1 2016-07-21 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/81303
4 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5 into account prologue and epilogue iterations when raising
6 min_profitable_iters to sth at least covering one vector iteration.
8 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10 * config/arm/arm.c (arm_test_cpu_arch_dat):
13 2017-07-20 Nathan Sidwell <nathan@acm.org>
16 * tree.h (TYPE_METHODS): Delete.
17 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19 (dbxout_type_methods): Scan TYPE_FIELDS.
20 (dbxout_type): Don't check TYPE_METHODS here.
21 * function.c (use_register_for_decl): Always ignore register for
22 class types when not optimizing.
23 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
24 * tree.c (free_lang_data_in_type): Stitch out member functions and
25 templates from TYPE_FIELDS.
26 (build_distinct_type_copy, verify_type_variant,
27 verify_type): Member fns are on TYPE_FIELDS.
28 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
29 * tree-pretty-print.c (dump_generic_node): Likewise.
31 2017-07-20 Jakub Jelinek <jakub@redhat.com>
34 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
35 V2TImode and V4TImode.
36 (ix86_expand_vector_extract): Likewise.
37 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
38 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
39 (ssescalarmode): Handle V4TImode and V2TImode.
40 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
41 (*vec_extractv2ti, *vec_extractv4ti): New insns.
42 (VEXTRACTI128_MODE): New mode iterator.
43 (splitter for *vec_extractv?ti first element): New.
44 (VEC_INIT_MODE): New mode iterator.
45 (vec_init<mode>): Consolidate 3 expanders into one using
46 VEC_INIT_MODE mode iterator.
48 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
50 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
51 non_spilled_static_chain_regno_p.
53 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
55 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
57 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
59 * bb-reorder.c (connect_traces): Allow copying of blocks within
62 2017-07-20 Richard Biener <rguenther@suse.de>
64 * gimple.h (gimple_phi_result): Add gphi * overload.
65 (gimple_phi_result_ptr): Likewise.
66 (gimple_phi_arg): Likewise. Adjust index assert to only
67 allow actual argument accesses rather than all slots available
69 (gimple_phi_arg_def): Add gphi * overload.
70 * tree-phinodes.c (make_phi_node): Initialize only actual
72 (resize_phi_node): Clear memory not covered by old node,
73 do not initialize excess argument slots.
74 (reserve_phi_args_for_new_edge): Initialize new argument slot
77 2017-07-20 Bin Cheng <bin.cheng@arm.com>
79 PR tree-optimization/81388
81 2016-07-21 Bin Cheng <bin.cheng@arm.com>
83 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
84 by removing computation of may_be_zero.
86 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
87 Tom de Vries <tom@codesourcery.com>
90 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
91 when gimple level profile disagrees with what RTL expander did.
93 2017-07-20 Richard Biener <rguenther@suse.de>
95 PR tree-optimization/61171
96 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
97 (vect_analyze_stmt): Add slp instance parameter.
98 (vectorizable_reduction): Likewise.
99 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
100 (vect_is_simple_reduction): Deal with chains not detected
101 as SLP reduction chain, specifically not properly associated
102 chains containing a mix of plus/minus.
103 (get_reduction_op): Remove.
104 (get_initial_defs_for_reduction): Simplify, pass in whether
105 this is a reduction chain, pass in the SLP node for the PHIs.
106 (vect_create_epilog_for_reduction): Get the SLP instance as
108 (vectorizable_reduction): Get the SLP instance as arg.
109 During analysis remember the SLP node with the PHIs in the
110 instance. Simplify getting at the vectorized reduction PHIs.
111 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
112 through SLP instance.
113 (vect_slp_analyze_operations): Likewise.
114 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
115 (vect_transform_stmt): Likewise.
117 2017-07-20 Tom de Vries <tom@codesourcery.com>
119 PR tree-optimization/81489
120 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
121 read of phi arg location to before loop that modifies phi.
123 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
125 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
128 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
131 * except.c (execute): Fix ordering issue.
133 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
135 PR rtl-optimization/81423
136 * combine.c (make_compound_operation_int): Don't try to optimize
137 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
139 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
141 PR rtl-optimization/81423
142 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
143 with a constant that is -1 in the truncated to mode.
145 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
147 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
148 (determine_unlikely_bbs): ... here.
149 * predict.h (propagate_unlikely_bbs_forward): Declare.
150 * cfgexpand.c (pass_expand::execute): Use it.
151 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
153 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
154 propagate_unlikely_bbs_forward.
156 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
159 * except.c (maybe_add_nop_after_section_switch): New function.
162 2017-07-19 Tom de Vries <tom@codesourcery.com>
164 * gimple.h (gimple_phi_set_arg): Make assert more strict.
166 2017-07-19 Tom de Vries <tom@codesourcery.com>
168 * gimple.h (gimple_phi_arg): Make assert more strict.
170 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
172 * config.gcc (powerpc*-*-*): Add mmintrin.h.
173 * config/rs6000/mmintrin.h: New file.
174 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
176 2017-07-19 Jakub Jelinek <jakub@redhat.com>
178 PR tree-optimization/81346
179 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
181 2017-07-19 Tom de Vries <tom@codesourcery.com>
183 * config/nvptx/nvptx.md (VECIM): Add V2DI.
185 2017-07-19 Tom de Vries <tom@codesourcery.com>
187 * config/nvptx/nvptx-modes.def: Add V2DImode.
188 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
189 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
190 (nvptx_output_mov_insn): Handle lack of mov.b128.
191 (nvptx_print_operand): Handle 'H' and 'L' codes.
192 (nvptx_vector_mode_supported): Allow V2DImode.
193 (nvptx_preferred_simd_mode): New function.
194 (nvptx_data_alignment): New function.
195 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
196 nvptx_preferred_simd_mode.
197 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
199 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
201 2017-07-19 Tom de Vries <tom@codesourcery.com>
203 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
204 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
205 (nvptx_vector_mode_supported): New function. Allow V2SImode.
206 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
207 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
208 (mov<VECIM>_insn): New define_insn.
209 (define_expand "mov<VECIM>): New define_expand.
211 2017-07-19 Tom de Vries <tom@codesourcery.com>
213 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
215 2017-07-19 Jakub Jelinek <jakub@redhat.com>
217 PR tree-optimization/81346
218 * fold-const.h (fold_div_compare, range_check_type): Declare.
219 * fold-const.c (range_check_type): New function.
