12475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12476 Alan Hayward <alan.hayward@arm.com>
12477 David Sherwood <david.sherwood@arm.com>
12479 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12480 in subroutines. Return the mode mask for non-integer modes.
12481 (cached_nonzero_bits): Change the type of the mode parameter
12482 to scalar_int_mode.
12483 (nonzero_bits1): Likewise. Remove early exit for other mode
12484 classes. Handle CONST_INT_P first and then check whether X
12485 also has a scalar integer mode.
12487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12492 Return a scalar_int_mode and assert that the size is greater than
12494 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12495 (op_by_pieces_d::op_by_pieces_d): Likewise.
12496 (op_by_pieces_d::run): Likewise.
12497 (can_store_by_pieces): Likewise.
12499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12500 Alan Hayward <alan.hayward@arm.com>
12501 David Sherwood <david.sherwood@arm.com>
12503 * combine.c (extract_left_shift): Add a mode argument and update
12505 (make_compound_operation_int): Change the type of the mode parameter
12506 to scalar_int_mode and update the call to extract_left_shift.
12508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12509 Alan Hayward <alan.hayward@arm.com>
12510 David Sherwood <david.sherwood@arm.com>
12512 * combine.c (simplify_and_const_int): Change the type of the mode
12513 parameter to scalar_int_mode.
12514 (simplify_and_const_int_1): Likewise. Update recursive call.
12516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12517 Alan Hayward <alan.hayward@arm.com>
12518 David Sherwood <david.sherwood@arm.com>
12520 * combine.c (simplify_compare_const): Check that the mode is a
12521 scalar_int_mode (rather than VOIDmode) before testing its
12523 (simplify_comparison): Move COMPARISON_P handling out of the
12524 loop and restrict the latter part of the loop to scalar_int_modes.
12525 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12526 and when considering SUBREG_REGs. Use is_int_mode instead of
12527 checking GET_MODE_CLASS against MODE_INT.
12529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12530 Alan Hayward <alan.hayward@arm.com>
12531 David Sherwood <david.sherwood@arm.com>
12533 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12534 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12535 shift_unit_mode and for modes involved in scalar shifts.
12537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12538 Alan Hayward <alan.hayward@arm.com>
12539 David Sherwood <david.sherwood@arm.com>
12541 * combine.c (force_int_to_mode): New function, split out from...
12542 (force_to_mode): ...here. Keep xmode up-to-date and use it
12543 instead of GET_MODE (x).
12545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12546 Alan Hayward <alan.hayward@arm.com>
12547 David Sherwood <david.sherwood@arm.com>
12549 * optabs-query.h (extraction_insn::struct_mode): Change type to
12550 opt_scalar_int_mode and update comment.
12551 (extraction_insn::field_mode): Change type to scalar_int_mode.
12552 (extraction_insn::pos_mode): Likewise.
12553 * combine.c (make_extraction): Update accordingly.
12554 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12555 (get_optab_extraction_insn): Likewise.
12556 * recog.c (simplify_while_replacing): Likewise.
12557 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12558 parameter to opt_scalar_int_mode.
12560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12561 Alan Hayward <alan.hayward@arm.com>
12562 David Sherwood <david.sherwood@arm.com>
12564 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12565 to a scalar_int_mode instead of a machine_mode.
12566 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12567 (get_best_mode): Return a boolean and use a pointer argument to store
12568 the selected mode. Replace the limit mode parameter with a bit limit.
12569 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12570 for the values returned by bit_field_mode_iterator::next_mode.
12571 (store_bit_field): Update call to get_best_mode.
12572 (store_fixed_bit_field): Likewise.
12573 (extract_fixed_bit_field): Likewise.
12574 * expr.c (optimize_bitfield_assignment_op): Likewise.
12575 * fold-const.c (optimize_bit_field_compare): Likewise.
12576 (fold_truth_andor_1): Likewise.
12577 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12578 Update for new type of m_mode.
12579 (get_best_mode): As above.
12581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12582 Alan Hayward <alan.hayward@arm.com>
12583 David Sherwood <david.sherwood@arm.com>
12585 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12586 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12587 (store_bit_field): Check is_a <scalar_int_mode> before calling
12588 strict_volatile_bitfield_p.
12589 (extract_bit_field): Likewise.
12591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12592 Alan Hayward <alan.hayward@arm.com>
12593 David Sherwood <david.sherwood@arm.com>
12595 * target.def (cstore_mode): Return a scalar_int_mode.
12596 * doc/tm.texi: Regenerate.
12597 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12598 * targhooks.h (default_cstore_mode): Likewise.
12599 * targhooks.c (default_cstore_mode): Likewise, using a forced
12601 * expmed.c (emit_cstore): Expect the target of the cstore to be
12604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12605 Alan Hayward <alan.hayward@arm.com>
12606 David Sherwood <david.sherwood@arm.com>
12608 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12610 (niter_desc): Likewise mode.
12611 (iv_analyze): Add a mode parameter.
12613 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12614 and change its type to scalar_int_mode.
12615 * loop-iv.c: Update commentary at head of file.
12616 (iv_constant): Pass the mode paraeter before the rtx it describes
12617 and change its type to scalar_int_mode. Remove VOIDmode handling.
12618 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12619 (iv_extend): Likewise.
12620 (shorten_into_mode): Likewise.
12621 (iv_add): Use scalar_int_mode.
12622 (iv_mult): Likewise.
12623 (iv_shift): Likewise.
12624 (canonicalize_iv_subregs): Likewise.
12625 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12626 it describes and change its mode to scalar_int_mode. Also change
12627 the type of the returned inner_mode to scalar_int_mode.
12628 (get_biv_step): Likewise, turning outer_mode from a pointer
12629 into a direct parameter. Update call to get_biv_step_1.
12630 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12631 iv_constant and get_biv_step.
12632 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12633 and change its type to scalar_int_mode. Don't initialise iv->mode
12634 to VOIDmode and remove later checks for its still being VOIDmode.
12635 Update calls to iv_analyze_op and iv_analyze_expr. Check
12636 is_a <scalar_int_mode> when changing the mode under consideration.
12637 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12638 Update call to iv_analyze_expr.
12639 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12640 inner register is not also a scalar_int_mode. Update call to
12642 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12643 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12644 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12645 separate mode class checks. Update calls to iv_analyze. Remove
12646 fix-up of VOIDmodes after iv_analyze_biv.
12647 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12648 don't have a scalar_int_mode. Update call to biv_p.
12650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12651 Alan Hayward <alan.hayward@arm.com>
12652 David Sherwood <david.sherwood@arm.com>
12654 * cfgexpand.c (convert_debug_memory_address): Use
12655 as_a <scalar_int_mode>.
12656 * combine.c (expand_compound_operation): Likewise.
12657 (make_extraction): Likewise.
12658 (change_zero_ext): Likewise.
12659 (simplify_comparison): Likewise.
12660 * cse.c (cse_insn): Likewise.
12661 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12662 (mem_loc_descriptor): Likewise.
12663 (loc_descriptor): Likewise.
12664 * expmed.c (init_expmed_one_mode): Likewise.
12665 (synth_mult): Likewise.
12666 (emit_store_flag_1): Likewise.
12667 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12668 of a comparison with size.
12669 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12670 (reduce_to_bit_field_precision): Likewise.
12671 * function.c (expand_function_end): Likewise.
12672 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12673 * loop-doloop.c (doloop_modify): Likewise.
12674 * optabs.c (expand_binop): Likewise.
12675 (expand_unop): Likewise.
12676 (expand_copysign_absneg): Likewise.
12677 (prepare_cmp_insn): Likewise.
12678 (maybe_legitimize_operand): Likewise.
12679 * recog.c (const_scalar_int_operand): Likewise.
12680 * rtlanal.c (get_address_mode): Likewise.
12681 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12682 (simplify_cond_clz_ctz): Likewise.
12683 * tree-nested.c (get_nl_goto_field): Likewise.
12684 * tree.c (build_vector_type_for_mode): Likewise.
12685 * var-tracking.c (use_narrower_mode): Likewise.
12687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12688 Alan Hayward <alan.hayward@arm.com>
12689 David Sherwood <david.sherwood@arm.com>
12691 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12692 * builtins.c (expand_builtin_signbit): Use it.
12693 * cfgexpand.c (expand_debug_expr): Likewise.
12694 * dojump.c (do_jump): Likewise.
12695 (do_compare_and_jump): Likewise.
12696 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12697 * expmed.c (make_tree): Likewise.
12698 * expr.c (expand_expr_real_2): Likewise.
12699 (expand_expr_real_1): Likewise.
12700 (try_casesi): Likewise.
12701 * fold-const-call.c (fold_const_call_ss): Likewise.
12702 * fold-const.c (unextend): Likewise.
12703 (extract_muldiv_1): Likewise.
12704 (fold_single_bit_test): Likewise.
12705 (native_encode_int): Likewise.
12706 (native_encode_string): Likewise.
12707 (native_interpret_int): Likewise.
12708 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12709 * internal-fn.c (expand_addsub_overflow): Likewise.
12710 (expand_neg_overflow): Likewise.
12711 (expand_mul_overflow): Likewise.
12712 (expand_arith_overflow): Likewise.
12713 * match.pd: Likewise.
12714 * stor-layout.c (layout_type): Likewise.
12715 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12716 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12717 * tree-ssanames.c (get_range_info): Likewise.
12718 * tree-switch-conversion.c (array_value_type) Likewise.
12719 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12720 (vect_recog_divmod_pattern): Likewise.
12721 (vect_recog_mixed_size_cond_pattern): Likewise.
12722 * tree-vrp.c (extract_range_basic): Likewise.
12723 (simplify_float_conversion_using_ranges): Likewise.
12724 * tree.c (int_fits_type_p): Likewise.
12725 * ubsan.c (instrument_bool_enum_load): Likewise.
12726 * varasm.c (mergeable_string_section): Likewise.
12727 (narrowing_initializer_constant_valid_p): Likewise.
12728 (output_constant): Likewise.
12730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12731 Alan Hayward <alan.hayward@arm.com>
12732 David Sherwood <david.sherwood@arm.com>
12734 * machmode.h (NARROWEST_INT_MODE): New macro.
12735 * expr.c (alignment_for_piecewise_move): Use it instead of
12736 GET_CLASS_NARROWEST_MODE (MODE_INT).
12737 (push_block): Likewise.
12738 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12740 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12743 Alan Hayward <alan.hayward@arm.com>
12744 David Sherwood <david.sherwood@arm.com>
12746 * postreload.c (move2add_valid_value_p): Change the type of the
12747 mode parameter to scalar_int_mode.
12748 (move2add_use_add2_insn): Add a mode parameter and use it instead
12750 (move2add_use_add3_insn): Likewise.
12751 (reload_cse_move2add): Update accordingly.
12753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12754 Alan Hayward <alan.hayward@arm.com>
12755 David Sherwood <david.sherwood@arm.com>
12757 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12759 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12760 * optabs.c (expand_unop): Likewise.
12762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12763 Alan Hayward <alan.hayward@arm.com>
12764 David Sherwood <david.sherwood@arm.com>
12766 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12767 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12768 (popcount_loc_descriptor): Likewise.
12769 (bswap_loc_descriptor): Likewise.
12770 (rotate_loc_descriptor): Likewise.
12771 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12772 calling the functions above.
12774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12775 Alan Hayward <alan.hayward@arm.com>
12776 David Sherwood <david.sherwood@arm.com>
12778 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12780 (try_combine): Likewise.
12781 (simplify_if_then_else): Likewise.
12782 * cse.c (cse_insn): Likewise.
12783 * dwarf2out.c (mem_loc_descriptor): Likewise.
12784 * emit-rtl.c (gen_lowpart_common): Likewise.
12785 * simplify-rtx.c (simplify_truncation): Likewise.
12786 (simplify_binary_operation_1): Likewise.
12787 (simplify_const_relational_operation): Likewise.
12788 (simplify_ternary_operation): Likewise.
12789 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12796 * reload.c (push_reload): Likewise.
12797 (find_reloads): Likewise.
12799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12800 Alan Hayward <alan.hayward@arm.com>
12801 David Sherwood <david.sherwood@arm.com>
12803 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12804 (make_compound_operation_int): Likewise.
12805 (change_zero_ext): Likewise.
12806 * expr.c (convert_move): Likewise.
12807 (convert_modes): Likewise.
12808 * fwprop.c (forward_propagate_subreg): Likewise.
12809 * loop-iv.c (get_biv_step_1): Likewise.
12810 * optabs.c (widen_operand): Likewise.
12811 * postreload.c (move2add_valid_value_p): Likewise.
12812 * recog.c (simplify_while_replacing): Likewise.
12813 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12814 (simplify_binary_operation_1): Likewise. Remove redundant
12815 mode equality check.
12817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12818 Alan Hayward <alan.hayward@arm.com>
12819 David Sherwood <david.sherwood@arm.com>
12821 * combine.c (combine_simplify_rtx): Add checks for
12822 is_a <scalar_int_mode>.
12823 (simplify_if_then_else): Likewise.
12824 (make_field_assignment): Likewise.
12825 (simplify_comparison): Likewise.
12826 * ifcvt.c (noce_try_bitop): Likewise.
12827 * loop-invariant.c (canonicalize_address_mult): Likewise.
12828 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12835 is_a <scalar_int_mode> instead of != BLKmode.
12837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12838 Alan Hayward <alan.hayward@arm.com>
12839 David Sherwood <david.sherwood@arm.com>
12841 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12842 instead of != VOIDmode.
12843 * combine.c (if_then_else_cond): Likewise.
12844 (change_zero_ext): Likewise.
12845 * dwarf2out.c (mem_loc_descriptor): Likewise.
12846 (loc_descriptor): Likewise.
12847 * rtlanal.c (canonicalize_condition): Likewise.
12848 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
12855 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12858 Alan Hayward <alan.hayward@arm.com>
12859 David Sherwood <david.sherwood@arm.com>
12861 * wide-int.h (int_traits<unsigned char>) New class.
12862 (int_traits<unsigned short>) Likewise.
12863 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12864 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12866 * combine.c (set_nonzero_bits_and_sign_copies): Use
12867 is_a <scalar_int_mode>.
12868 (find_split_point): Likewise.
12869 (combine_simplify_rtx): Likewise.
12870 (simplify_logical): Likewise.
12871 (expand_compound_operation): Likewise.
12872 (expand_field_assignment): Likewise.
12873 (make_compound_operation): Likewise.
12874 (extended_count): Likewise.
12875 (change_zero_ext): Likewise.
12876 (simplify_comparison): Likewise.
12877 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12878 (ucompare_loc_descriptor): Likewise.
12879 (minmax_loc_descriptor): Likewise.
12880 (mem_loc_descriptor): Likewise.
12881 (loc_descriptor): Likewise.
12882 * expmed.c (init_expmed_one_mode): Likewise.
12883 * lra-constraints.c (lra_constraint_offset): Likewise.
12884 * optabs.c (prepare_libcall_arg): Likewise.
12885 * postreload.c (move2add_note_store): Likewise.
12886 * reload.c (operands_match_p): Likewise.
12887 * rtl.h (load_extend_op): Likewise.
12888 * rtlhooks.c (gen_lowpart_general): Likewise.
12889 * simplify-rtx.c (simplify_truncation): Likewise.
12890 (simplify_unary_operation_1): Likewise.
12891 (simplify_binary_operation_1): Likewise.
12892 (simplify_const_binary_operation): Likewise.
12893 (simplify_const_relational_operation): Likewise.
12894 (simplify_subreg): Likewise.
12895 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12896 * var-tracking.c (adjust_mems): Likewise.
12897 (prepare_call_arguments): Likewise.
12899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12900 Alan Hayward <alan.hayward@arm.com>
12901 David Sherwood <david.sherwood@arm.com>
12903 * machmode.h (is_int_mode): New fuction.
12904 * combine.c (find_split_point): Use it.
12905 (combine_simplify_rtx): Likewise.
12906 (simplify_if_then_else): Likewise.
12907 (simplify_set): Likewise.
12908 (simplify_shift_const_1): Likewise.
12909 (simplify_comparison): Likewise.
12910 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12911 * cse.c (notreg_cost): Likewise.
12912 (cse_insn): Likewise.
12913 * cselib.c (cselib_lookup_1): Likewise.
12914 * dojump.c (do_jump_1): Likewise.
12915 (do_compare_rtx_and_jump): Likewise.
12916 * dse.c (get_call_args): Likewise.
12917 * dwarf2out.c (rtl_for_decl_init): Likewise.
12918 (native_encode_initializer): Likewise.
12919 * expmed.c (emit_store_flag_1): Likewise.
12920 (emit_store_flag): Likewise.
12921 * expr.c (convert_modes): Likewise.
12922 (store_field): Likewise.
12923 (expand_expr_real_1): Likewise.
12924 * fold-const.c (fold_read_from_constant_string): Likewise.
12925 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12926 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12927 * optabs.c (expand_binop): Likewise.
12928 (expand_unop): Likewise.
12929 (expand_abs_nojump): Likewise.
12930 (expand_one_cmpl_abs_nojump): Likewise.
12931 * simplify-rtx.c (mode_signbit_p): Likewise.
12932 (val_signbit_p): Likewise.
12933 (val_signbit_known_set_p): Likewise.
12934 (val_signbit_known_clear_p): Likewise.
12935 (simplify_relational_operation_1): Likewise.
12936 * tree.c (vector_type_mode): Likewise.
12938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12939 Alan Hayward <alan.hayward@arm.com>
12940 David Sherwood <david.sherwood@arm.com>
12942 * machmode.h (smallest_mode_for_size): Fix formatting.
12943 (smallest_int_mode_for_size): New function.
12944 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12945 instead of smallest_mode_for_size.
12946 * combine.c (make_extraction): Likewise.
12947 * config/arc/arc.c (arc_expand_movmem): Likewise.
12948 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12949 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12950 * config/s390/s390.c (s390_expand_insv): Likewise.
12951 * config/sparc/sparc.c (assign_int_registers): Likewise.
12952 * config/spu/spu.c (spu_function_value): Likewise.
12953 (spu_function_arg): Likewise.
12954 * coverage.c (get_gcov_type): Likewise.
12955 (get_gcov_unsigned_t): Likewise.
12956 * dse.c (find_shift_sequence): Likewise.
12957 * expmed.c (store_bit_field_1): Likewise.
12958 * expr.c (convert_move): Likewise.
12959 (store_field): Likewise.
12960 * internal-fn.c (expand_arith_overflow): Likewise.
12961 * optabs-query.c (get_best_extraction_insn): Likewise.
12962 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12963 * stor-layout.c (layout_type): Likewise.
12964 (initialize_sizetypes): Likewise.
12965 * targhooks.c (default_get_mask_mode): Likewise.
12966 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12969 Alan Hayward <alan.hayward@arm.com>
12970 David Sherwood <david.sherwood@arm.com>
12972 * machmode.h (opt_mode::else_blk): New function.
12973 (int_mode_for_mode): Declare.
12974 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12975 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12977 * cfgexpand.c (expand_debug_expr): Likewise.
12978 * combine.c (gen_lowpart_or_truncate): Likewise.
12979 (gen_lowpart_for_combine): Likewise.
12980 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12981 * config/avr/avr.c (avr_to_int_mode): Likewise.
12982 (avr_out_plus_1): Likewise.
12983 (avr_out_plus): Likewise.
12984 (avr_out_round): Likewise.
12985 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12986 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12987 (s390_expand_vcond): Likewise.
12988 * config/spu/spu.c (spu_split_immediate): Likewise.
12989 (spu_expand_mov): Likewise.
12990 * dse.c (get_stored_val): Likewise.
12991 * expmed.c (store_bit_field_1): Likewise.
12992 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12994 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12995 (extract_low_bits): Likewise.
12996 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12997 handling rather than repeating the check.
12998 (emit_group_store): Likewise.
12999 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13000 * optabs.c (expand_absneg_bit): Likewise.
13001 (expand_copysign_absneg): Likewise.
13002 (expand_copysign_bit): Likewise.
13003 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13004 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13005 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13006 * var-tracking.c (prepare_call_arguments): Likewise.
13007 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13008 int_mode_for_mode instead of mode_for_size.
13009 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13012 Alan Hayward <alan.hayward@arm.com>
13013 David Sherwood <david.sherwood@arm.com>
13015 * machmode.h (int_mode_for_size): New function.
13016 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13017 instead of mode_for_size.
13018 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13020 * combine.c (expand_field_assignment): Use int_mode_for_size
13021 instead of mode_for_size.
13022 (make_extraction): Likewise.
13023 (simplify_shift_const_1): Likewise.
13024 (simplify_comparison): Likewise.
13025 * dojump.c (do_jump): Likewise.
13026 * dwarf2out.c (mem_loc_descriptor): Likewise.
13027 * emit-rtl.c (init_derived_machine_modes): Likewise.
13028 * expmed.c (flip_storage_order): Likewise.
13029 (convert_extracted_bit_field): Likewise.
13030 * expr.c (copy_blkmode_from_reg): Likewise.
13031 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13032 * internal-fn.c (expand_mul_overflow): Likewise.
13033 * lower-subreg.c (simple_move): Likewise.
13034 * optabs-libfuncs.c (init_optabs): Likewise.
13035 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13036 * tree.c (vector_type_mode): Likewise.
13037 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13038 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13039 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13040 * tree-vect-stmts.c (vectorizable_load): Likewise.
13041 (vectorizable_store): Likewise.
13043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13044 Alan Hayward <alan.hayward@arm.com>
13045 David Sherwood <david.sherwood@arm.com>
13047 * coretypes.h (pod_mode): New type.
13048 (scalar_int_mode_pod): New typedef.
13049 * machmode.h (pod_mode): New class.
13050 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13051 * genmodes.c (emit_mode_int_n): Update accordingly.
13052 * lower-subreg.h (target_lower_subreg): Change type to
13053 scalar_int_mode_pod.
13054 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13055 scalar_int_mode_pod.
13057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13058 Alan Hayward <alan.hayward@arm.com>
13059 David Sherwood <david.sherwood@arm.com>
13061 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13062 machine_mode to scalar_int_mode.
13063 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13064 (rs6000_option_override_internal): Remove cast to int.
13065 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13066 machine_mode to scalar_int_mode.
13067 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13068 (rs6000_option_override_internal): Remove cast to int.
13069 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13070 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13072 * config/s390/s390.c (s390_expand_builtin): Likewise.
13073 * coretypes.h (scalar_int_mode): New type.
13074 (opt_scalar_int_mode): New typedef.
13075 * machmode.h (scalar_int_mode): New class.
13076 (scalar_int_mode::includes_p): New function.
13077 (byte_mode): Change type to scalar_int_mode.
13078 (word_mode): Likewise.
13079 (ptr_mode): Likewise.
13080 * emit-rtl.c (byte_mode): Likewise.
13081 (word_mode): Likewise.
13082 (ptr_mode): Likewise.
13083 (init_derived_machine_modes): Update accordingly.
13084 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13085 and MODE_PARTIAL_INT.
13086 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13087 opt_scalar_int_mode.
13089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13090 Alan Hayward <alan.hayward@arm.com>
13091 David Sherwood <david.sherwood@arm.com>
13093 * target.def (libgcc_floating_mode_supported_p): Take a
13095 * doc/tm.texi: Regenerate.
13096 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13098 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13099 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13103 Alan Hayward <alan.hayward@arm.com>
13104 David Sherwood <david.sherwood@arm.com>
13106 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13107 * doc/tm.texi: Regenerate.
13108 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13109 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13110 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13111 * targhooks.h (default_floatn_mode): Likewise.
13112 * targhooks.c (default_floatn_mode): Likewise.
13113 * tree.c (build_common_tree_nodes): Update accordingly.
13115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13116 Alan Hayward <alan.hayward@arm.com>
13117 David Sherwood <david.sherwood@arm.com>
13119 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13120 (mode_iterator::iterate_p): Likewise.
13121 (mode_iterator::get_wider): Likewise.
13122 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13125 Alan Hayward <alan.hayward@arm.com>
13126 David Sherwood <david.sherwood@arm.com>
13128 * coretypes.h (opt_scalar_float_mode): New typedef.
13129 * machmode.h (float_mode_for_size): New function.
13130 * emit-rtl.c (double_mode): Delete.
13131 (init_emit_once): Use float_mode_for_size.
13132 * stor-layout.c (layout_type): Likewise.
13133 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13136 Alan Hayward <alan.hayward@arm.com>
13137 David Sherwood <david.sherwood@arm.com>
13139 * output.h (assemble_real): Take a scalar_float_mode.
13140 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13141 * config/arm/arm.md (consttable_4): Likewise.
13142 (consttable_8): Likewise.
13143 (consttable_16): Likewise.
13144 * config/mips/mips.md (consttable_float): Likewise.
13145 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13146 * varasm.c (assemble_real): Take a scalar_float_mode.
13147 (output_constant_pool_2): Update accordingly.
13148 (output_constant): Likewise.
13150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13151 Alan Hayward <alan.hayward@arm.com>
13152 David Sherwood <david.sherwood@arm.com>
13154 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13155 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13156 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13157 (native_encode_real): Likewise.
13158 (native_interpret_real): Likewise.
13159 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13160 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13163 Alan Hayward <alan.hayward@arm.com>
13164 David Sherwood <david.sherwood@arm.com>
13166 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13167 <scalar_float_mode>. Simplify.
13168 (gen_extend_conv_libfunc): Likewise.
13170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13171 Alan Hayward <alan.hayward@arm.com>
13172 David Sherwood <david.sherwood@arm.com>
13174 * coretypes.h (scalar_float_mode): New type.
13175 * machmode.h (mode_traits::from_int): Use machine_mode if
13176 USE_ENUM_MODES is defined.
13177 (is_a): New function.
13179 (dyn_cast): Likewise.
13180 (scalar_float_mode): New class.
13181 (scalar_float_mode::includes_p): New function.
13182 (is_float_mode): Likewise.
13183 * gdbhooks.py (MachineModePrinter): New class.
13184 (build_pretty_printer): Use it for scalar_float_mode.
13185 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13186 (format_helper::format_helper): Turn into a template.
13187 * genmodes.c (get_mode_class): New function.
13188 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13189 or machine_mode if none.
13190 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13191 as_a <scalar_float_mode>.
13192 * dwarf2out.c (mem_loc_descriptor): Likewise.
13193 (insert_float): Likewise.
13194 (add_const_value_attribute): Likewise.
13195 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13196 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13197 (expand_unop): Update accordingly.
13198 (expand_abs_nojump): Likewise.
13199 (expand_copysign_absneg): Take a scalar_float_mode.
13200 (expand_copysign_bit): Likewise.
13201 (expand_copysign): Update accordingly.
13203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13204 Alan Hayward <alan.hayward@arm.com>
13205 David Sherwood <david.sherwood@arm.com>
13207 * coretypes.h (opt_mode): New class.
13208 * machmode.h (opt_mode): Likewise.
13209 (opt_mode::else_void): New function.
13210 (opt_mode::require): Likewise.
13211 (opt_mode::exists): Likewise.
13212 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13213 (GET_MODE_2XWIDER_MODE): Likewise.
13214 (mode_iterator::get_wider): Update accordingly.
13215 (mode_iterator::get_2xwider): Likewise.
13216 (mode_iterator::get_known_wider): Likewise, turning into a template.
13217 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13218 forcing a wider mode to exist.
13219 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13220 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13221 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13222 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13223 * lower-subreg.c (init_lower_subreg): Likewise.
13224 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13225 on the final iteration.
13226 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13227 a wider mode exists before asking for a move pattern.
13228 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13229 forcing a wider mode to exist.
13230 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13231 returning false if no such mode exists.
13232 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13233 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13234 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13235 Avoid checking for a MODE_INT if we already know the mode is not a
13237 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13238 forcing a wider mode to exist.
13239 (expmed_mult_highpart_optab): Likewise.
13240 (expmed_mult_highpart): Likewise.
13241 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13243 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13244 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13245 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13246 * internal-fn.c (expand_mul_overflow): Update use of
13247 GET_MODE_2XWIDER_MODE.
13248 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13249 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13250 GET_MODE_WIDER_MODE.
13251 (convert_plusminus_to_widen): Likewise.
13252 * tree-switch-conversion.c (array_value_type): Likewise.
13253 * var-tracking.c (emit_note_insn_var_location): Likewise.
13254 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13255 Return false inside rather than outside the loop if no wider mode
13257 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13258 and GET_MODE_2XWIDER_MODE
13259 (can_compare_p): Use else_void.
13260 * gdbhooks.py (OptMachineModePrinter): New class.
13261 (build_pretty_printer): Use it for opt_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 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13268 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13271 Alan Hayward <alan.hayward@arm.com>
13272 David Sherwood <david.sherwood@arm.com>
13274 * machmode.h (mode_traits): New structure.
13275 (get_narrowest_mode): New function.
13276 (mode_iterator::start): Likewise.
13277 (mode_iterator::iterate_p): Likewise.
13278 (mode_iterator::get_wider): Likewise.
13279 (mode_iterator::get_known_wider): Likewise.
13280 (mode_iterator::get_2xwider): Likewise.
13281 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13282 (FOR_EACH_MODE): Likewise.
13283 (FOR_EACH_MODE_FROM): Likewise.
13284 (FOR_EACH_MODE_UNTIL): Likewise.
13285 (FOR_EACH_WIDER_MODE): Likewise.
13286 (FOR_EACH_2XWIDER_MODE): Likewise.
13287 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13288 * combine.c (simplify_comparison): Likewise
13289 * config/i386/i386.c (type_natural_mode): Likewise.
13290 * cse.c (cse_insn): Likewise.
13291 * dse.c (find_shift_sequence): Likewise.
13292 * emit-rtl.c (init_derived_machine_modes): Likewise.
13293 (init_emit_once): Likewise.
13294 * explow.c (hard_function_value): Likewise.
13295 * expmed.c (extract_fixed_bit_field_1): Likewise.
13296 (extract_bit_field_1): Likewise.
13297 (expand_divmod): Likewise.
13298 (emit_store_flag_1): Likewise.
13299 * expr.c (init_expr_target): Likewise.
13300 (convert_move): Likewise.
13301 (alignment_for_piecewise_move): Likewise.
13302 (widest_int_mode_for_size): Likewise.
13303 (emit_block_move_via_movmem): Likewise.
13304 (copy_blkmode_to_reg): Likewise.
13305 (set_storage_via_setmem): Likewise.
13306 (compress_float_constant): Likewise.
13307 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13308 * optabs-query.c (get_best_extraction_insn): Likewise.
13309 * optabs.c (expand_binop): Likewise.
13310 (expand_twoval_unop): Likewise.
13311 (expand_twoval_binop): Likewise.
13312 (widen_leading): Likewise.
13313 (widen_bswap): Likewise.
13314 (expand_parity): Likewise.
13315 (expand_unop): Likewise.
13316 (prepare_cmp_insn): Likewise.
13317 (prepare_float_lib_cmp): Likewise.
13318 (expand_float): Likewise.
13319 (expand_fix): Likewise.
13320 (expand_sfix_optab): Likewise.
13321 * postreload.c (move2add_use_add2_insn): Likewise.
13322 * reg-stack.c (reg_to_stack): Likewise.
13323 * reginfo.c (choose_hard_reg_mode): Likewise.
13324 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13325 * stor-layout.c (mode_for_size): Likewise.
13326 (smallest_mode_for_size): Likewise.
13327 (mode_for_vector): Likewise.
13328 (finish_bitfield_representative): Likewise.
13329 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13330 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13331 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13332 * var-tracking.c (prepare_call_arguments): Likewise.
13334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13335 Alan Hayward <alan.hayward@arm.com>
13336 David Sherwood <david.sherwood@arm.com>
13338 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13339 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13340 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13341 * machmode.h (mode_size): Move earlier in file.
13342 (mode_precision): Likewise.
13343 (mode_inner): Likewise.
13344 (mode_nunits): Likewise.
13345 (mode_unit_size): Likewise.
13346 (unit_unit_precision): Likewise.
13347 (mode_wider): Likewise.
13348 (mode_2xwider): Likewise.
13349 (machine_mode): New class.
13350 (mode_to_bytes): New function.
13351 (mode_to_bits): Likewise.
13352 (mode_to_precision): Likewise.
13353 (mode_to_inner): Likewise.
13354 (mode_to_unit_size): Likewise.
13355 (mode_to_unit_precision): Likewise.
13356 (mode_to_nunits): Likewise.
13357 (GET_MODE_SIZE): Use mode_to_bytes.
13358 (GET_MODE_BITSIZE): Use mode_to_bits.
13359 (GET_MODE_PRECISION): Use mode_to_precision.
13360 (GET_MODE_INNER): Use mode_to_inner.
13361 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13362 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13363 (GET_MODE_NUNITS): Use mode_to_nunits.
13364 * system.h (ALWAYS_INLINE): New macro.
13365 * config/powerpcspe/powerpcspe-c.c
13366 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13367 int for arg1_mode and arg2_mode.
13369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13370 Alan Hayward <alan.hayward@arm.com>
13371 David Sherwood <david.sherwood@arm.com>
13373 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13374 Prefix mode names with E_ in case statements.
13375 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13376 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13377 (aarch64_split_simd_move): Likewise.
13378 (aarch64_gen_storewb_pair): Likewise.
13379 (aarch64_gen_loadwb_pair): Likewise.
13380 (aarch64_gen_store_pair): Likewise.
13381 (aarch64_gen_load_pair): Likewise.
13382 (aarch64_get_condition_code_1): Likewise.
13383 (aarch64_constant_pool_reload_icode): Likewise.
13384 (get_rsqrte_type): Likewise.
13385 (get_rsqrts_type): Likewise.
13386 (get_recpe_type): Likewise.
13387 (get_recps_type): Likewise.
13388 (aarch64_gimplify_va_arg_expr): Likewise.
13389 (aarch64_simd_container_mode): Likewise.
13390 (aarch64_emit_load_exclusive): Likewise.
13391 (aarch64_emit_store_exclusive): Likewise.
13392 (aarch64_expand_compare_and_swap): Likewise.
13393 (aarch64_gen_atomic_cas): Likewise.
13394 (aarch64_emit_bic): Likewise.
13395 (aarch64_emit_atomic_swap): Likewise.
13396 (aarch64_emit_atomic_load_op): Likewise.
13397 (aarch64_evpc_trn): Likewise.
13398 (aarch64_evpc_uzp): Likewise.
13399 (aarch64_evpc_zip): Likewise.
13400 (aarch64_evpc_ext): Likewise.
13401 (aarch64_evpc_rev): Likewise.
13402 (aarch64_evpc_dup): Likewise.
13403 (aarch64_gen_ccmp_first): Likewise.
13404 (aarch64_gen_ccmp_next): Likewise.
13405 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13406 (alpha_emit_xfloating_libcall): Likewise.
13407 (emit_insxl): Likewise.
13408 (alpha_arg_type): Likewise.
13409 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13410 (arc_preferred_simd_mode): Likewise.
13411 (arc_secondary_reload): Likewise.
13412 (get_arc_condition_code): Likewise.
13413 (arc_print_operand): Likewise.
13414 (arc_legitimate_constant_p): Likewise.
13415 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13416 * config/arc/arc.md (casesi_load): Likewise.
13417 (casesi_compact_jump): Likewise.
13418 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13419 (cc_use_register): Likewise.
13420 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13421 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13422 (arm_init_iwmmxt_builtins): Likewise.
13423 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13424 (neon_expand_vector_init): Likewise.
13425 (arm_attr_length_move_neon): Likewise.
13426 (maybe_get_arm_condition_code): Likewise.
13427 (arm_emit_vector_const): Likewise.
13428 (arm_preferred_simd_mode): Likewise.
13429 (arm_output_iwmmxt_tinsr): Likewise.
13430 (thumb1_output_casesi): Likewise.
13431 (thumb2_output_casesi): Likewise.
13432 (arm_emit_load_exclusive): Likewise.
13433 (arm_emit_store_exclusive): Likewise.
13434 (arm_expand_compare_and_swap): Likewise.
13435 (arm_evpc_neon_vuzp): Likewise.
13436 (arm_evpc_neon_vzip): Likewise.
13437 (arm_evpc_neon_vrev): Likewise.
13438 (arm_evpc_neon_vtrn): Likewise.
13439 (arm_evpc_neon_vext): Likewise.
13440 (arm_validize_comparison): Likewise.
13441 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13442 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13443 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13444 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13445 (c6x_preferred_simd_mode): Likewise.
13446 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13447 (epiphany_rtx_costs): Likewise.
13448 * config/epiphany/predicates.md (proper_comparison_operator):
13450 * config/frv/frv.c (condexec_memory_operand): Likewise.
13451 (frv_emit_move): Likewise.
13452 (output_move_single): Likewise.
13453 (output_condmove_single): Likewise.
13454 (frv_hard_regno_mode_ok): Likewise.
13455 (frv_matching_accg_mode): Likewise.
13456 * config/h8300/h8300.c (split_adds_subs): Likewise.
13457 (h8300_rtx_costs): Likewise.
13458 (h8300_print_operand): Likewise.
13459 (compute_mov_length): Likewise.
13460 (output_logical_op): Likewise.
13461 (compute_logical_op_length): Likewise.
13462 (compute_logical_op_cc): Likewise.
13463 (h8300_shift_needs_scratch_p): Likewise.
13464 (output_a_shift): Likewise.
13465 (compute_a_shift_length): Likewise.
13466 (compute_a_shift_cc): Likewise.
13467 (expand_a_rotate): Likewise.
13468 (output_a_rotate): Likewise.
13469 * config/i386/i386.c (classify_argument): Likewise.
13470 (function_arg_advance_32): Likewise.
13471 (function_arg_32): Likewise.
13472 (function_arg_64): Likewise.
13473 (function_value_64): Likewise.
13474 (ix86_gimplify_va_arg): Likewise.
13475 (ix86_legitimate_constant_p): Likewise.
13476 (put_condition_code): Likewise.
13477 (split_double_mode): Likewise.
13478 (ix86_avx256_split_vector_move_misalign): Likewise.
13479 (ix86_expand_vector_logical_operator): Likewise.
13480 (ix86_split_idivmod): Likewise.
13481 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13482 (ix86_build_const_vector): Likewise.
13483 (ix86_build_signbit_mask): Likewise.
13484 (ix86_match_ccmode): Likewise.
13485 (ix86_cc_modes_compatible): Likewise.
13486 (ix86_expand_branch): Likewise.
13487 (ix86_expand_sse_cmp): Likewise.
13488 (ix86_expand_sse_movcc): Likewise.
13489 (ix86_expand_int_sse_cmp): Likewise.
13490 (ix86_expand_vec_perm_vpermi2): Likewise.
13491 (ix86_expand_vec_perm): Likewise.
13492 (ix86_expand_sse_unpack): Likewise.
13493 (ix86_expand_int_addcc): Likewise.
13494 (ix86_split_to_parts): Likewise.
13495 (ix86_vectorize_builtin_gather): Likewise.
13496 (ix86_vectorize_builtin_scatter): Likewise.
13497 (avx_vpermilp_parallel): Likewise.
13498 (inline_memory_move_cost): Likewise.
13499 (ix86_tieable_integer_mode_p): Likewise.
13500 (x86_maybe_negate_const_int): Likewise.
13501 (ix86_expand_vector_init_duplicate): Likewise.
13502 (ix86_expand_vector_init_one_nonzero): Likewise.
13503 (ix86_expand_vector_init_one_var): Likewise.
13504 (ix86_expand_vector_init_concat): Likewise.
13505 (ix86_expand_vector_init_interleave): Likewise.
13506 (ix86_expand_vector_init_general): Likewise.
13507 (ix86_expand_vector_set): Likewise.
13508 (ix86_expand_vector_extract): Likewise.
13509 (emit_reduc_half): Likewise.
13510 (ix86_emit_i387_round): Likewise.
13511 (ix86_mangle_type): Likewise.
13512 (ix86_expand_round_sse4): Likewise.
13513 (expand_vec_perm_blend): Likewise.
13514 (canonicalize_vector_int_perm): Likewise.
13515 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13516 (expand_vec_perm_1): Likewise.
13517 (expand_vec_perm_interleave3): Likewise.
13518 (expand_vec_perm_even_odd_pack): Likewise.
13519 (expand_vec_perm_even_odd_1): Likewise.
13520 (expand_vec_perm_broadcast_1): Likewise.
13521 (ix86_vectorize_vec_perm_const_ok): Likewise.
13522 (ix86_expand_vecop_qihi): Likewise.
13523 (ix86_expand_mul_widen_hilo): Likewise.
13524 (ix86_expand_sse2_abs): Likewise.
13525 (ix86_expand_pextr): Likewise.
13526 (ix86_expand_pinsr): Likewise.
13527 (ix86_preferred_simd_mode): Likewise.
13528 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13529 * config/i386/sse.md (*andnot<mode>3): Likewise.
13530 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13531 (*<code><mode>3): Likewise.
13532 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13533 (ia64_expand_atomic_op): Likewise.
13534 (ia64_arg_type): Likewise.
13535 (ia64_mode_to_int): Likewise.
13536 (ia64_scalar_mode_supported_p): Likewise.
13537 (ia64_vector_mode_supported_p): Likewise.
13538 (expand_vec_perm_broadcast): Likewise.
13539 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13540 (iq2000_function_arg_advance): Likewise.
13541 (iq2000_function_arg): Likewise.
13542 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13543 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13544 (m68k_libcall_value): Likewise.
13545 (m68k_function_value): Likewise.
13546 (sched_attr_op_type): Likewise.
13547 * config/mcore/mcore.c (mcore_output_move): Likewise.
13548 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13550 (microblaze_function_arg): Likewise.
13551 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13552 (mips_print_operand): Likewise.
13553 (mips_mode_ok_for_mov_fmt_p): Likewise.
13554 (mips_vector_mode_supported_p): Likewise.
13555 (mips_preferred_simd_mode): Likewise.
13556 (mips_expand_vpc_loongson_even_odd): Likewise.
13557 (mips_expand_vec_unpack): Likewise.
13558 (mips_expand_vi_broadcast): Likewise.
13559 (mips_expand_vector_init): Likewise.
13560 (mips_expand_vec_reduc): Likewise.
13561 (mips_expand_msa_cmp): Likewise.
13562 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13563 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13564 (cc_flags_for_mode): Likewise.
13565 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13566 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13567 (nds32_output_casesi_pc_relative): Likewise.
13568 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13569 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13570 (nvptx_gen_unpack): Likewise.
13571 (nvptx_gen_pack): Likewise.
13572 (nvptx_gen_shuffle): Likewise.
13573 (nvptx_gen_wcast): Likewise.
13574 (nvptx_preferred_simd_mode): Likewise.
13575 * config/pa/pa.c (pa_secondary_reload): Likewise.
13576 * config/pa/predicates.md (base14_operand): Likewise.
13577 * config/powerpcspe/powerpcspe-c.c
13578 (altivec_resolve_overloaded_builtin): Likewise.
13579 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13581 (rs6000_preferred_simd_mode): Likewise.
13582 (output_vec_const_move): Likewise.
13583 (rs6000_expand_vector_extract): Likewise.
13584 (rs6000_split_vec_extract_var): Likewise.
13585 (reg_offset_addressing_ok_p): Likewise.
13586 (rs6000_legitimate_offset_address_p): Likewise.
13587 (rs6000_legitimize_address): Likewise.
13588 (rs6000_emit_set_const): Likewise.
13589 (rs6000_const_vec): Likewise.
13590 (rs6000_emit_move): Likewise.
13591 (spe_build_register_parallel): Likewise.
13592 (rs6000_darwin64_record_arg_recurse): Likewise.
13593 (swap_selector_for_mode): Likewise.
13594 (spe_init_builtins): Likewise.
13595 (paired_init_builtins): Likewise.
13596 (altivec_init_builtins): Likewise.
13597 (do_load_for_compare): Likewise.
13598 (rs6000_generate_compare): Likewise.
13599 (rs6000_expand_float128_convert): Likewise.
13600 (emit_load_locked): Likewise.
13601 (emit_store_conditional): Likewise.
13602 (rs6000_output_function_epilogue): Likewise.
13603 (rs6000_handle_altivec_attribute): Likewise.
13604 (rs6000_function_value): Likewise.
13605 (emit_fusion_gpr_load): Likewise.
13606 (emit_fusion_p9_load): Likewise.
13607 (emit_fusion_p9_store): Likewise.
13608 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13609 (fusion_gpr_mem_load): Likewise.
13610 (fusion_addis_mem_combo_load): Likewise.
13611 (fusion_addis_mem_combo_store): Likewise.
13612 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13613 (fusion_gpr_mem_load): Likewise.
13614 (fusion_addis_mem_combo_load): Likewise.
13615 (fusion_addis_mem_combo_store): Likewise.
13616 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13618 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13619 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13620 (rs6000_preferred_simd_mode): Likewise.
13621 (output_vec_const_move): Likewise.
13622 (rs6000_expand_vector_extract): Likewise.
13623 (rs6000_split_vec_extract_var): Likewise.
13624 (reg_offset_addressing_ok_p): Likewise.
13625 (rs6000_legitimate_offset_address_p): Likewise.
13626 (rs6000_legitimize_address): Likewise.
13627 (rs6000_emit_set_const): Likewise.
13628 (rs6000_const_vec): Likewise.
13629 (rs6000_emit_move): Likewise.
13630 (rs6000_darwin64_record_arg_recurse): Likewise.
13631 (swap_selector_for_mode): Likewise.
13632 (paired_init_builtins): Likewise.
13633 (altivec_init_builtins): Likewise.
13634 (rs6000_expand_float128_convert): Likewise.
13635 (emit_load_locked): Likewise.
13636 (emit_store_conditional): Likewise.
13637 (rs6000_output_function_epilogue): Likewise.
13638 (rs6000_handle_altivec_attribute): Likewise.
13639 (rs6000_function_value): Likewise.
13640 (emit_fusion_gpr_load): Likewise.
13641 (emit_fusion_p9_load): Likewise.
13642 (emit_fusion_p9_store): Likewise.
13643 * config/rx/rx.c (rx_gen_move_template): Likewise.
13644 (flags_from_mode): Likewise.
13645 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13646 (s390_slb_comparison): Likewise.
13647 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13648 (s390_vector_mode_supported_p): Likewise.
13649 (s390_cc_modes_compatible): Likewise.
13650 (s390_match_ccmode_set): Likewise.
13651 (s390_canonicalize_comparison): Likewise.
13652 (s390_emit_compare_and_swap): Likewise.
13653 (s390_branch_condition_mask): Likewise.
13654 (s390_rtx_costs): Likewise.
13655 (s390_secondary_reload): Likewise.
13656 (__SECONDARY_RELOAD_CASE): Likewise.
13657 (s390_expand_cs): Likewise.
13658 (s390_preferred_simd_mode): Likewise.
13659 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13660 * config/sh/sh.c (sh_print_operand): Likewise.
13661 (dump_table): Likewise.
13662 (sh_secondary_reload): Likewise.
13663 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13664 * config/sh/sh.md (casesi_worker_1): Likewise.
13665 (casesi_worker_2): Likewise.
13666 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13667 (fcc_comparison_operator): Likewise.
13668 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13669 (emit_soft_tfmode_cvt): Likewise.
13670 (sparc_preferred_simd_mode): Likewise.
13671 (output_cbranch): Likewise.
13672 (sparc_print_operand): Likewise.
13673 (sparc_expand_vec_perm_bmask): Likewise.
13674 (vector_init_bshuffle): Likewise.
13675 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13676 (spu_vector_mode_supported_p): Likewise.
13677 (spu_expand_insv): Likewise.
13678 (spu_emit_branch_or_set): Likewise.
13679 (spu_handle_vector_attribute): Likewise.
13680 (spu_builtin_splats): Likewise.
13681 (spu_builtin_extract): Likewise.
13682 (spu_builtin_promote): Likewise.
13683 (spu_expand_sign_extend): Likewise.
13684 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13685 (tilegx_simd_int): Likewise.
13686 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13687 (tilepro_simd_int): Likewise.
13688 * config/v850/v850.c (const_double_split): Likewise.
13689 (v850_print_operand): Likewise.
13690 (ep_memory_offset): Likewise.
13691 * config/vax/vax.c (vax_rtx_costs): Likewise.
13692 (vax_output_int_move): Likewise.
13693 (vax_output_int_add): Likewise.
13694 (vax_output_int_subtract): Likewise.
13695 * config/visium/predicates.md (visium_branch_operator): Likewise.
13696 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13697 (visium_print_operand_address): Likewise.
13698 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13699 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13700 (xtensa_expand_conditional_branch): Likewise.
13701 (xtensa_copy_incoming_a7): Likewise.
13702 (xtensa_output_literal): Likewise.
13703 * dfp.c (decimal_real_maxval): Likewise.
13704 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13707 Alan Hayward <alan.hayward@arm.com>
13708 David Sherwood <david.sherwood@arm.com>
13710 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13711 (mode_nunits_inline): Likewise.
13712 (mode_inner_inline): Likewise.
13713 (mode_unit_size_inline): Likewise.
13714 (mode_unit_precision_inline): Likewise.
13715 (emit_insn_modes_h): Likewise. Also emit a #define of the
13717 (emit_mode_wider): Add an E_ prefix to mode names.
13718 (emit_mode_complex): Likewise.
13719 (emit_mode_inner): Likewise.
13720 (emit_mode_adjustments): Likewise.
13721 (emit_mode_int_n): Likewise.
13722 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13723 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13724 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13725 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13726 (CRC32_BUILTIN, ENTRY): Likewise.
13727 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13728 (aarch64_pop_regs): Likewise.
13729 (aarch64_process_components): Likewise.
13730 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13731 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13732 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13733 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13734 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13735 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13737 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13738 (pcmpistr): Likewise.
13739 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13740 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13741 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13743 * config/rl78/rl78.c (mduc_regs): Likewise.
13744 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13745 (htm_expand_builtin): Likewise.
13746 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13747 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13748 * config/xtensa/xtensa.c (print_operand): Likewise.
13749 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13750 (NUM_MODE_VECTOR_INT): Likewise.
13751 * genoutput.c (null_operand): Likewise.
13752 (output_operand_data): Likewise.
13753 * genrecog.c (print_parameter_value): Likewise.
13754 * lra.c (debug_operand_data): Likewise.
13756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13758 * dwarf2out.c (scompare_loc_descriptor_wide)
13759 (scompare_loc_descriptor_narrow): New functions, split out from...
13760 (scompare_loc_descriptor): ...here.
13761 * expmed.c (emit_store_flag_int): New function, split out from...
13762 (emit_store_flag): ...here.
13764 2017-08-30 Richard Biener <rguenther@suse.de>
13766 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13767 (dwarf2out_early_finish): Move setting of AT_pubnames from
13768 early debug output to early finish.
13770 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13772 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13773 and -mdata-region to the assembler.
13775 2017-08-30 Richard Biener <rguenther@suse.de>
13777 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13779 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13780 (dwarf2out_early_global_decl): Only generate a DIE for the
13781 abstract origin if it doesn't already exist or is a declaration DIE.
13782 (resolve_addr): Do not add the linkage name twice when
13783 generating a stub DIE for the DW_TAG_GNU_call_site target.
13785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13787 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13788 Use machine_mode rather than int for arg1_mode.
13790 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13793 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13794 that the second argument of the built-in functions to unpack
13795 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13796 switch statement instead a lot of if statements.
13797 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13798 Allow 64-bit values to be in Altivec registers as well as
13799 traditional floating point registers.
13800 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13802 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13804 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13805 MAX_REGS_PER_ADDRESS == 1.
13807 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13809 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13810 * config/i386/i386.c (ix86_option_override_internal): Simplify
13811 setting of opts->x_flag_entry.
13813 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13814 Jakub Jelinek <jakub@redhat.com>
13815 Richard Biener <rguenther@suse.de>
13817 PR tree-optimization/81503
13818 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13819 folded constant fits in the target type; reorder tests for clarity.
13821 2017-08-29 Martin Liska <mliska@suse.cz>
13823 * passes.def: Include pass_lower_switch.
13824 * stmt.c (dump_case_nodes): Remove and move to
13825 tree-switch-conversion.
13826 (case_values_threshold): Likewise.
13827 (expand_switch_as_decision_tree_p): Likewise.
13828 (emit_case_decision_tree): Likewise.
13829 (expand_case): Likewise.
13830 (balance_case_nodes): Likewise.
13831 (node_has_low_bound): Likewise.
13832 (node_has_high_bound): Likewise.
13833 (node_is_bounded): Likewise.
13834 (emit_case_nodes): Likewise.
13835 (struct simple_case_node): New struct.
13836 (add_case_node): Remove.
13837 (emit_case_dispatch_table): Use vector instead of case_list.
13838 (reset_out_edges_aux): Remove.
13839 (compute_cases_per_edge): Likewise.
13840 (expand_case): Build list of simple_case_node.
13841 (expand_sjlj_dispatch_table): Use it.
13842 * tree-switch-conversion.c (struct case_node): Moved from
13843 stmt.c and adjusted.
13844 (emit_case_nodes): Likewise.
13845 (node_has_low_bound): Likewise.
13846 (node_has_high_bound): Likewise.
13847 (node_is_bounded): Likewise.
13848 (case_values_threshold): Likewise.
13849 (reset_out_edges_aux): Likewise.
13850 (compute_cases_per_edge): Likewise.
13851 (add_case_node): Likewise.
13852 (dump_case_nodes): Likewise.
13853 (balance_case_nodes): Likewise.
13854 (expand_switch_as_decision_tree_p): Likewise.
13855 (emit_jump): Likewise.
13856 (emit_case_decision_tree): Likewise.
13857 (try_switch_expansion): Likewise.
13858 (do_jump_if_equal): Likewise.
13859 (emit_cmp_and_jump_insns): Likewise.
13860 (fix_phi_operands_for_edge): New function.
13861 (record_phi_operand_mapping): Likewise.
13862 (class pass_lower_switch): New pass.
13863 (pass_lower_switch::execute): New function.
13864 (make_pass_lower_switch): Likewise.
13865 (conditional_probability):
13866 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13867 * tree-pass.h: Add make_pass_lower_switch.
13869 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13872 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13875 2017-08-29 Richard Biener <rguenther@suse.de>
13877 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13878 we do not add a DW_AT_inline attribute twice.
13879 (gen_subprogram_die): Remove code setting DW_AT_inline on
13880 DECL_ABSTRACT_P nodes.
13882 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13884 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13885 calls to internal functions.
13886 (gimplify_modify_expr): Likewise.
13887 * tree-call-cdce.c (use_internal_fn): Likewise.
13888 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13889 (convert_to_divmod): Set the nothrow flag.
13890 * tree-if-conv.c (predicate_mem_writes): Likewise.
13891 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13892 (vectorizable_call): Likewise.
13893 (vectorizable_store): Likewise.
13894 (vectorizable_load): Likewise.
13895 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13896 (vect_recog_mask_conversion_pattern): Likewise.
13898 2017-08-29 Martin Liska <mliska@suse.cz>
13901 * gcc.c (driver_handle_option): Add new argument.
13902 * opts-common.c (handle_option): Pass
13903 target_option_override_hook.
13904 * opts-global.c (lang_handle_option): Add new option.
13905 (set_default_handlers): Add new argument.
13906 (decode_options): Likewise.
13907 * opts.c (target_handle_option): Likewise.
13908 (common_handle_option): Call target_option_override_hook.
13909 * opts.h (struct cl_option_handler_func): Add hook for
13910 target option override.
13911 (struct cl_option_handlers): Likewise.
13912 (set_default_handlers): Add new argument.
13913 (decode_options): Likewise.
13914 (common_handle_option): Likewise.
13915 (target_handle_option): Likewise.
13916 * toplev.c (toplev::main): Pass targetm.target_option.override
13919 2017-08-29 Richard Biener <rguenther@suse.de>
13920 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13922 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13923 life to the active subtree.
13925 2017-08-28 Jeff Law <law@redhat.com>
13927 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13928 derive_equivalences.
13929 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13930 Code moved into....
13931 (edge_info::derive_equivalences): New private member function
13933 * tree-ssa-dom.c (class edge_info): Changed from a struct
13934 to a class. Add ctor/dtor, methods and data members.
13935 (edge_info::edge_info): Renamed from allocate_edge_info.
13936 Initialize additional members.
13937 (edge_info::~edge_info): New.
13938 (free_dom_edge_info): Delete the edge info.
13939 (record_edge_info): Use new class & associated member functions.
13940 Tighten forms for testing for edge equivalences.
13941 (record_temporary_equivalences): Iterate over the simple
13942 equivalences rather than assuming there's only one per edge.
13943 (cprop_into_successor_phis): Iterate over the simple
13944 equivalences rather than assuming there's only one per edge.
13945 (optimize_stmt): Use operand_equal_p rather than pointer
13946 equality for mini-DSE code.
13948 2017-08-28 Nathan Sidwell <nathan@acm.org>
13950 * gcc.c (execute): Fold SIGPIPE handling into switch
13951 statement. Adjust internal error message.
13953 2017-08-28 Richard Biener <rguenther@suse.de>
13956 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13957 Do nothing for removed DIEs.
13959 2017-08-28 Richard Biener <rguenther@suse.de>
13961 PR tree-optimization/81977
13962 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13965 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13968 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13969 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13970 using targetm.gen_mem_thread_fence.
13972 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13975 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13976 predicate to register_operand. Reorder operands.
13977 (*btr<mode>): Ditto.
13978 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13979 (*btr<mode>_mask): Ditto.
13981 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13983 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13984 * config/rs6000/xmmintrin.h: New file.
13985 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13987 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13990 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13991 parameter and_insn and return it.
13992 (recombine_lvx_pattern): Insert a copy to ensure availability of
13993 the base register of the copied masking operation at the point of
13994 the instruction replacement.
13995 (recombine_stvx_pattern): Likewise.
13997 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13999 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14000 undocumented switches.
14001 (-mpower9-dform-vector): Likewise.
14002 (-mpower9-dform): Likewise.
14003 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14004 comments to delete references to -mpower9-dform* switches.
14005 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14006 Delete reference to -mpower9-dform* switches, test for
14007 -mpower9-vector instead.
14008 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14009 (OTHER_P9_VECTOR_MASKS): Likewise.
14010 (POWERPC_MASKS): Likewise.
14011 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14012 tests against -mpower9-dform* to -mpower9-vector. Delete code
14013 that checked for -mpower9-dform* consistancy with other options.
14014 Add test for -mpower9-misc to enable other power9 switches.
14015 (rs6000_init_hard_regno_mode_ok): Likewise.
14016 (rs6000_option_override_internal): Likewise.
14017 (rs6000_emit_prologue): Likewise.
14018 (rs6000_emit_epilogue): Likewise.
14019 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14020 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14021 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14023 (emit_fusion_p9_store): Likewise.
14024 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14025 resetting these macros if the assembler does not support ISA 3.0
14027 (TARGET_P9_DFORM_VECTOR): Likewise.
14028 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14029 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14031 2017-08-25 Alan Modra <amodra@gmail.com>
14033 PR rtl-optimization/81747
14034 * cse.c (cse_extended_basic_block): Don't attempt to record
14035 equivalences for degenerate conditional branches that branch
14036 to their fall-through.
14038 2017-08-24 Martin Sebor <msebor@redhat.com>
14040 PR middle-end/81908
14041 * gimple-fold.c (size_must_be_zero_p): New function.
14042 (gimple_fold_builtin_memory_op): Call it.
14044 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14046 * config/rs6000/mm_malloc.h: New file.
14048 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14050 PR tree-optimization/81913
14051 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14052 analysis when either IVs in condition can wrap.
14054 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14056 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14057 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14059 2017-08-24 Richard Biener <rguenther@suse.de>
14062 * targhooks.c (default_target_can_inline_p): Properly
14063 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14064 is present and always compare.
14065 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14066 infer -mfpmath=sse from TARGET_SSE_P.
14067 (ix86_can_inline_p): Properly use target_option_default_node when
14068 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14070 2017-08-24 Richard Biener <rguenther@suse.de>
14073 * dwarf2out.c (output_die): Handle flag_generate_offload like
14075 (output_comp_unit): Likewise.
14076 (gen_array_type_die): Likewise.
14077 (dwarf2out_early_finish): Likewise.
14078 (note_variable_value_in_expr): Likewise.
14079 (dwarf2out_finish): Likewise. Adjust assert.
14080 * cgraphunit.c (symbol_table::compile): Move setting of
14081 flag_generate_offload earlier ...
14082 (symbol_table::finalize_compilation_unit): ... here, before
14083 early debug finalization.
14085 2017-08-24 Richard Biener <rguenther@suse.de>
14087 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14088 and ipa-fnsummary.h.
14089 (ix86_can_inline_p): When ix86_fpmath flags do not match
14090 check whether the callee uses FP math at all.
14092 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14094 PR middle-end/81931
14095 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14096 instead of TYPE_PRECISION.
14098 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14099 Alan Hayward <alan.hayward@arm.com>
14100 David Sherwood <david.sherwood@arm.com>
14102 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14104 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14106 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14108 (ix86_frame::stack_realign_allocate): New field.
14109 (struct machine_frame_state): Modify comments.
14110 (machine_frame_state::sp_realigned_fp_end): New field.
14111 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14112 layout calculation.
14113 (sp_valid_at): Add assertion to assure no attempt to access invalid
14114 offset of a realigned stack.
14115 (fp_valid_at): Likewise.
14116 (choose_baseaddr): Modify comments.
14117 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14118 ix86_expand_prologue.
14119 (ix86_expand_prologue): Modify stack realignment and allocation.
14120 (ix86_expand_epilogue): Modify comments.
14121 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14122 avx2_runtime, avx512f, and avx512f_runtime.
14124 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14126 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14127 (mstackrealign): Do not init to -1.
14128 * config/i386/i386.c (ix86_option_override_internal):
14129 Check opts_set, not opts when setting default value of
14130 opts->x_ix86_force_align_arg_pointer.
14132 2017-08-23 Richard Biener <rguenther@suse.de>
14134 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14135 lang_hooks.decl_printable_name.
14136 * print-rtl-function.c (print_rtx_function): Likewise.
14137 * tree-pretty-print.c (dump_function_header): Likewise.
14139 2017-08-23 Richard Biener <rguenther@suse.de>
14142 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14143 -g0 at compile-time.
14145 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14147 PR middle-end/19706
14148 * doc/sourcebuild.texi (Other hardware attributes):
14151 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14153 PR middle-end/19706
14154 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14155 Fix single-use check.
14157 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14159 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14161 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14163 * doc/install.texi: Modify to add more details on running selected
14166 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14168 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14169 is combined with -mabi=ms.
14170 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14173 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14175 PR tree-optimization/81488
14176 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14177 and cached_basis fields.
14178 (MAX_SPREAD): New constant.
14179 (alloc_cand_and_find_basis): Initialize new fields.
14180 (clear_visited): New function.
14181 (create_phi_basis_1): Rename from create_phi_basis, set visited
14182 and cached_basis fields.
14183 (create_phi_basis): New wrapper function.
14184 (phi_add_costs_1): Rename from phi_add_costs, add spread
14185 parameter, set visited field, short-circuit when limits reached.
14186 (phi_add_costs): New wrapper function.
14187 (record_phi_increments_1): Rename from record_phi_increments, set
14189 (record_phi_increments): New wrapper function.
14190 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14191 (phi_incr_cost): New wrapper function.
14192 (all_phi_incrs_profitable_1): Rename from
14193 all_phi_incrs_profitable, set visited field.
14194 (all_phi_incrs_profitable): New wrapper function.
14196 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14197 Alan Hayward <alan.hayward@arm.com>
14198 David Sherwood <david.sherwood@arm.com>
14200 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14201 that takes the outer and inner modes.
14202 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14203 comparison as the canonical test for a paradoxical subreg.
14204 * combine.c (simplify_set): Use paradoxical_subreg_p.
14205 (make_extraction): Likewise.
14206 (force_to_mode): Likewise.
14207 (rtx_equal_for_field_assignment_p): Likewise.
14208 (gen_lowpart_for_combine): Likewise.
14209 (simplify_comparison): Likewise.
14210 * cse.c (equiv_constant): Likewise.
14211 * expmed.c (store_bit_field_1): Likewise.
14212 * final.c (alter_subreg): Likewise.
14213 * fwprop.c (propagate_rtx): Likewise.
14214 (forward_propagate_subreg): Likewise.
14215 * ira-conflicts.c (ira_build_conflicts): Likewise.
14216 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14217 * lra-constraints.c (curr_insn_transform): Likewise.
14218 (split_reg): Likewise.
14219 * lra-eliminations.c (move_plus_up): Likewise.
14220 (lra_eliminate_regs_1): Likewise.
14221 * recog.c (general_operand): Likewise.
14222 * ree.c (combine_reaching_defs): Likewise.
14223 * reload.c (push_reload): Likewise.
14224 (find_reloads): Likewise.
14225 * reload1.c (elimination_effects): Likewise.
14226 (compute_reload_subreg_offset): Likewise.
14227 (choose_reload_regs): Likewise.
14228 * rtlanal.c (subreg_lsb_1): Likewise.
14229 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14230 (simplify_subreg): Likewise.
14231 * var-tracking.c (track_loc_p): Likewise.
14232 * emit-rtl.c (byte_lowpart_offset): Likewise.
14233 (paradoxical_subreg_p): Delete out-of-line definition.
14235 2017-08-22 Jeff Law <law@redhat.com>
14237 PR tree-optimization/81741
14238 PR tree-optimization/71947
14239 * tree-ssa-dom.c: Include tree-inline.h.
14240 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14241 equivalences if one is more expensive to compute than the other.
14242 * tree-ssa-scopedtables.h (class const_or_copies): Make
14243 record_const_or_copy_raw method private.
14244 (class avail_exprs_stack): New method simplify_binary_operation.
14245 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14246 avail_exprs_stack::simplify_binary_operation as needed.
14247 (avail_exprs_stack::simplify_binary_operation): New function.
14249 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14251 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14252 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14253 (DOT_SYMBOLS): Likewise.
14254 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14255 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14256 (RS6000_ABI_NAME): Likewise.
14257 (TARGET_CMODEL): Likewise.
14258 (TOC_SECTION_ASM_OP): Likewise.
14259 (SET_CMODEL): New macro.
14260 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14262 2017-08-22 Richard Biener <rguenther@suse.de>
14264 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14265 to allow for free-lang-data replacements similar to verify_type_variant.
14267 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14268 Alan Hayward <alan.hayward@arm.com>
14269 David Sherwood <david.sherwood@arm.com>
14271 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14272 VOIDmode for the LABEL_REF.
14274 2017-08-22 Richard Biener <rguenther@suse.de>
14276 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14278 2017-08-22 Richard Biener <rguenther@suse.de>
14280 * common.opt (feliminate-dwarf2-dups): Ignore.
14281 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14282 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14283 same_die_p_wrap, compute_section_prefix,
14284 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14285 (comdat_symbol_id, comdat_symbol_number): Likewise.
14286 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14288 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14289 (output_die): Mark unreachable path unreachable.
14290 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14291 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14292 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14293 (dwarf2out_early_finish): Likewise.
14295 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14297 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14299 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14302 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14303 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14304 (avr_attribute_table) <io, io_low, address>: Initialize
14305 .decl_required with true.
14307 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14309 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14310 undocumented debugging options.
14311 (-mvsx-scalar-double): Likewise.
14312 (-mallow-df-permute): Likewise.
14313 (-mvectorize-builtins): Likewise.
14314 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14315 (rs6000_builtin_vectorized_function): Likewise.
14316 (rs6000_builtin_md_vectorized_function): Likewise.
14317 (rs6000_opt_vars): Likewise.
14319 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14322 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14323 (*btrq_imm): Rename from *btrq.
14324 (*btcq_imm): Rename from *btcq.
14325 (btsc): New code attribute.
14326 (*<btsc><mode>): New insn pattern.
14327 (*btr<mode>): Ditto.
14328 (*<btsc><mode>_mask): New insn_and_split pattern.
14329 (*btr<mode>_mask): Ditto.
14331 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14332 Alan Hayward <alan.hayward@arm.com>
14333 David Sherwood <david.sherwood@arm.com>
14335 * function.c (pad_below): Simplify padding calculation.
14337 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14338 Alan Hayward <alan.hayward@arm.com>
14339 David Sherwood <david.sherwood@arm.com>
14341 * target.def (function_prologue): Remove frame size argument.
14342 (function_epilogue): Likewise.
14343 * doc/tm.texi: Regenerate.
14344 * final.c (final_start_function): Update call to function_prologue.
14345 (final_end_function): Update call to function_epilogue.
14346 (default_function_pro_epilogue): Remove frame size argument.
14347 * output.h (default_function_pro_epilogue): Likewise.
14348 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14349 (arm_output_function_prologue): Likewise.
14350 * config/frv/frv.c (frv_function_prologue): Likewise.
14351 (frv_function_epilogue): Likewise.
14352 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14353 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14354 (ia64_output_function_epilogue): Likewise.
14355 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14356 (m32r_output_function_epilogue): Likewise.
14357 * config/microblaze/microblaze.c (microblaze_function_prologue)
14358 (microblaze_function_epilogue): Likewise.
14359 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14360 (mips_output_function_epilogue): Likewise.
14361 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14362 (mmix_target_asm_function_epilogue): Likewise.
14363 * config/msp430/msp430.c (msp430_start_function): Likewise.
14364 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14365 (nds32_asm_function_epilogue): Likewise.
14366 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14367 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14368 (pa_output_function_epilogue): Likewise.
14369 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14370 (rs6000_output_function_epilogue): Likewise.
14371 * config/rl78/rl78.c (rl78_start_function): Likewise.
14372 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14373 (rs6000_output_function_epilogue): Likewise.
14374 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14375 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14376 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14377 (sparc_asm_function_epilogue): Likewise.
14379 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14381 * tree.h (type_has_mode_precision_p): New function.
14382 * convert.c (convert_to_integer_1): Use it.
14383 * expr.c (expand_expr_real_2): Likewise.
14384 (expand_expr_real_1): Likewise.
14385 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14386 * match.pd: Likewise.
14387 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14388 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14389 * tree-tailcall.c (process_assignment): Likewise.
14390 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14391 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14392 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14393 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14394 (vectorizable_assignment): Likewise.
14395 (vectorizable_shift): Likewise.
14396 (vectorizable_operation): Likewise.
14397 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14399 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14401 * match.pd: Add pow (C, x) simplification.
14403 2017-08-21 Richard Biener <rguenther@suse.de>
14405 PR tree-optimization/81900
14406 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14407 for blocks with abnormal predecessors.
14408 (compute_antic): Do not set visited flag prematurely.
14410 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14413 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14415 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14417 * stor-layout.h (vector_type_mode): Move to...
14418 * tree.h (vector_type_mode): ...here.
14419 * stor-layout.c (vector_type_mode): Move to...
14420 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14422 2017-08-21 Richard Biener <rguenther@suse.de>
14424 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14425 register_external_die hooks.
14426 (debug_false_tree_charstarstar_uhwistar): Declare.
14427 (debug_nothing_tree_charstar_uhwi): Likewise.
14428 * debug.c (do_nothing_debug_hooks): Adjust.
14429 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14430 (debug_nothing_tree_charstar_uhwi): Likewise.
14431 * dbxout.c (dbx_debug_hooks): Adjust.
14432 (xcoff_debug_hooks): Likewise.
14433 * sdbout.c (sdb_debug_hooks): Likewise.
14434 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14435 * dwarf2out.c (macinfo_label_base): New global.
14436 (dwarf2out_register_external_die): New function for the
14437 register_external_die hook.
14438 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14439 (dwarf2_debug_hooks): Use them.
14440 (dwarf2_lineno_debug_hooks): Adjust.
14441 (struct die_struct): Add with_offset flag.
14442 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14443 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14444 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14445 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14446 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14447 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14448 defining section names for the early LTO debug variants.
14449 (reset_indirect_string): New helper.
14450 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14451 (print_dw_val): Add support for offsetted symbol references.
14452 (get_ultimate_context): Split out from is_cxx.
14453 (is_cxx): Use get_ultimate_context.
14454 (is_fortran): Add decl overload.
14455 (compute_comp_unit_symbol): Split out worker from
14456 compute_section_prefix.
14457 (compute_section_prefix): Call compute_comp_unit_symbol and
14458 set comdat_type_p here.
14459 (output_die): Skip DIE symbol output for the LTO added one.
14460 Handle DIE symbol references with offset.
14461 (output_comp_unit): Guard section name mangling properly.
14462 For LTO debug sections emit a symbol at the section beginning
14463 which we use to refer to its DIEs.
14464 (add_abstract_origin_attribute): For DIEs registered via
14465 dwarf2out_register_external_die directly refer to the early
14466 DIE rather than indirectly through the shadow one we created.
14467 Remove obsolete call to dwarf2out_abstract_function for
14468 non-function/block origins.
14469 (gen_array_type_die): When generating early LTO debug do
14470 not emit DW_AT_string_length.
14471 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14472 late when in LTO. As suggested place a gcc_unreachable for
14473 the DECL_ABSTRACT_P case.
14474 (gen_subprogram_die): Avoid another specification DIE
14475 for early built declarations/definitions for the late LTO case.
14476 (gen_variable_die): Add type references for late duplicated VLA dies
14478 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14479 we have the abstract instance already.
14480 (process_scope_var): Adjust decl DIE contexts in LTO which
14481 first puts them in limbo.
14482 (gen_decl_die): Do not generate type DIEs late apart from
14483 types for VLAs or for decls we do not yet have a DIE. Do not
14484 call dwarf2out_abstract_function late.
14485 (dwarf2out_early_global_decl): Make sure to create DIEs
14486 for abstract instances of a decl first.
14487 (dwarf2out_late_global_decl): Adjust comment.
14488 (output_macinfo_op): With multiple macro sections use
14489 macinfo_label_base to distinguish labels.
14490 (output_macinfo): Likewise. Update macinfo_label_base.
14491 Pass in the line info label.
14492 (note_variable_value_in_expr): When generating LTO resolve
14493 all variable values here by generating DIEs as needed.
14494 (init_sections_and_labels): Add early LTO debug flag parameter
14495 and generate different sections and names if set. Add generation
14496 counter for the labels so we can have multiple of them.
14497 (reset_dies): Helper to allow DIEs to be output multiple times.
14498 (dwarf2out_finish): When outputting DIEs to the fat part of an
14499 LTO object first reset DIEs.
14500 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14501 (modified_type_die): Check for decl_ultimate_origin being self
14503 (gen_type_die_with_usage): Likewise.
14504 (gen_typedef_die): Allow decl_ultimate_origin being self.
14505 (set_decl_abstract_flags): Remove.
14506 (set_block_abstract_flags): Likewise.
14507 (dwarf2out_abstract_function): Treat the early generated DIEs
14508 as the abstract copy and only add DW_AT_inline and
14509 DW_AT_artificial here and call set_decl_origin_self.
14510 If the DIE has an abstract origin don't do anything.
14511 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14512 if we have none yet (Go fails to build one, PR78628).
14513 (variably_modified_type_p): Prevent endless recursion for Ada
14514 cyclic pointer types.
14515 * lto-streamer-in.c: Include debug.h.
14516 (dref_queue): New global.
14517 (lto_read_tree_1): Stream in DIE references.
14518 (lto_input_tree): Register DIE references.
14519 (input_function): Stream DECL_DEBUG_ARGS.
14520 * lto-streamer-out.c: Include debug.h.
14521 (lto_write_tree_1): Output DIE references.
14522 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14523 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14524 (output_function): Stream DECL_DEBUG_ARGS.
14525 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14526 Stream DECL_ABSTRACT_ORIGIN.
14527 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14528 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14529 DECL_CONTEXT for file-scope decls.
14530 * lto-streamer.h (struct dref_entry): Declare.
14531 (dref_queue): Likewise.
14532 * cfgexpand.c (pass_expand::execute): Do not call the
14533 outlining_inline_function hook here.
14534 * lto-wrapper.c (debug_obj): New global.
14535 (tool_cleanup): Unlink it if required.
14536 (debug_objcopy): New function.
14537 (run_gcc): Handle early debug sections in the IL files by
14538 extracting them to separate files, partially linkin them and
14539 feeding the result back as result to the linker.
14540 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14541 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14542 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14543 sections into a separate segment.
14544 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14546 (darwin_asm_dwarf_section): Likewise.
14547 (darwin_asm_output_dwarf_offset): Likewise.
14548 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14550 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14551 Alan Hayward <alan.hayward@arm.com>
14552 David Sherwood <david.sherwood@arm.com>
14554 * read-md.h (md_reader::record_potential_iterator_use): Replace
14555 pointer argument with an rtx and an index.
14556 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14557 (apply_mode_iterator): Likewise.
14558 (apply_code_iterator): Likewise.
14559 (apply_int_iterator): Likewise.
14560 (apply_subst_iterator): Likewise.
14561 (record_iterator_use): Likewise.
14562 (record_attribute_use): Likewise.
14563 (md_reader::record_potential_iterator_use): Likewise. Update calls
14564 to record_iterator_use and apply_iterator.
14565 (iterator_use): Replace ptr with x and index.
14566 (attribute_use): Likewise.
14567 (apply_attribute_uses): Update calls to apply_iterator.
14568 (apply_iterators): Likewise. Update initialization of iterator_use.
14569 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14570 and record_potential_iterator_use.
14571 (rtx_reader::read_rtx_operand): Likewise.
14573 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14574 Alan Hayward <alan.hayward@arm.com>
14575 David Sherwood <david.sherwood@arm.com>
14577 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14580 2017-08-21 Richard Biener <rguenther@suse.de>
14582 PR middle-end/81884
14583 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14584 at struct end conservatively when comparing common bases.
14586 2017-08-21 Richard Biener <rguenther@suse.de>
14588 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14589 (mem_ref_in_stmt): Remove.
14590 (determine_max_movement): Use ref index to get at the reference.
14591 (invariantness_dom_walker::before_dom_children): Deal with
14592 lim data already initialized.
14593 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14595 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14597 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14598 (TARGET_ISA_ROUND): Ditto.
14599 (TARGET_ROUND): Ditto.
14600 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14601 * config/i386/i386.md: Ditto.
14602 * config/i386/sse.md: Ditto.
14603 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14604 with OPTION_MASK_ISA_SSE4_1.
14606 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14609 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14610 __builtin_ia32_lzcnt_u16.
14612 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14615 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14616 (rs6000_set_current_function): Rewrite function to use it.
14618 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14621 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14622 and TYPE_WARN_IF_NOT_ALIGN.
14623 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14624 (handle_warn_if_not_align): New.
14625 (place_union_field): Call handle_warn_if_not_align.
14626 (place_field): Call handle_warn_if_not_align.
14627 Copy TYPE_WARN_IF_NOT_ALIGN.
14628 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14629 (layout_type): Likewise.
14630 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14632 (tree_decl_common): Add warn_if_not_align.
14633 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14634 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14635 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14636 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14637 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14638 * doc/extend.texi: Document warn_if_not_aligned attribute.
14639 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14641 2017-08-17 Martin Liska <mliska@suse.cz>
14644 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14645 (get_data_dependence): Use it as pointer type.
14646 (distribute_loop): Likewise.
14648 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14650 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14651 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14652 (p8_vmrgow_<mode>_direct): New define_insn.
14653 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14654 handle endianness for vmrgew and vmrgow permute patterns.
14656 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14658 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14659 * config/rs6000/rs6000-cpus.def: Remove comment.
14660 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14661 (POWERPC_MASKS): Likewise.
14662 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14663 use of TARGET_VSX_TIMODE.
14664 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14665 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14666 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14667 (rs6000_option_override_internal): Remove dead code.
14668 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14669 (rs6000_legitimize_reload_address): Likewise.
14670 (rs6000_legitimate_address_p): Likewise.
14671 (rs6000_opt_masks): Delete "vsx-timode".
14672 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14673 from function comment.
14674 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14675 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14676 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14678 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14679 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14680 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14681 (VSX_TI): Likewise.
14683 (define_peephole2): Likewise.
14685 2017-08-17 Martin Sebor <msebor@redhat.com>
14688 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14689 past the end of an array.
14690 (test_pp_format): Add test cases.
14692 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14694 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14695 missing ECF_NOTHROW flags.
14697 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14700 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14701 operands residing in integer registers.
14702 (*vsx_le_perm_load_<mode>): Likewise.
14703 (*vsx_le_perm_store_<mode>): Likewise.
14704 (define_peephole2): Add peepholes to optimize the above.
14706 2017-08-17 Marek Polacek <polacek@redhat.com>
14708 PR middle-end/81814
14709 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14710 to mimic what shorten_compare did. Change the return type to bool.
14711 (fold_cond_expr_with_comparison): Update call to
14712 operand_equal_for_comparison_p.
14713 (fold_ternary_loc): Likewise.
14715 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14717 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14719 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14721 2017-08-17 Richard Biener <rguenther@suse.de>
14723 * tree-ssa-structalias.c (solve_graph): When propagating
14724 to successors update the graphs succ edges and avoid duplicate work.
14726 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14729 * config/i386/i386.c (ix86_option_override_internal): Save target
14730 specific options after ix86_stack_protector_guard_reg was changed.
14732 2017-08-17 Richard Biener <rguenther@suse.de>
14734 PR tree-optimization/81827
14735 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14736 (new_var_info): Initialize it conservatively.
14737 (get_call_vi): Mark register vars.
14738 (new_scalar_tmp_constraint_exp): Likewise.
14739 (handle_rhs_call): Likewise.
14740 (handle_const_call): Likewise.
14741 (create_function_info_for): Likewise.