220 (build_range_check): Use range_check_type.
221 (fold_div_compare): No longer static, rewritten into
222 a match.pd helper function.
223 (fold_comparison): Don't call fold_div_compare here.
224 * match.pd (X / C1 op C2): New optimization using fold_div_compare
227 2017-07-19 Nathan Sidwell <nathan@acm.org>
229 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
230 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
231 * tree.c (find_decls_types_r, verify_type): Use
232 TYPE_{MIN,MAX}_VALUE_RAW.
233 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
234 (hash_tree): Likewise.
235 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
237 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
240 2017-07-18 Tom de Vries <tom@codesourcery.com>
243 * omp-expand.c (expand_omp_for_static_chunk): Handle
244 equal-argument loop exit phi.
246 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
249 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
250 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
252 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
254 (ror,rol -> rorx splitters): Use const_int_operand as
257 2017-06-18 Richard Biener <rguenther@suse.de>
259 PR tree-optimization/81410
260 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
261 the gap in the ! slp_perm SLP case after each group.
263 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
266 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
269 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
272 * predict.c (set_even_probabilities): Cleanup; do not affect
273 probabilities that are already known.
274 (combine_predictions_for_bb): Call even when count is set.
276 2017-07-18 Nathan Sidwell <nathan@acm.org>
278 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
281 2017-07-18 Bin Cheng <bin.cheng@arm.com>
284 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
285 optimization for loop niter analysis.
287 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
290 * config/avr/avr.c (avr_optimize_casesi): Don't use
291 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
293 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
295 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
296 body_cost_vec from _vect_peel_extended_info.
297 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
298 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
301 2017-07-18 Bin Cheng <bin.cheng@arm.com>
303 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
305 2017-07-18 Richard Biener <rguenther@suse.de>
307 PR tree-optimization/80620
308 PR tree-optimization/81403
309 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
310 info when re-using a VN table entry.
312 2017-07-18 Richard Biener <rguenther@suse.de>
314 PR tree-optimization/81418
315 * tree-vect-loop.c (vectorizable_reduction): Properly compute
316 vectype_in. Verify that with lane-reducing reduction operations
317 we have a single def-use cycle.
319 2017-07-17 Carl Love <cel@us.ibm.com>
321 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
323 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
324 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
325 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
326 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
327 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
328 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
329 VMULOSW): New enum "unspec" values.
330 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
331 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
332 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
333 altivec_vmulosw): New patterns.
334 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
335 VMULOSW): Add definitions.
336 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
338 * config/alpha/alpha.c: Include predict.h.
340 2017-07-17 Yury Gribov <tetra2005@gmail.com>
342 * tree-vrp.c (compare_assert_loc): Fix comparison function
343 to return predictable results.
345 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
347 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
350 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
351 * doc/invoke.texi (mexpand-adddi): Update text.
353 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
355 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
356 that also clobbers the CC register. The old expand code is moved
358 (*arc_clzsi2): ... here.
359 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
360 the CC register. The old expand code is moved to ...
361 (arc_ctzsi2): ... here.
363 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
365 * config/arc/arc.opt (mindexed-loads): Use initial value
366 TARGET_INDEXED_LOADS_DEFAULT.
367 (mauto-modify-reg): Use initial value
368 TARGET_AUTO_MODIFY_REG_DEFAULT.
369 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
370 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
371 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
372 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
374 2017-07-17 Martin Liska <mliska@suse.cz>
377 * opts.c (finish_options): Do not allow -fgnu-tm
378 w/ -fsanitize={kernel-,}address. Say sorry.
380 2017-07-17 Bin Cheng <bin.cheng@arm.com>
383 * tree-loop-distribution.c (classify_partition): Only assert on
385 (merge_dep_scc_partitions): Delete prameter. Update function call.
386 (distribute_loop): Remove code handling loop with unknown niters.
387 (pass_loop_distribution::execute): Skip loop with unknown niters.
389 2017-07-17 Bin Cheng <bin.cheng@arm.com>
392 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
393 function sort_partitions_by_post_order.
395 2017-07-17 Bin Cheng <bin.cheng@arm.com>
397 PR tree-optimization/81374
398 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
399 the max index of basic blocks, rather than number of basic blocks.
401 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
403 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
405 (arc_legitimate_pic_operand_p): Likewise.
406 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
408 (arc_needs_pcl_p): Likewise.
409 (arc_legitimate_pc_offset_p): Likewise.
410 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
411 function is also used in constrains.md.
412 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
413 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
414 PLUS. Only return true/false in known cases, otherwise assert.
415 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
416 is already called in arc_legitimate_constant_p.
417 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
419 (LEGITIMATE_PIC_OPERAND_P): Use
420 arc_raw_symbolic_reference_mentioned_p function.
421 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
426 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
427 Andrew Burgess <andrew.burgess@embecosm.com>
429 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
430 (arc_return_address_register): New function.
431 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
432 (arc_handle_fndecl_attribute): Add naked attribute.
433 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
434 (TARGET_WARN_FUNC_RETURN): Likewise.
435 (arc_allocate_stack_slots_for_args): New function.
436 (arc_warn_func_return): Likewise.
437 (machine_function): Change type fn_type.
438 (arc_compute_function_type): Consider new naked function type,
439 change function return type.
440 (arc_must_save_register): Adapt to handle new
441 arc_compute_function_type's return type.
442 (arc_expand_prologue): Likewise.
443 (arc_expand_epilogue): Likewise.
444 (arc_return_address_regs): Delete.
445 (arc_return_address_register): New function.
446 (arc_epilogue_uses): Use above function.
447 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
448 (arc_function_type): Change encoding, add naked type.
449 (ARC_INTERRUPT_P): Change to handle the new encoding.
450 (ARC_FAST_INTERRUPT_P): Likewise.
451 (ARC_NORMAL_P): Define.
452 (ARC_NAKED_P): Likewise.
453 (arc_compute_function_type): Delete prototype.
454 * config/arc/arc.md (in_ret_delay_slot): Use
455 arc_return_address_register function.
456 (simple_return): Likewise.
457 (p_return_i): Likewise.
459 2017-07-17 Jakub Jelinek <jakub@redhat.com>
461 PR tree-optimization/81428
462 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
463 can't be built for those types.
465 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
467 Remove stuff dead since r239246.
469 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
470 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
471 (avr_inform_devices): Remove dead stuff.