14742 (solve_constraints): Sort varinfos to separate register from
14743 non-register vars to pack points-to solution bitmaps during
14746 2017-08-17 Marek Polacek <polacek@redhat.com>
14748 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14750 2017-08-17 Richard Biener <rguenther@suse.de>
14752 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14753 to true when overflow is undefined and we saturated the result.
14755 2017-08-17 Alan Modra <amodra@gmail.com>
14758 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14759 Don't use store multiple if only one reg needs saving.
14760 (interesting_frame_related_regno): New function.
14761 (rs6000_frame_related): Don't emit frame info for regs that
14763 (rs6000_emit_epilogue): Likewise.
14765 2017-08-16 Nathan Sidwell <nathan@acm.org>
14767 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14768 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14769 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14770 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14771 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14772 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14773 (verify_type): Adjust for TYPE_BINFO move.
14774 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14775 process TYPE_BINFO directly.
14776 (hash_tree): Likewise.
14777 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14779 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14782 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14784 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14786 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14789 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14790 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14791 Add ix86_binary_operator_ok to insn constraint.
14792 (*iordi_1_bts): Ditto.
14793 (*xordi_1_btc): Ditto.
14794 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14795 Update corresponding peephole2 pattern.
14799 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14801 PR tree-optimization/81832
14802 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14803 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14805 2017-08-16 Marek Polacek <polacek@redhat.com>
14808 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14809 regarding size_zero_node.
14811 2017-08-16 Martin Liska <mliska@suse.cz>
14814 * config.gcc: Respect previously set extra_objs in case
14817 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14819 PR tree-optimization/81835
14820 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14821 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14822 not depend on the phi.
14824 2017-08-16 Alan Modra <amodra@gmail.com>
14826 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14829 2017-08-16 Alan Modra <amodra@gmail.com>
14831 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14832 (save_reg_p): ..into this. Update all callers.
14833 (first_reg_to_save): Simplify.
14835 2017-08-16 Alan Modra <amodra@gmail.com>
14837 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14840 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14844 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14845 constant count if that count is less than 32.
14847 2017-08-15 Nathan Sidwell <nathan@acm.org>
14849 * gcc.c (execute): Emit friendlier message if inferior is killed
14850 by an external cause.
14852 2017-08-15 Richard Biener <rguenther@suse.de>
14854 PR tree-optimization/81790
14855 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14856 CONSTRUCTORs from simplifying and VN.
14858 2017-08-14 Martin Sebor <msebor@redhat.com>
14860 * builtin-attrs.def: Add comments.
14862 2017-08-14 Martin Sebor <msebor@redhat.com>
14865 * doc/extend.texi (attribute nonstring): Document new attribute.
14867 2017-08-14 Martin Sebor <msebor@redhat.com>
14870 * tree-diagnostic.c (default_tree_printer): Handle %G.
14871 * gimple-pretty-print.h (percent_G_format): Declare new function.
14872 * gimple-pretty-print.c (percent_G_format): Define.
14873 * tree-pretty-print.c (percent_K_format): Add argument.
14875 2017-08-14 Martin Sebor <msebor@redhat.com>
14877 PR translation/79998
14878 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14879 Remove a stray space.
14881 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14884 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14885 (*iordi_1_bts): Ditto.
14886 (*xordi_1_btc): Ditto.
14888 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14891 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14892 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14894 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14895 quoted strings, and make more translator-friendly.
14896 (darwin_rs6000_override_options): Likewise.
14897 (rs6000_option_override_internal): Likewise.
14898 (rs6000_return_in_memory): Fix overlong line.
14899 (init_cmulative_args): Use quoted strings, and make more
14900 translator-friendly.
14901 (rs6000_pass_by_reference): Fix overlong line.
14902 (def_builtin): Use quoted strings.
14903 (altivec_expand_predicate_builtin): Use quoted strings, and make
14904 more translator-friendly.
14905 (htm_expand_builtin): Use quoted strings.
14906 (cpu_expand_builtin): Use quoted strings, and make more
14907 translator-friendly.
14908 (altivec_expand_builtin): Likewise.
14909 (paired_expand_predicate_builtin): Likewise.
14910 (rs6000_invalid_builtin): Likewise.
14911 (builtin_function_type): Use quoted strings.
14912 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14913 more translator-friendly.
14914 (rs6000_trampoline_init): Likewise.
14915 (rs6000_handle_altivec_attribute): Likewise.
14916 (rs6000_inner_target_options): Use quoted strings.
14917 (rs6000_disable_incompatible_switches): Likewise.
14918 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14919 strings, and make more translator-friendly.
14920 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14922 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14924 PR tree-optimization/81799
14925 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14926 cond_expr to simple gimple operand.
14928 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14930 PR middle-end/46932
14931 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14933 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14937 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14938 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14939 TARGET_GASISR_PROLOGUES.
14940 * config/avr/avr.c (avr_option_override): Same.
14941 (avr_pass_pre_proep::execute): Same.
14943 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14946 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14947 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14949 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14951 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14952 define_insn_and_split. Split to a memory load from 0 in
14953 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14954 using PTR mode iterator.
14955 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14956 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14957 (*add_tp_<mode>): Redefine as define_insn_and_split.
14958 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14959 address space. Merge with *add_tp_x32 using PTR mode iterator.
14960 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14961 Split to an add with a memory load from 0 in
14962 DEFAULT_TLS_SEG_REG address space.
14964 2017-08-12 Andrew Pinski <apinski@cavium.com>
14966 * config/aarch64/aarch64-option-extensions.def (rdma):
14967 Fix feature string to what Linux prints out in /proc/cpuinfo.
14969 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14972 * dwarf2out.c (modified_type_die): For C typedef types that have
14973 an ultimate origin, process the ultimate origin instead of the
14975 (gen_typedef_die): Assert that input DECLs have no ultimate
14977 (gen_type_die_with_usage): For typedef variants that have an
14978 ultimate origin, just call gen_decl_die on the original DECL.
14979 (process_scope_var): Avoid creating DIEs for local typedefs and
14980 concrete static variables.
14982 2017-08-12 Alan Modra <amodra@gmail.com>
14986 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14987 match gnu-user.h startfile.
14988 (ENDFILE_LINUX_SPEC): Similarly.
14990 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14993 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14995 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14997 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14998 * config/aarch64/aarch64.md (mov<mode>): Change.
14999 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15000 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15001 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15003 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15005 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15006 for storage order barriers.
15008 2017-08-11 Martin Liska <mliska@suse.cz>
15011 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15012 variables of void type.
15014 2017-08-11 Martin Liska <mliska@suse.cz>
15016 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15017 TARGET_SUPPORTS_ALIASES.
15018 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15019 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15020 (optimize_weakref): Likewise.
15021 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15022 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15023 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15025 2017-08-11 Martin Liska <mliska@suse.cz>
15028 * config/i386/i386.c (make_resolver_func): Do complete
15029 refactoring of the function.
15031 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15034 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15035 * config/i386/i386.c (ix86_stack_protect_guard): Use
15036 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15037 * doc/invoke.texi (x86 Options): Document
15038 -mstack-protector-guard-symbol= option.
15040 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15042 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15043 * config/i386/i386.c (ix86_split_stack_guard): New function.
15044 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15045 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15046 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15047 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15048 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15049 (split_stack_space_check): Call ix86_split_stack_guard.
15051 2017-08-10 Martin Sebor <msebor@redhat.com>
15053 * print-tree.c (print_node): Print location using the established
15055 Replace spaces with colons.
15056 (debug_raw, debug): Ditto.
15058 2017-08-10 Martin Sebor <msebor@redhat.com>
15061 * pretty-print.c (pp_format): Correct the handling of %s precision.
15063 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15066 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15068 (ix86_finalize_stack_frame_flags): This. Also clear
15069 frame_pointer_needed if -fno-omit-frame-pointer is used without
15071 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15072 with ix86_finalize_stack_frame_flags.
15073 (ix86_expand_epilogue): Likewise.
15074 (ix86_expand_split_stack_prologue): Likewise.
15075 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15077 2017-08-10 Martin Liska <mliska@suse.cz>
15080 * c-attribs.c (handle_target_attribute):
15081 Report warning for an empty string argument of target attribute.
15083 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15086 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15088 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15089 or DECL_NONLOCAL labels.
15090 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15091 or DECL_NONLOCAL labels here.
15093 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15095 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15096 to indicate when early gimple folding has been disabled.
15097 (rs6000_gimple_fold_builtin): Add debug content.
15098 (rs6000_invalid_builtin): Fix whitespace.
15099 (rs6000_expand_builtin): Fix whitespace.
15100 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15102 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15105 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15106 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15108 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15110 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15111 (qdf24xx): Likewise.
15112 * config/aarch64/aarch64.md: Include falkor.md.
15113 * config/aarch64/falkor.md: New.
15115 2017-08-09 Marek Polacek <polacek@redhat.com>
15118 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15119 * diagnostic.c (emit_diagnostic): Add a comment.
15120 (emit_diagnostic_valist): New function.
15122 2017-08-09 Marek Polacek <polacek@redhat.com>
15125 * input.c (make_location): New overload.
15126 * input.h (make_location): Declare.
15128 2017-08-08 Alan Modra <amodra@gmail.com>
15129 H.J. Lu <hongjiu.lu@intel.com>
15132 * gcc.c (NO_PIE_SPEC): Delete.
15133 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15135 (LINK_PIE_SPEC): ..to here.
15136 (LINK_COMMAND_SPEC): Support -no-pie.
15137 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15138 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15140 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15141 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15142 (ENDFILE_CRTEND_SPEC): Similarly.
15144 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15147 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15148 (mstack-protector-guard-offset=): Ditto.
15149 * config/i386/i386.c (ix86_option_override): Handle
15150 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15152 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15153 ix86_stack_protect_guard_offset variables.
15154 (TARGET_STACK_PROTECT_GUARD): Always define.
15155 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15156 and -mstack-protector-guard-offset= options.
15158 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15160 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15161 boundary case for the last candidate.
15163 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15165 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15166 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15168 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15170 PR middle-end/19706
15171 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15172 * config/aarch64/aarch64-builtins.c
15173 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15174 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15175 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15177 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15178 Andrew Pinski <pinskia@gmail.com>
15180 PR middle-end/19706
15181 * internal-fn.def (XORSIGN): New.
15182 * optabs.def (xorsign_optab): New.
15183 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15184 (convert_expand_mult_copysign): New.
15185 (pass_optimize_widening_mul::execute): Call
15186 convert_expand_mult_copysign.
15188 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15190 PR tree-optimization/81354
15191 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15192 Insert on edges rather than explicitly creating landing pads.
15193 (analyze_candidates_and_replace): Commit edge inserts.
15195 2017-08-08 Richard Biener <rguenther@suse.de>
15197 PR middle-end/81719
15198 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15199 (expand_simple_operations): Also look through ADDR_EXPRs with
15200 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15202 2017-08-08 Richard Biener <rguenther@suse.de>
15204 PR tree-optimization/81723
15205 * tree-vect-slp.c (struct bst_traits): New hash traits.
15206 (bst_fail): New global.
15207 (vect_build_slp_tree_2): New worker, split out from ...
15208 (vect_build_slp_tree): ... this now wrapping it with using
15209 bst_fail set to cache SLP tree build fails. Properly handle
15211 (vect_analyze_slp_instance): Allocate and free bst_fail.
15213 2017-08-08 Martin Liska <mliska@suse.cz>
15216 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15217 LABEL_DECLs that can be from a different function.
15219 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15221 PR tree-optimization/81744
15222 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15223 loop's number of iterations.
15225 2017-08-08 Martin Liska <mliska@suse.cz>
15227 * asan.c: Include header files.
15228 * attribs.c (build_decl_attribute_variant): New function moved
15230 (build_type_attribute_qual_variant): Likewise.
15231 (cmp_attrib_identifiers): Likewise.
15232 (simple_cst_list_equal): Likewise.
15233 (omp_declare_simd_clauses_equal): Likewise.
15234 (attribute_value_equal): Likewise.
15235 (comp_type_attributes): Likewise.
15236 (build_type_attribute_variant): Likewise.
15237 (lookup_ident_attribute): Likewise.
15238 (remove_attribute): Likewise.
15239 (merge_attributes): Likewise.
15240 (merge_type_attributes): Likewise.
15241 (merge_decl_attributes): Likewise.
15242 (merge_dllimport_decl_attributes): Likewise.
15243 (handle_dll_attribute): Likewise.
15244 (attribute_list_equal): Likewise.
15245 (attribute_list_contained): Likewise.
15246 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15247 (lookup_attribute_by_prefix): Likewise.
15248 * bb-reorder.c: Include header files.
15249 * builtins.c: Likewise.
15250 * calls.c: Likewise.
15251 * cfgexpand.c: Likewise.
15252 * cgraph.c: Likewise.
15253 * cgraphunit.c: Likewise.
15254 * convert.c: Likewise.
15255 * dwarf2out.c: Likewise.
15256 * final.c: Likewise.
15257 * fold-const.c: Likewise.
15258 * function.c: Likewise.
15259 * gimple-expr.c: Likewise.
15260 * gimple-fold.c: Likewise.
15261 * gimple-pretty-print.c: Likewise.
15262 * gimple.c: Likewise.
15263 * gimplify.c: Likewise.
15264 * hsa-common.c: Likewise.
15265 * hsa-gen.c: Likewise.
15266 * internal-fn.c: Likewise.
15267 * ipa-chkp.c: Likewise.
15268 * ipa-cp.c: Likewise.
15269 * ipa-devirt.c: Likewise.
15270 * ipa-fnsummary.c: Likewise.
15271 * ipa-inline.c: Likewise.
15272 * ipa-visibility.c: Likewise.
15274 * lto-cgraph.c: Likewise.
15275 * omp-expand.c: Likewise.
15276 * omp-general.c: Likewise.
15277 * omp-low.c: Likewise.
15278 * omp-offload.c: Likewise.
15279 * omp-simd-clone.c: Likewise.
15280 * opts-global.c: Likewise.
15281 * passes.c: Likewise.
15282 * predict.c: Likewise.
15283 * sancov.c: Likewise.
15284 * sanopt.c: Likewise.
15285 * symtab.c: Likewise.
15286 * toplev.c: Likewise.
15287 * trans-mem.c: Likewise.
15288 * tree-chkp.c: Likewise.
15289 * tree-eh.c: Likewise.
15290 * tree-into-ssa.c: Likewise.
15291 * tree-object-size.c: Likewise.
15292 * tree-parloops.c: Likewise.
15293 * tree-profile.c: Likewise.
15294 * tree-ssa-ccp.c: Likewise.
15295 * tree-ssa-live.c: Likewise.
15296 * tree-ssa-loop.c: Likewise.
15297 * tree-ssa-sccvn.c: Likewise.
15298 * tree-ssa-structalias.c: Likewise.
15299 * tree-ssa.c: Likewise.
15300 * tree-streamer-in.c: Likewise.
15301 * tree-vectorizer.c: Likewise.
15302 * tree-vrp.c: Likewise.
15303 * tsan.c: Likewise.
15304 * ubsan.c: Likewise.
15305 * varasm.c: Likewise.
15306 * varpool.c: Likewise.
15307 * tree.c: Remove functions moved to attribs.[ch].
15308 * tree.h: Likewise.
15309 * config/aarch64/aarch64.c: Add attrs.h header file.
15310 * config/alpha/alpha.c: Likewise.
15311 * config/arc/arc.c: Likewise.
15312 * config/arm/arm.c: Likewise.
15313 * config/avr/avr.c: Likewise.
15314 * config/bfin/bfin.c: Likewise.
15315 * config/c6x/c6x.c: Likewise.
15316 * config/cr16/cr16.c: Likewise.
15317 * config/cris/cris.c: Likewise.
15318 * config/darwin.c: Likewise.
15319 * config/epiphany/epiphany.c: Likewise.
15320 * config/fr30/fr30.c: Likewise.
15321 * config/frv/frv.c: Likewise.
15322 * config/ft32/ft32.c: Likewise.
15323 * config/h8300/h8300.c: Likewise.
15324 * config/i386/winnt.c: Likewise.
15325 * config/ia64/ia64.c: Likewise.
15326 * config/iq2000/iq2000.c: Likewise.
15327 * config/lm32/lm32.c: Likewise.
15328 * config/m32c/m32c.c: Likewise.
15329 * config/m32r/m32r.c: Likewise.
15330 * config/m68k/m68k.c: Likewise.
15331 * config/mcore/mcore.c: Likewise.
15332 * config/microblaze/microblaze.c: Likewise.
15333 * config/mips/mips.c: Likewise.
15334 * config/mmix/mmix.c: Likewise.
15335 * config/mn10300/mn10300.c: Likewise.
15336 * config/moxie/moxie.c: Likewise.
15337 * config/msp430/msp430.c: Likewise.
15338 * config/nds32/nds32-isr.c: Likewise.
15339 * config/nds32/nds32.c: Likewise.
15340 * config/nios2/nios2.c: Likewise.
15341 * config/nvptx/nvptx.c: Likewise.
15342 * config/pa/pa.c: Likewise.
15343 * config/pdp11/pdp11.c: Likewise.
15344 * config/powerpcspe/powerpcspe.c: Likewise.
15345 * config/riscv/riscv.c: Likewise.
15346 * config/rl78/rl78.c: Likewise.
15347 * config/rx/rx.c: Likewise.
15348 * config/s390/s390.c: Likewise.
15349 * config/sh/sh.c: Likewise.
15350 * config/sol2.c: Likewise.
15351 * config/sparc/sparc.c: Likewise.
15352 * config/spu/spu.c: Likewise.
15353 * config/stormy16/stormy16.c: Likewise.
15354 * config/tilegx/tilegx.c: Likewise.
15355 * config/tilepro/tilepro.c: Likewise.
15356 * config/v850/v850.c: Likewise.
15357 * config/vax/vax.c: Likewise.
15358 * config/visium/visium.c: Likewise.
15359 * config/xtensa/xtensa.c: Likewise.
15361 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15364 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15365 constraints since the -mupper-regs-* switches have been
15367 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15368 into a vector from a double word element that was extracted from
15369 another vector, and eliminate extra XXPERMDI instructions.
15370 (vsx_concat_<mode>_2): Likewise.
15371 (vsx_concat_<mode>_3): Likewise.
15372 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15373 concat to allow optimizing inserts from previous extracts.
15375 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15377 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15378 memory reference to a SSP offset in TLS address space.
15379 (ix86_print_operand) <case '@'>: Remove.
15380 (ix86_print_operand_punct_valid_p): Remove '@' code.
15381 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15382 UNSPEC_SP_TLS_TEST.
15383 (stack_tls_protect_set_<mode>): Remove.
15384 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15385 (stack_tls_protect_test_<mode>): Remove.
15386 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15388 2017-08-07 Olivier Hainque <hainque@adacore.com>
15391 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15393 2017-08-07 Douglas Rupp <rupp@adacore.com>
15395 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15396 variable was referenced as multidir in command.
15398 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15401 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15404 2017-08-07 Martin Liska <mliska@suse.cz>
15406 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15407 * config/rl78/rl78.c: Add include of attribs.h.
15408 * config/sh/sh.c: Likewise.
15409 * config/v850/v850.c: Likewise.
15411 2017-08-07 Tom de Vries <tom@codesourcery.com>
15413 PR middle-end/78266
15414 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15416 2017-08-07 Martin Liska <mliska@suse.cz>
15418 * config/mips/mips.c: Include attribs.h.
15420 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15423 * doc/invoke.texi: Document change in behvaior for -Ofast for
15426 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15428 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15430 (aarch64_pop_regs): Likewise.
15431 (aarch64_gen_load_pair): Likewise.
15432 (aarch64_save_callee_saves): Likewise.
15433 (aarch64_restore_callee_saves): Likewise.
15435 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15437 * config/i386/i386.c: Revert the last change.
15439 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15442 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15444 (ix86_finalize_stack_frame_flags): This. Also clear
15445 frame_pointer_needed if -fno-omit-frame-pointer is used without
15447 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15448 with ix86_finalize_stack_frame_flags.
15449 (ix86_expand_epilogue): Likewise.
15450 (ix86_expand_split_stack_prologue): Likewise.
15452 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15455 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15456 to P_AES for Westmere.
15458 2017-08-07 Jonathan Yong <10walls@gmail.com>
15460 * config/i386/mingw.opt (fset-stack-executable): Removed.
15461 * config/i386/cygming.opt (fset-stack-executable): Moved
15463 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15465 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15467 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15469 2017-08-07 Marek Polacek <polacek@redhat.com>
15471 PR middle-end/81737
15472 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15474 2017-08-07 Martin Liska <mliska@suse.cz>
15476 * attribs.h (canonicalize_attr_name): New function.
15477 (cmp_attribs): Move from c-format.c and adjusted.
15478 (is_attribute_p): Moved from tree.h.
15479 * tree-inline.c: Add new includes.
15480 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15481 (private_is_attribute_p): Remove.
15482 (private_lookup_attribute): Likewise.
15483 (private_lookup_attribute_by_prefix): Simplify.
15484 (remove_attribute): Use is_attribute_p.
15485 * tree.h: Remove removed declarations.
15487 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15489 PR middle-end/81698
15490 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15491 instead of computing it in the function. Formatting fix.
15492 (expand_case): Don't rely on default_edge being the first edge,
15493 clear it if removing it, pass default_edge to
15494 emit_case_dispatch_table.
15495 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15498 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15500 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15501 insn in the function, emit NOP after the insn.
15503 2017-08-06 Tom de Vries <tom@codesourcery.com>
15505 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15508 2017-08-06 Tom de Vries <tom@codesourcery.com>
15510 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15513 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15515 PR tree-optimization/57371
15516 * match.pd: New pattern.
15518 2017-08-04 Marek Polacek <polacek@redhat.com>
15520 PR middle-end/81695
15521 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15522 perform the computation in offset_int.
15524 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15526 PR tree-optimization/81136
15527 * tree-vectorizer.h: Include tree-hash-traits.h.
15528 (vec_base_alignments): New typedef.
15529 (vec_info): Add a base_alignments field.
15530 (vect_record_base_alignments): Declare.
15531 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15533 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15534 (create_data_ref): Add an is_conditional_in_stmt argument.
15535 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15536 the is_conditional_in_stmt field.
15537 (data_ref_loc): Add an is_conditional_in_stmt field.
15538 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15539 (find_data_references_in_stmt): Update call to create_data_ref.
15540 (graphite_find_data_references_in_stmt): Likewise.
15541 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15542 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15543 (vect_record_base_alignment): New function.
15544 (vect_record_base_alignments): Likewise.
15545 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15546 for nested statements even if we fail to compute a misalignment.
15547 Use pooled base alignments for unconditional references.
15548 (vect_find_same_alignment_drs): Compare base addresses instead
15550 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15551 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15553 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15555 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15556 Add an explicit name for the enum. Use auto_vec for slp_instances
15557 and grouped_stores.
15558 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15560 (_bb_vec_info): Add a constructor and destructor.
15561 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15562 (destroy_loop_vec_info): Delete.
15563 (vect_destroy_datarefs): Likewise.
15564 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15565 (vec_info::vec_info): New function.
15566 (vec_info::~vec_info): Likewise.
15567 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15568 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15569 destroy_loop_vec_info.
15570 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15571 (_loop_vec_info::_loop_vec_info): ...this.
15572 (destroy_loop_vec_info): Replace with...
15573 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15574 the stmt_vec_infos. Leave handling of vec_info information to its
15575 destructor. Remove explicit vector releases.
15576 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15577 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15578 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15579 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15580 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15581 (destroy_bb_vec_info): Replace with...
15582 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15583 information to its destructor.
15584 (vect_slp_analyze_bb_1): Use new and delete instead of
15585 new_bb_vec_info and destroy_bb_vec_info.
15586 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15589 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15591 * tree-data-ref.h (subscript): Add access_fn field.
15592 (data_dependence_relation): Add could_be_independent_p.
15593 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15594 (same_access_functions): Move to tree-data-ref.c.
15595 * tree-data-ref.c (ref_contains_union_access_p): New function.
15596 (access_fn_component_p): Likewise.
15597 (access_fn_components_comparable_p): Likewise.
15598 (dr_analyze_indices): Add a reference to access_fn_component_p.
15599 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15601 (constant_access_functions): Likewise.
15602 (add_other_self_distances): Likewise.
15603 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15604 (initialize_data_dependence_relation): Use XCNEW and remove
15605 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15606 of access functions that have the same type. Allow the
15607 subsequence to end with different bases in some circumstances.
15608 Record the chosen access functions in SUB_ACCESS_FN.
15609 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15610 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15611 (subscript_dependence_tester_1): Likewise dra and drb.
15612 (build_classic_dist_vector): Update calls accordingly.
15613 (subscript_dependence_tester): Likewise.
15614 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15615 DDR_COULD_BE_INDEPENDENT_P.
15616 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15617 comp_alias_ddrs instead of may_alias_ddrs.
15618 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15620 (vect_analyze_data_ref_dependence): Use it if
15621 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15622 distance vectors if that fails.
15623 (dependence_distance_ge_vf): New function.
15624 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15625 LOOP_VINFO_MAY_ALIAS_DDRS.
15627 2017-08-04 Richard Biener <rguenther@suse.de>
15629 PR middle-end/81705
15630 * fold-const.c (fold_binary_loc): Properly restrict
15631 minus_var0 && minus_var1 case when associating undefined overflow
15634 2017-08-04 Tom de Vries <tom@codesourcery.com>
15636 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15638 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15640 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15641 Don't start diagnostic messages with a capital letter.
15642 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15644 (rs6000_invalid_builtin): Likewise.
15645 (rs6000_trampoline_init): Likewise.
15647 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15650 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15651 after setting changeable df flags.
15653 2017-08-03 Richard Biener <rguenther@suse.de>
15655 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15656 up if the use is in USE - X.
15658 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15660 * toplev.c (dumpfile.h): New include.
15661 (internal_error_reentered): New static function. Use it...
15662 (internal_error_function): ...here to handle reentered internal_error.
15664 2017-08-03 Richard Biener <rguenther@suse.de>
15666 PR middle-end/81148
15667 * fold-const.c (split_tree): Add minus_var and minus_con
15668 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15669 here but always use minus_*.
15670 (associate_trees): Assert we never associate with MINUS_EXPR
15671 and NULL first operand. Do not recurse for PLUS_EXPR operands
15672 when associating as MINUS_EXPR either.
15673 (fold_binary_loc): Track minus_var and minus_con.
15675 2017-08-03 Tom de Vries <tom@codesourcery.com>
15678 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15679 ACCEL_COMPILER, apply finish_options on
15680 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15682 2017-08-03 Tom de Vries <tom@codesourcery.com>
15685 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15686 function_entry_patch_area_size > 0.
15688 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15691 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15692 instead of 10??LU, perform unit multiplication in wide_int,
15693 don't change alloc_object_size_limit if the limit is larger
15696 PR tree-optimization/81655
15697 PR tree-optimization/81588
15698 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15699 the case when ranges[i].low and high are 1 for unsigned type with
15702 PR middle-end/81052
15703 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15704 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15706 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15708 * tree-vrp.h: Add include guard.
15710 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15713 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15714 (ud2): New insn pattern.
15715 * config/i386/i386.c (ix86_expand_epilogue):
15716 For naked functions, generate ud2 instead of trap insn.
15718 2017-08-02 Marek Polacek <polacek@redhat.com>
15721 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15723 2017-08-02 Tom de Vries <tom@codesourcery.com>
15724 Cesar Philippidis <cesar@codesourcery.com>
15726 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15727 Add missing edge probabilities.
15729 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15731 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15732 Correct endianness.
15734 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15736 PR middle-end/79499
15737 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15738 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15739 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15740 prologue_seq sequences - if any.
15742 2017-08-02 Richard Biener <rguenther@suse.de>
15744 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15745 via vectors if supported, integer extracts via punning if supported
15746 or otherwise vector extracts.
15748 2017-08-02 Richard Biener <rguenther@suse.de>
15750 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15752 (bitmap_insert_into_set): ... this.
15754 2017-08-02 Richard Biener <rguenther@suse.de>
15756 PR tree-optimization/81633
15758 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15760 PR tree-optimization/71752
15761 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15763 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15765 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15766 (machine_function::call_ms2sysv_pad_out): Remove field.
15767 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15768 (ix86_compute_frame_layout): Likewise.
15770 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15773 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15774 attribute with interrupt attribute.
15776 2017-08-01 Andrew Pinski <apinski@cavium.com>
15778 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15779 BIT_INSERT_EXPR's operand 1
15780 to see if the types precision matches.
15782 2017-08-01 Martin Liska <mliska@suse.cz>
15784 PR middle-end/70140
15785 * builtins.c (expand_builtin_memcpy_args): Remove.
15786 (expand_builtin_memcpy): Call newly added function
15787 expand_builtin_memory_copy_args.
15788 (expand_builtin_memcpy_with_bounds): Likewise.
15789 (expand_builtin_mempcpy): Remove last argument.
15790 (expand_builtin_mempcpy_with_bounds): Likewise.
15791 (expand_builtin_memory_copy_args): New function created from
15792 expand_builtin_mempcpy_args with small modifications.
15793 (expand_builtin_mempcpy_args): Remove.
15794 (expand_builtin_stpcpy): Remove unused argument.
15795 (expand_builtin): Likewise.
15796 (expand_builtin_with_bounds): Likewise.
15798 2017-08-01 Martin Liska <mliska@suse.cz>
15801 Make mempcpy more optimal (PR middle-end/70140).
15803 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15806 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15807 __builtin_vec_cmpne verify both arguments are compatible vectors
15808 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15809 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15810 move computation of aligned to after checking the argument types.
15814 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15815 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15818 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15819 Doug Rupp <rupp@adacore.com>
15820 Olivier Hainque <hainque@adacore.com>
15822 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15823 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15825 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15826 for TARGET_OS_CPP_BUILTIN.
15827 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15828 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15830 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15831 passing required abi options to the assembler for EABI configurations.
15832 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15833 of .text.hot and .text.unlikely sections for kernel modules when
15834 using ARM style exceptions.
15835 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15836 options. Add EXTRA_CC1_SPEC.
15837 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15839 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15841 (ARM_TARGET2_DWARF_FORMAT): Define.
15842 * config/arm/t-vxworks: Adjust multilib control to removal of the
15843 Diab command line options.
15845 2017-08-01 Martin Liska <mliska@suse.cz>
15847 PR gcov-profile/81561
15848 * gcov.c (unblock): Make unblocking safe as we need to preserve
15849 index correspondence of blocks and block_lists.
15851 2017-08-01 Richard Biener <rguenther@suse.de>
15853 PR tree-optimization/81181
15854 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15855 (compute_antic): ... end of iteration here.
15857 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15859 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15860 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15861 (ftree-slp-vectorize): Likewise.
15862 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15863 can no longer be set independent of flag_tree_loop_vectorize.
15864 * omp-general.c (emp_max_vf): Likewise.
15865 * opts.c (enable_fdo_optimizations): Remove references to
15866 flag_tree_vectorize, these are now implicit.
15867 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15868 and leave it for the options machinery.
15870 2017-08-01 Martin Liska <mliska@suse.cz>
15872 PR middle-end/70140
15873 * builtins.c (expand_builtin_memcpy_args): Remove.
15874 (expand_builtin_memcpy): Call newly added function
15875 expand_builtin_memory_copy_args.
15876 (expand_builtin_memcpy_with_bounds): Likewise.
15877 (expand_builtin_mempcpy): Remove last argument.
15878 (expand_builtin_mempcpy_with_bounds): Likewise.
15879 (expand_builtin_memory_copy_args): New function created from
15880 expand_builtin_mempcpy_args with small modifications.
15881 (expand_builtin_mempcpy_args): Remove.
15882 (expand_builtin_stpcpy): Remove unused argument.
15883 (expand_builtin): Likewise.
15884 (expand_builtin_with_bounds): Likewise.
15886 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15889 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15890 print "ds:" only for immediates in generic address space.
15892 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15895 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15896 (ix86_function_ok_for_sibcall): Return false for naked functions.
15898 2017-08-01 Richard Biener <rguenther@suse.de>
15900 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15901 (compute_antic): Seed worklist with exit block predecessors.
15902 * cfganal.c (dfs_find_deadend): For a cycle return the source
15903 of the edge closing it.
15905 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15907 * config/aarch64/aarch64.c
15908 (aarch64_can_const_movi_rtx_p): Move 0 check.
15910 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15912 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15913 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15915 * match.pd: Ditto in address comparison pattern.
15917 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15919 PR tree-optimization/81627
15920 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15921 closed ssa form for store-store chain.
15923 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15925 PR tree-optimization/81620
15926 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15927 for store-store chain.
15929 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15931 PR tree-optimization/81588
15932 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15933 ranges[i].in_p, invert comparison code ccode. For >/>=,
15934 swap rhs1 and rhs2 and comparison code unconditionally,
15935 for </<= don't do that. Don't swap rhs1/rhs2 again if
15936 ranges[i].in_p, instead invert comparison code ccode if
15937 opcode or oe->rank is BIT_IOR_EXPR.
15940 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15941 a direct optab to conversion optab.
15942 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15943 with GET_MODE_INNER as last argument instead of optab_handler.
15944 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15945 vector extraction if possible and optab is available.
15946 * expr.c (store_constructor): Use convert_optab_handler instead
15947 of optab_handler. Use vector initialization from smaller
15948 vectors if possible and optab is available.
15949 * tree-vect-stmts.c (vectorizable_load): Likewise.
15950 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15951 now have two modes.
15952 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15953 of vec_init from half-sized vectors with the same element mode.
15954 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15955 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15956 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15957 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15958 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15959 after mode in gen_vec_extract* calls.
15960 (vec_extract<mode>): Renamed to ...
15961 (vec_extract<mode><ssescalarmodelower>): ... this.
15962 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15963 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15964 element mode after mode in gen_vec_init* calls.
15965 (VEC_INIT_HALF_MODE): New mode iterator.
15966 (vec_init<mode>): Renamed to ...
15967 (vec_init<mode><ssescalarmodelower>): ... this.
15968 (vec_init<mode><ssehalfvecmodelower>): New expander.
15969 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15970 (vec_extractv2sfsf): ... this.
15971 (vec_initv2sf): Renamed to ...
15972 (vec_initv2sfsf): ... this.
15973 (vec_extractv2si): Renamed to ...
15974 (vec_extractv2sisi): ... this.
15975 (vec_initv2si): Renamed to ...
15976 (vec_initv2sisi): ... this.
15977 (vec_extractv4hi): Renamed to ...
15978 (vec_extractv4hihi): ... this.
15979 (vec_initv4hi): Renamed to ...
15980 (vec_initv4hihi): ... this.
15981 (vec_extractv8qi): Renamed to ...
15982 (vec_extractv8qiqi): ... this.
15983 (vec_initv8qi): Renamed to ...
15984 (vec_initv8qiqi): ... this.
15985 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15986 (vec_init<mode>): Renamed to ...
15987 (vec_init<mode><VEC_base_l>): ... this.
15988 (vec_extract<mode>): Renamed to ...
15989 (vec_extract<mode><VEC_base_l>): ... this.
15990 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15991 (vec_initv2sfsf): ... this.
15992 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15993 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15994 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15995 element mode after mode in gen_vec_init* calls.
15996 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15997 (vec_init<mode><Vel>): ... this.
15998 (vec_extract<mode>): Renamed to ...
15999 (vec_extract<mode><Vel>): ... this.
16000 * config/aarch64/iterators.md (Vel): New mode attribute.
16001 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16002 Add element mode after mode in gen_vec_extract* calls.
16003 * config/s390/vector.md (non_vec_l): New mode attribute.
16004 (vec_extract<mode>): Renamed to ...
16005 (vec_extract<mode><non_vec_l>): ... this.
16006 (vec_init<mode>): Renamed to ...
16007 (vec_init<mode><non_vec_l>): ... this.
16008 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16009 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16011 * config/arm/iterators.md (V_elem_l): New mode attribute.
16012 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16013 (vec_extract<mode><V_elem_l>): ... this.
16014 (vec_extractv2di): Renamed to ...
16015 (vec_extractv2didi): ... this.
16016 (vec_init<mode>): Renamed to ...
16017 (vec_init<mode><V_elem_l>): ... this.
16018 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16019 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16020 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16021 Add element mode after gen_vec_extract* calls.
16022 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16023 (vec_init<mode><unitmode>): ... this.
16024 (vec_extract<mode>): Renamed to ...
16025 (vec_extract<mode><unitmode>): ... this.
16026 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16027 (vec_init<mode><unitmode>): ... this.
16028 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16029 (vec_initv2sfsf): ... this.
16030 (vec_extractv2sf): Renamed to ...
16031 (vec_extractv2sfsf): ... this.
16032 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16033 Add element mode after gen_vec_extract* calls.
16034 * config/mips/mips.md (unitmode): New mode iterator.
16035 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16036 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16037 * config/spu/spu.md (inner_l): New mode attribute.
16038 (vec_init<mode>): Renamed to ...
16039 (vec_init<mode><inner_l>): ... this.
16040 (vec_extract<mode>): Renamed to ...
16041 (vec_extract<mode><inner_l>): ... this.
16042 * config/sparc/sparc.md (veltmode): New mode iterator.
16043 (vec_init<VMALL:mode>): Renamed to ...
16044 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16045 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16046 (vec_initv2sisi): ... this.
16047 (vec_initv2sf): Renamed to ...
16048 (vec_initv2sfsf): ... this.
16049 (vec_extractv2sf): Renamed to ...
16050 (vec_extractv2sfsf): ... this.
16051 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16052 (vec_init<mode>): Renamed to ...
16053 (vec_init<mode><VEC_base_l>): ... this.
16054 (vec_extract<mode>): Renamed to ...
16055 (vec_extract<mode><VEC_base_l>): ... this.
16056 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16057 (vec_initv2sfsf): ... this.
16058 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16059 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16060 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16061 gen_vec_init* calls.
16063 2017-08-01 Richard Biener <rguenther@suse.de>
16065 PR tree-optimization/81297
16066 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16067 TREE_OVERFLOW from INTEGER_CSTs.
16069 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16071 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16073 2017-07-31 Carl Love <cel@us.ibm.com>
16075 * config/rs6000/rs6000-c: Add support for built-in functions
16076 vector signed char vec_xl_be (signed long long, signed char *);
16077 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16078 vector signed int vec_xl_be (signed long long, signed int *);
16079 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16080 vector signed long long vec_xl_be (signed long long, signed long long *);
16081 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16082 vector signed short vec_xl_be (signed long long, signed short *);
16083 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16084 vector double vec_xl_be (signed long long, double *);
16085 vector float vec_xl_be (signed long long, float *);
16086 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16087 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16088 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16090 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16091 (altivec_expand_builtin): Add switch statement to call
16092 altivec_expand_xl_be for each builtin.
16093 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16094 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16095 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16096 __builtin_vsx_le_be_v16qi.
16097 * doc/extend.texi: Update the built-in documentation file for the
16098 new built-in functions.
16100 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16103 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16105 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16107 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16109 * config.gcc: Add z14.
16110 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16111 CPU model numbers for z13s and z14.
16112 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16114 * config/s390/s390-opts.h (enum processor_type): Rename
16115 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16116 * config/s390/s390.c (processor_table): Add field for CPU name to
16117 be passed to Binutils.
16118 (s390_asm_output_machine_for_arch): Use the new field in
16119 processor_table for Binutils.
16120 (s390_expand_builtin): Replace arch12 with z14.
16121 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16122 (s390_get_sched_attrmask): Likewise.
16123 (s390_get_unit_mask): Likewise.
16124 * config/s390/s390.opt: Add z14 to processor_type enum.
16126 2017-07-31 Martin Jambor <mjambor@suse.cz>
16129 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16130 regardless of optimization level.
16132 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16133 Martin Liska <mliska@suse.cz>
16135 * predict.def: Remove old comment and adjust probability.
16136 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16137 PREDICT statements.
16139 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16142 * config/i386/i386.c (ix86_function_naked): New function.
16143 (ix86_can_use_return_insn_p): Return false for naked functions.
16144 (ix86_expand_prologue): Skip prologue for naked functions.
16145 (ix86_expand_epilogue): Skip epilogue for naked functions
16146 and emit trap instruction.
16147 (ix86_warn_func_return): New function.
16148 (ix86_attribute_table): Add "naked" attribute specification.
16149 (TARGET_WARN_FUNC_RETURN): Define.
16150 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16152 2017-07-31 Martin Liska <mliska@suse.cz>
16154 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16155 (dump_gimple_bb_header): Always dump BB info.
16156 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16158 2017-07-31 Martin Liska <mliska@suse.cz>
16161 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16162 also with current_function_decl non-null equality.
16164 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16167 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16168 change type to the element type, instead add eltype variable and
16169 use it where we are interested in the element type.
16171 PR tree-optimization/81603
16172 * ipa-polymorphic-call.c
16173 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16174 offset arithmetic in offset_int, bail out if the resulting bit offset
16175 doesn't fit into shwi.
16177 2017-07-31 Martin Liska <mliska@suse.cz>
16179 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16180 (gimplify_save_expr): Fix comment.
16182 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16185 * config/i386/i386.c (ix86_function_arg): Update arguments for
16187 (ix86_compute_frame_layout): Set the initial stack offset to
16188 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16189 INCOMING_FRAME_SP_OFFSET.
16190 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16191 stack before exception handler returns.
16192 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16193 the 'ERROR_CODE' for exception handler.
16195 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16197 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16198 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16199 (ASM_OUTPUT_REG_POP): Ditto.
16200 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16201 instead of asm_fprintf to output pure string.
16203 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16205 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16206 to imported_module_or_decl hook.
16207 (debug_nothing_tree_tree_tree_bool): Remove.
16208 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16209 * debug.c (do_nothing_debug_hooks): Use
16210 debug_nothing_tree_tree_tree_bool_bool instead of
16211 debug_nothing_tree_tree_tree_bool.
16212 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16213 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16214 * sdbout.c (sdb_debug_hooks): Likewise.
16215 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16216 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16218 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16219 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16220 attribute, don't add anything.
16222 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16224 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16225 (fold_build2_stat_loc): Likewise.
16226 (fold_build3_stat_loc): Likewise.
16227 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16228 (fold_build1_loc): Remove macro.
16229 (fold_build2_loc): Likewise.
16230 (fold_build3_loc): Likewise.
16232 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16234 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16235 (gimple_build_debug_bind_source_stat): Likewise.
16236 * gimple.h (gimple_build_debug_bind): Remove macro.
16237 (gimple_build_debug_bind_source): Likewise.
16239 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16241 * bitmap.c (bitmap_alloc): Adjust.
16242 (bitmap_gc_alloc): Likewise.
16243 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16245 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16247 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16248 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16249 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16250 (bitmap_gc_alloc_stat): Likewise.
16251 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16253 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16255 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16256 * rtl.h (shallow_copy_rtx): Remove macro.
16258 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16260 * emit-rtl.c (gen_raw_REG): Adjust.
16261 * gengenrtl.c (gendef): Likewise.
16262 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16263 * rtl.h (rtx_alloc): Remove macro.
16265 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16267 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16268 (build_tree_list_stat): Likewise.
16269 * tree.h (build_tree_list): Remove macro.
16270 (build_tree_list_vec): Likewise.
16272 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16274 * tree.c (make_vector_stat): Remove _stat from name.
16275 (build_vector_stat): Likewise.
16276 * tree.h (make_vector_stat): Remove macro.
16277 (build_vector_stat): Likewise.
16279 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16281 * tree.h (build_var_debug_value): Remove prototype.
16283 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16285 * tree.c (tree_cons_stat): Remove _stat from name.
16286 * tree.h (tree_cons): Remove macro.
16288 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16290 * tree.c (build_vl_exp_stat): Remove _stat from name.
16291 * tree.h (build_vl_exp): Remove macro.
16293 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16295 * tree.c (build_decl_stat): Remove _stat from name.
16296 * tree.h (build_decl): Remove macro.
16298 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16300 * gimple.c (gimple_build_with_ops_stat): Adjust.
16301 (gimple_alloc_stat): Remove _stat from name.
16302 * gimple.h (gimple_alloc): Remove macro.
16304 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16306 * tree.c (make_tree_vec_stat): Remove _stat from name.
16307 (grow_tree_vec_stat): Likewise.
16308 * tree.h (make_tree_vec_stat): Adjust prototype.
16309 (grow_tree_vec_stat): Likewise.
16310 (make_tree_vec): Remove macro.
16311 (grow_tree_vec): Likewise.
16313 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16315 * fold-const.c (fold_build1_stat_loc): Adjust.
16316 (fold_build2_stat_loc): Likewise.
16317 (fold_build3_stat_loc): Likewise.
16318 * tree.c (build0_stat): Remove _stat from name.
16319 (build1_stat): Likewise.
16320 (build2_stat): Likewise.
16321 (build3_stat): Likewise.
16322 (build4_stat): Likewise.
16323 (build5_stat): Likewise.
16324 * tree.h (build1_loc): Remove macro, and rename _stat function
16326 (build2_loc): Likewise.
16327 (build3_loc): Likewise.
16328 (build4_loc): Likewise.
16329 (build5_loc): Likewise.
16331 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16333 * tree.c (make_int_cst_stat): Remove _stat from name.
16334 * tree.h (make_int_cst_stat): Adjust prototype.
16335 (make_int_cst): Remove macro.
16337 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16339 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16340 * tree.h (make_tree_binfo_stat): Adjust prototype.
16341 (make_tree_binfo): Remove.
16343 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16345 * tree.c (copy_node_stat): Rename to copy_node.
16346 (build_distinct_type_copy): Adjust.
16347 * tree.h (copy_node_stat): Adjust prototype.
16348 (copy_node): Remove macro.
16350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16352 * tree.c (make_node_stat): rename to make_node.
16353 (build_tree_list_stat): Adjust.
16354 (build0_stat): Likewise.
16355 (build2_stat): Likewise.
16356 (build3_stat): Likewise.
16357 (build4_stat): Likewise.
16358 (build5_stat): Likewise.
16359 (build_decl_stat): Likewise.
16360 * tree.h (make_node_stat): Adjust prototype.
16361 (make_node): remove macro.
16363 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16365 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16366 (PPC_FEATURE2_SCV): Likewise.
16367 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16369 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16371 * config/aarch64/aarch64.c
16372 (aarch64_internal_mov_immediate): Add new special pattern.
16373 * config/aarch64/aarch64.md (*movdi_aarch64):
16374 Add reg/32bit const mov case.
16376 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16377 Richard Sandiford <richard.sandiford@linaro.org>
16379 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16380 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16381 Add integer and movi cases.
16382 (movi-split-hf-df-sf split, fp16): New.
16383 (enabled): Added TARGET_FP_F16INST.
16384 * config/aarch64/iterators.md (GPF_HF): New.
16385 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16387 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16389 * config/aarch64/aarch64.c
16390 (aarch64_simd_container_mode): Add prototype.
16391 (aarch64_expand_mov_immediate): Add HI support.
16392 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16393 (aarch64_can_const_movi_rtx_p): New.
16394 (aarch64_preferred_reload_class):
16395 Remove restrictions of using FP registers for certain SIMD operations.
16396 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16397 (aarch64_valid_floating_const): Add integer move validation.
16398 (aarch64_simd_imm_scalar_p): Remove.
16399 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16400 (aarch64_legitimate_constant_p): Expand list of supported cases.
16401 * config/aarch64/aarch64-protos.h
16402 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16403 (aarch64_reinterpret_float_as_int): New.
16404 (aarch64_simd_imm_scalar_p): Remove.
16405 * config/aarch64/constraints.md (Uvi): New.
16406 (Dd): Split into Ds and new Dd.
16407 * config/aarch64/aarch64.md (*movsi_aarch64):
16409 (*movdi_aarch64): Add SIMD mov case.
16411 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16413 * tree-predcom.c: (struct chain): Handle store-store chain in which
16414 stores for elimination only store loop invariant values.
16415 (execute_pred_commoning_chain): Ditto.
16416 (prepare_initializers_chain_store_elim): Ditto.
16417 (prepare_finalizers): Ditto.
16418 (is_inv_store_elimination_chain): New function.
16419 (initialize_root_vars_store_elim_1): New function.
16421 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16423 * tree-predcom.c: Revise general description of the pass.
16424 (enum chain_type): New enum type for store elimination.
16425 (struct chain): New field supporting store elimination.
16426 (struct component): Ditto.
16427 (dump_chain): Dump store-stores chain.
16428 (release_chain): Release resources.
16429 (split_data_refs_to_components): Compute and create component
16430 contains only stores for elimination.
16431 (get_chain_last_ref_at): New function.
16432 (make_invariant_chain): Initialization.
16433 (make_rooted_chain): Specify chain type in parameter and record it.
16434 (add_looparound_copies): Skip for store-stores chain.
16435 (determine_roots_comp): Compute type of chain and pass it to
16437 (initialize_root_vars_store_elim_2): New function.
16438 (finalize_eliminated_stores): New function.
16439 (remove_stmt): Handle store for elimination.
16440 (execute_pred_commoning_chain): Execute predictive commoning on
16441 store-store chains.
16442 (determine_unroll_factor): Skip unroll for store-stores chain.
16443 (prepare_initializers_chain_store_elim): New function.
16444 (prepare_initializers_chain): Hanlde store-store chain.
16445 (prepare_finalizers_chain, prepare_finalizers): New function.
16446 (tree_predictive_commoning_loop): Return integer value indicating
16447 if loop is unrolled or lcssa form is corrupted.
16448 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16450 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16452 * tree-predcom.c (initialize_root): Delete.
16453 (execute_pred_commoning_chain): Initialize root vars and replace
16454 reference of non-combined chain directly, rather than call above
16457 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16459 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16460 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16462 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16464 * tree-predcom.c (struct chain): New field init_seq.
16465 (release_chain): Release init_seq.
16466 (prepare_initializers_chain): Record intialization stmts in above
16468 (insert_init_seqs): New function.
16469 (tree_predictive_commoning_loop): Call insert_init_seqs.
16471 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16473 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16475 2017-07-28 Richard Biener <rguenther@suse.de>
16477 * match.pd: Remove superfluous :c.
16478 * genmatch.c (simplify::id): Add member.
16479 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16481 (current_id): New global.
16482 (dt_node::parent): Move from ...
16483 (dt_operand::parent): ... here. Add for_id member.
16484 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16485 (decision_tree::find_node): Relax order requirement when
16486 merging DT_TRUE nodes to ones inbetween the current simplify
16487 and the one we try to merge with. Add diagnostic whenever
16488 we need to enforce pattern order by not merging.
16489 (decision_tree::insert): Set current_id.
16490 (decision_tree::print_node): Dump parent node and for_id.
16491 (parser::last_id): Add member.
16492 (parser::push_simplify): Assign unique id.
16493 (parser::parser): Initialize last_id.
16495 2017-07-28 Martin Liska <mliska@suse.cz>
16498 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16499 gimple_build_debug_bind.
16501 2017-07-28 Richard Biener <rguenther@suse.de>
16503 PR tree-optimization/81502
16504 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16506 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16508 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16509 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16510 for BIT_FIELD_REF args.
16511 * fold-const.c (make_bit_field_ref): Likewise.
16512 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16514 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16517 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16518 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16519 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16520 Or it into SANITIZER_UNDEFINED.
16521 * ubsan.c: Include gimple-fold.h and varasm.h.
16522 (ubsan_expand_ptr_ifn): New function.
16523 (instrument_pointer_overflow): New function.
16524 (maybe_instrument_pointer_overflow): New function.
16525 (instrument_object_size): Formatting fix.
16526 (pass_ubsan::execute): Call instrument_pointer_overflow
16527 and maybe_instrument_pointer_overflow.
16528 * internal-fn.c (expand_UBSAN_PTR): New function.
16529 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16530 * sanitizer.def (__ubsan_handle_pointer_overflow,
16531 __ubsan_handle_pointer_overflow_abort): New builtins.
16532 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16533 * internal-fn.def (UBSAN_PTR): New internal function.
16534 * opts.c (sanitizer_opts): Add pointer-overflow.
16535 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16536 * fold-const.c (build_range_check): Compute pointer range check in
16537 integral type if pointer arithmetics would be needed. Formatting
16540 2017-07-28 Martin Liska <mliska@suse.cz>
16543 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16544 parameters that are of a variable-length.
16546 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16548 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16550 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16551 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16552 (CRT_CALL_STATIC_FUNCTION): Likewise.
16553 (ASM_DEFAULT_SPEC): New define.
16554 (ASM_SPEC32): Likewise.
16555 (ASM_SPEC64): Likewise.
16556 (ASM_SPEC_COMMON): Likewise.
16557 (ASM_SPEC): Likewise.
16558 (INVALID_64BIT): Likewise.
16559 (LINK_OS_DEFAULT_SPEC): Likewise.
16560 (LINK_OS_SPEC32): Likewise.
16561 (LINK_OS_SPEC64): Likewise.
16562 (POWERPC_LINUX): Likewise.
16563 (PTRDIFF_TYPE): Likewise.
16564 (RESTORE_FP_PREFIX): Likewise.
16565 (RESTORE_FP_SUFFIX): Likewise.
16566 (SAVE_FP_PREFIX): Likewise.
16567 (SAVE_FP_SUFFIX): Likewise.
16568 (SIZE_TYPE): Likewise.
16569 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16570 (TARGET_64BIT): Likewise.
16571 (TARGET_64BIT): Likewise.
16572 (TARGET_AIX): Likewise.
16573 (WCHAR_TYPE_SIZE): Likewise.
16574 (WCHAR_TYPE): Undefine.
16575 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16576 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16577 (CPP_OS_RTEMS_SPEC): Delete.
16578 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16579 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16581 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16583 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16585 PR tree-optimization/81578
16586 * tree-parloops.c (build_new_reduction): Bail out if
16587 reduction_code isn't one of the standard OpenMP reductions.
16588 Move the details printing after that decision.
16590 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16592 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16593 related to reload_in_progress.
16594 (splat_input_operand): Likewise.
16595 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16597 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16599 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16600 (TARGET_INSTANTIATE_DECLS): Likewise.
16601 (legitimate_indexed_address_p): Delete reload_in_progress code.
16602 (rs6000_debug_legitimate_address_p): Likewise.
16603 (rs6000_eliminate_indexed_memrefs): Likewise.
16604 (rs6000_emit_le_vsx_store): Likewise.
16605 (rs6000_emit_move_si_sf_subreg): Likewise.
16606 (rs6000_emit_move): Likewise.
16607 (register_to_reg_type): Likewise.
16608 (rs6000_pre_atomic_barrier): Likewise.
16609 (rs6000_machopic_legitimize_pic_address): Likewise.
16610 (rs6000_allocate_stack_temp): Likewise.
16611 (rs6000_address_for_fpconvert): Likewise.
16612 (rs6000_address_for_altivec): Likewise.
16613 (rs6000_secondary_memory_needed_rtx): Delete function.
16614 (rs6000_check_sdmode): Likewise.
16615 (rs6000_alloc_sdmode_stack_slot): Likewise.
16616 (rs6000_instantiate_decls): Likewise.
16617 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16618 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16619 Delete reload_in_progress.
16620 (*vec_reload_and_plus_<mptrsize>): Likewise.
16621 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16622 (vsx_div_v2di): Likewise.
16623 (vsx_udiv_v2di): Likewise.
16625 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16627 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16628 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16629 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16630 (rs6000_debug_reg_global): Delete print of LRA status.
16631 (rs6000_option_override_internal): Delete dead LRA related code.
16632 (rs6000_lra_p): Delete function.
16633 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16635 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16637 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16638 * config/riscv/rtems.h: New file.
16640 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16641 Sudakshina Das <sudi.das@arm.com>
16643 * config/aarch64/aarch64.md
16644 (define_split for and<mode>3nr_compare): Move
16645 non aarch64_logical_operand to a register.
16646 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16647 register immediate operand to a register.
16648 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16650 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16652 PR middle-end/81564
16653 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16655 2017-07-27 Richard Biener <rguenther@suse.de>
16657 PR tree-optimization/81573
16658 PR tree-optimization/81494
16659 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16660 multi defuse cycle case.
16662 2017-07-27 Richard Biener <rguenther@suse.de>
16664 PR tree-optimization/81571
16665 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16668 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16670 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16671 earlier and only if MASK_FPU is set. Adjust formatting.
16673 2017-07-27 Martin Liska <mliska@suse.cz>
16675 * opt-functions.awk: Add validation of value of Init.
16676 * optc-gen.awk: Pass new argument.
16678 2017-07-27 Martin Liska <mliska@suse.cz>
16680 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16681 Fix wrong condition.
16683 2017-07-27 Martin Liska <mliska@suse.cz>
16685 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16686 BBs and edges seen by autoFDO.
16688 2017-07-27 Richard Biener <rguenther@suse.de>
16690 PR tree-optimization/81502
16691 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16692 with incompatible but same sized type.
16693 (execute_update_addresses_taken): Likewise.
16695 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16697 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16698 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16700 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16703 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16704 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16705 Change s_operand to memory_operand.
16707 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16709 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16710 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16711 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16712 Emit instructions rather than returning an expression. Handle TFmode
16713 and KFmode by casting to TImode.
16714 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16715 (rs6000_emit_le_vsx_store): Likewise.
16716 * config/rs6000/vsx.md (VSX_TI): New iterator.
16717 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16718 (*vsx_le_undo_permute_<mode>): Likewise.
16719 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16720 emit the split sequence.
16721 (*vsx_le_perm_store_<mode>): Likewise.
16723 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16725 PR tree-optimization/81555
16726 PR tree-optimization/81556
16727 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16728 if true, force CHANGED for the recursive invocation.
16729 (reassociate_bb): Remember original length of ops array, pass
16730 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16732 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16733 attributes for noipa attribute. For naked attribute use
16734 lookup_attribute first before lookup_attribute_spec.
16735 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16737 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16738 for functions with noipa attribute.
16739 (cgraph_externally_visible_p): Return true for functions with noipa
16741 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16742 for functions with noipa attribute.
16743 * doc/extend.texi: Document noipa function attribute.
16744 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16745 also for functions with noipa attribute.
16746 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16748 2017-07-26 Andrew Pinski <apinski@cavium.com>
16750 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16751 vec_unalign_load_cost and vec_unalign_store_cost.
16753 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16755 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16756 -mvsx-small-integer option.
16757 (ISA_3_0_MASKS_IEEE): Likewise.
16758 (OTHER_VSX_VECTOR_MASKS): Likewise.
16759 (POWERPC_MASKS): Likewise.
16760 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16761 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16762 code, only testing for DImode being allowed in non-VSX floating
16764 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16765 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16767 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16768 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16769 TARGET_P8_VECTOR test.
16770 (rs6000_secondary_reload_simple_move): Likewise.
16771 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16772 since TARGET_P9_VECTOR was already tested.
16773 (rs6000_opt_masks): Remove -mvsx-small-integer.
16774 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16775 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16777 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16778 test for TARGET_VEXTRACTUB was used, and that uses
16780 (p9 extract splitter): Likewise.
16781 (vsx_extract_<mode>_di_p9): Likewise.
16782 (vsx_extract_<mode>_store_p9): Likewise.
16783 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16784 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16785 the elimination of TARGET_VSX_SMALL_INTEGER.
16786 (vsx_extract_<mode>_p8): Likewise.
16787 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16788 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16789 (vsx_set_<mode>_p9): Likewise.
16790 (vsx_set_v4sf_p9): Likewise.
16791 (vsx_set_v4sf_p9_zero): Likewise.
16792 (vsx_insert_extract_v4sf_p9): Likewise.
16793 (vsx_insert_extract_v4sf_p9_2): Likewise.
16794 * config/rs6000/rs6000.md (sign extend splitter): Change
16795 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16796 (floatsi<mode>2_lfiwax_mem): Likewise.
16797 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16798 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16799 since a test for TARGET_P9_VECTOR was used.
16800 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16801 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16802 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16803 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16804 TARGET_P8_VECTOR test.
16805 (fix_trunc<mode>si2_stfiwx): Likewise.
16806 (fix_trunc<mode>si2_internal): Likewise.
16807 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16808 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16810 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16811 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16812 TARGET_P8_VECTOR test.
16813 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16814 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16815 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16817 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16818 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16819 since a test for TARGET_P9_VECTOR was used.
16820 (splitter for loading small constants): Likewise.
16822 2017-07-26 Andrew Pinski <apinski@cavium.com>
16824 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16827 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16830 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16831 (fp_valid_at): Likewise.
16833 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16835 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16836 (qdf24xx_addrcost_table): Likewise.
16837 (cortexa57_tunings): Update to use generic_branch_cost.
16838 (cortexa72_tunings): Likewise.
16839 (cortexa73_tunings): Likewise.
16840 (qdf24xx_tunings): Likewise.
16842 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16844 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16845 (thunderx2t99_branch_cost): Likewise.
16846 (cortexa35_tunings): Update to use generic_branch_cost.
16847 (cortexa53_tunings): Likewise.
16848 (cortexa57_tunings): Likewise.
16849 (cortexa72_tunings): Likewise.
16850 (cortexa73_tunings): Likewise.
16851 (thunderx2t99_tunings): Likewise.
16853 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16855 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16856 (sparc_option_override): Honour MASK_FSMULD.
16857 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16858 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16859 * config/sparc/sparc.opt (mfsmuld): New option.
16860 * doc/invoke.texi (mfsmuld): Document option.
16862 2017-07-26 Marek Polacek <polacek@redhat.com>
16864 PR middle-end/70992
16865 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16867 2017-07-26 Richard Biener <rguenther@suse.de>
16869 * gimple-match-head.c (do_valueize): Return OP if valueize
16871 (get_def): New helper to get at the def stmt of a SSA name
16872 if valueize allows.
16873 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16874 do_valueize to get at the def stmt.
16875 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16877 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16879 PR middle-end/46932
16880 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16882 2017-07-26 Martin Liska <mliska@suse.cz>
16885 * function.c (expand_function_start): Make expansion of
16886 nonlocal_goto_save_area after parm_birth_insn.
16888 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16890 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16891 from all CPU target flags enable members.
16893 2017-07-26 Richard Biener <rguenther@suse.de>
16895 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16896 (decision_tree::gen): Make 'type' const.
16897 (write_predicate): Likewise.
16899 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16901 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16902 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16903 (rs6000_option_override_internal): Likewise.
16904 (rs6000_expand_vector_set): Likewise.
16905 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16906 (TARGET_UPPER_REGS_SF): Likewise.
16907 (TARGET_UPPER_REGS_DI): Likewise.
16908 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16909 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16910 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16911 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16912 (Splitters for DI constants in Altivec registers): Likewise.
16913 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16914 (vsx_set_v4sf_p9): Likewise.
16915 (vsx_set_v4sf_p9_zero): Likewise.
16916 (vsx_insert_extract_v4sf_p9): Likewise.
16917 (vsx_insert_extract_v4sf_p9_2): Likewise.
16919 2017-07-25 Carl Love <cel@us.ibm.com>
16921 * doc/extend.texi: Update the built-in documentation file for the
16922 existing built-in functions
16923 vector signed char vec_cnttz (vector signed char);
16924 vector unsigned char vec_cnttz (vector unsigned char);
16925 vector signed short vec_cnttz (vector signed short);
16926 vector unsigned short vec_cnttz (vector unsigned short);
16927 vector signed int vec_cnttz (vector signed int);
16928 vector unsigned int vec_cnttz (vector unsigned int);
16929 vector signed long long vec_cnttz (vector signed long long);
16930 vector unsigned long long vec_cnttz (vector unsigned long long);
16932 2017-07-25 Andrew Pinski <apinski@cavium.com>
16934 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16935 accesses where the use is for the first operand of a BIT_INSERT.
16937 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16940 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16941 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16943 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16945 * config/i386/gstabs.h: Delete.
16946 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16948 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16950 * config/i386/i386.c (ix86_decompose_address): Do not check for
16951 register RTX when looking at index_reg or base_reg.
16952 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16954 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16956 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16957 to update EH info here.
16959 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16961 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16963 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16965 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16967 2017-07-25 Torsten Duwe <duwe@suse.de>
16969 * common.opt: Introduce -fpatchable-function-entry
16970 command line option, and its variables function_entry_patch_area_size
16971 and function_entry_patch_area_start.
16972 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16973 including a two-value parser.
16974 * target.def (print_patchable_function_entry): New target hook.
16975 * targhooks.h (default_print_patchable_function_entry): New function.
16976 * targhooks.c (default_print_patchable_function_entry): Likewise.
16977 * toplev.c (process_options): Switch off IPA-RA if
16978 patchable function entries are being generated.
16979 * varasm.c (assemble_start_function): Look at the
16980 patchable-function-entry command line switch and current
16981 function attributes and maybe generate NOP instructions by
16982 calling the print_patchable_function_entry hook.
16983 * doc/extend.texi: Document patchable_function_entry attribute.
16984 * doc/invoke.texi: Document -fpatchable_function_entry
16985 command line option.
16986 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16988 * doc/tm.texi: Re-generate.
16990 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16993 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16994 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16996 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16998 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17000 2017-07-25 Richard Biener <rguenther@suse.de>
17002 PR tree-optimization/81455
17003 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17004 not walk in cycles when looking for guards.
17006 2017-07-25 Richard Biener <rguenther@suse.de>
17008 PR tree-optimization/81529
17009 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17010 when optimizing backedge uses.
17012 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17014 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17016 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17017 to dl_section_ref. On AIX, append an expression to subtract
17018 the size of the section length to dl_section_ref.
17020 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17022 * configure.ac: If any of the config.* scripts fail, exit 1.
17023 * configure: Regenerate.
17025 2017-07-25 Richard Biener <rguenther@suse.de>
17027 PR middle-end/81546
17028 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17029 of immediate uses, be more verbose on errors.
17031 2017-07-25 Richard Biener <rguenther@suse.de>
17033 PR tree-optimization/81510
17034 * tree-vect-loop.c (vect_is_simple_reduction): When the
17035 reduction stmt is not inside the loop bail out.
17037 2017-07-25 Richard Biener <rguenther@suse.de>
17039 PR tree-optimization/81303
17040 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17041 profitability check against LOOP_VINFO_NITERSM1.
17043 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17045 * domwalk.c (cmp_bb_postorder): Simplify.
17046 (sort_bbs_postorder): New function. Use it...
17047 (dom_walker::walk): ...here to optimize common cases.
17049 2017-07-25 Martin Liska <mliska@suse.cz>
17052 * ipa-visibility.c (function_and_variable_visibility): Make the
17053 redirection just on target that supports aliasing.
17054 Fix GNU coding style.
17056 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17059 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17061 * config/arm/rtems.h: Likewise.
17062 * config/bfin/rtems.h: Likewise.
17063 * config/i386/rtemself.h: Likewise.
17064 * config/lm32/rtems.h: Likewise.
17065 * config/m32c/rtems.h: Likewise.
17066 * config/m68k/rtemself.h: Likewise.
17067 * config/microblaze/rtems.h: Likewise.
17068 * config/mips/rtems.h: Likewise.
17069 * config/moxie/rtems.h: Likewise.
17070 * config/nios2/rtems.h: Likewise.
17071 * config/powerpcspe/rtems.h: Likewise.
17072 * config/rs6000/rtems.h: Likewise.
17073 * config/rtems.h: Likewise.
17074 * config/sh/rtems.h: Likewise.
17075 * config/sh/rtemself.h: Likewise.
17076 * config/sparc/rtemself.h: Likewise.
17078 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17081 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17082 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17083 * tree-ssa-structalias.c (alias_get_name): Same.
17085 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17088 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17089 instructions if no du chain is found.
17091 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17093 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17095 2017-07-25 Richard Biener <rguenther@suse.de>
17097 PR middle-end/81505
17098 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17101 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17103 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17104 upper-regs options.
17105 (ISA_2_7_MASKS_SERVER): Likewise.
17106 (ISA_3_0_MASKS_IEEE): Likewise.
17107 (OTHER_P8_VECTOR_MASKS): Likewise.
17108 (OTHER_VSX_VECTOR_MASKS): Likewise.
17109 (POWERPC_MASKS): Likewise.
17110 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17111 duplicate list of options.
17112 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17113 explicit -mupper-regs options.
17114 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17115 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17116 alias for -mupper-regs-df.
17117 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17118 (rs6000_init_hard_regno_mode_ok): Likewise.
17119 (rs6000_option_override_internal): Likewise.
17120 (rs6000_opt_masks): Likewise.
17121 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17122 options in terms of whether -mvsx or -mpower8-vector was used.
17123 (TARGET_UPPER_REGS_DI): Likewise.
17124 (TARGET_UPPER_REGS_SF): Likewise.
17125 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17126 -mupper-regs-* options.
17128 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17130 * passes.c (emergency_dump_function): Print some empty lines and a
17131 header before the RTL dump.
17133 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17135 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17137 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17140 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17141 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17143 2017-07-24 Carl Love <cel@us.ibm.com>
17145 * config/rs6000/rs6000-c.c: Add support for built-in functions
17146 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17147 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17148 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17149 vec_extract_fp_from_shortl): Add defines for the two builtins.
17150 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17151 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17153 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17154 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17155 * doc/extend.texi: Update the built-in documentation file for the
17156 new built-in function.
17158 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17161 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17163 * doc/generic.texi: Likewise.
17164 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17165 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17167 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17169 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17170 (aarch64_mls_elt_merge<mode>): Likewise.
17172 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17174 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17175 having __cxa_atexit.
17177 2017-07-23 Michael Collison <michael.collison@arm.com>
17179 * config/arm/arm.c (arm_option_override): Deprecate
17180 use of -mstructure-size-boundary.
17181 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17182 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17184 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17187 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17188 Reduce cost estimate for direct moves.
17190 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17193 * config/i386/i386.c (ix86_option_override_internal): Disable
17194 BMI, BMI2 and TBM instructions for -m16.
17196 2017-07-21 Carl Love <cel@us.ibm.com>
17198 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17199 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17200 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17201 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17202 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17203 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17204 VMULOSW): New enum "unspec" values.
17205 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17206 altivec_vmulosw): New patterns.
17207 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17208 VMULOSW): Add definitions.
17210 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17212 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17213 (qdf24xx): Likewise.
17214 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17215 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17216 (AARCH64_FL_V8_1): Renumber.
17217 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17218 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17219 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17220 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17221 rdma to feature modifiers list.
17223 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17225 PR middle-end/56727
17226 * ipa-visibility (function_and_variable_visibility): Convert
17227 recursive PLT call to direct call if appropriate.
17229 2017-07-21 Andrew Pinski <apinski@cavium.com>
17231 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17232 operand 1 to see if the types precision matches.
17233 * fold-const.c (operand_equal_p): Likewise.
17235 2017-07-21 Richard Biener <rguenther@suse.de>
17237 PR tree-optimization/81303
17238 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17239 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17240 (vect_peeling_hash_get_lowest_cost): Adjust.
17241 (vect_enhance_data_refs_alignment): Likewise. Use
17242 vect_get_peeling_costs_all_drs to compute the penalty for no
17243 peeling to match up costs.
17245 2017-07-21 Richard Biener <rguenther@suse.de>
17247 PR tree-optimization/81500
17248 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17249 we didn't identify a reduction path.
17251 2017-07-21 Tom de Vries <tom@codesourcery.com>
17252 Cesar Philippidis <cesar@codesourcery.com>
17254 PR gcov-profile/81442
17255 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17258 2017-07-21 Tom de Vries <tom@codesourcery.com>
17261 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17263 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17264 nvptx_override_options_after_change.
17266 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17268 * dwarf2out.c (output_file_names): Avoid double testing for
17269 dwarf_version >= 5.
17271 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17273 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17275 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17277 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17279 (try_crossjump_to_edge): Do not punt on partitioned functions.
17281 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17283 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17284 Put all BBs reachable only via paths crossing cold region to cold
17286 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17288 2016-07-21 Richard Biener <rguenther@suse.de>
17290 PR tree-optimization/81303
17291 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17292 into account prologue and epilogue iterations when raising
17293 min_profitable_iters to sth at least covering one vector iteration.
17295 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17297 * config/arm/arm.c (arm_test_cpu_arch_dat):
17300 2017-07-20 Nathan Sidwell <nathan@acm.org>
17302 Remove TYPE_METHODS.
17303 * tree.h (TYPE_METHODS): Delete.
17304 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17305 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17306 (dbxout_type_methods): Scan TYPE_FIELDS.
17307 (dbxout_type): Don't check TYPE_METHODS here.
17308 * function.c (use_register_for_decl): Always ignore register for
17309 class types when not optimizing.
17310 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17311 * tree.c (free_lang_data_in_type): Stitch out member functions and
17312 templates from TYPE_FIELDS.
17313 (build_distinct_type_copy, verify_type_variant,
17314 verify_type): Member fns are on TYPE_FIELDS.
17315 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17316 * tree-pretty-print.c (dump_generic_node): Likewise.
17318 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17321 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17322 V2TImode and V4TImode.
17323 (ix86_expand_vector_extract): Likewise.
17324 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17325 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17326 (ssescalarmode): Handle V4TImode and V2TImode.
17327 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17328 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17329 (VEXTRACTI128_MODE): New mode iterator.
17330 (splitter for *vec_extractv?ti first element): New.
17331 (VEC_INIT_MODE): New mode iterator.
17332 (vec_init<mode>): Consolidate 3 expanders into one using
17333 VEC_INIT_MODE mode iterator.
17335 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17337 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17338 non_spilled_static_chain_regno_p.
17340 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17342 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17344 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17346 * bb-reorder.c (connect_traces): Allow copying of blocks within
17349 2017-07-20 Richard Biener <rguenther@suse.de>
17351 * gimple.h (gimple_phi_result): Add gphi * overload.
17352 (gimple_phi_result_ptr): Likewise.
17353 (gimple_phi_arg): Likewise. Adjust index assert to only
17354 allow actual argument accesses rather than all slots available
17356 (gimple_phi_arg_def): Add gphi * overload.
17357 * tree-phinodes.c (make_phi_node): Initialize only actual
17359 (resize_phi_node): Clear memory not covered by old node,
17360 do not initialize excess argument slots.
17361 (reserve_phi_args_for_new_edge): Initialize new argument slot
17364 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17366 PR tree-optimization/81388
17368 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17370 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17371 by removing computation of may_be_zero.
17373 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17374 Tom de Vries <tom@codesourcery.com>
17376 PR middle-end/81030
17377 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17378 when gimple level profile disagrees with what RTL expander did.
17380 2017-07-20 Richard Biener <rguenther@suse.de>
17382 PR tree-optimization/61171
17383 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17384 (vect_analyze_stmt): Add slp instance parameter.
17385 (vectorizable_reduction): Likewise.
17386 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17387 (vect_is_simple_reduction): Deal with chains not detected
17388 as SLP reduction chain, specifically not properly associated
17389 chains containing a mix of plus/minus.
17390 (get_reduction_op): Remove.
17391 (get_initial_defs_for_reduction): Simplify, pass in whether
17392 this is a reduction chain, pass in the SLP node for the PHIs.
17393 (vect_create_epilog_for_reduction): Get the SLP instance as
17395 (vectorizable_reduction): Get the SLP instance as arg.
17396 During analysis remember the SLP node with the PHIs in the
17397 instance. Simplify getting at the vectorized reduction PHIs.
17398 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17399 through SLP instance.
17400 (vect_slp_analyze_operations): Likewise.
17401 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17402 (vect_transform_stmt): Likewise.
17404 2017-07-20 Tom de Vries <tom@codesourcery.com>
17406 PR tree-optimization/81489
17407 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17408 read of phi arg location to before loop that modifies phi.
17410 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17412 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17415 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17417 PR middle-end/81331
17418 * except.c (execute): Fix ordering issue.
17420 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17422 PR rtl-optimization/81423
17423 * combine.c (make_compound_operation_int): Don't try to optimize
17424 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17426 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17428 PR rtl-optimization/81423
17429 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17430 with a constant that is -1 in the truncated to mode.
17432 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17434 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17435 (determine_unlikely_bbs): ... here.
17436 * predict.h (propagate_unlikely_bbs_forward): Declare.
17437 * cfgexpand.c (pass_expand::execute): Use it.
17438 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17440 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17441 propagate_unlikely_bbs_forward.
17443 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17445 PR middle-end/81331
17446 * except.c (maybe_add_nop_after_section_switch): New function.
17449 2017-07-19 Tom de Vries <tom@codesourcery.com>
17451 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17453 2017-07-19 Tom de Vries <tom@codesourcery.com>
17455 * gimple.h (gimple_phi_arg): Make assert more strict.
17457 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17459 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17460 * config/rs6000/mmintrin.h: New file.
17461 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17463 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17465 PR tree-optimization/81346
17466 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17468 2017-07-19 Tom de Vries <tom@codesourcery.com>
17470 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17472 2017-07-19 Tom de Vries <tom@codesourcery.com>
17474 * config/nvptx/nvptx-modes.def: Add V2DImode.
17475 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17476 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17477 (nvptx_output_mov_insn): Handle lack of mov.b128.
17478 (nvptx_print_operand): Handle 'H' and 'L' codes.
17479 (nvptx_vector_mode_supported): Allow V2DImode.
17480 (nvptx_preferred_simd_mode): New function.
17481 (nvptx_data_alignment): New function.
17482 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17483 nvptx_preferred_simd_mode.
17484 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17486 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17488 2017-07-19 Tom de Vries <tom@codesourcery.com>
17490 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17491 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17492 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17493 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17494 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17495 (mov<VECIM>_insn): New define_insn.
17496 (define_expand "mov<VECIM>): New define_expand.
17498 2017-07-19 Tom de Vries <tom@codesourcery.com>
17500 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17502 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17504 PR tree-optimization/81346
17505 * fold-const.h (fold_div_compare, range_check_type): Declare.
17506 * fold-const.c (range_check_type): New function.
17507 (build_range_check): Use range_check_type.
17508 (fold_div_compare): No longer static, rewritten into
17509 a match.pd helper function.
17510 (fold_comparison): Don't call fold_div_compare here.
17511 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17512 as helper function.
17514 2017-07-19 Nathan Sidwell <nathan@acm.org>
17516 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17517 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17518 * tree.c (find_decls_types_r, verify_type): Use
17519 TYPE_{MIN,MAX}_VALUE_RAW.
17520 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17521 (hash_tree): Likewise.
17522 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17524 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17527 2017-07-18 Tom de Vries <tom@codesourcery.com>
17529 PR middle-end/81464
17530 * omp-expand.c (expand_omp_for_static_chunk): Handle
17531 equal-argument loop exit phi.
17533 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17536 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17537 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17538 operand 2 predicate.
17539 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17540 operand 2 predicate.
17541 (ror,rol -> rorx splitters): Use const_int_operand as
17542 operand 2 predicate.
17544 2017-06-18 Richard Biener <rguenther@suse.de>
17546 PR tree-optimization/81410
17547 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17548 the gap in the ! slp_perm SLP case after each group.
17550 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17552 PR middle-end/81463
17553 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17556 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17558 PR middle-end/81462
17559 * predict.c (set_even_probabilities): Cleanup; do not affect
17560 probabilities that are already known.
17561 (combine_predictions_for_bb): Call even when count is set.
17563 2017-07-18 Nathan Sidwell <nathan@acm.org>
17565 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17568 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17571 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17572 optimization for loop niter analysis.
17574 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17577 * config/avr/avr.c (avr_optimize_casesi): Don't use
17578 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17580 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17582 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17583 body_cost_vec from _vect_peel_extended_info.
17584 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17585 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17588 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17590 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17592 2017-07-18 Richard Biener <rguenther@suse.de>
17594 PR tree-optimization/80620
17595 PR tree-optimization/81403
17596 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17597 info when re-using a VN table entry.
17599 2017-07-18 Richard Biener <rguenther@suse.de>
17601 PR tree-optimization/81418
17602 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17603 vectype_in. Verify that with lane-reducing reduction operations
17604 we have a single def-use cycle.
17606 2017-07-17 Carl Love <cel@us.ibm.com>
17608 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17610 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17611 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17612 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17613 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17614 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17615 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17616 VMULOSW): New enum "unspec" values.
17617 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17618 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17619 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17620 altivec_vmulosw): New patterns.
17621 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17622 VMULOSW): Add definitions.
17623 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17625 * config/alpha/alpha.c: Include predict.h.
17627 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17629 * tree-vrp.c (compare_assert_loc): Fix comparison function
17630 to return predictable results.
17632 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17634 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17636 (subdi3): Likewise.
17637 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17638 * doc/invoke.texi (mexpand-adddi): Update text.
17640 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17642 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17643 that also clobbers the CC register. The old expand code is moved
17645 (*arc_clzsi2): ... here.
17646 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17647 the CC register. The old expand code is moved to ...
17648 (arc_ctzsi2): ... here.
17650 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17652 * config/arc/arc.opt (mindexed-loads): Use initial value
17653 TARGET_INDEXED_LOADS_DEFAULT.
17654 (mauto-modify-reg): Use initial value
17655 TARGET_AUTO_MODIFY_REG_DEFAULT.
17656 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17657 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17658 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17659 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17661 2017-07-17 Martin Liska <mliska@suse.cz>
17664 * opts.c (finish_options): Do not allow -fgnu-tm
17665 w/ -fsanitize={kernel-,}address. Say sorry.
17667 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17670 * tree-loop-distribution.c (classify_partition): Only assert on
17671 numer of iterations.
17672 (merge_dep_scc_partitions): Delete prameter. Update function call.
17673 (distribute_loop): Remove code handling loop with unknown niters.
17674 (pass_loop_distribution::execute): Skip loop with unknown niters.
17676 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17679 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17680 function sort_partitions_by_post_order.
17682 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17684 PR tree-optimization/81374
17685 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17686 the max index of basic blocks, rather than number of basic blocks.
17688 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17690 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17692 (arc_legitimate_pic_operand_p): Likewise.
17693 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17695 (arc_needs_pcl_p): Likewise.
17696 (arc_legitimate_pc_offset_p): Likewise.
17697 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17698 function is also used in constrains.md.
17699 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17700 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17701 PLUS. Only return true/false in known cases, otherwise assert.
17702 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17703 is already called in arc_legitimate_constant_p.
17704 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17706 (LEGITIMATE_PIC_OPERAND_P): Use
17707 arc_raw_symbolic_reference_mentioned_p function.
17708 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17713 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17714 Andrew Burgess <andrew.burgess@embecosm.com>
17716 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17717 (arc_return_address_register): New function.
17718 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17719 (arc_handle_fndecl_attribute): Add naked attribute.
17720 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17721 (TARGET_WARN_FUNC_RETURN): Likewise.
17722 (arc_allocate_stack_slots_for_args): New function.
17723 (arc_warn_func_return): Likewise.
17724 (machine_function): Change type fn_type.
17725 (arc_compute_function_type): Consider new naked function type,
17726 change function return type.
17727 (arc_must_save_register): Adapt to handle new
17728 arc_compute_function_type's return type.
17729 (arc_expand_prologue): Likewise.
17730 (arc_expand_epilogue): Likewise.
17731 (arc_return_address_regs): Delete.
17732 (arc_return_address_register): New function.
17733 (arc_epilogue_uses): Use above function.
17734 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17735 (arc_function_type): Change encoding, add naked type.
17736 (ARC_INTERRUPT_P): Change to handle the new encoding.
17737 (ARC_FAST_INTERRUPT_P): Likewise.
17738 (ARC_NORMAL_P): Define.
17739 (ARC_NAKED_P): Likewise.
17740 (arc_compute_function_type): Delete prototype.
17741 * config/arc/arc.md (in_ret_delay_slot): Use
17742 arc_return_address_register function.
17743 (simple_return): Likewise.
17744 (p_return_i): Likewise.
17746 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17748 PR tree-optimization/81428
17749 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17750 can't be built for those types.
17752 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17754 Remove stuff dead since r239246.
17756 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17757 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17758 (avr_inform_devices): Remove dead stuff.
17760 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17762 * config/arm/arm_neon.h: Fix softp typo.
17764 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17766 PR tree-optimization/81365
17767 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17768 aggregate moves onto bb predecessor edges, make sure there are no
17769 loads that could alias the lhs in between the start of bb and the
17772 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17775 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17776 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17777 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17779 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17781 PR tree-optimization/81396
17782 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17783 (init_symbolic_number): Initialize it to 1.
17784 (perform_symbolic_merge): Add n_ops from both operands into the new
17786 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17787 without base_addr as useless if they need more than one operation.
17788 (bswap_replace): Handle !bswap case for NULL base_addr.
17790 2017-07-17 Tom de Vries <tom@codesourcery.com>
17793 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17796 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17798 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17799 conditional builtin define __FIX_LEON3FT_B2BST.
17801 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17803 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17804 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17807 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17809 PR rtl-optimization/81424
17810 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17811 to remove potential trapping from operands if -fnon-call-exceptions.
17813 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17815 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17816 profile_proability for scalling.
17817 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17819 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17821 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17823 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17825 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17826 fixpoint arithmetics.
17828 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17830 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17831 fixpoint arithmetics.
17833 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17835 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17836 fixpoint arithmetics.
17838 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17840 * profile-count.h (profile_probability::from_reg_br_prob_note,
17841 profile_probability::to_reg_br_prob_note): New functions.
17842 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17843 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17844 * predict.c (probability_reliable_p): Update.
17845 (edge_probability_reliable_p): Update.
17846 (br_prob_note_reliable_p): Update.
17847 (invert_br_probabilities): Update.
17848 (add_reg_br_prob_note): New function.
17849 (combine_predictions_for_insn): Update.
17850 * asan.c (asan_clear_shadow): Update.
17851 * cfgbuild.c (compute_outgoing_frequencies): Update.
17852 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17853 (update_br_prob_note): Update.
17854 (rtl_verify_edges): Update.
17855 (purge_dead_edges): Update.
17856 (fixup_reorder_chain): Update.
17857 * emit-rtl.c (try_split): Update.
17858 * ifcvt.c (cond_exec_process_insns): Update.
17859 (cond_exec_process_if_block): Update.
17860 (dead_or_predicable): Update.
17861 * internal-fn.c (expand_addsub_overflow): Update.
17862 (expand_neg_overflow): Update.
17863 (expand_mul_overflow): Update.
17864 * loop-doloop.c (doloop_modify): Update.
17865 * loop-unroll.c (compare_and_jump_seq): Update.
17866 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17867 * predict.h: Update.
17868 * reorg.c (mostly_true_jump): Update.
17870 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17871 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17872 * config/arc/arc.c: (emit_unlikely_jump): Update.
17873 * config/arm/arm.c: (emit_unlikely_jump): Update.
17874 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17875 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17876 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17877 (ix86_print_operand): Update.
17878 (ix86_split_fp_branch): Update.
17879 (predict_jump): Update.
17880 * config/ia64/ia64.c (ia64_print_operand): Update.
17881 * config/mmix/mmix.c (mmix_print_operand): Update.
17882 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17883 (rs6000_expand_split_stack_prologue): Update.
17884 * config/rs6000/rs6000.c: Update.
17885 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17886 (s390_expand_vec_movstr): Update.
17887 (s390_expand_cs_tdsi): Update.
17888 (s390_expand_split_stack_prologue): Update.
17889 * config/sh/sh.c (sh_print_operand): Update.
17890 (expand_cbranchsi4): Update.
17891 (expand_cbranchdi4): Update.
17892 * config/sparc/sparc.c (output_v9branch): Update.
17893 * config/spu/spu.c (get_branch_target): Update.
17894 (ea_load_store_inline): Update.
17895 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17896 * config/tilepro/tilepro.c: Update.
17898 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17900 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17901 (gimplify_save_expr): Likewise.
17903 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17905 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17907 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17909 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17911 * timevar.def (TV_IPA_FNSUMMARY): Define.
17913 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17915 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17916 to back store errata sensitive sequence from being generated.
17917 (sqrtdf2_fix): Likewise.
17919 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17921 * tree-ssa-threadupdate.c (compute_path_counts,
17922 update_joiner_offpath_counts): Use profile_probability.
17924 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17927 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17929 * config/arm/arm-c.c (arm_cpu_builtins): Define
17930 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17932 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17934 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17935 array entries to represent __ieee128 versions of the
17936 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17937 scalar_extract_sig, and scalar_insert_exp built-in functions.
17938 (altivec_resolve_overloaded_builtin): Add special case handling
17939 for the __builtin_scalar_insert_exp function, as represented by
17940 the P9V_BUILTIN_VEC_VSIEDP constant.
17941 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17942 exponent support for __ieee128 argument.
17943 (VSESQP): Add scalar extract signature support for __ieee128
17945 (VSTDCNQP): Add scalar test negative support for __ieee128
17947 (VSIEQP): Add scalar insert exponent support for __int128 argument
17948 with __ieee128 result.
17949 (VSIEQPF): Add scalar insert exponent support for __ieee128
17950 argument with __ieee128 result.
17951 (VSTDCQP): Add scalar test data class support for __ieee128
17953 (VSTDCNQP): Add overload support for scalar test negative with
17954 __ieee128 argument.
17955 (VSTDCQP): Add overload support for scalar test data class
17956 __ieee128 argument.
17957 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17958 UNSPEC_VSX_SXSIGDP.
17959 (UNSPEC_VSX_SIEXPQP): New constant.
17960 (xsxexpqp): New insn for VSX scalar extract exponent quad
17962 (xsxsigqp): New insn for VSX scalar extract significand quad
17964 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17965 precision with floating point argument.
17966 (xststdcqp): New expand for VSX scalar test data class quad
17968 (xststdcnegqp): New expand for VSX scalar test negative quad
17970 (xststdcqp): New insn to match expansions for VSX scalar test data
17971 class quad precision and VSX scalar test negative quad precision.
17972 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17973 special case operand checking to enforce that second operand of
17974 VSX scalar test data class with quad precision argument is a 7-bit
17976 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17977 prototypes and descriptions of __ieee128 versions of
17978 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17979 scalar_test_data_class, and scalar_test_neg built-in functions.
17981 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17983 PR tree-optimization/81162
17984 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17985 replace a negate with an add.
17987 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17989 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17991 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17993 * config/arm/arm-c.c (arm_cpu_builtins): Define
17994 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17996 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17998 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17999 (armv8-r): Set ARM Cortex-R52 as default CPU.
18000 * config/arm/arm-tables.opt: Regenerate.
18001 * config/arm/arm-tune.md: Regenerate.
18002 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18004 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18005 extension for -mcpu=cortex-r52.
18007 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18009 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18010 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18011 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18012 (fp-armv8): Define it as FP_ARMv8 only.
18013 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18014 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18016 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18017 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18018 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18019 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18020 than TARGET_FPU_ARMV8.
18021 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18022 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18023 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18025 * config/arm/neon.md (neon_vrint): Likewise.
18026 (neon_vcvt): Likewise.
18027 (neon_<fmaxmin_op><mode>): Likewise.
18028 (<fmaxmin><mode>3): Likewise.
18029 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18030 * config/arm/predicates.md (arm_cond_move_operator): Check against
18031 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18033 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18035 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18036 to top of function.
18038 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18040 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18041 loop in comment with memset.
18043 2017-07-14 Martin Liska <mliska@suse.cz>
18045 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18046 * dwarf2out.c (is_java): Remove the function.
18047 (output_pubname): Remove usage of the function.
18048 (lower_bound_default): Remove usage of DW_LANG_Java.
18049 (gen_compile_unit_die): Likewise.
18050 * gcc.c: Remove compiler defaults for .java and .zip files.
18051 * gimple-expr.c (remove_suffix): Change as there's no longer
18052 extension than 4-letter one.
18053 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18054 (gimplify_save_expr): Likewise.
18055 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18056 as it's possible even for other languages than Java.
18057 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18058 * lto-opts.c (lto_write_options): Remove reference to Java.
18059 * opts.c (strip_off_ending): Update file extension handling.
18060 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18061 * tree-eh.c (lower_resx): Likewise.
18062 * tree.c (free_lang_data_in_type): Remove dead code.
18063 (find_decls_types_r): Likewise.
18064 (build_common_builtin_nodes): Remove Java from a comment.
18065 (verify_type): Remove dead code.
18066 * varasm.c (assemble_external): Remove Java from a comment.
18068 2017-07-14 Martin Liska <mliska@suse.cz>
18070 * opts.c (finish_options): Add quotes.
18071 (common_handle_option): Likewise.
18073 2017-07-14 Martin Liska <mliska@suse.cz>
18075 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18076 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18077 Remove N_SO_PASCAL.
18078 * dwarf2out.c (lower_bound_default): Do not handle
18080 (gen_compile_unit_die): Likewise.
18081 * gcc.c: Remove default extension binding for GNU Pascal.
18082 * stmt.c: Remove Pascal language from a comment.
18083 * xcoffout.c: Likewise.
18085 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18088 * diagnostic-show-locus.c (fixit_cmp): New function.
18089 (layout::layout): Sort m_fixit_hints.
18090 (column_range::column_range): Assert that the values are valid.
18091 (struct char_span): New struct.
18092 (correction::overwrite): New method.
18093 (struct source_line): New struct.
18094 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18095 calls in terms of classes source_line and char_span, and
18096 correction::overwrite.
18097 (selftest::test_overlapped_fixit_printing_2): New function.
18098 (selftest::diagnostic_show_locus_c_tests): Call it.
18100 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18102 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18103 early if there is no lhs.
18105 2017-07-13 Martin Liska <mliska@suse.cz>
18107 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18108 (gen_reference_type_die): Likewise.
18109 * stor-layout.c: Remove Pascal-related comment.
18111 2017-07-13 Martin Liska <mliska@suse.cz>
18113 * opts.c (finish_options): Add quotes to error messages.
18114 (parse_sanitizer_options): Likewise.
18116 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18118 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18120 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18122 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18124 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18126 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18128 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18130 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18133 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18134 provides the hardware capability bits, define the macro
18135 __BUILTIN_CPU_SUPPORTS__.
18136 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18137 if GLIBC does not provide the hardware capability bits. Add a
18138 gcc_unreachable call if the built-in cpu function is neither
18139 __builtin_cpu_is nor __builtin_cpu_supports.
18140 (rs6000_get_function_versions_dispatcher): Change the warning
18141 that an old GLIBC is used which does not export the capability
18142 bits to be an error.
18143 * doc/extend.texi (target_clones attribute): Document the
18144 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18145 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18146 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18147 the macros defined by GCC if the newer GLIBC is available.
18149 2017-07-12 Jeff Law <law@redhat.com>
18151 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18152 remaining includes slightly.
18153 * config/riscv/riscv-builtins.c: Include profile-count.h.
18155 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18158 * config/avr/avr.c (avr_set_current_function): In diagnostic
18159 messages: Quote keywords and (parts of) identifiers.
18160 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18163 2017-07-12 Carl Love <cel@us.ibm.com>
18165 * config/rs6000/rs6000-c.c: Add support for built-in functions
18166 vector bool char vec_revb (vector bool char);
18167 vector bool short vec_revb (vector short char);
18168 vector bool int vec_revb (vector bool int);
18169 vector bool long long vec_revb (vector bool long long);
18170 * doc/extend.texi: Update the built-in documentation file for the
18171 new built-in functions.
18173 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18175 * config/s390/s390.md: Remove movcc splitter.
18177 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18179 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18180 load/store on condition.
18182 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18185 * config/avr/avr.c (avr_encode_section_info)
18186 [progmem && !TREE_READONLY]: Error if progmem object needs
18189 2017-07-11 Michael Collison <michael.collison@arm.com>
18191 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18194 2017-07-11 Carl Love <cel@us.ibm.com>
18196 * config/rs6000/rs6000-c.c: Add support for builtins
18197 vector unsigned int vec_parity_lsbb (vector signed int);
18198 vector unsigned int vec_parity_lsbb (vector unsigned int);
18199 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18200 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18201 vector unsigned long long vec_parity_lsbb (vector signed long long);
18202 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18203 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18204 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18205 * doc/extend.texi: Update the built-in documentation file for the
18206 new built-in functions.
18208 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18210 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18211 (layout::m_primary_loc): New field.
18212 (layout::layout): Initialize new field. Move location filtering
18213 logic from here to...
18214 (layout::maybe_add_location_range): ...this new method. Add
18215 support for filtering to just the lines already specified by other
18217 (layout::will_show_line_p): New method.
18218 (gcc_rich_location::add_location_if_nearby): New method.
18219 (selftest::test_add_location_if_nearby): New test function.
18220 (selftest::diagnostic_show_locus_c_tests): Call it.
18221 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18224 2017-07-11 Tom de Vries <tom@codesourcery.com>
18226 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18227 (bb_first_real_insn): New function.
18228 (nvptx_single): Add extra initialization of broadcasted condition
18231 2017-07-11 Nathan Sidwell <nathan@acm.org>
18233 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18235 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18237 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18238 Binutils doc as TEXI will mess them up.
18239 * doc/invoke.texi (AVR Options): Same here.
18241 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18243 * config/sparc/sparc.opt (mfix-ut700): New option.
18244 (mfix-gr712rc): Likewise.
18245 (sparc_fix_b2bst): New variable.
18246 * doc/invoke.texi (SPARC options): Document them.
18247 (ARM options): Fix warnings.
18248 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18249 instructions to prevent sequences that can trigger the store-store
18250 errata for certain LEON3FT processors.
18251 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18252 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18253 * config/sparc/sparc.md (fix_b2bst): New attribute.
18254 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18256 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18259 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18261 (*rsqrtsf2_sse): Ditto.
18263 (div<mode>3): Macroize insn from divdf3 and divsf3
18264 using MODEF mode iterator.
18266 2017-07-10 Martin Sebor <msebor@redhat.com>
18268 PR tree-optimization/80397
18269 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18270 instead of testing for equality to INTEGER_TYPE.
18272 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18274 * config.gcc: Remove uclibc from arc target spec.
18276 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18278 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18280 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18283 * lto-wrapper.c (remove_option): New function.
18284 (merge_and_complain): Merge PIC/PIE options more realistically.
18286 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18288 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18292 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18293 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18294 * config.in: Regenerate.
18295 * configure: Regenerate.
18296 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18297 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18298 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18299 (TARGET_GASISR_PROLOGUES): ...target mask.
18300 * common/config/avr/avr-common.c
18301 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18302 Set -mgas-isr-prologues.
18303 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18304 INSERT_PASS_BEFORE for it.
18305 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18306 * config/avr/avr.c (avr_option_override)
18307 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18308 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18309 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18310 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18311 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18313 (make_avr_pass_pre_proep): New function.
18314 (emit_push_sfr) <treg>: Add argument to function and use it
18315 instead of TMP_REG.
18316 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18317 and set machine->gasisr.yes.
18318 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18319 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18320 __gcc_isr.n_pushed to .L__stack_usage.
18321 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18322 (avr_asm_final_postscan_insn): ...this new static function.
18323 * config/avr/avr.h (machine_function)
18324 <is_no_gccisr, use_L__stack_usage>: New fields.
18325 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18326 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18327 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18328 (gasisr, *gasisr): New expander and insn.
18329 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18330 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18331 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18333 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18335 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18338 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18340 Move jump-tables out of .text again.
18343 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18344 (ASM_OUTPUT_ADDR_VEC): New function.
18345 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18346 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18347 INSN_ADDRESSes as asm comment.
18348 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18349 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18350 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18351 * config/avr/avr.md (*tablejump): Adjust comment.
18352 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18353 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18355 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18356 (avr_output_addr_vec): New proto.
18357 (avr_log_t) <insn_addresses>: New field.
18359 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18362 * config/i386/i386.c (ix86_function_arg_advance): Set
18363 outgoing_args_on_stack to true if there are outgoing arguments
18365 (ix86_function_arg): Likewise.
18366 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18367 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18368 * config/i386/i386.h (machine_function): Add
18369 outgoing_args_on_stack.
18371 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18373 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18374 supporting pthreds.
18375 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18377 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18379 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18380 (REAL_H): Remove $(MACHMODE_H).
18381 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18383 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18384 $(MACHMODE_H) and double-int.h.
18385 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18387 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18390 2017-07-07 Andrew Pinski <apinski@cavium.com>
18392 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18393 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18395 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18397 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18398 Add warning if GCC was not configured to link against a GLIBC that
18399 exports the hardware capability bits.
18400 (make_resolver_func): Make resolver function private and not a
18401 COMDAT function. Create the name with clone_function_name instead
18402 of make_unique_name.
18405 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18406 correct operand in doing the split.
18408 2017-07-07 Carl Love <cel@us.ibm.com>
18410 * config/rs6000/rs6000-c: Add support for built-in function
18411 vector unsigned short vec_pack_to_short_fp32 (vector float,
18413 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18414 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18415 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18416 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18417 (convert_4f32_8i16): Add define_expand.
18418 * doc/extend.texi: Update the built-in documentation file for the
18419 new built-in function.
18421 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18423 * config/sparc/m8.md: New file.
18424 * config/sparc/sparc.md: Include m8.md.
18426 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18428 * config/sparc/sparc.opt: New option -mvis4b.
18429 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18430 (sparc_option_override): Handle VIS4B.
18431 (enum sparc_builtins): Define
18432 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18433 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18434 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18435 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18436 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18437 (check_constant_argument): New function.
18438 (sparc_vis_init_builtins): Define builtins
18439 __builtin_vis_dictunpack{8,16,32},
18440 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18441 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18442 __builtin_vis_fpcmpde{8,16,32}shl and
18443 __builtin_vis_fpcmpur{8,16,32}shl.
18444 (sparc_expand_builtin): Check that the constant operands to
18445 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18446 constant and in range.
18447 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18449 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18450 (SPARC_IMM5_P): Likewise.
18451 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18452 (enabled): Handle vis4b.
18453 (UNSPEC_DICTUNPACK): New unspec.
18454 (UNSPEC_FPCMPSHL): Likewise.
18455 (UNSPEC_FPUCMPSHL): Likewise.
18456 (UNSPEC_FPCMPDESHL): Likewise.
18457 (UNSPEC_FPCMPURSHL): Likewise.
18458 (cpu_feature): New CPU feature `vis4b'.
18459 (dictunpack{8,16,32}): New insns.
18460 (FPCSMODE): New mode iterator.
18461 (fpcscond): New code iterator.
18462 (fpcsucond): Likewise.
18463 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18464 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18465 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18466 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18467 * config/sparc/constraints.md: Define constraints `q' for unsigned
18468 2-bit integer constants and `t' for unsigned 5-bit integer
18470 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18472 (imm5_operand_dictunpack16): Likewise.
18473 (imm5_operand_dictunpack32): Likewise.
18474 (imm2_operand): Likewise.
18475 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18476 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18477 ditunpack* and fpcmp*shl builtins.
18479 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18481 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18482 * config.in: Add HAVE_AS_SPARC6 define.
18483 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18485 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18487 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18488 (CPP_CPU_SPEC): Handle m8.
18489 (ASM_CPU_SPEC): Likewise.
18490 * config/sparc/sparc-opts.h (enum processor_type): Add
18492 * config/sparc/sparc.c (m8_costs): New struct.
18493 (sparc_option_override): Handle TARGET_CPU_m8.
18494 (sparc32_initialize_trampoline): Likewise.
18495 (sparc64_initialize_trampoline): Likewise.
18496 (sparc_issue_rate): Likewise.
18497 (sparc_register_move_cost): Likewise.
18498 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18499 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18500 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18501 (CPP_CPU_SPEC): Handle M8.
18502 (ASM_CPU_SPEC): Likewise.
18503 (AS_M8_FLAG): Define.
18504 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18505 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18506 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18508 * configure: Regenerate.
18509 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18512 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18514 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18516 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18517 ("*movdi_insn_sp32"): Do not set v3pipe.
18518 ("*movsi_insn"): Likewise.
18519 ("*movdi_insn_sp64"): Likewise.
18520 ("*movsf_insn"): Likewise.
18521 ("*movdf_insn_sp32"): Likewise.
18522 ("*movdf_insn_sp64"): Likewise.
18523 ("*zero_extendsidi2_insn_sp64"): Likewise.
18524 ("*sign_extendsidi2_insn"): Likewise.
18525 ("*mov<VM32:mode>_insn"): Likewise.
18526 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18527 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18528 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18529 ("<vlop:code><VL:mode>3"): Likewise.
18530 ("*not_<vlop:code><VL:mode>3"): Likewise.
18531 ("*nand<VL:mode>_vis"): Likewise.
18532 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18533 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18534 ("one_cmpl<VL:mode>2"): Likewise.
18535 ("faligndata<VM64:mode>_vis"): Likewise.
18536 ("alignaddrsi_vis"): Likewise.
18537 ("alignaddrdi_vis"): Likweise.
18538 ("alignaddrlsi_vis"): Likewise.
18539 ("alignaddrldi_vis"): Likewise.
18540 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18541 ("bmaskdi_vis"): Likewise.
18542 ("bmasksi_vis"): Likewise.
18543 ("bshuffle<VM64:mode>_vis"): Likewise.
18544 ("cmask8<P:mode>_vis"): Likewise.
18545 ("cmask16<P:mode>_vis"): Likewise.
18546 ("cmask32<P:mode>_vis"): Likewise.
18547 ("pdistn<P:mode>_vis"): Likewise.
18548 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18550 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18552 * config/sparc/sparc.md ("subtype"): New insn attribute.
18553 ("*wrgsr_sp64"): Set insn subtype.
18554 ("*rdgsr_sp64"): Likewise.
18555 ("alignaddrsi_vis"): Likewise.
18556 ("alignaddrdi_vis"): Likewise.
18557 ("alignaddrlsi_vis"): Likewise.
18558 ("alignaddrldi_vis"): Likewise.
18559 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18560 ("fexpand_vis"): Likewise.
18561 ("fpmerge_vis"): Likewise.
18562 ("faligndata<VM64:mode>_vis"): Likewise.
18563 ("bshuffle<VM64:mode>_vis"): Likewise.
18564 ("cmask8<P:mode>_vis"): Likewise.
18565 ("cmask16<P:mode>_vis"): Likewise.
18566 ("cmask32<P:mode>_vis"): Likewise.
18567 ("fchksm16_vis"): Likewise.
18568 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18569 ("fmean16_vis"): Likewise.
18570 ("fp<plusminus_insn>64_vis"): Likewise.
18571 ("<plusminus_insn>v8qi3"): Likewise.
18572 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18573 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18574 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18575 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18576 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18577 ("*movqi_insn"): Likewise.
18578 ("*movhi_insn"): Likewise.
18579 ("*movsi_insn"): Likewise.
18580 ("movsi_pic_gotdata_op"): Likewise.
18581 ("*movdi_insn_sp32"): Likewise.
18582 ("*movdi_insn_sp64"): Likewise.
18583 ("movdi_pic_gotdata_op"): Likewise.
18584 ("*movsf_insn"): Likewise.
18585 ("*movdf_insn_sp32"): Likewise.
18586 ("*movdf_insn_sp64"): Likewise.
18587 ("*zero_extendhisi2_insn"): Likewise.
18588 ("*zero_extendqihi2_insn"): Likewise.
18589 ("*zero_extendqisi2_insn"): Likewise.
18590 ("*zero_extendqidi2_insn"): Likewise.
18591 ("*zero_extendhidi2_insn"): Likewise.
18592 ("*zero_extendsidi2_insn_sp64"): Likewise.
18593 ("ldfsr"): Likewise.
18594 ("prefetch_64"): Likewise.
18595 ("prefetch_32"): Likewise.
18596 ("tie_ld32"): Likewise.
18597 ("tie_ld64"): Likewise.
18598 ("*tldo_ldub_sp32"): Likewise.
18599 ("*tldo_ldub1_sp32"): Likewise.
18600 ("*tldo_ldub2_sp32"): Likewise.
18601 ("*tldo_ldub_sp64"): Likewise.
18602 ("*tldo_ldub1_sp64"): Likewise.
18603 ("*tldo_ldub2_sp64"): Likewise.
18604 ("*tldo_ldub3_sp64"): Likewise.
18605 ("*tldo_lduh_sp32"): Likewise.
18606 ("*tldo_lduh1_sp32"): Likewise.
18607 ("*tldo_lduh_sp64"): Likewise.
18608 ("*tldo_lduh1_sp64"): Likewise.
18609 ("*tldo_lduh2_sp64"): Likewise.
18610 ("*tldo_lduw_sp32"): Likewise.
18611 ("*tldo_lduw_sp64"): Likewise.
18612 ("*tldo_lduw1_sp64"): Likewise.
18613 ("*tldo_ldx_sp64"): Likewise.
18614 ("*mov<VM32:mode>_insn"): Likewise.
18615 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18616 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18618 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18620 * config/sparc/sparc.md ("type"): New insn type viscmp.
18621 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18623 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18624 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18625 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18626 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18628 ("n7_vis_logical_11cycle"): Likewise.
18629 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18630 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18631 * config/sparc/niagara.md ("niag_vis"): Likewise.
18632 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18633 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18635 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18637 * config/sparc/sparc.md: New instruction type `bmask'.
18638 (bmaskdi_vis): Use the `bmask' type.
18639 (bmasksi_vis): Likewise.
18640 * config/sparc/ultra3.md (us3_array): Likewise.
18641 * config/sparc/niagara7.md (n7_array): Likewise.
18642 * config/sparc/niagara4.md (n4_array): Likewise.
18643 * config/sparc/niagara2.md (niag2_vis): Likewise.
18644 (niag3_vis): Likewise.
18645 * config/sparc/niagara.md (niag_vis): Likewise.
18647 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18649 * ipa-comdats.c: Remove optimize check from gate.
18650 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18651 for functions not optimized.
18652 (ipa_fn_summary_read): Skip optimize check.
18653 (ipa_fn_summary_write): Likewise.
18654 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18656 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18658 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18659 (check_callers): Give up on caller which is not optimized.
18660 (inline_small_functions): Likewise.
18661 (ipa_inline): Do not give up when not optimizing.
18662 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18663 away unoptimizes cdtors.
18664 (whole_program_function_and_variable_visibility): Do
18665 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18666 * ipa.c (process_references): Do not check optimize.
18667 (symbol_table::remove_unreachable_nodes): Update optimize check.
18668 (set_writeonly_bit): Update optimize check.
18669 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18670 (pass_ipa_single_use::gate): Remove.
18672 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18674 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18675 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18676 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18677 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18678 permute_load, permute_store, adjust_extract, adjust_splat,
18679 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18680 replace_swap_with_copy, dump_swap_insn_table,
18681 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18682 recombine_lvx_pattern, recombine_stvx_pattern,
18683 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18684 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18685 to file rs6000-p8swap.c.
18686 * config/rs6000/rs6000-p8swap.c: New file.
18687 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18688 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18689 and rs6000*-*-* targets.
18691 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18693 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18695 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18697 * lto-wrapper.c (merge_and_complain): Do not merge
18698 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18699 fsigned_zeros, ftrapping_math, fwrapv.
18700 (append_compiler_options): Do not track these options.
18701 (append_linker_options): Likewie
18703 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18705 * cgraphunit.c (cgraph_node::finalize_function): When
18706 !flag_toplevel_reorde set no_reorder flag.
18707 (varpool_node::finalize_decl): Likewise.
18708 (symbol_table::compile): Drop no toplevel reorder path.
18710 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18712 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18713 edges; zero probability is not better than uninitialized.
18715 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18717 * asan.h (asan_sanitize_allocas_p): Declare.
18718 * asan.c (asan_sanitize_allocas_p): New function.
18719 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18720 (handle_builtin_alloca): Likewise.
18721 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18722 if !asan_sanitize_allocas_p.
18723 * params.def (asan-instrument-allocas): Add new option.
18724 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18725 * opts.c (common_handle_option): Disable allocas sanitization for
18728 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18730 * asan.c: Include gimple-fold.h.
18731 (get_last_alloca_addr): New function.
18732 (handle_builtin_stackrestore): Likewise.
18733 (handle_builtin_alloca): Likewise.
18734 (asan_emit_allocas_unpoison): Likewise.
18735 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18736 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18737 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18738 (instrument_builtin_call): Pass gimple iterator to
18739 get_mem_refs_of_builtin_call.
18740 (last_alloca_addr): New global.
18741 * asan.h (asan_emit_allocas_unpoison): Declare.
18742 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18743 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18744 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18745 if function calls alloca.
18746 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18747 * gimple-fold.h (replace_call_with_value): Declare.
18748 * internal-fn.c: Include asan.h.
18749 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18750 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18752 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18754 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18755 (C_SELFTEST_FLAGS): New.
18756 (CPP_SELFTEST_FLAGS): New.
18757 (SELFTEST_DEPS): New, from deps of s-selftest.
18758 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18759 (CPP_SELFTEST_DEPS): New.
18760 (selftest): Add dependency on s-selftest-c++.
18761 (s-selftest): Rename to...
18762 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18763 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18764 than SELFTEST_FLAGS.
18765 (selftest-gdb): Rename to...
18766 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18768 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18769 (selftest-valgrind): Rename to...
18770 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18772 (selftest-valgrind): Reintroduce as an alias for
18773 selftest-c-valgrind.
18774 (s-selftest-c++): New.
18775 (selftest-c++-gdb): New.
18776 (selftest-c++-valgrind): New.
18778 2017-07-06 Olivier Hainque <hainque@adacore.com>
18780 * gcc.c (process_command): When deciding if undefined variables
18781 should be ignored when processing specs, accept "gcc -v" as well.
18783 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18785 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18786 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18788 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18790 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18791 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18792 * config/arm/arm-tables.opt: Regenerate.
18793 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18795 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18797 2017-07-06 Carl Love <cel@us.ibm.com>
18799 * ChangeLog: Clean up from mid air collision
18801 2017-07-06 Carl Love <cel@us.ibm.com>
18803 * config/rs6000/rs6000-c.c: Add support for built-in functions
18804 vector signed int vec_subc (vector signed int, vector signed int);
18805 vector signed __int128 vec_subc (vector signed __int128,
18806 vector signed __int128);
18807 vector unsigned __int128 vec_subc (vector unsigned __int128,
18808 vector unsigned __int128);
18809 vector signed int vec_sube (vector signed int, vector signed int,
18810 vector signed int);
18811 vector unsigned int vec_sube (vector unsigned int,
18812 vector unsigned int,
18813 vector unsigned int);
18814 vector signed __int128 vec_sube (vector signed __int128,
18815 vector signed __int128,
18816 vector signed__int128);
18817 vector unsigned __int128 vec_sube (vector unsigned __int128,
18818 vector unsigned __int128,
18819 vector unsigned __int128);
18820 vector signed int vec_subec (vector signed int, vector signed int,
18821 vector signed int);
18822 vector unsigned int vec_subec (vector unsigned int,
18823 vector unsigned int,
18824 vector unsigned int);
18825 vector signed __int128 vec_subec (vector signed __int128,
18826 vector signed __int128,
18827 vector signed__int128);
18828 vector unsigned __int128 vec_subec (vector unsigned __int128,
18829 vector unsigned __int128,
18830 vector unsigned __int128);
18831 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18832 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18833 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18834 BU_ALTIVEC_OVERLOAD_X definitions.
18835 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18836 * doc/extend.texi: Update the built-in documentation file for the new
18837 built-in functions.
18839 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18842 * diagnostic-show-locus.c (layout::layout): Use start and finish
18843 spelling location for the start and finish of each range.
18844 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18845 Add unused aspect param.
18846 * input.c (expand_location_1): Add "aspect" param, and use it
18847 to access the correct part of the location.
18848 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18850 (expand_location_to_spelling_point): Likewise.
18851 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18852 param, and pass it to expand_location_1.
18854 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18856 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18857 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18858 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18859 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18860 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18861 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18862 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18863 _mm_maskz_getmant_ss): New intrinsics.
18864 (__builtin_ia32_getexpss128_mask): Changed to ...
18865 __builtin_ia32_getexpss128_round ... this.
18866 (__builtin_ia32_getexpsd128_mask): Changed to ...
18867 __builtin_ia32_getexpsd128_round ... this.
18868 * config/i386/i386-builtin-types.def
18869 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18870 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18871 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18872 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18873 __builtin_ia32_getmantss_mask_round): New builtins.
18874 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18875 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18876 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18877 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18878 * config/i386/sse.md
18879 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18880 avx512f_sgetexp<mode><mask_scalar_name>
18881 <round_saeonly_scalar_name> ... this.
18882 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18883 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18884 vgetexp<ssescalarmodesuffix>
18885 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18886 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18887 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18888 avx512f_vgetmant<mode><mask_scalar_name>
18889 <round_saeonly_scalar_name> ... this.
18890 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18891 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18892 vgetmant<ssescalarmodesuffix>
18893 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18894 %0<mask_scalar_operand4>, %1, %2
18895 <round_saeonly_scalar_mask_op4>, %3} ... this.
18896 * config/i386/subst.md (mask_scalar_operand4,
18897 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18898 round_saeonly_scalar_nimm_predicate): New subst attributes.
18900 2017-07-06 Julia Koval <julia.koval@intel.com>
18902 * config/i386/i386.c (ix86_erase_embedded_rounding):
18903 Remove code for old rounding pattern.
18905 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18907 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18909 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18911 * doc/sourcebuild.texi (Test Directives, Variants of
18912 dg-require-support): Add documentation for dg-require-stack-check.
18914 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18916 * config/i386/subst.md (mask_scalar, round_scalar,
18917 round_saeonly_scalar): New meta-templates.
18918 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18919 round_scalar_mask_operand3, round_scalar_mask_op3,
18920 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18921 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18922 round_saeonly_scalar_constraint,
18923 round_saeonly_scalar_prefix): New subst attribute.
18924 * config/i386/sse.md
18925 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18926 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18927 <round_scalar_name> ... this.
18928 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18929 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18930 <round_scalar_name> ... this.
18931 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18932 <sse>_vm<code><mode>3<mask_scalar_name>
18933 <round_saeonly_scalar_name> ... this.
18934 (v<plusminus_mnemonic><ssescalarmodesuffix>
18935 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18936 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18937 v<plusminus_mnemonic><ssescalarmodesuffix>
18938 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18939 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18940 (v<multdiv_mnemonic><ssescalarmodesuffix>
18941 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18942 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18943 v<multdiv_mnemonic><ssescalarmodesuffix>
18944 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18945 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18946 (v<maxmin_float><ssescalarmodesuffix>
18947 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18948 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18949 v<maxmin_float><ssescalarmodesuffix>
18950 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18951 %0<mask_scalar_operand3>, %1, %<iptr>2
18952 <round_saeonly_scalar_mask_op3>} ... this.
18954 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18956 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18957 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18959 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18960 Alan Hayward <alan.hayward@arm.com>
18961 David Sherwood <david.sherwood@arm.com>
18963 * combine.c (simplify_if_then_else): Remove "enum" before
18965 * compare-elim.c (can_eliminate_compare): Likewise.
18966 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18968 (aarch64_lookup_simd_builtin_type): Likewise.
18969 (aarch64_simd_builtin_type): Likewise.
18970 (aarch64_init_simd_builtin_types): Likewise.
18971 (aarch64_simd_expand_args): Likewise.
18972 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18974 (aarch64_reverse_mask): Likewise.
18975 (aarch64_simd_emit_reg_reg_move): Likewise.
18976 (aarch64_gen_adjusted_ldpstp): Likewise.
18977 (aarch64_ccmp_mode_to_code): Likewise.
18978 (aarch64_operands_ok_for_ldpstp): Likewise.
18979 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18980 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18982 (aarch64_min_divisions_for_recip_mul): Likewise.
18983 (aarch64_reassociation_width): Likewise.
18984 (aarch64_get_condition_code_1): Likewise.
18985 (aarch64_simd_emit_reg_reg_move): Likewise.
18986 (aarch64_simd_attr_length_rglist): Likewise.
18987 (aarch64_reverse_mask): Likewise.
18988 (aarch64_operands_ok_for_ldpstp): Likewise.
18989 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18990 (aarch64_gen_adjusted_ldpstp): Likewise.
18991 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18993 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18994 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18995 (arm_lookup_simd_builtin_type): Likewise.
18996 (arm_simd_builtin_type): Likewise.
18997 (arm_init_simd_builtin_types): Likewise.
18998 (arm_expand_builtin_args): Likewise.
18999 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19000 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19001 (ft32_setup_incoming_varargs): Likewise.
19002 (ft32_function_arg): Likewise.
19003 (ft32_function_arg_advance): Likewise.
19004 (ft32_pass_by_reference): Likewise.
19005 (ft32_arg_partial_bytes): Likewise.
19006 (ft32_valid_pointer_mode): Likewise.
19007 (ft32_addr_space_pointer_mode): Likewise.
19008 (ft32_addr_space_legitimate_address_p): Likewise.
19009 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19011 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19012 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19013 (iamcu_alignment): Likewise.
19014 (canonicalize_vector_int_perm): Likewise.
19015 (ix86_noce_conversion_profitable_p): Likewise.
19016 (ix86_mpx_bound_mode): Likewise.
19017 (ix86_operands_ok_for_move_multiple): Likewise.
19018 * config/microblaze/microblaze-protos.h
19019 (microblaze_expand_conditional_branch_reg): Likewise.
19020 * config/microblaze/microblaze.c
19021 (microblaze_expand_conditional_branch_reg): Likewise.
19022 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19024 (rs6000_reassociation_width): Likewise.
19025 (rs6000_invalid_binary_op): Likewise.
19026 (fusion_p9_p): Likewise.
19027 (emit_fusion_p9_load): Likewise.
19028 (emit_fusion_p9_store): Likewise.
19029 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19031 (riscv_hard_regno_mode_ok_p): Likewise.
19032 (riscv_address_insns): Likewise.
19033 (riscv_split_symbol): Likewise.
19034 (riscv_legitimize_move): Likewise.
19035 (riscv_function_value): Likewise.
19036 (riscv_hard_regno_nregs): Likewise.
19037 (riscv_expand_builtin): Likewise.
19038 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19039 (riscv_build_integer): Likewise.
19040 (riscv_split_integer): Likewise.
19041 (riscv_legitimate_constant_p): Likewise.
19042 (riscv_cannot_force_const_mem): Likewise.
19043 (riscv_regno_mode_ok_for_base_p): Likewise.
19044 (riscv_valid_base_register_p): Likewise.
19045 (riscv_valid_offset_p): Likewise.
19046 (riscv_valid_lo_sum_p): Likewise.
19047 (riscv_classify_address): Likewise.
19048 (riscv_legitimate_address_p): Likewise.
19049 (riscv_address_insns): Likewise.
19050 (riscv_load_store_insns): Likewise.
19051 (riscv_force_binary): Likewise.
19052 (riscv_split_symbol): Likewise.
19053 (riscv_force_address): Likewise.
19054 (riscv_legitimize_address): Likewise.
19055 (riscv_move_integer): Likewise.
19056 (riscv_legitimize_const_move): Likewise.
19057 (riscv_legitimize_move): Likewise.
19058 (riscv_address_cost): Likewise.
19059 (riscv_subword): Likewise.
19060 (riscv_output_move): Likewise.
19061 (riscv_canonicalize_int_order_test): Likewise.
19062 (riscv_emit_int_order_test): Likewise.
19063 (riscv_function_arg_boundary): Likewise.
19064 (riscv_pass_mode_in_fpr_p): Likewise.
19065 (riscv_pass_fpr_single): Likewise.
19066 (riscv_pass_fpr_pair): Likewise.
19067 (riscv_get_arg_info): Likewise.
19068 (riscv_function_arg): Likewise.
19069 (riscv_function_arg_advance): Likewise.
19070 (riscv_arg_partial_bytes): Likewise.
19071 (riscv_function_value): Likewise.
19072 (riscv_pass_by_reference): Likewise.
19073 (riscv_setup_incoming_varargs): Likewise.
19074 (riscv_print_operand): Likewise.
19075 (riscv_elf_select_rtx_section): Likewise.
19076 (riscv_save_restore_reg): Likewise.
19077 (riscv_for_each_saved_reg): Likewise.
19078 (riscv_register_move_cost): Likewise.
19079 (riscv_hard_regno_mode_ok_p): Likewise.
19080 (riscv_hard_regno_nregs): Likewise.
19081 (riscv_class_max_nregs): Likewise.
19082 (riscv_memory_move_cost): Likewise.
19083 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19084 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19085 (rl78_addr_space_address_mode): Likewise.
19086 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19088 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19089 (rs6000_reassociation_width): Likewise.
19090 (rs6000_invalid_binary_op): Likewise.
19091 (fusion_p9_p): Likewise.
19092 (emit_fusion_p9_load): Likewise.
19093 (emit_fusion_p9_store): Likewise.
19094 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19095 (ok_for_simple_move_operands): Likewise.
19096 (ok_for_simple_move_strict_operands): Likewise.
19097 (ok_for_simple_arith_logic_operands): Likewise.
19098 (visium_legitimize_reload_address): Likewise.
19099 (visium_select_cc_mode): Likewise.
19100 (output_cbranch): Likewise.
19101 (visium_split_double_move): Likewise.
19102 (visium_expand_copysign): Likewise.
19103 (visium_expand_int_cstore): Likewise.
19104 (visium_expand_fp_cstore): Likewise.
19105 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19106 (visium_function_arg): Likewise.
19107 (visium_function_arg_advance): Likewise.
19108 (visium_libcall_value): Likewise.
19109 (visium_setup_incoming_varargs): Likewise.
19110 (visium_legitimate_constant_p): Likewise.
19111 (visium_legitimate_address_p): Likewise.
19112 (visium_legitimize_address): Likewise.
19113 (visium_secondary_reload): Likewise.
19114 (visium_register_move_cost): Likewise.
19115 (visium_memory_move_cost): Likewise.
19116 (prepare_move_operands): Likewise.
19117 (ok_for_simple_move_operands): Likewise.
19118 (ok_for_simple_move_strict_operands): Likewise.
19119 (ok_for_simple_arith_logic_operands): Likewise.
19120 (visium_function_value_1): Likewise.
19121 (rtx_ok_for_offset_p): Likewise.
19122 (visium_legitimize_reload_address): Likewise.
19123 (visium_split_double_move): Likewise.
19124 (visium_expand_copysign): Likewise.
19125 (visium_expand_int_cstore): Likewise.
19126 (visium_expand_fp_cstore): Likewise.
19127 (visium_split_cstore): Likewise.
19128 (visium_select_cc_mode): Likewise.
19129 (visium_split_cbranch): Likewise.
19130 (output_cbranch): Likewise.
19131 (visium_print_operand_address): Likewise.
19132 * expmed.c (flip_storage_order): Likewise.
19133 * expmed.h (emit_cstore): Likewise.
19134 (flip_storage_order): Likewise.
19135 * genrecog.c (validate_pattern): Likewise.
19136 * hsa-gen.c (gen_hsa_addr): Likewise.
19137 * internal-fn.c (expand_arith_overflow): Likewise.
19138 * ira-color.c (allocno_copy_cost_saving): Likewise.
19139 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19140 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19141 (process_invariant_for_inheritance): Likewise.
19142 * lra-eliminations.c (move_plus_up): Likewise.
19143 * omp-low.c (lower_oacc_reductions): Likewise.
19144 * simplify-rtx.c (simplify_subreg): Likewise.
19145 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19146 (TARGET_CHKP_BOUND_MODE): Likewise..
19147 * targhooks.c (default_chkp_bound_mode): Likewise.
19148 (default_setup_incoming_vararg_bounds): Likewise.
19149 * targhooks.h (default_chkp_bound_mode): Likewise.
19150 (default_setup_incoming_vararg_bounds): Likewise.
19151 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19152 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19153 (have_whole_vector_shift): Likewise.
19154 * tree-vect-stmts.c (vectorizable_load): Likewise.
19155 * doc/tm.texi: Regenerate.
19157 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19159 Graceful degrade if Binutils PR21472 is not available.
19162 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19163 .rodata in flash test fails.
19164 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19165 * confgure: Regenerate.
19166 * config.in: Regenerate.
19167 * config/avr/avr.c (avr_asm_named_section)
19168 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19169 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19170 (avr_asm_init_sections): Same.
19172 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19174 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19175 (fma<VH:mode>4_intrinsic): Likewise.
19176 (*fmsub<VCVTF:mode>4): Likewise.
19177 (*fmsub<VH:mode>4_intrinsic): Likewise.
19179 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19182 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19183 Don't depend on "optimize > 0".
19184 (out_movhi_r_mr, out_movqi_mr_r): Same.
19185 (out_movhi_mr_r, out_movqi_r_mr): Same.
19186 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19187 io_address_operand on "optimize > 0".
19189 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19191 * tree-loop-distribution.c: Add general explanantion on the pass.
19192 (generate_loops_for_partition): Mark distributed loop.
19193 (pg_add_dependence_edges): New parameter. Handle alias data
19194 dependence specially and record it in the parameter if asked.
19195 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19196 (init_partition_graph_vertices, add_partition_graph_edge): New.
19197 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19198 (free_partition_graph_vdata, build_partition_graph): New.
19199 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19200 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19201 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19202 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19203 (version_for_distribution_p, finalize_partitions): New.
19204 (distribute_loop): Handle alias data dependence specially. Factor
19205 out loop fusion code as functions and call these functions.
19207 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19209 * tree-loop-distribution.c (classify_partition): New parameter and
19210 better handle reduction statement.
19211 (rdg_build_partitions): Revise comment.
19212 (distribute_loop): Compute statements in all partitions and pass it
19213 to classify_partition.
19215 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19217 * tree-loop-distribution.c (enum partition_type): New.
19218 (struct partition): New field type.
19219 (partition_merge_into): Add parameter. Update partition type.
19220 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19221 (build_rdg_partition_for_vertex): Compute partition type.
19222 (rdg_build_partitions): Dump partition type.
19223 (distribute_loop): Update calls to partition_merge_into.
19225 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19227 * tree-loop-distribution.c (struct ddr_hasher): New.
19228 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19230 (classify_partition): Call get_data_dependence.
19231 (pg_add_dependence_edges): Ditto.
19232 (distribute_loop): Release data dependence hash table.
19234 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19236 * tree-loop-distribution.c (ref_base_address): Delete.
19237 (similar_memory_accesses): Rename ...
19238 (share_memory_accesses): ... to this. Check if partitions access
19239 the same memory reference.
19240 (distribute_loop): Call share_memory_accesses.
19242 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19244 * tree-loop-distribution.c (struct partition): New field recording
19245 its data reference.
19246 (partition_alloc, partition_free): Init and release data refs.
19247 (partition_merge_into): Merge data refs.
19248 (build_rdg_partition_for_vertex): Collect data refs for partition.
19249 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19251 (distribute_loop): Remve data refs from vertice data of partition
19254 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19256 * tree-loop-distribution.c (params.h): Include header file.
19257 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19258 (datarefs_vec): New global var.
19259 (create_rdg_vertices): Use datarefs_vec directly.
19260 (free_rdg): Don't free data references.
19261 (build_rdg): Update use. Don't free data references.
19262 (distribute_loop): Compute global variable for data references.
19263 Bail out if there are too many data references.
19265 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19267 * tree-loop-distribution.c (loop_nest): New global var.
19268 (build_rdg): Use loop directly, rather than loop nest.
19269 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19271 (distribute_loop): Compute global variable loop nest. Update use.
19273 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19275 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19276 (partition_merge_into): New parameter. Dump reason for fusion.
19277 (distribute_loop): Update use of partition_merge_into.
19279 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19281 * tree-loop-distribution.c (bb_top_order_index): New.
19282 (bb_top_order_index_size, bb_top_order_cmp): New.
19283 (stmts_from_loop): Use topological order.
19284 (pass_loop_distribution::execute): Compute and release topological
19285 order for basic blocks.
19287 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19289 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19292 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19294 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19295 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19296 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19297 * internal-fn.def (LOOP_DIST_ALIAS): New.
19298 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19299 (fold_loop_internal_call): ... this.
19300 (vect_loop_dist_alias_call): New function.
19301 (set_uid_loop_bbs): Call fold_loop_internal_call.
19302 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19305 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19308 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19309 Require dead FLAGS_REG at the beginning of a peephole.
19311 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19314 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19315 arguments in the call to __builtin_ia32_sbb_u32.
19316 (_subborrow_u64): Swap _X and _Y arguments in the call to
19317 __builtin_ia32_sbb_u64.
19319 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19322 * tree-vrp.c (compare_assert_loc): Turn into a function template
19323 with stable template parameter. Only test if a->e is NULL,
19324 !a->e == !b->e has been verified already. Use e == NULL or
19325 e != NULL instead of e or ! e tests. If stable is true, don't use
19326 iterative_hash_expr, on the other side allow a or b or both NULL
19327 and sort the NULLs last.
19328 (process_assert_insertions): Sort using compare_assert_loc<false>
19329 instead of compare_assert_loc, later sort using
19330 compare_assert_loc<true> before calling process_assert_insertions_for
19331 in a loop. Use break instead of continue once seen NULL pointer.
19333 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19335 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19336 Cortex-R7 and Cortex-R8 processors.
19338 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19340 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19341 uninitialized while src is not.
19343 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19345 * common/config/arm/arm-common.c: Adjust include path for
19347 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19348 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19349 (arm-cpu-data.h): Likewise.
19350 (arm-cpu-cdata.h): Likewise.
19351 * config/arm/arm-cpu.h: Delete.
19352 * config/arm/arm-cpu-cdata.h: Delete.
19353 * config/arm/arm-cpu-data.h: Delete.
19355 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19357 * config/arm/arm-cpus.in (cortex-a55): New.
19358 (cortex-a75): Likewise.
19359 (cortex-a75.cortex-a55): Likewise.
19360 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19362 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19363 * config/arm/arm-cpu-cdata.h: Regenerate.
19364 * config/arm/arm-cpu-data.h: Regenerate.
19365 * config/arm/arm-cpu.h: Regenerate.
19366 * config/arm/arm-tables.opt: Regenerate.
19367 * config/arm/arm-tune.md: Regenerate.
19369 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19371 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19373 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19375 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19378 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19380 PR tree-optimization/81292
19381 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19382 full_string_p, also call adjust_related_strinfos if the adjustment
19383 is simple, otherwise invalidate related strinfos.
19385 2017-07-04 Martin Liska <mliska@suse.cz>
19388 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19389 newly created variable as DECL_IGNORED_P.
19391 2017-07-04 Martin Liska <mliska@suse.cz>
19394 * ipa-inline.c (inline_small_functions):
19395 Use xstrdup_for_dump.
19397 2017-07-04 Tom de Vries <tom@codesourcery.com>
19399 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19401 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19404 * config/darwin.c (darwin_function_switched_text_sections):
19407 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19409 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19411 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19413 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19415 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19417 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19418 min_profitable_iters, and th as inclusive lower bounds.
19419 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19420 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19421 for min_profitable_iters and min_profitable_estimate.
19422 (vect_transform_loop): Treat th as an inclusive lower bound.
19423 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19425 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19428 * config/darwin.c (darwin_function_switched_text_sections):
19429 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19430 in two pieces, and suppress the use of buf.
19432 2017-07-03 Nathan Sidwell <nathan@acm.org>
19434 * hash-table.h (hash_table_mod1): Fix indentation.
19436 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19438 PR middle-end/81290
19439 * predict.c (force_edge_cold): Be more careful about propagation
19441 * profile-count.h (profile_probability::guessed,
19442 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19444 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19446 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19448 * doc/invoke.texi (rcpc architecture extension): Document it.
19450 2017-07-03 Richard Biener <rguenther@suse.de>
19452 PR tree-optimization/60510
19453 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19454 the scalar reduction PHI and use it.
19455 (vectorizable_reduction): Properly guard the single_defuse_cycle
19456 path for non-SLP reduction chains where we cannot use it.
19457 Rework reduc_def/index and vector type deduction. Rework
19458 vector operand gathering during reduction op code-gen.
19459 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19460 chains dissolve the chain and leave it to non-SLP reduction
19463 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19465 * tree-data-ref.h (dr_alignment): Declare.
19466 * tree-data-ref.c (dr_alignment): New function.
19467 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19468 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19470 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19472 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19474 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19475 and base_misalignment fields.
19476 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19477 * tree-data-ref.c: Include builtins.h.
19478 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19479 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19480 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19481 * tree-vect-data-refs.c: Include tree-cfg.h.
19482 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19483 fields instead of calculating an alignment here.
19484 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19485 innermost_loop_behavior fields.
19487 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19489 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19491 (DR_STEP_ALIGNMENT): New macro.
19492 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19493 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19494 (create_data_ref): Print it.
19495 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19496 to tell whether the step preserves vector (mis)alignment.
19497 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19498 Move the check for an integer step and generalise to all INTEGER_CST.
19499 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19500 Print the outer step alignment.
19502 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19504 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19505 with offset_alignment.
19506 (DR_ALIGNED_TO): Delete.
19507 (DR_OFFSET_ALIGNMENT): New macro.
19508 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19509 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19510 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19511 (create_data_ref): Likewise.
19512 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19513 (vect_analyze_data_refs): Likewise.
19514 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19515 creating dummy innermost behavior.
19517 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19519 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19520 with a "innermost_loop_behavior *" and refeence tree.
19521 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19522 (create_data_ref): Update call accordingly.
19523 * tree-predcom.c (find_looparound_phi): Likewise.
19525 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19527 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19528 fields with dr_wrt_vec_loop.
19529 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19530 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19531 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19532 (vect_dr_behavior): New function.
19533 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19534 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19535 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19536 track whether the step preserves the misalignment.
19537 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19538 Use vect_dr_behavior.
19539 (vect_setup_realignment): Update call accordingly.
19540 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19541 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19542 call to vect_create_addr_base_for_vector_ref.
19543 (vect_create_cond_for_align_checks): Likewise.
19544 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19545 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19546 (vect_recog_mask_conversion_pattern): Likewise.
19547 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19548 (new_stmt_vec_info): Remove redundant zeroing.
19550 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19552 * common/config/arm/arm-common.c (arm_be8_option): New function.
19553 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19554 (ISA_ARMv6): Add isa_bit_be8.
19555 * config/arm/arm.h (arm_be8_option): Add prototype.
19556 (BE8_SPEC_FUNCTION): New define.
19557 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19558 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19559 (mlittle-endian): Similarly.
19560 (mbe8, mbe32): New options.
19561 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19562 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19564 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19566 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19568 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19570 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19571 (cleanup_tree_cfg_bb): Use it.
19572 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19574 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19576 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19579 * loop-doloop.c (add_test): Update profile.
19581 2017-07-03 Martin Liska <mliska@suse.cz>
19584 * sanopt.c (rewrite_usage_of_param): New function.
19585 (sanitize_rewrite_addressable_params): Likewise.
19586 (pass_sanopt::execute): Call rewrite_usage_of_param.
19588 2017-07-03 Richard Biener <rguenther@suse.de>
19590 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19591 back to using VIEW_CONVERT_EXPR.
19593 2017-07-03 Martin Liska <mliska@suse.cz>
19596 * doc/extend.texi: Document when a resolver function is
19597 generated for target_clones.
19599 2017-07-03 Martin Liska <mliska@suse.cz>
19601 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19602 and shadow memory of auto variables which are subject of
19603 use-after-scope sanitization.
19604 (asan_expand_mark_ifn): Add do set only when is_poison.
19606 2016-07-03 Richard Biener <rguenther@suse.de>
19608 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19610 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19611 (vectorizable_reduction): Perform reduction PHI creation when
19612 visiting a reduction PHI and adjust and simplify code generation
19613 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19614 (vect_transform_loop): Visit reduction PHIs.
19615 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19616 defs into the SLP tree.
19617 (vect_build_slp_tree): Reduction defs terminate the recursion.
19618 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19620 (vect_get_vec_defs_for_stmt_copy): Export.
19621 (vect_get_vec_defs): Likewise.
19622 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19624 (vect_get_vec_defs_for_stmt_copy): Declare.
19625 (vect_get_vec_defs): Likewise.
19627 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19629 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19630 parameter with a "loop" parameter and use it instead of the
19631 loop containing DR_STMT. Don't check simple_iv when doing
19632 BB analysis. Describe the two analysis modes in the comment.
19634 2017-07-03 Tom de Vries <tom@codesourcery.com>
19636 PR tree-optimization/69468
19637 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19638 (find_same_succ_bb): Handle ignore_edge_flags.
19640 2017-07-03 Tom de Vries <tom@codesourcery.com>
19642 PR tree-optimization/81192
19643 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19645 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19647 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19649 2017-07-03 Tom de Vries <tom@codesourcery.com>
19651 PR tree-optimization/81192
19652 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19653 BB_SAME_SUCC (bb) == NULL.
19655 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19657 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19660 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19662 * dumpfile.c: Include profile-count.h
19663 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19665 (insert_cond_bb): Update profile.
19666 * tree-cfg.h (insert_cond_bb): Update prototype.
19667 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19668 * tree-dump.c: Do not include tree-cfg.
19670 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19672 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19674 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19676 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19679 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19681 * tree-complex.c (expand_complex_div_wide): update profile.
19683 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19684 Alan Hayward <alan.hayward@arm.com>
19685 David Sherwood <david.sherwood@arm.com>
19687 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19688 (CORETYPES_H): New define.
19689 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19690 (insn-modes-inline.h, s-modes-inline-h): New rules.
19691 (generated_files): Add insn-modes-inline.h.
19692 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19693 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19694 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19695 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19696 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19697 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19698 (build/gencodes.o, build/genconditions.o): Likewise.
19699 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19700 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19701 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19702 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19703 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19704 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19705 * coretypes.h: Include everything up to real.h for generators.
19706 Include insn-modes.h first. Include wide-int-print.h after
19707 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19708 * machmode.h: Don't include insn-modes.h here.
19709 * function-tests.c: Remove includes of signop.h, machmode.h,
19710 double-int.h and wide-int.h.
19712 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19714 * optc-save-gen.awk: Likewise.
19715 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19716 * godump.c: Remove include of wide-int-print.h.
19717 * pretty-print.h: Likewise.
19718 * wide-int-print.cc: Likewise.
19719 * wide-int.cc: Likewise.
19720 * hash-map-tests.c: Remove include of signop.h.
19721 * hash-set-tests.c: Likewise.
19722 * rtl-tests.c: Likewise.
19723 * mkconfig.sh: Remove include of machmode.h.
19724 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19726 (emit_insn_modes_inline_h): ...this new function. Emit the code
19727 into an insn-modes-inline.h header file, adding appropriate
19728 include guards and end comments.
19729 (emit_insn_modes_c_header): Remove include of machmode.h.
19730 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19732 (main): Handle -i flag and call emit_insn_modes_inline_h when
19735 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19737 * tree-ssa-strlen.c (strinfo): Rename the length field to
19738 nonzero_chars. Add a full_string_p field.
19739 (compare_nonzero_chars, zero_length_string_p): New functions.
19740 (get_addr_stridx): Add an offset_out parameter.
19741 Use compare_nonzero_chars.
19742 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19743 (new_strinfo): Update after above changes to strinfo.
19744 (set_endptr_and_length): Set full_string_p.
19745 (get_string_length): Update after above changes to strinfo.
19746 (unshare_strinfo): Update call to new_strinfo.
19747 (maybe_invalidate): Likewise.
19748 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19749 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19750 as a uhwi instead of an shwi. Update after above changes to
19751 strinfo and new_strinfo.
19752 (zero_length_string): Assert that chainsi contains full strings.
19753 Use zero_length_string_p. Update call to new_strinfo.
19754 (adjust_related_strinfos): Update after above changes to strinfo.
19755 Copy full_string_p from origsi.
19756 (adjust_last_stmt): Use zero_length_string_p.
19757 (handle_builtin_strlen): Update after above changes to strinfo and
19758 new_strinfo. Install the lhs as the string length if the previous
19759 entry didn't describe a full string.
19760 (handle_builtin_strchr): Update after above changes to strinfo
19762 (handle_builtin_strcpy): Likewise.
19763 (handle_builtin_strcat): Likewise.
19764 (handle_builtin_malloc): Likewise.
19765 (handle_pointer_plus): Likewise.
19766 (handle_builtin_memcpy): Likewise. Track nonzero characters
19767 that aren't necessarily followed by a nul terminator.
19768 (handle_char_store): Likewise.
19770 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19772 PR tree-optimization/80769
19773 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19774 for malloc and calloc. Document the new invariant that all related
19775 strinfos have delayed lengths or none do.
19776 (verify_related_strinfos): Move earlier in file.
19777 (set_endptr_and_length): New function, split out from...
19778 (get_string_length): ...here. Also set the lengths of related
19780 (zero_length_string): Assert that chainsi has known (rather than
19782 (adjust_related_strinfos): Likewise.
19784 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19786 PR tree-optimization/81136
19787 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19788 assert that two references with the same misalignment have the same
19789 compile-time misalignment if those compile-time misalignments
19792 2017-07-01 Andi Kleen <ak@linux.intel.com>
19794 * print-tree.c (print_node): Print all attributes.
19796 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19798 * cfg.c (scale_bbs_frequencies): New function.
19799 * cfg.h (scale_bbs_frequencies): Declare it.
19800 * cfgloopanal.c (single_likely_exit): Cleanup.
19801 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19803 (scale_loop_profile): Likewise.
19804 (loop_version): Likewise.
19805 (create_empty_loop_on_edge): Update.
19806 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19807 scale_loop_frequencies, scale_loop_profile, loopify,
19808 loop_version): Update prototypes.
19809 * modulo-sched.c (sms_schedule): Update.
19810 * predict.c (unlikely_executed_edge_p): Also check probability.
19811 (probably_never_executed_edge_p): Fix typo.
19812 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19813 * tree-parloops.c (gen_parallel_loop): Update.
19814 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19815 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19816 * tree-ssa-loop-split.c (split_loop): Update.
19817 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19818 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19819 (vect_loop_versioning): Update.
19820 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19822 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19824 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19826 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19828 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19829 to keep profile consistent.
19831 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19833 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19834 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19835 * profile-count.h (max_safe_multiplier): Make unsigned.
19836 (profile_count::guessed_zero): New.
19838 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19840 * bb-reorder.c (fix_up_crossing_landing_pad,
19841 fix_crossing_conditional_branches): Use make_single_succ_edge
19842 to keep profile consistent.
19844 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19846 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19849 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19852 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19853 the right scopes, make sure cond_jump isn't preserved between multiple
19854 iterations. Search for fallthru edge whenever there are 3+ edges and
19855 use find_fallthru_edge for it.
19857 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19859 Patch by Alexander Monakov <amonakov@ispras.ru>
19860 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19861 probabilities consistently.
19863 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19865 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19866 emit_cmp_and_jump_insns.
19868 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19871 * tree-inline.c (expand_call_inline): Combine profile statuses.
19873 2017-06-30 Andrew Pinski <apinski@cavium.com>
19875 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19876 fold_stmt returned true.
19878 2017-06-30 Nathan Sidwell <nathan@acm.org>
19880 * ggc.h (empty_string): Delete.
19881 * cfgexpand.c (expand_asm_stmt): Use plain "".
19882 * optabs.c (expand_asm_memory_barrier): Likewise.
19883 * stringpool.c (empty_string): Delete.
19884 (digit_vector, digit_string): Delete.
19885 (ggc_alloc_string): Use plain "", don't optimize single digit
19886 strings. Use ggc_alloc_atomic.
19888 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19890 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19891 comparison set and one other set, use the cost of the non-comparison
19894 2017-06-30 Nathan Sidwell <nathan@acm.org>
19896 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19899 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19901 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19902 loops. Remove now unneeded calls to gimple_switch_set_label() that
19903 just set removed labels to NULL_TREE.
19905 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19907 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19908 (set_range_info): ...here. Only call set_range_info_raw if domain
19910 (set_nonzero_bits): Call set_range_info_raw.
19911 * tree-ssanames.h (set_range_info_raw): New.
19913 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19916 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19917 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19918 of nonimmediate_operand and <store_mask_constraint> instead of m
19919 for the input operand. For V8FI iterator, always split if input
19920 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19921 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19922 <store_mask_predicate> instead of register_operand and
19923 <store_mask_constraint> instead of v for the input operand. Make
19924 sure both operands aren't MEMs for if not <mask_applied>.
19926 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19928 * lto-wrapper.c (copy_file) Close both file descriptors before
19931 2017-06-30 Martin Liska <mliska@suse.cz>
19934 * multiple_target.c (create_dispatcher_calls): Make ifunc
19935 also for function that don't have calls or are not referenced.
19937 2017-06-30 Richard Biener <rguenther@suse.de>
19939 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19940 analyze the first scalar stmt. Move vector type computation
19941 for the BB case here from ...
19942 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19943 live operation processing in the SLP case properly.
19945 2017-06-30 Richard Biener <rguenther@suse.de>
19947 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19949 2017-06-30 Martin Liska <mliska@suse.cz>
19952 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19953 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19955 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19957 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19960 2017-06-30 Richard Biener <rguenther@suse.de>
19962 PR tree-optimization/81249
19963 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19964 condition reduction result to original scalar type.
19966 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19968 * profile-count.h (enum profile_quality): Fix typos and whitespace
19971 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19973 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19974 type for branch probabilities.
19976 2017-06-29 Julian Brown <julian@codesourcery.com>
19977 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19979 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19980 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19981 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19982 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19984 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19986 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19987 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19988 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19989 CC usage from generic code to here.
19990 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19991 CC usage into the target macros.
19993 2017-06-29 Maya Rashish <coypu@sdf.org>
19995 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19998 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20000 * arm/arm-builtins.c: Include profile-count.h
20001 * except.c (sjlj_emit_function_enter): Use
20002 profile_probability::unlikely.
20004 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20006 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20007 and tocrel_offset be pointer args rather than implicitly using
20009 (legitimate_constant_pool_address_p, rs6000_emit_move,
20010 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20011 tocrel_offset and use in toc_relative_expr_p call.
20012 (print_operand, print_operand_address): Use static tocrel_base_oac
20013 and tocrel_offset_oac.
20014 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20017 2017-06-29 Maya Rashish <coypu@sdf.org>
20019 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20021 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20023 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20024 objects, take into account only the alignment of 'op0' and 'mode1' if
20027 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20029 * ccmp.c (ccmp_tree_comparison_p): New function.
20030 (ccmp_candidate_p): Update to use above function.
20031 (get_compare_parts): New function.
20032 (expand_ccmp_next): Update to use new functions.
20033 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20035 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20036 take mode as argument.
20037 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20038 * expr.c (expand_expr_real_1): Pass mode as argument.
20040 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20042 * combine.c (combine_instructions): Print insns to dump_file, together
20045 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20047 * asan.c (asan_emit_stack_protection): Update.
20048 (create_cond_insert_point): Update.
20049 * auto-profile.c (afdo_propagate_circuit): Update.
20050 * basic-block.h (struct edge_def): Turn probability to
20051 profile_probability.
20052 (EDGE_FREQUENCY): Update.
20053 * bb-reorder.c (find_traces_1_round): Update.
20054 (better_edge_p): Update.
20055 (sanitize_hot_paths): Update.
20056 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20057 (make_single_succ_edge): Update.
20058 (check_bb_profile): Update.
20059 (dump_edge_info): Update.
20060 (update_bb_profile_for_threading): Update.
20061 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20062 probabilitycount to 0.
20063 * cfgbuild.c (compute_outgoing_frequencies): Update.
20064 * cfgcleanup.c (try_forward_edges): Update.
20065 (outgoing_edges_match): Update.
20066 (try_crossjump_to_edge): Update.
20067 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20068 (expand_gimple_tailcall): Update.
20069 (construct_init_block): Use make_single_succ_edge.
20070 (construct_exit_block): Use make_single_succ_edge.
20071 * cfghooks.c (verify_flow_info): Update.
20072 (redirect_edge_succ_nodup): Update.
20073 (split_edge): Update.
20074 (account_profile_record): Update.
20075 * cfgloopanal.c (single_likely_exit): Update.
20076 * cfgloopmanip.c (scale_loop_profile): Update.
20077 (set_zero_probability): Remove.
20078 (duplicate_loop_to_header_edge): Update.
20079 * cfgloopmanip.h (loop_version): Update prototype.
20080 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20081 (force_nonfallthru_and_redirect): Update.
20082 (update_br_prob_note): Update.
20083 (rtl_verify_edges): Update.
20084 (purge_dead_edges): Update.
20085 (rtl_lv_add_condition_to_bb): Update.
20086 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20087 * cgraphunit.c (init_lowered_empty_function): Update.
20088 (cgraph_node::expand_thunk): Update.
20089 * cilk-common.c: Include profile-count.h
20090 * dojump.c (inv): Remove.
20091 (jumpifnot): Update.
20092 (jumpifnot_1): Update.
20093 (do_jump_1): Update.
20095 (do_jump_by_parts_greater_rtx): Update.
20096 (do_compare_rtx_and_jump): Update.
20097 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20098 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20099 * dwarf2cfi.c: Include profile-count.h
20100 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20101 (sjlj_emit_dispatch_table): Likewise.
20102 * explow.c: Include profile-count.h
20103 * expmed.c (emit_store_flag_force): Update.
20104 (do_cmp_and_jump): Update.
20105 * expr.c (compare_by_pieces_d::generate): Update.
20106 (compare_by_pieces_d::finish_mode): Update.
20107 (emit_block_move_via_loop): Update.
20108 (store_expr_with_bounds): Update.
20109 (store_constructor): Update.
20110 (expand_expr_real_2): Update.
20111 (expand_expr_real_1): Update.
20112 * expr.h (try_casesi, try_tablejump): Update prototypes.
20113 * gimple-pretty-print.c (dump_probability): Update.
20114 (dump_profile): New.
20115 (dump_gimple_label): Update.
20116 (dump_gimple_bb_header): Update.
20117 * graph.c (draw_cfg_node_succ_edges): Update.
20118 * hsa-gen.c (convert_switch_statements): Update.
20119 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20120 (find_if_case_1): Update.
20121 (find_if_case_2): Update.
20122 * internal-fn.c (expand_arith_overflow_result_store): Update.
20123 (expand_addsub_overflow): Update.
20124 (expand_neg_overflow): Update.
20125 (expand_mul_overflow): Update.
20126 (expand_vector_ubsan_overflow): Update.
20127 * ipa-cp.c (good_cloning_opportunity_p): Update.
20128 * ipa-split.c (split_function): Use make_single_succ_edge.
20129 * ipa-utils.c (ipa_merge_profiles): Update.
20130 * loop-doloop.c (add_test): Update.
20131 (doloop_modify): Update.
20132 * loop-unroll.c (compare_and_jump_seq): Update.
20133 (unroll_loop_runtime_iterations): Update.
20134 * lra-constraints.c (lra_inheritance): Update.
20135 * lto-streamer-in.c (input_cfg): Update.
20136 * lto-streamer-out.c (output_cfg): Update.
20137 * mcf.c (adjust_cfg_counts): Update.
20138 * modulo-sched.c (sms_schedule): Update.
20139 * omp-expand.c (expand_omp_for_init_counts): Update.
20140 (extract_omp_for_update_vars): Update.
20141 (expand_omp_ordered_sink): Update.
20142 (expand_omp_for_ordered_loops): Update.
20143 (expand_omp_for_generic): Update.
20144 (expand_omp_for_static_nochunk): Update.
20145 (expand_omp_for_static_chunk): Update.
20146 (expand_cilk_for): Update.
20147 (expand_omp_simd): Update.
20148 (expand_omp_taskloop_for_outer): Update.
20149 (expand_omp_taskloop_for_inner): Update.
20150 * omp-simd-clone.c (simd_clone_adjust): Update.
20151 * optabs.c (expand_doubleword_shift): Update.
20152 (expand_abs): Update.
20153 (emit_cmp_and_jump_insn_1): Update.
20154 (expand_compare_and_swap_loop): Update.
20155 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20156 * predict.c (predictable_edge_p): Update.
20157 (edge_probability_reliable_p): Update.
20158 (set_even_probabilities): Update.
20159 (combine_predictions_for_insn): Update.
20160 (combine_predictions_for_bb): Update.
20161 (propagate_freq): Update.
20162 (estimate_bb_frequencies): Update.
20163 (force_edge_cold): Update.
20164 * profile-count.c (profile_count::dump): Add missing space into dump.
20165 (profile_count::debug): Add newline.
20166 (profile_count::differs_from_p): Explicitly convert to unsigned.
20167 (profile_count::stream_in): Update.
20168 (profile_probability::dump): New member function.
20169 (profile_probability::debug): New member function.
20170 (profile_probability::differs_from_p): New member function.
20171 (profile_probability::differs_lot_from_p): New member function.
20172 (profile_probability::stream_in): New member function.
20173 (profile_probability::stream_out): New member function.
20174 * profile-count.h (profile_count_quality): Rename to ...
20175 (profile_quality): ... this one.
20176 (profile_probability): New.
20177 (profile_count): Update.
20178 * profile.c (compute_branch_probabilities): Update.
20179 * recog.c (peep2_attempt): Update.
20180 * sched-ebb.c (schedule_ebbs): Update.
20181 * sched-rgn.c (find_single_block_region): Update.
20182 (compute_dom_prob_ps): Update.
20183 (schedule_region): Update.
20184 * sel-sched-ir.c (compute_succs_info): Update.
20185 * stmt.c (struct case_node): Update.
20186 (do_jump_if_equal): Update.
20187 (get_outgoing_edge_probs): Update.
20188 (conditional_probability): Update.
20189 (emit_case_dispatch_table): Update.
20190 (expand_case): Update.
20191 (expand_sjlj_dispatch_table): Update.
20192 (emit_case_nodes): Update.
20193 * targhooks.c: Update.
20194 * tracer.c (better_p): Update.
20195 (find_best_successor): Update.
20196 * trans-mem.c (expand_transaction): Update.
20197 * tree-call-cdce.c: Update.
20198 * tree-cfg.c (gimple_split_edge): Upate.
20199 (move_sese_region_to_fn): Upate.
20200 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20201 * tree-eh.c (lower_resx): Upate.
20202 (cleanup_empty_eh_move_lp): Upate.
20203 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20204 * tree-inline.c (copy_edges_for_bb): Update.
20205 (copy_cfg_body): Update.
20206 * tree-parloops.c (gen_parallel_loop): Update.
20207 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20208 (gimple_gen_time_profiler): Update.
20209 * tree-ssa-dce.c (remove_dead_stmt): Update.
20210 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20211 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20212 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20213 (unloop_loops): Update.
20214 (try_peel_loop): Update.
20215 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20216 * tree-ssa-loop-split.c (connect_loops): Update.
20217 (split_loop): Update.
20218 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20219 (hoist_guard): Update.
20220 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20221 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20222 (value_replacement): Update.
20223 * tree-ssa-reassoc.c (branch_fixup): Update.
20224 * tree-ssa-tail-merge.c (replace_block_by): Update.
20225 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20226 (create_edge_and_update_destination_phis): Update.
20227 (compute_path_counts): Update.
20228 (recompute_probabilities): Update.
20229 (update_joiner_offpath_counts): Update.
20230 (freqs_to_counts_path): Update.
20231 (duplicate_thread_path): Update.
20232 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20233 (struct switch_conv_info): Update.
20234 (gen_inbound_check): Update.
20235 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20236 (vect_do_peeling): Update.
20237 (vect_loop_versioning): Update.
20238 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20239 (optimize_mask_stores): Update.
20240 * ubsan.c (ubsan_expand_null_ifn): Update.
20241 * value-prof.c (gimple_divmod_fixed_value): Update.
20242 (gimple_divmod_fixed_value_transform): Update.
20243 (gimple_mod_pow2): Update.
20244 (gimple_mod_pow2_value_transform): Update.
20245 (gimple_mod_subtract): Update.
20246 (gimple_mod_subtract_transform): Update.
20247 (gimple_ic): Update.
20248 (gimple_stringop_fixed_value): Update.
20249 (gimple_stringops_transform): Update.
20250 * value-prof.h: Update.
20252 2017-06-29 Carl Love <cel@us.ibm.com>
20254 * config/rs6000/rs6000-c.c: Add support for built-in functions
20255 vector signed int vec_signed (vector float);
20256 vector signed long long vec_signed (vector double);
20257 vector signed int vec_signed2 (vector double, vector double);
20258 vector signed int vec_signede (vector double);
20259 vector signed int vec_signedo (vector double);
20260 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20261 instruction generator.
20262 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20263 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20264 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20266 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20267 vunsignede_v2df): Add define_expands.
20268 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20269 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20270 VEC_UNSIGNEDO): Add definitions.
20271 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20272 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20273 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20274 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20275 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20276 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20277 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20278 vec_unsignede, vec_unsignedo): Add builtin defines.
20279 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20281 * doc/extend.texi: Update the built-in documentation file for the
20282 new built-in functions.
20284 2017-06-29 Richard Biener <rguenther@suse.de>
20286 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20287 reduction chains to LOOP_VINFO_REDUCTIONS.
20288 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20289 SLP reductions after processing reduction chains.
20291 2017-06-29 Nathan Sidwell <nathan@acm.org>
20293 * builtins.c (fold_builtin_FUNCTION): Use
20294 lang_hooks.decl_printable_name.
20296 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20298 PR middle-end/81194
20299 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20300 with only one label.
20301 * stmt.c (expand_case): Assert NCASES is greater than one.
20303 2017-06-29 Richard Biener <rguenther@suse.de>
20305 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20307 (group_case_labels): Likewise.
20308 (find_taken_edge): Push sanity checking on val to workers...
20309 (find_taken_edge_cond_expr): ... here
20310 (find_taken_edge_switch_expr): ... and here, handle cases
20311 with just a default label.
20312 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20313 (group_case_labels): Likewise.
20314 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20315 group_case_labels does anything cleanup the CFG again.
20317 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20319 PR tree-optimization/81196
20320 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20321 exit condition comparing two IVs.
20323 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20325 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20326 profile to the dummy entry at the end of the list of architectures.
20327 * config/arm/arm-cpu-cdata.h: Regenerated.
20329 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20330 Michael Collison <michael.collison@arm.com>
20333 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20335 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20336 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20337 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20338 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20339 of shift when the shift amount is masked with constant equal to
20340 the size of the mode.
20341 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20344 2017-06-29 Martin Liska <mliska@suse.cz>
20346 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20348 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20351 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20354 (edge_growth_cache_entry::edge_growth_cache_entry): New
20356 (reset_edge_growth_cache): Update to use constructor.
20358 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20360 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20361 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20362 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20364 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20366 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20367 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20369 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20371 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20372 (*-linux-uclibc*): Add t-uclibc tmake_file.
20373 * config/t-musl: New.
20374 * config/t-uclibc: New.
20376 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20378 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20380 (gen_comm_data): Emit architectural setting of arch_prof.
20381 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20383 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20384 (armv8-m.base, armv8-m.main): Likewise.
20385 * arm-protos.h (arm_build_target): Add profile field.
20386 (arch_option): Likewise.
20387 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20389 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20390 arm_active_target.profile.
20392 2017-06-28 Richard Biener <rguenther@suse.de>
20394 PR middle-end/81227
20395 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20396 TYPE_OVERFLOW_WRAPS.
20397 * match.pd (negate_expr_p): Likewise.
20398 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20399 fold_build2, not fold_binary.
20401 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20403 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20404 Convert memory address to Pmode.
20405 (aarch64_print_operand): Assert MEM operands are always Pmode.
20407 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20410 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20411 Remove redundant if.