473 2017-07-17 Tamar Christina <tamar.christina@arm.com>
475 * config/arm/arm_neon.h: Fix softp typo.
477 2017-07-17 Jakub Jelinek <jakub@redhat.com>
479 PR tree-optimization/81365
480 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
481 aggregate moves onto bb predecessor edges, make sure there are no
482 loads that could alias the lhs in between the start of bb and the
485 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
488 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
489 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
490 [LSHIFTRT, outer_code = TRUNCATE]: Same.
492 2017-07-17 Jakub Jelinek <jakub@redhat.com>
494 PR tree-optimization/81396
495 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
496 (init_symbolic_number): Initialize it to 1.
497 (perform_symbolic_merge): Add n_ops from both operands into the new
499 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
500 without base_addr as useless if they need more than one operation.
501 (bswap_replace): Handle !bswap case for NULL base_addr.
503 2017-07-17 Tom de Vries <tom@codesourcery.com>
506 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
509 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
511 * gcc/config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
512 conditional builtin define __FIX_LEON3FT_B2BST.
514 2017-07-17 Daniel Cederman <cederman@gaisler.com>
516 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
517 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
520 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
522 PR rtl-optimization/81424
523 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
524 to remove potential trapping from operands if -fnon-call-exceptions.
526 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
528 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
529 profile_proability for scalling.
530 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
532 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
534 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
536 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
538 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
539 fixpoint arithmetics.
541 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
543 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
544 fixpoint arithmetics.
546 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
548 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
549 fixpoint arithmetics.
551 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
553 * profile-count.h (profile_probability::from_reg_br_prob_note,
554 profile_probability::to_reg_br_prob_note): New functions.
555 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
556 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
557 * predict.c (probability_reliable_p): Update.
558 (edge_probability_reliable_p): Update.
559 (br_prob_note_reliable_p): Update.
560 (invert_br_probabilities): Update.
561 (add_reg_br_prob_note): New function.
562 (combine_predictions_for_insn): Update.
563 * asan.c (asan_clear_shadow): Update.
564 * cfgbuild.c (compute_outgoing_frequencies): Update.
565 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
566 (update_br_prob_note): Update.
567 (rtl_verify_edges): Update.
568 (purge_dead_edges): Update.
569 (fixup_reorder_chain): Update.
570 * emit-rtl.c (try_split): Update.
571 * ifcvt.c (cond_exec_process_insns): Update.
572 (cond_exec_process_if_block): Update.
573 (dead_or_predicable): Update.
574 * internal-fn.c (expand_addsub_overflow): Update.
575 (expand_neg_overflow): Update.
576 (expand_mul_overflow): Update.
577 * loop-doloop.c (doloop_modify): Update.
578 * loop-unroll.c (compare_and_jump_seq): Update.
579 * optabs.c (emit_cmp_and_jump_insn_1): Update.
581 * reorg.c (mostly_true_jump): Update.
583 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
584 * config/alpha/alpha.c (emit_unlikely_jump): Update.
585 * config/arc/arc.c: (emit_unlikely_jump): Update.
586 * config/arm/arm.c: (emit_unlikely_jump): Update.
587 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
588 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
589 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
590 (ix86_print_operand): Update.
591 (ix86_split_fp_branch): Update.
592 (predict_jump): Update.
593 * config/ia64/ia64.c (ia64_print_operand): Update.
594 * config/mmix/mmix.c (mmix_print_operand): Update.
595 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
596 (rs6000_expand_split_stack_prologue): Update.
597 * config/rs6000/rs6000.c: Update.
598 * config/s390/s390.c (s390_expand_vec_strlen): Update.
599 (s390_expand_vec_movstr): Update.
600 (s390_expand_cs_tdsi): Update.
601 (s390_expand_split_stack_prologue): Update.
602 * config/sh/sh.c (sh_print_operand): Update.
603 (expand_cbranchsi4): Update.
604 (expand_cbranchdi4): Update.
605 * config/sparc/sparc.c (output_v9branch): Update.
606 * config/spu/spu.c (get_branch_target): Update.
607 (ea_load_store_inline): Update.
608 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
609 * config/tilepro/tilepro.c: Update.
611 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
613 * gimplify.c (mostly_copy_tree_r): Revert latest change.
614 (gimplify_save_expr): Likewise.
616 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
618 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
620 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
622 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
624 * timevar.def (TV_IPA_FNSUMMARY): Define.
626 2017-07-16 Daniel Cederman <cederman@gaisler.com>
628 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
629 to back store errata sensitive sequence from being generated.
630 (sqrtdf2_fix): Likewise.
632 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
634 * tree-ssa-threadupdate.c (compute_path_counts,
635 update_joiner_offpath_counts): Use profile_probability.
637 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
640 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
642 * config/arm/arm-c.c (arm_cpu_builtins): Define
643 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
645 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
647 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
648 array entries to represent __ieee128 versions of the
649 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
650 scalar_extract_sig, and scalar_insert_exp built-in functions.
651 (altivec_resolve_overloaded_builtin): Add special case handling
652 for the __builtin_scalar_insert_exp function, as represented by
653 the P9V_BUILTIN_VEC_VSIEDP constant.
654 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
655 exponent support for __ieee128 argument.
656 (VSESQP): Add scalar extract signature support for __ieee128
658 (VSTDCNQP): Add scalar test negative support for __ieee128
660 (VSIEQP): Add scalar insert exponent support for __int128 argument
661 with __ieee128 result.
662 (VSIEQPF): Add scalar insert exponent support for __ieee128
663 argument with __ieee128 result.
664 (VSTDCQP): Add scalar test data class support for __ieee128
666 (VSTDCNQP): Add overload support for scalar test negative with
668 (VSTDCQP): Add overload support for scalar test data class
670 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
672 (UNSPEC_VSX_SIEXPQP): New constant.
673 (xsxexpqp): New insn for VSX scalar extract exponent quad
675 (xsxsigqp): New insn for VSX scalar extract significand quad
677 (xsiexpqpf): New insn for VSX scalar insert exponent quad
678 precision with floating point argument.
679 (xststdcqp): New expand for VSX scalar test data class quad
681 (xststdcnegqp): New expand for VSX scalar test negative quad
683 (xststdcqp): New insn to match expansions for VSX scalar test data
684 class quad precision and VSX scalar test negative quad precision.