20412 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20413 * config/arm/aarch-common-protos.h
20414 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20415 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20417 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20420 * multiple_target.c (create_dispatcher_calls): Set the default
20421 clone to be static, not public.
20423 2017-06-28 Richard Biener <rguenther@suse.de>
20425 * tree-vect-loop.c (vectorizable_reduction): Move special
20426 cond reduction IV var creation ...
20427 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20428 parameter. Use STMT_VINFO_VECTYPE.
20429 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20432 2017-06-28 Martin Liska <mliska@suse.cz>
20435 * ipa-visibility.c (non_local_p): Handle visibility.
20437 2017-06-28 Martin Liska <mliska@suse.cz>
20440 * common.opt: Add IntegerRange to various options.
20441 * opt-functions.awk (integer_range_info): New function.
20442 * optc-gen.awk: Add integer_range_info to cl_options struct.
20443 * opts-common.c (decode_cmdline_option): Handle
20444 CL_ERR_INT_RANGE_ARG.
20445 (cmdline_handle_error): Likewise.
20446 * opts.c (print_filtered_help): Show valid interval in
20447 when --help is provided.
20448 * opts.h (struct cl_option): Add range_min and range_max fields.
20449 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20451 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20453 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20454 (x * C EQ/NE y * C): New transformation.
20456 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20458 * genmultilib (combination_space): Accept '+' in option names.
20460 2017-06-28 Martin Liska <mliska@suse.cz>
20463 * asan.c (instrument_derefs): Bail out inner references
20464 that are hard register variables.
20466 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20469 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20470 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20472 2017-06-28 Richard Biener <rguenther@suse.de>
20474 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20475 (vect_get_slp_defs): Adjust.
20476 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20478 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20480 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20481 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20482 (vectorizable_reduction): Adjust.
20483 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20485 (vect_get_slp_defs): Likewise.
20486 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20487 (vectorizable_bswap): Adjust.
20488 (vectorizable_call): Likewise.
20489 (vectorizable_conversion): Likewise.
20490 (vectorizable_assignment): Likewise.
20491 (vectorizable_shift): Likewise.
20492 (vectorizable_operation): Likewise.
20493 (vectorizable_store): Likewise.
20494 (vectorizable_condition): Likewise.
20495 (vectorizable_comparison): Likewise.
20497 2017-06-28 Michael Collison <michael.collison@arm.com>
20500 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20502 (arm_gen_movmemqi): Removed unused variable 'i'.
20503 Convert 'for' loop into 'while' loop.
20504 (arm_expand_prologue): Remove last unnecessary set of insn.
20505 (thumb_pop): Remove unused variable 'pushed_words'.
20506 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20508 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20510 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20511 * config/s390/s390-protos.h: Add prototype of
20512 s390_rel_address_ok_p.
20513 * config/s390/s390.c (s390_got_symbol): New function.
20514 (s390_rel_address_ok_p): New function.
20515 (legitimize_pic_address): Use s390_rel_address_ok_p.
20516 (s390_load_got): Use s390_got_symbol.
20517 (s390_option_override): Issue error if
20518 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20519 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20521 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20523 2017-06-27 Andrew Pinski <apinski@cavium.com>
20525 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20526 (X * copysign (1.0, X)): New pattern.
20527 (X * copysign (1.0, -X)): New pattern.
20528 (copysign (-1.0, CST)): New pattern.
20530 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20532 * genmultilib (combination_space): Remove variable.
20533 Validate reuse rules against regular expression for any sequence
20534 of multilib options in any order.
20536 2017-06-27 Michael Collison <michael.collison@arm.com>
20538 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20539 call aarch64_split_simd_combine.
20540 * (aarch64_combine_internal<mode>): Delete pattern.
20541 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20542 Allow register and subreg operands.
20544 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20546 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20547 specific need, just fallback on defaults.
20548 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20550 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20551 Olivier Hainque <hainque@adacore.com>
20553 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20555 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20556 targets. Pick a default if no particular attempt applied.
20557 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20560 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20562 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20563 (x86_64-wrs-vxworks7): Likewise.
20565 2017-06-27 Marek Polacek <polacek@redhat.com>
20568 * ubsan.c (instrument_null): Check get_base_address's result for null.
20570 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20572 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20574 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20576 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20577 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20578 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20579 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20580 New function types.
20581 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20582 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20583 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20584 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20585 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20586 BUILT_IN_FEUPDATEENV): New builtins.
20587 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20588 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20589 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20590 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20592 (builtin_structptr_types): Adjust size.
20593 * tree.c (builtin_structptr_types): Add four entries.
20595 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20596 Olivier Hainque <hainque@adacore.com>
20598 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20599 (TLS_SYM): New local macro, forcing reference to __tls__ on
20600 link command lines for VxWorks 7 RTPs, triggering initialization
20602 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20603 OS features TLS support, true for RTPs on VxWorks 7.
20604 * config/vxworks.c (vxworks_override_options): Setup emutls
20607 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20609 * predict.c (test_prediction_value_range): Use -1U instead of -1
20610 to avoid narrowing conversion warning.
20611 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20612 to avoid narrowing conversion warning.
20613 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20615 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20617 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20619 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20620 64bit configurations.
20621 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20622 (SIZE_TYPE): Likewise.
20623 * config/vxworks.c (vxworks_emutls_var_fields): Use
20624 long_unsigned_type_node instead of unsigned_type_node as the offset
20625 field type, which is "pointer" mode in emutls.c.
20627 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20630 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20632 PR middle-end/81207
20633 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20634 gimple_vuse copying separately from gimple_vdef copying.
20636 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20638 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20641 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20642 Olivier Hainque <hainque@adacore.com>
20644 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20645 for all vxworks7 targets.
20646 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20647 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20648 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20649 variations for VX6/VX7 and 32/64bits later on in ...
20650 (VXWORKS_LIB_SPEC): Leverage new macros.
20651 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20652 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20654 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20656 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20657 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20659 2017-06-26 Carl Love <cel@us.ibm.com>
20661 * config/rs6000/rs6000-c.c: Add support for built-in functions
20662 vector bool char vec_reve (vector bool char);
20663 vector signed char vec_reve (vector signed char);
20664 vector unsigned char vec_reve (vector unsigned char);
20665 vector bool int vec_reve (vector bool int);
20666 vector signed int vec_reve (vector signed int);
20667 vector unsigned int vec_reve (vector unsigned int);
20668 vector bool long long vec_reve (vector bool long long);
20669 vector signed long long vec_reve (vector signed long long);
20670 vector unsigned long long vec_reve (vector unsigned long long);
20671 vector bool short vec_reve (vector bool short);
20672 vector signed short vec_reve (vector signed short);
20673 vector double vec_reve (vector double);
20674 vector float vec_reve (vector float);
20675 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20676 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20677 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20678 (altivec_vreve): New pattern.
20679 * config/rs6000/altivec.h (vec_reve): New define.
20680 * doc/extend.texi (vec_rev): Update the built-in documentation file
20681 for the new built-in functions.
20683 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20685 PR tree-optimization/71815
20686 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20688 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20690 (slsr_process_phi): Likewise.
20691 (replace_uncond_cands_and_profitable_phis): Don't replace a
20692 multiply candidate with a stride of 1 (copy or cast).
20693 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20695 (lowest_cost_path): Likewise.
20696 (total_savings): Likewise.
20698 2017-06-26 Richard Biener <rguenther@suse.de>
20701 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20702 Use def_builtin_pure for all gather builtins.
20704 2017-06-26 Richard Biener <rguenther@suse.de>
20706 PR tree-optimization/81203
20707 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20708 non-dominating BBs.
20710 2017-06-26 Marek Polacek <polacek@redhat.com>
20713 * doc/invoke.texi: Document -Wmultistatement-macros.
20715 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20717 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20718 arm_neon_ok_no_float_abi effective target.
20720 2017-06-26 Richard Biener <rguenther@suse.de>
20722 PR tree-optimization/80928
20723 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20724 (copy_bbs): Set BB_DUPLICATED flag early.
20725 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20727 (execute_on_shrinking_pred): Likewise.
20728 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20729 BB_DUPLICATED blocks.
20730 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20731 iterate over all PHIs considering removal of *gsi.
20733 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20735 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20738 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20740 * config/rs6000/rs6000-string.c: (expand_block_clear,
20741 do_load_for_compare, select_block_compare_mode,
20742 compute_current_alignment, expand_block_compare,
20743 expand_strncmp_align_check, expand_strn_compare,
20744 expand_block_move, rs6000_output_load_multiple)
20745 Move functions related to string/block move/compare
20746 to a separate file.
20747 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20748 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20749 for this function which is now used in two files.
20750 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20751 * config.gcc: Add rs6000-string.o to extra_objs for
20752 targets powerpc*-*-* and rs6000*-*-*.
20754 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20757 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20758 32-bit, since indexed is not valid for DImode.
20759 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20760 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20761 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20762 (define_peephole2 for Altivec d-form store): Likewise.
20765 * multiple_target.c (create_dispatcher_calls): Only create the
20766 dispatcher call if the function is the default clone of a
20767 versioned function.
20769 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20771 PR middle-end/80902
20772 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20773 a call, force the call to not be a tail call.
20775 2017-06-23 Jeff Law <law@redhat.com>
20777 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20780 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20782 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20783 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20784 handling for early expansion of vector shifts (sl,sr,sra,rl).
20785 (builtin_function_type): Add vector shift right instructions
20786 to the unsigned argument list.
20788 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20790 rtl-optimizatoin/79286
20791 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20792 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20793 trap. PIC register plus a const unspec without offset can never trap.
20795 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20797 * tree.h (builtin_structptr_type): New type.
20798 (builtin_structptr_types): Declare new array.
20799 * tree.c (builtin_structptr_types): New array.
20800 (free_lang_data, build_common_tree_nodes): Use it.
20802 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20805 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20808 2017-06-22 Matt Turner <mattst88@gmail.com>
20810 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20811 Lake models to skylake case. Assume skylake for unknown
20812 models with clflushopt.
20814 2017-06-22 Jeff Law <law@redhat.com>
20816 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20817 frame sizes that do not satisfy aarch64_uimm12_shift.
20819 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20821 * profile-count.h (apply_probability,
20822 apply_scale, probability_in): Fix checks for zero.
20824 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20826 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20827 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20829 2016-06-22 Richard Biener <rguenther@suse.de>
20831 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20832 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20833 REDUC_MAX_EXPR support.
20834 (vectorizable_reduction): Likewise.
20835 (vect_create_epilog_for_reduction): Likewise.
20837 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20839 * match.pd (A / (1 << B) -> A >> B): New.
20840 * generic-match-head.c: Include optabs-tree.h.
20841 * gimple-match-head.c: Likewise.
20842 * optabs-tree.h (target_supports_op_p): New.
20843 * optabs-tree.c (target_supports_op_p): New.
20845 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20847 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20848 $gcc_cv_ld --help output.
20849 (gcc_cv_ld_demangle): Likewise.
20850 (gcc_cv_ld_eh_frame_hdr): Likewise.
20851 (gcc_cv_ld_pie): Likewise.
20852 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20853 (gcc_cv_ld_buildid): Likewise.
20854 (gcc_cv_ld_sysroot): Likewise.
20855 (ld_bndplt_support): Likewise.
20856 (ld_pushpopstate_support): Likewise.
20857 * configure: Regenerate.
20858 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20860 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20863 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20864 operands indexes to avoid gap between operands and match_dups.
20866 2017-06-21 Andrew Pinski <apinski@cavium.com>
20868 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20869 Increment Arith_shift and Arith_shift_reg by 1.
20870 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20872 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20873 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20874 (aarch64_strip_extend): Add new argument and test for it.
20875 (aarch64_cheap_mult_shift_p): New function.
20876 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20877 add a cost if it is true.
20878 Update calls to aarch64_strip_extend.
20879 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20881 2017-06-21 Andrew Pinski <apinski@cavium.com>
20883 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20885 (thunderxt88): Likewise.
20886 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20887 (thunderx_prefetch_tune): New variable.
20888 (thunderx2t99_prefetch_tune): Update for the correct values.
20889 (thunderxt88_tunings): New variable.
20890 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20891 generic_prefetch_tune.
20892 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20894 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20896 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20897 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20898 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20899 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20900 (aarch64_atomic_cas<mode>, GPI): Likewise.
20902 2017-06-21 Martin Liska <mliska@suse.cz>
20904 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20905 statements on cold and hot labels.
20906 * predict.c (tree_estimate_probability_bb): Remove the
20907 prediction from this place.
20909 2017-06-21 Martin Liska <mliska@suse.cz>
20911 PR tree-optimization/79489
20912 * gimplify.c (maybe_add_early_return_predict_stmt): New
20914 (gimplify_return_expr): Call the function.
20915 * predict.c (tree_estimate_probability_bb): Remove handling
20917 * predict.def: Update comment about early return predictor.
20918 * gimple-predict.h (is_gimple_predict): New function.
20919 * predict.def: Change default value of early return to 66.
20920 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20922 * passes.def: Put pass_strip_predict_hints to the beginning of
20925 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20927 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20928 FUNCTION_DECL declarations.
20929 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20931 (dwaf2out_decl): Likewise.
20932 * godump.c (go_early_global_decl): Skip call to the real debug hook
20933 for FUNCTION_DECL declarations.
20934 * passes.c (rest_of_decl_compilation): Skip call to the
20935 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20936 -fdump-go-spec is passed.
20938 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20940 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20941 Reorder for compactness.
20942 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20943 (def_builtin_pure, def_builtin_pure2): New functions.
20944 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20946 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20948 * match.pd (nop_convert): New predicate.
20949 ((A +- CST1) +- CST2): Allow some NOP conversions.
20951 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20954 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20955 with ctors/dtors if GOVD_SHARED is set.
20957 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20959 * config/aarch64/aarch64.md (movti_aarch64):
20960 Emit mov rather than orr.
20961 (movtf_aarch64): Likewise.
20962 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20963 Emit mov rather than orr.
20965 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20967 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20968 Swap alternatives, make integer dup more expensive.
20970 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20972 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20973 Return true for non-tls symbols.
20975 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20977 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20978 (cortex-a75): Likewise.
20979 (cortex-a75.cortex-a55): Likewise.
20980 * config/aarch64/aarch64-tune.md: Regenerate.
20981 * doc/invoke.texi (-mtune): Document new values for -mtune.
20983 2017-06-21 Tom de Vries <tom@codesourcery.com>
20985 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20986 stack_size feature.
20987 (Effective-Target Keywords, Other attributes): Suggest using
20988 dg-add-options stack_size feature to get stack limit in stack_size
20989 effective target documentation.
20991 2017-06-21 Julian Brown <julian@codesourcery.com>
20992 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20994 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20995 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20996 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20998 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20999 attribute type list for neon_multiply.
21000 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21001 attribute type list for neon_multiply.
21002 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21003 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21004 attribute type list for neon_multiply.
21005 * config/arm/types.md (crypto_pmull): Add.
21006 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21007 attribute type list.
21009 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21011 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21014 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21016 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21017 to make sure not to dereference a NULL cost_classes_ptr pointer.
21019 2017-06-20 Carl Love <cel@us.ibm.com>
21021 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21022 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21023 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21024 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21025 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21026 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21027 VMULOSW): New enum "unspec" values.
21028 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21029 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21030 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21031 altivec_vmulosw): New patterns.
21032 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21033 VMULOSW): Add definitions.
21035 2017-06-20 Julia Koval <julia.koval@intel.com>
21037 * config/i386/i386.c: Fix rounding expand for new pattern.
21038 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21040 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21042 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21043 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21045 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21047 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21050 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21052 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21053 architecture, then by implementer ID.
21054 * config/aarch64/aarch64-tune.md: Regenerate.
21056 2017-06-20 Richard Biener <rguenther@suse.de>
21058 PR middle-end/81097
21059 * fold-const.c (split_tree): Fold to type before negating.
21061 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21063 * diagnostic-show-locus.c
21064 (selftest::test_fixit_deletion_affecting_newline): New function.
21065 (selftest::diagnostic_show_locus_c_tests): Call it.
21067 2017-06-20 Andreas Schwab <schwab@suse.de>
21070 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21073 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21075 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21076 __ARM_FEATURE_COPROC according to support.
21078 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21080 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21081 Rewritten to avoid overflow for > 32-bit pointers.
21084 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21085 by removing enum keyword.
21086 (ubsan_type_descriptor): Likewise. Formatting fix.
21089 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21090 splitter): Require TARGET_SSE2 in the condition.
21092 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21095 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21096 for doing vector set of SFmode on ISA 3.0.
21097 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21098 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21100 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21101 SFmode value into a V4SF variable that was extracted from another
21102 V4SF variable without converting the element to double precision
21103 and back to single precision vector format.
21104 (vsx_insert_extract_v4sf_p9_2): Likewise.
21106 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21108 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21109 in UWHI to avoid undefined overflow.
21112 * ubsan.h (enum ubsan_encode_value_phase): New.
21113 (ubsan_encode_value): Change second argument to
21114 enum ubsan_encode_value_phase with default value of
21115 UBSAN_ENCODE_VALUE_GENERIC.
21116 * ubsan.c (ubsan_encode_value): Change second argument to
21117 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21118 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21119 create_tmp_var_raw instead of create_tmp_var and use a
21121 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21122 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21123 ubsan_encode_value callers.
21126 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21127 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21128 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21130 2017-06-19 Richard Biener <rguenther@suse.de>
21132 PR middle-end/81118
21133 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21134 estimates if we changed anything.
21136 2017-06-19 Richard Biener <rguenther@suse.de>
21138 PR tree-optimization/80887
21139 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21140 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21141 simplified lookups, then reset mprts_hook.
21142 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21144 (try_to_simplify): Likewise.
21146 2017-06-19 Martin Liska <mliska@suse.cz>
21149 * gimplify.c (gimplify_switch_expr):
21150 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21152 2017-06-19 Martin Liska <mliska@suse.cz>
21154 * doc/install.texi: Document that PGO runs in 4 stages.
21156 2017-06-19 Martin Liska <mliska@suse.cz>
21159 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21160 to dispatcher function name.
21161 * multiple_target.c (replace_function_decl): New function.
21162 (create_dispatcher_calls): Redirect both edges and references.
21164 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21166 * profile-count.c (profile_count::dump): Dump quality.
21167 (profile_count::differs_from_p): Update for unsigned val.
21168 * profile-count.h (profile_count_quality): New enum.
21169 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21171 2017-06-19 Richard Biener <rguenther@suse.de>
21173 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21174 struct function as arg.
21175 (estimate_numbers_of_iterations): Export overload with loop arg.
21176 (free_numbers_of_iterations_estimates_loop): Use an overload of
21177 free_numbers_of_iterations_estimates instead.
21178 * tree-cfg.c (remove_bb): Adjust.
21179 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21180 * tree-parloops.c (gen_parallel_loop): Likewise.
21181 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21183 (tree_unroll_loops_completely): Likewise.
21184 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21185 Use an overload instead and export.
21186 (estimated_loop_iterations): Adjust.
21187 (max_loop_iterations): Likewise.
21188 (likely_max_loop_iterations): Likewise.
21189 (estimate_numbers_of_iterations): Take struct function as arg
21191 (loop_exits_before_overflow): Adjust.
21192 (free_numbers_of_iterations_estimates_loop): Use an overload.
21193 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21194 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21196 2017-06-19 Richard Biener <rguenther@suse.de>
21199 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21200 RANGE_EXPR conservatively.
21202 2017-06-16 Carl Love <cel@us.ibm.com>
21204 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21205 definitions for vec_float, vec_float2, vec_floato,
21206 vec_floate built-ins.
21207 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21208 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21210 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21211 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21212 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21213 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21214 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21215 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21216 vec_floato): Add builtin defines.
21217 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21218 Update the built-in documentation file for the new built-in
21221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21223 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21224 (mthumb): Mark as the negative of -marm.
21226 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21228 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21230 (ARM Options, -mtune): Document that this accepts the same
21231 extension options as -mcpu.
21232 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21236 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21237 permitted extensions.
21239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21241 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21242 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21243 (armv8-m.main): Add option +nodsp.
21244 * config/arm/arm-cpu-cdata.h: Regenerated.
21246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21248 * config/arm/t-fuchsia: New file.
21249 * config.gcc (arm*-*-fuchsia*): Use it.
21251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21253 * config/arm/t-symbian: Rewrite for new option infrastructure.
21255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21257 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21258 (MULTILIB_REQUIRED): Likewise.
21260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21262 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21263 (MULTILIB_RESUE): Likewise.
21264 (MULTILIB_MATCHES): Likewise.
21265 (MULTLIB_REQUIRED): Likewise.
21267 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21269 * config/arm/t-rtems: Rewrite for new option framework.
21271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21273 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21274 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21275 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21276 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21277 * config/arm/t-multilib: ... here.
21278 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21279 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21280 armv7-a and armv8*-a when A-profile libraries have not been built.
21281 * config/arm/t-rmprofile: Rewrite.
21283 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21285 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21286 with a backslash. Remove the backslash after substituting unescaped
21288 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21292 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21293 the driver to pass through the default setting of -mfloat-abi.
21294 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21296 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21297 (all_feat_combs): New rule.
21298 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21300 * config/arm/t-aprofile: Rewrite.
21302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21304 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21305 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21306 fpu is not specified by the user/command-line.
21307 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21308 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21309 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21310 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21311 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21312 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21314 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21316 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21317 * config/arm/t-arm-elf: Rewritten.
21319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21321 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21322 have some floating-point instructions.
21323 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21324 (TARGET_MAYBE_HARD_FLOAT): New macro.
21325 * config/arm/arm-builtins.c (arm_init_builtins): Use
21326 TARGET_MAYBE_HARD_FLOAT.
21327 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21331 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21332 (configargs.h): Include it.
21333 (arm_print_hint_for_fpu_option): New function.
21334 (arm_parse_fpu_option): New function.
21335 (candidate_extension): New class.
21336 (arm_canon_for_multilib): New function.
21337 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21338 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21339 (ARCH_CANONICAL_SPECS): New macro.
21340 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21342 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21344 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21345 are set after handling multilib fragments. Set target_cpu_default2
21348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21350 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21352 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21353 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21355 (check_arch): Likewise.
21356 * config/arm/arm.c (arm_configure_build_target): Handle
21357 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21358 options in the default.
21360 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21362 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21363 when an option is an alias of another.
21364 * config/arm/parsecpu.awk (optalias): New parser token.
21365 (gen_comm_data): Mark non-alias options as such. Emit entries
21366 for extension aliases.
21367 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21368 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21369 (armv6kz, armv6zk, armv6t2): Likewise.
21370 (armv7): Make vfpv3-d16 an alias.
21371 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21373 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21374 Sort in canonical order.
21375 (armv8-a): Sort in canonical order.
21376 (armv8.1-a, armv8.2-a): Likewise.
21377 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21379 (cortex-a9): Sort in canonical order.
21380 * config/arm/arm.c (selftests.h): Include it.
21381 (arm_test_cpu_arch_data): New function.
21382 (arm_run_self_tests): New function.
21383 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21384 (targetm): Move declaration to the end of the file.
21385 * arm-cpu-cdata.h: Regenerated.
21387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21389 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21390 call to target_mode_check describing the type of option passed.
21391 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21392 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21393 arm_parse_cpu_option_name to match parameters against list of
21395 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21396 arm_arch_core_flags data structure.
21397 * config/arm/arm-cpu_cdata.h: Regenerated.
21399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21401 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21403 (arm_print_hint_for_cpu_option): Likewise.
21404 (arm_print_hint_for_arch_option): Likewise.
21405 (arm_parse_cpu_option_name): Likewise.
21406 (arm_parse_arch_option_name): Likewise.
21407 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21408 of entries in the all_fpus list.
21409 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21410 (arm_parse_cpu_option_name): Declare.
21411 (arm_parse_arch_option_name): Declare.
21412 (arm_parse_option_features): Declare.
21413 (arm_intialize_isa): Declare.
21414 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21416 (gen_comm_data): ... here. Make definitions non-static.
21417 * config/arm/arm-cpu-data.h: Regenerated.
21418 * config/arm/arm-cpu-cdata.h: Regenerated.
21420 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21422 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21423 (cpu_arch_extension): New structure.
21424 (cpu_arch_option, arch_option, cpu_option): New structures.
21425 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21426 architecture types.
21427 (gen_data): Generate new format data tables.
21428 * config/arm/arm.c (cpu_tune): New structure.
21429 (cpu_option, processors): Delete.
21430 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21431 (arm_print_hint_for_cpu_option): ... this and ...
21432 (arm_print_hint_for_arch_option): ... this.
21433 (arm_parse_arch_cpu_name): Delete. Replace with ...
21434 (arm_parse_cpu_option_name): ... this and ...
21435 (arm_parse_arch_option_name): ... this.
21436 (arm_unrecognized_feature): Change type of target parameter to
21438 (arm_parse_arch_cpu_features): Delete. Replace with ...
21439 (arm_parse_option_features): ... this.
21440 (arm_configure_build_target): Rework to use new configuration data
21442 (arm_print_tune_info): Rework for new configuration data tables.
21443 * config/arm/arm-cpu-data.h: Regenerated.
21444 * config/arm/arm-cpu.h: Regenerated.
21446 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21448 * Makefile.in (OBJS): Move sbitmap.o from here ...
21449 (OBJS-libcommon): ... to here.
21451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21453 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21454 (ISA_ALL_CRYPTO): New macro.
21455 (ISA_ALL_SIMD): New macro
21456 (ISA_ALL_FP): New macro.
21457 * config/arm/arm.c (fpu_bitlist): Update initializer.
21458 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21460 (arm9e): Add fpu. Add option for nofp
21461 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21462 (arm926ej-s, arm1026ej-s): Likewise.
21463 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21464 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21465 neon-fp16, neon-vfpv4, nofp and nosimd.
21466 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21467 (cortex-a8): Add fpu. Add option for nofp.
21468 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21469 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21470 (cortex-r4f): Add fpu.
21471 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21472 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21474 (cortex-r8): Likewise.
21475 (cortex-m4): Add fpu. Add option for nofp.
21476 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21477 (cortex-a17.cortex-a7): Likewise.
21478 (cortex-a32): Add fpu. Add options for crypto and nofp.
21479 (cortex-a35, cortex-a53): Likewise.
21480 (cortex-a57): Add fpu. Add option for crypto.
21481 (cortex-a72, cortex-a73): Likewise.
21482 (exynos-m1): Likewise.
21483 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21484 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21485 (cortex-m33): Add fpu. Add option for nofp.
21486 * config/arm/arm-cpu-cdata.h: Regenerated
21487 * config/arm/arm-cpu-data.h: Regenerated.
21489 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21491 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21492 (armv5te, armv5tej): Likewise.
21493 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21494 (armv7): Add options fp and vfpv3-d16.
21495 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21496 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21498 (armv7ve): Likewise.
21499 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21500 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21501 (armv8-a): Add nocrypto option.
21502 (armv8.1-a, armv8.2-a): Likewise.
21503 (armv8-m.main): add options fp, fp.dp and nofp.
21505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21507 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21509 (armv8-a+crc): Delete.
21510 (armv8.1-a): Add options simd, crypto and nofp.
21511 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21512 (armv8.2-a+fp16): Delete.
21513 (armv8-m.main): Add option dsp.
21514 (armv8-m.main+dsp): Delete.
21515 (cortex-a8): Add fpu. Add nofp option.
21516 (cortex-a9): Add fpu. Add nofp and nosimd options.
21517 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21518 link to main cpu and architecture data structures.
21519 (gen_comm_data): Only put isa attributes from the main architecture
21521 (option): New statement for architecture and CPU entries.
21522 * arm.c (struct cpu_option): New structure.
21523 (struct processors): Add entry for options.
21524 (arm_unrecognized_feature): New function.
21525 (arm_parse_arch_cpu_name): Ignore any characters after the first
21527 (arm_parse_arch_cpu_feature): New function.
21528 (arm_configure_build_target): Separate out any CPU and architecture
21529 features and parse separately. Don't error out if -mfpu=auto is
21530 used with only an architecture string.
21531 (arm_print_asm_arch_directives): New function.
21532 (arm_file_start): Call it.
21533 * config/arm/arm-cpu-cdata.h: Regenerated.
21534 * config/arm/arm-cpu-data.h: Likewise.
21535 * config/arm/arm-tables.opt: Likewise.
21537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21539 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21540 assembler when it is not -mfpu=auto.
21542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21544 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21545 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21546 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21547 (ASM_CPU_SPEC): Rewrite.
21548 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21549 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21550 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21551 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21552 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21553 copied string is NUL-terminated. Also strip any characters prefixed
21555 (arm_rewrite_selected_arch): New function.
21556 (arm_rewrite_march): New function.
21558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21560 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21561 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21562 (march, mcpu, mtune): Convert to string-based options.
21563 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21564 (arm_parse_arch_cpu_name): New function.
21565 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21566 identify selected architecture or CPU.
21567 (arm_option_save): New function.
21568 (TARGET_OPTION_SAVE): Redefine.
21569 (arm_option_restore): Restore string options.
21570 (arm_option_print): Print string options.
21572 2017-06-16 Martin Sebor <msebor@redhat.com>
21574 PR tree-optimization/80933
21575 PR tree-optimization/80934
21576 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21577 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21578 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21579 (gimple_fold_builtin): Call them.
21581 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21583 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21584 as unlikely; update profile.
21586 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21588 * predict.c (force_edge_cold): Handle declaring edges impossible
21591 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21593 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21595 (try_unroll_loop_completely): Fix reporting.
21597 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21599 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21601 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21604 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21605 if given a non-constant argument for an intrinsic which requires a
21608 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21610 * profile.c (compare_freqs): New function.
21611 (branch_prob): Sort edge list.
21612 (find_spanning_tree): Assume that the list is priority sorted.
21614 2017-06-16 Richard Biener <rguenther@suse.de>
21616 PR tree-optimization/81090
21617 * passes.def (pass_record_bounds): Remove.
21618 * tree-pass.h (make_pass_record_bounds): Likewise.
21619 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21620 make_pass_record_bounds): Likewise.
21621 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21622 not free niter estimates at the beginning but at the end.
21623 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21625 2017-06-16 Richard Biener <rguenther@suse.de>
21627 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21628 initializer to workaround ICE in host GCC 4.8.
21630 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21632 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21634 (clone_inlined_nodes): Update.
21636 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21638 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21639 prefetch settings, and enable prefetching by default at -O3.
21641 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21643 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21644 Set flag_prefetch_loop_arrays according to tuning data.
21646 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21648 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21649 New tune structure.
21650 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21651 [Unrelated to main purpose of the patch] Place the pointer field last
21652 to enable type checking errors when tune structure are wrongly merged.
21653 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21654 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21655 (thunderx2t99_prefetch_tune): New tune constants.
21656 (tune_params *_tunings): Update all tunings (no functional change).
21657 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21658 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21659 from tunings structures.
21661 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21664 * ubsan.c (instrument_null): Add T argument, use it instead
21665 of computing it based on IS_LHS.
21666 (instrument_object_size): Likewise.
21667 (pass_ubsan::execute): Adjust instrument_null and
21668 instrument_object_size callers to pass gimple_get_lhs or
21669 gimple_assign_rhs1 result to it. Use instrument_null instead of
21670 calling get_base_address and instrument_mem_ref. Handle
21671 aggregate call arguments for object-size sanitization.
21673 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21675 PR tree-optimization/81089
21676 * tree-vrp.c (is_masked_range_test): Validate operands of
21679 2017-06-15 Martin Sebor <msebor@redhat.com>
21682 * dumpfile.c (dump_register): Avoid calling memset to initialize
21683 a class with a default ctor.
21684 * gcc.c (struct compiler): Remove const qualification.
21685 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21686 * hash-table.h: Ditto.
21687 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21689 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21690 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21692 * params.h (struct param_info): Make struct members non-const.
21693 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21694 with default initialization.
21695 * vec.h (vec_copy_construct, vec_default_construct): New helper
21697 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21698 with vec_copy_construct.
21699 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21700 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21701 * doc/invoke.texi (-Wclass-memaccess): Document.
21703 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21705 * emit-rtl.h (is_leaf): Update comment about local
21706 register allocator.
21708 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21711 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21712 for a variable to have a section before checking if the section has a
21714 Set section to.persistent if persistent attribute is set.
21715 Warn if .persistent attribute is used on an automatic variable.
21717 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21719 PR rtl-optimization/80474
21720 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21722 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21724 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21727 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21729 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21730 (casesi): Emit bounds checking as RTL.
21731 (casesi_internal_mips16_<mode>): Remove bounds checking.
21733 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21735 * config/xtensa/xtensa.c (xtensa_option_override): Append
21736 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21737 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21738 xtensa_doloop_hooks): Define unconditionally.
21739 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21741 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21742 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21743 for it in xtensa_option_override.
21744 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21745 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21747 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21749 * doc/cppopts.texi: Document '-' special value to -MF.
21751 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21753 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21754 (cortex_a53_fconst): Likewise.
21755 (cortex_a53_fpmul): Likewise.
21756 (cortex_a53_f_load_64): Likewise.
21757 (cortex_a53_f_load_many): Likewise.
21758 (cortex_a53_advsimd_alu): Likewise.
21759 (cortex_a53_advsimd_alu_q): Likewise.
21760 (cortex_a53_advsimd_mul): Likewise.
21761 (cortex_a53_advsimd_mul_q): Likewise.
21762 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21763 Add missing fmul, r2f_cvt and fconst cases.
21765 2017-06-14 Richard Biener <rguenther@suse.de>
21767 PR middle-end/81088
21768 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21770 (fold_binary_loc): When associating do not treat pre-existing
21771 TREE_OVERFLOW on literal constants as a reason to allow
21772 TREE_OVERFLOW on associated literal constants.
21774 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21776 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21777 (MASK_FEATURES): New macro.
21778 * config/sparc/sparc.c (sparc_option_override): Remove the special
21779 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21781 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21783 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21784 a division of 0 if non-call exceptions are enabled.
21786 2017-06-14 Andrew Pinski <apinski@cavium.com>
21787 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21790 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21791 Improve vector initialization code gen for only variable case.
21793 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21795 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21797 2017-06-14 Richard Biener <rguenther@suse.de>
21799 PR tree-optimization/81083
21800 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21803 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21805 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21806 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21807 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21808 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21809 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21810 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21812 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21814 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21815 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21817 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21819 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21821 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21823 * config/rs6000/t-rtems: Don't handle SPE.
21825 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21827 * config/rs6000/t-linux: Don't handle SPE.
21829 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21831 * config/rs6000/eabispe.h: Delete file.
21833 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21835 * config/rs6000/t-spe: Delete file.
21837 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21839 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21840 (rs6000_legitimate_offset_address_p): Return false for anything in
21841 V2SImode or V2SFmode.
21843 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21845 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21846 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21847 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21849 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21850 (rs6000_legitimate_offset_address_p): Ditto.
21851 (rs6000_emit_move): Ditto.
21852 (rs6000_init_builtins): Remove V4HI_type_node.
21854 2017-06-13 Martin Liska <mliska@suse.cz>
21857 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21858 (gate_asan): Likewise.
21859 * asan.h (asan_no_sanitize_address_p): Remove the function.
21860 (sanitize_flags_p): New function.
21861 * builtins.def: Fix coding style.
21862 * common.opt: Use renamed enum value.
21863 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21864 * doc/extend.texi: Document no_sanitize attribute.
21865 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21866 to SANITIZE_UNDEFINED_NONDEFAULT.
21867 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21868 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21869 Use sanitize_flags_p.
21870 * gimplify.c (gimplify_function_tree): Likewise.
21871 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21872 * opts.c (parse_no_sanitize_attribute): New function.
21873 (common_handle_option): Use renamed enum value.
21874 * opts.h (parse_no_sanitize_attribute): Declare.
21875 * tree.c (sanitize_flags_p): New function.
21876 * tree.h: Declared here.
21877 * tsan.c: Use sanitize_flags_p.
21878 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21879 (instrument_mem_ref): Likewise.
21880 (instrument_bool_enum_load): Likewise.
21881 (do_ubsan_in_current_function): Remove the function.
21882 (pass_ubsan::execute): Use sanitize_flags_p.
21883 * ubsan.h: Remove do_ubsan_in_current_function
21884 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21885 (dump_function_to_file): Use it here.
21887 2017-06-13 Martin Jambor <mjambor@suse.cz>
21889 PR tree-optimization/80803
21890 PR tree-optimization/81063
21891 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21892 (propagate_subaccesses_across_link): Enqueue subtree whenever
21893 necessary instead of relying on the caller.
21895 2017-06-13 Martin Jambor <mjambor@suse.cz>
21897 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21898 that have a first_link.
21899 (sort_and_splice_var_accesses): Do not check first_link before
21901 (subtree_mark_written_and_enqueue): Likewise.
21902 (propagate_all_subaccesses): Likewise and do not stop at first
21903 parent with a first_link.
21905 2017-06-13 Martin Jambor <mjambor@suse.cz>
21907 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21910 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21912 * match.pd: New pattern.
21914 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21916 * tree-vrp.c (is_masked_range_test): New function.
21917 (register_edge_assert_for): Determine ranges for
21920 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21922 PR tree-optimization/67328
21923 * fold-const.c (maskable_range_p): New function.
21924 (build_range_check): Generate bittests if possible.
21926 2017-06-13 Martin Liska <mliska@suse.cz>
21928 * gimple-pretty-print.c (dump_probability): Add new argument.
21929 (dump_edge_probability): Dump both probability and count.
21930 (dump_gimple_label): Likewise.
21931 (dump_gimple_bb_header): Likewise.
21933 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21936 * config/avr/avr-devices.c: Fix indentation.
21937 * config/avr/gen-avr-mmcu-specs.c: Dito.
21939 2017-06-13 Richard Biener <rguenther@suse.de>
21941 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21942 instead get vector type from stmt_info.
21943 (vectorizable_reduction): Adjust. Remove dead code.
21945 2017-06-13 Richard Biener <rguenther@suse.de>
21947 PR middle-end/81065
21948 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21949 case of C * (x * C2 + C3).
21950 (fold_addr_of_array_ref_difference): Properly fold index difference.
21952 2017-06-12 David S. Miller <davem@davemloft.net>
21955 * config/sparc/sparc.md (return expander): Emit frame blockage if
21956 function uses alloca.
21958 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21960 * combine.c (make_field_assignment): Check len rather than the mode
21961 precision when calling force_to_mode.
21963 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21965 Support multilibs and devices that see flash in RAM address range.
21968 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21969 (avr_mcu_t) <flash_pm_offset>: New field.
21970 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21971 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21972 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21973 (AVR_TINY_PM_OFFSET): Remove macro.
21974 * config/avr/avr.opt (-mshort-calls): New option.
21975 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21976 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21977 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21978 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21979 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21980 instead of avr_arch->have_jmp_call.
21981 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21982 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21983 avr_arch->flash_pm_offset to define.
21984 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21985 new field flash_pm_offset. Add entry for avrxmega3.
21986 (avr_texinfo): Add entry for avrxmega3.
21987 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21988 attiny212, attiny214,
21989 attiny412, attiny414, attiny416, attiny417,
21990 attiny814, attiny816, attiny817,
21991 attiny1614, attiny1616, attiny1617,
21992 attiny3214, attiny3216, attiny3217.
21993 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21994 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21995 (avr_print_operand_address) [AVR_TINY]: Same.
21996 (avr_asm_init_sections) <readonly_data_section>: Only patch
21997 callback if avr_arch->flash_pm_offset = 0.
21998 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21999 for rodata if avr_arch->flash_pm_offset != 0.
22000 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22001 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22002 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22003 (m_options): Append opt_rcall.
22004 (m_dirnames): Append dir_rcall.
22005 * config/avr/t-multilib: Regenerate.
22007 * configure.ac [target=avr]: Check whether avrxmega3 default
22008 linker description file works as needed.
22009 * configure: Regenerate.
22010 * doc/avr-mmcu.texi: Regenerate.
22011 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22012 <__AVR_ARCH__>: Document avrxmega3 and 103.
22013 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22014 <__AVR_SHORT_CALLS__>: Document it.
22015 <__AVR_PM_BASE_ADDRESS__>: Document it.
22016 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22017 (AVR Variable Attributes) <progmem>: Document this is
22018 not needed for avrxmega3.
22019 (AVR Named Address Spaces) <__flash>: Dito.
22021 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22023 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22025 2017-06-12 Doug Rupp <rupp@adacore.com>
22027 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22028 Append vxworks-stdint.h to the tm_file list.
22029 * config/vxworks-stdint.h: New file.
22031 2017-06-12 Martin Liska <mliska@suse.cz>
22033 PR tree-optimization/81041
22034 * tree-profile.c (gimple_gen_ic_func_profiler):
22035 Create an extra BB in profile-generate
22036 (gimple_gen_time_profiler): Likewise.
22038 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22040 PR tree-optimization/81003
22041 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22042 (update_range_test): Use it instead of force_gimple_operand_gsi.
22044 2017-06-12 Richard Biener <rguenther@suse.de>
22046 PR tree-optimization/81053
22047 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22048 with backedge value not defined in loop. Simplify def stmt
22051 2017-06-11 Tom de Vries <tom@codesourcery.com>
22054 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22056 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22057 nvptx_cannot_force_const_mem.
22059 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22061 * opts.c (finish_options): Move test for flag_split_stack after
22062 it has been initialized.
22064 2017-06-11 Jason Merrill <jason@redhat.com>
22066 * tree.h (id_equal): New.
22067 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22068 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22069 instead of strcmp of IDENTIFIER_POINTER.
22071 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22073 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22074 (mark_all_inlined_calls_cdtor): Fix formating.
22075 (inline_transform): Rescale profile before inlining.
22077 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22079 * cgraph.h (cgraph_edge::clone): Update prototype.
22080 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22081 (cgraph_node::create_clone): Update.
22082 (cgraph_node::create_version_clone): Update.
22083 * tree-inline.c (copy_bb): Update.
22084 (expand_call_inline): Update.
22086 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22088 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22089 factored out from ...
22090 (rs6000_emit_prologue): ... here.
22092 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22094 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22095 factored out from ...
22096 (rs6000_emit_prologue): ... here.
22098 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22100 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22102 (handle_missing_profiles): Fix computation of tp_first_run.
22103 (counts_to_freqs): Do not touch freqs when count is 0.
22105 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22107 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22110 2017-06-10 Tom de Vries <tom@codesourcery.com>
22112 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22113 attributes): Document signal effective target.
22115 2017-06-10 Tom de Vries <tom@codesourcery.com>
22117 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22118 Document effective target stack_size.
22120 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22122 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22123 to the edit_context if they can be auto-applied.
22125 2017-06-9 Ian Lance Taylor <iant@golang.org>
22127 * opts.c (finish_options): If -fsplit-stack, disable implicit
22128 -forder-blocks-and-partition.
22129 * doc/invoke.texi (Optimize Options): Document that when using
22130 -fsplit-stack -forder-blocks-and-partition is not implicitly
22133 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22135 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22136 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22137 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22138 * builtins.def (abort, trap, unreachable): Declare cold.
22139 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22140 * tree-core.h (ECF_COLD): New.
22141 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22142 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22144 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22146 * predict.c (unlikely_executed_stmt_p): Cleanup.
22148 2017-06-09 Richard Biener <rguenther@suse.de>
22150 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22151 model if the ref is always written to.
22153 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22155 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22157 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22159 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22162 2017-06-09 Tom de Vries <tom@codesourcery.com>
22165 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22166 "target cannot support label values" when encountering LABEL_REF.
22168 2017-06-09 Martin Liska <mliska@suse.cz>
22170 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22171 (gimple_gen_ic_func_profiler): Emit direct comparison
22172 of __gcov_indirect_call_callee with NULL.
22173 (gimple_gen_time_profiler): Change probability from
22174 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22176 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22178 * profile.c (edge_gcov_counts): Turn to pointer.
22179 (compute_branch_probabilities, compute_branch_probabilities): Update.
22180 (branch_prob): Do not clear edge_gcov_count.
22181 * profile.h (edge_gcov_counts): Turn to pointer.
22182 (edge_gcov_count): Update.
22184 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22186 * gimple.h (gimple_check_failed): Mark cold.
22188 2017-06-09 Richard Biener <rguenther@suse.de>
22190 PR tree-optimization/66623
22191 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22192 refactor check_reduction into two parts, properly computing
22193 whether we have to check reduction validity for outer loop
22196 2017-06-09 Richard Biener <rguenther@suse.de>
22198 PR tree-optimization/79483
22199 * graphite-scop-detection.c (order): New global.
22200 (get_order): Compute bb to order mapping that satisfies code
22201 generation constraints.
22202 (cmp_pbbs): New helper.
22203 (build_scops): Start domwalk at entry block, sort generated
22206 2017-06-09 Richard Biener <rguenther@suse.de>
22208 PR middle-end/81007
22209 * ipa-polymorphic-call.c
22210 (ipa_polymorphic_call_context::restrict_to_inner_class):
22211 Skip FIELD_DECLs with error_mark_node type.
22212 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22215 2017-06-09 Martin Liska <mliska@suse.cz>
22217 * predict.c (struct branch_predictor): New struct.
22218 (test_prediction_value_range): New test.
22219 (predict_c_tests): New function.
22220 * selftest-run-tests.c (selftest::run_tests): Run the function.
22221 * selftest.h: Declare new tests.
22223 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22226 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22227 gen_add3_insn did not fail.
22228 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22229 r0, construct that number in a temporary reg and add that reg to r0.
22230 If asked to put the result in r0 as well, fail.
22232 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22234 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22235 for early expansion of vec_eqv.
22237 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22239 PR middle-end/81005
22240 * ubsan.c (instrument_null): Avoid pointless code temporary.
22241 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22243 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22247 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22250 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22253 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22255 * predict.c (unlikely_executed_edge_p): Move ahead.
22256 (probably_never_executed_edge_p): Use it.
22258 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22260 PR middle-end/79988
22261 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22262 gimple_call_builtin_p call.
22264 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22266 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22267 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22268 rtl_check_failed_type2, rtl_check_failed_code1,
22269 rtl_check_failed_code2, rtl_check_failed_code_mode,
22270 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22271 rtvec_check_failed_bounds, rtl_check_failed_flag,
22272 _fatal_insn_not_found, _fatal_insn): Likewise.
22273 * tree.h (tree_contains_struct_check_failed,
22274 tree_check_failed, tree_not_check_failed,
22275 tree_class_check_failed, tree_range_check_failed,
22276 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22277 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22278 tree_operand_check_failed, omp_clause_check_failed,
22279 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22282 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22284 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22285 flag_branch_probabilities.
22286 * ipa-inline.c (edge_badness): Likewise.
22287 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22288 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22289 * predict.c (maybe_hot_frequency_p): Likewise.
22290 (probably_never_executed): Likewise.
22291 * sched-ebb.c (schedule_ebbs): Likewise.
22292 * sched-rgn.c (find_single_block_region): Likewise.
22293 * tracer.c (tail_duplicate): Likewise.
22295 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22297 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22298 longer requires x_flag_profile_use.
22300 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22302 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22303 instead of flag_reorder_blocks_and_partition.
22304 * dbxout.c (dbxout_function_end): Likewise.
22305 * dwarf2out.c (gen_subprogram_die): Likewise.
22306 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22307 * hw-doloop.c (reorg_loops): Likewise.
22308 * varasm.c (assemble_start_function,
22309 assemble_end_function): Likewise.
22310 (decide_function_section): Do not check for
22311 flag_reorder_blocks_and_partition.
22313 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22315 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22317 (chkp_get_hard_register_fake_addr_expr): Ditto.
22318 (chkp_build_addr_expr): Add check for hard reg case.
22319 (chkp_parse_array_and_component_ref): Ditto.
22320 (chkp_find_bounds_1): Ditto.
22321 (chkp_process_stmt): Don't generate bounds store for
22324 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22326 * predict.c (maybe_hot_bb_p): Do not check profile status.
22327 (maybe_hot_edge_p): Likewise.
22328 (probably_never_executed): Check for zero counts even if profile
22330 (unlikely_executed_edge_p): New function.
22331 (unlikely_executed_stmt_p): New function.
22332 (unlikely_executed_bb_p): New function.
22333 (set_even_probabilities): Use unlikely predicates.
22334 (combine_predictions_for_bb): Likewise.
22335 (predict_paths_for_bb): Likewise.
22336 (predict_paths_leading_to_edge): Likewise.
22337 (determine_unlikely_bbs): New function.
22338 (estimate_bb_frequencies): Use it.
22339 (compute_function_frequency): Use zero counts even if profile is
22341 * profile-count.h: Fix typo.
22343 2017-08-08 Julia Koval <julia.koval@intel.com>
22345 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22346 _mm512_mask_cvtsepi16_storeu_epi8,
22347 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22348 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22349 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22350 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22351 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22352 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22353 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22354 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22355 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22356 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22357 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22358 __builtin_ia32_pmovuswb256mem_mask,
22359 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22360 __builtin_ia32_pmovwb512mem_mask): New builtins.
22362 2017-08-08 Julia Koval <julia.koval@intel.com>
22364 PR target/73350,80862
22365 * config/i386/subst.md (round): Fix round pattern.
22366 * config/i386/i386.c (ix86_erase_embedded_rounding):
22367 Fix erasing rounding for the fixed pattern.
22369 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22371 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22373 2017-06-08 Martin Liska <mliska@suse.cz>
22375 PR gcov-profile/80911
22376 * gcov.c (block_info::block_info): New constructor.
22378 2017-06-07 Carl Love <cel@us.ibm.com>
22380 * config/rs6000/rs6000-c: The return type of the following
22381 built-in functions was implemented as int not long long. Fix sign
22382 of return value for the unsigned version of vec_mulo and vec_mule.
22383 vector unsigned long long vec_bperm (vector unsigned long long,
22384 vector unsigned char)
22385 vector signed long long vec_mule (vector signed int,
22387 vector unsigned long long vec_mule (vector unsigned int,
22388 vector unsigned int)
22389 vector signed long long vec_mulo (vector signed int,
22391 vector unsigned long long vec_mulo (vector unsigned int,
22392 vector unsigned int)
22393 * doc/extend.texi: Fix the documentation for the built-in
22396 2017-06-07 Carl Love <cel@us.ibm.com>
22399 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22402 2017-06-07 Carl Love <cel@us.ibm.com>
22404 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22405 support, Generate doublehv for signed int/float for BE case only.
22407 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22409 * doc/invoke.texi (mcx16): Rewrite.
22411 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22413 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22414 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22415 *mov<mode>_softfloat, and an anonymous splitter): Use
22416 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22418 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22420 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22422 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22423 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22424 (rs6000_debug_reg_global): Adjust.
22425 (rs6000_init_hard_regno_mode_ok): Adjust.
22426 (rs6000_dbx_register_number): Adjust.
22427 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22428 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22429 Remove SPE_ACC and SPEFSCR.
22430 (REG_ALLOC_ORDER): Ditto.
22431 (FRAME_POINTER_REGNUM): Change to 111.
22432 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22433 (REG_CLASS_NAMES): Ditto.
22434 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22435 (REGISTER_NAMES): Ditto.
22436 (ADDITIONAL_REG_NAMES): Ditto.
22437 (rs6000_reg_names): Ditto.
22438 * config/rs6000/rs6000.md: Renumber some register number
22441 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22443 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22445 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22446 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22448 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22449 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22450 Delete the SPE high registers.
22451 (REG_ALLOC_ORDER): Ditto.
22452 (enum reg_class): Remove SPE_HIGH_REGS.
22453 (REG_CLASS_NAMES): Ditto.
22454 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22455 (REGISTER_NAMES): Ditto.
22456 (rs6000_reg_names): Ditto.
22457 * doc/tm.texi.in: Remove SPE as example.
22458 * doc/tm.texi: Regenerate.
22460 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22462 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22463 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22464 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22465 * config/rs6000/rs6000.md (type): Remove "brinc".
22467 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22469 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22470 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22471 * config/rs6000/linuxspe.h: Delete file.
22472 * config/rs6000/rs6000.md: Don't include spe.md.
22473 * config/rs6000/spe.h: Delete file.
22474 * config/rs6000/spe.md: Delete file.
22475 * config/rs6000/t-rs6000: Remove spe.md.
22477 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22479 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22480 (reg_or_none500mem_operand): Delete.
22481 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22482 instead of reg_or_none500mem_operand.
22484 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22486 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22487 handling of SPE flags.
22488 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22490 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22492 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22494 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22495 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22496 paired_divv2sf3): Similar.
22497 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22498 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22499 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22501 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22502 Rename the paired_* instruction patterns.
22503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22505 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22506 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22507 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22508 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22509 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22510 PAIRED_VECTOR_MODE.
22511 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22512 (spe_func_has_64bit_regs_p): Delete.
22513 (spe_expand_predicate_builtin): Delete.
22514 (spe_expand_evsel_builtin): Delete.
22515 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22516 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22517 (invalid_e500_subreg): Delete.
22518 (rs6000_legitimize_address): Always force_reg op2 as well, for
22519 paired single memory accesses.
22520 (rs6000_member_type_forces_blk): Delete.
22521 (rs6000_spe_function_arg): Delete.
22522 (rs6000_expand_unop_builtin): Delete SPE handling.
22523 (rs6000_expand_binop_builtin): Ditto.
22524 (spe_expand_stv_builtin): Delete.
22525 (bdesc_2arg_spe): Delete.
22526 (spe_expand_builtin): Delete.
22527 (spe_expand_predicate_builtin): Delete.
22528 (spe_expand_evsel_builtin): Delete.
22529 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22530 (spe_init_builtins): Delete.
22531 (spe_func_has_64bit_regs_p): Delete.
22532 (savres_routine_name): Delete "info" parameter. Adjust callers.
22533 (rs6000_emit_stack_reset): Ditto.
22534 (rs6000_dwarf_register_span): Delete.
22535 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22536 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22537 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22539 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22541 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22542 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22543 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22544 mulv2sf3, divv2sf3): Delete expanders.
22546 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22548 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22550 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22552 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22553 * config/rs6000/rs6000.c: Ditto.
22555 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22557 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22558 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22559 comparison_operator.
22561 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22563 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22564 * config/rs6000/rs6000.opt: Ditto.
22565 * config/rs6000/t-rtems: Ditto.
22567 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22569 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22570 TARGET_E500_SINGLE by 0, simplify.
22571 * config/rs6000/rs6000.c: Ditto.
22572 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22573 (spe_build_register_parallel): Delete.
22574 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22575 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22576 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22577 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22578 (E500_CONVERT): Delete.
22579 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22581 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22583 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22584 * config/rs6000/dfp.md: Ditto.
22585 (negdd2, *negdd2_fpr): Merge.
22586 (absdd2, *absdd2_fpr): Merge.
22587 (negtd2, *negtd2_fpr): Merge.
22588 (abstd2, *abstd2_fpr): Merge.
22589 * config/rs6000/e500.h: Delete file.
22590 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22591 TARGET_FPRS by 1 and simplify.
22592 * config/rs6000/rs6000-c.c: Ditto.
22593 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22594 TARGET_DF_SPE by 0.
22595 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22597 * config/rs6000/rs6000.md: Ditto.
22598 (floatdidf2, *floatdidf2_fpr): Merge.
22599 (move_from_CR_gt_bit): Delete.
22600 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22601 (E500_CR_IOR_COMPARE): Delete.
22602 (All patterns that require !TARGET_FPRS): Delete.
22603 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22605 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22607 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22609 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22611 * graphds.c (add_edge): Intitialize edge's attached data.
22612 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22613 pointer parameter. Call pointed function on each edge during
22614 graph traversing. Skip traversing the edge when the function
22616 (graphds_dfs, graphds_scc): Ditto.
22617 (for_each_edge): New parameter. Pass the new parameter to callback
22619 * graphds.h (skip_edge_callback): New function pointer type.
22620 (graphds_dfs, graphds_scc): New function pointer parameter.
22621 (graphds_edge_callback, for_each_edge): New parameter.
22623 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22625 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22626 out code checking if runtime alias check is possible to below ...
22627 Call the new function.
22628 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22629 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22631 2017-06-07 Marek Polacek <polacek@redhat.com>
22634 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22635 TYPE_OVERFLOW_WRAPS check.
22637 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22639 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22640 if versioning is required.
22641 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22642 peeling with the check for versioning.
22644 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22646 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22647 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22648 Set true to new parameter if new ssa variable is defined.
22649 (vect_gen_vector_loop_niters): Refactor. Set range information
22650 for the new vector loop bound variable.
22651 (vect_do_peeling): Ditto.
22653 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22655 * tree-affine.c (ssa.h): Include header file.
22656 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22657 has wrapping overflow behavior.
22659 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22661 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22663 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22665 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22666 (tree_to_aff_combination): ... here.
22668 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22670 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22671 reg_pressure model function.
22672 (ivopts_global_cost_for_size): Delete.
22673 (determine_set_costs, iv_ca_recount_cost): Call new model function
22674 ivopts_estimate_reg_pressure.
22676 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22678 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22679 expensive than udiv. Remove floating point cases from mod.
22681 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22683 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22684 Increase idiv cost.
22686 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22688 * config/aarch64/aarch64.md
22689 (copysignsf3): Fix mask generation.
22691 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22693 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22695 (class dump_manager): Add register_dumps method.
22696 * dumpfile.c: Include langhooks.h.
22697 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22698 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22699 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22700 (dump_manager::dump_register): Start with 512 entries instead of 32.
22701 (dump_manager::register_dumps): New method.
22702 * toplev.c (general_init): Instead of invoking register_dumps
22703 langhook, invoke register_dumps method on the dump manager.
22704 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22707 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22709 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22710 Say that # requires an associated define_split to exist, and that
22711 the define_split must be suitable for use after register allocation.
22713 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22715 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22716 (compute_outgoing_frequencies): Also initialize zero counts.
22717 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22718 around loops; preserve more of profile when nothing changes.
22720 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22722 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22724 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22725 * config/arm/arm-cpu-cdata.h: Regenerate.
22726 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22727 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22728 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22729 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22730 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22732 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22734 * config/arm/t-rmprofile: Likewise.
22735 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22737 2017-06-06 David S. Miller <davem@davemloft.net>
22740 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22741 blockage if function uses alloca.
22743 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22745 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22746 New "uid" fields to hold pretty-print IDs of group and ref.
22747 Memory references are now identified as <group_id>:<ref_id>
22748 instead of using [random] addresses.
22749 (dump_mem_details): Simplify, no functional change.
22750 (dump_mem_ref): Simplify and make output more concise.
22751 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22752 (find_or_create_group): Initialize group uid.
22753 (record_ref): Initialize ref uid. Improve debug output.
22754 (prune_group_by_reuse, should_issue_prefetch_p,)
22755 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22756 (mark_nontemporal_store, determine_loop_nest_reuse):
22757 Improve debug output.
22759 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22761 * dbgcnt.def (prefetch): New debug counter.
22762 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22763 (schedule_prefetches): Stop issueing prefetches if debug counter
22766 2017-06-06 Tom de Vries <tom@codesourcery.com>
22768 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22769 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22771 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22773 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22774 Use aarch64_reg_or_zero predicate for operand 4.
22775 (aarch64_compare_and_swap<mode> define_insn_and_split):
22776 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22777 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22779 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22781 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22782 (arm_compute_save_core_reg_mask): This.
22783 (thumb1_compute_save_reg_mask): Rename into ...
22784 (thumb1_compute_save_core_reg_mask): This.
22785 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22786 (arm_compute_frame_layout): Likewise.
22788 2017-06-06 Richard Biener <rguenther@suse.de>
22790 PR tree-optimization/80974
22791 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22792 keep or clear leaders SSA info.
22794 2017-06-06 Tom de Vries <tom@codesourcery.com>
22796 * config/nvptx/nvptx.c (split_mode_p): New function.
22797 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22799 2017-06-06 Tom de Vries <tom@codesourcery.com>
22801 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22803 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22806 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22809 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22811 * shrink-wrap.c (handle_simple_exit): Update profile.
22812 (try_shrink_wrapping): Upate profile.
22814 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22816 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22817 (tree_guess_outgoing_edge_probabilities): New.
22818 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22819 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22821 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22823 * ipa-split.c (split_function): Initialize return bb profile.
22825 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22827 * profile.c (compute_branch_probabilities): Also initialize
22828 EXIT_BLOCK profile.
22830 2017-06-06 Richard Biener <rguenther@suse.de>
22832 PR tree-optimization/80928
22833 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22834 (vect_analyze_loop_operations): Properly guard analysis for
22836 (vect_transform_loop): Likewise.
22837 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22838 (vect_model_induction_cost): Do not cost for pure SLP.
22839 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22840 of induction in inner loop vectorization.
22841 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22842 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22843 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22845 (vect_analyze_slp_cost_1): Cost induction.
22846 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22847 (vect_get_slp_vect_defs): Likewise.
22848 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22849 (vect_transform_stmt): Handle SLP reductions.
22850 * tree-vectorizer.h (vectorizable_induction): Adjust.
22852 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22854 * config/rs6000/rs6000.c (make_resolver_func): Update
22855 init_lowered_empty_function call.
22857 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22859 * doc/invoke.texi: Document the -fprofile-abs-path option.
22860 * common.opt (fprofile-abs-path): New option.
22861 * gcov-io.h (gcov_write_filename): Declare.
22862 * gcov-io.c (gcov_write_filename): New function.
22863 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22864 * profile.c (output_location): Likewise.
22866 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22868 * shring-wrap.c: Revert accidental commit.
22870 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22872 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22874 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22876 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22878 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22879 profile in callgraph edge.
22880 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22881 (apply_scale): Likewise.
22882 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22883 Also scale profile when inlining function with zero profile.
22884 (initialize_cfun): Update exit block profile even when it is zero.
22885 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22886 when profile is read.
22888 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22890 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22891 (CLONE_*): New constants to define the processors we can generate
22892 code for with the target_clone attribute.
22893 (rs6000_clone_map): New array to identify which clone processors
22894 the current program is running on.
22895 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22896 target_clone attribute.
22897 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22898 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22899 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22900 (cpu_expand_builtin): Add support for target_clone attribute.
22901 (rs6000_valid_attribute_p): Allow "default" attribute.
22902 (get_decl_name): New debug function to simplify printing the
22903 current function name in debugging statements.
22904 (rs6000_clone_priority): New functions to support the target_clone
22905 attribute, and be able to generate code to switch between ISA 2.05
22906 through ISA 3.0 (power6 through power9).
22907 (rs6000_compare_version_priority): Likewise.
22908 (rs6000_get_function_versions_dispatcher): Likewise.
22909 (make_resolver_func): Likewise.
22910 (add_condition_to_bb): Likewise.
22911 (dispatch_function_versions): Likewise.
22912 (rs6000_generate_version_dispatcher_body): Likewise.
22913 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22914 (fusion_gpr_load_p): Fix a spacing issue.
22915 * doc/extend.texi (Common Function Attributes): Document that the
22916 PowerPC supports the target_clone attribute.
22918 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22920 * config/arm/arm.h: explain F symbol found in description of ARM
22921 register allocation in its legend.
22923 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22925 * config/mips/frame-header-opt.c: Include profile-count.h.
22926 * config/riscv/riscv.c: Include profile-count.h
22928 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22930 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22932 (sm_set_flag_if_changed): Add bbs field.
22933 (execute_sm_if_changed_flag_set): Pass BBS.
22934 (execute_sm): Update.
22936 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22938 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22941 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22943 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22944 (peephole2): New peephole2 to emit the above.
22945 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22947 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22949 * config/aarch64/aarch64.c (define_peephole2 above
22950 *sub_<shift>_<mode>): New peephole.
22952 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22954 * config/i386/i386.c (make_resolver_func): Update.
22955 * Makefile.in: Add profile-count.h and profile-count.o
22956 * auto-profile.c (afdo_indirect_call): Update to new API.
22957 (afdo_set_bb_count): Update.
22958 (afdo_propagate_edge): Update.
22959 (afdo_propagate_circuit): Update.
22960 (afdo_calculate_branch_prob): Update.
22961 (afdo_annotate_cfg): Update.
22962 * basic-block.h: Include profile-count.h
22963 (struct edge_def): Turn count to profile_count.
22964 (struct basic_block_def): Likewie.
22965 (REG_BR_PROB_BASE): Move to profile-count.h
22966 (RDIV): Move to profile-count.h
22967 * bb-reorder.c (max_entry_count): Turn to profile_count.
22968 (find_traces): Update.
22969 (rotate_loop):Update.
22970 (connect_traces):Update.
22971 (sanitize_hot_paths):Update.
22972 * bt-load.c (migrate_btr_defs): Update.
22973 * cfg.c (RDIV): Remove.
22974 (init_flow): Use alloc_block.
22975 (alloc_block): Uninitialize count.
22976 (unchecked_make_edge): Uninitialize count.
22977 (check_bb_profile): Update.
22978 (dump_edge_info): Update.
22979 (dump_bb_info): Update.
22980 (update_bb_profile_for_threading): Update.
22981 (scale_bbs_frequencies_int): Update.
22982 (scale_bbs_frequencies_gcov_type): Update.
22983 (scale_bbs_frequencies_profile_count): New.
22984 * cfg.h (update_bb_profile_for_threading): Update.
22985 (scale_bbs_frequencies_profile_count): Declare.
22986 * cfgbuild.c (compute_outgoing_frequencies): Update.
22987 (find_many_sub_basic_blocks): Update.
22988 * cfgcleanup.c (try_forward_edges): Update.
22989 (try_crossjump_to_edge): Update.
22990 * cfgexpand.c (expand_gimple_tailcall): Update.
22991 (construct_exit_block): Update.
22992 * cfghooks.c (verify_flow_info): Update.
22993 (dump_bb_for_graph): Update.
22994 (split_edge): Update.
22995 (make_forwarder_block): Update.
22996 (duplicate_block): Update.
22997 (account_profile_record): Update.
22998 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22999 (get_estimated_loop_iterations): Update.
23000 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23001 (single_likely_exit): Update.
23002 * cfgloopmanip.c (scale_loop_profile): Update.
23004 (set_zero_probability): Update.
23005 (lv_adjust_loop_entry_edge): Update.
23006 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23007 (purge_dead_edges): Update.
23008 (rtl_account_profile_record): Update.
23009 * cgraph.c (cgraph_node::create): Uninitialize count.
23010 (symbol_table::create_edge): Uninitialize count.
23011 (cgraph_update_edges_for_call_stmt_node): Update.
23012 (cgraph_edge::dump_edge_flags): Update.
23013 (cgraph_node::dump): Update.
23014 (cgraph_edge::maybe_hot_p): Update.
23015 * cgraph.h: Include profile-count.h
23016 (create_clone), create_edge, create_indirect_edge): Update.
23017 (cgraph_node): Turn count to profile_count.
23018 (cgraph_edge0: Likewise.
23019 (make_speculative, clone): Update.
23020 (create_edge): Update.
23021 (init_lowered_empty_function): Update.
23022 * cgraphclones.c (cgraph_edge::clone): Update.
23023 (duplicate_thunk_for_node): Update.
23024 (cgraph_node::create_clone): Update.
23025 * cgraphunit.c (cgraph_node::analyze): Update.
23026 (cgraph_node::expand_thunk): Update.
23027 * final.c (dump_basic_block_info): Update.
23028 * gimple-streamer-in.c (input_bb): Update.
23029 * gimple-streamer-out.c (output_bb): Update.
23030 * graphite.c (print_global_statistics): Update.
23031 (print_graphite_scop_statistics): Update.
23032 * hsa-brig.c: Include basic-block.h.
23033 * hsa-dump.c: Include basic-block.h.
23034 * hsa-gen.c (T sum_slice): Update.
23035 (convert_switch_statements):Update.
23036 * hsa-regalloc.c: Include basic-block.h.
23037 * ipa-chkp.c (chkp_produce_thunks): Update.
23038 * ipa-cp.c (struct caller_statistics): Update.
23039 (init_caller_stats): Update.
23040 (gather_caller_stats): Update.
23041 (ipcp_cloning_candidate_p): Update.
23042 (good_cloning_opportunity_p): Update.
23043 (get_info_about_necessary_edges): Update.
23044 (dump_profile_updates): Update.
23045 (update_profiling_info): Update.
23046 (update_specialized_profile): Update.
23047 (perhaps_add_new_callers): Update.
23048 (decide_about_value): Update.
23049 (ipa_cp_c_finalize): Update.
23050 * ipa-devirt.c (struct odr_type_warn_count): Update.
23051 (struct decl_warn_count): Update.
23052 (struct final_warning_record): Update.
23053 (possible_polymorphic_call_targets): Update.
23054 (ipa_devirt): Update.
23055 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23056 * ipa-icf.c (sem_function::merge): Update.
23057 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23058 * ipa-inline.c (compute_uninlined_call_time): Update.
23059 (compute_inlined_call_time): Update.
23060 (want_inline_small_function_p): Update.
23061 (want_inline_self_recursive_call_p): Update.
23062 (edge_badness): Update.
23063 (lookup_recursive_calls): Update.
23064 (recursive_inlining): Update.
23065 (inline_small_functions): Update.
23066 (dump_overall_stats): Update.
23067 (dump_inline_stats): Update.
23068 * ipa-profile.c (ipa_profile_generate_summary): Update.
23069 (ipa_propagate_frequency): Update.
23070 (ipa_profile): Update.
23071 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23072 * ipa-utils.c (ipa_merge_profiles): Update.
23073 * loop-doloop.c (doloop_modify): Update.
23074 * loop-unroll.c (report_unroll): Update.
23075 (unroll_loop_runtime_iterations): Update.
23076 * lto-cgraph.c (lto_output_edge): Update.
23077 (lto_output_node): Update.
23078 (input_node): Update.
23079 (input_edge): Update.
23080 (merge_profile_summaries): Update.
23081 * lto-streamer-in.c (input_cfg): Update.
23082 * lto-streamer-out.c (output_cfg): Update.
23083 * mcf.c (create_fixup_graph): Update.
23084 (adjust_cfg_counts): Update.
23085 (sum_edge_counts): Update.
23086 * modulo-sched.c (sms_schedule): Update.
23087 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23088 * predict.c (maybe_hot_count_p): Update.
23089 (probably_never_executed): Update.
23090 (dump_prediction): Update.
23091 (combine_predictions_for_bb): Update.
23092 (propagate_freq): Update.
23093 (handle_missing_profiles): Update.
23094 (counts_to_freqs): Update.
23095 (rebuild_frequencies): Update.
23096 (force_edge_cold): Update.
23097 * predict.h: Include profile-count.h
23098 (maybe_hot_count_p, counts_to_freqs): UPdate.
23099 * print-rtl-function.c: Do not include cfg.h
23100 * print-rtl.c: Include basic-block.h
23101 * profile-count.c: New file.
23102 * profile-count.h: New file.
23103 * profile.c (is_edge_inconsistent): Update.
23104 (correct_negative_edge_counts): Update.
23105 (is_inconsistent): Update.
23106 (set_bb_counts): Update.
23107 (read_profile_edge_counts): Update.
23108 (compute_frequency_overlap): Update.
23109 (compute_branch_probabilities): Update; Initialize and deinitialize
23111 (branch_prob): Update.
23112 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23113 (edge_gcov_count): New.
23114 (bb_gcov_count): New.
23115 * shrink-wrap.c (try_shrink_wrapping): Update.
23116 * tracer.c (better_p): Update.
23117 * trans-mem.c (expand_transaction): Update.
23118 (ipa_tm_insert_irr_call): Update.
23119 (ipa_tm_insert_gettmclone_call): Update.
23120 * tree-call-cdce.c: Update.
23121 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23122 (gimple_duplicate_sese_tail): Update.
23123 (gimple_account_profile_record): Update.
23124 (execute_fixup_cfg): Update.
23125 * tree-inline.c (copy_bb): Update.
23126 (copy_edges_for_bb): Update.
23127 (initialize_cfun): Update.
23128 (freqs_to_counts): Update.
23129 (copy_cfg_body): Update.
23130 (expand_call_inline): Update.
23131 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23132 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23133 (try_unroll_loop_completely): Update.
23134 (try_peel_loop): Update.
23135 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23136 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23137 * tree-ssa-loop-split.c (connect_loops): Update.
23138 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23139 * tree-ssa-reassoc.c (branch_fixup): Update.
23140 * tree-ssa-tail-merge.c (replace_block_by): Update.
23141 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23142 (compute_path_counts): Update.
23143 (update_profile): Update.
23144 (recompute_probabilities): Update.
23145 (update_joiner_offpath_counts): Update.
23146 (estimated_freqs_path): Update.
23147 (freqs_to_counts_path): Update.
23148 (clear_counts_path): Update.
23149 (ssa_fix_duplicate_block_edges): Update.
23150 (duplicate_thread_path): Update.
23151 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23152 (struct switch_conv_info): Update.
23153 * tree-tailcall.c (decrease_profile): Update.
23154 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23155 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23156 * value-prof.c (check_counter): Update.
23157 (gimple_divmod_fixed_value): Update.
23158 (gimple_mod_pow2): Update.
23159 (gimple_mod_subtract): Update.
23160 (gimple_ic_transform): Update.
23161 (gimple_stringop_fixed_value): Update.
23162 * value-prof.h (gimple_ic): Update.
23164 2017-06-02 Carl Love <cel@us.ibm.com>
23166 * config/rs6000/rs6000-c: Add support for built-in functions
23167 vector double vec_doublee (vector signed int);
23168 vector double vec_doublee (vector unsigned int);
23169 vector double vec_doublee (vector float);
23170 vector double vec_doubleh (vector signed int);
23171 vector double vec_doubleh (vector unsigned int);
23172 vector double vec_doubleh (vector float);
23173 vector double vec_doublel (vector signed int);
23174 vector double vec_doublel (vector unsigned int);
23175 vector double vec_doublel (vector float);
23176 vector double vec_doubleo (vector signed int);
23177 vector double vec_doubleo (vector unsigned int);
23178 vector double vec_doubleo (vector float);.
23179 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23180 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23182 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23183 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23184 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23186 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23187 vec_doublel, vec_doubleh.
23188 * doc/extend.texi: Update the built-in documentation file for the
23189 new built-in functions.
23191 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23194 * ipa-inline-analysis.c (free_growth_caches): Set
23195 edge_removal_hook_holder to NULL after removing it.
23197 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23199 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23200 comparision with zero.
23202 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23203 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23204 for early expansion of vec_min and vec_max builtins.
23205 (builtin_function_type): Add min/max unsigned variants to those
23206 identified as having unsigned arguments.
23208 2017-06-02 Olivier Hainque <hainque@adacore.com>
23210 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23212 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23214 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23215 Use VALL_F16 iterator rather than VALL.
23217 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23219 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23220 Emit CBNZ inside loop when doing a strong exchange and comparing
23221 against zero. Generate the CC flags after the loop.
23223 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23225 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23226 (dl_section_ref): New.
23227 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23228 On AIX, append an expression to subtract the size of the
23229 section length to dl_section_ref.
23231 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23233 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23234 for early expansion of vector absolute builtins.
23236 2017-06-02 Richard Biener <rguenther@suse.de>
23238 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23239 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23241 2017-06-02 Richard Biener <rguenther@suse.de>
23243 PR tree-optimization/80948
23244 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23245 stmt order as well.
23247 2017-06-02 Richard Biener <rguenther@suse.de>
23249 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23251 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23252 for inductions as relevant.
23254 2017-06-02 Richard Biener <rguenther@suse.de>
23256 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23257 (vectorizable_induction): ... this. Remove dead code.
23259 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23261 * builtins. (expand_builtin_alloca): Remove second parameter and
23262 infer its value from the first parameter instead.
23263 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23265 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23267 PR rtl-optimization/80903
23268 * loop-doloop.c (add_test): Unshare sequence.
23270 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23272 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23274 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23276 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23278 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23279 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23280 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23281 (xlogue_layout::get_stub_rtx): Make static.
23282 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23283 (xlogue_layout::compute_stub_managed_regs): Rename to...
23284 (xlogue_layout::count_stub_managed_regs): ...this.
23285 (xlogue_layout::is_stub_managed_reg): New function.
23286 (xlogue_layout::m_stub_names): Rename to...
23287 (xlogue_layout::s_stub_names): ...this, make static.
23288 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23289 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23290 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23291 xlogue_layout::s_stub_names): Instantiate statics.
23292 (stub_managed_regs): Remove.
23293 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23294 (disable_call_ms2sysv_xlogues): Rename to...
23295 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23296 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23298 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23299 change after reload_completed.
23300 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23302 (ix86_expand_prologue): Likewise.
23303 (ix86_expand_epilogue): Likewise.
23304 (ix86_expand_split_stack_prologue): Likewise.
23305 (ix86_compute_frame_layout): Remove frame parameter ...
23306 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23307 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23309 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23310 (ix86_frame): Move from here ...
23311 * config/i386/i386.h (ix86_frame): ... to here.
23312 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23313 complete ix86_frame data structure instead. Remove some_ld_name.
23315 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23317 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23318 symbols that hold a DECL_VALUE_EXPR.
23320 2017-06-01 Martin Jambor <mjambor@suse.cz>
23322 PR tree-optimization/80898
23323 * tree-sra.c (process_subtree_disqualification): Removed.
23324 (disqualify_candidate): Do not acll
23325 process_subtree_disqualification.
23326 (subtree_mark_written_and_enqueue): New function.
23327 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23328 RHS has been disqualified and re-queue LHS if necessary. Apart
23329 from that, ignore disqualified RHS.
23331 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23333 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23334 address fetch for z10 or later.
23336 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23338 * config/arc/arc.md (tst_movb): Add guard when splitting.
23340 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23342 * config/arc/arc.c (arc_can_eliminate): Test against
23343 arc_frame_pointer_needed.
23345 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23347 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23348 to prevent store reordering.
23349 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23350 (type): Add block type.
23351 (stack_tie): Define special instruction to be used in
23354 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23356 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23357 constraint. It is not valid for the pattern.
23358 (noncommutative_binary_comparison): Likewise.
23360 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23362 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23365 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23367 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23368 be used by the reg-alloc.
23370 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23372 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23373 reg-alloc when having mul64 or mul32x16 instructions.
23374 (mulsidi3): Likewise.
23375 (umulsidi3): Likewise.
23376 (mulsi32x16): New pattern.
23377 (mulsi64): Likewise.
23378 (mulsidi64): Likewise.
23379 (umulsidi64): Likewise.
23380 (MUL32x16_REG): Define.
23381 (mul64_600): Use MUL32x16_REG.
23382 (mac64_600): Likewise.
23383 (umul64_600): Likewise.
23384 (umac64_600): Likewise.
23386 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23388 * config/arc/arc.md (mulsi3_700): Make it commutative.
23390 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23392 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23394 (*sign_extendsidi2_insn): Likewise for movstosw.
23396 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23398 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23399 the type of the input discriminant value. Convert the
23400 discriminant value of signedness vary.
23402 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23404 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23405 Add to -Wall section.
23407 2017-06-01 Richard Biener <rguenther@suse.de>
23409 PR middle-end/66313
23410 * fold-const.c (fold_plusminus_mult_expr): If the factored
23411 factor may be zero use a wrapping type for the inner operation.
23412 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23413 and handle moved defs.
23414 (process_assignment): Properly guard the unary op case. Return a
23415 tri-state indicating that moving the stmt before the call may allow
23416 to continue. Pass through to_move.
23417 (find_tail_calls): Handle moving unrelated defs before
23420 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23423 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23424 splitter result in the canonical way.
23426 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23428 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23429 also for 32bit target. Update insn attributes.
23430 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23432 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23434 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23435 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23436 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23437 (_mm_maskz_min_ss): New intrinsics.
23439 2017-05-31 Martin Liska <mliska@suse.cz>
23441 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23442 Change comment style to one we normally use.
23443 (vectorizable_reduction): Likewise.
23444 (vectorizable_induction): Likewise.
23445 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23446 (vectorizable_call): Likewise.
23447 (vectorizable_simd_clone_call): Likewise.
23448 (vectorizable_conversion): Likewise.
23449 (vectorizable_assignment): Likewise.
23450 (vectorizable_shift): Likewise.
23451 (vectorizable_operation): Likewise.
23452 (vectorizable_store): Likewise.
23453 (vectorizable_load): Likewise.
23454 * tree-vectorizer.h: Likewise.
23456 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23458 * passes.c (emergency_dump_function): New.
23459 * tree-pass.h (emergency_dump_function): Declare.
23460 * plugin.c (plugins_internal_error_function): Remove.
23461 * plugin.h (plugins_internal_error_function): Remove declaration.
23462 * toplev.c (internal_error_function): New static function. Use it...
23463 (general_init): ...here.
23465 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23467 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23468 (arc_rtx_costs): Add costs for new patterns.
23469 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23470 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23472 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23474 * tree-ssa-strlen.c (get_next_strinfo): New function.
23475 (get_stridx_plus_constant): Use it.
23476 (zero_length_string): Likewise.
23477 (adjust_related_strinfos): Likewise.
23478 (adjust_last_stmt): Likewise.
23480 2017-05-31 Richard Biener <rguenther@suse.de>
23483 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23484 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23486 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23488 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23489 loop_vinfo argument and use of dependence distance vectors.
23490 Check instead whether the two references differ only in their
23491 initial value and assume that they have the same alignment if the
23492 difference is a multiple of the vector alignment.
23493 (vect_analyze_data_refs_alignment): Update call accordingly.
23495 2017-05-31 Martin Liska <mliska@suse.cz>
23498 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23500 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23502 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23503 (create_intersect_range_checks): Move from ...
23504 * tree-data-ref.c (create_intersect_range_checks_index)
23505 (create_intersect_range_checks): ... to here.