685 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
686 special case operand checking to enforce that second operand of
687 VSX scalar test data class with quad precision argument is a 7-bit
689 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
690 prototypes and descriptions of __ieee128 versions of
691 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
692 scalar_test_data_class, and scalar_test_neg built-in functions.
694 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
696 PR tree-optimization/81162
697 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
698 replace a negate with an add.
700 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
702 * doc/invoke.texi (arm/-mcpu): Document +crypto.
704 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
706 * config/arm/arm-c.c (arm_cpu_builtins): Define
707 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
709 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
711 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
712 (armv8-r): Set ARM Cortex-R52 as default CPU.
713 * config/arm/arm-tables.opt: Regenerate.
714 * config/arm/arm-tune.md: Regenerate.
715 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
717 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
718 extension for -mcpu=cortex-r52.
720 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
722 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
723 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
724 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
725 (fp-armv8): Define it as FP_ARMv8 only.
726 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
727 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
729 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
730 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
731 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
732 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
733 than TARGET_FPU_ARMV8.
734 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
735 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
736 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
738 * config/arm/neon.md (neon_vrint): Likewise.
739 (neon_vcvt): Likewise.
740 (neon_<fmaxmin_op><mode>): Likewise.
741 (<fmaxmin><mode>3): Likewise.
742 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
743 * config/arm/predicates.md (arm_cond_move_operator): Check against
744 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
746 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
748 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
751 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
753 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
754 loop in comment with memset.
756 2017-07-14 Martin Liska <mliska@suse.cz>
758 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
759 * dwarf2out.c (is_java): Remove the function.
760 (output_pubname): Remove usage of the function.
761 (lower_bound_default): Remove usage of DW_LANG_Java.
762 (gen_compile_unit_die): Likewise.
763 * gcc.c: Remove compiler defaults for .java and .zip files.
764 * gimple-expr.c (remove_suffix): Change as there's no longer
765 extension than 4-letter one.
766 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
767 (gimplify_save_expr): Likewise.
768 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
769 as it's possible even for other languages than Java.
770 * langhooks.h (struct lang_hooks): Remove Java from a comment.
771 * lto-opts.c (lto_write_options): Remove reference to Java.
772 * opts.c (strip_off_ending): Update file extension handling.
773 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
774 * tree-eh.c (lower_resx): Likewise.
775 * tree.c (free_lang_data_in_type): Remove dead code.
776 (find_decls_types_r): Likewise.
777 (build_common_builtin_nodes): Remove Java from a comment.
778 (verify_type): Remove dead code.
779 * varasm.c (assemble_external): Remove Java from a comment.
781 2017-07-14 Martin Liska <mliska@suse.cz>
783 * opts.c (finish_options): Add quotes.
784 (common_handle_option): Likewise.
786 2017-07-14 Martin Liska <mliska@suse.cz>
788 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
789 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
791 * dwarf2out.c (lower_bound_default): Do not handle
793 (gen_compile_unit_die): Likewise.
794 * gcc.c: Remove default extension binding for GNU Pascal.
795 * stmt.c: Remove Pascal language from a comment.
796 * xcoffout.c: Likewise.
798 2017-07-13 David Malcolm <dmalcolm@redhat.com>
801 * diagnostic-show-locus.c (fixit_cmp): New function.
802 (layout::layout): Sort m_fixit_hints.
803 (column_range::column_range): Assert that the values are valid.
804 (struct char_span): New struct.
805 (correction::overwrite): New method.
806 (struct source_line): New struct.
807 (line_corrections::add_hint): Add assertions. Reimplement memcpy
808 calls in terms of classes source_line and char_span, and
809 correction::overwrite.
810 (selftest::test_overlapped_fixit_printing_2): New function.
811 (selftest::diagnostic_show_locus_c_tests): Call it.
813 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
815 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
816 early if there is no lhs.
818 2017-07-13 Martin Liska <mliska@suse.cz>
820 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
821 (gen_reference_type_die): Likewise.
822 * stor-layout.c: Remove Pascal-related comment.
824 2017-07-13 Martin Liska <mliska@suse.cz>
826 * opts.c (finish_options): Add quotes to error messages.
827 (parse_sanitizer_options): Likewise.
829 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
831 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
833 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
835 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
837 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
839 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
841 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
843 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
846 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
847 provides the hardware capability bits, define the macro
848 __BUILTIN_CPU_SUPPORTS__.
849 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
850 if GLIBC does not provide the hardware capability bits. Add a
851 gcc_unreachable call if the built-in cpu function is neither
852 __builtin_cpu_is nor __builtin_cpu_supports.
853 (rs6000_get_function_versions_dispatcher): Change the warning
854 that an old GLIBC is used which does not export the capability
856 * doc/extend.texi (target_clones attribute): Document the
857 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
858 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
859 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
860 the macros defined by GCC if the newer GLIBC is available.
862 2017-07-12 Jeff Law <law@redhat.com>
864 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
865 remaining includes slightly.
866 * config/riscv/riscv-builtins.c: Include profile-count.h.
868 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
871 * config/avr/avr.c (avr_set_current_function): In diagnostic
872 messages: Quote keywords and (parts of) identifiers.
873 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
876 2017-07-12 Carl Love <cel@us.ibm.com>
878 * config/rs6000/rs6000-c.c: Add support for built-in functions
879 vector bool char vec_revb (vector bool char);
880 vector bool short vec_revb (vector short char);
881 vector bool int vec_revb (vector bool int);
882 vector bool long long vec_revb (vector bool long long);
883 * doc/extend.texi: Update the built-in documentation file for the
884 new built-in functions.
886 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
888 * config/s390/s390.md: Remove movcc splitter.
890 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
892 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
893 load/store on condition.
895 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
898 * config/avr/avr.c (avr_encode_section_info)
899 [progmem && !TREE_READONLY]: Error if progmem object needs
902 2017-07-11 Michael Collison <michael.collison@arm.com>
904 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
907 2017-07-11 Carl Love <cel@us.ibm.com>
909 * config/rs6000/rs6000-c.c: Add support for builtins
910 vector unsigned int vec_parity_lsbb (vector signed int);
911 vector unsigned int vec_parity_lsbb (vector unsigned int);
912 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
913 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
914 vector unsigned long long vec_parity_lsbb (vector signed long long);
915 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
916 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
917 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
918 * doc/extend.texi: Update the built-in documentation file for the
919 new built-in functions.
921 2017-07-11 David Malcolm <dmalcolm@redhat.com>
923 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
924 (layout::m_primary_loc): New field.