23506 (create_runtime_alias_checks): New function factored from ...
23507 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23508 here. Call above function.
23509 * tree-data-ref.h (create_runtime_alias_checks): New function.
23511 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23513 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23514 segment length for dr_b and compute it in wide_int.
23516 2017-05-31 Richard Biener <rguenther@suse.de>
23518 PR tree-optimization/80906
23519 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23520 and pass through iv_map.
23521 (copy_bb_and_scalar_dependences): Adjust.
23522 (translate_pending_phi_nodes): Likewise.
23523 (copy_loop_close_phi_args): Handle code-generating IVs instead
23526 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23528 * diagnostic-color.c (color_dict): Add "type-diff".
23529 (parse_gcc_colors): Update comment.
23530 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23531 -fdiagnostics-show-template-tree and -fno-elide-type.
23532 (GCC_COLORS): Add type-diff to example.
23534 (-fdiagnostics-show-template-tree): New.
23535 (-fno-elide-type): New.
23536 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23537 the pp_format_decoder callback. Call any m_format_postprocessor's
23539 (pretty_printer::pretty_printer): Initialize
23540 m_format_postprocessor.
23541 (pretty_printer::~pretty_printer): Delete any
23542 m_format_postprocessor.
23543 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23544 (class format_postprocessor): New class.
23545 (struct pretty_printer::format_decoder): Document the new parameters.
23546 (struct pretty_printer::m_format_postprocessor): New field.
23547 * tree-diagnostic.c (default_tree_printer): Update for new
23548 bool and const char ** params.
23549 * tree-diagnostic.h (default_tree_printer): Likewise.
23551 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23553 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23554 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23555 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23556 rs6000_gen_cell_microcode code.
23557 (rs6000_final_prescan_insn): Delete.
23558 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23559 "warn-cell-microcode" entries.
23560 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23561 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23562 throughout. Change cc_reg_not_micro_cr0_operand to
23563 cc_reg_not_cr0_operand throughout.
23564 (*extendhi<mode>2_noload): Delete.
23565 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23566 (mwarn-cell-microcode): Delete.
23567 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23568 -mgen-cell-microcode and -mwarn-cell-microcode.
23570 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23573 * config/i386/constraints.md (Yd): New constraint.
23575 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23576 and (?Yd, r) alternatives. Update insn attributes.
23577 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23578 and (?*Yd, r) alternatives. Update insn attributes.
23579 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23581 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23583 * gimplify.c (gimplify_modify_expr): Don't create a
23584 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23587 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23589 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23591 2017-05-30 Richard Biener <rguenther@suse.de>
23593 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23594 and reduc_def fields.
23595 (STMT_VINFO_REDUC_TYPE): New define.
23596 (STMT_VINFO_REDUC_DEF): Likewise.
23597 (vect_force_simple_reduction): Adjust prototype.
23598 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23599 (vect_is_simple_reduction): Remove check_reduction argument.
23600 (vect_force_simple_reduction): Adjust and set
23601 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23602 (vectorizable_reduction): Do not re-do reduction analysis
23603 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23604 * tree-parloops.c (gather_scalar_reductions): Adjust.
23606 2017-05-30 Richard Biener <rguenther@suse.de>
23608 PR middle-end/80901
23609 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23612 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23614 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23615 Introduce unknown_misalignment parameter and remove vf.
23616 (vect_peeling_hash_get_lowest_cost):
23617 Pass unknown_misalignment parameter.
23618 (vect_enhance_data_refs_alignment):
23619 Fix unsupportable data ref treatment.
23621 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23623 * tree-vect-data-refs.c (vect_get_data_access_cost):
23624 Workaround for SLP handling.
23625 (vect_enhance_data_refs_alignment):
23626 Compute costs for doing no peeling at all, compare to the best
23627 peeling costs so far and avoid peeling if cheaper.
23629 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23631 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23632 Return peeling info and set costs to zero for unlimited cost
23634 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23635 with unknown misalignment. Compute and costs for unknown
23636 misalignment, compare them to the costs for known misalignment
23637 and choose the cheapest for peeling.
23639 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23641 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23642 (vect_get_peeling_costs_all_drs): Create function.
23643 (vect_peeling_hash_get_lowest_cost):
23644 Use vect_get_peeling_costs_all_drs.
23645 (vect_peeling_supportable): Create function.
23646 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23648 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23650 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23651 DR_HAS_NEGATIVE_STEP.
23652 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23653 (vect_enhance_data_refs_alignment): Use.
23654 (vect_duplicate_ssa_name_ptr_info): Use.
23655 * tree-vectorizer.h (dr_misalignment): Use.
23656 (known_alignment_for_access_p): Use.
23658 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23661 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23663 (has_section_name): New function.
23665 2017-05-30 Martin Liska <mliska@suse.cz>
23668 * auto-profile.c (get_function_decl_from_block): Fix
23671 2017-05-30 Richard Biener <rguenther@suse.de>
23673 PR middle-end/80876
23674 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23676 2017-05-30 Martin Liska <mliska@suse.cz>
23678 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23679 * dumpfile.h (struct dump_file_info): Remove ctors.
23681 2017-05-30 Martin Liska <mliska@suse.cz>
23683 * predict.def: Fix GNU coding style.
23685 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23687 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23688 Mark 'to' argument with ATTRIBUTE_UNUSED.
23690 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23692 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23693 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23694 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23697 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23699 * doc/install.texi (Options specification): Restore entry of
23700 --enable-sjlj-exceptions.
23702 2017-05-27 Michael Eager <eager@eagercon.com>
23705 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23707 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23709 * config/microblaze/microblaze.h
23710 (FIXED_REGISTERS): Update in macro.
23711 (CALL_USED_REGISTERS): Update in macro.
23713 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23715 * doc/install.texi: Add links to macOS binary distributions.
23717 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23721 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23723 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23725 2017-05-26 Martin Liska <mliska@suse.cz>
23727 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23729 2017-05-26 Martin Liska <mliska@suse.cz>
23731 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23732 always leading ';; '.
23733 (dump_bb_info): Likewise.
23734 (brief_dump_cfg): Likewise.
23735 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23736 * dumpfile.c: Remove usage of TDF_VERBOSE.
23737 * dumpfile.h (enum dump_kind): Likewise.
23738 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23739 * print-tree.c (debug_verbose): Remove.
23740 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23741 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23742 * tree-diagnostic.c (default_tree_printer): Replace
23743 TDF_DIAGNOSTIC with TDF_SLIM.
23745 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23747 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23748 in parameter loop, rather than loop_vinfo.
23749 (create_intersect_range_checks): Ditto.
23750 (vect_create_cond_for_alias_checks): Update call to above functions.
23752 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23754 PR tree-optimization/80815
23755 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23756 for merging runtime alias checks. Handle negative DR_STEPs.
23758 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23760 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23762 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23763 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23764 out code pruning runtime alias checks.
23765 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23766 factored out from above.
23767 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23769 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23771 (prune_runtime_alias_test_list): New decalaration.
23773 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23775 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23776 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23777 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23778 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23779 (operator==, comp_dr_with_seg_len_pair): Ditto.
23780 (vect_prune_runtime_alias_test_list): Ditto.
23782 2017-05-26 Martin Liska <mliska@suse.cz>
23785 * params.def: Bound partial-inlining-entry-probability param.
23787 2017-05-26 Marek Polacek <polacek@redhat.com>
23790 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23793 2017-05-26 Richard Biener <rguenther@suse.de>
23795 PR tree-optimization/80842
23796 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23799 2017-05-26 Richard Biener <rguenther@suse.de>
23801 PR tree-optimization/80844
23802 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23804 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23806 * doc/md.texi (Machine Constraints): Update x86 family
23807 machine constraints section to match 'config/i386/constraints.md'.
23809 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23811 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23813 2017-05-25 Nathan Sidwell <nathan@acm.org>
23815 * doc/invoke.texi (--enable-languages): Update documentation.
23817 2017-05-25 Martin Liska <mliska@suse.cz>
23819 * dumpfile.c: Add TDF_FOLDING.
23820 * dumpfile.h (enum dump_kind): Likewise.
23821 * genmatch.c (dt_simplify::gen_1): Use it.
23823 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23825 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23827 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23829 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23830 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23832 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23834 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23835 * match.pd (X == C): Rewrite it here.
23836 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23837 with_certain_nonzero_bits2): New predicates.
23838 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23840 2017-05-24 Nathan Sidwell <nathan@acm.org>
23842 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23845 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23848 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23850 * config/powerpcspe: New port. Files are copied from the rs6000
23851 port, with "rs6000" in filenames replaced by "powerpcspe".
23853 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23855 PR rtl-optimization/80754
23856 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23858 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23860 * config/sparc/sparc.md (length): Return the correct value for -mflat
23861 sibcalls to match output_sibcall.
23863 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23867 * config/rs6000/rs6000.c (struct machine_function): Add new field
23869 (rs6000_get_separate_components): Init that field, use it.
23870 (rs6000_components_for_bb): Use the field.
23872 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23874 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23876 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23878 PR middle-end/80823
23879 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23881 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23884 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23885 address against address_operand predicate.
23886 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23888 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23890 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23891 parameters passed indirectly.
23893 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23895 * config/i386/i386.md (*movdi_internal): Remove SSE4
23896 alternative 18 (?r, *v). Update insn attributes.
23897 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23898 Update insn attributes.
23899 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23900 Update insn attributes.
23901 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23902 alternative 1 (r, v). Remove isa attribute.
23903 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23904 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23905 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23907 2017-05-23 Tom de Vries <tom@codesourcery.com>
23909 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23912 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23914 * cgraphunit.c (symbol_table::process_new_functions): Update.
23915 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23916 (inline_generate_summary): Rename to ...
23917 (ipa_fn_summary_generate): ... this one.
23918 (inline_read_summary): Rename to ...
23919 (ipa_fn_summary_read): ... this one.
23920 (inline_write_summary): Rename to ...
23921 (ipa_fn_summary_write): ... this one.
23922 (inline_free_summary): Rename to ...
23923 (ipa_free_fn_summary): ... this one.
23924 (pass_data_local_fn_summary, pass_local_fn_summary,
23925 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23926 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23927 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23928 make_pass_ipa_fn_summary): New.
23929 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23930 inline_write_summary, inline_free_summary): Remove.
23931 (ipa_free_fn_summary) : New.
23932 * ipa-inline.c (ipa_inline): Update.
23933 (pass_ipa_inline): Do not generate summaries.
23934 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23936 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23937 and add pass_ipa_fn_summary.
23938 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23940 (make_pass_inline_parameters): Remove.
23942 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23944 * omp-low.c (struct omp_context): Remove "default_kind" member.
23947 * omp-offload.c (execute_oacc_device_lower): Remove the
23948 parallelism dimensions function attributes for unparallelized
23949 OpenACC kernels constructs.
23951 2017-05-23 Martin Liska <mliska@suse.cz>
23953 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23955 (cgraph_edge::make_speculative): Likewise.
23956 (cgraph_edge::resolve_speculation): Likewise.
23957 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23958 (cgraph_node::dump): Likewise.
23959 * cgraph.h: Likewise.
23960 * cgraphunit.c (analyze_functions): Likewise.
23961 (symbol_table::compile): Likewise.
23962 * ipa-cp.c (print_all_lattices): Likewise.
23963 (determine_versionability): Likewise.
23964 (initialize_node_lattices): Likewise.
23965 (ipcp_verify_propagated_values): Likewise.
23966 (estimate_local_effects): Likewise.
23967 (update_profiling_info): Likewise.
23968 (create_specialized_node): Likewise.
23969 (perhaps_add_new_callers): Likewise.
23970 (decide_about_value): Likewise.
23971 (decide_whether_version_node): Likewise.
23972 (identify_dead_nodes): Likewise.
23973 (ipcp_store_bits_results): Likewise.
23974 * ipa-devirt.c (dump_targets): Likewise.
23975 (ipa_devirt): Likewise.
23976 * ipa-icf.c (sem_item::dump): Likewise.
23977 (sem_function::equals): Likewise.
23978 (sem_variable::equals): Likewise.
23979 (sem_item_optimizer::read_section): Likewise.
23980 (sem_item_optimizer::execute): Likewise.
23981 (congruence_class::dump): Likewise.
23982 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23983 (dump_inline_summary): Likewise.
23984 (estimate_node_size_and_time): Likewise.
23985 (inline_analyze_function): Likewise.
23986 * ipa-inline-transform.c (inline_call): Likewise.
23987 * ipa-inline.c (report_inline_failed_reason): Likewise.
23988 (want_early_inline_function_p): Likewise.
23989 (edge_badness): Likewise.
23990 (update_edge_key): Likewise.
23991 (inline_small_functions): Likewise.
23992 * ipa-profile.c (ipa_profile): Likewise.
23993 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23994 (ipa_make_edge_direct_to_target): Likewise.
23995 (remove_described_reference): Likewise.
23996 (ipa_impossible_devirt_target): Likewise.
23997 (propagate_controlled_uses): Likewise.
23998 (ipa_print_node_params): Likewise.
23999 (ipcp_transform_function): Likewise.
24000 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24001 (propagate_pure_const): Likewise.
24002 * ipa-reference.c (generate_summary): Likewise.
24003 (read_write_all_from_decl): Likewise.
24004 (propagate): Likewise.
24005 (ipa_reference_read_optimization_summary): Likewise.
24006 * ipa-utils.c (ipa_merge_profiles): Likewise.
24007 * ipa.c (walk_polymorphic_call_targets): Likewise.
24008 (symbol_table::remove_unreachable_nodes): Likewise.
24009 (ipa_single_use): Likewise.
24010 * passes.c (execute_todo): Likewise.
24011 * predict.c (drop_profile): Likewise.
24012 * symtab.c (symtab_node::get_dump_name): New function.
24013 (symtab_node::dump_name): Likewise.
24014 (symtab_node::dump_asm_name): Likewise.
24015 (symtab_node::dump_references): Likewise.
24016 (symtab_node::dump_referring): Likewise.
24017 (symtab_node::dump_base): Likewise.
24018 (symtab_node::debug_symtab): Likewise.
24019 * tree-sra.c (convert_callers_for_node): Likewise.
24020 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24021 * value-prof.c (init_node_map): Likewise.
24023 2017-05-23 Martin Liska <mliska@suse.cz>
24025 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24026 and symtab_node::debug_symtab to symbol_table::debug.
24027 * cgraphunit.c (analyze_functions): Use the renamed function.
24028 (symbol_table::compile): Likewise.
24029 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24030 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24031 * passes.c (execute_todo): Likewise.
24032 * symtab.c (symbol_table::dump): New function.
24033 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24035 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24037 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24038 that nonconst implies exec.
24040 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24042 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24043 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24044 (inline_edge_summary_vec): Turn into ...
24045 (ipa_call_summaries): ... this one.
24046 (redirect_to_unreachable, edge_set_predicate,
24047 evaluate_properties_for_edge, inline_summary_alloc,
24048 reset_ipa_call_summary, reset_inline_summary,
24049 inline_summary_t::duplicate): Update.
24050 (inline_edge_duplication_hook): Turn to ...
24051 (ipa_call_summary_t::duplicate): ... this one.
24052 (inline_edge_removal_hook): Turn to ...
24053 (ipa_call_summary_t::remove): ... this one.
24054 (dump_inline_edge_summary): Turn to ...
24055 (dump_ipa_call_summary): ... this one.
24056 (estimate_function_body_sizes): Update.
24057 (inline_update_callee_summaries): Update.
24058 (remap_edge_change_prob): Update.
24059 (remap_edge_summaries): Update.
24060 (inline_merge_summary): Update.
24061 (do_estimate_edge_time): Update.
24062 (inline_generate_summary): Update.
24063 (inline_read_section): Update.
24064 (inline_read_summary): Update.
24065 (inline_free_summary): Update.
24066 * ipa-inline.c (can_inline_edge_p): Update.
24067 (compute_inlined_call_time): Update.
24068 (want_inline_small_function_p): Update.
24069 (edge_badness): Update.
24070 (early_inliner): Update.
24071 * ipa-inline.h (inline_edge_summary): Turn to ...
24072 (ipa_call_summary): ... this one.
24073 (ipa_call_summary_t): New class.
24074 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24075 (ipa_call_summaries): New.
24076 (inline_edge_summary): Remove.
24077 (estimate_edge_growth): Update.
24078 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24079 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24080 * ipa-split.c (execute_split_functions): Update.
24081 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24083 2017-05-23 Tom de Vries <tom@codesourcery.com>
24085 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24086 attributes): Document rdrand effective target.
24088 2017-05-23 Tom de Vries <tom@codesourcery.com>
24090 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24091 attributes): Sort alphabetically.
24093 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24095 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24097 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24100 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24101 V2DF/V2DI splat into two separate patterns, one that handles
24102 registers, and the other that only handles memory. Drop support
24103 for splatting from a GPR on ISA 2.07 and then splitting the
24104 splat into direct move and splat.
24105 (vsx_splat_<mode>_reg): Likewise.
24106 (vsx_splat_<mode>_mem): Likewise.
24108 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24110 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24112 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24114 PR middle-end/80809
24115 * omp-low.c (finish_taskreg_remap): New function.
24116 (finish_taskreg_scan): If unit size of ctx->record_type
24117 is non-constant, unshare the size expression and replace
24118 decls in it with possible outer var refs.
24120 PR middle-end/80809
24121 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24122 GOVD_SHARED rather than GOVD_PRIVATE with it.
24123 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24124 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24126 PR middle-end/80853
24127 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24128 as last argument to build_outer_var_ref for pointer bases of array
24129 section reductions.
24131 2017-05-19 Martin Sebor <msebor@redhat.com>
24133 * print-tree.c (print_node): Print DECL_READ_P flag.
24135 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24137 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24138 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24139 * cgraph.c: Likewise.
24140 * cgraphunit.c: Likewise.
24141 * gengtype.c: Likewise.
24142 * ipa-cp.c: Likewise.
24143 * ipa-devirt.c: Likewise.
24144 * ipa-icf.c: Likewise.
24145 * ipa-predicate.c: Likewise.
24146 * ipa-profile.c: Likewise.
24147 * ipa-prop.c: Likewise.
24148 * ipa-split.c: Likewise.
24150 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24151 edge_predicate_pool, dump_inline_hints,
24152 inline_summary::account_size_time, redirect_to_unreachable,
24153 edge_set_predicate, set_hint_predicate,
24154 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24155 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24156 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24157 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24158 ipa_call_summary_t::remove, initialize_growth_caches,
24159 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24160 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24161 mark_modified, unmodified_parm_1, unmodified_parm,
24162 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24163 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24164 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24165 will_be_nonconstant_predicate, record_modified_bb_info,
24166 get_minimal_bb, record_modified, param_change_prob,
24167 phi_result_unknown_predicate, predicate_for_phi_result,
24168 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24169 estimate_function_body_sizes, compute_inline_parameters,
24170 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24171 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24172 inline_update_callee_summaries, remap_edge_change_prob,
24173 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24174 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24175 inline_analyze_function, inline_summary_t::insert,
24176 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24177 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24178 inline_free_summary): Move to ipa-fnsummary.h
24179 (predicate_t): Remove.
24180 * ipa-fnsummary.c: New file.
24181 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24182 (enum inline_hints_vals, inline_hints, agg_position_info,
24183 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24184 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24185 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24186 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24187 inline_read_summary, inline_write_summary, inline_free_summary,
24188 inline_analyze_function, initialize_inline_failed,
24189 inline_merge_summary, inline_update_overall_summary,
24190 compute_inline_parameters): Move to ipa-fnsummary.h
24191 * ipa-fnsummary.h: New file.
24192 * ipa-inline-transform.h: Include ipa-inline.h.
24193 * ipa-inline.c: LIkewise.
24195 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24197 * ipa-inline.c (edge_badness): Use inlined_time instead of
24198 inline_summaries->get.
24200 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24202 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24204 2017-05-22 Nathan Sidwell <nathan@acm.org>
24206 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24207 (fdump-lang): Document 'raw' option.
24208 * dumpfile.h (TDI_tu): Delete.
24209 * dumpfile.c (dump_files): Remove translation-unit.
24210 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24212 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24214 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24215 command option from $(AWK) call.
24216 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24217 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24218 [FORMAT]: Remove handling of variable.
24219 * config/avr/t-multilib: Regenerate.
24221 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24223 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24225 (dump_inline_summary): Do not print self_time.
24226 (estimate_function_body_sizes): Do not set self_time.
24227 (compute_inline_parameters): Likewise.
24228 (inline_read_section, inline_write_summary): Do not stream self_time.
24229 * ipa-inline.h (inline_summary): Drop self_time.
24231 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24233 * ipa-inline-analysis.c (account_size_time): Rename to ...
24234 (inline_summary::account_size_time): ... this one.
24235 (reset_ipa_call_summary): Turn to ...
24236 (ipa_call_summary::reset): ... this one.
24237 (reset_inline_summary): Turn to ...
24238 (inline_summary::reset): ... this one.
24239 (inline_summary_t::remove): Update.
24240 (inline_summary_t::duplicate): Update.
24241 (ipa_call_summary_t::remove): Update.
24242 (dump_inline_summary): Update.
24243 (estimate_function_body_sizes): Update.
24244 (compute_inline_parameters): Update.
24245 (estimate_node_size_and_time): Update.
24246 (inline_merge_summary): Update.
24247 (inline_update_overall_summary): Update.
24248 (inline_read_section): Update.
24249 (inline_write_summary): Update.
24250 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24251 add account_size_time and reset member functions.
24252 (ipa_call_summary): Add reset function.
24253 * ipa-predicate.h (predicate::operator &): Constify.
24255 2017-05-22 Richard Biener <rguenther@suse.de>
24257 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24259 2017-05-19 Jason Merrill <jason@redhat.com>
24261 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24263 2017-05-19 Marek Polacek <polacek@redhat.com>
24266 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24267 TYPE_OVERFLOW_WRAPS checks.
24269 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24271 * tree-core.h (enum omp_clause_default_kind): Add
24272 "OMP_CLAUSE_DEFAULT_PRESENT".
24273 * tree-pretty-print.c (dump_omp_clause): Handle it.
24274 * gimplify.c (enum gimplify_omp_var_data): Add
24275 "GOVD_MAP_FORCE_PRESENT".
24276 (gimplify_adjust_omp_clauses_1): Map it to
24277 "GOMP_MAP_FORCE_PRESENT".
24278 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24280 * gimplify.c (oacc_default_clause): Clarify.
24282 2017-05-19 Nathan Sidwell <nathan@acm.org>
24284 LANG_HOOK_REGISTER_DUMPS
24285 * toplev.c (general_init): Call register dump lang hook.
24286 * doc/invoke.texi: Document -fdump-lang option family.
24287 * dumpfile.c (dump_files): Remove class dump here.
24288 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24289 * dumpfile.h (tree_dump_index): Remove TDI_class.
24290 * langhooks-def.h (lhd_register_dumps): Declare.
24291 (LANG_HOOKS_REGISTER_DUMPS): Define.
24292 (LANG_HOOKS_INITIALIZER): Add it.
24293 * langhooks.c (lhd_register_dumps): Define.
24294 * langhooks.h (struct lang_hooks): Add register_dumps.
24296 2017-05-19 Nathan Sidwell <nathan@acm.org>
24298 * context.h (context::set_passes): New.
24299 * context.c (context::context): Do not create pass manager.
24300 * toplev.c (general_init): Create pass manager here.
24302 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24304 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24305 use this splitter if two add or or instructions would also work for
24306 the constant we want to generate.
24308 2017-05-19 Richard Biener <rguenther@suse.de>
24311 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24312 predicate evaluation.
24314 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24316 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24318 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24321 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24323 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24324 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24325 (inline_edge_summary_vec): Turn into ...
24326 (ipa_call_summaries): ... this one.
24327 (redirect_to_unreachable, edge_set_predicate,
24328 evaluate_properties_for_edge, inline_summary_alloc,
24329 reset_ipa_call_summary, reset_inline_summary,
24330 inline_summary_t::duplicate): Update.
24331 (inline_edge_duplication_hook): Turn to ...
24332 (ipa_call_summary_t::duplicate): ... this one.
24333 (inline_edge_removal_hook): Turn to ...
24334 (ipa_call_summary_t::remove): ... this one.
24335 (dump_inline_edge_summary): Turn to ...
24336 (dump_ipa_call_summary): ... this one.
24337 (estimate_function_body_sizes): Update.
24338 (inline_update_callee_summaries): Update.
24339 (remap_edge_change_prob): Update.
24340 (remap_edge_summaries): Update.
24341 (inline_merge_summary): Update.
24342 (do_estimate_edge_time): Update.
24343 (inline_generate_summary): Update.
24344 (inline_read_section): Update.
24345 (inline_read_summary): Update.
24346 (inline_free_summary): Update.
24347 * ipa-inline.c (can_inline_edge_p): Update.
24348 (compute_inlined_call_time): Update.
24349 (want_inline_small_function_p): Update.
24350 (edge_badness): Update.
24351 (early_inliner): Update.
24352 * ipa-inline.h (inline_edge_summary): Turn to ...
24353 (ipa_call_summary): ... this one.
24354 (ipa_call_summary_t): New class.
24355 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24356 (ipa_call_summaries): New.
24357 (inline_edge_summary): Remove.
24358 (estimate_edge_growth): Update.
24359 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24360 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24361 * ipa-split.c (execute_split_functions): Update.
24362 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24364 2017-05-19 Richard Biener <rguenther@suse.de>
24366 PR middle-end/80764
24367 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24369 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24371 * config/rs6000/rs6000.c (struct machine_function): Add field
24372 fpr_is_wrapped_separately.
24373 (rs6000_get_separate_components): Use 64 components. Handle the
24374 new FPR components.
24375 (rs6000_components_for_bb): Handle the FPR components.
24376 (rs6000_emit_prologue_components): Handle the FPR components.
24377 (rs6000_emit_epilogue_components): Handle the FPR components.
24378 (rs6000_set_handled_components): Handle the FPR components.
24379 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24380 that are already separately shrink-wrapped.
24381 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24382 that are already separately shrink-wrapped.
24384 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24387 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24390 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24391 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24392 cases where the register allocator uses a move and an offsettable
24393 memory operation to/from a FPR register on ISA 2.06/2.07.
24394 (define_peephole2 for Altivec d-form store): Likewise.
24396 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24399 * config/i386/mmx.md (*mov<mode>_internal): Enable
24400 alternatives 11, 12, 13 and 14 also for 32bit targets.
24401 Remove alternatives 15, 16, 17 and 18.
24402 * config/i386/sse.md (vec_concatv2di): Change
24403 alternative (!x, *y) to (x, ?!*Yn).
24405 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24407 * dumpfile.h (enum dump_kind): Remove stray comma.
24409 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24411 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24412 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24413 predicate::num_conditions
24414 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24415 (CHANGED): turn into predicate::changed.
24416 (agg_position_info): Move to ipa-predicate.h
24417 (add_condition, predicate::add_clause, predicate::operator &=,
24418 predicate::or_with, predicate::evaluate, predicate::probability,
24419 dump_condition, dump_clause, predicate::dump,
24420 predicate::remap_after_duplication, predicate::remap_after_inlining,
24421 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24422 (evaluate_conditions_for_known_args): Update.
24423 (set_cond_stmt_execution_predicate): Update.
24424 * ipa-inline.h: Include ipa-predicate.h
24425 (condition, inline_param_summary, conditions, agg_position_info,
24426 predicate): Move to ipa-predicate.h
24427 * ipa-predicate.c: New file.
24428 * ipa-predicate.h: New file.
24430 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24432 * final.c (leaf_function_p): Check we are not in a sequence.
24434 2017-05-18 Martin Liska <mliska@suse.cz>
24436 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24437 * dumpfile.c (dump_register): Use new enum dump_kind.
24438 (get_dump_file_name): Likewise.
24439 (dump_enable_all): Likewise.
24440 (dump_switch_p_1): Likewise.
24441 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24442 * dumpfile.h (enum dump_kind): New enum type.
24443 (struct dump_file_info): Create constructor and
24444 format fields and comments.
24445 * passes.c (pass_manager::register_one_dump_file):
24447 * statistics.c (statistics_early_init): Likewise.
24448 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24449 TDF_TREE with TDF_SLIM.
24450 (gather_memory_references_ref): Likewise.
24452 2017-05-18 Martin Liska <mliska@suse.cz>
24454 * vec.h (struct vnull): Use it.
24456 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24458 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24459 (true_predicate, false_predicate, true_predicate_p,
24460 false_predicate_p): Remove.
24461 (single_cond_predicate, not_inlined_predicate): Turn to member function
24463 (add_condition): Update.
24464 (add_clause): Turn to...
24465 (predicate::add_clause): ... this one; update; allow passing NULL
24467 (and_predicates): Turn to ...
24468 (predicate::operator &=): ... this one.
24469 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24470 (or_predicates): Turn to ...
24471 (predicate::or_with): ... this one.
24472 (evaluate_predicate): Turn to ...
24473 (predicate::evaluate): ... this one.
24474 (predicate_probability): Turn to ...
24475 (predicate::probability): ... this one.
24476 (dump_condition): Update.
24477 (dump_predicate): Turn to ...
24478 (predicate::dump): ... this one.
24479 (account_size_time): Update.
24480 (edge_set_predicate): Update.
24481 (set_hint_predicate): UPdate.
24482 (evaluate_conditions_for_known_args): Update.
24483 (evaluate_properties_for_edge): Update.
24484 (remap_predicate_after_duplication): Turn to...
24485 (predicate::remap_after_duplication): ... this one.
24486 (remap_hint_predicate_after_duplication): Update.
24487 (inline_summary_t::duplicate): UPdate.
24488 (dump_inline_edge_summary): Update.
24489 (dump_inline_summary): Update.
24490 (set_cond_stmt_execution_predicate): Update.
24491 (set_switch_stmt_execution_predicate): Update.
24492 (compute_bb_predicates): Update.
24493 (will_be_nonconstant_expr_predicate): Update.
24494 (will_be_nonconstant_predicate): Update.
24495 (phi_result_unknown_predicate): Update.
24496 (predicate_for_phi_result): Update.
24497 (array_index_predicate): Update.
24498 (estimate_function_body_sizes): Update.
24499 (estimate_node_size_and_time): Update.
24500 (estimate_ipcp_clone_size_and_time): Update.
24501 (remap_predicate): Rename to ...
24502 (predicate::remap_after_inlining): ... this one.
24503 (remap_hint_predicate): Update.
24504 (inline_merge_summary): Update.
24505 (inline_update_overall_summary): Update.
24506 (estimate_size_after_inlining): Update.
24507 (read_predicate): Rename to ...
24508 (predicate::stream_in): ... this one.
24509 (read_inline_edge_summary): Update.
24510 (write_predicate): Rename to ...
24511 (predicate::stream_out): ... this one.
24512 (write_inline_edge_summary): Update.
24513 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24514 (clause_t): Turn to uint32_t
24515 (predicate): Turn to class; implement constructor and operators
24517 (size_time_entry): Update.
24518 (inline_summary): Update.
24519 (inline_edge_summary): Update.
24521 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24523 * fold-const.c (fold_binary_loc): Move transformation...
24524 * match.pd (C - X CMP X): ... here.
24526 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24528 * config/sparc/sparc.c (sparc_option_override): Set function
24529 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24530 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24532 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24534 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24536 2017-05-18 Marek Polacek <polacek@redhat.com>
24539 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24540 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24541 gimple_assign_load_p.
24543 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24545 PR middle-end/80692
24546 * real.c (do_compare): Give decimal_do_compare preference over
24547 comparing just the signs.
24549 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24551 * doc/md.texi (Canonicalization of Instructions): Describe the
24552 canonical form of instructions that inherently set a condition
24555 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24557 PR middle-end/80775
24558 * tree-cfg.c: Move deletion of unreachable case statements to after
24559 the merging of consecutive case labels.
24561 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24563 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24564 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24565 restoring of callee-saved registers.
24567 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24569 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24570 * config/visium/visium.c (single_set_and_flags): Likewise.
24571 * config/visium/visium.md (Substitutions): Likewise.
24573 2017-05-17 Martin Liska <mliska@suse.cz>
24575 * cfg.c: Introduce dump_flags_t type and
24576 use it instead of int type.
24578 * cfghooks.c: Likewise.
24579 * cfghooks.h (struct cfg_hooks): Likewise.
24580 * cfgrtl.c: Likewise.
24581 * cfgrtl.h: Likewise.
24582 * cgraph.c (cgraph_node::get_body): Likewise.
24583 * coretypes.h: Likewise.
24584 * domwalk.c: Likewise.
24585 * domwalk.h: Likewise.
24586 * dumpfile.c (struct dump_option_value_info): Likewise.
24587 (dump_enable_all): Likewise.
24588 (dump_switch_p_1): Likewise.
24589 (opt_info_switch_p): Likewise.
24590 * dumpfile.h (enum tree_dump_index): Likewise.
24591 (struct dump_file_info): Likewise.
24592 * genemit.c: Likewise.
24593 * generic-match-head.c: Likewise.
24594 * gengtype.c (open_base_files): Likewise.
24595 * gimple-pretty-print.c: Likewise.
24596 * gimple-pretty-print.h: Likewise.
24597 * graph.c (print_graph_cfg): Likewise.
24598 * graphite-scop-detection.c (dot_all_sese): Likewise.
24599 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24600 * loop-unroll.c (report_unroll): Likewise.
24601 * passes.c (pass_manager::register_one_dump_file): Likewise.
24602 * print-tree.c: Likewise.
24603 * statistics.c: Likewise.
24604 * tree-cfg.c: Likewise.
24605 * tree-cfg.h: Likewise.
24606 * tree-dfa.c: Likewise.
24607 * tree-dfa.h: Likewise.
24608 * tree-dump.c (dump_function): Likewise.
24609 * tree-dump.h (struct dump_info): Likewise.
24610 * tree-pretty-print.c: Likewise.
24611 * tree-pretty-print.h: Likewise.
24612 * tree-ssa-live.c: Likewise.
24613 * tree-ssa-live.h: Likewise.
24614 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24615 * tree-vect-loop.c: Likewise.
24616 * tree-vect-slp.c: Likewise.
24618 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24619 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24621 PR tree-optimization/80457
24622 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24623 of all arguments to a statement as scalar_to_vec operations.
24624 (vectorizable_call): Adjust call to vect_model_simple_cost for
24626 (vectorizable_conversion): Likewise.
24627 (vectorizable_assignment): Likewise.
24628 (vectorizable_shift): Likewise.
24629 (vectorizable_operation): Likewise.
24630 (vectorizable_comparison): Likewise.
24631 (vect_is_simple_cond): Record the def types for operands.
24632 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24633 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24634 for statement argument count.
24636 2017-05-16 Carl Love <cel@us.ibm.com>
24638 * config/rs6000/rs6000-c: Add support for built-in functions
24639 vector unsigned long long vec_bperm (vector unsigned long long,
24640 vector unsigned char)
24641 vector signed long long vec_mule (vector signed int,
24643 vector unsigned long long vec_mule (vector unsigned int,
24644 vector unsigned int)
24645 vector signed long long vec_mulo (vector signed int,
24647 vector unsigned long long vec_mulo (vector unsigned int,
24648 vector unsigned int)
24649 vector signed char vec_sldw (vector signed char,
24650 vector signed char,
24652 vector unsigned char vec_sldw (vector unsigned char,
24653 vector unsigned char,
24655 vector signed short vec_sldw (vector signed short,
24656 vector signed short,
24658 vector unsigned short vec_sldw (vector unsigned short,
24659 vector unsigned short,
24661 vector signed int vec_sldw (vector signed int,
24664 vector unsigned int vec_sldw (vector unsigned int,
24665 vector unsigned int,
24667 vector signed long long vec_sldw (vector signed long long,
24668 vector signed long long,
24670 vector unsigned long long vec_sldw (vector unsigned long long,
24671 vector unsigned long long,
24673 * config/rs6000/rs6000-c: Add support for built-in functions
24674 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24675 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24676 * doc/extend.texi: Update the built-in documentation for the
24677 new built-in functions.
24679 2017-05-16 Marek Polacek <polacek@redhat.com>
24683 * tree.c (save_expr): Don't fold the expression.
24685 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24687 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24688 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24689 and (?*y,m). Update insn attributes.
24691 2017-05-16 Martin Liska <mliska@suse.cz>
24693 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24694 flags argument of print_gimple_stmt, print_gimple_expr,
24695 print_generic_stmt and print_generic_expr.
24696 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24697 * coretypes.h: Likewise.
24698 * except.c (dump_eh_tree): Likewise.
24699 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24700 * gimple-pretty-print.h: Likewise.
24701 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24702 (backprop::push_to_worklist): Likewise.
24703 (backprop::pop_from_worklist): Likewise.
24704 (backprop::process_use): Likewise.
24705 (backprop::intersect_uses): Likewise.
24706 (note_replacement): Likewise.
24707 * gimple-ssa-store-merging.c
24708 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24709 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24710 (pass_store_merging::execute): Likewise.
24711 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24712 (ssa_base_cand_dump_callback): Likewise.
24713 (dump_incr_vec): Likewise.
24714 (replace_refs): Likewise.
24715 (replace_mult_candidate): Likewise.
24716 (create_add_on_incoming_edge): Likewise.
24717 (create_phi_basis): Likewise.
24718 (insert_initializers): Likewise.
24719 (all_phi_incrs_profitable): Likewise.
24720 (introduce_cast_before_cand): Likewise.
24721 (replace_one_candidate): Likewise.
24722 * gimplify.c (gimplify_expr): Likewise.
24723 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24724 (set_rename): Likewise.
24725 (rename_uses): Likewise.
24726 (copy_loop_phi_nodes): Likewise.
24727 (add_close_phis_to_merge_points): Likewise.
24728 (copy_loop_close_phi_args): Likewise.
24729 (copy_cond_phi_args): Likewise.
24730 (graphite_copy_stmts_from_block): Likewise.
24731 (translate_pending_phi_nodes): Likewise.
24732 * graphite-poly.c (print_pdr): Likewise.
24733 (dump_gbb_cases): Likewise.
24734 (dump_gbb_conditions): Likewise.
24735 (print_scop_params): Likewise.
24736 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24737 (build_cross_bb_scalars_use): Likewise.
24738 (gather_bbs::before_dom_children): Likewise.
24739 * hsa-dump.c (dump_hsa_immed): Likewise.
24740 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24741 (get_replacement_map): Likewise.
24742 * ipa-inline-analysis.c (dump_condition): Likewise.
24743 (estimate_function_body_sizes): Likewise.
24744 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24745 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24746 * ipa-prop.c (ipa_dump_param): Likewise.
24747 (ipa_print_node_jump_functions_for_edge): Likewise.
24748 (ipa_modify_call_arguments): Likewise.
24749 (ipa_modify_expr): Likewise.
24750 (ipa_dump_param_adjustments): Likewise.
24751 (ipa_dump_agg_replacement_values): Likewise.
24752 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24753 * ipa-pure-const.c (check_stmt): Likewise.
24754 (pass_nothrow::execute): Likewise.
24755 * ipa-split.c (execute_split_functions): Likewise.
24756 * omp-offload.c (dump_oacc_loop_part): Likewise.
24757 (dump_oacc_loop): Likewise.
24758 * trans-mem.c (tm_log_emit): Likewise.
24759 (tm_memopt_accumulate_memops): Likewise.
24760 (dump_tm_memopt_set): Likewise.
24761 (dump_tm_memopt_transform): Likewise.
24762 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24763 (print_loop): Likewise.
24764 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24765 (chkp_gather_checks_info): Likewise.
24766 (chkp_get_check_result): Likewise.
24767 (chkp_remove_check_if_pass): Likewise.
24768 (chkp_use_outer_bounds_if_possible): Likewise.
24769 (chkp_reduce_bounds_lifetime): Likewise.
24770 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24771 (chkp_mark_completed_bounds): Likewise.
24772 (chkp_register_incomplete_bounds): Likewise.
24773 (chkp_mark_invalid_bounds): Likewise.
24774 (chkp_maybe_copy_and_register_bounds): Likewise.
24775 (chkp_build_returned_bound): Likewise.
24776 (chkp_get_bound_for_parm): Likewise.
24777 (chkp_build_bndldx): Likewise.
24778 (chkp_get_bounds_by_definition): Likewise.
24779 (chkp_generate_extern_var_bounds): Likewise.
24780 (chkp_get_bounds_for_decl_addr): Likewise.
24781 * tree-chrec.c (chrec_apply): Likewise.
24782 * tree-data-ref.c (dump_data_reference): Likewise.
24783 (dump_subscript): Likewise.
24784 (dump_data_dependence_relation): Likewise.
24785 (analyze_overlapping_iterations): Likewise.
24786 * tree-inline.c (expand_call_inline): Likewise.
24787 (tree_function_versioning): Likewise.
24788 * tree-into-ssa.c (dump_defs_stack): Likewise.
24789 (dump_currdefs): Likewise.
24790 (dump_names_replaced_by): Likewise.
24791 (dump_update_ssa): Likewise.
24792 (update_ssa): Likewise.
24793 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24794 * tree-parloops.c (build_new_reduction): Likewise.
24795 (try_create_reduction_list): Likewise.
24796 (ref_conflicts_with_region): Likewise.
24797 (oacc_entry_exit_ok_1): Likewise.
24798 (oacc_entry_exit_single_gang): Likewise.
24799 * tree-pretty-print.h: Likewise.
24800 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24801 (get_scalar_evolution): Likewise.
24802 (add_to_evolution): Likewise.
24803 (get_loop_exit_condition): Likewise.
24804 (analyze_evolution_in_loop): Likewise.
24805 (analyze_initial_condition): Likewise.
24806 (analyze_scalar_evolution): Likewise.
24807 (instantiate_scev): Likewise.
24808 (number_of_latch_executions): Likewise.
24809 (gather_chrec_stats): Likewise.
24810 (final_value_replacement_loop): Likewise.
24811 (scev_const_prop): Likewise.
24812 * tree-sra.c (dump_access): Likewise.
24813 (disqualify_candidate): Likewise.
24814 (create_access): Likewise.
24815 (reject): Likewise.
24816 (maybe_add_sra_candidate): Likewise.
24817 (create_access_replacement): Likewise.
24818 (analyze_access_subtree): Likewise.
24819 (analyze_all_variable_accesses): Likewise.
24820 (sra_modify_assign): Likewise.
24821 (initialize_constant_pool_replacements): Likewise.
24822 (find_param_candidates): Likewise.
24823 (decide_one_param_reduction): Likewise.
24824 (replace_removed_params_ssa_names): Likewise.
24825 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24826 * tree-ssa-copy.c (dump_copy_of): Likewise.
24827 (copy_prop_visit_cond_stmt): Likewise.
24828 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24829 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24830 (record_equivalences_from_stmt): Likewise.
24831 * tree-ssa-dse.c (compute_trims): Likewise.
24832 (delete_dead_call): Likewise.
24833 (delete_dead_assignment): Likewise.
24834 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24835 (forward_propagate_into_cond): Likewise.
24836 (pass_forwprop::execute): Likewise.
24837 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24838 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24840 (move_computations_worker): Likewise.
24841 (execute_sm): Likewise.
24842 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24843 (remove_exits_and_undefined_stmts): Likewise.
24844 (remove_redundant_iv_tests): Likewise.
24845 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24846 (adjust_iv_update_pos): Likewise.
24847 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24848 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24849 (value_replacement): Likewise.
24850 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24851 * tree-ssa-pre.c (print_pre_expr): Likewise.
24852 (get_representative_for): Likewise.
24853 (create_expression_by_pieces): Likewise.
24854 (insert_into_preds_of_block): Likewise.
24855 (eliminate_insert): Likewise.
24856 (eliminate_dom_walker::before_dom_children): Likewise.
24857 (eliminate): Likewise.
24858 (remove_dead_inserted_code): Likewise.
24859 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24860 * tree-ssa-reassoc.c (get_rank): Likewise.
24861 (eliminate_duplicate_pair): Likewise.
24862 (eliminate_plus_minus_pair): Likewise.
24863 (eliminate_not_pairs): Likewise.
24864 (undistribute_ops_list): Likewise.
24865 (eliminate_redundant_comparison): Likewise.
24866 (update_range_test): Likewise.
24867 (optimize_range_tests_var_bound): Likewise.
24868 (optimize_vec_cond_expr): Likewise.
24869 (rewrite_expr_tree): Likewise.
24870 (rewrite_expr_tree_parallel): Likewise.
24871 (linearize_expr): Likewise.
24872 (break_up_subtract): Likewise.
24873 (linearize_expr_tree): Likewise.
24874 (attempt_builtin_powi): Likewise.
24875 (attempt_builtin_copysign): Likewise.
24876 (transform_stmt_to_copy): Likewise.
24877 (transform_stmt_to_multiply): Likewise.
24878 (dump_ops_vector): Likewise.
24879 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24880 (print_scc): Likewise.
24881 (set_ssa_val_to): Likewise.
24882 (visit_reference_op_store): Likewise.
24883 (visit_use): Likewise.
24884 (sccvn_dom_walker::before_dom_children): Likewise.
24885 (run_scc_vn): Likewise.
24886 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24888 (expr_hash_elt::print): Likewise.
24889 (const_and_copies::pop_to_marker): Likewise.
24890 (const_and_copies::record_const_or_copy_raw): Likewise.
24891 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24892 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24893 (dump_predicates): Likewise.
24894 (find_uninit_use): Likewise.
24895 (warn_uninitialized_phi): Likewise.
24896 (pass_late_warn_uninitialized::execute): Likewise.
24897 * tree-ssa.c (verify_vssa): Likewise.
24898 (verify_ssa): Likewise.
24899 (maybe_optimize_var): Likewise.
24900 * tree-vrp.c (dump_value_range): Likewise.
24901 (dump_all_value_ranges): Likewise.
24902 (dump_asserts_for): Likewise.
24903 (register_edge_assert_for_2): Likewise.
24904 (vrp_visit_cond_stmt): Likewise.
24905 (vrp_visit_switch_stmt): Likewise.
24906 (vrp_visit_stmt): Likewise.
24907 (vrp_visit_phi_node): Likewise.
24908 (simplify_cond_using_ranges_1): Likewise.
24909 (fold_predicate_in): Likewise.
24910 (evrp_dom_walker::before_dom_children): Likewise.
24911 (evrp_dom_walker::push_value_range): Likewise.
24912 (evrp_dom_walker::pop_value_range): Likewise.
24913 (execute_early_vrp): Likewise.
24915 2017-05-16 Richard Biener <rguenther@suse.de>
24917 * dwarf2out.c (loc_list_from_tree_1): Do not create
24918 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24920 2017-05-16 Richard Biener <rguenther@suse.de>
24922 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24924 (note_variable_value_in_expr): If we resolved the decl ref
24925 do not push to the stack.
24927 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24929 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24930 operations in fast-math mode.
24931 (vaddq_f16): Likewise.
24932 (vmul_f16): Likewise.
24933 (vmulq_f16): Likewise.
24934 (vsub_f16): Likewise.
24935 (vsubq_f16): Likewise.
24936 * config/arm/neon.md (add<mode>3): New.
24938 (fma:<VH:mode>3): New. Also remove outdated comment.
24941 2017-05-16 Martin Liska <mliska@suse.cz>
24945 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24946 (odr_types_equivalent_p): Likewise.
24948 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24950 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24952 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24955 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24956 non-interunit SSE move alternatives with '?'.
24957 (zero-extendsidi peephole2): New peephole to skip intermediate
24958 general register in SSE zero-extend sequence.
24960 2017-05-15 Jeff Law <law@redhat.com>
24962 * reorg.c (relax_delay_slots): Create a new variable to hold
24963 the temporary target rather than clobbering TARGET_LABEL.
24965 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24966 missing argument to extract_bit_field call.
24967 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24969 2017-05-15 Martin Liska <mliska@suse.cz>
24972 * gcc.c (process_command): Do not allow empty argument of -o option.
24974 2017-05-15 Renlin Li <renlin.li@arm.com>
24976 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24977 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24978 * config/aarch64/constraints.md (Usf): Add long call check.
24979 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24980 (call_value): Likewise.
24981 (sibcall): Likewise.
24982 (sibcall_value): Likewise.
24984 (call_value_insn): New.
24985 (sibcall_insn): Update rtx pattern.
24986 (sibcall_value_insn): Likewise.
24987 (call_internal): Remove.
24988 (call_value_internal): Likewise.
24989 (sibcall_internal): Likewise.
24990 (sibcall_value_internal): Likewise.
24991 (call_reg): Likewise.
24992 (call_symbol): Likewise.
24993 (call_value_reg): Likewise.
24994 (call_value_symbol): Likewise.
24996 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24999 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25001 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25003 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25004 compatible with CCGOCmode and with CCZmode.
25006 2017-05-14 Martin Sebor <msebor@redhat.com>
25008 PR middle-end/77671
25009 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25010 (gimple_fold_builtin_snprintf): Same.
25011 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25012 (gimple_fold_builtin_snprintf): Same.
25013 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25014 of character types.
25015 (is_call_safe): New function.
25016 (try_substitute_return_value): Call it.
25017 (try_simplify_call): New function.
25018 (pass_sprintf_length::handle_gimple_call): Call it.
25020 2017-05-14 Martin Sebor <msebor@redhat.com>
25022 PR middle-end/80669
25023 * builtins.c (expand_builtin_stpncpy): Simplify.
25025 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25027 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25028 * config/i386/i386.h
25029 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25030 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25031 (struct machine_function): Add new members call_ms2sysv,
25032 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25033 (struct machine_frame_state): New fields sp_realigned and
25034 sp_realigned_offset.
25035 * config/i386/i386.c
25036 (enum xlogue_stub): New enum.
25037 (enum xlogue_stub_sets): New enum.
25038 (class xlogue_layout): New class.
25039 (struct ix86_frame): New fields stack_realign_allocate_offset,
25040 stack_realign_offset and outlined_save_offset. Modify comments to
25041 detail stack layout when using out-of-line stubs.
25042 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25043 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25044 -mcall-ms2sysv-xlogues.
25045 (stub_managed_regs): New static variable.
25046 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25047 registers managed by out-of-line stub.
25048 (disable_call_ms2sysv_xlogues): New function.
25049 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25050 m->call_ms2sysv when appropriate and compute frame layout for
25052 (sp_valid_at, fp_valid_at): New inline functions.
25053 (choose_basereg): New function.
25054 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25056 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25057 Use align parameter of choose_baseaddr to generated aligned SSE movs
25059 (pro_epilogue_adjust_stack): Modify to track
25060 machine_frame_state::sp_realigned.
25061 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25062 (ix86_nsaved_sseregs): Likewise.
25063 (ix86_emit_save_regs): Likewise.
25064 (ix86_emit_save_regs_using_mov): Likewise.
25065 (ix86_emit_save_sse_regs_using_mov): Likewise.
25066 (get_scratch_register_on_entry): Likewise.
25067 (gen_frame_set): New function.
25068 (gen_frame_load): Likewise.
25069 (gen_frame_store): Likewise.
25070 (emit_outlined_ms2sysv_save): Likewise.
25071 (emit_outlined_ms2sysv_restore): Likewise.
25072 (ix86_expand_prologue): Modify stack re-alignment code and call
25073 emit_outlined_ms2sysv_save when appropriate.
25074 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25075 parameter rtx_insn *insn, which allows the function to be used to only
25076 generate the notes.
25077 (ix86_expand_epilogue): Modify validity checks of frame and stack
25078 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25079 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25080 * config/i386/predicates.md
25081 (save_multiple): New predicate.
25082 (restore_multiple): Likewise.
25083 * config/i386/sse.md
25084 (save_multiple<mode>): New pattern.
25085 (save_multiple_realign<mode>): Likewise.
25086 (restore_multiple<mode>): Likewise.
25087 (restore_multiple_and_return<mode>): Likewise.
25088 (restore_multiple_leave_return<mode>): Likewise.
25089 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25091 2017-05-14 Julia Koval <julia.koval@intel.com>
25093 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25094 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25095 (__builtin_ia32_xsetbv): New builtins.
25096 * config/i386/i386.c (ix86_expand_special_args_builtin):
25098 (ix86_expand_builtin): Special expand for new intrinsics.
25099 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25100 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25101 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25103 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25105 * cfganal.c (inverted_post_order_compute): Change argument type
25107 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25108 * df-core.c (rest_of_handle_df_initialize): Adjust.
25109 (rest_of_handle_df_finish): Likewise.
25110 (df_analyze_1): Likewise.
25111 (df_analyze): Likewise.
25112 (loop_inverted_post_order_compute): Change argument to be a vec *.
25113 (df_analyze_loop): Adjust.
25114 (df_get_n_blocks): Likewise.
25115 (df_get_postorder): Likewise.
25116 * df.h (struct df_d): Change field to be a vec.
25117 * lcm.c (compute_laterin): Adjust.
25118 (compute_available): Likewise.
25119 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25120 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25121 * tree-ssa-pre.c (compute_antic): Likewise.
25123 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25125 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25126 (depth_first_search::depth_first_search): Change structure init
25127 function to this constructor.
25128 (depth_first_search::add_bb): Rename function to this member.
25129 (depth_first_search::execute): Likewise.
25130 (flow_dfs_compute_reverse_finish): Adjust.
25132 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25134 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25135 (longest_simple_path): Likewise.
25136 * shrink-wrap.c (spread_components): Likewise.
25137 (disqualify_problematic_components): Likewise.
25138 (emit_common_heads_for_components): Likewise.
25139 (emit_common_tails_for_components): Likewise.
25140 (insert_prologue_epilogue_for_components): Likewise.
25142 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25144 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25147 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25149 * df-core.c (df_set_blocks): Start using auto_bitmap.
25150 (df_compact_blocks): Likewise.
25151 * df-problems.c (df_rd_confluence_n): Likewise.
25152 * df-scan.c (df_insn_rescan_all): Likewise.
25153 (df_process_deferred_rescans): Likewise.
25154 (df_update_entry_block_defs): Likewise.
25155 (df_update_exit_block_uses): Likewise.
25156 (df_entry_block_bitmap_verify): Likewise.
25157 (df_exit_block_bitmap_verify): Likewise.
25158 (df_scan_verify): Likewise.
25159 * lra-constraints.c (lra_constraints): Likewise.
25160 (undo_optional_reloads): Likewise.
25161 (lra_undo_inheritance): Likewise.
25162 * lra-remat.c (calculate_gen_cands): Likewise.
25163 (do_remat): Likewise.
25164 * lra-spills.c (assign_spill_hard_regs): Likewise.
25165 (spill_pseudos): Likewise.
25166 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25167 (bitmap_set_subtract_values): Likewise.
25169 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25171 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25172 management with auto_bitmap.
25173 (fix_inter_tick): Likewise.
25174 (fix_recovery_deps): Likewise.
25175 * ira.c (add_store_equivs): Likewise.
25176 (find_moveable_pseudos): Likewise.
25177 (split_live_ranges_for_shrink_wrap): Likewise.
25178 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25179 (rtx_reuse_manager::seen_def_p): Likewise.
25180 (rtx_reuse_manager::set_seen_def): Likewise.
25181 * print-rtl.h (class rtx_reuse_manager): Likewise.
25183 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25185 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25187 (migrate_btr_def): Likewise.
25188 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25189 * df-core.c (loop_post_order_compute): Likewise.
25190 (loop_inverted_post_order_compute): Likewise.
25191 * hsa-common.h: Likewise.
25192 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25193 * init-regs.c (initialize_uninitialized_regs): Likewise.
25194 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25195 (inline_small_functions): Likewise.
25196 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25197 * ira.c (combine_and_move_insns): Likewise.
25198 (build_insn_chain): Likewise.
25199 * loop-invariant.c (find_invariants): Likewise.
25200 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25201 * predict.c (tree_predict_by_opcode): Likewise.
25202 (predict_paths_leading_to): Likewise.
25203 (predict_paths_leading_to_edge): Likewise.
25204 (estimate_loops_at_level): Likewise.
25205 (estimate_loops): Likewise.
25206 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25207 (spread_components): Likewise.
25208 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25209 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25210 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25211 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25212 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25213 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25214 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25215 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25216 (mark_threaded_blocks): Likewise.
25217 (thread_through_all_blocks): Likewise.
25218 * tree-ssa.c (verify_ssa): Likewise.
25219 (execute_update_addresses_taken): Likewise.
25220 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25222 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25224 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25226 (post_order_compute): Likewise.
25227 (inverted_post_order_compute): Likewise.
25228 (pre_and_rev_post_order_compute_fn): Likewise.
25230 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25232 * genrecog.c (int_set::int_set): Explicitly construct our
25233 auto_vec base class.
25234 * vec.h (auto_vec::auto_vec): New constructor.
25236 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25238 * bitmap.h (class auto_bitmap): New constructor taking
25239 bitmap_obstack * argument.
25241 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25243 * bitmap.h (class auto_bitmap): Change type of m_bits to
25244 bitmap_head, and adjust ctor / dtor and member operators.
25246 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25248 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25249 when returned register mode doesn't match original mode.
25251 2017-05-12 Jeff Law <law@redhat.com>
25252 Jakub Jelinek <jakub@redhat.com>
25254 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25255 we look for cc setter after the compare-elim changes.
25256 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25257 within the vector to match what compare-elim now expects.
25258 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25259 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25261 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25262 after the compare-elim changes.
25263 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25264 the vector to match what compare-elim now expects.
25265 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25266 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25267 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25268 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25269 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25271 * config/visium/visium.c (single_set_and_flags): Fix where
25272 we look for cc setter after the compare-elim changes.
25273 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25274 with the vector to match what compare-elim now expects.
25275 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25276 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25277 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25278 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25279 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25280 (neg<mode>2_insn_set_overflow): Likewise.
25282 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25284 PR middle-end/79794
25285 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25286 maybe_expand_insn call, set ops[0].target. If still set after call,
25287 set alt_rtl. Add extra arg to recursive calls.
25288 (extract_bit_field): Add alt_rtl argument. Pass to
25290 * expmed.h (extract_bit_field): Fix prototype.
25291 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25292 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25293 to extract_bit_field_calls.
25294 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25295 Pass alt_rtl to extract_bit_field calls.
25296 * calls.c (store_unaligned_arguments_into_psuedos)
25297 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25298 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25300 * optabs.h (struct expand_operand): Add target bitfield.
25302 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25304 * compare-elim.c (try_eliminate_compare): Canonicalize
25305 operation with embedded compare to
25306 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25307 (set (reg) (operation)].
25309 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25311 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25314 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25315 cost of adding a carry flag for ADC instruction.
25316 [case MINUS]: Ignore the cost of subtracting a carry flag
25317 for SBB instruction.
25319 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25321 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25323 * config/rs6000/bmiintrin.h: New file.
25324 * config/rs6000/bmi2intrin.h: New file.
25325 * config/rs6000/x86intrin.h: New file.
25327 2017-05-12 Jeff Law <law@redhat.com>
25329 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25332 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25334 PR middle-end/80707
25335 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25337 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25339 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25340 early expansion of vector divide builtins.
25341 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25342 builtins identified as having unsigned arguments.
25344 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25346 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25347 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25348 expansion of vector logical operations (and, andc, or, xor,
25351 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25353 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25354 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25356 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25358 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25359 early GIMPLE expansion of vector multiplies.
25361 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25363 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25364 TARGET_HAVE_MOVT conditional.
25365 (movt splitter): Likewise.
25367 2017-05-12 Richard Biener <rguenther@suse.de>
25369 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25370 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25371 Fold all stmts not inplace.
25373 2017-05-12 Richard Biener <rguenther@suse.de>
25375 PR tree-optimization/80713
25376 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25377 inserted_exprs bit for not removed stmts.
25379 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25381 PR middle-end/69921
25382 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25383 parallelized" attribute for parallelized OpenACC kernels.
25384 * omp-offload.c (execute_oacc_device_lower): Use it.
25386 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25387 Set "oacc kernels" attribute.
25388 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25389 parameter. Adjust all users.
25390 (oacc_fn_attrib_kernels_p): Remove function.
25391 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25392 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25393 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25394 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25395 assert "oacc kernels" attribute is set.
25397 2017-05-11 Carl Love <cel@us.ibm.com>
25399 * config/rs6000/rs6000-c: Add support for built-in functions
25400 vector unsigned char vec_popcnt (vector signed char)
25401 vector unsigned char vec_popcnt (vector unsigned char)
25402 vector unsigned short vec_popcnt (vector signed short)
25403 vector unsigned short vec_popcnt (vector unsigned short)
25404 vector unsigned int vec_popcnt (vector signed int)
25405 vector unsigned int vec_popcnt (vector unsigned int)
25406 vector unsigned long long vec_popcnt (vector signed long long)
25407 vector unsigned long long vec_popcnt (vector unsigned long long)
25408 vector signed long long vec_slo (vector signed long long,
25409 vector signed char)
25410 vector signed long long vec_slo (vector signed long long,
25411 vector unsigned char)
25412 vector unsigned long long vec_slo (vector unsigned long long,
25413 vector signed char)
25414 vector unsigned long long vec_slo (vector unsigned long long,
25415 vector unsigned char)
25416 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25417 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25418 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25419 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25420 * doc/extend.texi: Update the built-in documentation file for the
25421 new built-in functions.
25423 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25425 * attribs.h (sorted_attr_string): Move machine independent
25426 functions for target clone support from the i386 port to common
25427 code. Rename ix86_function_versions to common_function_versions.
25428 Rename make_name to make_unique_name.
25429 (common_function_versions): Likewise.
25430 (make_unique_name): Likewise.
25431 (make_dispatcher_decl): Likewise.
25432 (is_function_default_version): Likewise.
25433 * attribs.c (attr_strcmp): Likewise.
25434 (sorted_attr_string): Likewise.
25435 (common_function_versions): Likewise.
25436 (make_unique_name): Likewise.
25437 (make_dispatcher_decl): Likewise.
25438 (is_function_default_version): Likewise.
25439 * config/i386/i386.c (attr_strcmp): Likewise.
25440 (sorted_attr_string): Likewise.
25441 (ix86_function_versions): Likewise.
25442 (make_name): Likewise.
25443 (make_dispatcher_decl): Likewise.
25444 (is_function_default_version): Likewise.
25445 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25447 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25450 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25451 Account for direct move costs for vec_construct of integer
25454 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25457 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25458 (UNSPEC_STX_ATOMIC): Ditto.
25459 (loaddi_via_sse): New insn.
25460 (storedi_via_sse): Ditto.
25461 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25462 Update corresponding peephole2 patterns.
25463 (atomic_storedi_fpu): Ditto.
25465 2017-05-11 Julia Koval <julia.koval@intel.com>
25467 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25468 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25470 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25471 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25472 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25474 2017-05-11 Nathan Sidwell <nathan@acm.org>
25476 * graphite-poly.c: Include dumpfile.h.
25478 * dumpfle.h (dump_function): Declare here ...
25479 * tree-dump.h (dump_function): ... not here.
25480 * dumpfile.c: #include tree-cfg.h.
25481 (dump_function): Move here from ...
25482 * tree-dump.c (dump_function): ... here.
25483 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25484 * graphite-poly.c: Don't include tree-dump.h.
25485 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25486 * print-tree.c: Likewise.
25487 * stor-layout.c: Likewise.
25488 * tree-nested.c: Likewise.
25490 * dumpfile.c (dump_start): Use TDF_FLAGS.
25491 (dump_enable_all): Fix TDF_KIND check thinko.
25493 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25495 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25496 array entries to represent two legal parameterizations of the
25497 overloaded __builtin_cmpb function, as represented by the
25498 P6_OV_BUILTIN_CMPB constant.
25499 (altivec_resolve_overloaded_builtin): Add special case handling
25500 for the __builtin_cmpb function, as represented by the
25501 P6_OV_BUILTIN_CMPB constant.
25502 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25503 (BU_P6_64BIT_2): New macro.
25504 (BU_P6_OVERLOAD_2): New macro
25505 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25506 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25507 (CMPB): Add overload support to represent both 32-bit and 64-bit
25508 compare-bytes function.
25509 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25510 support for TARGET_CMPB.
25511 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25512 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25513 documentation of the __builtin_cmpb overloaded built-in function.
25515 2017-05-11 Richard Biener <rguenther@suse.de>
25517 PR tree-optimization/80705
25518 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25519 bases are not vectorizable.
25521 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25523 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25524 when counting register pressure.
25526 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25528 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25529 (iv_ca_more_deps): Renamed to ...
25530 (iv_ca_compare_deps): ... this.
25531 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25533 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25535 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25537 (determine_group_iv_costs): ... here.
25538 (find_inv_vars_cb): Record inv var if it's not recorded before.
25540 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25542 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25543 (get_shiftadd_cost): Ditto.
25545 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25547 * tree-ssa-address.c: Include header file.
25548 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25550 (add_to_parts): Refactor.
25551 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25552 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25555 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25557 PR tree-optimization/53090
25558 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25560 (extract_cond_operands): Detect condition with IV on both sides
25561 and return COMP_IV_EXPR_2.
25562 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25563 (rewrite_use_compare): Simplify by removing call to function
25564 extract_cond_operands.
25566 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25568 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25569 (extract_cond_operands): Detect condition comparing against non-
25570 invariant bound and return appropriate enum value.
25571 (find_interesting_uses_cond): Update use of extract_cond_operands.
25572 Handle its return value accordingly.
25573 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25575 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25577 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25578 nonlinear iv_use computation in loop invariant sensitive way.
25580 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25582 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25583 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25585 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25587 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25588 (dump_cand): Support iv_cand.inv_exprs.
25589 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25591 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25594 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25596 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25598 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25599 as local function. Include necessary header files.
25600 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25602 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25604 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25606 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25608 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25609 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25610 RSHIFT_EXPR and BIT_NOT_EXPR.
25612 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25614 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25615 (adjust_setup_cost): New parameter supporting round up adjustment.
25616 (struct address_cost_data): Delete.
25617 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25618 (split_address_cost, ptr_difference_cost): Delete.
25619 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25620 (struct ainc_cost_data): New struct.
25621 (get_address_cost_ainc): New function.
25622 (get_address_cost, get_computation_cost): Reimplement.
25623 (determine_group_iv_cost_address): Record inv_expr for all uses of
25625 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25626 (iv_ca_has_deps): Reimplemented to ...
25627 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25629 (iv_ca_extend): Call iv_ca_more_deps.
25631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25633 * tree-ssa-address.c (struct mem_address): Move to header file.
25634 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25635 * tree-ssa-address.h (struct mem_address): Move from C file.
25636 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25638 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25640 * tree-affine.h (aff_combination_type): New interface.
25641 (aff_combination_zero_p): Remove static.
25642 (aff_combination_const_p): New interface.
25643 (aff_combination_singleton_var_p): New interfaces.
25645 2017-05-11 Richard Biener <rguenther@suse.de>
25647 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25648 Skip unreachable blocks and destinations.
25649 (eliminate): Move stmt removal and fixup ...
25650 (fini_eliminate): ... here. Skip inserted exprs.
25651 (pass_pre::execute): Move fini_pre after fini_eliminate.
25652 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25653 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25654 PRE to get rid of dead code that has invalid SSA form and
25655 split critical edges again.
25657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25659 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25661 2017-05-11 Richard Biener <rguenther@suse.de>
25663 * passes.c (execute_function_todo): Verify loops if they are
25664 said to be up-to-date.
25665 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25666 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25668 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25671 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25672 handle calling assemble_external ourself.
25675 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25676 modes with zero size. Enhance comment.
25678 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25680 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25681 built-ins for vec_xl and vec_xst with short and char pointer
25684 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25686 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25687 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25688 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25689 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25690 (_mm_maskz_min_round_ss): New intrinsics.
25691 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25692 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25693 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25694 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25695 (__builtin_ia32_minss_mask_round): New builtins.
25696 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25697 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25698 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25700 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25701 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25703 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25706 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25708 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25709 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25710 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25711 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25712 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25713 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25714 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25715 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25716 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25717 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25718 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25719 (__builtin_ia32_mulss_mask_round): New builtins.
25720 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25721 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25722 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25724 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25725 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25727 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25730 2017-05-10 Julia Koval <julia.koval@intel.com>
25732 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25733 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25734 (_mm256_setr_m128i): New intrinsics.
25736 2017-05-10 Julia Koval <julia.koval@intel.com>
25738 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25739 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25740 (_mm_maskz_rcp14_ss): New intrinsics.
25741 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25742 (__builtin_ia32_rcp14ss_mask): New builtins.
25743 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25745 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25747 PR tree-optimization/51513
25748 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25749 (assert_unreachable_fallthru_edge_p): Use it.
25750 (group_case_labels_stmt): Likewise.
25751 * tree-cfg.h: Prototype it.
25752 * stmt.c: Include cfghooks.h and tree-cfg.h.
25753 (emit_case_dispatch_table) <gap_label>: New local variable.
25754 Use it to fill dispatch table gaps.
25755 Test for default_label before updating probabilities.
25756 (expand_case) <default_label>: Remove unneeded initialization.
25757 Test for unreachable default case statement and remove its edge.
25758 Set default_label accordingly.
25759 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25761 2017-05-10 Carl Love <cel@us.ibm.com>
25763 * config/rs6000/rs6000-c: Add support for built-in functions
25764 vector signed char vec_neg (vector signed char)
25765 vector signed short int vec_neg (vector short int)
25766 vector signed int vec_neg (vector signed int)
25767 vector signed long long vec_neg (vector signed long long)
25768 vector float vec_neg (vector float)
25769 vector double vec_neg (vector double)
25770 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25772 * config/rs6000/altivec.h: Add define for vec_neg
25773 * doc/extend.texi: Update the built-in documentation for the
25774 new built-in functions.
25776 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25778 PR tree-optimization/77644
25779 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25781 2017-05-10 Nathan Sidwell <nathan@acm.org>
25783 * dumpfile.h (TDI_lang_all): New.
25784 (TDF_KIND): New. Renumber others
25785 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25787 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25789 (get_dump_file_name): Adjust suffix generation.
25790 (dump_enable_all): Use TDF_KIND.
25791 * doc/invoke.texi (-fdump-lang-all): Document.
25793 * dumpfile.h: Tabify.
25795 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25798 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25799 Move member access before delete.
25801 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25803 * tree-inline.c (expand_call_inline): Split block at stmt
25806 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25809 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25810 are now unused after splitting mov{sf,sd}_hardfloat.
25811 (f32_lr2): Likewise.
25812 (f32_lm): Likewise.
25813 (f32_lm2): Likewise.
25814 (f32_li): Likewise.
25815 (f32_li2): Likewise.
25816 (f32_lv): Likewise.
25817 (f32_sr): Likewise.
25818 (f32_sr2): Likewise.
25819 (f32_sm): Likewise.
25820 (f32_sm2): Likewise.
25821 (f32_si): Likewise.
25822 (f32_si2): Likewise.
25823 (f32_sv): Likewise.
25824 (f32_dm): Likewise.
25825 (f32_vsx): Likewise.
25826 (f32_av): Likewise.
25827 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25828 For movsf, order stores so the VSX stores occur before the GPR
25829 store which encourages the register allocator to use a traditional
25830 FPR instead of a GPR. For movsd, order the stores so that the GPR
25831 store comes before the VSX stores to allow the power6 to work.
25832 This is due to the power6 not having a 32-bit integer store
25833 instruction from a FPR.
25834 (movsf_hardfloat): Likewise.
25835 (movsd_hardfloat): Likewise.
25837 2017-05-09 Martin Sebor <msebor@redhat.com>
25839 PR translation/80280
25840 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25843 PR translation/80280
25844 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25845 data member added in r247778.
25846 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25848 2017-05-09 Nathan Sidwell <nathan@acm.org>
25850 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25852 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25855 2017-05-09 Marek Polacek <polacek@redhat.com>
25857 * doc/invoke.texi: Fix typo.
25859 2017-05-09 Richard Biener <rguenther@suse.de>
25861 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25862 (vrp_val_is_min): Likewise.
25863 (set_value_range_to_value): Likewise.
25864 (set_value_range_to_nonnegative): Likewise.
25865 (gimple_assign_nonzero_p): Likewise.
25866 (gimple_stmt_nonzero_p): Likewise.
25867 (vrp_int_const_binop): Likewise. Remove unreachable case.
25868 (adjust_range_with_scev): Adjust comments.
25869 (compare_range_with_value): Likewise.
25870 (extract_range_from_phi_node): Likewise.
25871 (test_for_singularity): Likewise.
25873 2017-05-09 Richard Biener <rguenther@suse.de>
25875 * tree-vrp.c (get_single_symbol): Add assert that we don't
25876 get overflowed constants as invariant part.
25877 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25878 checks. Use wi::cmp instead of recursing for integer constants.
25879 (compare_values): Just ignore whether we assumed undefined
25880 overflow instead of failing the compare.
25881 (extract_range_for_var_from_comparison_expr): Add comment before the
25882 TREE_NO_WARNING sets.
25883 (test_for_singularity): Likewise.
25884 (extract_range_from_comparison): Do not disable optimization
25885 when we assumed undefined overflow.
25886 (extract_range_basic): Remove init of unused var.
25888 2017-05-09 Richard Biener <rguenther@suse.de>
25890 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25891 (extract_range_from_multiplicative_op_1): Adjust.
25892 (extract_range_from_binary_expr_1): Use int_const_binop.
25894 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25897 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25898 rs6000_store_data_bypass_p in seven define_bypass directives and
25899 in several comments.
25900 * config/rs6000/rs6000-protos.h: Add prototype for
25901 rs6000_store_data_bypass_p function.
25902 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25903 function implements slightly different (rs6000-specific) semantics
25904 than store_data_bypass_p, returning false rather than aborting
25905 with assertion error when arguments do not satisfy the
25906 requirements of store data bypass.
25907 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25908 rs6000_store_data_bypass_p.
25910 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25912 * config/xtensa/xtensa-protos.h
25913 (xtensa_initial_elimination_offset): New declaration.
25914 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25915 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25916 macro definition, add case for FRAME_POINTER_REGNUM when
25917 FRAME_GROWS_DOWNWARD.
25918 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25919 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25920 xtensa_initial_elimination_offset.
25922 2017-05-08 Nathan Sidwell <nathan@acm.org>
25924 * doc/invoke.texi: Alphabetize -fdump options.
25926 2017-05-08 Martin Sebor <msebor@redhat.com>
25928 PR translation/80280
25929 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25931 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25933 * target.def (compute_frame_layout): New optional target hook.
25934 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25935 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25936 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25938 * reload1.c (verify_initial_elim_offsets): Likewise.
25939 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25940 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25941 (arm_get_frame_offsets): Split up into this ...
25942 (arm_compute_frame_layout): ... and this function.
25944 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25946 * config/aarch64/constraints.md (Usa): New constraint.
25947 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25949 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25951 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25952 with_multilib_list after it has been checked.
25954 2017-05-08 Richard Biener <rguenther@suse.de>
25956 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25957 (bitmap_set_subtract_values): Likewise.
25959 2017-05-08 Richard Biener <rguenther@suse.de>
25961 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25962 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25964 (gimple_stmt_nonzero_warnv_p): Rename to ...
25965 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25967 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25968 (extract_range_basic): Adjust, do not disable propagation on
25969 strict overflow sensitive simplification.
25970 (vrp_visit_cond_stmt): Likewise.
25972 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25974 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25975 body size unconditionally.
25977 2017-05-07 Jeff Law <law@redhat.com>
25980 2017-05-06 Jeff Law <law@redhat.com>
25981 PR tree-optimization/78496
25982 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25985 PR tree-optimization/78496
25986 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25987 (simplify_stmt_using_ranges): Call it.
25988 (vrp_dom_walker::before_dom_children): Extract equivalences
25989 from an ASSERT_EXPR with an equality comparison against a
25992 2017-05-06 Jeff Law <law@redhat.com>
25994 PR tree-optimization/78496
25995 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25998 PR tree-optimization/78496
25999 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26000 (simplify_stmt_using_ranges): Call it.
26001 (vrp_dom_walker::before_dom_children): Extract equivalences
26002 from an ASSERT_EXPR with an equality comparison against a
26005 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26007 * lra-constraints.c (lra_copy_reg_equiv): New function.
26008 (split_reg): Use it to copy equivalence information from the
26009 original register to the spill register.
26011 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26013 PR rtl-optimization/75964
26014 * simplify-rtx.c (simplify_const_relational_operation): Remove
26015 invalid handling of comparisons of integer ABS.
26017 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26019 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26020 initialize to zero.
26021 (init_regs): Remove declaration.
26022 (function_arg_advance_32): Initialize error_p as boolean variable.
26024 2017-05-05 Nathan Sidwell <nathan@acm.org>
26026 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26027 lines. Use for (;;).
26029 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26031 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26032 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26033 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26034 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26035 VF=2 that require versioning.
26037 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26039 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26042 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26044 * diagnostic.h (diagnostic_override_option_index): Convert from
26045 macro to inline function.
26047 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26049 * diagnostic.c (last_module_changed_p): New function.
26050 (set_last_module): New function.
26051 (diagnostic_report_current_module): Convert macro usage to
26052 the above functions.
26053 * diagnostic.h (diagnostic_context::last_module): Strengthen
26054 from const line_map * to const line_map_ordinary *.
26055 (diagnostic_last_module_changed): Delete macro.
26056 (diagnostic_set_last_module): Delete macro.
26058 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26060 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26061 with diagnostic_report_diagnostic.
26062 (diagnostic_n_impl_richloc): Likewise.
26063 * diagnostic.h (report_diagnostic): Delete macro.
26064 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26065 with diagnostic_report_diagnostic.
26066 * substring-locations.c (format_warning_va): Likewise.
26068 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26070 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26071 save/restor of format_spec. Move option-printing code to...
26072 (print_option_information): ...this new function, and
26073 reimplement by simply printing to the pretty_printer,
26074 rather than appending to the format string.
26076 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26078 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26079 handling logic into...
26080 (update_effective_level_from_pragmas): ...this new function.
26082 2017-05-04 Andrew Waterman <andrew@sifive.com>
26084 * config/riscv/riscv.opt (mstrict-align): New option.
26085 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26086 (SLOW_UNALIGNED_ACCESS): Define.
26087 (riscv_slow_unaligned_access): Declare.
26088 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26090 (riscv_slow_unaligned_access): New variable.
26091 (rocket_tune_info): Set slow_unaligned_access to true.
26092 (optimize_size_tune_info): Set slow_unaligned_access to false.
26093 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26094 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26095 (riscv_option_override): Set riscv_slow_unaligned_access.
26096 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26098 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26100 * config/riscv/riscv.md: Unify indentation.
26102 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26107 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26109 (extendsi<mode>2): Add support for doing sign extension via
26110 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26111 don't have ISA 3.0 instructions.
26112 (extendsi<mode>2 splitter): Likewise.
26113 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26114 generate the normal insns since SImode can now go in vector
26115 registers. Disallow the special UNSPECs needed for previous
26116 machines to hide SImode being used. Add new insns
26117 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26118 (fix_trunc<mode>si2_stfiwx): Likewise.
26119 (fix_trunc<mode>si2_internal): Likewise.
26120 (fixuns_trunc<mode>si2): Likewise.
26121 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26122 (fctiw<u>z_<mode>_smallint): Likewise.
26123 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26124 of floating point to 32-bit integer from doing a direct move to
26125 the GPR registers to do a store.
26126 (fctiwz_<mode>): Break long line.
26128 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26130 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26131 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26132 (addr_list, addr_offset_valid_p): New.
26133 (split_address_groups): Check offset validity with above function.
26134 (gt-tree-ssa-loop-ivopts.h): Include header file.
26136 2017-05-05 Nathan Sidwell <nathan@acm.org>
26138 * config.gcc (arm*-*-*): Add missing 'fi'.
26140 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26142 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26143 included in -fopt-info does not matter.
26144 * doc/optinfo.texi (-fopt-info): Fix description of default
26145 behavour. Explicitly say order of options included in -fopt-info
26148 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26150 * config.gcc: Allow combinations of aprofile and rmprofile values for
26151 --with-multilib-list.
26152 * config/arm/t-multilib: New file.
26153 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26154 variables. Remove setting of ISA and floating-point ABI in
26155 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26156 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26157 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26158 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26160 * config/arm/t-rmprofile: Likewise except for the matches changes.
26161 * doc/install.texi (--with-multilib-list): Document the combination of
26162 aprofile and rmprofile values and warn about pitfalls in doing that.
26164 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26166 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26167 (movdi_aarch64): Likewise.
26169 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26171 PR tree-optimization/80632
26172 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26174 (build_arrays): Initialize it for virtual phis.
26175 (fix_phi_nodes): Use it for virtual phis.
26177 PR tree-optimization/80558
26178 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26179 [x, y] op z into [x op, y op z] for op & or | if conditions
26182 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26183 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26186 * config/arm/arm.md (use_literal_pool): Remove.
26187 (64-bit immediate split): No longer takes cost into consideration
26188 if arm_disable_literal_pool is enabled.
26189 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26190 used when arm_disable_literal_pool is enabled.
26191 (arm_max_const_double_inline_cost): Remove use of
26192 arm_disable_literal_pool.
26193 (push_minipool_fix): Add assert.
26194 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26195 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26196 (no_literal_pool_sf_immediate): New.
26198 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26200 PR tree-optimization/80613
26201 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26202 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26204 2017-05-05 Richard Biener <rguenther@suse.de>
26206 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26208 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26210 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26211 of this flag from insn conditions due to removal from r247495.
26213 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26215 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26217 (arm_early_store_addr_dep_ptr): Likewise.
26218 * config/arm/aarch-common-protos.h
26219 (arm_early_load_addr_dep_ptr): Add prototype.