925 (layout::layout): Initialize new field. Move location filtering
926 logic from here to...
927 (layout::maybe_add_location_range): ...this new method. Add
928 support for filtering to just the lines already specified by other
930 (layout::will_show_line_p): New method.
931 (gcc_rich_location::add_location_if_nearby): New method.
932 (selftest::test_add_location_if_nearby): New test function.
933 (selftest::diagnostic_show_locus_c_tests): Call it.
934 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
937 2017-07-11 Tom de Vries <tom@codesourcery.com>
939 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
940 (bb_first_real_insn): New function.
941 (nvptx_single): Add extra initialization of broadcasted condition
944 2017-07-11 Nathan Sidwell <nathan@acm.org>
946 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
948 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
950 * doc/extend.texi (AVR Function Attributes): Remove weblink to
951 Binutils doc as TEXI will mess them up.
952 * doc/invoke.texi (AVR Options): Same here.
954 2017-07-11 Daniel Cederman <cederman@gaisler.com>
956 * config/sparc/sparc.opt (mfix-ut700): New option.
957 (mfix-gr712rc): Likewise.
958 (sparc_fix_b2bst): New variable.
959 * doc/invoke.texi (SPARC options): Document them.
960 (ARM options): Fix warnings.
961 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
962 instructions to prevent sequences that can trigger the store-store
963 errata for certain LEON3FT processors.
964 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
965 (sparc_option_override): Set sparc_fix_b2bst appropriately.
966 * config/sparc/sparc.md (fix_b2bst): New attribute.
967 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
969 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
972 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
974 (*rsqrtsf2_sse): Ditto.
976 (div<mode>3): Macroize insn from divdf3 and divsf3
977 using MODEF mode iterator.
979 2017-07-10 Martin Sebor <msebor@redhat.com>
981 PR tree-optimization/80397
982 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
983 instead of testing for equality to INTEGER_TYPE.
985 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
987 * config.gcc: Remove uclibc from arc target spec.
989 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
991 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
993 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
996 * lto-wrapper.c (remove_option): New function.
997 (merge_and_complain): Merge PIC/PIE options more realistically.
999 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1001 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1005 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1006 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1007 * config.in: Regenerate.
1008 * configure: Regenerate.
1009 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1010 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1011 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1012 (TARGET_GASISR_PROLOGUES): ...target mask.
1013 * common/config/avr/avr-common.c
1014 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1015 Set -mgas-isr-prologues.
1016 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1017 INSERT_PASS_BEFORE for it.
1018 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1019 * config/avr/avr.c (avr_option_override)
1020 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1021 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1022 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1023 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1024 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1026 (make_avr_pass_pre_proep): New function.
1027 (emit_push_sfr) <treg>: Add argument to function and use it
1029 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1030 and set machine->gasisr.yes.
1031 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1032 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1033 __gcc_isr.n_pushed to .L__stack_usage.
1034 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1035 (avr_asm_final_postscan_insn): ...this new static function.
1036 * config/avr/avr.h (machine_function)
1037 <is_no_gccisr, use_L__stack_usage>: New fields.
1038 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1039 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1040 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1041 (gasisr, *gasisr): New expander and insn.
1042 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1043 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1044 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1046 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1048 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1051 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1053 Move jump-tables out of .text again.
1056 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1057 (ASM_OUTPUT_ADDR_VEC): New function.
1058 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1059 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1060 INSN_ADDRESSes as asm comment.
1061 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1062 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1063 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1064 * config/avr/avr.md (*tablejump): Adjust comment.
1065 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1066 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1068 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1069 (avr_output_addr_vec): New proto.
1070 (avr_log_t) <insn_addresses>: New field.
1072 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1075 * config/i386/i386.c (ix86_function_arg_advance): Set
1076 outgoing_args_on_stack to true if there are outgoing arguments
1078 (ix86_function_arg): Likewise.
1079 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1080 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1081 * config/i386/i386.h (machine_function): Add
1082 outgoing_args_on_stack.
1084 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1086 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1088 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1090 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1092 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1093 (REAL_H): Remove $(MACHMODE_H).
1094 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1096 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1097 $(MACHMODE_H) and double-int.h.
1098 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1100 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1103 2017-07-07 Andrew Pinski <apinski@cavium.com>
1105 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1106 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1108 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1110 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1111 Add warning if GCC was not configured to link against a GLIBC that
1112 exports the hardware capability bits.
1113 (make_resolver_func): Make resolver function private and not a
1114 COMDAT function. Create the name with clone_function_name instead
1115 of make_unique_name.
1118 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1119 correct operand in doing the split.
1121 2017-07-07 Carl Love <cel@us.ibm.com>
1123 * config/rs6000/rs6000-c: Add support for built-in function
1124 vector unsigned short vec_pack_to_short_fp32 (vector float,
1126 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1127 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1128 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1129 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1130 (convert_4f32_8i16): Add define_expand.
1131 * doc/extend.texi: Update the built-in documentation file for the
1132 new built-in function.
1134 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1136 * config/sparc/m8.md: New file.
1137 * config/sparc/sparc.md: Include m8.md.
1139 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1141 * config/sparc/sparc.opt: New option -mvis4b.
1142 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1143 (sparc_option_override): Handle VIS4B.
1144 (enum sparc_builtins): Define
1145 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1146 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1147 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1148 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1149 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1150 (check_constant_argument): New function.
1151 (sparc_vis_init_builtins): Define builtins
1152 __builtin_vis_dictunpack{8,16,32},
1153 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1154 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1155 __builtin_vis_fpcmpde{8,16,32}shl and
1156 __builtin_vis_fpcmpur{8,16,32}shl.
1157 (sparc_expand_builtin): Check that the constant operands to
1158 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1159 constant and in range.
1160 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1162 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1163 (SPARC_IMM5_P): Likewise.
1164 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1165 (enabled): Handle vis4b.
1166 (UNSPEC_DICTUNPACK): New unspec.
1167 (UNSPEC_FPCMPSHL): Likewise.
1168 (UNSPEC_FPUCMPSHL): Likewise.
1169 (UNSPEC_FPCMPDESHL): Likewise.
1170 (UNSPEC_FPCMPURSHL): Likewise.
1171 (cpu_feature): New CPU feature `vis4b'.
1172 (dictunpack{8,16,32}): New insns.
1173 (FPCSMODE): New mode iterator.
1174 (fpcscond): New code iterator.