26220 (arm_early_store_addr_dep_ptr): Likewise.
26221 * config/arm/cortex-a53.md: Add new bypasses.
26223 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26225 * tree.c (next_type_uid): Change type to unsigned.
26226 (type_hash_canon): Decrement back next_type_uid if
26227 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26228 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26231 2017-05-04 Martin Sebor <msebor@redhat.com>
26233 * builtins.c: Fix a trivial typo in a comment.
26235 PR middle-end/79234
26236 * builtins.c (check_sizes): Adjust to handle reading past the end.
26237 Avoid printing excessive upper bound of ranges. Use %E to print
26238 tree nodes instead of converting them to %wu.
26239 (expand_builtin_memchr): New function.
26240 (compute_dest_size): Rename...
26241 (compute_objsize): ...to this.
26242 (expand_builtin_memcpy): Adjust.
26243 (expand_builtin_mempcpy): Adjust.
26244 (expand_builtin_strcat): Adjust.
26245 (expand_builtin_strcpy): Adjust.
26246 (check_strncat_sizes): Adjust.
26247 (expand_builtin_strncat): Adjust.
26248 (expand_builtin_strncpy): Adjust and simplify.
26249 (expand_builtin_memset): Adjust.
26250 (expand_builtin_bzero): Adjust.
26251 (expand_builtin_memcmp): Adjust.
26252 (expand_builtin): Handle memcmp.
26253 (maybe_emit_chk_warning): Check strncat just once.
26255 2017-05-04 Martin Sebor <msebor@redhat.com>
26257 PR preprocessor/79214
26258 PR middle-end/79222
26259 PR middle-end/79223
26260 * builtins.c (check_sizes): Add inlining context and issue
26261 warnings even when -Wno-system-headers is set.
26262 (check_strncat_sizes): Same.
26263 (expand_builtin_strncat): Same.
26264 (expand_builtin_memmove): New function.
26265 (expand_builtin_stpncpy): Same.
26266 (expand_builtin): Handle memmove and stpncpy.
26268 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26270 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26271 which is not used any more.
26273 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26275 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26277 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26279 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26280 (cortexa53_tunings): Likewise.
26281 (cortexa57_tunings): Likewise.
26282 (cortexa72_tunings): Likewise.
26283 (cortexa73_tunings): Likewise.
26285 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26287 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26288 Set loop alignment to 8.
26290 2017-05-04 Martin Sebor <msebor@redhat.com>
26292 PR translation/80280
26293 * builtins.c (expand_builtin_object_size): Add missing quoting to
26294 %D and like directives.
26295 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26296 (hsa_type_for_tree_type): Same.
26297 (verify_function_arguments): Same.
26298 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26299 * varasm.c (get_section): Same.
26302 2017-05-04 Martin Sebor <msebor@redhat.com>
26304 PR translation/80280
26305 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26307 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26309 * config/aarch64/aarch64.c (generic_addrcost_table):
26310 Change HI/TI mode setting.
26312 2017-05-04 Martin Jambor <mjambor@suse.cz>
26314 PR tree-optimization/80622
26315 * tree-sra.c (comes_initialized_p): New function.
26316 (build_accesses_from_assign): Only set write lazily when
26317 comes_initialized_p is false.
26318 (analyze_access_subtree): Use comes_initialized_p.
26319 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26320 instead of testing for PARM_DECL.
26322 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26324 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26325 constraint on operand 0 to allow more general addressing modes.
26326 Adjust output template.
26327 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26329 * config/aarch64/aarch64-protos.h
26330 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26331 * config/aarch64/constraints.md (Dp): New address constraint.
26332 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26335 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26337 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26338 update use of estimate_ipcp_clone_size_and_time.
26339 (estimate_local_effects): Update use of
26340 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26341 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26342 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26343 Return nonspecialized time.
26345 2017-05-04 Richard Biener <rguenther@suse.de>
26347 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26348 for the last VUSE which def dominates the PHI. Directly call
26350 (get_continuation_for_phi_1): Remove.
26352 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26354 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26355 to explain the use of truncating division. Cap the number of
26356 iterations to the maximum given by nb_iterations_upper_bound,
26359 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26361 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26362 * configure: Regenerate.
26363 * config.in: Regenerate.
26364 * config/i386/driver-mingw32.c: new file.
26365 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26366 * config.host: Link driver-mingw32.o on MinGW host.
26367 * doc/install.texi: Document new --enable-mingw-wildcard configure
26370 2017-05-04 Marek Polacek <polacek@redhat.com>
26372 PR tree-optimization/80612
26373 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26375 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26376 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26378 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26379 (movt splitter): Likewise.
26380 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26381 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26382 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26383 block for Thumb-1 with MOVT.
26384 (thumb2_legitimate_address_p): Move code block ...
26385 (can_avoid_literal_pool_for_label_p): ... into this new function.
26386 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26388 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26389 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26390 "M-profile targets with the MOVT instruction".
26392 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26394 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26395 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26396 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26398 2017-05-04 Martin Liska <mliska@suse.cz>
26400 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26401 variable cond_code.
26403 2017-05-04 Richard Biener <rguenther@suse.de>
26405 * tree.c (array_at_struct_end_p): Handle arrays at struct
26406 end with flexarrays more conservatively. Refactor and treat
26407 arrays of arrays or aggregates more strict. Fix
26408 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26409 * tree.c (array_at_struct_end_p): Adjust prototype.
26410 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26411 * gimple-fold.c (get_range_strlen): Likewise.
26412 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26414 2017-05-04 Richard Biener <rguenther@suse.de>
26416 PR tree-optimization/31130
26417 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26419 (supports_overflow_infinity): Likewise.
26420 (is_negative_overflow_infinity): Likewise.
26421 (is_positive_overflow_infinity): Likewise.
26422 (is_overflow_infinity): Likewise.
26423 (stmt_overflow_infinity): Likewise.
26424 (overflow_infinity_range_p): Likewise.
26425 (usable_range_p): Remove as always returning true.
26426 (make_overflow_infinity): Remove.
26427 (negative_overflow_infinity): Likewise.
26428 (positive_overflow_infinity): Likewise.
26429 (avoid_overflow_infinity): Likewise.
26430 (set_value_range): Adjust accordingly.
26431 (set_value_range_to_nonnegative): Likewise, remove now unused
26432 overflow_infinity arg.
26433 (vrp_operand_equal_p): Adjust.
26434 (update_value_range): Likewise.
26435 (range_int_cst_singleton_p): Likewise.
26436 (operand_less_p): Likewise.
26437 (compare_values_warnv): Likewise.
26438 (extract_range_for_var_from_comparison_expr): Likewise.
26439 (vrp_int_const_binop): Likewise.
26440 (zero_nonzero_bits_from_vr): Likewise.
26441 (extract_range_from_multiplicative_op_1): Likewise.
26442 (extract_range_from_binary_expr_1): Likewise.
26443 (extract_range_from_unary_expr): Likewise.
26444 (extract_range_from_comparison): Likewise.
26445 (extract_range_basic): Likewise.
26446 (adjust_range_with_scev): Likewise.
26447 (compare_ranges): Likewise.
26448 (compare_range_with_value): Likewise.
26449 (dump_value_range): Likewise.
26450 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26452 (simplify_cond_using_ranges): Adjust.
26454 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26456 * brig-builtins.def: Added a builtin for class_f64.
26457 * builtin-types.def: Added a builtin type needed by class_f64.
26459 2017-05-03 Jason Merrill <jason@redhat.com>
26461 * timevar.def: Add TV_CONSTEXPR.
26463 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26465 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26467 2017-05-03 Martin Jambor <mjambor@suse.cz>
26469 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26470 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26471 * ipa-cp.c (ipcp_propagate_stage): Do not call
26472 ipa_update_after_lto_read.
26473 * ipa-inline.c (ipa_inline): Likewise.
26475 2017-05-03 Martin Jambor <mjambor@suse.cz>
26477 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26478 tag. Added a default constructor and a destructor.
26479 (ipa_edge_args_sum_t): New class;
26480 (ipa_edge_args_sum): Declare.
26481 (ipa_edge_args_vector): Remove declaration.
26482 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26483 (ipa_free_edge_args_substructures): Remove declaration.
26484 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26485 (ipa_edge_args_info_available_for_edge_p): Likewise.
26486 * ipa-prop.c (ipa_edge_args_vector): Removed.
26487 (edge_removal_hook_holder): Likewise.
26488 (edge_duplication_hook_holder): Likewise.
26489 (ipa_edge_args_sum): New variable.
26490 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26491 ipa_edge_args_vector.
26492 (ipa_free_edge_args_substructures): Likewise.
26493 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26494 ipa_edge_args_vector.
26495 (ipa_edge_removal_hook): Turned into method
26496 ipa_edge_args_sum_t::remove.
26497 (ipa_edge_duplication_hook): Turned into method
26498 ipa_edge_args_sum_t::duplicate.
26499 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26500 registering edge hooks.
26501 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26502 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26503 ipa_edge_args_sum instead of ipa_edge_args_vector.
26504 * ipa-profile.c (ipa_profile): Likewise.
26506 2017-05-03 Martin Jambor <mjambor@suse.cz>
26508 * symbol-summary.h (function_summary): New method exists.
26509 (function_summary::symtab_removal): Deallocate through release.
26510 (call_summary): New class.
26511 (gt_ggc_mx): New overload.
26512 (gt_pch_nx): Likewise.
26513 (gt_pch_nx): Likewise.
26515 2017-05-03 Jeff Law <law@redhat.com>
26517 PR tree-optimization/78496
26518 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26519 from simplify_cond_using_ranges. Split off code to walk
26520 backwards through casts into ...
26521 (simplify_cond_using_ranges_2): New function.
26522 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26523 (execute_vrp): After identifying jump threads, call
26524 simplify_cond_using_ranges_2.
26526 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26529 * ipa-inline.h (inline_summary): Add ctor.
26530 (create_ggc): Do not use ggc_cleared_alloc.
26532 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26533 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26535 * gcc.c (handle_braces): Support escaping in switch matching
26537 * doc/invoke.texi (Spec Files): Document it.
26538 Remove superfluous @code markup in items.
26540 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26542 * diagnostic-show-locus.c (struct column_range): New struct.
26543 (get_affected_columns): New function.
26544 (get_printed_columns): New function.
26545 (struct correction): New struct.
26546 (correction::ensure_capacity): New function.
26547 (correction::ensure_terminated): New function.
26548 (struct line_corrections): New struct.
26549 (line_corrections::~line_corrections): New dtor.
26550 (line_corrections::add_hint): New function.
26551 (layout::print_trailing_fixits): Reimplement in terms of the new
26553 (selftest::test_overlapped_fixit_printing): New function.
26554 (selftest::diagnostic_show_locus_c_tests): Call it.
26556 2017-05-03 Nathan Sidwell <nathan@acm.org>
26558 Canonicalize canonical type hashing
26559 * tree.h (type_hash_canon_hash): Declare.
26560 * tree.c (type_hash_list, attribute_hash_list): Move into
26561 type_hash_canon_hash.
26562 (build_type_attribute_qual_variant): Break out hash code calc into
26563 type_hash_canon_hash.
26564 (type_hash_canon_hash): New. Generic type hash computation.
26565 (build_range_type_1, build_array_type_1, build_function_type,
26566 build_method_type_directly, build_offset_type, build_complex_type,
26567 make_vector_type): Call it.
26569 2017-05-03 Richard Biener <rguenther@suse.de>
26571 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26572 When all DRs have unknown misaligned do not always peel
26573 when there is a store but apply the same costing model as if
26574 there were only loads.
26576 2017-05-03 Richard Biener <rguenther@suse.de>
26579 PR tree-optimization/80492
26580 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26581 compare_base_decls returning dont-know properly.
26583 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26585 * config/arm/iterators.md (CCSI): New mode iterator.
26586 (arch): New mode attribute.
26587 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26588 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26589 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26590 code iterator for success result mode.
26591 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26592 the corresponding new insn generators.
26594 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26597 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26598 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26600 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26602 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26603 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26604 (DDR_A): Wrap DDR argument in brackets.
26605 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26606 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26607 (DDR_REVERSED_P): Likewise.
26609 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26611 PR tree-optimization/79472
26612 * tree-switch-conversion.c (struct switch_conv_info): Add
26613 contiguous_range and default_case_nonstandard fields.
26614 (collect_switch_conv_info): Compute contiguous_range and
26615 default_case_nonstandard fields, don't clear final_bb if
26616 contiguous_range and only the default case doesn't have the required
26618 (check_all_empty_except_final): Set default_case_nonstandard instead
26619 of failing if contiguous_range and the default case doesn't have empty
26621 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26622 and only the default case doesn't have the required constants. Skip
26624 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26625 if default_case_nonstandard.
26626 (build_constructors): Build constant 1 just once. Assert that default
26627 values aren't inserted in between cases if contiguous_range. Skip
26629 (build_arrays): Skip virtual phis.
26630 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26631 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26632 Handle virtual phis.
26633 (gen_inbound_check): Handle default_case_nonstandard case.
26634 (process_switch): Adjust check_final_bb caller. Call
26635 gather_default_values with the first non-default case instead of
26636 default case if default_case_nonstandard.
26638 2017-05-02 Nathan Sidwell <nathan@acm.org>
26640 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26641 check. Fix formatting.
26643 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26645 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26646 errors when comparing specialized and unspecialized times.
26648 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26650 * diagnostic-show-locus.c
26651 (layout::should_print_annotation_line_p): Make private.
26652 (layout::print_annotation_line): Make private.
26653 (layout::annotation_line_showed_range_p): Make private.
26654 (layout::show_ruler): Make private.
26655 (layout::print_source_line): Make private. Pass in line and
26656 line_width, rather than calling location_get_source_line. Drop
26658 (layout::print_leading_fixits): New method.
26659 (layout::print_any_fixits): Rename to...
26660 (layout::print_trailing_fixits): ...this, and make private.
26661 Don't print newline fixits.
26662 (diagnostic_show_locus): Move logic for printing one row into...
26663 (layout::print_line): ...this new function. Move the
26664 location_get_source_line call and error-handling from
26665 print_source_line to here. Call print_leading_fixits, and rename
26666 print_any_fixits to print_trailing_fixits.
26667 (selftest::test_fixit_insert_containing_newline): Update now that
26668 newlines are partially supported.
26669 (selftest::test_fixit_insert_containing_newline_2): New test.
26670 (selftest::test_fixit_replace_containing_newline): Update comments.
26671 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26672 * edit-context.c (class added_line): New class.
26673 (class edited_line): Describe newline handling in comment.
26674 (edited_line::actually_edited_p): New method.
26675 (edited_line::print_content): Delete redundant decl.
26676 (edited_line::m_predecessors): New field.
26677 (edited_file::print_content): Call edited_line::print_content.
26678 (edited_file::print_diff): Update to support newlines.
26679 (edited_file::print_diff_hunk): Likewise.
26680 (edited_file::print_run_of_changed_lines): New function.
26681 (edited_file::print_diff_line): Convert to...
26682 (print_diff_line): ...this.
26683 (edited_file::get_effective_line_count): New function.
26684 (edited_line::edited_line): Initialize new field m_predecessors.
26685 (edited_line::~edited_line): Clean up m_predecessors.
26686 (edited_line::apply_fixit): Handle newlines.
26687 (edited_line::get_effective_line_count): New function.
26688 (edited_line::print_content): New function.
26689 (edited_line::print_diff_lines): New function.
26690 (selftest::test_applying_fixits_insert_containing_newline): New
26692 (selftest::test_applying_fixits_replace_containing_newline): New
26694 (selftest::insert_line): New function.
26695 (selftest::test_applying_fixits_multiple_lines): Add example of
26697 (selftest::edit_context_c_tests): Call the new tests.
26699 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26701 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26702 parameter cand. Update dump information.
26703 (get_computation_cost): Update uses.
26705 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26707 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26708 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26709 (get_computation_at, rewrite_use_address): Update use of
26710 get_computation_aff.
26712 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26714 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26715 (get_computation): Delete.
26716 (get_computation_cost): Implement like get_computation_cost_at.
26717 Use get_computation_at.
26718 (get_computation_cost_at): Delete.
26719 (rewrite_use_nonlinear_expr): Use get_computation_at.
26720 (rewrite_use_compare, remove_unused_ivs): Ditto.
26722 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26724 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26726 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26728 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26729 (ivopts_global_cost_for_size): Rename parameter and update uses.
26730 (iv_ca_recount_cost): Update uses.
26731 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26732 candidates seperately in n_invs and n_cands.
26733 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26735 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26737 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26738 (find_inv_vars_cb): New.
26739 (find_depends): Renamed to ...
26740 (find_inv_vars): ... this.
26741 (add_candidate_1, force_var_cost): Call find_inv_vars.
26742 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26744 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26746 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26747 inv_vars. Add inv_exprs.
26748 (struct iv_cand): Rename depends_on to inv_vars.
26749 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26750 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26751 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26752 (dump_cand): Dump inv_vars.
26753 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26754 (record_invariant, find_depends, add_candidate_1): Ditto.
26755 (set_group_iv_cost, force_var_cost): Ditto.
26756 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26757 (get_computation_cost_at, get_computation_cost): Ditto.
26758 (determine_group_iv_cost_generic): Ditto.
26759 (determine_group_iv_cost_address): Ditto.
26760 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26761 (determine_group_iv_costs): Ditto.
26762 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26763 (iv_ca_set_remove_invariants): Renamed to ...
26764 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26765 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26766 (iv_ca_set_add_invariants): Renamed to ...
26767 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26768 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26769 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26770 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26771 (create_new_ivs): Remove useless dump.
26773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26775 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26777 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26778 (iv_ca_set_no_cp, create_new_iv): Ditto.
26780 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26782 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26784 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26786 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26787 function tree_check2.
26789 2017-05-02 Martin Liska <mliska@suse.cz>
26791 * doc/gcov.texi: Add missing preposition.
26792 * gcov.c (function_info::function_info): Properly fill up
26793 all member variables.
26795 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26797 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26799 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26801 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26803 2017-05-02 Martin Liska <mliska@suse.cz>
26806 * lto-streamer-in.c (lto_read_tree_1): Remove
26807 LTO_STREAMER_DEBUG.
26808 * lto-streamer.c (struct tree_hash_entry): Likewise.
26809 (struct tree_entry_hasher): Likewise.
26810 (tree_entry_hasher::hash): Likewise.
26811 (tree_entry_hasher::equal): Likewise.
26812 (lto_streamer_init): Likewise.
26813 (lto_orig_address_map): Likewise.
26814 (lto_orig_address_get): Likewise.
26815 (lto_orig_address_remove): Likewise.
26816 * lto-streamer.h: Likewise.
26817 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26818 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26820 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26822 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26823 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26824 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26825 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26826 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26827 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26828 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26829 (mm_maskz_sub_ss): New intrinsics.
26830 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26831 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26832 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26833 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26834 (__builtin_ia32_subss_mask_round): New builtins.
26835 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26836 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26837 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26839 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26840 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26842 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26845 2017-05-02 Martin Jambor <mjambor@suse.cz>
26847 PR tree-optimization/78687
26848 * tree-sra.c (access): New field parent.
26849 (process_subtree_disqualification): New function.
26850 (disqualify_candidate): Call it.
26851 (build_accesses_from_assign): Reset write flag if creating an
26853 (build_access_subtree): Fill in parent field and also prpagate
26854 down grp_write flag.
26855 (create_artificial_child_access): New parameter set_grp_write, set
26856 grp_write to its value.
26857 (propagate_subaccesses_across_link): Also propagate grp_write flag
26859 (propagate_all_subaccesses): Push the closest parent back to work
26860 queue if add_access_to_work_queue returned true.
26862 2017-05-02 Richard Biener <rguenther@suse.de>
26864 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26865 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26866 -fstrict-overflow documentation.
26867 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26868 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26869 flag_strict_overflow.
26870 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26871 * lto-opts.c (lto_write_options): Do not stream it.
26872 * lto-wrapper.c (merge_and_complain): Do not handle it.
26873 * opts.c (default_options_table): Do not set -fstrict-overflow.
26874 (finish_options): Likewise do not clear it when sanitizing.
26875 * simplify-rtx.c (simplify_const_relational_operation): Do not
26876 test flag_strict_overflow.
26878 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26880 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26881 using enabled attribute.
26882 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26883 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26884 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26885 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26886 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26887 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26888 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26889 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26890 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26891 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26893 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26895 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26897 2017-05-02 Richard Biener <rguenther@suse.de>
26899 PR tree-optimization/80591
26901 2017-04-10 Richard Biener <rguenther@suse.de>
26903 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26906 2017-05-02 Richard Biener <rguenther@suse.de>
26908 PR tree-optimization/80549
26909 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26910 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26911 headers if they do not have a preheader.
26913 2017-05-02 Martin Liska <mliska@suse.cz>
26916 * common.opt: Fix typo.
26917 * doc/invoke.texi: Likewise.
26919 2017-05-01 Jan Beulich <jbeulich@suse.com>
26921 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26922 swapping, add (x,x,m,x,n) alternative.
26924 2017-05-01 Nathan Sidwell <nathan@acm.org>
26926 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26927 unnecessary unadjusted_alignment check.
26929 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26932 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26934 * gimplify.c (gimplify_cilk_detach): New function.
26935 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26936 * tree-core.h: Document EXPR_CILK_SPAWN.
26937 * tree.h (EXPR_CILK_SPAWN): Define.
26939 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26941 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26942 to use new fixit_hint representation, using the "replace" logic.
26943 (get_line_span_for_fixit_hint): Likewise.
26944 (layout::print_any_fixits): Likewise.
26945 (selftest::test_one_liner_many_fixits): Rename to...
26946 (selftest::test_one_liner_many_fixits_1): ...this, and update
26947 comment and expected output to reflect that the multiple fix-it
26948 hints are now consolidated into one insertion.
26949 (selftest::test_one_liner_many_fixits_2): New test.
26950 (selftest::test_diagnostic_show_locus_one_liner): Update for
26952 (selftest::test_fixit_consolidation): Update for fix-it API
26954 * diagnostic.c (print_parseable_fixits): Likewise.
26955 * edit-context.c (edited_line::m_line_events): Convert from
26956 auto_vec <line_event *> to auto_vec <line_event>.
26957 (class line_event): Convert from abstract base class to a concrete
26958 class, taking over the role of replace_event.
26959 (class insert_event): Delete.
26960 (class replace_event): Rename to class line_event. Convert to
26962 (edit_context::add_fixits): Reimplement.
26963 (edit_context::apply_insert): Delete.
26964 (edit_context::apply_replace): Rename to...
26965 (edit_context::apply_fixit): ...this. Convert to half-open range.
26966 (edited_file::apply_insert): Delete.
26967 (edited_file::apply_replace): Rename to...
26968 (edited_file::apply_fixit): ...this.
26969 (edited_line::~edited_line): Drop deletion of events.
26970 (edited_line::apply_insert): Delete.
26971 (edited_line::apply_replace): Rename to...
26972 (edited_line::apply_fixit): ...this. Convert to half-open range.
26973 Update for change to type of m_line_events.
26974 * edit-context.h (edit_context::apply_insert): Delete.
26975 (edit_context::apply_replace): Rename to...
26976 (edit_context::apply_fixit): ...this.
26978 2017-05-01 Martin Sebor <msebor@redhat.com>
26980 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26983 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26986 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26987 __get_cpuid_max returns 0.
26988 (__get_cpuid_count): Ditto.
26990 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26992 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26993 replacement expression is another instance of one of its arguments.
26995 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26998 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26999 check for stack push/pop autoinc.
27000 * config/i386/i386.c (ix86_agi_dependent): Return false
27001 if the only reason why modified_in_p returned true is that
27002 addr is SP based and set_insn is a push or pop.
27004 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27006 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27009 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27012 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27013 (account_size_time): Use two predicates - exec_pred and
27015 (evaluate_conditions_for_known_args): Compute both clause and
27017 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27018 (inline_summary_t::duplicate): Update.
27019 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27021 (compute_inline_parameters): Likewise.
27022 (estimate_edge_size_and_time): Update caluclation of time.
27023 (estimate_node_size_and_time): Compute both time and nonspecialized
27025 (estimate_ipcp_clone_size_and_time): Update.
27026 (inline_merge_summary): Update.
27027 (do_estimate_edge_time): Update.
27028 (do_estimate_edge_size): Update.
27029 (do_estimate_edge_hints): Update.
27030 (inline_read_section, inline_write_summary): Stream both new predicates.
27031 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27033 (compute_inlined_call_time): Cleanup.
27034 (big_speedup_p): Update.
27035 (edge_badness): Update.
27036 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27037 (size_time_entry): Replace predicate by exec_predicate and
27038 nonconst_predicate.
27039 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27040 (estimate_edge_time): Return also nonspec_time.
27041 (reset_edge_growth_cache): Update.
27043 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27045 PR rtl-optimization/80491
27046 * ifcvt.c (noce_process_if_block): When looking for x setter
27047 with missing else_bb, don't check only the insn right before
27048 cond_earliest, but look for the last insn that x is modified in
27049 within the same bb.
27051 PR rtl-optimization/80491
27052 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27054 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27056 PR tree-optimization/80487
27057 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27059 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27061 PR tree-optimization/79697
27062 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27063 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27064 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27066 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27067 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27069 2017-04-28 Martin Sebor <msebor@redhat.com>
27071 PR tree-optimization/80523
27072 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27073 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27075 (maybe_warn, format_directive, parse_directive): Use new functions.
27076 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27078 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27080 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27082 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27084 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27085 target_header_dir): Set correctly.
27086 * configure: Regenerated.
27087 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27088 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27089 instead of SYSTEM_HEADER_DIR.
27091 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27093 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27094 (estimate_local_effects): Likewise.
27095 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27096 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27097 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27098 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27099 do_estimate_edge_time, estimate_edge_time): Likewise.
27100 * ipa-inline-analysis.c (estimate_node_size_and_time,
27101 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27102 (estimate_time_after_inlining): Remove.
27104 2017-04-28 Martin Liska <mliska@suse.cz>
27106 * doc/gcov.texi: Enhance documentation of gcov.
27108 2017-04-28 Martin Liska <mliska@suse.cz>
27110 * doc/gcov.texi: Sort options in alphabetic order.
27111 * doc/gcov-dump.texi: Likewise.
27112 * doc/gcov-tool.texi: Likewise.
27113 * gcov.c (print_usage): Likewise.
27114 * gcov-dump.c (print_usage): Likewise.
27115 * gcov-tool.c (print_merge_usage_message): Likewise.
27116 (print_rewrite_usage_message): Likewise.
27117 (print_overlap_usage_message): Likewise.
27119 2017-04-28 Martin Liska <mliska@suse.cz>
27121 PR gcov-profile/53915
27122 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27124 2017-04-28 Martin Liska <mliska@suse.cz>
27126 PR gcov-profile/79891
27127 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27128 is marked by compiler as living on a line.
27129 (get_cycles_count): Remove usage of the union.
27130 (output_intermediate_file): Likewise.
27131 (find_source): Fix GNU coding style.
27132 (accumulate_line_counts): Remove old non-all block mode.
27133 (output_lines): Remove usage of the union.
27134 * profile.c (output_location): Include all BBs, even if
27135 belonging to a same line (and file) as a previous BB.
27137 2017-04-28 Martin Liska <mliska@suse.cz>
27139 * gcov.c (process_args): Handle new argument 'w'.
27140 (read_graph_file): Assign ID to BBs.
27141 (output_branch_count): Display BB # if verbose flag is set.
27142 (output_lines): Likewise for arcs.
27143 (print_usage): Add '--verbose' option help.
27144 * doc/gcov.texi: Document --verbose (-w) option.
27146 2017-04-28 Martin Liska <mliska@suse.cz>
27148 * gcov.c (struct block_location_info): New struct.
27149 (process_file): Fill up the new structure.
27150 (read_graph_file): Replace usage of encoding by the newly added
27152 (add_line_counts): Likewise.
27153 (accumulate_line_counts): Remove usage of the union.
27154 (function_info::function_info): New function.
27155 (function_info::~function_info): Likewise.
27156 (process_file): Call delete instead of release_function.
27157 (release_function): Release the function.
27158 (release_structures): Call delete instead of release_function.
27159 (solve_flow_graph): Replace usage of num_blocks.
27160 (find_exception_blocks): Likewise.
27161 (output_lines): Fix GNU coding style.
27163 2017-04-28 Martin Liska <mliska@suse.cz>
27166 * coverage.c (coverage_remove_note_file): New function.
27167 * coverage.h: Declare the function.
27168 * toplev.c (finalize): Clean if an error has been seen.
27170 2017-04-28 Martin Liska <mliska@suse.cz>
27172 PR gcov-profile/80031
27173 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27174 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27175 * gcov.c (read_graph_file): Read just number of blocks.
27176 * profile.c (branch_prob): Do not stream 0 flags per a basic
27179 2017-04-28 Martin Liska <mliska@suse.cz>
27181 * gcov-dump.c (tag_*): Add new argument to declarations.
27182 (dump_gcov_file): Likewise.
27183 (tag_blocks): Add and use new argument depth.
27184 (tag_arcs): Likewise.
27185 (tag_lines): Likewise.
27186 (tag_counters): Likewise.
27187 (tag_summary): Likewise.
27188 (dump_working_sets): Use depth to do a proper indentation.
27190 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27193 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27194 * symtab.c (symtab_node::debug_symtab): Move definition here.
27196 2017-04-28 Richard Biener <rguenther@suse.de>
27198 * lto-streamer.h (LTO_major_version): Bump to 7.
27200 2017-04-28 Richard Biener <rguenther@suse.de>
27202 * tree-vrp.c (assert_info): New struct.
27203 (add_assert_info): New helper.
27204 (register_edge_assert_for_2): Refactor to add asserts to a vector
27206 (register_edge_assert_for_1): Likewise.
27207 (register_edge_assert_for): Likewise.
27208 (finish_register_edge_assert_for): New helper actually registering
27209 asserts where live on edge.
27210 (find_conditional_asserts): Adjust.
27211 (find_switch_asserts): Likewise.
27212 (evrp_dom_walker::try_find_new_range): Generalize.
27213 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27215 2017-04-27 Marek Polacek <polacek@redhat.com>
27218 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27219 arg10 and arg11 to itype.
27221 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27223 * doc/extend.texi (Object Size Checking): Improve grammar.
27225 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27228 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27229 that the logic for permitting reciprocal estimates matches that
27232 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27235 * tree.c (type_cache_hasher::equal): Only compare
27236 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27237 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27238 non-aggregate element types.
27239 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27240 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27242 2017-04-27 Richard Biener <rguenther@suse.de>
27244 PR middle-end/80533
27245 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27246 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27247 keeping a reference to a trailing array.
27249 2017-04-27 Richard Biener <rguenther@suse.de>
27251 PR middle-end/80539
27252 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27253 being in loop-closed SSA form conservatively.
27254 (chrec_fold_multiply_poly_poly): Likewise.
27256 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27258 PR middle-end/79665
27259 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27260 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27262 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27265 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27266 (aarch64_function_arg_alignment): Return unsigned int again, but still
27267 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27268 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27269 Don't emit -Wpsabi note.
27270 (aarch64_function_arg_boundary): Likewise.
27271 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27274 2017-04-26 Nathan Sidwell <nathan@acm.org>
27276 * tree.h (crc32_unsigned_n): Declare.
27277 (crc32_unsigned, crc32_unsigned): Make inline.
27278 * tree.c (crc32_unsigned_bits): Replace with ...
27279 (crc32_unsigned_n): ... this.
27280 (crc32_unsigned, crc32_byte): Remove.
27281 (crc32_string): Remove unnecessary braces.
27283 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27285 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27286 * ipa-inline-analysis.c (MAX_TIME): Remove.
27287 (account_size_time): Use sreal for time.
27288 (dump_inline_summary): Update.
27289 (estimate_function_body_sizes): Update.
27290 (estimate_edge_size_and_time): Update.
27291 (estimate_calls_size_and_time): Update.
27292 (estimate_node_size_and_time): Update.
27293 (inline_merge_summary): Update.
27294 (inline_update_overall_summary): Update.
27295 (estimate_time_after_inlining): Update.
27296 (inline_read_section): Update.
27297 (inline_write_summary): Update.
27298 * ipa-inline.c (compute_uninlined_call_time): Update.
27299 (compute_inlined_call_time): Update.
27300 (recursive_inlining): Update.
27301 (inline_small_functions): Update.
27302 (dump_overall_stats): Update.
27303 * ipa-inline.h: Include sreal.h.
27304 (size_time_entry): Turn time to sreal.
27305 (inline_summary): Turn self_time nad time to sreal.
27307 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27309 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27311 (sreal::stream_out, sreal::stream_in): New.
27312 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27314 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27316 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27319 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27322 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27324 (dimode_scalar_chain::compute_convert_gain): Ditto.
27325 (dimode_scalar_chain::make_vector_copies): Ditto.
27326 (dimode_scalar_chain::convert_reg): Ditto.
27327 (dimode_scalar_chain::convert_insn): Ditto.
27328 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27329 (VI248_AVX512BW_1): New mode iterator.
27330 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27331 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27334 2017-04-25 Martin Sebor <msebor@redhat.com>
27336 PR tree-optimization/80497
27337 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27338 constants are representable in HOST_WIDE_INT.
27339 (parse_directive): Ditto.
27341 2017-04-25 Martin Sebor <msebor@redhat.com>
27344 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27345 (new_zero_array): Adjust signature.
27346 (dom_info::dom_init): Used unsigned rather that size_t.
27347 (dom_info::dom_info): Same.
27349 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27350 Jakub Jelinek <jakub@redhat.com>
27353 * config/arm/arm.c: Include gimple.h.
27354 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27355 returns negative, increment ncrn only if it returned positive.
27356 (arm_needs_doubleword_align): Return int instead of bool,
27357 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27358 members, but if there is any such non-FIELD_DECL
27359 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27360 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27361 returns negative, increment nregs only if it returned positive.
27362 (arm_setup_incoming_varargs): Likewise.
27363 (arm_function_arg_boundary): Emit -Wpsabi note if
27364 arm_needs_doubleword_align returns negative, return
27365 DOUBLEWORD_ALIGNMENT only if it returned positive.
27367 2017-04-25 Marek Polacek <polacek@redhat.com>
27370 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27371 first argument to type.
27373 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27376 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27377 type checks to test for compatibility instead of equality.
27379 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27380 Jakub Jelinek <jakub@redhat.com>
27383 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27385 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27386 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27387 the alignment computation, but return their maximum in warn_alignment.
27388 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27389 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27391 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27392 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27395 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27397 * config/arc/simdext.md (dmpyh): Fix typo.
27399 2017-04-25 Richard Biener <rguenther@suse.de>
27401 PR tree-optimization/80492
27402 * alias.c (compare_base_decls): Handle registers with asm
27403 specification conservatively.
27404 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27405 compare_base_decls returning dont-know properly.
27407 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27409 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27410 (legitimate_offset_address_p): New function.
27411 (arc_legitimate_address_p): Use above function.
27413 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27415 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27417 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27419 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27420 ACCH registers whenever they are available.
27422 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27424 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27425 double regs fix when not used.
27427 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27429 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27431 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27432 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27434 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27436 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27437 when emitting short ADD instructions.
27439 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27441 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27443 (cmpsi_cc_c_insn): Likewise.
27444 (cbranchsi4_scratch): Compute proper instruction length using
27445 compact_hreg_operand.
27446 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27448 2017-04-25 Richard Biener <rguenther@suse.de>
27450 PR middle-end/80509
27451 * passes.c (pass_manager::pass_manager): Initialize
27452 m_name_to_pass_map.
27454 2017-04-25 Richard Biener <rguenther@suse.de>
27456 PR tree-optimization/79201
27457 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27459 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27462 * config/s390/vector.md: Split MEM->GPR vector moves for
27463 non-s_operand addresses.
27465 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27468 * config/s390/predicates.md (reload_const_wide_int_operand): New
27470 * config/s390/s390.md ("movti"): Remove d/P alternative.
27471 ("movti_bigconst"): New pattern definition.
27473 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27476 * s390-protos.h (s390_expand_cs_hqi): Removed.
27477 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27478 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27479 modes as well as CCZ1mode and CCZmode.
27480 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27481 signature of s390_emit_compare_and_swap.
27482 (s390_expand_cs_hqi): Likewise, make static.
27483 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27484 compare-and-swap, in some cases.
27485 (s390_expand_cs): Wrapper function.
27486 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27488 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27489 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27490 patterns for small and large integers. Forbid symref memory operands.
27491 Move expander to s390.c. Require cc register.
27492 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27493 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27494 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27495 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27496 symref memory operands. Remove CC mode and call s390_match_ccmode
27498 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27500 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27502 * config/s390/s390.md (define_peephole2): New peephole to help
27503 combining the load-and-test pattern with volatile memory.
27505 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27507 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27508 with CCZmode for TARGET_Z196.
27510 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27512 PR rtl-optimization/80501
27513 * combine.c (make_compound_operation_int): Set subreg_code to SET
27514 even for AND with mask of the sign bit of mode.
27516 PR rtl-optimization/80500
27517 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27518 sum's initial value.
27520 2017-04-25 Julian Brown <julian@codesourcery.com>
27521 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27523 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27525 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27527 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27529 2017-04-25 Julian Brown <julian@codesourcery.com>
27530 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27532 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27533 (thunderx2t99_sha): New Reservation.
27535 2017-04-25 Julian Brown <julian@codesourcery.com>
27536 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27538 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27539 type for 1-element load.
27541 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27543 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27545 2017-04-24 Martin Jambor <mjambor@suse.cz>
27547 PR tree-optimization/80293
27548 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27549 char arrays not totally scalarizable if it is false.
27550 (analyze_all_variable_accesses): Pass correct value in the new
27551 parameter. Add a statistics counter.
27553 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27555 PR middle-end/79931
27556 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27558 2017-04-24 Richard Biener <rguenther@suse.de>
27560 PR tree-optimization/80494
27561 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27562 out for complex types.
27564 2017-04-24 Richard Biener <rguenther@suse.de>
27566 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27567 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27568 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27569 (DFS): Adjust and never fail.
27570 (sccvn_dom_walker::fail): Remove.
27571 (sccvn_dom_walker::before_dom_children): Adjust.
27572 (run_scc_vn): Likewise and never fail.
27573 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27574 (pass_fre::execute): Likewise.
27576 2017-04-24 Richard Biener <rguenther@suse.de>
27578 PR tree-optimization/79725
27579 * tree-ssa-sink.c (statement_sink_location): Return whether
27580 failure reason was zero uses. Move that check later.
27581 (sink_code_in_bb): Deal with zero uses by removing the stmt
27584 2017-04-24 Richard Biener <rguenther@suse.de>
27587 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27588 pointer-based references.
27590 2017-04-24 Richard Biener <rguenther@suse.de>
27593 * pass_manager.h (pass_manager::operator new): Remove.
27594 (pass_manager::operator delete): Likewise.
27595 * passes.c (pass_manager::operator new): Remove.
27596 (pass_manager::operator delete): Likewise.
27597 (pass_manager::pass_manager): Zero individual pass members.
27599 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27602 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27603 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27604 Check "XEXP (src, 1)" operand here.
27605 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27606 Check "XEXP (src, 1)" operand here.
27607 (dimode_scalar_chain::make_vector_copies): Detect count register
27608 of a shift instruction. Zero extend count register from QImode
27609 to DImode to satisfy vector shift pattern count operand predicate.
27610 Substitute vector shift count operand with a DImode copy.
27611 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27614 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27616 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27617 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27618 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27619 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27620 (UNSPEC_NOREX_MEM): Remove definition.
27622 2017-04-21 Richard Biener <rguenther@suse.de>
27624 PR tree-optimization/79547
27625 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27626 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27627 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27628 without any constraints.
27630 2017-04-21 Richard Biener <rguenther@suse.de>
27632 PR tree-optimization/78847
27633 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27635 2017-04-21 Richard Biener <rguenther@suse.de>
27637 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27638 (build_distinct_type_copy): Likewise.
27639 (build_variant_type_copy): Likewise.
27640 * tree.c (build_qualified_type): Pass down mem-stat info.
27641 (build_distinct_type_copy): Likewise.
27642 (build_variant_type_copy): Likewise.
27644 2017-04-21 Richard Biener <rguenther@suse.de>
27646 PR tree-optimization/80237
27647 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27649 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27650 for a simplified result.
27652 2016-04-21 Richard Biener <rguenther@suse.de>
27654 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27655 sth as strict as a simple_iv but a chrec without symbols and an
27656 operand defined in the loop we are peeling (and not some subloop).
27657 (propagate_constants_for_unrolling): Propagate all constants.
27659 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27662 * config/i386/i386.c (print_reg): Remove assert for disalowed
27663 regno values, call output_operand_lossage instead.
27665 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27668 * config/i386/constraints.md (Yc): New register constraint.
27669 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27670 Use Yc constraint for alternative 2 of operand 0. Remove
27671 preferred_for_speed attribute.
27673 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27675 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27676 lastprivate clauses in SIMT case.
27678 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27680 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27682 2017-04-20 Richard Biener <rguenther@suse.de>
27684 PR tree-optimization/57796
27685 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27686 as N scalar stores.
27687 (vect_model_load_cost): Cost gathers as N scalar loads.
27689 2017-04-20 Richard Biener <rguenther@suse.de>
27691 * ggc-page.c (ggc_allocated_p): Rename to ...
27692 (safe_lookup_page_table_entry): ... this and return the lookup
27694 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27696 2017-04-20 Richard Biener <rguenther@suse.de>
27698 PR tree-optimization/80453
27699 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27700 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27701 from the conditions.
27702 (vn_phi_eq): Pass them down.
27703 (vn_phi_lookup): Record them.
27704 (vn_phi_insert): Likewise.
27706 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27708 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27709 uninitialized variable warning to avoid buffer overrun.
27711 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27714 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27715 is suppressed for '{ 0 }' in C.
27717 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27719 * BASE-VER: Set to 8.0.0.
27721 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27723 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27724 priority .init_array and .fini_array section with SECTION_NOTYPE
27727 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27729 PR middle-end/80423
27730 * tree.h (build_array_type): Add typeless_storage default argument.
27731 * tree.c (type_cache_hasher::equal): Also compare
27732 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27733 (build_array_type): Add typeless_storage argument, set
27734 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27736 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27737 (build_array_type): Likewise. Add typeless_storage argument.
27739 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27740 Jakub Jelinek <jakub@redhat.com>
27742 PR tree-optimization/80426
27743 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27744 operation on symbolic operands, also compute the overflow for the
27745 invariant part when the operation degenerates into a negation.
27747 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27750 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27751 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27754 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27756 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27759 * config/avr/avr.c (tree.h): Include it.
27760 (cgraph.h): Include it.
27761 (avr_encode_section_info): Don't warn for uninitialized progmem
27762 variable if it's just an alias.
27764 2017-04-19 Richard Biener <rguenther@suse.de>
27767 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27768 when needed by AutoPGO.
27770 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27773 * doc/lto.texi: Remove an extra 'that'.
27775 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27777 PR rtl-optimization/80429
27778 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27779 are only used in debug insns.
27781 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27782 Vladimir Makarov <vmakarov@redhat.com>
27784 * config/sparc/predicates.md (input_operand): Add comment. Return
27785 true for any memory operand when LRA is in progress.
27786 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27788 2017-04-18 Jeff Law <law@redhat.com>
27791 * mips.md ({return,simple_return}_internal): Do not overwrite
27794 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27796 PR tree-optimization/80443
27797 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27798 instead of adding 1, subtract -1 and similarly instead of subtracting
27801 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27803 PR rtl-optimization/80357
27804 * haifa-sched.c (tmp_bitmap): New variable.
27805 (model_recompute): Handle duplicate use records.
27806 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27807 (free_global_sched_pressure_data): Free it.
27809 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27812 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27813 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27814 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27815 instead of SYSTEM_HEADER_DIR.
27817 2017-04-18 Jeff Law <law@redhat.com>
27819 PR middle-end/80422
27820 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27821 predecessors after walking up the insn chain.
27823 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27826 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27827 sizetype type into debug info.
27829 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27832 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27833 unneeded test for TARGET_UPPER_REGS_SF.
27834 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27836 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27839 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27840 instead of gsi_after_labels.
27842 2017-04-18 Jeff Law <law@redhat.com>
27844 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27848 2017-04-13 Jeff Law <law@redhat.com>
27849 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27850 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27852 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27855 * config/avr/avr.c (intl.h): Include it.
27856 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27858 2017-04-18 Martin Liska <mliska@suse.cz>
27860 PR gcov-profile/78783
27861 * gcov-tool.c (gcov_output_files): Validate that destination
27862 file is either removed by the tool or by a user.
27864 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27865 Guy Benyei <guybe@mellanox.com>
27867 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27868 block, and do not negate it, the stored id is already negative.
27870 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27872 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27874 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27877 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27878 masks of options that should be turned off if the VSX vector
27879 options are turned off.
27880 (OTHER_P8_VECTOR_MASKS): Likewise.
27881 (OTHER_VSX_VECTOR_MASKS): Likewise.
27882 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27883 rs6000_disable_incompatible_switches to validate no type switches
27885 (rs6000_incompatible_switch): New function to disallow turning on
27886 other vector options if -mno-vsx, -mno-power8-vector, or
27887 -mno-power9-vector are specified.
27889 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27891 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27893 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27895 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27896 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27897 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27898 (ARG_POINTER_CFA_OFFSET): Likewise.
27900 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27902 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27903 conditions to take advantage of various optimizations.
27905 2017-04-13 Jeff Law <law@redhat.com>
27907 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27908 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27909 (zero_extendsidi2_dext): Likewise.
27911 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27914 * fold-const.c (fold_ternary_loc): Revert
27915 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27918 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27920 PR rtl-optimization/80343
27921 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27922 new scratch pseudo.
27924 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27927 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27928 to ubsan_encode_value.
27930 2017-04-13 Jeff Law <law@redhat.com>
27932 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27933 appearing in DEBUG_INSNs.
27935 2017-04-13 Martin Liska <mliska@suse.cz>
27937 PR gcov-profile/80413
27938 * gcov-io.c (gcov_write_string): Copy to buffer just when
27939 allocated size is greater than zero.
27941 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27944 * dwarf2out.c (decls_for_scope): Ignore declarations of
27945 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27947 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27950 * ipa-visibility.c (non_local_p): Fix typos.
27951 (localize_node): When localizing symbol in same comdat group,
27952 dissolve the group only when we know external symbols are going
27954 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27956 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27958 PR tree-optimization/79390
27959 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27960 order does not result in usable sequence, retry with reversed operand
27966 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27967 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27968 op0 instead of fold_convert_loc (loc, type, arg0).
27970 2017-04-12 Jeff Law <law@redhat.com>
27972 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27973 has a delay slot in the generated code.
27975 * config/cris/cris.md (cris_preferred_reload_class): Return
27976 GENNONACR_REGS rather than GENERAL_REGS.
27978 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27981 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27982 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27983 signedness of the result type.
27985 2017-04-12 Richard Biener <rguenther@suse.de>
27986 Jeff Law <law@redhat.com>
27988 PR tree-optimization/80359
27989 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27990 trim stores to TARGET_MEM_REFs.
27992 2017-04-12 Richard Biener <rguenther@suse.de>
27994 PR tree-optimization/79390
27995 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27996 threading case even more.
27998 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28001 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28002 for quad_address_p for TImode, instead of just not indexed_address.
28004 2017-04-12 Richard Biener <rguenther@suse.de>
28005 Bernd Edlinger <bernd.edlinger@hotmail.de>
28007 PR middle-end/79671
28008 * alias.c (component_uses_parent_alias_set_from): Handle
28009 TYPE_TYPELESS_STORAGE.
28010 (get_alias_set): Likewise.
28011 * tree-core.h (tree_type_common): Add typeless_storage flag.
28012 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28013 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28014 for types containing members with TYPE_TYPELESS_STORAGE.
28015 (place_field): Likewise.
28016 (layout_type): Likewise for ARRAY_TYPE.
28017 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28018 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28019 TYPE_TYPELESS_STORAGE.
28020 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28022 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28025 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28026 first argument to type.
28028 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28032 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28033 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28034 (rs6000_expand_binop_builtin): Likewise.
28035 (rs6000_expand_ternop_builtin): Likewise; also add missing
28036 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28037 vshasigma built-ins.
28038 * doc/extend.texi: Document that vec_xxpermdi's third argument
28039 must be a constant.
28041 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28043 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28044 Use shift_const cost parameter when calculating gain of STV shifts.
28046 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28048 PR rtl-optimization/70478
28049 * lra-constraints.c (process_alt_operands): Check memory for
28050 disfavoring memory insn operand.
28052 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28054 PR middle-end/80100
28055 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28056 left shift in unsigned HOST_WIDE_INT type.
28058 PR rtl-optimization/80385
28059 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28060 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28063 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28064 if they have any depend clauses.
28066 2017-04-11 Martin Liska <mliska@suse.cz>
28069 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28070 * ipa-split.c (split_function): Create a local comdat symbol
28071 if caller is in a comdat group.
28073 2017-04-11 Martin Liska <mliska@suse.cz>
28076 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28079 2017-04-11 Martin Sebor <msebor@redhat.com>
28081 PR middle-end/80364
28082 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28083 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28085 (directive::set_width, directive::set_precision, format_character):
28087 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28090 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28093 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28094 conflict, set target->arch_name instead of target->cpu_name.
28096 2017-04-11 Richard Biener <rguenther@suse.de>
28098 PR tree-optimization/80374
28099 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28100 build_zero_cst, remove fold_convertible_p check again.
28102 2017-04-11 Martin Liska <mliska@suse.cz>
28105 * ubsan.c (instrument_object_size): Do not instrument register
28108 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28111 * config/i386/i386-builtin-types.def
28112 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28113 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28114 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28115 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28116 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28117 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28118 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28119 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28120 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28121 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28122 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28123 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28124 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28125 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28126 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28127 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28128 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28129 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28130 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28131 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28132 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28133 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28134 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28135 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28136 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28137 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28138 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28139 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28140 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28141 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28142 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28143 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28144 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28145 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28146 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28147 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28148 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28149 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28150 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28151 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28152 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28153 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28154 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28155 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28156 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28157 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28159 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28160 flag to second_arg_count, handle 4 argument function type _COUNT
28161 aliases, handle second_arg_count on second argument rather than last.
28163 2017-04-10 Jeff Law <law@redhat.com>
28165 PR tree-optimization/80374
28166 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28167 record anything if we can not convert integer_zero_node to the
28170 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28173 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28174 Enhance special handling given to the TARGET_P9_MINMAX option in
28175 relation to certain other options.
28177 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28179 PR tree-optimization/80153
28180 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28181 remove POINTER_PLUS_EXPR's base part directly, rather than through
28184 2017-04-10 Richard Biener <rguenther@suse.de>
28185 Bin Cheng <bin.cheng@arm.com>
28187 PR tree-optimization/80153
28188 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28189 the first element of pointer type aff_tree. Build result expr in
28191 (add_elt_to_tree): Convert to type unconditionally. Remove other
28192 fold_convert calls.
28193 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28194 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28196 2017-04-10 Richard Biener <rguenther@suse.de>
28198 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28201 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28203 PR rtl-optimization/70478
28204 * lra-constraints.c (curr_small_class_check): New.
28205 (update_and_check_small_class_inputs): New.
28206 (process_alt_operands): Update curr_small_class_check. Disfavor
28207 alternative insn memory operands. Check available regs for small
28210 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28213 * config/mips/mips.opt (-mvirt): Update description.
28214 * doc/invoke.texi (-mvirt): Likewise.
28216 2017-04-10 Richard Biener <rguenther@suse.de>
28218 PR middle-end/80362
28219 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28220 looking for NEGATE_EXPR in -A / -B to A / B folding.
28222 2017-04-10 Martin Liska <mliska@suse.cz>
28224 PR gcov-profile/80224
28225 * gcov.c (print_usage): Fix usage string.
28226 (get_gcov_intermediate_filename): Remove.
28227 (output_gcov_file): Use both for normal and intermediate format.
28228 (generate_results): Do not initialize special file for
28229 intermediate format.
28231 2017-04-10 Richard Biener <rguenther@suse.de>
28233 PR tree-optimization/80304
28234 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28237 2017-04-10 Nathan Sidwell <nathan@acm.org>
28240 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28241 (rs6000_init_builtins): Use it.
28243 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28245 * config/arm/arm.md (<mrc>): Add mode to SET source.
28246 (<mrrc>): Likewise.
28248 2017-04-10 Richard Biener <rguenther@suse.de>
28250 PR middle-end/80344
28251 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28253 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28256 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28257 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28258 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28259 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28260 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28261 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28262 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28263 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28264 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28265 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28266 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28267 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28268 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28269 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28270 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28271 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28272 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28273 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28274 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28275 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28276 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28277 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28278 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28280 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28282 PR rtl-optimization/70478
28283 * lra-constraints.c: Reverse the last patch.
28285 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28287 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28288 Add comment for WCHAR_T.
28290 2017-04-08 Martin Liska <mliska@suse.cz>
28293 2017-04-07 Martin Liska <mliska@suse.cz>
28296 * ipa-split.c (split_function): Add function part to a same comdat
28299 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28302 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28304 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28306 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28307 (vec_pairing): ...to this.
28308 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28309 (rs6000_sched_init): Adjust for name change.
28310 (struct rs6000_sched_context): Likewise.
28311 (rs6000_init_sched_context): Likewise.
28312 (rs6000_set_sched_context): Likewise.
28314 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28320 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28322 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28323 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28324 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28326 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28328 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28330 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28332 PR rtl-optimization/70703
28333 * ira-color.c (update_conflict_hard_regno_costs): Use
28334 int64_t instead of HOST_WIDE_INT.
28336 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28338 PR rtl-optimization/70478
28339 * lra-constraints.c (process_alt_operands): Disfavor alternative
28340 insn memory operands.
28342 2017-04-07 Jeff Law <law@redhat.com>
28344 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28345 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28347 2017-04-07 Martin Liska <mliska@suse.cz>
28350 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28351 Show error message instead of an ICE.
28353 2017-04-07 Martin Liska <mliska@suse.cz>
28356 * ipa-split.c (split_function): Add function part to a same comdat
28359 2017-04-07 Richard Biener <rguenther@suse.de>
28361 PR middle-end/80341
28362 * tree.c (get_unwidened): Also handle ! for_type case for
28364 * convert.c (do_narrow): Split out from ...
28365 (convert_to_integer_1): ... here. Do not pass final truncation
28366 type to get_unwidened for TRUNC_DIV_EXPR.
28368 2017-04-07 Richard Biener <rguenther@suse.de>
28370 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28372 (aff_combination_const): Adjust.
28373 (aff_combination_scale): Likewise.
28374 (aff_combination_add_elt): Likewise.
28375 (aff_combination_add_cst): Likewise.
28376 (aff_combination_convert): Likewise.
28377 (add_elt_to_tree): Likewise. Remove unused argument.
28378 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28380 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28382 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28384 * config/arm/arm.c (arm_default_short_enums): Use
28385 ARM_DEFAULT_SHORT_ENUMS.
28386 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28388 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28391 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28392 members with redundant out-of-class redeclaration.
28394 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28397 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28398 * config/i386/i386.md (*zero_extendsidi2):
28399 Add (?*x,*x) and (?*v,*v) alternatives.
28401 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28404 * config/i386/i386.c (ix86_expand_builtin)
28405 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28406 mode from insn data. Convert operands to insn operand mode.
28407 Copy operands that don't satisfy insn predicate to a register.
28409 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28411 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28414 2017-04-06 Richard Biener <rguenther@suse.de>
28416 PR tree-optimization/80334
28417 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28418 preserve alignment of accesses.
28420 2017-04-06 Richard Biener <rguenther@suse.de>
28422 PR tree-optimization/80262
28423 * tree-sra.c (build_ref_for_offset): Preserve address-space
28425 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28426 Drop useless address-space information on MEM_REF offsets.
28428 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28430 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28432 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28434 PR rtl-optimization/70703
28435 * ira-color.c (update_conflict_hard_regno_costs): Use
28436 HOST_WIDE_INT instead of long.
28438 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28441 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28442 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28444 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28445 for x86_64 target. Handle -m3dnowa option.
28447 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28449 PR rtl-optimization/70703
28450 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28451 (update_conflict_hard_regno_costs): Use long instead of unsigned
28452 arithmetic for cost calculation.
28454 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28455 Bernd Edlinger <bernd.edlinger@hotmail.de>
28458 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28461 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28464 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28465 ptr_mode with Pmode throughout.
28466 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28467 into probe_stack_range and use DImode.
28469 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28472 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28473 call_eh_return is true.
28475 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28477 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28478 Initialize last_match_fntype_index.
28480 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28483 * tree-nvr.c: Include internal-fn.h.
28484 (pass_return_slot::execute): Ignore internal calls without
28487 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28488 Richard Biener <rguenther@suse.de>
28491 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28492 captures used multiple times, except for the last use.
28493 * generic-match-head.c: Include gimplify.h.
28495 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28497 PR tree-optimization/79390
28498 * target.h (struct noce_if_info): Declare.
28499 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28500 * target.def (noce_conversion_profitable_p): New target hook.
28501 * ifcvt.h (struct noce_if_info): New type, moved from ...
28502 * ifcvt.c (struct noce_if_info): ... here.
28503 (noce_conversion_profitable_p): Renamed to ...
28504 (default_noce_conversion_profitable_p): ... this. No longer
28506 (noce_try_store_flag_constants, noce_try_addcc,
28507 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28508 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28509 instead of noce_conversion_profitable_p.
28510 * config/i386/i386.c: Include ifcvt.h.
28511 (ix86_option_override_internal): Don't override
28512 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28513 (ix86_noce_conversion_profitable_p): New function.
28514 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28515 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28516 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28517 * doc/tm.texi: Regenerated.
28519 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28521 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28524 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28527 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28528 instructions for small multiply cores.
28530 2017-04-04 Jeff Law <law@redhat.com>
28532 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28534 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28535 that are not set by the loop over the elements.
28537 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28540 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28541 int mode, convert_modes it to mode as unsigned, otherwise use
28542 lowpart_subreg to mode rather than SImode.
28543 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28544 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28545 Use DImode instead of SImode for the shift count operand.
28546 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28549 2017-04-04 Richard Biener <rguenther@suse.de>
28551 PR middle-end/80281
28552 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28553 arithmetic done for the negate or the plus. Simplify.
28554 (A - (-B) -> A + B): Likewise.
28555 * fold-const.c (split_tree): Make sure to not negate pointers.
28557 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28559 PR rtl-optimization/60818
28560 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28561 a compare of comparisons with the thing compared if this results
28562 in a different machine mode.
28564 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28566 * alias.c (base_alias_check): Fix typo in comment.
28567 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28568 * cgraphunit.c (symbol_table::compile): Likewise.
28569 * collect2.c (maybe_run_lto_and_relink): Likewise.
28570 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28571 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28572 * config/avr/avr.c (avr_map_op_t): Likewise.
28573 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28574 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28575 * config/epiphany/epiphany.md (movcc): Likewise.
28576 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28577 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28579 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28580 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28581 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28582 * config/sh/sh.c (sh_rtx_costs): Likewise.
28583 * fold-const.c (fold_truth_andor): Likewise.
28584 * genautomata.c (collapse_flag): Likewise.
28585 * gengtype.h (struct type::u::s): Likewise.
28586 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28587 * input.c (FORMAT_AMOUNT): Likewise.
28588 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28589 (known_aggs_to_agg_replacement_list): Likewise.
28590 * ipa-inline-analysis.c: Likewise.
28591 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28592 * ipa-polymorphic-call.c
28593 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28594 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28595 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28597 * modulo-sched.c (apply_reg_moves): Likewise.
28598 * omp-expand.c (build_omp_regions_1): Likewise.
28599 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28600 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28601 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28602 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28603 * value-prof.c: Likewise.
28604 * var-tracking.c (val_reset): Likewise.
28606 2017-04-03 Richard Biener <rguenther@suse.de>
28608 PR tree-optimization/80275
28609 * fold-const.c (split_address_to_core_and_offset): Handle
28612 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28614 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28615 descriptors is at least equal to that of functions.
28617 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28619 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28621 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28624 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28625 (mov<IMOD4:mode>): New expander.
28626 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28628 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28630 PR rtl-optimization/79405
28631 * fwprop.c (propagations_left): New variable.
28632 (forward_propagate_into): Decrement it.
28633 (fwprop_init): Initialize it.
28634 (fw_prop): If the variable has reached zero, stop propagating.
28635 (fwprop_addr): Ditto.
28637 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28640 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28641 a FUNCTION_DECL, pass it as decl instead of origin to
28644 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28646 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28649 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28652 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28653 TARGET_VSX_SMALL_INTEGER.
28655 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28657 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28658 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28660 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28662 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28663 extraction from odd-numbered MSA register.
28665 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28667 PR middle-end/80173
28668 * expmed.c (store_bit_field_1): Don't attempt to create
28669 a word subreg out of hard registers wider than word if they
28670 have HARD_REGNO_NREGS of 1 for their mode.
28672 PR middle-end/80163
28673 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28674 conversions to integer types wider than word and pointer.
28677 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28678 (rtx_equal_for_cselib_p): Pass 0 to it.
28679 * cselib.c (cselib_hasher::equal): Likewise.
28680 (rtx_equal_for_cselib_1): Add depth argument. If depth
28681 is 128, don't look up VALUE locs and punt. Increment
28682 depth in recursive calls when walking VALUE locs.
28684 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28686 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28687 (make_gcov_file_name): Use the canonical path name for generating
28689 (read_line): Fix handling of files with ascii null bytes.
28691 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28693 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28694 to initialise a vector register instead
28695 of using a const_int.
28697 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28699 PR translation/80189
28700 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28701 diagnostic messages.
28703 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28706 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28707 (dfp_diex_<mode>): Update mode of operand 1.
28708 * doc/extend.texi (dxex, dxexq): Document change to return type.
28709 (diex, diexq): Document change to argument type.
28711 2017-03-30 Martin Jambor <mjambor@suse.cz>
28714 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28715 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28716 it reflects the signature changes performed at the callee side.
28717 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28718 to cgraph_build_function_type_skip_args.
28719 (build_function_decl_skip_args): Adjust call to the above function.
28721 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28724 * config/i386/sse.md
28725 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28726 register as dest whenever it is a MEM not rtx_equal_p to the
28727 corresponding dup operand, and when forcing into reg move the
28728 reg into the memory afterwards.
28729 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28730 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28731 for the force_reg mode.
28732 (avx512vl_vextractf128<mode>): Use register as dest either
28733 always when a MEM, or when it is a MEM not rtx_equal_p to the
28734 corresponding dup operand, or even not when it is a CONST_VECTOR
28735 depending on the mode and lo vs. hi.
28736 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28738 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28739 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28740 Likewise. Require that operands[2] is even.
28741 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28742 Remove extraneous parens. Require that operands[2] is a multiple
28744 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28745 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28746 disallow memory then.
28748 2017-03-30 Richard Biener <rguenther@suse.de>
28750 PR tree-optimization/77498
28751 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28752 to non-constants over backedges.
28754 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28756 PR rtl-optimization/80233
28757 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28758 as last_combined_insn. Do not test for BARRIER_P separately.
28760 2017-03-29 Andreas Schwab <schwab@suse.de>
28763 * calls.c (prepare_call_address): Convert funexp to Pmode before
28764 copying to temp reg.
28766 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28768 PR tree-optimization/80158
28769 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28770 Handle possible future case of more than one alternate
28772 (replace_rhs_if_not_dup): Likewise.
28773 (replace_one_candidate): Likewise.
28775 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28777 PR rtl-optimization/80193
28778 * ira.c (ira): Do not check allocation for LRA.
28780 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28782 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28783 (nvptx_output_simt_exit): Declare.
28784 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28785 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28786 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28787 (nvptx_declare_function_name): ...here. Emit declaration of local
28788 memory space buffer for omp_simt_enter insn.
28789 (nvptx_output_unisimt_switch): New.
28790 (nvptx_output_softstack_switch): New.
28791 (nvptx_output_simt_enter): New.
28792 (nvptx_output_simt_exit): New.
28793 * config/nvptx/nvptx.h (struct machine_function): New fields
28794 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28795 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28796 (UNSPECV_SIMT_EXIT): Ditto.
28797 (omp_simt_enter_insn): New insn.
28798 (omp_simt_enter): New expansion.
28799 (omp_simt_exit): New insn.
28800 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28802 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28803 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28804 (expand_GOMP_SIMT_EXIT): New.
28805 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28806 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28807 (GOMP_SIMT_EXIT): Ditto.
28808 * target-insns.def (omp_simt_enter): New insn.
28809 (omp_simt_exit): Ditto.
28810 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28812 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28813 (lower_rec_input_clauses): Likewise.
28814 (lower_lastprivate_clauses): Handle SIMT privatization.
28816 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28817 (ompdevlow_adjust_simt_enter): New.
28818 (find_simtpriv_var_op): New.
28819 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28820 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28822 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28823 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28824 (copy_decl_for_dup_finish): Ditto.
28826 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28828 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28831 * config/i386/i386.c (ix86_option_override_internal): Always
28832 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28834 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28836 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28838 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28840 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28841 mark new edge's irreducible flag accordign to it.
28842 (vect_do_peeling): Check loop preheader edge's irreducible flag
28843 and pass it to function slpeel_add_loop_guard.
28845 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28847 PR tree-optimization/80218
28848 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28849 Update block frequencies and counts.
28851 2017-03-28 Richard Biener <rguenther@suse.de>
28853 PR tree-optimization/78644
28854 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28855 of a simplification result we may not use it at all.
28857 2017-03-28 Richard Biener <rguenther@suse.de>
28860 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28861 without arguments, generate default definition of a SSA name.
28863 2017-03-28 Richard Biener <rguenther@suse.de>
28865 PR middle-end/80222
28866 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28867 TYPE_REF_CAN_ALIAS_ALL references.
28868 * fold-const.c (fold_indirect_ref_1): Likewise.
28870 2017-03-28 Martin Liska <mliska@suse.cz>
28873 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28874 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28876 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28877 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28879 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28880 (EXTRA_SPECS): Define.
28881 (SUBTARGET_EXTRA_SPECS): Likewise.
28882 (SUBTARGET_CPP_SPEC): Likewise.
28883 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28884 SUBTARGET_EXTRA_SPECS.
28885 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28887 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28889 * config/arc/simdext.md (vst64_insn): Update pattern.
28890 (vld32wh_insn): Likewise.
28891 (vld32wl_insn): Likewise.
28892 (vld64_insn): Likewise.
28893 (vld32_insn): Likewise.
28895 2017-03-28 Marek Polacek <polacek@redhat.com>
28898 * fold-const.c (fold_comparison): Use protected_set_expr_location
28899 instead of SET_EXPR_LOCATION.
28901 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28903 * tree.c (add_expr): Avoid name lookup warning.
28905 2017-03-27 Jeff Law <law@redhat.com>
28907 PR tree-optimization/80216
28908 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28909 function name. Limit recursion depth.
28910 (record_temporary_equivalences): Corresponding changes.
28912 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28914 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28917 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28920 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28922 * cfgcleanup.c (reg_note_cfa_p): New array.
28923 (insns_have_identical_cfa_notes): New function.
28924 (old_insns_match_p): Don't cross-jump in between /f
28925 and non-/f instructions. If both i1 and i2 are frame related,
28926 verify all CFA notes, their order and content.
28928 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28931 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28932 HImode and SImode with zero extend to DImode to one insn.
28933 (bswap<mode>2_extenddi): Likewise.
28934 (bswapsi2_extenddi): Likewise.
28935 (bswaphi2_extendsi): Likewise.
28936 (bswaphi2): Combine bswap HImode and SImode into one insn.
28937 Separate memory insns from swapping register.
28938 (bswapsi2): Likewise.
28939 (bswap<mode>2): Likewise.
28940 (bswaphi2_internal): Delete, no longer used.
28941 (bswapsi2_internal): Likewise.
28942 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28943 store, and gpr<-gpr swap insns.
28944 (bswap<mode>2_store): Likewise.
28945 (bswaphi2_reg): Register only splitter, combine with the splitter.
28946 (bswaphi2 splitter): Likewise.
28947 (bswapsi2_reg): Likewise.
28948 (bswapsi2 splitter): Likewise.
28949 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28950 the insns into load, store, and register/register insns.
28951 (bswapdi2_ldbrx): Likewise.
28952 (bswapdi2_load): Likewise.
28953 (bswapdi2_store): Likewise.
28954 (bswapdi2_reg): Likewise.
28956 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28958 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28959 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28961 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28964 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28966 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28967 special handling for target option conflicts between dform
28968 options (-mpower9-dform, -mpower9-dform-vector,
28969 -mpower9-dform-scalar) and -mno-direct-move.
28971 2017-03-27 Richard Biener <rguenther@suse.de>
28973 PR tree-optimization/80181
28974 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28976 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28978 * config/arc/predicates.md (move_double_src_operand): Replace the
28979 call to move_double_src_operand with a call to address_operand.
28981 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28983 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28984 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28985 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28987 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28989 * config/arc/predicates.md (long_immediate_loadstore_operand):
28990 Consider scaled addresses cases.
28992 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28994 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28995 restored when in interrupt.
28996 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28997 doesn't have delay slot.
28999 2017-03-27 Richard Biener <rguenther@suse.de>
29002 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29003 inlined thunk clones.
29005 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29008 * asan.c (instrument_derefs): Copy over last operand from
29009 original COMPONENT_REF to the new COMPONENT_REF with
29010 DECL_BIT_FIELD_REPRESENTATIVE.
29011 * ubsan.c (instrument_object_size): Likewise.
29013 2017-03-27 Richard Biener <rguenther@suse.de>
29015 PR tree-optimization/80170
29016 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29017 sure DR/SCEV didnt fold in constants we do not see when looking
29018 at the reference base alignment.
29020 2017-03-27 Richard Biener <rguenther@suse.de>
29022 PR middle-end/80171
29023 * gimple-fold.c (fold_ctor_reference): Properly guard against
29024 NULL return value from canonicalize_constructor_val.
29026 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29029 * config/i386/i386.c (ix86_expand_builtin)
29030 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29031 flags reg setting and flags reg using instructions.
29032 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29033 clobbering instructions to zero extend op2.
29035 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29037 * doc/install.texi (Configuration) <--with-aix-soname>:
29038 Update link to AIX ld.
29040 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29042 PR rtl-optimization/80160
29043 PR rtl-optimization/80159
29044 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29045 reg_alternate_class into account.
29047 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29050 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29051 to consider in curr_insn_transform.
29053 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29055 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29056 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29057 and emit_mode_inner.
29059 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29061 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29062 argument to the overloaded builtin variants. Use the new flag to
29063 deprecate certain builtin variants.
29064 * config/s390/s390-builtin-types.def: Add new builtin types.
29065 * config/s390/s390-builtins.h: Support new flags field for
29066 overloaded builtins.
29067 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29068 (s390_macro_to_expand): Enable vector float data type.
29069 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29070 builtins by incrementing the __VEC__ version number.
29071 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29073 (s390_resolve_overloaded_builtin): Emit error messages depending
29074 on the builtin flags.
29075 * config/s390/s390.c (s390_expand_builtin): Support additional
29076 flags argument. Change error message to match the messages
29077 emitted in s390-c.c.
29078 * config/s390/s390.md: New UNSPEC_* constants.
29079 (op_type): Add new instruction types.
29080 * config/s390/vecintrin.h: Add new builtins and test data class
29082 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29083 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29084 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29085 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29086 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29087 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29089 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29090 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29091 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29092 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29094 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29095 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29096 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29097 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29099 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29100 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29101 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29102 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29103 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29104 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29105 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29107 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29108 ("vec_scatter_element<V_HW_4:mode>_DI")
29109 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29110 ("vec_fpint<mode>", "vflls")
29111 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29112 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29113 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29114 ("*vec_cmphe<mode>_cc"): ... these.
29116 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29117 mode constant instead of magic value.
29119 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29121 * config/s390/s390.c (s390_expand_vec_compare): Support other
29122 vector floating point modes than just V2DF.
29123 (s390_expand_vcond): Likewise.
29124 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29125 (s390_cannot_change_mode_class): Prevent mode changes between TF
29126 and V1TF in vector registers.
29127 * config/s390/s390.md (DF, SF): New mode attributes.
29128 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29129 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29130 SFmode support for VRs.
29131 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29133 (VFT, VF_HW): New mode iterators.
29134 (vw, sdx): New mode attributes.
29135 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29136 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29137 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29138 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29139 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29140 also the new vector floating point modes. Renaming to ...
29142 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29143 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29144 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29145 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29146 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29147 ("vec_unordered<mode>"): ... these.
29149 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29150 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29151 ("*vec_extendv2df"): New insn definitions.
29153 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29155 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29156 ("mulditi3_2", "*muldi3_sign"): New patterns.
29157 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29158 rename the pattern definition.
29160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29162 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29164 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29166 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29168 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29169 instruction if possible.
29170 * config/s390/vector.md (vec_halfnumelts): New mode
29172 ("*vec_vllezlf<mode>"): New pattern.
29174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29176 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29177 ("popcountv4si2", "popcountv2di2"): Rename to ...
29178 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29179 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29181 ("popcount<mode>2_vxe"): New pattern.
29183 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29185 * common/config/s390/s390-common.c (processor_flags_table): Add
29187 * config.gcc: Add arch12.
29188 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29189 Default to arch12 for unknown CPU model numbers.
29190 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29191 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29192 PROCESSOR_max sanity check.
29193 * config/s390/s390-opts.h (enum processor_type): Add
29195 * config/s390/s390.c (processor_table): Add arch12.
29196 (s390_expand_builtin): Add check for B_VXE flag.
29197 (s390_issue_rate): Add PROCESSOR_ARCH12.
29198 (s390_get_sched_attrmask): Likewise.
29199 (s390_get_unit_mask): Likewise.
29200 (s390_sched_score): Enable z13 scheduling for arch12.
29201 (s390_sched_reorder): Likewise.
29202 (s390_sched_variable_issue): Likewise.
29203 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29205 (s390_tune_attr): Use z13 scheduling also for arch12.
29206 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29207 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29208 (TARGET_VXE_P): New macros.
29209 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29210 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29211 * config/s390/s390.opt: Add arch12 as processor_type.
29213 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29215 * config/s390/s390.md
29216 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29217 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29218 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29220 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29221 Rename expanders to ...
29223 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29224 ("fixuns_truncdddi2_emu"): ... these.
29226 ("fixuns_trunc<mode>si2_emu"): New expander.
29228 ("*fixuns_truncdfdi2_z13"): Rename to ...
29229 ("*fixuns_truncdfdi2_vx"): ... this.
29231 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29233 * config/s390/2964.md: Remove the single element vector compare
29234 instructions which are no longer used.
29235 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29237 (s390_canonicalize_comparison): Remove handling of DFmode
29239 (s390_expand_vec_compare_scalar): Remove function.
29240 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29241 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29243 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29245 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29247 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29249 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29250 will anyway by matched by mov<mode>_64dfp.
29252 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29254 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29255 vlef/vstef. Add missing operand to vleif.
29257 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29259 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29260 pair for all vector types with 64 bit elements.
29261 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29262 * config/s390/vector.md (V_HW_64): ... here.
29263 (V_128_NOSINGLE): New mode iterator.
29264 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29265 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29266 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29267 ("*vec_load_pairv2di"): Change to ...
29268 ("*vec_load_pair<mode>"): ... this one.
29270 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29272 * config/s390/constraints.md: Add comments.
29273 (jKK): Reject element sizes > 8 bytes.
29274 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29276 * config/s390/s390.md: Add the s_operand checks formerly in
29277 s390_split_ok_p to various splitters where they are still
29279 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29280 for 128 bit vectors. Plus two splitters.
29282 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29284 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29290 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29291 error if the boundary argument is not constant.
29293 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29295 PR rtl-optimization/80112
29296 * loop-doloop.c (doloop_condition_get): Don't check condition
29297 if cmp isn't SET with IF_THEN_ELSE src.
29299 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29301 PR tree-optimization/80158
29302 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29303 replacing a candidate statement, also replace it for the
29304 candidate's alternate interpretation.
29305 (replace_rhs_if_not_dup): Likewise.
29306 (replace_one_candidate): Likewise.
29308 2017-03-24 Richard Biener <rguenther@suse.de>
29310 PR tree-optimization/80167
29311 * graphite-isl-ast-to-gimple.c
29312 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29314 (translate_isl_ast_to_gimple::get_rename): Likewise.
29316 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29318 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29319 handling of certain combinations of target options, including the
29320 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29321 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29323 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29326 * config/arm/arm.md (*load_multiple): Add reload_completed to
29327 matching condition.
29329 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29330 Richard Biener <rguenth@suse.de>
29332 PR tree-optimization/79908
29333 PR tree-optimization/80136
29334 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29335 been cast away, gimplify_and_add suffices.
29337 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29339 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29341 2017-03-23 Richard Biener <rguenther@suse.de>
29343 PR tree-optimization/80032
29344 * gimplify.c (gimple_push_cleanup): Forced unconditional
29345 cleanups still have to go to the conditional_cleanups
29348 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29350 PR tree-optimization/80072
29351 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29353 (next_operand_entry_id): Change type to unsigned int.
29354 (sort_by_operand_rank): Make sure to return the right return value
29355 even if unsigned fields are bigger than INT_MAX.
29356 (struct oecount): Change cnt and id type to unsigned int.
29357 (oecount_hasher::equal): Formatting fix.
29358 (oecount_cmp): Make sure to return the right return value
29359 even if unsigned fields are bigger than INT_MAX.
29360 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29363 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29364 TREE_READONLY on result if writing it more than once.
29367 * doc/invoke.texi (-fsanitize=thread): Document that with
29368 -fnon-call-exceptions atomics are not able to throw
29372 * tsan.c: Include tree-eh.h.
29373 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29374 maybe_clean_or_replace_eh_stmt where needed.
29375 (instrument_memory_accesses): Add cfg_changed argument.
29376 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29377 if it returned true.
29378 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29380 PR rtl-optimization/63191
29381 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29382 wrapper function, moved the whole old content into ...
29383 (ix86_delegitimize_address_1): ... this. New inline function.
29384 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29385 true as last argument instead of ix86_delegitimize_address.
29387 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29389 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29390 cortexa57_branch_cost.
29392 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29394 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29396 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29399 * doc/md.texi (Constraints): Document wA constraint.
29400 * config/rs6000/constraints.md (wA): New.
29401 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29402 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29403 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29404 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29406 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29409 * gimplify.c (is_oacc_declared): New function.
29410 (oacc_default_clause): Use it to set default flags for acc declared
29411 variables inside parallel regions.
29412 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29413 declared variables.
29414 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29415 declare attribute to any decl as necessary.
29417 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29420 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29421 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29422 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29423 (arm_arch_lpae): This.
29424 * config/arm/arm.c (arm_arch7ve): Rename into ...
29425 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29426 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29429 2017-03-22 Martin Liska <mliska@suse.cz>
29432 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29433 error message instead of an ICE.
29435 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29437 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29439 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29441 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29443 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29446 2017-03-21 Martin Sebor <msebor@redhat.com>
29448 * doc/extend.texi: Use "cannot" instead of "can't."
29449 * doc/hostconfig.texi: Same.
29450 * doc/install.texi: Same.
29451 * doc/invoke.texi: Same.
29452 * doc/loop.texi: Same.
29453 * doc/md.texi: Same.
29454 * doc/objc.texi: Same.
29455 * doc/rtl.texi: Same.
29456 * doc/tm.texi: Same.
29457 * doc/tm.texi.in: Same.
29458 * doc/trouble.texi: Same.
29460 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29463 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29464 (collect_checksum_attributes): Set it.
29465 (die_checksum_ordered): Use it.
29467 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29469 PR tree-optimization/79908
29470 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29471 change: For a VA_ARG whose LHS has been cast away, use
29472 force_gimple_operand to construct the side effects.
29474 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29476 PR translation/80001
29477 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29478 more amenable to translation.
29479 (oacc_loop_auto_partitions): Likewise.
29481 2017-03-21 Marek Polacek <polacek@redhat.com>
29482 Martin Sebor <msebor@redhat.com>
29484 PR tree-optimization/80109
29485 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29486 on INTEGRAL_TYPE_P.
29488 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29489 Segher Boessenkool <segher@kernel.crashing.org>
29492 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29493 check reg_used_between_p between insn and one of succ or succ2
29494 depending on if succ is artificial insn not inserted into insn
29497 2017-03-21 Martin Liska <mliska@suse.cz>
29499 PR gcov-profile/80081
29500 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29501 * doc/gcc.texi: Include gcov-dump stuff.
29502 * doc/gcov-dump.texi: New file.
29504 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29506 PR rtl-optimization/79150
29507 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29508 conditional jump, if the jump is the last insn of the loop.
29510 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29511 Richard Biener <rguenth@suse.de>
29513 PR tree-optimization/79908
29514 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29515 been cast away, use force_gimple_operand to construct the side
29518 2017-03-21 Martin Liska <mliska@suse.cz>
29520 PR libfortran/79956
29521 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29524 2017-03-21 Brad Spengler <spender@grsecurity.net>
29527 * plugin.c (htab_hash_plugin): New function.
29528 (add_new_plugin): Use it and adjust.
29529 (parse_plugin_arg_opt): Adjust.
29530 (init_one_plugin): Likewise.
29532 2017-03-21 Richard Biener <rguenther@suse.de>
29534 PR tree-optimization/80032
29535 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29536 if set force the cleanup to happen unconditionally.