1175 (fpcsucond): Likewise.
1176 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1177 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1178 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1179 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1180 * config/sparc/constraints.md: Define constraints `q' for unsigned
1181 2-bit integer constants and `t' for unsigned 5-bit integer
1183 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1185 (imm5_operand_dictunpack16): Likewise.
1186 (imm5_operand_dictunpack32): Likewise.
1187 (imm2_operand): Likewise.
1188 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1189 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1190 ditunpack* and fpcmp*shl builtins.
1192 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1194 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1195 * config.in: Add HAVE_AS_SPARC6 define.
1196 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1198 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1200 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1201 (CPP_CPU_SPEC): Handle m8.
1202 (ASM_CPU_SPEC): Likewise.
1203 * config/sparc/sparc-opts.h (enum processor_type): Add
1205 * config/sparc/sparc.c (m8_costs): New struct.
1206 (sparc_option_override): Handle TARGET_CPU_m8.
1207 (sparc32_initialize_trampoline): Likewise.
1208 (sparc64_initialize_trampoline): Likewise.
1209 (sparc_issue_rate): Likewise.
1210 (sparc_register_move_cost): Likewise.
1211 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1212 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1213 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1214 (CPP_CPU_SPEC): Handle M8.
1215 (ASM_CPU_SPEC): Likewise.
1216 (AS_M8_FLAG): Define.
1217 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1218 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1219 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1221 * configure: Regenerate.
1222 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1225 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1227 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1229 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1230 ("*movdi_insn_sp32"): Do not set v3pipe.
1231 ("*movsi_insn"): Likewise.
1232 ("*movdi_insn_sp64"): Likewise.
1233 ("*movsf_insn"): Likewise.
1234 ("*movdf_insn_sp32"): Likewise.
1235 ("*movdf_insn_sp64"): Likewise.
1236 ("*zero_extendsidi2_insn_sp64"): Likewise.
1237 ("*sign_extendsidi2_insn"): Likewise.
1238 ("*mov<VM32:mode>_insn"): Likewise.
1239 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1240 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1241 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1242 ("<vlop:code><VL:mode>3"): Likewise.
1243 ("*not_<vlop:code><VL:mode>3"): Likewise.
1244 ("*nand<VL:mode>_vis"): Likewise.
1245 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1246 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1247 ("one_cmpl<VL:mode>2"): Likewise.
1248 ("faligndata<VM64:mode>_vis"): Likewise.
1249 ("alignaddrsi_vis"): Likewise.
1250 ("alignaddrdi_vis"): Likweise.
1251 ("alignaddrlsi_vis"): Likewise.
1252 ("alignaddrldi_vis"): Likewise.
1253 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1254 ("bmaskdi_vis"): Likewise.
1255 ("bmasksi_vis"): Likewise.
1256 ("bshuffle<VM64:mode>_vis"): Likewise.
1257 ("cmask8<P:mode>_vis"): Likewise.
1258 ("cmask16<P:mode>_vis"): Likewise.
1259 ("cmask32<P:mode>_vis"): Likewise.
1260 ("pdistn<P:mode>_vis"): Likewise.
1261 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1263 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1265 * config/sparc/sparc.md ("subtype"): New insn attribute.
1266 ("*wrgsr_sp64"): Set insn subtype.
1267 ("*rdgsr_sp64"): Likewise.
1268 ("alignaddrsi_vis"): Likewise.
1269 ("alignaddrdi_vis"): Likewise.
1270 ("alignaddrlsi_vis"): Likewise.
1271 ("alignaddrldi_vis"): Likewise.
1272 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1273 ("fexpand_vis"): Likewise.
1274 ("fpmerge_vis"): Likewise.
1275 ("faligndata<VM64:mode>_vis"): Likewise.
1276 ("bshuffle<VM64:mode>_vis"): Likewise.
1277 ("cmask8<P:mode>_vis"): Likewise.
1278 ("cmask16<P:mode>_vis"): Likewise.
1279 ("cmask32<P:mode>_vis"): Likewise.
1280 ("fchksm16_vis"): Likewise.
1281 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1282 ("fmean16_vis"): Likewise.
1283 ("fp<plusminus_insn>64_vis"): Likewise.
1284 ("<plusminus_insn>v8qi3"): Likewise.
1285 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1286 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1287 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1288 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1289 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1290 ("*movqi_insn"): Likewise.
1291 ("*movhi_insn"): Likewise.
1292 ("*movsi_insn"): Likewise.
1293 ("movsi_pic_gotdata_op"): Likewise.
1294 ("*movdi_insn_sp32"): Likewise.
1295 ("*movdi_insn_sp64"): Likewise.
1296 ("movdi_pic_gotdata_op"): Likewise.
1297 ("*movsf_insn"): Likewise.
1298 ("*movdf_insn_sp32"): Likewise.
1299 ("*movdf_insn_sp64"): Likewise.
1300 ("*zero_extendhisi2_insn"): Likewise.
1301 ("*zero_extendqihi2_insn"): Likewise.
1302 ("*zero_extendqisi2_insn"): Likewise.
1303 ("*zero_extendqidi2_insn"): Likewise.
1304 ("*zero_extendhidi2_insn"): Likewise.
1305 ("*zero_extendsidi2_insn_sp64"): Likewise.
1306 ("ldfsr"): Likewise.
1307 ("prefetch_64"): Likewise.
1308 ("prefetch_32"): Likewise.
1309 ("tie_ld32"): Likewise.
1310 ("tie_ld64"): Likewise.
1311 ("*tldo_ldub_sp32"): Likewise.
1312 ("*tldo_ldub1_sp32"): Likewise.
1313 ("*tldo_ldub2_sp32"): Likewise.
1314 ("*tldo_ldub_sp64"): Likewise.
1315 ("*tldo_ldub1_sp64"): Likewise.
1316 ("*tldo_ldub2_sp64"): Likewise.
1317 ("*tldo_ldub3_sp64"): Likewise.
1318 ("*tldo_lduh_sp32"): Likewise.
1319 ("*tldo_lduh1_sp32"): Likewise.
1320 ("*tldo_lduh_sp64"): Likewise.
1321 ("*tldo_lduh1_sp64"): Likewise.
1322 ("*tldo_lduh2_sp64"): Likewise.
1323 ("*tldo_lduw_sp32"): Likewise.
1324 ("*tldo_lduw_sp64"): Likewise.
1325 ("*tldo_lduw1_sp64"): Likewise.
1326 ("*tldo_ldx_sp64"): Likewise.
1327 ("*mov<VM32:mode>_insn"): Likewise.
1328 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1329 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1331 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1333 * config/sparc/sparc.md ("type"): New insn type viscmp.
1334 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1336 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1337 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1338 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1339 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1341 ("n7_vis_logical_11cycle"): Likewise.
1342 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1343 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1344 * config/sparc/niagara.md ("niag_vis"): Likewise.
1345 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1346 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1348 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1350 * config/sparc/sparc.md: New instruction type `bmask'.
1351 (bmaskdi_vis): Use the `bmask' type.
1352 (bmasksi_vis): Likewise.
1353 * config/sparc/ultra3.md (us3_array): Likewise.
1354 * config/sparc/niagara7.md (n7_array): Likewise.
1355 * config/sparc/niagara4.md (n4_array): Likewise.
1356 * config/sparc/niagara2.md (niag2_vis): Likewise.
1357 (niag3_vis): Likewise.
1358 * config/sparc/niagara.md (niag_vis): Likewise.
1360 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1362 * ipa-comdats.c: Remove optimize check from gate.
1363 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1364 for functions not optimized.
1365 (ipa_fn_summary_read): Skip optimize check.
1366 (ipa_fn_summary_write): Likewise.
1367 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1369 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1371 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1372 (check_callers): Give up on caller which is not optimized.
1373 (inline_small_functions): Likewise.
1374 (ipa_inline): Do not give up when not optimizing.
1375 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1376 away unoptimizes cdtors.
1377 (whole_program_function_and_variable_visibility): Do
1378 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1379 * ipa.c (process_references): Do not check optimize.
1380 (symbol_table::remove_unreachable_nodes): Update optimize check.
1381 (set_writeonly_bit): Update optimize check.
1382 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1383 (pass_ipa_single_use::gate): Remove.
1385 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1387 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1388 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1389 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1390 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1391 permute_load, permute_store, adjust_extract, adjust_splat,
1392 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1393 replace_swap_with_copy, dump_swap_insn_table,
1394 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1395 recombine_lvx_pattern, recombine_stvx_pattern,
1396 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1397 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1398 to file rs6000-p8swap.c.
1399 * config/rs6000/rs6000-p8swap.c: New file.
1400 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1401 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1402 and rs6000*-*-* targets.
1404 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1406 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1408 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1410 * lto-wrapper.c (merge_and_complain): Do not merge
1411 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
1412 fsigned_zeros, ftrapping_math, fwrapv.
1413 (append_compiler_options): Do not track these options.
1414 (append_linker_options): Likewie
1416 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1418 * cgraphunit.c (cgraph_node::finalize_function): When
1419 !flag_toplevel_reorde set no_reorder flag.
1420 (varpool_node::finalize_decl): Likewise.
1421 (symbol_table::compile): Drop no toplevel reorder path.
1423 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1425 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
1426 edges; zero probability is not better than uninitialized.
1428 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1430 * asan.h (asan_sanitize_allocas_p): Declare.
1431 * asan.c (asan_sanitize_allocas_p): New function.
1432 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
1433 (handle_builtin_alloca): Likewise.
1434 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
1435 if !asan_sanitize_allocas_p.
1436 * params.def (asan-instrument-allocas): Add new option.
1437 * params.h (ASAN_PROTECT_ALLOCAS): Define.
1438 * opts.c (common_handle_option): Disable allocas sanitization for
1441 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1443 * asan.c: Include gimple-fold.h.
1444 (get_last_alloca_addr): New function.
1445 (handle_builtin_stackrestore): Likewise.
1446 (handle_builtin_alloca): Likewise.
1447 (asan_emit_allocas_unpoison): Likewise.
1448 (get_mem_refs_of_builtin_call): Add new parameter, remove const
1449 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
1450 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
1451 (instrument_builtin_call): Pass gimple iterator to
1452 get_mem_refs_of_builtin_call.
1453 (last_alloca_addr): New global.
1454 * asan.h (asan_emit_allocas_unpoison): Declare.
1455 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
1456 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
1457 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
1458 if function calls alloca.
1459 * gimple-fold.c (replace_call_with_value): Remove static keyword.
1460 * gimple-fold.h (replace_call_with_value): Declare.
1461 * internal-fn.c: Include asan.h.
1462 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
1463 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
1465 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1467 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
1468 (C_SELFTEST_FLAGS): New.
1469 (CPP_SELFTEST_FLAGS): New.
1470 (SELFTEST_DEPS): New, from deps of s-selftest.
1471 (C_SELFTEST_DEPS): New, from deps of s-selftest.
1472 (CPP_SELFTEST_DEPS): New.
1473 (selftest): Add dependency on s-selftest-c++.
1474 (s-selftest): Rename to...
1475 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
1476 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
1477 than SELFTEST_FLAGS.
1478 (selftest-gdb): Rename to...
1479 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
1481 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
1482 (selftest-valgrind): Rename to...
1483 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
1485 (selftest-valgrind): Reintroduce as an alias for
1486 selftest-c-valgrind.
1487 (s-selftest-c++): New.
1488 (selftest-c++-gdb): New.
1489 (selftest-c++-valgrind): New.
1491 2017-07-06 Olivier Hainque <hainque@adacore.com>
1493 * gcc.c (process_command): When deciding if undefined variables
1494 should be ignored when processing specs, accept "gcc -v" as well.
1496 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1498 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
1499 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
1501 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1503 * config/arm/arm-cpus.in (armv8-r): Add new entry.
1504 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
1505 * config/arm/arm-tables.opt: Regenerate.