29537 (gimplify_target_expr): Push inserted clobbers with force_uncond
29538 to avoid them being removed by control-dependent DCE.
29540 2017-03-21 Richard Biener <rguenther@suse.de>
29542 PR tree-optimization/80122
29543 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29544 va_arg_pack_len when the inlined call stmt requires pack
29546 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29548 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29551 * tsan.c (instrument_builtin_call): If the memory model argument
29552 is not a constant, assume it is valid.
29555 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29558 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29560 PR rtl-optimization/79910
29561 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29562 if its dest is used by an insn before I2 (other than the combined
29563 insns themselves, which are properly handled already).
29565 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29568 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29570 * combine.c (record_used_regs): New static function.
29571 (try_combine): Handle situations where there is an additional
29572 instruction between I2 and I3 which needs to have a LOG_LINK
29576 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29578 * combine.c (try_combine): Delete redundant i1 test. Call
29579 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29581 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29584 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29585 alternatives 13/14.
29587 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29589 PR tree-optimization/80054
29590 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29591 the optimization if a PHI or any of its arguments is not dominated
29592 by the candidate's basis. Use gphi* rather than gimple* as
29594 (replace_profitable_candidates): Clean up a gimple* variable that
29595 should be a gphi* variable.
29597 2017-03-20 Martin Sebor <msebor@redhat.com>
29600 * doc/extend.texi (attribute constructor): Document present limitation.
29602 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29605 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29606 __POWER9_VECTOR__ #ifdef control, change template definition to
29607 use Power9-specific built-in function.
29608 (vec_any_eq): Likewise.
29609 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29610 to control outcomes from this test.
29611 (vector_ae_<mode>p): For VEC_F modes, likewise.
29613 2017-03-20 Ian Lance Taylor <iant@google.com>
29615 * config/i386/i386.c (ix86_function_regparm): Save an extra
29616 register for -fsplit-stack with DECL_STATIC_CHAIN.
29618 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29621 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29622 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29624 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29626 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29628 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29631 2017-03-20 Marek Polacek <polacek@redhat.com>
29634 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29636 2017-03-20 Richard Biener <rguenther@suse.de>
29638 PR tree-optimization/80113
29639 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29640 allocate extra SSA name for PHI def.
29641 (add_close_phis_to_outer_loops): Likewise.
29642 (add_close_phis_to_merge_points): Likewise.
29643 (copy_loop_close_phi_args): Likewise.
29644 (copy_cond_phi_nodes): Likewise.
29646 2017-03-20 Martin Liska <mliska@suse.cz>
29648 PR middle-end/79753
29649 * tree-chkp.c (chkp_build_returned_bound): Do not build
29650 returned bounds for a LHS that's not a BOUNDED_P type.
29652 2017-03-20 Martin Liska <mliska@suse.cz>
29656 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29657 COMPLEX_CST and VECTOR_CST.
29659 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29662 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29663 target operand. A new splitter adds the clobber statement in case
29664 the target operand is dead anyway.
29666 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29668 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29669 to age-old versions of binutils and glibc.
29671 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29673 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29675 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29677 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29679 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29681 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29682 requirement for binutils 2.13.
29684 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29686 * combine.c (try_combine): Delete redundant i1 test. Call
29687 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29689 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29691 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29692 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29694 <riscv64-*-elf>: Re-arrange section
29695 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29696 <riscv32-*-linux>: Likewise.
29697 <riscv64-*-elf>: Likewise
29698 <riscv64-*-linux>: Likewise.
29700 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29703 * aarch64.opt(verbose-cost-dump): Fix typo.
29705 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29708 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29709 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29711 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29713 * reload.c (find_reloads): When reloading a nonoffsettable address,
29714 use RELOAD_OTHER for it and its address reloads.
29716 PR rtl-optimization/79910
29717 * combine.c (record_used_regs): New static function.
29718 (try_combine): Handle situations where there is an additional
29719 instruction between I2 and I3 which needs to have a LOG_LINK
29722 2017-03-17 Jeff Law <law@redhat.com>
29724 PR tree-optimization/71437
29725 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29726 conditional in the hash table first.
29727 (vrp_dom_walker::before_dom_children): Extract condition from
29728 ASSERT_EXPR. Record condition, its inverion and any implied
29729 conditions as well.
29731 2017-03-17 Marek Polacek <polacek@redhat.com>
29732 Markus Trippelsdorf <markus@trippelsdorf.de>
29734 PR tree-optimization/80079
29735 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29738 2017-03-17 Richard Biener <rguenther@suse.de>
29740 PR middle-end/80075
29741 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29742 Properly verify the LHS before the RHS possibly claims to be
29744 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29747 2017-03-17 Martin Jambor <mjambor@suse.cz>
29749 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29750 (List of -O2 options): Likewise.
29751 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29754 2017-03-17 Tom de Vries <tom@codesourcery.com>
29756 * gcov-dump.c (print_usage): Print bug_report_url.
29758 2017-03-17 Richard Biener <rguenther@suse.de>
29760 PR middle-end/80050
29761 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29762 (parser::peek): Likewise.
29764 2017-03-17 Richard Biener <rguenther@suse.de>
29766 PR tree-optimization/80048
29767 * sese.c (free_sese_info): Properly release rename_map and
29768 copied_bb_map elements.
29770 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29772 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29773 Add linked-list forward and backlinks. Insert on
29774 construction, remove on destruction.
29775 (class pass_store_merging): Add m_stores_head field.
29776 (pass_store_merging::terminate_and_process_all_chains):
29777 Iterate over m_stores_head list.
29778 (pass_store_merging::terminate_all_aliasing_chains):
29780 (pass_store_merging::execute): Check for debug stmts first.
29781 Push new chains onto the m_stores_head stack.
29783 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29786 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29787 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29788 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29790 2017-03-16 Jeff Law <law@redhat.com>
29792 PR tree-optimization/71437
29793 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29794 member function. Implementation moved into after_dom_children
29795 member function and into the threader's thread_outgoing_edges
29797 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29798 some code into new thread_outgoing_edges.
29799 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29800 definition. Simplify marker handling (do it here). Assume we always
29801 have the available expression and the const/copies tables.
29802 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29804 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29805 * tree-vrp.c (equiv_stack): No longer file scoped.
29806 (vrp_dom_walker): New class.
29807 (vrp_dom_walker::before_dom_children): New member function.
29808 (vrp_dom_walker::after_dom_children): Likewise.
29809 (identify_jump_threads): Setup domwalker. Use it rather than
29810 walking edges in a random order by hand. Simplify setup/finalization.
29811 (finalize_jump_threads): Remove.
29812 (vrp_finalize): Do not call identify_jump_threads here.
29813 (execute_vrp): Do it here instead and call thread_through_all_blocks
29816 PR tree-optimization/71437
29817 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29819 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29821 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29822 (dom_opt_dom_walker::thread_across_edge): Remove
29823 handle_dominating_asserts argument. All callers changed.
29824 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29825 changes. Remove calls to lhs_of_dominating_assert. Other
29826 uses of handle_dominating_asserts turn into unconditional code
29827 (simplify_control_stmt_condition_1): Likewise.
29828 (simplify_control_stmt_condition): Likewise.
29829 (thread_through_normal_block, thread_across_edge): Likewise.
29830 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29831 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29832 object if it is not an SSA_NAME.
29833 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29834 before calling into the VRP specific simplifiers.
29835 (identify_jump_threads): Remove handle_dominating_asserts
29838 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29841 * tree-diagnostic.c (default_tree_printer): No longer static.
29842 * tree-diagnostic.h (default_tree_printer): New prototype.
29844 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29846 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29847 Change ins into fmov.
29849 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29851 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29852 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29853 Use h_con constraint for operand 1.
29854 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29855 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29857 2017-03-15 Jeff Law <law@redhat.com>
29859 PR tree-optimization/71437
29860 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29861 (record_temporary_equivalences): Use it.
29863 PR tree-optimization/71437
29864 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29865 tree-ssa-scopedtables.
29866 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29867 (record_conditions, record_cond, vuse_eq): Likewise.
29868 (record_edge_info): Adjust to API tweak of record_conditions.
29869 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29870 (record_temporary_equivalences, optimize_stmt): Likewise.
29871 (eliminate_redundant_computations): Likewise.
29872 (record_equivalences_from_stmt): Likewise.
29873 * tree-ssa-scopedtables.c: Include options.h and params.h.
29874 (vuse_eq): New function, moved from tree-ssa-dom.c
29875 (build_and_record_new_cond): Likewise.
29876 (record_conditions): Likewise. Accept vector of conditions rather
29877 than edge_equivalence structure for first argument.
29878 for the first argument.
29879 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29880 from tree-ssa-dom.c.
29881 (avail_exprs_stack::record_cond): Likewise.
29882 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29883 from tree-ssa-dom.c.
29884 (avail_exprs_stack): Add new member functions lookup_avail_expr
29886 (record_conditions): Declare.
29888 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29891 * lra-constraints.c (process_alt_operands): Increase reject for
29892 reloading an input/output operand.
29894 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29897 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29898 insns to convert from signed/unsigned char/short to IEEE 128-bit
29900 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29902 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29905 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29906 subreg of inner mode for values already in registers.
29908 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29910 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29911 iteration reg is used after the loop.
29913 2017-03-14 Martin Sebor <msebor@redhat.com>
29915 PR tree-optimization/79800
29916 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29917 precision in negative-positive range.
29918 (format_floating): Call non-const overload with adjusted precision.
29920 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29923 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29926 2017-03-14 Martin Sebor <msebor@redhat.com>
29928 PR middle-end/80020
29929 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29930 * builtins.def (aligned_alloc): Use it.
29933 * Makefile.in (GTFILES): Add calls.c.
29934 * calls.c: Include "gt-calls.h".
29936 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29938 PR rtl-optimization/79728
29939 * regs.h (struct target_regs): New field
29940 x_contains_allocatable_regs_of_mode.
29941 (contains_allocatable_regs_of_mode): New macro.
29942 * reginfo.c (init_reg_sets_1): Initialize it, and change
29943 contains_reg_of_mode so it includes global regs as well.
29944 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29945 rather than contains_regs_of_mode.
29947 2017-03-14 Martin Liska <mliska@suse.cz>
29949 * doc/invoke.texi: Document options that can't be combined with
29950 -fcheck-pointer-bounds.
29952 2017-03-14 Martin Liska <mliska@suse.cz>
29954 PR middle-end/79831
29955 * doc/invoke.texi (-Wchkp): Document the option.
29957 2017-03-14 Martin Liska <mliska@suse.cz>
29959 * Makefile.in: Install gcov-dump.
29961 2017-03-14 Martin Liska <mliska@suse.cz>
29963 * multiple_target.c (expand_target_clones): Bail out for
29964 an invalid attribute.
29966 2017-03-14 Richard Biener <rguenther@suse.de>
29968 * alias.c (struct alias_set_entry): Pack properly.
29969 * cfgloop.h (struct loop): Likewise.
29970 * cse.c (struct set): Likewise.
29971 * ipa-utils.c (struct searchc_env): Likewise.
29972 * loop-invariant.c (struct invariant): Likewise.
29973 * lra-remat.c (struct cand): Likewise.
29974 * recog.c (struct change_t): Likewise.
29975 * rtl.h (struct address_info): Likewise.
29976 * symbol-summary.h (function_summary): Likewise.
29977 * tree-loop-distribution.c (struct partition): Likewise.
29978 * tree-object-size.c (struct object_size_info): Likewise.
29979 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29980 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29981 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29982 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29983 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29984 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29985 (struct _stmt_vec_info): Likewise.
29987 2017-03-14 Martin Liska <mliska@suse.cz>
29990 * multiple_target.c (create_dispatcher_calls): Check that
29991 a target can create a function dispatcher.
29993 2017-03-14 Martin Liska <mliska@suse.cz>
29996 * multiple_target.c (expand_target_clones): Drop local.local
29997 flag for default implementation.
29999 2017-03-14 Richard Biener <rguenther@suse.de>
30001 PR tree-optimization/80030
30002 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30004 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30006 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30007 gcc_fallthrough() instead of __attribute__((fallthrough));
30009 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30011 * doc/gcc.texi: Remove "up" link to (DIR).
30012 * doc/gccint.texi: Ditto.
30014 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30016 * doc/install.texi (Specific) <avr>: Remove reference to
30019 2017-03-13 Jeff Law <law@redhat.com>
30021 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30022 attribute rather than comments.
30024 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30025 match_scratch operand is highest.
30027 2017-03-13 Martin Liska <mliska@suse.cz>
30029 PR middle-end/78339
30030 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30031 is a CHKP clone, use original declaration.
30033 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30035 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30036 (arc_conditional_register_usage): Use a different allocation order
30037 when optimizing for size.
30038 * common/config/arc/arc-common.c (arc_option_optimization_table):
30039 Section anchors default on when optimizing for size.
30041 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30043 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30045 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30047 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30048 * config/arc/arc.md (cpu_facility): Add cd variant.
30049 (*movqi_insn): Add code density variant.
30050 (*movhi_insn): Likewise.
30051 (*movqi_insn): Likewise.
30052 (*addsi3_mixed): Likewise.
30053 (subsi3_insn): Likewise.
30055 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30057 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30059 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30061 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30062 expressions with MINUS and UNARY ops.
30064 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30067 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30069 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30070 between vec_select and vector argument.
30071 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30072 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30073 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30074 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30075 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30076 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30078 2017-03-13 Richard Biener <rguenther@suse.de>
30081 * params.def (vect-max-peeling-for-alignment): Fix typo.
30083 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30085 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30086 issue that only occurred with binutils below 2.18.
30088 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30090 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30091 refer to binutils 2.11/2.12 minimum.
30093 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30095 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30096 ftp.kernel.org and simplify binutils requirement.
30098 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30100 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30102 (Optimize Options): Ditto. Also remove redundancy.
30104 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30106 PR translation/79848
30107 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30109 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30110 to G_ to avoid double translation.
30112 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30114 PR translation/79923
30115 * auto-profile.c (get_combined_location): Convert leading
30116 character of diagnostics to lower case and remove trailing period.
30117 (read_profile): Likewise for various diagnostics.
30118 * config/arm/arm.c (arm_option_override): Remove trailing period
30119 from various diagnostics.
30120 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30121 (msp430_expand_delay_cycles): Likewise.
30123 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30126 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30127 full command-line argument, rather than just "str".
30128 (aarch64_validate_march): Likewise.
30129 (aarch64_validate_mtune): Likewise.
30131 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30133 PR rtl-optimization/78911
30134 * lra-assigns.c (must_not_spill_p): New function.
30135 (spill_for): Use it.
30137 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30139 PR tree-optimization/79981
30140 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30141 ATOMIC_COMPARE_EXCHANGE ifn result.
30142 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30143 IFN_ATOMIC_COMPARE_EXCHANGE.
30145 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30148 * opts.c (parse_sanitizer_options): Add missing question mark to
30149 "did you mean" message.
30151 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30153 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30155 (VMULEUH_UNS): Likewise.
30156 (VMULOUB_UNS): Likewise.
30157 (VMULOUH_UNS): Likewise.
30158 * config/rs6000/rs6000.c (builtin_function_type): Remove
30159 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30161 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30164 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30165 x with result of extra_parsing_for_operand_code_0.
30166 (function_reader::extra_parsing_for_operand_code_0): Convert
30167 return type from void to rtx, returning x. When reading
30168 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30169 larger size containing struct block_symbol.
30171 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30173 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30174 -mfloat128-hardware without -m64.
30176 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30179 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30180 entries to the case statement that marks unsigned arguments to
30181 overloaded functions.
30183 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30185 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30186 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30188 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30191 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30192 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30194 2017-03-10 Martin Liska <mliska@suse.cz>
30198 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30199 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30202 2017-03-10 Olivier Hainque <hainque@adacore.com>
30204 * tree-switch-conversion (array_value_type): Start by resetting
30205 candidate type to it's main variant.
30207 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30209 PR rtl-optimization/79909
30210 * combine.c (try_combine): Use simplify_replace_rtx on individual
30211 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30212 of the whole CALL_INSN_FUNCTION_USAGE.
30214 PR tree-optimization/79972
30215 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30216 get_range_info on SSA_NAMEs. Formatting fixes.
30218 2017-03-10 Richard Biener <rguenther@suse.de>
30219 Jakub Jelinek <jakub@redhat.com>
30221 PR tree-optimization/77975
30222 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30223 edge to be constant.
30224 (get_val_for): For constant x return it. Formatting fix.
30225 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30226 would use the same bases as the current one.
30228 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30230 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30231 instead of vec_select for V1TImode.
30232 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30234 (VSX_LE_128): Add V1TI to this mode iterator.
30235 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30236 (*vsx_le_perm_store_<mode>): Likewise.
30237 (pre-reload splitter for VSX stores): Likewise.
30238 (post-reload splitter for VSX stores): Likewise.
30239 (*vsx_xxpermdi2_le_<mode>): Likewise.
30240 (*vsx_lxvd2x2_le_<mode>): Likewise.
30241 (*vsx_stxvd2x2_le_<mode>): Likewise.
30243 2017-03-09 Michael Eager <eager@eagercon.com>
30245 Correct failures with --enable-checking=yes,rtl.
30247 * config/microblaze/microblaze.c (microblaze_expand_shift):
30248 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30249 test for const0_rtx.
30250 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30251 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30253 2017-03-09 Richard Biener <rguenther@suse.de>
30255 PR tree-optimization/79977
30256 * graphite-scop-detection.c (scop_detection::merge_sese):
30257 Handle the case of extra exits to blocks dominating the entry.
30259 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30261 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30264 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30266 PR rtl-optimization/79949
30267 * lra-constraints.c (process_alt_operands): Check memory when
30268 trying to predict a cycle. Print about the overall increase.
30270 2017-03-09 Richard Biener <rguenther@suse.de>
30272 PR middle-end/79971
30273 * gimple-expr.c (useless_type_conversion_p): Preserve
30274 TYPE_SATURATING for fixed-point types.
30276 2017-03-09 Richard Biener <rguenther@suse.de>
30279 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30280 alignment of BLKmode params.
30282 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30285 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30286 (VALL_NO_V2Q): Likewise.
30288 * config/aarch64/aarch64-simd.md
30289 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30291 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30292 VALL_NO_V2Q mode iterator.
30293 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30295 2017-03-09 Martin Liska <mliska@suse.cz>
30297 PR tree-optimization/79631
30298 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30299 tree_int_cst_sign_bit just for INTEGER constants.
30301 2017-03-09 Martin Liska <mliska@suse.cz>
30305 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30308 2017-03-09 Marek Polacek <polacek@redhat.com>
30311 * tree.c (inchash::add_expr): Handle TREE_VEC.
30313 2017-03-09 Martin Liska <mliska@suse.cz>
30316 (chkp_narrow_size_and_offset): New function.
30317 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30318 (void chkp_parse_bit_field_ref): New function.
30319 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30320 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30322 2017-03-09 Martin Liska <mliska@suse.cz>
30325 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30326 (chkp_find_bounds_1): Remove gcc_unreachable.
30328 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30331 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30332 BUILT_IN_SYNC*, determine the access type from the size suffix and
30333 always build a MEM_REF with that type. Handle forgotten
30334 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30337 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30338 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30339 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30340 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30341 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30342 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30343 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30344 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30345 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30346 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30347 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30348 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30349 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30350 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30351 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30352 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30353 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30354 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30355 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30356 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30357 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30358 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30359 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30360 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30361 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30362 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30363 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30364 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30365 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30366 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30367 definitions outside of __OPTIMIZE__ guarded section.
30370 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30371 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30372 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30373 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30376 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30378 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30379 ("vfenez<mode>"): Add missing constraints.
30381 2017-03-08 Martin Sebor <msebor@redhat.com>
30384 * config/nds32/nds32.c (nds32_option_override):
30385 Fix misspelled diagnostic.
30387 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30390 * gimplify.c (gimplify_omp_for): Replace index var in outer
30391 taskloop statement with an artificial variable and add
30392 OMP_CLAUSE_PRIVATE clause for it.
30394 2017-03-08 Richard Biener <rguenther@suse.de>
30396 PR tree-optimization/79955
30397 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30398 for accesses that are completely outside of the variable.
30400 2017-03-08 Andrew Haley <aph@redhat.com>
30402 PR tree-optimization/79943
30403 * tree-ssa-loop-split.c (compute_new_first_bound): When
30404 calculating the new upper bound, (END-BEG) should be added, not
30407 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30409 * config/avr/avr.md (setmemhi): Make sure match_dup
30410 operand number comes before match_scratch.
30412 2017-03-08 Richard Biener <rguenther@suse.de>
30414 PR tree-optimization/79920
30415 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30416 with ncopies == 1 to ...
30417 (vect_transform_slp_perm_load): ... here. Properly compute
30418 all element loads by iterating VF times over the group. Do
30419 not handle ncopies (computed in a broken way) in
30420 vect_create_mask_and_perm.
30422 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30425 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30426 is a uniform vector, use uniform_vector_p return value instead of
30427 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30429 2017-03-07 Marek Polacek <polacek@redhat.com>
30431 PR middle-end/79809
30432 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30433 (alloca_call_type): Likewise.
30435 2017-03-07 Martin Liska <mliska@suse.cz>
30437 * gcov.c (process_args): Put comment to correct location.
30439 2017-03-07 Martin Liska <mliska@suse.cz>
30441 PR middle-end/68270
30442 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30443 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30444 (chkp_narrow_bounds_for_field): Likewise.
30445 (chkp_parse_array_and_component_ref): Pass one more argument to
30448 2017-03-07 Richard Biener <rguenther@suse.de>
30450 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30453 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30455 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30456 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30458 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30461 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30462 to end of description.
30463 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30465 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30467 PR rtl-optimization/79901
30468 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30470 (*avx512f_<code><mode>3<mask_name>): ... this.
30471 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30472 iterator instead of VI8_AVX2_AVX512BW.
30474 PR rtl-optimization/79901
30475 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30476 min/max expander, expand it using expand_vec_cond_expr.
30479 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30482 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30485 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30486 to void * for PCH reasons.
30487 * dwarf2out.c (output_loc_operands, output_die): Cast
30488 v.val_vec.array to unsigned char *.
30490 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30493 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30496 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30498 PR rtl-optimization/79571
30499 * lra-constraints.c (process_alt_operands): Calculate static
30500 reject and subtract it from overall when only addresses will be
30503 2017-03-06 Julia Koval <julia.koval@intel.com>
30506 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30507 incoming stack boundary to 128 for 64-bit targets.
30509 2017-03-06 Richard Biener <rguenther@suse.de>
30511 PR tree-optimization/79894
30512 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30513 to NULL after folding it.
30515 2017-03-06 Richard Biener <rguenther@suse.de>
30517 PR tree-optimization/79824
30518 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30519 check disabling peeling for gaps.
30521 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30523 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30524 attributes): Document gettimeofday.
30526 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30528 * config/s390/s390.c (s390_option_override_internal): Set
30529 PARAM_MIN_VECT_LOOP_BOUND
30531 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30533 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30534 * config/s390/s390.md: Likewise.
30536 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30539 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30540 (<avx2_avx512>_perm<mode>): Rename to ...
30541 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30543 (<avx512>_perm<mode>_mask): Rename to ...
30544 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30546 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30547 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30548 instead of VI8F_256_512.
30549 (avx512f_perm<mode>): New define_expand.
30550 (avx512f_perm<mode>_mask): Likewise.
30551 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30552 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30554 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30556 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30557 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30559 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30561 2017-03-06 Martin Liska <mliska@suse.cz>
30564 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30565 when having a SSA NAME w/o VAR_DECL assigned to it.
30567 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30569 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30570 msa_dpsub_<su>_d): Fix MODE for vec_select.
30572 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30574 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30576 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30578 2017-03-06 Richard Biener <rguenther@suse.de>
30580 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30581 * plugin.c (register_plugin_info): Likewise.
30582 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30584 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30586 * config/i386/sse.md (sse_storehps, sse_storelps,
30587 avx_<castmode><avxsizesuffix>_<castmode>,
30588 avx512f_<castmode><avxsizesuffix>_<castmode>,
30589 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30590 in condition that at least one operand is not a MEM.
30592 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30594 PR middle-end/79805
30595 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30596 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30598 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30599 gimple_call_nothrow_p flag based on whether original builtin can throw.
30600 If it can, emit following stmts on the fallthrough edge.
30601 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30602 don't create new bb if inserting just debug stmts on the edge, try to
30603 insert them on the fallthru bb or just reset debug stmts.
30605 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30608 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30609 restore recog_data (including the operand rtxes inside it) around
30610 the call to get_insn_template.
30612 2017-03-03 Martin Sebor <msebor@redhat.com>
30614 PR tree-optimization/79699
30615 * context.c (context::~context): Free MPFR caches to avoid
30616 a memory leak on program exit.
30618 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30620 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30621 Use wide_int::ulow () instead of .elt (0).
30623 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30625 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30626 (*pushxf): Limit oF constraint to 32bit targets and add oC
30627 constraint for 64bit targets.
30628 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30629 (*pushdf): Change rmF constraint to rmC.
30631 2017-03-03 Martin Liska <mliska@suse.cz>
30633 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30634 Remove unused variable.
30636 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30639 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30640 is a memory operand, increase num_memory.
30641 (ix86_expand_args_builtin): Likewise.
30643 2017-03-03 Jan Hubicka <jh@suse.cz>
30646 * ipa-devirt.c (maybe_record_node): Properly handle
30647 __cxa_pure_virtual visibility.
30649 2017-03-03 Martin Liska <mliska@suse.cz>
30651 PR tree-optimization/79803
30652 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30654 (pass_loop_prefetch::execute): Disabled optimization if an
30655 assumption about L1 cache size is not met.
30657 2017-03-03 Martin Liska <mliska@suse.cz>
30659 PR rtl-optimization/79574
30660 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30661 (hash_scan_set): Likewise.
30662 (dump_hash_table): Likewise.
30663 (hoist_code): Likewise.
30665 2017-03-03 Richard Biener <rguenther@suse.de>
30667 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30668 in place of uhigh ().
30669 (fixed_convert_from_real): Likewise.
30671 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30674 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30676 2017-03-03 Richard Biener <rguenther@suse.de>
30678 PR middle-end/79818
30679 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30680 TYPE_OVERFLOW_UNDEFINED check.
30682 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30684 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30686 (vector_ae_<mode>_p): Likewise.
30687 (vector_nez_<mode>_p): Likewise.
30688 (vector_ne_v2di_p): Likewise.
30689 (vector_ae_v2di_p): Likewise.
30690 (vector_ne_<mode>_p): Likewise.
30691 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30693 (vsx_tsqrt<mode>2_fe): Likewise.
30695 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30698 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30700 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30702 PR rtl-optimization/79780
30703 * cprop.c (one_cprop_pass): When second and further conditional trap
30704 in a single basic block is turned into an unconditional trap, turn it
30705 into a deleted note to avoid RTL verification failures.
30707 2017-03-02 Richard Biener <rguenther@suse.de>
30709 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30711 2017-03-02 Richard Biener <rguenther@suse.de>
30713 PR tree-optimization/79345
30715 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30716 param and abort the walk, returning -1 if it is hit.
30717 (walk_aliased_vdefs): Take a limit param and pass it on.
30718 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30719 defaulting to 0 and return a signed int.
30720 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30721 (check_defs): New helper.
30722 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30723 about uninitialized memory.
30724 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30725 bogus uninitialized warning.
30726 (fixed_convert_from_real): Likewise.
30728 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30730 PR tree-optimization/66768
30731 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30732 iv_use if base object can't be determined.
30734 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30736 PR tree-optimization/79345
30737 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30738 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30739 (get_pattern_stats): Initialize it.
30740 * genemit.c (gen_expand): Verify match_scratch numbers come after
30741 match_operand/match_dup numbers.
30742 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30743 match_scratch numbers.
30744 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30746 * config/s390/s390.md (trunctdsd2): Likewise.
30748 2017-03-02 Richard Biener <rguenther@suse.de>
30750 * wide-int.h (wide_int_storage::operator=): Implement in terms
30753 2017-03-02 Richard Biener <rguenther@suse.de>
30755 PR tree-optimization/79777
30756 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30757 the to insert expression to sth existing.
30759 2017-03-01 Martin Sebor <msebor@redhat.com>
30761 PR middle-end/79692
30762 * gimple-ssa-sprintf.c
30763 (directive::known_width_and_precision): New function.
30764 (format_integer): Use it.
30765 (get_mpfr_format_length): Consider the full range of precision
30766 when computing %g output with the # flag. Set the likely byte
30767 count to 3 rather than 1 when precision is indeterminate.
30768 (format_floating): Correct the lower bound of precision.
30770 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30772 * doc/invoke.texi: Document default code model for 64-bit Linux.
30774 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30777 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30778 udiv rather than div since input pattern is unsigned.
30780 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30782 * config/i386/i386.c (print_reg): Warn for values of
30783 unsupported size in integer register.
30785 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30788 * config/rs6000/predicates.md (current_file_function_operand): Do
30789 not allow self calls to be local if the function is replaceable.
30791 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30794 * config/rs6000/altivec.h (vec_ctz and others): Change the
30795 preprocessor macro that controls conditional compilation from
30796 _ARCH_PWR9 to __POWER9_VECTOR__.
30797 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30798 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30799 control (instead of _ARCH_PWR9 control) so that template
30800 definition uses power9-specific function.
30801 (vec_any_eq): Likewise.
30802 (vec_all_ne): Change macro definition to use a power9-specific
30803 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30804 _ARCH_PWR9 control).
30805 (vec_any_eq) Likewise.
30806 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30807 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30808 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30809 support for xvcmpnedp instruction.
30810 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30811 macro expansion so that Power9 implementation of vec_all_ne does
30812 not use the AltiVec predicate framework.
30813 (VCMPNEH_P): Likewise.
30814 (VCMPNEW_P): Likewise.
30815 (VCMPNED_P): Likewise.
30816 (VCMPNEFP_P): Likewise.
30817 (VCMPNEDP_P): Likewise.
30818 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30819 implementation of vec_any_eq to not use AltiVec predicate
30821 (VCMPAEH_P): Likewise.
30822 (VCMPAEW_P): Likewise.
30823 (VCMPAED_P): Likewise.
30824 (VCMPAEFP_P): Likewise.
30825 (VCMPAEDP_P): Likewise.
30826 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30827 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30828 not use the AltiVec predicate framework.
30829 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30830 of vec_any_eq to not use AltiVec predicate framework.
30831 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30832 support for predefined __POWER9_VECTOR__ macro to indicate that
30833 Power9 instruction selection is enabled.
30834 (altivec_overloaded_builtins): Remove extraneous
30835 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30836 function argument types RS6000_BTI_bool_V16QI and
30837 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30838 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30839 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30840 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30841 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30842 Power9 for implementations of vec_cmpne. Change the signature for
30843 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30844 (representing vec_all_ne) to remove the previously described first
30845 argument of type RS6000_BTI_INTSI, as this was an artifact of
30846 reliance on the AltiVec predicate framework, which is no longer
30847 used in the implementation of these functions. Add
30848 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30849 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30850 since, unlike the AltiVec predicate framework implementation, we
30851 do not share function descriptors between vec_alle and vec_anyeq.
30852 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30853 set of modes that receive special treatment even when
30854 TARGET_P9_VECTOR is true. The special treatment emits code that
30855 does not depend on Power9 instructions.
30856 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30857 define_expand to not rely on AltiVec predicate framework.
30858 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30860 (vector_ne_v2di_p): Change this define_expand to not rely on
30861 AltiVec predicate framework.
30862 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30864 (vector_ne_<mode>_p): Change this define_expand to not rely on
30865 AltiVec predicate framework.
30866 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30868 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30869 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30870 define_insn pattern.
30871 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30872 define_insn pattern because the xvcmpne<VSs>. instruction is not
30874 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30875 instruction is not supported.
30877 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30879 * config/nvptx/nvptx.c: Include intl.h.
30881 2017-03-01 Martin Jambor <mjambor@suse.cz>
30884 * ipa-prop.h (ipa_bits): Removed field known.
30885 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30886 to pointers. Adjusted their comments to warn about their sharing.
30887 (ipcp_transformation_summary): Change bits to a vector of pointers.
30888 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30889 (ipa_get_ipa_bits_for_value): Declare.
30890 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30891 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30892 (ipa_bits_hash_table): Likewise.
30893 (ipa_vr_ggc_hash_traits): Likewise.
30894 (ipa_vr_hash_table): Likewise.
30895 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30896 being pointers and vr_known being removed.
30897 (ipa_set_jf_unknown): Likewise.
30898 (ipa_get_ipa_bits_for_value): New function.
30899 (ipa_set_jfunc_bits): Likewise.
30900 (ipa_get_value_range): New overloaded functions.
30901 (ipa_set_jfunc_vr): Likewise.
30902 (ipa_compute_jump_functions_for_edge): Use the above functions to
30903 construct bits and vr parts of jump functions.
30904 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30905 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30907 (ipcp_grow_transformations_if_necessary): Also allocate
30908 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30910 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30911 them. Fix too long lines.
30912 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30913 vr_known being removed.
30914 (ipa_read_jump_function): Use new setter functions to construct bits
30915 and vr parts of jump functions or set them to NULL.
30916 (write_ipcp_transformation_info): Adjust for bits being pointers.
30917 (read_ipcp_transformation_info): Likewise.
30918 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30920 Include gt-ipa-prop.h.
30921 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30923 (ipcp_store_bits_results): Likewise.
30924 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30925 Do not write to existing jump functions but use a temporary instead.
30927 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30930 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30931 attempt to use its first operand as BIT_FIELD_REF base.
30933 2017-03-01 Richard Biener <rguenther@suse.de>
30935 PR middle-end/79721
30936 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30937 interpolating formula in wrapping arithmetic.
30938 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30940 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30942 PR tree-optimization/79734
30943 * tree-vect-generic.c (expand_vector_condition): Optimize
30944 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30945 Handle VEC_COND_EXPR where comparison has different inner width from
30946 type's inner width.
30948 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30950 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30951 markup, and similar issues. Remove @opindex entries for things
30952 that aren't options. Add missing -mmpy-option entries.
30954 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30956 PR tree-optimization/79737
30957 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30958 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30959 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30960 instead of byte_size. Formatting fix.
30961 (shift_bytes_in_array_right): Formatting fix.
30963 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30966 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30967 condition on optimize for the leaf function test.
30969 2017-02-28 Martin Liska <mliska@suse.cz>
30972 * read-rtl-function.c (function_reader::handle_unknown_directive):
30973 Bail out when one uses -flto.
30975 2017-02-28 Martin Liska <mliska@suse.cz>
30977 * common.opt: Replace space with tabular for options of <number>
30979 * config/i386/i386.opt: Show <number> value for
30980 -mlarge-data-threshold.
30981 * opts.c (print_filtered_help): Do not display number in hexadecimal
30984 2017-02-28 Martin Liska <mliska@suse.cz>
30986 * common.opt: Fix --help=option -Q for options which are of
30989 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30991 * config/i386/i386.c (print_reg): Error out for values
30992 of 8-bit size in invalid integer register.
30994 2017-02-28 Martin Sebor <msebor@redhat.com>
30996 PR tree-optimization/79691
30997 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30999 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31002 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31003 gcc_unreachable with output_operand_lossage.
31005 2017-02-28 Richard Biener <rguenther@suse.de>
31007 PR tree-optimization/79740
31008 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31010 (visit_nary_op): Insert the nary into the hashtable if we
31011 pattern-matched sth.
31012 * tree-ssa-pre.c (eliminate_insert): Robustify.
31014 2017-02-28 Richard Biener <rguenther@suse.de>
31016 PR middle-end/79731
31017 * fold-const.c (decode_field_reference): Reject out-of-bound
31020 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31022 * config/i386/i386.c: Include intl.h.
31023 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31024 instead of just cond ? "..." : "...".
31025 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31026 * coverage.c (read_counts_file): Likewise.
31027 * omp-offload.c: Include intl.h.
31028 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31029 of just cond ? "..." : "...".
31030 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31031 of just cond ? "..." : "...".
31033 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31036 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31038 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31039 'tune for' CPU name.
31040 * config/arm/arm-cpu-data.h: Regenerated.
31042 2017-02-28 Richard Biener <rguenther@suse.de>
31044 PR tree-optimization/79732
31045 * tree-inline.c (expand_call_inline): Do not shadow var.
31047 2017-02-28 Richard Biener <rguenther@suse.de>
31049 PR tree-optimization/79723
31050 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31051 address-space properly.
31053 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31055 * doc/optinfo.texi (Optimization groups): Fix option used for
31057 * doc/invoke.texi (-fopt-info): Document "omp".
31058 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31059 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31060 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31061 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31062 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31064 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31066 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31067 OPTGROUP_OMP with "omp".
31069 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31072 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31073 for arithmetic shift of unsigned V2DI.
31075 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31077 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31078 arc/linux.h headers.
31079 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31080 (LINK_SPEC): Likewise.
31081 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31082 (EXTRA_SPECS): Likewise.
31083 (STARTFILE_SPEC): Likewise.
31084 (ENDFILE_SPEC): Likewise.
31085 (LIB_SPEC): Likewise.
31086 (TARGET_SDATA_DEFAULT): Likewise.
31087 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31088 (MULTILIB_DEFAULTS): Likewise.
31089 (DWARF2_UNWIND_INFO): Likewise.
31090 * config/arc/big.h: New file.
31091 * config/arc/elf.h: Likewise.
31092 * config/arc/linux.h: Likewise.
31093 * config/arc/t-uClibc: Remove.
31095 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31097 PR tree-optimization/77536
31098 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31099 (tree_transform_and_unroll_loop): Use above function to compute the
31100 estimated niter of unrolled loop and use it when scaling profile.
31101 Also use count info rather than frequency if it's non-zero.
31102 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31103 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31104 (vect_transform_loop): Call above function.
31106 2017-02-27 Richard Biener <rguenther@suse.de>
31108 PR tree-optimization/45397
31109 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31110 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31111 (visit_nary_op): Add pattern matching for CSEing sign-changed
31112 or truncated operations with wider ones.
31114 2017-02-27 Richard Biener <rguenther@suse.de>
31116 PR tree-optimization/79690
31117 * tree-vect-stmts.c (vectorizable_store): Use vector type
31118 built from the DR with address-space.
31120 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31122 * doc/invoke.texi (Optimize Options): Refine the description
31123 of asan-use-after-return.
31125 2017-02-25 Alan Modra <amodra@gmail.com>
31127 PR rtl-optimization/79584
31128 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31129 base, not ad->base_term, the reg within base. Remove assertion
31130 that ad->base == ad->base_term. Replace gen_int_mode using
31131 bogus mode with const0_rtx.
31133 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31135 PR middle-end/79396
31136 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31137 FMA_EXPR like tcc_binary or tcc_unary.
31139 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31142 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31144 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31145 (output_loc_operands): Handle DW_OP_call_ref and
31146 DW_OP_GNU_variable_value.
31147 (struct variable_value_struct): New type.
31148 (struct variable_value_hasher): Likewise.
31149 (variable_value_hash): New variable.
31150 (string_types): Remove.
31151 (copy_loc_descr): New function.
31152 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31153 (prepend_loc_descr_to_each): New function.
31154 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31155 instead of add_loc_descr_to_each if the first argument is single
31156 location list and the second has multiple.
31157 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31158 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31159 when looking for variable value which doesn't have other location info.
31160 (loc_list_from_tree): Formatting fix.
31161 (gen_array_type_die): Simplify DW_AT_string_length handling.
31162 (adjust_string_types): Remove.
31163 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31164 string_types. Call resolve_variable_values.
31165 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31166 (resolve_addr_in_expr): Likewise. Add A argument.
31167 (copy_deref_exprloc): Remove deref argument. Adjust for the
31168 original expression being DW_OP_GNU_variable_value with optionally
31169 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31170 optionally after it.
31171 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31172 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31173 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31174 (variable_value_hasher::hash, variable_value_hasher::equal): New
31176 (resolve_variable_value_in_expr, resolve_variable_value,
31177 resolve_variable_values, note_variable_value_in_expr,
31178 note_variable_value): New functions.
31179 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31182 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31185 * opts.h (handle_generated_option): Add GENERATED_P argument.
31186 * opts-common.c (handle_option): Adjust function comment.
31187 (handle_generated_option): Add GENERATED_P argument, pass it to
31189 (control_warning_option): Pass false to handle_generated_option
31191 * opts.c (maybe_default_option): Pass true to handle_generated_option
31193 * optc-gen.awk: Likewise.
31195 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31197 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31198 a REG, look at the REG it is a SUBREG of.
31199 (splitter for cmpeqsi_t): Ditto.
31201 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31203 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31204 the special USEs with the pattern of the insn, not the insn itself.
31206 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31209 * doc/invoke.texi: Document -mload-store-pairs.
31211 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31212 Sandra Loosemore <sandra@codesourcery.com>
31214 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31215 argument isn't a CONST_INT.
31216 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31217 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31218 (nios2_validate_compare): Bypass alternate compare logic if *op2
31219 is not a CONST_INT.
31220 (ldstwm_operation_p): Return false if first_base is not a REG or
31221 if first_offset is not a CONST_INT.
31223 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31225 * config/cris/cris.md: Use correct operand in a define_peephole2.
31227 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31229 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31231 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31233 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31234 this_insn if it is an INSN or JUMP_INSN.
31235 (force_offsettable): Look at base, not at addr.
31236 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31237 on things that aren't necessarily CONST_INTs.
31239 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31241 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31242 -mfpmath=sse is the default also for x86-32 targets with SSE2
31243 instruction set when @option{-ffast-math} is enabled
31245 2017-02-24 Jeff Law <law@redhat.com>
31247 PR rtl-optimizatoin/79286
31248 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31251 2017-02-24 Richard Biener <rguenther@suse.de>
31253 PR tree-optimization/79389
31254 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31257 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31259 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31260 function comment to reflect reality.
31261 (loop_exits_before_overflow): Fix typo in function description.
31263 2017-02-24 Richard Biener <rguenther@suse.de>
31265 PR tree-optimization/79389
31266 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31267 properly that a threading opportunity exists. Detect conditional
31268 copy/constant propagation opportunities.
31270 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31272 * config/visium/visium.md (type): Add trap.
31273 (b): New mode attribute.
31274 (*btst): Rename into...
31275 (*btst<mode>): ...this and adjust.
31276 (*cbranchsi4_btst_insn): Rename into...
31277 (*cbranch<mode>4_btst_insn): ...this and adjust.
31278 (trap): New define_insn.
31280 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31282 PR tree-optimization/79389
31283 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31284 (noce_reversed_cond_code): New function.
31285 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31286 reversed_comparison_code. Formatting fix.
31287 (noce_try_store_flag): Test rev_cond != NULL in addition to
31288 reversed_comparison_code.
31289 (noce_try_store_flag_constants): Likewise.
31290 (noce_try_store_flag_mask): Likewise.
31291 (noce_try_addcc): Use rev_cond if non-NULL instead of
31292 reversed_comparison_code.
31293 (noce_try_cmove_arith): Likewise. Formatting fixes.
31294 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31295 (noce_find_if_block): Initialize rev_cond.
31296 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31297 instead of false as last argument never attempt to reverse it
31300 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31302 PR tree-optimization/79663
31303 * tree-predcom.c (combine_chains): Process refs in reverse order
31304 only for ZERO length chains, and add explaining comment.
31306 2017-02-23 Jeff Law <law@redhat.com>
31308 PR tree-optimization/79578
31309 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31310 in call to operand_equal_p.
31312 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31315 * config/i386/cpuid.h: Fix another undefined behavior.
31317 2017-02-23 Richard Biener <rguenther@suse.de>
31319 PR tree-optimization/79683
31320 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31321 vector types for data-refs.
31323 2017-02-23 Martin Liska <mliska@suse.cz>
31325 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31327 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31329 PR middle-end/79665
31330 * internal-fn.c (get_range_pos_neg): Moved to ...
31331 * tree.c (get_range_pos_neg): ... here. No longer static.
31332 * tree.h (get_range_pos_neg): New prototype.
31333 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31334 are known to be in between 0 and signed maximum inclusive, try to
31335 expand both unsigned and signed divmod and use the cheaper one from
31338 2017-02-22 Jeff Law <law@redhat.com>
31340 PR tree-optimization/79578
31341 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31342 to compare base operands.
31344 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31347 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31348 gpc_reg_operand instead of fpr_reg_operand.
31350 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31352 * config/mips/mips.c (mips_return_in_memory): Force FP
31353 vector types to be returned in memory for o32 ABI.
31355 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31357 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31358 instead of DW_TAG_member for static data member declarations and don't
31359 set no_linkage_name for static inline data members.
31360 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31363 2017-02-22 Martin Liska <mliska@suse.cz>
31365 * doc/invoke.texi: Replace inequality signs with square brackets
31368 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31371 * lra-constraints.c (simplify_operand_subreg): Handle
31372 WORD_REGISTER_OPERATIONS targets.
31374 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31377 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31378 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31379 elimination by swapping fld*.
31381 2017-02-22 Richard Biener <rguenther@suse.de>
31383 PR tree-optimization/79673
31384 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31385 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31386 irrelevant address-space qualifiers and avoiding a
31387 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31389 2017-02-22 Richard Biener <rguenther@suse.de>
31391 PR tree-optimization/79666
31392 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31393 to not symbolically negate if that may introduce undefined
31396 2017-02-22 Martin Liska <mliska@suse.cz>
31399 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31400 * data-streamer-out.c (streamer_write_gcov_count_stream):
31402 * value-prof.c (stream_out_histogram_value): Make assert more
31403 precise based on type of counter.
31405 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31408 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31409 Use nonimmediate_operand instead of memory_operand for operand 1.
31410 (float-extend standard_x87sse_constant_load splitter): Ditto.
31412 2017-02-21 Jeff Law <law@redhat.com>
31414 PR tree-optimization/79621
31415 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31416 blocks with edges to themselves.
31418 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31421 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31422 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31423 Use gimple_call_builtin_p.
31426 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31427 on temporarily removed DEBUG_INSNs.
31429 PR tree-optimization/79649
31430 * tree-loop-distribution.c (classify_partition): Give up on
31431 non-generic address space loads/stores.
31433 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31435 * doc/loop.texi (Loop manipulation): Remove nonexistent
31436 tree_ssa_loop_version from the documentation.
31437 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31439 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31442 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31443 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31444 * config/rs6000/rs6000.c: Include except.h.
31445 (rs6000_expand_split_stack_prologue): Call
31446 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31448 2017-02-21 Martin Jambor <mjambor@suse.cz>
31451 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31452 have been analyzed.
31454 2017-02-21 Martin Jambor <mjambor@suse.cz>
31456 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31457 for backward compatibility only.
31458 * doc/invoke.texi (Option Summary): Remove all references to
31459 -fipa-cp-alignment.
31461 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31465 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31467 * lra-constraints.c (curr_insn_transform): Handle
31468 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31470 2017-02-21 Martin Liska <mliska@suse.cz>
31472 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31474 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31476 PR translation/79638
31477 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31479 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31482 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31483 (arm_function_ok_for_sibcall): Return false for an indirect call by
31484 descriptor if all the argument registers are used.
31485 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31486 alignment of the function.
31488 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31490 PR tree-optimization/61441
31491 * simplify-rtx.c (simplify_const_unary_operation): For
31492 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31493 the sNaN unmodified.
31495 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31497 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31498 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31499 instead of SYSTEM_HEADER_DIR.
31501 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31502 Martin Liška <mliska@suse.cz>
31504 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31505 Fix typos and grammar, use active voice, and clarify.
31507 2017-02-20 Marek Polacek <polacek@redhat.com>
31509 PR middle-end/79537
31510 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31513 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31515 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31518 * config/i386/i386.c (ix86_expand_builtin): Handle
31519 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31520 ix86_builtins_isa[fcode].isa as a requirement of those
31521 flags and any other flag in the bitmask.
31522 (ix86_init_mmx_sse_builtins): Use 0 instead of
31523 ~OPTION_MASK_ISA_64BIT as mask.
31524 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31525 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31526 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31527 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31529 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31532 * lra-constraints.c (split_reg): Check requested split mode
31533 is supported by the register.
31535 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31537 * lra-constraints.c (simplify_operand_subreg): Remove early
31540 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31543 * lra-constraints.c (curr_insn_transform): Tighten condition
31544 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31546 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31549 * lra-constraints.c (curr_insn_transform): Handle
31550 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31552 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31555 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31557 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31559 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31562 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31565 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31567 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31570 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31572 * final.c (last_columnnum, override_columnnum): New variables.
31573 (final_start_function): Set last_columnnum, pass it to begin_prologue
31574 hook and pass 0 to dwarf2out_begin_prologue.
31575 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31576 to source_line debug hook.
31577 (notice_source_line): Compute last_columnnum and for debug_column_info
31578 return true on column changes.
31579 * debug.h (struct gcc_debug_hooks): Add column argument to
31580 source_line and begin_prologue hooks.
31581 (debug_nothing_int_charstar_int_bool): Remove prototype.
31582 (debug_nothing_int_int_charstar,
31583 debug_nothing_int_int_charstar_int_bool): New prototypes.
31584 (dwarf2out_begin_prologue): Add column argument.
31585 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31586 begin_prologue hooks.
31587 (debug_nothing_int_charstar_int_bool): Remove.
31588 (debug_nothing_int_int_charstar,
31589 debug_nothing_int_int_charstar_int_bool): New functions.
31590 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31591 through to dwarf2out_source_line.
31592 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31593 (dwarf2out_source_line): Add column argument, emit it if requested.
31594 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31596 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31597 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31598 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31599 through to dwarf2out_begin_prologue.
31600 (vmsdbgout_source_line): Add column argument, pass it through to
31601 dwarf2out_source_line.
31602 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31603 dbxout_source_line caller.
31604 (dbxout_source_line): Add column argument.
31606 * common.opt (gno-column-info, gcolumn-info): New options.
31607 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31608 (check_die): Also test for multiple DW_AT_decl_column attributes.
31609 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31610 DW_AT_decl_column if requested.
31611 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31613 (gen_variable_die): Likewise.
31614 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31615 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31618 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31619 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31620 (ix86_handle_option): Handle OPT_m3dnowa.
31621 * doc/invoke.texi (-m3dnowa): Document.
31622 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31623 -m3dnowa instead of -m3dnow -march=athlon.
31626 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31627 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31629 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31632 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31633 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31634 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31635 generator for vsx_xxpermdi_<mode>_be.
31636 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31637 force big-endian semantics.
31638 (vsx_xxpermdi_<mode>_be): New define_expand with same
31639 implementation as previous version of vsx_xxpermdi_<mode>.
31641 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31643 PR tree-optimization/79327
31644 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31645 variable, its initialization and use.
31647 2017-02-17 Julia Koval <julia.koval@intel.com>
31649 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31650 (OPTION_MASK_ISA_PKU_UNSET): New.
31651 (ix86_handle_option): Handle -mrdpid.
31652 * config/i386/cpuid.h (bit_RDPID): New.
31653 * config/i386/driver-i386.c (host_detect_local_cpu):
31654 Detect RDPID feature.
31655 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31656 * config/i386/i386-c.c (ix86_target_macros_internal):
31658 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31659 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31660 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31661 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31662 * config/i386/i386.md (define_insn "rdpid"): New.
31663 * config/i386/i386.opt Add -mrdpid.
31664 * config/i386/immintrin.h (_rdpid_u32): New.
31666 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31668 PR rtl-optimization/79541
31669 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31670 instead of transforming it into USE.
31672 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31674 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31675 If HONOR_SNANS (SFmode) force the input to a register.
31676 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31677 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31678 an frsp or similar insn.
31680 2017-02-17 Martin Liska <mliska@suse.cz>
31682 PR rtl-optimization/79577
31683 * params.def (selsched-max-sched-times): Increase minimum to 1.
31685 2017-02-17 Martin Liska <mliska@suse.cz>
31687 PR rtl-optimization/79574
31688 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31690 2017-02-17 Martin Liska <mliska@suse.cz>
31692 PR tree-optimization/79529
31693 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31694 ssa_defined_default_def_p to handle cases which are implicitly
31696 * tree-ssa.c (ssa_defined_default_def_p): New function.
31697 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31698 which are implicitly defined.
31699 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31701 2017-02-17 Richard Biener <rguenther@suse.de>
31703 PR middle-end/79576
31704 * params.def (max-ssa-name-query-depth): Limit to 10.
31706 2017-02-17 Richard Biener <rguenther@suse.de>
31708 PR tree-optimization/79552
31709 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31712 2017-02-17 Richard Biener <rguenther@suse.de>
31715 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31717 2017-02-17 Marek Polacek <polacek@redhat.com>
31719 PR middle-end/79536
31720 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31721 (fold_negate_expr): New wrapper.
31723 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31725 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31726 Correct terminology and de-emphasize pre-standard behavior.
31728 2017-02-16 Alan Modra <amodra@gmail.com>
31730 PR rtl-optimization/79286
31731 * ira.c (def_dominates_uses): New function.
31732 (update_equiv_regs): Don't create an equivalence for insns that
31733 may trap where the register def does not dominate the use.
31735 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31737 PR rtl-optimization/78127
31738 * lra.c (lra): Call lra_eliminate before finish the loop after
31741 2017-02-16 Richard Biener <rguenther@suse.de>
31743 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31745 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31746 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31747 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31748 (isl_val_int_from_wi): New function.
31749 (extract_affine_gmp): Rename to ...
31750 (extract_affine_wi): ... this, take a widest_int.
31751 (extract_affine_int): Just wrap extract_affine_wi.
31752 (add_param_constraints): Use isl_val_int_from_wi.
31753 (add_loop_constraints): Likewise, and extract_affine_wi.
31755 2017-02-15 Jeff Law <law@redhat.com>
31757 PR middle-end/79521
31758 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31759 ira_init_register_move_cost_if_necessary.
31761 2017-02-15 Martin Sebor <msebor@redhat.com>
31763 PR middle-end/32003
31764 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31765 removed in a prior commit.
31767 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31769 PR tree-optimization/79347
31770 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31771 counters during peeling.
31773 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31775 * Makefile.in (site.exp): Remove "set ISLVER".
31777 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31780 * real.c (real_from_integer): Call real_convert even for decimal.
31782 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31785 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31787 2017-02-14 Andrew Pinski <apinski@cavium.com>
31789 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31790 cores and change the partno/implementer to be correct.
31791 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31792 the 'B" as the implementer.
31793 * config/aarch64/aarch64-tune.md: Regenerate.
31795 2017-02-14 Carl Love <cel@us.ibm.com>
31797 * config/rs6000/rs6000.c: Add case statement entry to make the
31798 xvcvuxdsp built-in argument unsigned.
31799 * config/rs6000/vsx.md: Fix the source and return operand types so they
31800 match the instruction definitions from the ISA document. Fix typo
31801 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31804 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31807 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31808 member early_clobber_alts.
31809 * lra-lives.c (reg_early_clobber_p): New.
31810 (process_bb_lives): Use it.
31811 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31812 (debug_operand_data): Initialize early_clobber_alts.
31813 (setup_operand_alternative): Set up early_clobber_alts.
31814 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31815 alternatives to new_insn_reg.
31816 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31818 (lra_update_insn_regno_info): Pass the new arg.
31820 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31822 PR middle-end/79505
31823 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31824 (new_oacc_loop_raw): Don't clear already cleared fields.
31827 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31828 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31829 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31831 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31834 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31836 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31839 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31840 the extra instruction to the right place to store 128-bit constant
31843 2017-02-14 Martin Sebor <msebor@redhat.com>
31845 PR middle-end/79448
31846 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31847 warning for strings of unknown length.
31849 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31851 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31853 2017-02-14 Jeff Law <law@redhat.com>
31856 * ira-costs.c (scan_one_insn): Initialize register move costs
31857 for pseudos seen in USE/CLOBBER insns.
31859 PR tree-optimization/79095
31860 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31861 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31862 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31863 if the operands are known to be not equal, then the resulting range
31865 (intersect_ranges): If the new range is ~[0,0] and the old range is
31866 wide, then prefer ~[0,0].
31867 * tree-vrp.c (overflow_comparison_p_1): New function.
31868 (overflow_comparison_p): New function.
31869 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31870 if NAME is used in an overflow test.
31871 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31872 overflow check that can be expressed as an equality test, then adjust
31873 ops to be that equality test.
31875 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31877 * config/s390/s390-builtin-types.def: Remove flags argument.
31878 * config/s390/s390.c (s390_init_builtins): Likewise.
31880 2017-02-14 Martin Liska <mliska@suse.cz>
31882 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31883 vector. Fix trailing white spaces.
31885 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31887 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31890 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31892 PR rtl-optimization/68664
31893 * config/arm/arm.c (arm_sched_can_speculate_insn):
31894 New function. Declare prototype.
31895 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31897 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31899 PR rtl-optimization/68664
31900 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31902 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31904 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31906 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31907 max skip bytes for function, loop and jump.
31909 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31911 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31912 ABS_EXPR for gimple dump.
31914 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31917 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31919 PR tree-optimization/79408
31920 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31921 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31922 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31923 also if rhs1 is INTEGER_CST.
31925 2017-02-14 Richard Biener <rguenther@suse.de>
31927 PR middle-end/79432
31928 * tree-into-ssa.c (insert_phi_nodes): When the function can
31929 have abnormal edges rewrite SSA names with broken use-def
31930 dominance out of SSA and register them for PHI insertion.
31932 2017-02-13 Martin Sebor <msebor@redhat.com>
31934 PR middle-end/79496
31935 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31936 clearing info.nowrite flag when snprintf size argument is a range.
31938 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31940 * cprop.c (cprop_jump): Add missing space in string literal.
31941 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31942 (get_constraint_for_component_ref): Likewise.
31943 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31944 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31945 * lra-constraints.c (process_alt_operands): Likewise.
31946 * ipa-inline.c (inline_small_functions): Likewise.
31947 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31948 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31949 * trans-mem.c (diagnose_tm_1_op): Likewise.
31950 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31951 (grid_parallel_clauses_gridifiable): Likewise.
31953 * config/nvptx/mkoffload.c (process): Add space in between
31956 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31957 "MOD4_SSE_REGS" and "ALL_REGS".
31959 * spellcheck.c (test_data): Add , in between "foo" and "food".
31961 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31964 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31965 boundary crossing check and subsequent code generation agree.
31967 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31969 * config/aarch64/aarch64.c (has_memory_op): Delete.
31970 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31973 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31975 PR rtl-optimization/79388
31976 PR rtl-optimization/79450
31977 * combine.c (distribute_notes): When removing TEM_INSN for which
31978 corresponding dest has last value recorded, invalidate that last
31981 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31983 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31984 of explicit '@'. Add missing assembly comment marker on branch costs
31987 2017-02-13 Nathan Sidwell <nathan@acm.org>
31989 * gengtype-lex.l (<in_struct>): Add '/'.
31991 2017-02-13 Martin Liska <mliska@suse.cz>
31994 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31996 2017-02-13 Richard Biener <rguenther@suse.de>
31998 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32000 * configure: Re-generate.
32001 * config.in: Likewise.
32002 * graphite-dependences.c: Simplify as if
32003 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32004 * graphite-isl-ast-to-gimple.c: Likewise.
32005 * graphite-optimize-isl.c: Likewise.
32006 * graphite-poly.c: Likewise.
32007 * graphite-sese-to-poly.c: Likewise.
32008 * graphite.h: Likewise.
32009 * toplev.c: Include isl/version.h and use isl_version () for
32010 printing the ISL version.
32011 * doc/install.texi: Update ISL requirement.
32013 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32015 * doc/standards.texi (Standards): Update reference to
32018 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32020 * doc/extend.texi (Named Address Spaces): sourceware.org now
32022 * doc/install.texi (Binaries): Ditto.
32025 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32027 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32028 standard terminology "stringize"/"stringizing" throughout.
32029 * doc/cppinternals.texi: Likewise.
32031 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32033 * doc/extend.texi: Fix some spelling mistakes and typos.
32034 * doc/invoke.texi: Likewise.
32036 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32039 * params.def (inline-min-speedup) Change from 10 to 8.
32041 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32043 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32046 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32049 * ipa-inline-analysis.c (get_minimal_bb): New function.
32050 (record_modified): Use it.
32051 (remap_edge_change_prob): Handle also ancestor functions.
32053 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32055 * doc/contrib.texi (Contributors): Remove broken link into
32056 the Mauve CVS repository.
32058 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32060 PR middle-end/79454
32061 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32062 result computation whenever lhs doesn't have vector mode, not
32063 just when it has BLKmode.
32065 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32067 * doc/makefile.texi (profiledbootstrap): Refer to the
32068 installation instructions only in textual form.
32070 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32073 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32075 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32077 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32078 (Specific): Update mingw-w64 reference.
32080 (Specific): Remove broken link to Renesas RX processor.
32082 2017-02-10 Richard Biener <rguenther@suse.de>
32084 * toplev.c (process_options): Do not mention obsolete graphite
32085 options when printing sorry message about missing graphite support.
32086 Mention -floop-nest-optimize.
32088 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32090 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32091 (vtst_p16): Likewise.
32092 (vtstq_p8): Likewise.
32093 (vtstq_p16): Likewise.
32095 (vtstq_p64): Likewise.
32096 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32097 (vset_lane_p64): New.
32098 (vsetq_lane_p64): New.
32100 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32102 PR tree-optimization/79411
32103 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32104 stmt operands are SSA_NAMEs used in abnormal phis.
32105 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32108 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32111 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32114 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32116 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32118 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32120 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32124 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32125 not arbitrary TREE_CONSTANT.
32128 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32129 "omp declare target link" attribute unless is_global_var.
32130 * omp-offload.c (find_link_var_op): Likewise.
32132 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32133 Chung-Lin Tang <cltang@codesourcery.com>
32135 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32137 (gimplify_adjust_omp_clauses): Don't delete TILE.
32138 (gimplify_omp_for): Deal with TILE.
32139 * internal-fn.c (expand_GOACC_TILE): New function.
32140 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32142 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32143 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32145 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32146 avoid DIV for outermost collapse var.
32147 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32148 Remove out of date comments, fix whitespace.
32149 * omp-general.c (omp_extract_for_data): Deal with tiling.
32150 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32151 adjust OLF_DIM_BASE value.
32152 (struct omp_for_data): Add tiling field.
32153 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32154 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32155 for auto loops. Remove default auto determining, moved to
32156 oacc_loop_fixed_partitions.
32157 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32158 stmts, add e_mask field.
32159 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32160 (oacc_thread_numbers): Use oacc_dim_call.
32161 (oacc_xform_tile): New.
32162 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32163 (finish_oacc_loop): Adjust for ifns vector.
32164 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32165 add case for GOACC_TILE fns.
32166 (oacc_loop_xform_loop): Delete.
32167 (oacc_loop_process): Iterate over call list directly, and add
32168 handling for GOACC_TILE fns.
32169 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32171 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32172 vector partitioning to outer loops. Assign 2 partitions to loops
32173 when available. Add TILE handling.
32174 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32175 (execite_oacc_device_lower): Process GOACC_TILE fns,
32176 ignore unknown specs.
32177 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32178 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32179 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32181 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32183 * configure.ac (ACX_BUGURL): Update.
32184 * configure: Regenerate.
32186 2017-02-09 Richard Biener <rguenther@suse.de>
32188 PR tree-optimization/69823
32189 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32190 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32192 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32194 * config/arc/arc-c.def: Add __NPS400__ definition.
32195 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32196 (TARGET_NPS400): Define.
32198 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32200 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32202 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32203 pointer, arch_info.
32204 (arc_cpu_types): Fill the arch_info field with a pointer into the
32205 arc_arch_types table.
32206 (arc_selected_cpu): Declare.
32207 * config/arc/arc.c (arc_selected_cpu): Make global.
32208 (arc_selected_arch): Delete.
32209 (arc_base_cpu): Delete.
32210 (arc_override_options): Remove references to deleted variables,
32211 update access to arch information.
32212 (ARC_OPT): Update access to arch information.
32213 (ARC_OPTX): Likewise.
32214 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32215 (TARGET_ARC600): Update access to arch information.
32216 (TARGET_ARC601): Likewise.
32217 (TARGET_ARC700): Likewise.
32218 (TARGET_EM): Likewise.
32219 (TARGET_HS): Likewise.
32220 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32223 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32226 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32227 condition/operands for integer GE/LE/GEU/LEU operations.
32229 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32231 PR translation/79397
32232 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32235 2017-02-08 Martin Jambor <mjambor@suse.cz>
32238 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32239 whether allocation happened.
32240 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32241 nothing was allocated.
32243 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32245 PR tree-optimization/79408
32246 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32247 constant, but SSA_NAME with a known integer range, use the minimum
32248 of that range instead of op1 to determine if modulo can be replaced
32249 with its first operand.
32251 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32253 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32255 2017-02-08 Richard Biener <rguenther@suse.de>
32257 PR tree-optimization/71824
32258 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32259 Check all loops contained in the merged region.
32261 2017-02-07 Andrew Pinski <apinski@cavium.com>
32263 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32265 2017-02-07 Andrew Pinski <apinski@cavium.com>
32267 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32268 (thunderxt88): Likewise.
32269 (thunderxt81): Disable LSE and change v8.1 to v8.
32270 (thunderxt83): Likewise.
32272 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32273 Richard Biener <rguenther@suse.de>
32275 PR middle-end/79399
32276 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32277 type from int to size_t.
32278 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32280 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32282 PR rtl-optimization/79386
32283 * cprop.c (bypass_conditional_jumps): Initialize
32284 bypass_last_basic_block already before splitting bbs after
32285 unconditional traps...
32286 (bypass_conditional_jumps): ... rather than here.
32289 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32290 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32291 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32292 fix -masm=intel patterns.
32294 2017-02-07 Richard Biener <rguenther@suse.de>
32296 PR tree-optimization/79256
32297 PR middle-end/79278
32298 * builtins.c (get_object_alignment_2): Use min_align_of_type
32299 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32300 and ADJUST_FIELD_ALIGN.
32302 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32304 * doc/tm.texi: Regenerate.
32305 * stor-layout.c (layout_decl): Adjust.
32306 (update_alignment_for_field): Likewise.
32307 (place_field): Likewise.
32308 (min_align_of_type): Likewise.
32309 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32310 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32311 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32312 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32313 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32314 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32315 * config/i386/i386.c (x86_field_alignment): Likewise.
32316 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32317 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32318 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32319 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32320 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32321 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32325 2017-01-30 Richard Biener <rguenther@suse.de>
32327 PR tree-optimization/79256
32328 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32329 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32332 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32334 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32335 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32336 builtins to SImode and emit a zero-extend, if necessary.
32338 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32340 * docs/invoke.texi (RISC-V Options): Alphabetize.
32342 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32344 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32347 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32349 * config/riscv/riscv.c: New file.
32350 * common/config/riscv/riscv-common.c: Likewise.
32351 * config.gcc: Likewise.
32352 * config/riscv/constraints.md: Likewise.
32353 * config/riscv/elf.h: Likewise.
32354 * config/riscv/generic.md: Likewise.
32355 * config/riscv/linux.h: Likewise.
32356 * config/riscv/multilib-generator: Likewise.
32357 * config/riscv/peephole.md: Likewise.
32358 * config/riscv/pic.md: Likewise.
32359 * config/riscv/predicates.md: Likewise.
32360 * config/riscv/riscv-builtins.c: Likewise.
32361 * config/riscv/riscv-c.c: Likewise.
32362 * config/riscv/riscv-ftypes.def: Likewise.
32363 * config/riscv/riscv-modes.def: Likewise.
32364 * config/riscv/riscv-opts.h: Likewise.
32365 * config/riscv/riscv-protos.h: Likewise.
32366 * config/riscv/riscv.h: Likewise.
32367 * config/riscv/riscv.md: Likewise.
32368 * config/riscv/riscv.opt: Likewise.
32369 * config/riscv/sync.md: Likewise.
32370 * config/riscv/t-elf-multilib: Likewise.
32371 * config/riscv/t-linux: Likewise.
32372 * config/riscv/t-linux-multilib: Likewise.
32373 * config/riscv/t-riscv: Likewise.
32374 * configure.ac: Likewise.
32375 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32376 Waterman as RISC-V maintainers.
32377 * doc/install.texi: Add RISC-V entries.
32378 * doc/invoke.texi: Add RISC-V options section.
32379 * doc/md.texi: Add RISC-V constraints section.
32380 * configure: Regenerated.
32382 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32385 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32386 false values to be constant vectors with all 0 or all 1 bits set.
32387 (vcondu<mode><mode>): Likewise.
32388 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32390 (fpmask_comparison_operator): Update comment.
32391 (vecint_comparison_operator): New predicate.
32392 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32393 vector conditionals when the true and false values are constant
32394 vectors with all 0 bits or all 1 bits set.
32396 2017-02-06 Martin Sebor <msebor@redhat.com>
32398 PR tree-optimization/79376
32399 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32401 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32403 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32404 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32405 to simplify split condition.
32407 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32409 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32410 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32413 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32415 PR rtl-optimization/68664
32416 * target.def (can_speculate_insn): New hook.
32417 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32418 * doc/tm.texi: Regenerate.
32419 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32420 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32421 (rs6000_sched_can_speculate_insn): New function.
32423 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32425 PR tree-optimization/79284
32426 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32427 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32428 vectorizable_mask_load_store, vectorizable_operation,
32429 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32430 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32431 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32432 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32433 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32434 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32435 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32436 is_gimple_assign (stmt). Replace another such test with
32437 is_gimple_assign (stmt).
32439 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32442 * config/avr/avr.c (rtl-iter.h): Include it.
32443 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32444 (avr_legitimate_combined_insn): ...and implementation.
32446 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32448 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32449 * config/s390/s390.c (s390_const_operand_ok)
32450 (s390_canonicalize_comparison, s390_extract_part)
32451 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32452 (s390_contiguous_bitmask_p, s390_rtx_costs)
32453 (legitimize_pic_address): Likewise.
32454 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32455 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32456 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32457 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32458 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32460 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32462 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32463 REGNO($0) == REGNO($1).
32465 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32467 * config/s390/linux.h(SIZE_TYPE): Add comment.
32469 2017-02-06 Julian Brown <julian@codesourcery.com>
32470 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32471 Virendra Pathak <virendra.pathak@broadcom.com>
32473 * config/aarch64/aarch64-cores.def: Change the scheduler
32475 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32476 * config/aarch64/thunderx2t99.md: New file.
32478 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32480 * doc/standards.texi (Go Language): Update link to language
32483 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32485 * tree-eh.c (lower_resx): Sanitize profile.
32486 (cleanup_empty_eh_move_lp): Likewise.
32488 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32491 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32493 * cfgloopmanip.h (loop_version): Update prototype.
32494 * modulo-sched.c (sms_schedule): Update call of loop_version.
32495 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32496 * tree-parloops.c (gen_parallel_loop): Likewise.
32497 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32498 * tree-ssa-loop-split.c (split_loop): Likewise.
32499 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32500 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32502 2017-02-05 Martin Liska <mliska@suse.cz>
32505 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32507 (print_operand_address): Initialize a struct to zero.
32509 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32511 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32512 garbage collector only in textual form.
32514 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32516 * doc/extend.texi (x86 specific memory model extensions for
32517 transactional memory): Simplify a phrase.
32519 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32522 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32523 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32524 (atomic_storedi_1): Likewise.
32526 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32528 PR tree-optimization/79338
32529 * tree-parloops.c (gather_scalar_reductions): Don't call
32530 vect_analyze_loop_form for loop->inner before destroying loop's
32533 2017-02-03 Martin Sebor <msebor@redhat.com>
32535 PR tree-optimization/79327
32536 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32537 when precision has resulted in leading zeros.
32538 (format_integer): Adjust the likely counter to assume an unknown
32539 argument that may be zero is non-zero.
32541 2017-02-03 Jason Merrill <jason@redhat.com>
32544 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32545 avoid copying non-taken branch.
32547 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32549 PR tree-optimization/79340
32550 * tree-vect-loop.c (vectorizable_reduction): Release
32551 vec_defs elements after safe_splicing them into other vectors.
32554 PR tree-optimization/79327
32555 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32556 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32558 (format_integer): Use wide_int_to_tree instead of build_int_cst
32559 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32560 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32561 of shortest and longest sequence.
32563 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32565 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32566 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32568 2017-02-03 Walter Lee <walt@tilera.com>
32571 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32572 after initial stackframe link reg save.
32573 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32575 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32578 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32579 wu for stxssp alternative.
32581 2017-02-03 Martin Sebor <msebor@redhat.com>
32583 PR tree-optimization/79352
32584 * gimple-fold.c (get_range_strlen): Add argument.
32585 (get_range_strlen): Change return type to bool.
32586 (get_maxval_strlen): Pass in a dummy argument.
32587 * gimple-fold.h (get_range_strlen): Change return type to bool.
32588 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32589 * tree.h (array_at_struct_end_p): Add argument.
32590 * tree.c (array_at_struct_end_p): Handle it.
32592 2017-02-03 Martin Liska <mliska@suse.cz>
32595 * multiple_target.c (create_dispatcher_calls): Redirect edge
32596 from a caller of a dispatcher.
32597 (expand_target_clones): Make the clones local.
32598 (ipa_target_clone): Do both target clones and resolvers.
32599 (ipa_dispatcher_calls): Remove the pass.
32600 (pass_dispatcher_calls::gate): Likewise.
32601 (make_pass_dispatcher_calls): Likewise.
32602 * passes.def (pass_target_clone): Put as very first IPA early
32605 2017-02-03 Martin Liska <mliska@suse.cz>
32607 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32608 in case of a function with ifunc attribute.
32610 2017-02-03 Martin Liska <mliska@suse.cz>
32612 * cgraph.c (cgraph_node::dump): Dump function version info.
32613 * symtab.c (symtab_node::dump_base): Add missing new line.
32615 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32617 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32618 (ifcombine_ifandif): Use it.
32620 2017-02-03 Martin Liska <mliska@suse.cz>
32622 * doc/invoke.texi: Document default value for
32623 use-after-scope-direct-emission-threshold.
32625 2017-02-03 Martin Liska <mliska@suse.cz>
32627 PR tree-optimization/79339
32628 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32629 (format_floating): Likewise.
32631 2017-02-03 Martin Liska <mliska@suse.cz>
32634 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32636 (ipa_node_params_t::remove): Likewise.
32637 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32638 initialization from removed ipa_node_params_t::insert.
32639 (ipa_node_params::~ipa_node_params): Move from removed
32640 ipa_node_params_t::release.
32641 * symbol-summary.h (symbol_summary::m_released): New member.
32642 Do not release a summary twice. Do not allow to call finalizer
32643 for types of a summary that live in GGC memory.
32645 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32647 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32650 2017-02-02 Martin Sebor <msebor@redhat.com>
32652 PR middle-end/79275
32653 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32654 (format_string): Tighten up the range of output for non-constant
32655 strings and correct the expected range for wide non-constant strings.
32657 2017-02-02 Martin Sebor <msebor@redhat.com>
32659 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32661 PR middle-end/32003
32662 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32664 (-fdump-tree-@var): Add to index and document how to come up
32665 with pass-specific option and dump file names.
32666 (-fdump-passes): Clarify where to look for output.
32668 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32670 PR middle-end/77445
32671 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32672 statistics of the analyzed path; allow threading for speed when
32673 any of BBs along the path are optimized for speed.
32675 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32677 PR middle-end/78468
32678 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32679 settings of the virtual registers.
32682 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32684 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32685 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32688 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32690 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32691 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32693 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32695 * config/s390/s390.md: Add missing comments with the expanded
32697 * config/s390/vector.md: Likewise.
32698 * config/s390/vx-builtins.md: Likewise.
32700 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32703 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32704 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32705 conditions on a single line.
32707 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32709 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32710 __S390_VX__ to __VX__.
32712 2017-02-01 Andrew Pinski <apinski@cavium.com>
32714 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32715 stmt_info to record_stmt_cost.
32716 (vect_get_known_peeling_cost): Pass stmt_info if known to
32718 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32719 cpu_vector_cost field into
32720 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32721 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32722 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32723 splitting of scalar_stmt_cost and vec_stmt_cost.
32724 (thunderx_vector_cost): Likewise.
32725 (cortexa57_vector_cost): LIkewise.
32726 (exynosm1_vector_cost): Likewise.
32727 (xgene1_vector_cost): Likewise.
32728 (thunderx2t99_vector_cost): Improve after the splitting of the two
32730 (aarch64_builtin_vectorization_cost): Update for the splitting of
32731 scalar_stmt_cost and vec_stmt_cost.
32733 2017-02-01 Torvald Riegel <triegel@redhat.com>
32734 Richard Henderson <rth@redhat.com>
32736 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32737 conditional on existance of a fast atomic load.
32738 * optabs-query.c (can_atomic_load_p): New function.
32739 * optabs-query.h (can_atomic_load_p): Declare it.
32740 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32741 no fast atomic load is available for the particular size of access.
32742 (expand_atomic_compare_and_swap): Likewise.
32743 (expand_atomic_load): Likewise.
32744 (expand_atomic_store): Likewise.
32745 (expand_atomic_fetch_op): Likewise.
32746 * testsuite/lib/target-supports.exp
32747 (check_effective_target_sync_int_128): Remove x86 because it provides
32748 no fast atomic load.
32749 (check_effective_target_sync_int_128_runtime): Likewise.
32751 2017-02-01 Richard Biener <rguenther@suse.de>
32753 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32754 (graphite_transform_loops): Provide opt-info for optimized nests.
32755 * tree-parloop.c (parallelize_loops): Provide opt-info for
32756 parallelized loops.
32758 2017-02-01 Richard Biener <rguenther@suse.de>
32760 PR middle-end/79315
32761 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32762 was not set before.
32764 2017-02-01 Richard Biener <rguenther@suse.de>
32766 PR tree-optimization/71824
32767 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32768 Verify the loops are valid in the merged SESE region.
32769 (scop_detection::can_represent_loop_1): Check analyzing the
32770 evolution of the number of iterations in the region succeeds.
32772 2017-01-31 Ian Lance Taylor <iant@golang.org>
32774 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32775 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32777 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32779 PR preprocessor/79210
32780 * input.c (get_substring_ranges_for_loc): Replace line_width
32781 assertion with error-handling.
32783 2017-01-31 Richard Biener <rguenther@suse.de>
32785 PR tree-optimization/77318
32786 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32787 (create_pw_aff_from_tree): Take loop parameter.
32788 (add_condition_to_pbb): Pass loop of the condition to
32789 create_pw_aff_from_tree.
32791 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32793 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32794 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32796 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32800 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32802 (convert_int_to_float128): Likewise.
32803 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32804 (convert_int_to_float128): Likewise.
32805 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32806 (UNSPEC_IEEE128_CONVERT): Likewise.
32807 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32808 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32809 Use local variables for IBM extended format.
32810 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32811 (fix_trunc<mode>si2_fprs): Likewise.
32812 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32813 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32814 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32815 to know that we can now have integers of all sizes in vector
32817 (fix<uns>_<mode>di2_hw): Likewise.
32818 (float<uns>_<mode>si2_hw): Likewise.
32819 (fix_<mode>si2_hw): Likewise.
32820 (fixuns_<mode>si2_hw): Likewise.
32821 (float<uns>_<mode>di2_hw): Likewise.
32822 (float_<mode>di2_hw): Likewise.
32823 (float_<mode>si2_hw): Likewise.
32824 (floatuns_<mode>di2_hw): Likewise.
32825 (floatuns_<mode>si2_hw): Likewise.
32826 (xscvqp<su>wz_<mode>): Delete, no longer used.
32827 (xscvqp<su>dz_<mode>): Likewise.
32828 (xscv<su>dqp_<mode>): Likewise.
32829 (ieee128_mfvsrd_64bit): Likewise.
32830 (ieee128_mfvsrd_32bit): Likewise.
32831 (ieee128_mfvsrwz): Likewise.
32832 (ieee128_mtvsrw): Likewise.
32833 (ieee128_mtvsrd_64bit): Likewise.
32834 (ieee128_mtvsrd_32bit): Likewise.
32836 2017-01-31 Martin Liska <mliska@suse.cz>
32839 * ipa-prop.c (ipa_free_all_node_params): Call release method
32840 instead of ~sumbol_summary to not to trigger double times
32843 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32845 PR tree-optimization/71691
32846 * bitmap.h (class auto_bitmap): New.
32847 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32848 is_maybe_undefined instead of ssa_undefined_value_p.
32850 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32852 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32853 __S390_ARCH_LEVEL__ to __ARCH__.
32855 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32857 PR tree-optimization/79267
32858 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32859 if should_remove_lhs_p is true.
32861 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32864 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32865 (add_alignment_attribute): New.
32866 (base_type_die): Add alignment attribute.
32867 (subrange_type_die): Likewise.
32868 (modified_type_die): Likewise.
32869 (gen_array_type_die): Likewise.
32870 (gen_descr_array_type_die: Likewise.
32871 (gen_enumeration_type_die): Likewise.
32872 (gen_subprogram_die): Likewise.
32873 (gen_variable_die): Likewise.
32874 (gen_field_die): Likewise.
32875 (gen_ptr_to_mbr_type_die): Likewise.
32876 (gen_struct_or_union_type_die): Likewise.
32877 (gen_subroutine_type_die): Likewise.
32878 (gen_typedef_die): Likewise.
32879 (base_type_cmp): Compare alignment attribute.
32881 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32884 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32885 (setb_unsigned) New pattern for setb with CCUNS.
32886 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32887 subfc./subfe sequence to avoid overflow problems. Generate a
32888 shorter sequence with cmpld/setb for power9.
32889 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32890 for generating subfc. instruction.
32891 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32892 now uses this instruction.
32894 2017-01-30 Ian Lance Taylor <iant@google.com>
32897 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32898 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32900 2017-01-30 Martin Sebor <msebor@redhat.com>
32902 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32903 Move constant to the right of a relational operator.
32904 (get_mpfr_format_length, format_character, format_string): Ditto.
32905 (should_warn_p, maybe_warn): Same.
32907 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32909 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32912 * asan.c (get_translation_unit_decl): Remove function.
32913 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32915 2017-01-30 Martin Liska <mliska@suse.cz>
32917 PR gcov-profile/79259
32918 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32919 -fprofile-generate.
32921 2017-01-30 Martin Liska <mliska@suse.cz>
32924 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32925 Initialize variables with NULL value.
32927 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32930 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32932 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32934 2017-01-30 Richard Biener <rguenther@suse.de>
32936 PR tree-optimization/79276
32937 * tree-vrp.c (process_assert_insertions): Properly adjust common
32938 when removing a duplicate.
32940 2017-01-30 Richard Biener <rguenther@suse.de>
32942 PR tree-optimization/79256
32943 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32944 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32946 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32948 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32951 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32952 ("*r<noxa>sbg_<mode>_sll_bitmask")
32953 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32954 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32955 Use contiguous_bitmask_nowrap_operand.
32957 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32960 * config/rs6000/altivec.h (vec_xl): Revise #define.
32961 (vec_xst): Likewise.
32963 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32965 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32967 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32969 PR rtl-optimization/79194
32970 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32971 traps before call to bypass_conditional_jumps.
32973 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32975 PR tree-optimization/71374
32976 * lra-constraints.c (check_conflict_input_operands): New.
32977 (match_reload): Use it.
32979 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32982 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32983 account to calculate conflict_set.
32985 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32987 PR rtl-optimization/78559
32988 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32989 other_insn in combine.
32991 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32993 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32994 uint16_type_node for BT_UINT16.
32996 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32998 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32999 "RTL Tests" to menu.
33000 (GIMPLE Tests): New node.
33001 (RTL Tests): New node.
33003 2017-01-27 Richard Biener <rguenther@suse.de>
33005 PR tree-optimization/79245
33006 * tree-loop-distribution.c (distribute_loop): Apply cost
33007 modeling also to detected patterns.
33009 2017-01-27 Richard Biener <rguenther@suse.de>
33011 PR tree-optimization/71433
33012 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33013 (compare_assert_loc): New function.
33014 (process_assert_insertions): Sort and optimize assert locations
33015 to remove duplicates and push down identical assertions on
33016 edges to their destination block.
33018 2017-01-27 Richard Biener <rguenther@suse.de>
33020 PR tree-optimization/79244
33021 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33022 out SSA names even if abnormal.
33024 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33026 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33027 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33028 instead of MPFR_RNDN.
33030 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33033 * arm.c (arm_option_override): Don't call build_target_option_node
33034 until after doing all option overrides.
33035 (arm_valid_target_attribute_tree): Likewise.
33037 2017-01-27 Martin Liska <mliska@suse.cz>
33039 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33040 for {cd}tors and C++ {cd}tors.
33042 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33044 * config/s390/s390.md ("*setmem_long_and")
33045 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33047 2017-01-26 Martin Sebor <msebor@redhat.com>
33049 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33052 2017-01-26 Martin Sebor <msebor@redhat.com>
33054 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33055 HAVE_DFmode before using XFmode or DFmode.
33056 (parse_directive): Avoid using the z length modifier to avoid
33057 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33059 PR middle-end/78703
33060 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33061 to accept adjustment as an array.
33062 (get_int_range): New function.
33063 (struct directive): Make width and prec arrays.
33064 (directive::set_width, directive::set_precision): Call get_int_range.
33065 (format_integer, format_floating): Handle width and precision ranges.
33066 (format_string, parse_directive): Same.
33068 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33071 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33072 comdat_type_p set, just clone them, but keep the children in the
33076 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33077 which have direct callers with -fvar-tracking-assignments enabled
33079 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33080 inside of type units.
33082 2017-01-26 Martin Sebor <msebor@redhat.com>
33084 PR middle-end/78703
33085 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33087 (struct format_result): Replace number_chars, number_chars_min,
33088 and number_chars_max with a single member of struct result_range.
33090 (format_result::operator+=): Adjust.
33091 (struct fmtresult): Remove bounded. Handle likely and unlikely
33093 (fmtresult::adjust_for_width_or_precision): New function.
33094 (fmtresult:type_max_digits): New function.
33095 (bytes_remaining): Handle likely and unlikely counters.
33096 (min_bytes_remaining): Remove.
33097 (format_percent): Simplify.
33098 (format_integer, format_floating): Set likely and unlikely counters.
33099 (get_string_length, format_character, format_string): Same.
33100 (format_plain, should_warn_p): New function.
33101 (maybe_warn): Call should_warn_p. Update diagnostic messages
33102 and handle those for all directives, including plain strings.
33103 (format_directive): Handle likely and unlikely counters.
33104 Remove unnecessary quoting from diagnostics. Add an informational
33106 (add_bytes): Remove.
33107 (pass_sprintf_length::compute_format_length): Simplify.
33108 (try_substitute_return_value): Handle likely and unlikely counters.
33110 2017-01-26 Carl Love <cel@us.ibm.com>
33112 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33113 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33115 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33118 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33119 endianess for subregs into account.
33120 * lra-constraints.c (lra_constraints): Do risky transformations
33121 always on the first iteration.
33122 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33123 last_call_used_reg_set.
33124 (process_bb_lives): Define and use last_call_used_reg_set.
33125 * lra.c (lra): Always continue after lra_constraints on the first
33128 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33130 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33132 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33134 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33136 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33137 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33138 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33139 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33140 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33141 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33142 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33143 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33144 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33146 2017-01-26 Marek Polacek <polacek@redhat.com>
33149 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33150 for the third operand.
33152 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33154 PR middle-end/79236
33155 * omp-low.c (struct omp_context): Add simt_stmt field.
33156 (scan_omp_for): Return omp_context *.
33157 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33158 context to the _simt_ SIMD stmt.
33159 (lower_omp_for): For combined SIMD with sibling _simt_
33160 SIMD, make sure to use the same decls in _looptemp_
33161 clauses as in the sibling.
33163 2017-01-26 David Sherwood <david.sherwood@arm.com>
33165 PR middle-end/79212
33166 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33169 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33172 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33173 emit fld b; fld a; if possible.
33175 * brig-builtins.def: Update copyright years.
33176 * config/arm/arm_acle_builtins.def: Update copyright years.
33178 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33181 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33182 constraint instead of o for the stxsd instruction.
33184 2017-01-25 Carl Love <cel@us.ibm.com>
33186 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33187 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33189 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33191 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33193 2017-01-25 Richard Biener <rguenther@suse.de>
33195 PR tree-optimization/69264
33196 * target.def (vector_alignment_reachable): Improve documentation.
33197 * doc/tm.texi: Regenerate.
33198 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33200 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33201 earlier changes with respect to TYPE_USER_ALIGN.
33202 (vector_alignment_reachable_p): Likewise. Improve dumping.
33204 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33207 * config/arm/arm.md (xordi3): Force constant operand into a register
33210 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33212 * doc/invoke.texi (-fstore-merging): Correct default optimization
33213 levels at which it is enabled.
33214 (-O): Move -fstore-merging from list to...
33217 2017-01-25 Richard Biener <rguenther@suse.de>
33220 * omp-expand.c: Include debug.h.
33221 (expand_omp_taskreg): Make sure to generate early debug before
33222 outlining anything from a function.
33223 (expand_omp_target): Likewise.
33224 (grid_expand_target_grid_body): Likewise.
33226 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33229 * asan.c (get_translation_unit_decl): New function.
33230 (asan_add_global): Extract modules file name from globals
33231 TRANSLATION_UNIT_DECL name.
33233 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33236 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33237 for long calls with APCS frame and VFP.
33239 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33241 * cfg.c (original_copy_tables_initialized_p): New function.
33242 * cfg.h (original_copy_tables_initialized_p): New decl.
33243 * cfgrtl.c (relink_block_chain): Guard the call to
33244 free_original_copy_tables with a call to
33245 original_copy_tables_initialized_p.
33246 * cgraph.h (symtab_node::native_rtl_p): New decl.
33247 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33248 (symtab_node::needed_p): Don't assert for early assembly output
33249 for __RTL functions.
33250 (cgraph_node::finalize_function): Set "force_output" for __RTL
33252 (cgraph_node::analyze): Bail out early for __RTL functions.
33253 (analyze_functions): Update assertion to support __RTL functions.
33254 (cgraph_node::expand): Bail out early for __RTL functions.
33255 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33257 * function.h (struct function): Update comment for field
33259 * gimple-expr.c: Include "tree-pass.h".
33260 (gimple_has_body_p): Return false for __RTL functions.
33261 * Makefile.in (OBJS): Add run-rtl-passes.o.
33262 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33264 (gcc::pass_manager::get_clean_slate): New accessor.
33265 * passes.c: Include "insn-addr.h".
33266 (should_skip_pass_p): Add logging. Update logic for running
33267 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33268 property-provider override so it is only done for gimple passes.
33270 (skip_pass): New function.
33271 (execute_one_pass): Call skip_pass when skipping passes.
33272 * read-md.c (md_reader::read_char): Support filtering
33273 the input to a subset of line numbers.
33274 (md_reader::md_reader): Initialize fields
33275 m_first_line and m_last_line.
33276 (md_reader::read_file_fragment): New function.
33277 * read-md.h (md_reader::read_file_fragment): New decl.
33278 (md_reader::m_first_line): New field.
33279 (md_reader::m_last_line): New field.
33280 * read-rtl-function.c (function_reader::create_function): Only
33281 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33282 curr_properties. Set DECL_INITIAL to a dummy block.
33283 (read_rtl_function_body_from_file_range): New function.
33284 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33286 * run-rtl-passes.c: New file.
33287 * run-rtl-passes.h: New file.
33289 2017-01-24 Jeff Law <law@redhat.com>
33291 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33294 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33296 PR tree-optimization/79159
33297 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33298 (record_nonwrapping_iv): Improve boundary using above function if no
33299 value range information.
33301 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33302 Martin Jambor <mjambor@suse.cz>
33304 * brig-builtins.def: New file.
33305 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33306 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33307 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33308 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33309 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33310 * builtin-types.def (BT_INT8): New.
33311 (BT_INT16): Likewise.
33312 (BT_UINT8): Likewise.
33313 (BT_UINT16): Likewise.
33314 (BT_FN_ULONG): Likewise.
33315 (BT_FN_UINT_INT): Likewise.
33316 (BT_FN_UINT_ULONG): Likewise.
33317 (BT_FN_UINT_LONG): Likewise.
33318 (BT_FN_UINT_PTR): Likewise.
33319 (BT_FN_ULONG_PTR): Likewise.
33320 (BT_FN_INT8_FLOAT): Likewise.
33321 (BT_FN_INT16_FLOAT): Likewise.
33322 (BT_FN_UINT32_FLOAT): Likewise.
33323 (BT_FN_UINT16_FLOAT): Likewise.
33324 (BT_FN_UINT8_FLOAT): Likewise.
33325 (BT_FN_UINT64_FLOAT): Likewise.
33326 (BT_FN_UINT16_UINT32): Likewise.
33327 (BT_FN_UINT32_UINT16): Likewise.
33328 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33329 (BT_FN_INT_PTR_INT): Likewise.
33330 (BT_FN_UINT_PTR_UINT): Likewise.
33331 (BT_FN_LONG_PTR_LONG): Likewise.
33332 (BT_FN_ULONG_PTR_ULONG): Likewise.
33333 (BT_FN_VOID_UINT64_UINT64): Likewise.
33334 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33335 (BT_FN_INT8_INT8_INT8): Likewise.
33336 (BT_FN_INT16_INT16_INT16): Likewise.
33337 (BT_FN_INT_INT_INT): Likewise.
33338 (BT_FN_UINT_FLOAT_UINT): Likewise.
33339 (BT_FN_FLOAT_UINT_UINT): Likewise.
33340 (BT_FN_ULONG_UINT_UINT): Likewise.
33341 (BT_FN_ULONG_UINT_PTR): Likewise.
33342 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33343 (BT_FN_UINT_UINT_UINT): Likewise.
33344 (BT_FN_VOID_UINT_PTR): Likewise.
33345 (BT_FN_UINT_UINT_PTR: Likewise.
33346 (BT_FN_UINT32_UINT64_PTR): Likewise.
33347 (BT_FN_INT_INT_UINT_UINT): Likewise.
33348 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33349 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33350 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33351 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33352 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33353 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33354 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33355 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33356 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33357 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33358 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33359 * doc/frontends.texi: List BRIG FE.
33360 * doc/install.texi (Testing): Add BRIG tesring requirements.
33361 * doc/invoke.texi (Overall Options): Mention BRIG.
33362 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33364 2017-01-24 Richard Biener <rguenther@suse.de>
33366 PR translation/79208
33367 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33369 2017-01-24 Martin Jambor <mjambor@suse.cz>
33372 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33373 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33374 and known_contexts.
33376 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33378 PR middle-end/79123
33379 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33380 casts from signed to unsigned really don't have a range.
33382 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33384 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33385 GMP_RNDx for compatiblity.
33387 2017-01-24 Martin Liska <mliska@suse.cz>
33390 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33391 that would prevent us to call alloca with -1 as argument.
33393 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33395 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33396 Avoid -Wformat-security warning.
33398 2017-01-23 Andrew Pinski <apinski@cavium.com>
33400 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33403 2017-01-23 Martin Sebor <msebor@redhat.com>
33405 PR middle-end/78703
33406 * gimple-ssa-sprintf.c (warn_level): New global.
33407 (format_integer): Use it here and throughout the rest of the file.
33408 Use the same switch to compute sign as base.
33409 (maybe_warn): New function.
33410 (format_directive): Factor out warnings into maybe_warn.
33411 Add debugging output. Use warn_level.
33412 (add_bytes): Use warn_level.
33413 (pass_sprintf_length::compute_format_length): Add debugging output.
33414 (try_substitute_return_value): Same.
33415 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33417 PR middle-end/78703
33418 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33419 (struct fmtresult, format_integer, format_floating): Adjust.
33420 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33421 (get_string_length, format_string,format_directive): Same.
33422 (pass_sprintf_length::compute_format_length): Same.
33423 (try_substitute_return_value): Simplify slightly.
33425 PR middle-end/78703
33426 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33427 (fmtresult::operator+=): Outlined.
33428 (struct fmtresult): Add ctors.
33429 (struct conversion_spec): Rename...
33430 (struct directive): ...to this. Add and remove data members.
33431 (directive::set_width, directive::set_precision): New functions.
33432 (format_percent): Use fmtresult ctor.
33433 (get_width_and_precision): Remove.
33434 (format_integer): Make naming changes. Avoid computing width and
33436 (format_floating): Same. Adjust indentation.
33437 (format_character, format_none): New functions.
33438 (format_string): Moved character handling to format_character.
33439 (format_directive): Remove arguments, change return type.
33440 (parse_directive): New function.
33441 (pass_sprintf_length::compute_format_length): Move directive
33442 parsing to parse_directive.
33444 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33446 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33447 (assign_assembler_name_if_needed): ... this.
33448 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33449 (assign_assembler_name_if_needed): ... this.
33450 (free_lang_data_in_cgraph): Adjust callers.
33451 * cgraphunit.c (cgraph_node::analyze): Likewise.
33452 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33455 2017-01-23 Richard Biener <rguenther@suse.de>
33457 PR tree-optimization/79088
33458 PR tree-optimization/79188
33459 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33460 resetting loop bounds after last path deletion. Reset loop
33461 bounds of the target loop, make code match the comments.
33462 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33463 Make sure loops need no fixups.
33465 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33467 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33468 exponent support with double type for first argument.
33469 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33470 type returned by __builtin_vec_extract_sig,
33471 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33472 functions from "vector int" to "vector unsigned int" or from
33473 "vector long long int" to "vector unsigned long long int".
33474 Changed type returned by __builtin_vec_extract_exp,
33475 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33476 functions from "vector int" to "vector unsigned int" or from
33477 "vector long long int" to "vector unsigned long long int".
33478 Changed return type of __builtin_vec_test_data_class,
33479 __builtin_vec_test_data_class_sp, and
33480 __builtin_vec_test_data_class_dp from "vector int" to
33481 "vector bool int" or from "vector long long int" to "vector bool
33482 long long int" and changed second argument type from "unsigned
33483 int" to "int". Added new overloaded function forms "vector float
33484 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33485 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33486 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33487 double, vector unsigned long long int)" and "vector double
33488 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33489 long int)". Changed return type of
33490 __builtin_scalar_test_data_class and
33491 __builtin_scalar_test_data_class_sp and
33492 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33493 int" and changed second argument from "unsigned int" to "int".
33494 Changed type returned by __builtin_scalar_test_neg,
33495 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33496 from "int" to "bool int". Added new overloaded function form
33497 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33498 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33499 exponent double-precision with floating point first argument.
33500 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33501 documentation of scalar_test_data_class, scalar_test_neg,
33502 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33503 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33504 vec_test_data_class built-in functions to reflect refinements in
33505 their type signatures.
33507 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33509 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33511 (aarch64_elf_asm_destructor): Likewise.
33513 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33515 PR rtl-optimization/78634
33516 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33517 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33518 * ifcvt.c (noce_try_cmove): Add missing cost check.
33520 PR rtl-optimization/71724
33521 * combine.c (if_then_else_cond): Look for situations where it is
33522 beneficial to undo the work of one of the recursive calls.
33524 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33526 PR tree-optimization/70754
33527 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33528 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33529 combined stmt before it if not NULL.
33530 (combine_chains): Process refs reversely and compute dominance point
33533 2017-01-23 Martin Liska <mliska@suse.cz>
33535 PR tree-optimization/79196
33536 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33537 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33539 (strlen_optimize_stmt): Call the renamed function.
33541 2017-01-23 Michael Matz <matz@suse.de>
33543 PR tree-optimization/78384
33544 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33546 2017-01-23 Richard Biener <rguenther@suse.de>
33548 PR tree-optimization/79186
33549 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33550 both incoming edges before moving an assert.
33552 2017-01-23 Martin Jambor <mjambor@suse.cz>
33554 * ipa-prop.c (load_from_param_1): Removed.
33555 (load_from_unmodified_param): Bits from load_from_param_1 put back
33557 (load_from_param): Removed.
33558 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33559 with stmt. Reverted back to use of load_from_unmodified_param.
33561 2017-01-23 Martin Jambor <mjambor@suse.cz>
33564 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33565 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33566 field a pointer to garbage collected vector, mark lattices and
33567 ipcp_orig_node with GTY((skip)).
33568 (ipa_get_param_count): Adjust to descriptors being a pointer.
33569 (ipa_get_param): Likewise.
33570 (ipa_get_type): Likewise.
33571 (ipa_get_param_move_cost): Likewise.
33572 (ipa_set_param_used): Likewise.
33573 (ipa_get_controlled_uses): Likewise.
33574 (ipa_set_controlled_uses): Likewise.
33575 (ipa_is_param_used): Likewise.
33576 (ipa_node_params_t): Move into garbage collector. New methods insert
33578 (ipa_node_params_sum): Annotate wth GTY(()).
33579 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33581 (ipa_load_from_parm_agg): Adjust declaration.
33582 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33583 * ipa-profile.c (ipa_profile): Likewise.
33584 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33585 (ipa_populate_param_decls): Make descriptors parameter garbage
33587 (ipa_dump_param): Adjust to descriptors being a pointer.
33588 (ipa_alloc_node_params): Likewise.
33589 (ipa_initialize_node_params): Likewise.
33590 (load_from_param_1): Make descriptors parameter garbage collected.
33591 (load_from_unmodified_param): Likewise.
33592 (load_from_param): Likewise.
33593 (ipa_load_from_parm_agg): Likewise.
33594 (ipa_node_params::~ipa_node_params): Removed.
33595 (ipa_free_all_node_params): Remove call to delete operator.
33596 (ipa_node_params_t::insert): New.
33597 (ipa_node_params_t::remove): Likewise.
33598 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33599 copy known_csts and known_contexts vectors.
33600 (ipa_read_node_info): Adjust to descriptors being a pointer.
33601 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33603 (ipcp_transform_function): Make descriptors variable garbage
33606 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33608 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33609 * config/i386/avx512dqintrin.h: Ditto.
33610 * config/i386/avx512fintrin.h: Ditto.
33611 * config/i386/i386.c: Handle new builtins.
33612 * config/i386/i386-builtin.def: Add new builtins.
33613 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33614 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33616 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33617 Martin Liska <mliska@suse.cz>
33619 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33620 * asan.c (asan_expand_poison_ifn): Support stores and use
33621 appropriate ASAN report function.
33622 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33623 * internal-fn.def (ASAN_POISON_USE): Declare.
33624 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33625 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33626 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33627 ASAN_POISON calls w/o LHS.
33628 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33629 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33630 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33631 * gimplify.c (asan_poison_variables): Add attribute
33632 use_after_scope_memory to variables that really needs to live
33634 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33635 having the attribute.
33637 2017-01-23 Martin Liska <mliska@suse.cz>
33639 * asan.c (create_asan_shadow_var): New function.
33640 (asan_expand_poison_ifn): Likewise.
33641 * asan.h (asan_expand_poison_ifn): New declaration.
33642 * internal-fn.c (expand_ASAN_POISON): Likewise.
33643 * internal-fn.def (ASAN_POISON): New builtin.
33644 * sanopt.c (pass_sanopt::execute): Expand
33645 asan_expand_poison_ifn.
33646 * tree-inline.c (copy_decl_for_dup_finish): Make function
33648 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33649 * tree-ssa.c (is_asan_mark_p): New function.
33650 (execute_update_addresses_taken): Rewrite local variables
33651 (identified just by use-after-scope as addressable) into SSA.
33653 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33655 * doc/install.texi (Specific): opensource.apple.com uses https
33656 now. Remove trailing slash.
33658 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33660 * README.Portability: Remove note on an Irix compatibility issue.
33662 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33664 * gcov.c (INCLUDE_ALGORITHM): Define.
33665 (INCLUDE_VECTOR): Define.
33666 No longer include <vector> and <algorithm> directly.
33668 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33670 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33672 * doc/invoke.texi (Code Gen Options): Ditto.
33674 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33677 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33679 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33681 rtl-optimization/79125
33682 * cprop.c (local_cprop_pass): Handle cases where we make an
33683 unconditional trap.
33685 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33689 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33690 read from, for big endian.
33692 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33694 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33695 register pauth builtins for LP64 only.
33697 2017-01-20 Marek Polacek <polacek@redhat.com>
33700 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33703 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33705 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33707 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33708 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33709 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33711 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33714 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33715 in big-endian mode when they are not a single duplicated value.
33717 2017-01-20 Richard Biener <rguenther@suse.de>
33719 * BASE-VER: Bump to 7.0.1.
33721 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33723 * omp-low.c (omplow_simd_context): New struct. Use it...
33724 (lower_rec_simd_input_clauses): ...here and...
33725 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33726 references to idx, lane, max_vf, is_simt.
33728 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33730 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33733 2017-01-20 Martin Jambor <mjambor@suse.cz>
33735 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33736 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33738 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33739 (GTFILES): Rename hsa.c to hsa-common.c.
33740 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33741 * hsa-dump.c: Likewise.
33742 * hsa-gen.c: Likewise.
33743 * hsa-regalloc.c: Likewise.
33744 * ipa-hsa.c: Likewise.
33745 * omp-expand.c: Likewise.
33746 * omp-low.c: Likewise.
33747 * toplev.c: Likewise.
33749 2017-01-20 Marek Polacek <polacek@redhat.com>
33752 * doc/invoke.texi: Document -Wduplicated-branches.
33753 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33754 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33755 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33756 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33757 return 0 only when not OEP_LEXICOGRAPHIC.
33758 (fold_build_cleanup_point_expr): Use the expression
33759 location when building CLEANUP_POINT_EXPR.
33760 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33761 * tree.c (add_expr): Handle error_mark_node.
33763 2017-01-20 Martin Liska <mliska@suse.cz>
33766 * tree-profile.c (init_ic_make_global_vars): Do not call
33768 (gimple_init_gcov_profiler): Likewise.
33770 2017-01-20 Martin Liska <mliska@suse.cz>
33773 * cgraph.h (maybe_create_reference): Remove argument and
33775 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33777 * ipa-cp.c (create_specialized_node): Likewise.
33778 * symtab.c (symtab_node::maybe_create_reference): Handle
33779 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33781 2017-01-20 Martin Liska <mliska@suse.cz>
33783 * read-rtl-function.c (function_reader::create_function): Use
33784 build_decl instread of build_decl_stat.
33786 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33788 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33789 * config/i386/avx512dqintrin.h: Ditto.
33790 * config/i386/avx512fintrin.h: Ditto.
33791 * config/i386/i386-builtin-types.def: Add new types.
33792 * config/i386/i386.c: Handle new types.
33793 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33794 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33795 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33796 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33797 (__builtin_ia32_kshiftridi): New.
33798 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33800 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33804 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33805 define to rs6000_init_stack_protect_guard.
33806 (rs6000_init_stack_protect_guard): New function.
33808 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33809 Yunqiang Su <yunqiang.su@imgtec.com>
33811 * config.gcc (supported_defaults): Add madd4.
33812 (with_madd4): Add validation.
33813 (all_defaults): Add madd4.
33814 * config/mips/mips.opt (mmadd4): New option.
33815 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33817 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33819 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33820 (ISA_HAS_FUSED_MADD4): Likewise.
33821 * doc/invoke.texi (-mmadd4): Document the new option.
33822 * doc/install.texi (--with-madd4): Document the new option.
33824 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33826 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33827 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33828 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33829 (aarch64_init_pauth_hint_builtins): New.
33830 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33831 (aarch64_expand_builtin): Expand new builtins.
33833 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33835 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33836 * combine-stack-adj.c (no_unhandled_cfa): Handle
33837 REG_CFA_TOGGLE_RA_MANGLE.
33838 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33839 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33840 info for return address signing.
33841 (aarch64_expand_epilogue): Likewise.
33843 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33845 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33846 * config/aarch64/aarch64-protos.h
33847 (aarch64_return_address_signing_enabled): New declaration.
33848 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33850 (aarch64_expand_prologue): Sign return address before it's pushed onto
33852 (aarch64_expand_epilogue): Authenticate return address fetched from
33854 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33855 (aarch64_attributes): New function attributes for "sign-return-address".
33856 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33857 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33858 ("*do_return"): Generate combined instructions according to key index.
33859 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33860 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33862 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33863 * config/aarch64/aarch64.opt (msign-return-address=): New.
33864 * doc/extend.texi (AArch64 Function Attributes): Documents
33865 "sign-return-address=".
33866 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33868 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33870 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33871 overall option summary.
33873 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33875 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33876 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33877 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33878 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33880 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33882 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33883 -mpower9-minmax by default for -mcpu=power9.
33884 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33885 128-bit floating point.
33887 2017-01-20 Alan Modra <amodra@gmail.com>
33889 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33890 optimizing for size.
33892 2017-01-20 Alan Modra <amodra@gmail.com>
33895 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33896 for strcmp and strncmp from corresponding builtin decl.
33898 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33900 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33901 instead of i386/rtems-64.h.
33902 * config/i386/rtems-64.h: Remove.
33904 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33908 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33910 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33912 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33914 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33915 Change int to HOST_WIDE_INT.
33916 * config/aarch64/aarch64-protos.h
33917 (aarch64_simd_gen_const_vector_dup): Likewise.
33918 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33920 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33922 * langhooks-def.h (lhd_type_for_size): New decl.
33923 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33924 * langhooks.c (lhd_type_for_size): New function, taken from
33927 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33929 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33930 define_bypass for CR latency.
33931 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33932 (power9-alu2): Add define_bypass for CR latency.
33934 (power9-mul): Update insn latency.
33935 (power9-mul-compare): Update insn latency, bypass latency and remove
33938 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33940 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33942 * config/aarch64/aarch64.md
33943 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33944 aarch64_nopcrelative_literal_loads.
33945 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33947 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33949 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33950 TARGET_LOONGSON_3A.
33951 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33953 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33956 * config.gcc (supported_defaults): Add lxc1-sxc1.
33957 (with_lxc1_sxc1): Add validation.
33958 (all_defaults): Add lxc1-sxc1.
33959 * config/mips/mips.opt (mlxc1-sxc1): New option.
33960 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33962 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33963 __mips_no_lxc1_sxc1.
33964 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33965 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33966 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33968 2017-01-19 Richard Biener <rguenther@suse.de>
33970 PR tree-optimization/72488
33971 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33972 sure to restore SSA info.
33973 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33975 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33977 PR rtl-optimization/79121
33978 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33979 of the inner type when shifting an extended value.
33981 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33984 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33985 interposable aliases.
33987 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33990 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33991 Use the evmergelohi instruction.
33992 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33993 (mov_sitf_e500_subreg8_2_be): Likewise.
33994 (mov_sitf_e500_subreg12_2_le): Likewise.
33995 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33996 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33997 (mov_sitf_e500_subreg8_2_le): Likewise.
33998 (mov_sitf_e500_subreg12_2_be): Likewise.
34000 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34002 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34003 attribute from vecsimple to vecperm.
34004 (altivec_vbpermq2): Likewise.
34006 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34009 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34011 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34012 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34013 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34014 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34015 case where N arg is SIZE_MAX.
34016 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34017 (cmpstrsi): Add pattern.
34019 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34021 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34022 __builtin_vec_revb builtins.
34023 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34024 built-in functions to support generation of the ISA 3.0 XXBR<x>
34025 vector byte reverse instructions.
34026 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34027 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34028 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34029 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34030 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34031 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34032 (P9V_BUILTIN_VEC_REVB): Likewise.
34033 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34034 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34035 (p9_xxbrq_v16qi): Likewise.
34036 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34037 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34038 (p9_xxbrh_v8hi): Likewise.
34039 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34040 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34041 vec_revb built-in functions.
34043 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34045 PR rtl-optimization/78952
34046 * config/i386/i386.md (any_extract): New code iterator.
34047 (*insvqi_2): Use any_extract for source operand.
34048 (*insvqi_3): Use any_shiftrt for source operand.
34050 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34052 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34054 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34056 2017-01-18 Matthias Klose <doko@ubuntu.com>
34058 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34060 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34062 * config/rs6000/altivec.h (vec_bperm): Change #define.
34063 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34064 (altivec_vbpermq2): New define_insn.
34065 (altivec_vbpermd): Likewise.
34066 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34067 function interface.
34068 (VBPERMD): Likewise.
34069 (VBPERM): New polymorphic function interface.
34070 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34071 Add entries for P9V_BUILTIN_VEC_VBPERM.
34072 * doc/extend.texi: Add interfaces for vec_bperm.
34074 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34076 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34077 first letter of error messages.
34078 (s390_resolve_overloaded_builtin): Likewise.
34079 * config/s390/s390.c (s390_expand_builtin): Likewise.
34080 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34081 (s390_valid_target_attribute_inner_p): Likewise.
34082 * config/s390/s390.md ("tabort"): Likewise.
34084 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34086 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34087 (ISA_AVOID_DIV_HILO): New macro.
34088 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34089 (ISA_HAS_DDIV): Likewise.
34091 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34093 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34095 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34097 * doc/invoke.texi (fabi-version): Spelling fix.
34099 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34102 * doc/invoke.texi (fabi-version): Mention mangling fix for
34105 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34108 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34110 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34113 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34114 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34116 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34118 (stack_protect_test): Ditto.
34119 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34120 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34122 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34123 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34124 -mstack-protector-guard-offset=.
34125 (RS/6000 and PowerPC Options): Ditto.
34127 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34129 * config/i386/i386.h (MASK_CLASS_P): New define.
34130 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34131 there are no registers from different register sets also when
34132 mask registers are used. Update function comment.
34133 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34134 to (*k/*r) and (*k/*km) alternatives.
34136 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34138 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34139 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34140 (EH_RETURN_HANDLER_RTX): New define.
34141 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34142 Force frame pointer in EH return functions.
34143 (aarch64_expand_epilogue): Add barrier for eh_return.
34144 (aarch64_final_eh_return_addr): Remove.
34145 (aarch64_eh_return_handler_rtx): New function.
34146 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34148 (aarch64_eh_return_handler_rtx): New prototype.
34150 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34152 * config/rs6000/altivec.h (vec_rlmi): New #define.
34153 (vec_vrlnm): Likewise.
34154 (vec_rlnm): Likewise.
34155 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34156 (UNSPEC_VRLNM): Likewise.
34157 (VIlong): New mode iterator.
34158 (altivec_vrl<VI_char>mi): New define_insn.
34159 (altivec_vrl<VI_char>nm): Likewise.
34160 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34162 (VRLDNM): Likewise.
34163 (RLNM): New polymorphic function entry.
34164 (VRLWMI): New monomorphic function entry.
34165 (VRLDMI): Likewise.
34166 (RLMI): New polymorphic function entry.
34167 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34168 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34169 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34172 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34175 * dwarf2out.c (field_byte_offset): Restore the
34176 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34177 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34178 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34181 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34184 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34186 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34189 * dwarf2out.c (add_data_member_location_attribute): For constant
34190 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34191 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34192 DW_AT_byte_size attributes.
34194 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34196 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34197 after forcing to constant memory when the code model is medium.
34199 2017-01-17 Julia Koval <julia.koval@intel.com>
34202 * config/i386/avx512fintrin.h
34203 (_mm512_i32gather_ps): Change __addr type to void const*.
34204 (_mm512_mask_i32gather_ps): Ditto.
34205 (_mm512_i32gather_pd): Ditto.
34206 (_mm512_mask_i32gather_pd): Ditto.
34207 (_mm512_i64gather_ps): Ditto.
34208 (_mm512_mask_i64gather_ps): Ditto.
34209 (_mm512_i64gather_pd): Ditto.
34210 (_mm512_mask_i64gather_pd): Ditto.
34211 (_mm512_i32gather_epi32): Ditto.
34212 (_mm512_mask_i32gather_epi32): Ditto.
34213 (_mm512_i32gather_epi64): Ditto.
34214 (_mm512_mask_i32gather_epi64): Ditto.
34215 (_mm512_i64gather_epi32): Ditto.
34216 (_mm512_mask_i64gather_epi32): Ditto.
34217 (_mm512_i64gather_epi64): Ditto.
34218 (_mm512_mask_i64gather_epi64): Ditto.
34219 (_mm512_i32scatter_ps): Change __addr type to void*.
34220 (_mm512_mask_i32scatter_ps): Ditto.
34221 (_mm512_i32scatter_pd): Ditto.
34222 (_mm512_mask_i32scatter_pd): Ditto.
34223 (_mm512_i64scatter_ps): Ditto.
34224 (_mm512_mask_i64scatter_ps): Ditto.
34225 (_mm512_i64scatter_pd): Ditto.
34226 (_mm512_mask_i64scatter_pd): Ditto.
34227 (_mm512_i32scatter_epi32): Ditto.
34228 (_mm512_mask_i32scatter_epi32): Ditto.
34229 (_mm512_i32scatter_epi64): Ditto.
34230 (_mm512_mask_i32scatter_epi64): Ditto.
34231 (_mm512_i64scatter_epi32): Ditto.
34232 (_mm512_mask_i64scatter_epi32): Ditto.
34233 (_mm512_i64scatter_epi64): Ditto.
34234 (_mm512_mask_i64scatter_epi64): Ditto.
34235 * config/i386/avx512pfintrin.h
34236 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34237 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34238 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34239 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34240 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34241 (_mm512_prefetch_i32scatter_ps): Ditto.
34242 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34243 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34244 (_mm512_prefetch_i64scatter_pd): Ditto.
34245 (_mm512_prefetch_i64scatter_ps): Ditto.
34246 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34247 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34248 * config/i386/avx512vlintrin.h
34249 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34250 (_mm_mmask_i32gather_ps): Ditto.
34251 (_mm256_mmask_i32gather_pd): Ditto.
34252 (_mm_mmask_i32gather_pd): Ditto.
34253 (_mm256_mmask_i64gather_ps): Ditto.
34254 (_mm_mmask_i64gather_ps): Ditto.
34255 (_mm256_mmask_i64gather_pd): Ditto.
34256 (_mm_mmask_i64gather_pd): Ditto.
34257 (_mm256_mmask_i32gather_epi32): Ditto.
34258 (_mm_mmask_i32gather_epi32): Ditto.
34259 (_mm256_mmask_i32gather_epi64): Ditto.
34260 (_mm_mmask_i32gather_epi64): Ditto.
34261 (_mm256_mmask_i64gather_epi32): Ditto.
34262 (_mm_mmask_i64gather_epi32): Ditto.
34263 (_mm256_mmask_i64gather_epi64): Ditto.
34264 (_mm_mmask_i64gather_epi64): Ditto.
34265 (_mm256_i32scatter_ps): Change __addr type to void*.
34266 (_mm256_mask_i32scatter_ps): Ditto.
34267 (_mm_i32scatter_ps): Ditto.
34268 (_mm_mask_i32scatter_ps): Ditto.
34269 (_mm256_i32scatter_pd): Ditto.
34270 (_mm256_mask_i32scatter_pd): Ditto.
34271 (_mm_i32scatter_pd): Ditto.
34272 (_mm_mask_i32scatter_pd): Ditto.
34273 (_mm256_i64scatter_ps): Ditto.
34274 (_mm256_mask_i64scatter_ps): Ditto.
34275 (_mm_i64scatter_ps): Ditto.
34276 (_mm_mask_i64scatter_ps): Ditto.
34277 (_mm256_i64scatter_pd): Ditto.
34278 (_mm256_mask_i64scatter_pd): Ditto.
34279 (_mm_i64scatter_pd): Ditto.
34280 (_mm_mask_i64scatter_pd): Ditto.
34281 (_mm256_i32scatter_epi32): Ditto.
34282 (_mm256_mask_i32scatter_epi32): Ditto.
34283 (_mm_i32scatter_epi32): Ditto.
34284 (_mm_mask_i32scatter_epi32): Ditto.
34285 (_mm256_i32scatter_epi64): Ditto.
34286 (_mm256_mask_i32scatter_epi64): Ditto.
34287 (_mm_i32scatter_epi64): Ditto.
34288 (_mm_mask_i32scatter_epi64): Ditto.
34289 (_mm256_i64scatter_epi32): Ditto.
34290 (_mm256_mask_i64scatter_epi32): Ditto.
34291 (_mm_i64scatter_epi32): Ditto.
34292 (_mm_mask_i64scatter_epi32): Ditto.
34293 (_mm256_i64scatter_epi64): Ditto.
34294 (_mm256_mask_i64scatter_epi64): Ditto.
34295 (_mm_i64scatter_epi64): Ditto.
34296 (_mm_mask_i64scatter_epi64): Ditto.
34297 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34298 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34299 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34300 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34301 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34302 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34303 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34304 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34305 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34306 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34307 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34308 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34309 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34310 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34311 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34312 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34313 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34314 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34315 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34316 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34317 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34318 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34319 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34320 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34321 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34322 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34323 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34324 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34325 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34326 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34327 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34328 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34329 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34330 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34331 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34332 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34333 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34334 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34335 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34336 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34337 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34338 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34339 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34340 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34341 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34342 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34343 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34344 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34345 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34346 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34347 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34348 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34349 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34350 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34351 definitions accordingly.
34353 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34354 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34357 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34360 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34363 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34364 hard regs for inner regno.
34366 2017-01-17 Martin Liska <mliska@suse.cz>
34369 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34370 assumption and add comment.
34372 2017-01-17 Nathan Sidwell <nathan@acm.org>
34374 * ipa-visibility.c (localize_node): New function, broken out of ...
34375 (function_and_variable_visibility): ... here. Call it.
34377 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34379 PR middle-end/77445
34380 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34381 correctly set frequency of oudgoing edge.
34382 (duplicate_thread_path): Fix profile updating.
34384 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34387 * configure.ac: Add GCC_BASE_VER.
34388 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34389 version from BASE-VER file.
34390 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34391 (gcc.o): Depend on $(BASEVER).
34392 * common.opt (dumpfullversion): New option.
34393 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34394 * doc/invoke.texi: Document -dumpfullversion.
34395 * doc/install.texi: Document --with-gcc-major-version-only.
34396 * configure: Regenerated.
34398 2017-01-17 Richard Biener <rguenther@suse.de>
34400 PR tree-optimization/71433
34401 * tree-vrp.c (register_new_assert_for): Merge same asserts
34402 on all incoming edges.
34403 (process_assert_insertions_for): Handle insertions at the
34406 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34408 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34409 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34411 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34414 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34417 2017-01-17 Alan Modra <amodra@gmail.com>
34420 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34421 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34422 symbolic stack limit when pic.
34424 2017-01-16 Martin Sebor <msebor@redhat.com>
34426 PR tree-optimization/78608
34427 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34429 2017-01-16 Jeff Law <law@redhat.com>
34432 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34433 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34434 for several include directories that may be relative to sysroot.
34435 * config/i386/x-mingw32 (gplus_includedir): Define.
34436 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34437 (native_system_includedir): Likewise.
34438 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34439 override if TARGET_SYSTEM_ROOT is defined.
34440 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34442 PR tree-optimization/79090
34443 PR tree-optimization/33562
34444 PR tree-optimization/61912
34445 PR tree-optimization/77485
34446 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34447 and computed trims into the dump file.
34449 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34451 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34453 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34456 * gimplify.c (gimplify_init_constructor): If want_value and
34457 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34461 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34462 sequence. Formatting fixes.
34463 (doloop_optimize): Formatting fixes.
34466 * gcc.c (debug_level_greater_than_spec_func): New function.
34467 (static_spec_functions): Add debug-level-gt spec function.
34468 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34470 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34471 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34472 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34473 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34474 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34475 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34477 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34479 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34480 QImode fixups to general and mask registers only.
34482 2017-01-16 Carl Love <cel@us.ibm.com>
34484 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34485 for built-in functions
34486 vector signed char vec_nabs (vector signed char)
34487 vector signed short vec_nabs (vector signed short)
34488 vector signed int vec_nabs (vector signed int)
34489 vector signed long long vec_nabs (vector signed long long)
34490 vector float vec_nabs (vector float)
34491 vector double vec_nabs (vector double)
34492 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34494 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34495 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34496 * doc/extend.texi: Update the documentation file for the new built-in
34499 2017-01-16 Martin Sebor <msebor@redhat.com>
34501 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34504 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34506 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34507 UNSPEC_VSX__XXSPLTD to require special splat handling.
34509 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34512 * system.h: Poison strndup.
34514 2017-01-16 Alan Modra <amodra@gmail.com>
34517 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34520 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34522 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34524 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34526 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34527 call recog here. Assert that INSN_CODE (insn) is non-negative.
34529 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34532 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34534 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34535 in the currently scheduled RTL fragment.
34537 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34539 PR rtl-optimization/78751
34540 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34543 2017-01-14 Jeff Law <law@redhat.com>
34545 PR tree-optimization/79090
34546 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34547 variable length stores.
34548 (compute_trims): Delete dead assignment to *trim_tail.
34549 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34552 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34554 PR rtl-optimization/78626
34555 PR rtl-optimization/78727
34556 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34557 of a block, and split such blocks after everything else is finished.
34559 2017-01-14 Alan Modra <amodra@gmail.com>
34562 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34563 target legitimate_combined_insn.
34564 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34565 (rs6000_legitimate_combined_insn): New function.
34566 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34568 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34569 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34570 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34572 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34574 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34576 2017-01-13 Jeff Law <law@redhat.com>
34578 PR tree-optimization/33562
34579 PR tree-optimization/61912
34580 PR tree-optimization/77485
34581 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34583 (delete_dead_assignment): Likewise.
34584 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34585 statement to delete_dead_call and delete_dead_assignment.
34587 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34590 * substring-locations.c (format_warning_va): Strengthen case 1 so
34591 that both endpoints of the substring must be within the format
34592 range for just the substring to be printed.
34594 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34596 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34597 * config/i386/i386.c (ix86_target_string): Add missing options
34598 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34599 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34600 flags_other and ix86_target_other to flags2_other. Display unknown
34602 (ix86_valid_target_attribute_inner_p): Add missing options and
34603 reorder options by implied ISAs, as in ix86_target_string.
34605 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34607 * hash-table.h (hash_table::too_empty_p): New function.
34608 (hash_table::expand): Use it.
34609 (hash_table::traverse): Likewise.
34610 (hash_table::empty_slot): Use sizeof (value_type) instead of
34611 sizeof (PTR) to convert bytes to elements. Shrink the table
34612 if the current size is excessive for the current number of
34615 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34617 * ira-costs.c (record_reg_classes): Break from the inner loop
34618 early once alt_fail is known to be true. Update outer loop
34619 handling accordingly.
34621 2017-01-13 Jeff Law <law@redhat.com>
34623 * tree-ssa-dse.c (decrement_count): New function.
34624 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34625 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34626 when we know the partially dead statement is a mem* function.
34628 PR tree-optimization/61912
34629 PR tree-optimization/77485
34630 * tree-ssa-dse.c: Include expr.h.
34631 (maybe_trim_constructor_store): New function.
34632 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34634 PR tree-optimization/33562
34635 PR tree-optimization/61912
34636 PR tree-optimization/77485
34637 * doc/invoke.texi: Document new dse-max-object-size param.
34638 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34639 * tree-ssa-dse.c: Include params.h.
34640 (dse_store_status): New enum.
34641 (initialize_ao_ref_for_dse): New, partially extracted from
34643 (valid_ao_ref_for_dse, normalize_ref): New.
34644 (setup_live_bytes_from_ref, compute_trims): Likewise.
34645 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34646 (maybe_trim_partially_dead_store): Likewise.
34647 (maybe_trim_complex_store): Likewise.
34648 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34649 Track what bytes live from the original store. Return tri-state
34650 for dead, partially dead or live.
34651 (dse_dom_walker): Add constructor, destructor and new private members.
34652 (delete_dead_call, delete_dead_assignment): New extracted from
34654 (dse_optimize_stmt): Make a member of dse_dom_walker.
34655 Use initialize_ao_ref_for_dse.
34657 PR tree-optimization/33562
34658 PR tree-optimization/61912
34659 PR tree-optimization/77485
34660 * sbitmap.h (bitmap_count_bits): Prototype.
34661 (bitmap_clear_range, bitmap_set_range): Likewise.
34662 * sbitmap.c (bitmap_clear_range): New function.
34663 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34665 2017-01-13 Martin Liska <mliska@suse.cz>
34668 * function.c (set_cfun): Add new argument force.
34669 * function.h (set_cfun): Likewise.
34670 * ipa-inline-transform.c (inline_call): Use the function when
34671 strict alising from is dropped for function we inline to.
34673 2017-01-13 Richard Biener <rguenther@suse.de>
34675 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34676 for dumping GIMPLE INTEGER_CSTs.
34678 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34680 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34681 to 201112L since C++17.
34683 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34686 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34687 if -fsanitize=kernel-address is present.
34689 2017-01-13 Richard Biener <rguenther@suse.de>
34691 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34692 as _Literal ( type ) number in case usual suffixes do not
34693 preserve all information.
34695 2017-01-13 Richard Biener <rguenther@suse.de>
34697 PR tree-optimization/77283
34698 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34699 and ssa-iterators.h.
34700 (is_feasible_trace): Implement a cost model based on joiner
34703 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34706 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34707 char or short to __float128/_Float128 directly.
34709 2017-01-12 Martin Sebor <msebor@redhat.com>
34711 to -Wformat-overflow.
34712 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34713 (min_bytes_remaining): Same.
34714 (get_string_length): Same.
34715 (format_string): Same.
34716 (format_directive): Same.
34718 (pass_sprintf_length::handle_gimple_call): Same.
34720 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34722 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34723 info.nowrite calls with no lhs that can't throw. Return bool
34724 whether gsi_remove has been called or not.
34725 (pass_sprintf_length::handle_gimple_call): Return bool whether
34726 try_substitute_return_value called gsi_remove. Formatting fix.
34727 (pass_sprintf_length::execute): Don't use gsi_remove if
34728 handle_gimple_call returned true.
34731 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34732 be removed due to side-effects, don't remove following barrier nor
34733 turn the successor edge into fallthru edge.
34735 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34738 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34739 element-reversing loads and stores as not swappable.
34741 2017-01-12 Nathan Sidwell <nathan@acm.org>
34742 Nicolai Stange <nicstange@gmail.com>
34744 * combine.c (try_combine): Don't ignore result of overlap checking
34745 loop. Combine overlap & asm check into single loop.
34747 2017-01-12 Richard Biener <rguenther@suse.de>
34749 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34750 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34752 2017-01-12 Richard Biener <rguenther@suse.de>
34754 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34755 and TS_TARGET_OPTION directly derive from TS_BASE.
34756 * tree-core.h (tree_optimization_option): Derive from tree_base.
34757 (tree_target_option): Likewise.
34759 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34761 * config/i386/i386.c (memory_address_length): Increase len
34762 only when rip_relative_addr_p returns false.
34764 2017-01-11 Julia Koval <julia.koval@intel.com>
34766 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34767 (OPTION_MASK_ISA_SGX_SET): New.
34768 (ix86_handle_option): Handle OPT_msgx.
34769 * config.gcc: Added sgxintrin.h.
34770 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34771 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34772 * config/i386/i386.c (ix86_target_string): Add -msgx.
34774 (ix86_option_override_internal): Handle new options.
34775 (ix86_valid_target_attribute_inner_p): Add sgx.
34776 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34777 * config/i386/i386.opt: Add msgx.
34778 * config/i386/sgxintrin.h: New file.
34779 * config/i386/x86intrin.h: Add sgxintrin.h.
34781 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34784 * fold-const.c (maybe_nonzero_address): Return 1 for function
34786 (tree_single_nonzero_warnv_p): Don't handle function local objects
34790 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34793 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34796 * opts.c: Include "spellcheck.h"
34797 (struct string_fragment): New struct.
34798 (struct edit_distance_traits<const string_fragment &>): New
34800 (get_closest_sanitizer_option): New function.
34801 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34803 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34805 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34807 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34808 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34809 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34810 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34811 for initial die_offset if dwarf_split_debug_info.
34812 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34813 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34815 (output_skeleton_debug_sections): Formatting fix. Use
34816 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34817 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34819 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34821 * config/arm/cortex-a53.md: Add bypasses for
34822 cortex_a53_r2f_cvt.
34823 (cortex_a53_r2f): Only use for transfers.
34824 (cortex_a53_f2r): Likewise.
34825 (cortex_a53_r2f_cvt): Add reservation for conversions.
34826 (cortex_a53_f2r_cvt): Likewise.
34828 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34830 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34831 to all inlined functions, change static to extern.
34833 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34836 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34838 (arm_assemble_integer): Likewise.
34840 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34842 * config.gcc: Use new awk script to check CPU, FPU and architecture
34843 parameters for --with-... options.
34844 * config/arm/parsecpu.awk: New file
34845 * config/arm/arm-cpus.in: New file.
34846 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34848 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34850 * config/arm/t-arm: Update dependency rules.
34851 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34852 of processing .def files.
34853 * config/arm/genopt.sh: Deleted.
34854 * config/arm/gentune.sh: Deleted.
34855 * config/arm/arm-cores.def: Deleted.
34856 * config/arm/arm-arches.def: Deleted.
34857 * config/arm/arm-fpus.def: Deleted.
34858 * config/arm/arm-tune.md: Regenerated.
34859 * config/arm/arm-tables.opt: Regenerated.
34860 * config/arm/arm-cpu.h: New generated file.
34861 * config/arm/arm-cpu-data.h: New generated file.
34862 * config/arm/arm-cpu-cdata.h: New generated file.
34864 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34867 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34869 (input_varpool_node): Unpack dynamically_initialized bit.
34871 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34873 PR rtl-optimization/79032
34874 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34875 the alignment of the adjusted memory reference against that of MODE,
34876 instead of the alignment of the original memory reference.
34878 2017-01-11 Martin Jambor <mjambor@suse.cz>
34880 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34882 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34883 decorated functions.
34885 2017-01-11 Richard Biener <rguenther@suse.de>
34887 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34888 set range/nonnull info for PHI results. Do not set it on
34889 stmts marked for removal.
34891 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34893 * expr.c (store_field): In the bitfield case, fetch the return value
34894 from the registers before applying a single big-endian adjustment.
34895 Always do a final load for a BLKmode value not larger than a word.
34897 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34900 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34901 that we correctly handle column numbers greater than
34902 LINE_MAP_MAX_COLUMN_NUMBER.
34904 2017-01-10 Martin Sebor <msebor@redhat.com>
34906 PR middle-end/78245
34907 * gimple-ssa-sprintf.c (get_destination_size): Call
34908 {init,fini}object_sizes.
34909 * tree-object-size.c (addr_object_size): Adjust.
34910 (pass_through_call): Adjust.
34911 (pass_object_sizes::execute): Adjust.
34912 * tree-object-size.h (fini_object_sizes): Declare.
34914 2017-01-10 Martin Sebor <msebor@redhat.com>
34916 PR tree-optimization/78775
34917 * builtins.c (get_size_range): Move...
34918 * calls.c: ...to here.
34919 (alloc_max_size): Accept zero argument.
34920 (operand_signed_p): Remove.
34921 (maybe_warn_alloc_args_overflow): Call get_size_range.
34922 * calls.h (get_size_range): Declare.
34924 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34926 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34927 from TI's devices.csv file as of September 2016.
34928 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34930 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34932 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34933 * doc/invoke.texi: Likewise.
34934 * doc/md.texi: Likewise.
34935 * doc/objc.texi: Likewise.
34937 2017-01-10 Joshua Conner <joshconner@google.com>
34939 * config/arm/fuchsia-elf.h: New file.
34940 * config/fuchsia.h: New file.
34941 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34942 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34944 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34946 2016-01-10 Richard Biener <rguenther@suse.de>
34948 PR tree-optimization/79034
34949 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34950 Propagate out degenerate PHIs in the joiner.
34952 2017-01-10 Martin Liska <mliska@suse.cz>
34954 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34955 (sort_congruence_classes_by_decl_uid): Likewise.
34956 (sort_congruence_class_groups_by_decl_uid): Likewise.
34957 (sem_item_optimizer::merge_classes): Sort class, groups in these
34958 classes and members in the groups by DECL_UID of declarations.
34959 This would make merge operations stable.
34961 2017-01-10 Martin Liska <mliska@suse.cz>
34963 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34964 usage of m_classes_vec.
34965 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34966 (sem_item_optimizer::get_group_by_hash): Likewise.
34967 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34968 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34969 (sem_item_optimizer::verify_classes): Likewise.
34970 (sem_item_optimizer::process_cong_reduction): Likewise.
34971 (sem_item_optimizer::dump_cong_classes): Likewise.
34972 (sem_item_optimizer::merge_classes): Likewise.
34973 * ipa-icf.h (congruence_class_hash): Rename from
34974 congruence_class_group_hash. Remove declaration of m_classes_vec.
34976 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34978 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34979 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34980 * config.gcc: Add avx512vpopcntdqintrin.h.
34981 * config/i386/avx512vpopcntdqintrin.h: New.
34982 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34983 * config/i386/i386-builtin-types.def: Add new types.
34984 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34985 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34986 __builtin_ia32_vpopcountq_v8di_mask): New.
34987 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34988 __AVX512VPOPCNTDQ__.
34989 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34990 (PTA_AVX512VPOPCNTDQ): Define.
34991 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34992 TARGET_AVX512VPOPCNTDQ_P): Define.
34993 * config/i386/i386.opt: Add mavx512vpopcntdq.
34994 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34995 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34997 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34999 PR middle-end/77484
35000 * predict.def (PRED_CALL): Set to 67.
35002 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35004 * expr.c (store_field): In the bitfield case, if the value comes from
35005 a function call and is of an aggregate type returned in registers, do
35006 not modify the field mode; extract the value in all cases if the mode
35007 is BLKmode and the size is not larger than a word.
35009 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35012 * config/i386/cpuid.h: Fix undefined behavior.
35014 2017-01-04 Jeff Law <law@redhat.com>
35016 PR tree-optimization/79007
35017 PR tree-optimization/67955
35018 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35019 conservative for pt.null when flag_non_call_exceptions is on.
35021 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35023 PR translation/79019
35024 PR translation/79020
35025 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35026 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35027 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35029 * config/avr/avr.opt (maccumulate-args): Likewise.
35030 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35031 * common.opt (freport-bug): Likewise.
35032 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35033 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35034 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35035 translatable string.
35036 * config/i386/i386.c (function_value_32): Likewise.
35037 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35038 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35040 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35041 * common/config/msp430/msp430-common.c (msp430_handle_option):
35043 * symtab.c (symtab_node::verify_base): Likewise.
35044 * opts.c (set_debug_level): Likewise.
35045 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35046 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35047 missing whitespace to translatable strings.
35048 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35049 * config/sh/superh.h: Likewise.
35050 * config/i386/xopintrin.h: Likewise.
35051 * config/i386/znver1.md: Likewise.
35052 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35053 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35054 * double-int.h (struct double_int): Likewise.
35055 * double-int.c (div_and_round_double): Likewise.
35056 * wide-int.cc: Likewise.
35057 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35058 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35059 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35060 (crossjumps_occurred): ... this.
35061 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35064 PR tree-optimization/78899
35065 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35066 returning bool return struct loop *, NULL for failure and the new
35068 (versionable_outer_loop_p): Don't version outer loop if it has
35069 dont_vectorized bit set.
35070 (tree_if_conversion): When versioning outer loop, ensure
35071 tree_if_conversion is performed also on the inner loop of the
35072 non-vectorizable outer loop copy.
35073 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35074 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35075 prevent vectorization of it.
35076 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35077 the outer loop vectorization of the non-scalar version is attempted
35078 before vectorization of the inner loop in scalar version. If
35079 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35080 vectorization of its inner loop.
35081 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35082 has 2 inner loops, rename also on edges from bb whose single pred
35083 is outer_loop->header. Fix typo in function comment.
35085 2017-01-09 Martin Sebor <msebor@redhat.com>
35088 * asan.c (asan_emit_stack_protection): Increase local buffer size
35089 to avoid snprintf truncation warning.
35091 2017-01-09 Andrew Pinski <apinski@cavium.com>
35093 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35094 to reference thunderx2t99 for the tuning structure
35095 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35097 (thunderx2t99_extra_costs): This.
35098 * config/aarch64/aarch64-tune.md: Regenerate.
35099 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35100 (vulcan_addrcost_table): This.
35101 (vulcan_regmove_cost): Rename to ...
35102 (thunderx2t99_regmove_cost): This.
35103 (vulcan_vector_cost): Rename to ...
35104 (thunderx2t99_vector_cost): this.
35105 (vulcan_branch_cost): Rename to ...
35106 (thunderx2t99_branch_cost): This.
35107 (vulcan_tunings): Rename to ...
35108 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35109 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35111 2017-01-09 Martin Jambor <mjambor@suse.cz>
35115 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35116 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35117 (propagate_vr_accross_jump_function): Use the above function for all
35118 value range computations for pass-through jump functions and type
35119 converasion from explicit value range values.
35120 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35121 parameters from TYPE_ARG_TYPES.
35122 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35123 (ipa_write_node_info): Stream type of the actual argument.
35124 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35126 2017-01-09 Martin Liska <mliska@suse.cz>
35129 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35130 (lookup_compiler): Do not show error message with have_E.
35132 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35134 PR tree-optimization/78938
35135 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35136 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35137 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35138 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35141 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35143 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35146 2017-01-09 Richard Biener <rguenther@suse.de>
35148 PR tree-optimization/78997
35149 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35150 name condition properly.
35152 2017-01-09 Richard Biener <rguenther@suse.de>
35155 * dwarf2out.c (is_cxx): New overload with context.
35156 (is_naming_typedef_decl): Use it.
35158 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35160 * invoke.texi (Option Summary): Correct spacing in option lists
35161 and add line breaks to fix over-long lines.
35163 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35165 PR middle-end/17660
35167 * extend.texi (Common Variable Attributes): Add xref to GCC
35168 Internals manual to explain mode attribute keywords.
35170 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35173 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35174 and Preprocessor Options.
35175 (Options for Linking): Document -pthread here....
35176 (RS/6000 and PowerPC Options): ...not here.
35177 (Solaris 2 Options): ...or here.
35178 * doc/cppopts.texi: Document -pthread.
35180 2017-01-08 Martin Sebor <msebor@redhat.com>
35182 PR middle-end/77708
35183 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35184 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35185 New member functions.
35186 (format_directive): Used them.
35188 (pass_sprintf_length::handle_gimple_call): Same.
35189 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35190 to avoid truncation for any argument.
35191 (extract_affine_mul): Same.
35192 * tree.c (get_file_function_name): Same.
35194 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35196 PR middle-end/77484
35197 * predict.def (PRED_INDIR_CALL): Set to 86.
35199 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35201 PR preprocessor/54124
35202 * doc/cppopts.texi: Reformat -d subtable to list the full name
35203 of the options. Add cross-reference to the docs for the general
35204 compiler -d options.
35205 * doc/invoke.texi (Developer Options): Add cross-reference to the
35206 preprocessor-specific -d option documentation.
35208 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35210 PR preprocessor/13498
35211 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35212 redudant material, and reflect new command-line options.
35213 (System Headers): Likewise.
35215 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35217 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35218 -isystem, and -idirafter. Copy-edit.
35219 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35220 default for -ftrack-macro-expansion. Delete obsolete and
35221 badly-formatted implementation details about -fdebug-cpp output.
35222 * doc/cppwarnopts.texi: Copy-edit.
35224 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35227 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35228 that the transition from a max line width >= 1<<10 to narrower
35229 lines works correctly.
35231 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35233 * doc/options.texi (PerFunction): New.
35234 * opt-functions.awk (switch_flags): Map both Optimization and
35235 PerFunction to CL_OPTIMIZATION.
35236 * opth-gen.awk: Test for PerFunction flag along with
35238 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35239 it only when the latter is present. Skip those that don't in
35240 the hash function generator.
35241 * common.opt (fvar-tracking): Mark as PerFunction instead of
35243 (fvar-tracking-assignments): Likewise.
35244 (fvar-tracking-assignments-toggle): Likewise.
35245 (fvar-tracking-uninit): Likewise.
35247 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35249 PR translation/79018
35250 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35253 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35256 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35257 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35258 TARGET_LONG_JUMP_TABLE_OFFSETS.
35259 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35260 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35261 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35262 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35263 * config/m68k/m68k.md (tablejump expander): Likewise.
35264 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35265 TARGET_LONG_JUMP_TABLE_OFFSETS.
35266 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35267 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35269 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35270 David Holsgrove <david.holsgrove@xilinx.com>
35272 * common/config/microblaze/microblaze-common.c
35273 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35274 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35276 * config/microblaze/microblaze.c (microblaze_must_save_register)
35277 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35279 (microblaze_eh_return): New function.
35280 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35281 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35282 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35283 * config/microblaze/microblaze.md (eh_return): New pattern.
35285 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35287 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35288 GCC_DIAGNOSTIC_STRINGIFY): Define.
35290 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35292 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35294 * config/arm/arm.md (<mcrr>): New.
35296 * config/arm/arm.c (arm_arch5te): New.
35297 (arm_option_override): Set arm_arch5te.
35298 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35300 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35301 (arm_mcrr_qualifiers): ... this. New.
35302 (MRRC_QUALIFIERS): Define to...
35303 (arm_mrrc_qualifiers): ... this. New.
35304 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35306 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35307 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35308 (MRRCI, mrrc, MRRC): New.
35309 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35310 VUNSPEC_MRRC2): New.
35312 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35314 * config/arm/arm.md (<mcr>): New.
35316 * config/arm/arm.c (arm_coproc_builtin_available): Add
35317 support for mcr, mrc, mcr2 and mrc2.
35318 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35319 (arm_mcr_qualifiers): ... this. New.
35320 (MRC_QUALIFIERS): Define to ...
35321 (arm_mrc_qualifiers): ... this. New.
35322 (MCR_QUALIFIERS): Define to ...
35323 (arm_mcr_qualifiers): ... this. New.
35324 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35326 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35327 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35328 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35329 VUNSPEC_MRC2): New.
35331 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35333 * config/arm/arm.md (*ldc): New.
35337 * config/arm/arm.c (arm_coproc_builtin_available): Add
35338 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35339 (arm_coproc_ldc_stc_legitimate_address): New.
35340 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35341 'qualifier_const_pointer'.
35342 (LDC_QUALIFIERS): Define to...
35343 (arm_ldc_qualifiers): ... this. New.
35344 (STC_QUALIFIERS): Define to...
35345 (arm_stc_qualifiers): ... this. New.
35346 * config/arm/arm-protos.h
35347 (arm_coproc_ldc_stc_legitimate_address): New.
35348 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35349 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35350 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35351 stc2, stcl, stc2l): New.
35352 * config/arm/constraints.md (Uz): New.
35353 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35354 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35355 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35356 VUNSPEC_STC2L): New.
35358 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35360 * config/arm/arm.md (<cdp>): New.
35361 * config/arm/arm.c (neon_const_bounds): Rename this ...
35362 (arm_const_bounds): ... this.
35363 (arm_coproc_builtin_available): New.
35364 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35365 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35366 (CDP_QUALIFIERS): Define to...
35367 (arm_cdp_qualifiers): ... this. New.
35369 (arm_expand_builtin_args): Add case for 6 arguments.
35370 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35371 (arm_const_bounds): ... this.
35372 (arm_coproc_builtin_available): New.
35373 * config/arm/arm_acle.h (__arm_cdp): New.
35375 * config/arm/arm_acle_builtins.def (cdp): New.
35377 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35378 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35379 'arm_const_bounds'.
35380 * config/arm/types.md (coproc): New.
35381 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35382 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35383 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35384 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35386 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35388 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35389 (UBINOP_QUALIFIERS): New.
35391 (acle_builtin_data): New. Change comment.
35392 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35393 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35394 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35395 arm_acle_builtins.def.
35396 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35397 (arm_init_acle_builtins): New.
35398 (CRC32_BUILTIN): Remove.
35399 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35400 crc32cb, crc32ch and crc32cw.
35401 (arm_init_crc32_builtins): Remove.
35402 (arm_init_builtins): Use arm_init_acle_builtins rather
35403 than arm_init_crc32_builtins.
35404 (arm_expand_acle_builtin): New.
35405 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35406 * config/arm/arm_acle_builtins.def: New.
35408 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35410 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35411 (arm_builtin_datum): ... this.
35412 (arm_init_neon_builtin): Rename to ...
35413 (arm_init_builtin): ... this. Add a new parameters PREFIX
35414 and USE_SIG_IN_NAME.
35415 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35416 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35417 'arm_builtin_datum'.
35418 (arm_init_vfp_builtins): Likewise.
35419 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35420 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35421 (arm_expand_neon_args): Rename to ...
35422 (arm_expand_builtin_args): ... this. Rename builtin_arg
35423 enum values and differentiate between ARG_BUILTIN_MEMORY
35424 and ARG_BUILTIN_NEON_MEMORY.
35425 (arm_expand_neon_builtin_1): Rename to ...
35426 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35427 values, arm_expand_builtin_args and add bool parameter NEON.
35428 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35429 (arm_expand_vfp_builtin): Likewise.
35430 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35432 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35434 PR middle-end/77484
35435 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35436 * predict.c (tree_estimate_probability_bb): Reverse direction of
35437 polymorphic call predictor.
35439 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35441 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35442 (determine_pass_name_match): ...this new function and...
35443 (should_skip_pass_p): ...this new function.
35445 2017-01-06 Nathan Sidwell <nathan@acm.org>
35447 * ipa-visibility.c (function_and_variable_visibility): Reformat
35448 comments and long lines. Remove extrneous if.
35449 * symtab.c (symtab_node::make_decl_local): Fix code format.
35450 (symtab_node::set_section_for_node): Fix comment typo.
35452 2017-01-06 Martin Liska <mliska@suse.cz>
35455 * lra-constraints.c: Rename invariant to lra_invariant.
35456 * predict.c (set_even_probabilities): Initialize e to NULL.
35458 2017-01-05 Martin Sebor <msebor@redhat.com>
35460 PR tree-optimization/78910
35461 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35462 (format_integer): Correct off-by-one error in the handling
35463 of precision with negative numbers in signed conversions..
35465 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35467 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35469 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35471 PR tree-optimization/71016
35472 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35473 factor_out_conditional_conversion. Formatting fix.
35474 (factor_out_conditional_conversion): Add cond_stmt argument.
35475 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35476 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35479 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35481 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35482 read-rtl-function.o, and selftest-rtl.o.
35483 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35484 (selftest::aarch64_test_loading_full_dump): New function.
35485 (selftest::aarch64_run_selftests): New function.
35486 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35487 selftest::aarch64_run_selftests.
35488 * config/i386/i386.c
35489 (selftest::ix86_test_loading_dump_fragment_1): New function.
35490 (selftest::ix86_test_loading_call_insn): New function.
35491 (selftest::ix86_test_loading_full_dump): New function.
35492 (selftest::ix86_test_loading_unspec): New function.
35493 (selftest::ix86_run_selftests): Call the new functions.
35494 * emit-rtl.c (maybe_set_max_label_num): New function.
35495 * emit-rtl.h (maybe_set_max_label_num): New decl.
35496 * function.c (instantiate_decls): Guard call to
35497 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35498 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35500 * gensupport.c (gen_reader::gen_reader): Pass "false"
35501 for new "compact" param of rtx_reader.
35502 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35503 rather than an empty string for NULL strings.
35504 * read-md.c: Potentially include config.h rather than bconfig.h.
35505 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35506 (have_error): New global, copied from errors.c.
35507 (md_reader::read_name): Rename to...
35508 (md_reader::read_name_1): ...this, adding "out_loc" param,
35509 and converting "missing name or number" to returning false, rather
35511 (md_reader::read_name): Reimplement in terms of read_name_1.
35512 (md_reader::read_name_or_nil): New function.
35513 (md_reader::read_string): Handle "(nil)" by returning NULL.
35514 (md_reader::md_reader): Add new param "compact".
35515 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35516 (md_reader::read_file): New method.
35517 * read-md.h (md_reader::md_reader): Add new param "compact".
35518 (md_reader::read_file): New method.
35519 (md_reader::is_compact): New accessor.
35520 (md_reader::read_name): Convert return type from void to file_location.
35521 (md_reader::read_name_or_nil): New decl.
35522 (md_reader::read_name_1): New decl.
35523 (md_reader::m_compact): New field.
35524 (noop_reader::noop_reader): Pass "false" for new "compact" param
35526 (rtx_reader::rtx_reader): Add new "compact" param.
35527 (rtx_reader::read_rtx_operand): Make virtual and convert return
35528 type from void to rtx.
35529 (rtx_reader::read_until): New decl.
35530 (rtx_reader::handle_any_trailing_information): New virtual function.
35531 (rtx_reader::postprocess): New virtual function.
35532 (rtx_reader::finalize_string): New virtual function.
35533 (rtx_reader::m_in_call_function_usage): New field.
35534 (rtx_reader::m_reuse_rtx_by_id): New field.
35535 * read-rtl-function.c: New file.
35536 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35537 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35538 (selftest::verify_three_block_rtl_cfg): New decl.
35539 * read-rtl-function.h: New file.
35540 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35541 For host, include function.h, memmodel.h, and emit-rtl.h.
35542 (one_time_initialization): New function.
35543 (struct compact_insn_name): New struct.
35544 (compact_insn_names): New array.
35545 (find_code): Handle insn codes in compact dumps.
35546 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35547 (bind_subst_iter_and_attr): Likewise.
35548 (add_condition_to_string): Likewise.
35549 (add_condition_to_rtx): Likewise.
35550 (apply_attribute_uses): Likewise.
35551 (add_current_iterators): Likewise.
35552 (apply_iterators): Likewise.
35553 (initialize_iterators): Guard usage of apply_subst_iterator with
35554 #ifdef GENERATOR_FILE.
35555 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35556 (md_reader::read_mapping): Likewise.
35557 (add_define_attr_for_define_subst): Likewise.
35558 (add_define_subst_attr): Likewise.
35559 (read_subst_mapping): Likewise.
35560 (check_code_iterator): Likewise.
35561 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35563 (one_time_initialization): New function.
35564 (rtx_reader::read_until): New method.
35565 (read_flags): New function.
35566 (parse_reg_note_name): New function.
35567 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35568 Handle reuse_rtx ids.
35569 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35570 Add parsing support for RTL dumps, mirroring the special-cases in
35571 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35572 values, and calling handle_any_trailing_information.
35573 (rtx_reader::read_rtx_operand): Convert return type from void
35574 to rtx, returning return_rtx. Handle case 'e'. Call
35575 finalize_string on XSTR and XTMPL fields.
35576 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35577 "(nil)" values were omitted. Call the postprocess vfunc on the
35579 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35580 class ctor. Initialize m_in_call_function_usage. Call
35581 one_time_initialization.
35582 * rtl-tests.c (selftest::test_uncond_jump): Call
35583 set_new_first_and_last_insn.
35584 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35585 * selftest-rtl.c: New file.
35586 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35587 (selftest::get_insn_by_uid): New decl.
35588 * selftest-run-tests.c (selftest::run_tests): Call
35589 read_rtl_function_c_tests.
35590 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35591 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35594 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35596 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35597 operands in a special way. Assert that pos+len <= mode precision.
35599 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35601 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35602 3 argument Alias with unlimited for the negative form.
35603 (fno-vect-cost-model): Removed.
35605 2017-01-05 Martin Liska <mliska@suse.cz>
35607 * hsa-gen.c (gen_hsa_divmod): New function.
35608 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35610 2017-01-05 Martin Liska <mliska@suse.cz>
35613 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35616 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35618 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35619 small constant length operands.
35621 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35623 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35624 between loop iterations.
35626 2017-01-05 Martin Liska <mliska@suse.cz>
35629 * gimplify.c (gimplify_decl_expr): Compare to
35630 asan_poisoned_variables instread of checking flags.
35631 (gimplify_target_expr): Likewise.
35632 (gimplify_expr): Likewise.
35633 (gimplify_function_tree): Conditionally initialize
35634 asan_poisoned_variables.
35636 2017-01-04 Jeff Law <law@redhat.com>
35638 PR tree-optimizatin/78812
35639 * rtl.h (contains_mem_rtx_p): Prototype.
35640 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35641 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35642 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35643 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35644 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35646 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35648 * input.c (assert_char_at_range): Default-initialize actual_range.
35650 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35652 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35653 to match the caller.
35655 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35657 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35658 insns after final jump in test to emit dummy move.
35660 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35662 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35663 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35665 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35667 * multiple_target.c (create_dispatcher_calls): Init e_next.
35668 * tree-ssa-loop-split.c (split_loop): Init border.
35669 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35672 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35677 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35678 (altivec_register_operand): Do not return true if the operand
35679 contains a SUBREG mixing SImode and SFmode.
35680 (vsx_register_operand): Likewise.
35681 (vsx_reg_sfsubreg_ok): New predicate.
35682 (vfloat_operand): Do not return true if the operand contains a
35683 SUBREG mixing SImode and SFmode.
35684 (vint_operand): Likewise.
35685 (vlogical_operand): Likewise.
35686 (gpc_reg_operand): Likewise.
35687 (int_reg_operand): Likewise.
35688 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35689 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35690 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35692 (rs6000_emit_move_si_sf_subreg): New helper function.
35693 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35694 fixup SUBREGs involving SImode and SFmode.
35695 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35696 numbers for the new peephole2 optimization.
35697 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35698 the GLIBC math library that do AND/IOR/XOR operations on single
35699 precision floating point.
35700 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35701 target macros to say whether we need to avoid SUBREGs mixing
35703 (TARGET_ALLOW_SF_SUBREG): Likewise.
35704 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35705 (UNSPEC_SI_FROM_SF): Likewise.
35706 (iorxor): Change spacing.
35707 (and_ior_xor): New iterator for AND, IOR, and XOR.
35708 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35709 (movdi_from_sf_zero_ext): Likewise.
35710 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35711 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35712 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35713 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35714 (fms<mode>4): Likewise.
35715 (fnma<mode>4): Likewise.
35716 (fnms<mode>4): Likewise.
35717 (nfma<mode>4): Likewise.
35718 (nfms<mode>4): Likewise.
35720 2017-01-04 Marek Polacek <polacek@redhat.com>
35723 * doc/invoke.texi: Document -Wpointer-compare.
35725 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35727 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35730 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35731 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35734 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35737 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35738 documentation of the powerpc_popcntb_ok attribute.
35739 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35740 code to issue warning messages if a requested CPU configuration is
35741 not supported by the binary (assembler and loader) toolchain.
35742 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35743 made to define a built-in function that has been disabled.
35744 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35745 made to define a built-in function that has been disabled.
35746 (altivec_init_builtins): Add comment explaining why definition
35747 of the DST built-in functions is not preceded by an assertion
35748 check. Add assertions to prevent ICE if attempts are made to
35749 define an altivec predicate or an abs* built-in function that has
35751 (htm_init_builtins): Add comment explaining why definition of the
35752 htm built-in functions is not preceded by an assertion check.
35754 2017-01-04 Jeff Law <law@redhat.com>
35756 PR tree-optimizatin/67955
35757 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35758 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35759 the points-to solution does not include pt_null. Use DECL_PT_UID
35762 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35764 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35765 Use gen_int_mode instead of gen_lopwart for const_int operands.
35767 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35769 PR tree-optimization/71563
35770 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35771 out of range value - has low bits known to be zero.
35773 2017-01-04 Alan Modra <amodra@gmail.com>
35775 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35776 * configure: Regenerate.
35777 * config.in: Regenerate.
35779 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35782 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35783 a substring of the message, but strcmp with the whole message. Ifdef
35784 ENABLE_NLS, translate the message first using dgettext.
35786 2017-01-03 Jeff Law <law@redhat.com>
35788 PR tree-optimizatin/78856
35789 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35790 (mark_threaded_blocks): Remove code to truncate thread paths that
35791 cross multiple loop headers. Instead invalidate the cached loop
35792 iteration information and handle case of a thread path walking
35793 into an irreducible region.
35795 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35798 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35799 assertions. Add support for doing the signbit if the IEEE 128-bit
35800 floating point value is in a GPR.
35801 * config/rs6000/rs6000.md (Fsignbit): Delete.
35802 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35803 Update the length attribute if the value is in a GPR.
35804 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35805 the sign or zero extension instruction, since the value is always 0/1.
35806 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35809 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35810 extracting SImode to a GPR register so that we can generate a
35811 store, limit the vector to be in a traditional Altivec register
35812 for the vextuwrx instruction.
35814 2017-01-03 Ian Lance Taylor <iant@google.com>
35816 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35818 2017-01-03 Martin Sebor <msebor@redhat.com>
35820 PR tree-optimization/78696
35821 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35822 precision. Use MPFR for %f for greater fidelity. Correct handling
35824 (pass_sprintf_length::compute_format_length): Set width and precision
35825 specified by asrerisk to void_node for vararg functions.
35826 (try_substitute_return_value): Adjust dump output.
35828 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35830 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35832 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35834 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35835 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35836 -mlra/-mno-lra was passed to the compiler.
35838 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35840 PR rtl-optimization/65618
35841 * emit-rtl.c (try_split): Move initialization of "before" and
35842 "after" to just before the call to emit_insn_after_setloc.
35844 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35846 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35848 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35850 * dwarf2out.c (gen_enumeration_type_die): When
35851 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35853 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35855 PR tree-optimization/78965
35856 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35857 Change first argument from const call_info & to call_info &. For %n
35858 set info.nowrite to false.
35860 PR middle-end/78901
35861 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35862 possibly throwing calls.
35864 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35865 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35866 and fns handling, rather than in a separate case SSA_NAME.
35868 2017-01-02 Jeff Law <law@redhat.com>
35870 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35871 fixes for first_period and second_period variables.
35873 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35876 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35877 (*insvqi_1): New insn pattern.
35878 (*insvqi_1_mem_rex64): Ditto.
35879 (*insvqi_2): Ditto.
35880 (*insvqi_3): Rename from *insvqi.
35882 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35884 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35886 * doc/cfg.texi (Edges): Remove reference to Java.
35887 (Maintaining the CFG): Ditto.
35889 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35891 PR middle-end/77674
35892 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35893 transparent aliases.
35895 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35897 PR middle-end/77484
35898 * predict.def (PRED_CALL): Update hitrate.
35899 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35900 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35901 into direct/indirect/polymorphic variants.
35903 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35905 Update copyright years.
35907 * gcc.c (process_command): Update copyright notice dates.
35908 * gcov-dump.c (print_version): Ditto.
35909 * gcov.c (print_version): Ditto.
35910 * gcov-tool.c (print_version): Ditto.
35911 * gengtype.c (create_file): Ditto.
35912 * doc/cpp.texi: Bump @copying's copyright year.
35913 * doc/cppinternals.texi: Ditto.
35914 * doc/gcc.texi: Ditto.
35915 * doc/gccint.texi: Ditto.
35916 * doc/gcov.texi: Ditto.
35917 * doc/install.texi: Ditto.
35918 * doc/invoke.texi: Ditto.
35920 Copyright (C) 2017 Free Software Foundation, Inc.
35922 Copying and distribution of this file, with or without modification,
35923 are permitted in any medium without royalty provided the copyright
35924 notice and this notice are preserved.