1506 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
1508 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
1510 2017-07-06 Carl Love <cel@us.ibm.com>
1512 * ChangeLog: Clean up from mid air collision
1514 2017-07-06 Carl Love <cel@us.ibm.com>
1516 * config/rs6000/rs6000-c.c: Add support for built-in functions
1517 vector signed int vec_subc (vector signed int, vector signed int);
1518 vector signed __int128 vec_subc (vector signed __int128,
1519 vector signed __int128);
1520 vector unsigned __int128 vec_subc (vector unsigned __int128,
1521 vector unsigned __int128);
1522 vector signed int vec_sube (vector signed int, vector signed int,
1524 vector unsigned int vec_sube (vector unsigned int,
1525 vector unsigned int,
1526 vector unsigned int);
1527 vector signed __int128 vec_sube (vector signed __int128,
1528 vector signed __int128,
1529 vector signed__int128);
1530 vector unsigned __int128 vec_sube (vector unsigned __int128,
1531 vector unsigned __int128,
1532 vector unsigned __int128);
1533 vector signed int vec_subec (vector signed int, vector signed int,
1535 vector unsigned int vec_subec (vector unsigned int,
1536 vector unsigned int,
1537 vector unsigned int);
1538 vector signed __int128 vec_subec (vector signed __int128,
1539 vector signed __int128,
1540 vector signed__int128);
1541 vector unsigned __int128 vec_subec (vector unsigned __int128,
1542 vector unsigned __int128,
1543 vector unsigned __int128);
1544 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
1545 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
1546 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
1547 BU_ALTIVEC_OVERLOAD_X definitions.
1548 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
1549 * doc/extend.texi: Update the built-in documentation file for the new
1552 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1555 * diagnostic-show-locus.c (layout::layout): Use start and finish
1556 spelling location for the start and finish of each range.
1557 * genmatch.c (linemap_client_expand_location_to_spelling_point):
1558 Add unused aspect param.
1559 * input.c (expand_location_1): Add "aspect" param, and use it
1560 to access the correct part of the location.
1561 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
1563 (expand_location_to_spelling_point): Likewise.
1564 (linemap_client_expand_location_to_spelling_point): Add "aspect"
1565 param, and pass it to expand_location_1.
1567 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
1569 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
1570 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
1571 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
1572 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
1573 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
1574 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
1575 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
1576 _mm_maskz_getmant_ss): New intrinsics.
1577 (__builtin_ia32_getexpss128_mask): Changed to ...
1578 __builtin_ia32_getexpss128_round ... this.
1579 (__builtin_ia32_getexpsd128_mask): Changed to ...
1580 __builtin_ia32_getexpsd128_round ... this.
1581 * config/i386/i386-builtin-types.def
1582 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
1583 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
1584 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
1585 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
1586 __builtin_ia32_getmantss_mask_round): New builtins.
1587 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
1588 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
1589 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
1590 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
1591 * config/i386/sse.md
1592 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
1593 avx512f_sgetexp<mode><mask_scalar_name>
1594 <round_saeonly_scalar_name> ... this.
1595 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
1596 %0, %1, %2<round_saeonly_op3>}): Changed to ...
1597 vgetexp<ssescalarmodesuffix>
1598 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1599 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
1600 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
1601 avx512f_vgetmant<mode><mask_scalar_name>
1602 <round_saeonly_scalar_name> ... this.
1603 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
1604 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
1605 vgetmant<ssescalarmodesuffix>
1606 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
1607 %0<mask_scalar_operand4>, %1, %2
1608 <round_saeonly_scalar_mask_op4>, %3} ... this.
1609 * config/i386/subst.md (mask_scalar_operand4,
1610 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
1611 round_saeonly_scalar_nimm_predicate): New subst attributes.
1613 2017-07-06 Julia Koval <julia.koval@intel.com>
1615 * gcc/config/i386/i386.c (ix86_erase_embedded_rounding):
1616 Remove code for old rounding pattern.
1618 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
1620 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
1622 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
1624 * doc/sourcebuild.texi (Test Directives, Variants of
1625 dg-require-support): Add documentation for dg-require-stack-check.
1627 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
1629 * config/i386/subst.md (mask_scalar, round_scalar,
1630 round_saeonly_scalar): New meta-templates.
1631 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
1632 round_scalar_mask_operand3, round_scalar_mask_op3,
1633 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
1634 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
1635 round_saeonly_scalar_constraint,
1636 round_saeonly_scalar_prefix): New subst attribute.
1637 * config/i386/sse.md
1638 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
1639 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
1640 <round_scalar_name> ... this.
1641 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
1642 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
1643 <round_scalar_name> ... this.
1644 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
1645 <sse>_vm<code><mode>3<mask_scalar_name>
1646 <round_saeonly_scalar_name> ... this.
1647 (v<plusminus_mnemonic><ssescalarmodesuffix>
1648 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
1649 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
1650 v<plusminus_mnemonic><ssescalarmodesuffix>
1651 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1652 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
1653 (v<multdiv_mnemonic><ssescalarmodesuffix>
1654 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
1655 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
1656 v<multdiv_mnemonic><ssescalarmodesuffix>
1657 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1658 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
1659 (v<maxmin_float><ssescalarmodesuffix>
1660 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
1661 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
1662 v<maxmin_float><ssescalarmodesuffix>
1663 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1664 %0<mask_scalar_operand3>, %1, %<iptr>2
1665 <round_saeonly_scalar_mask_op3>} ... this.
1667 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
1669 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
1670 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
1672 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
1673 Alan Hayward <alan.hayward@arm.com>
1674 David Sherwood <david.sherwood@arm.com>
1676 * combine.c (simplify_if_then_else): Remove "enum" before
1678 * compare-elim.c (can_eliminate_compare): Likewise.
1679 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1681 (aarch64_lookup_simd_builtin_type): Likewise.
1682 (aarch64_simd_builtin_type): Likewise.
1683 (aarch64_init_simd_builtin_types): Likewise.
1684 (aarch64_simd_expand_args): Likewise.
1685 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
1687 (aarch64_reverse_mask): Likewise.
1688 (aarch64_simd_emit_reg_reg_move): Likewise.
1689 (aarch64_gen_adjusted_ldpstp): Likewise.
1690 (aarch64_ccmp_mode_to_code): Likewise.
1691 (aarch64_operands_ok_for_ldpstp): Likewise.
1692 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1693 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
1695 (aarch64_min_divisions_for_recip_mul): Likewise.
1696 (aarch64_reassociation_width): Likewise.
1697 (aarch64_get_condition_code_1): Likewise.
1698 (aarch64_simd_emit_reg_reg_move): Likewise.
1699 (aarch64_simd_attr_length_rglist): Likewise.
1700 (aarch64_reverse_mask): Likewise.
1701 (aarch64_operands_ok_for_ldpstp): Likewise.
1702 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1703 (aarch64_gen_adjusted_ldpstp): Likewise.
1704 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
1706 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
1707 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1708 (arm_lookup_simd_builtin_type): Likewise.
1709 (arm_simd_builtin_type): Likewise.
1710 (arm_init_simd_builtin_types): Likewise.
1711 (arm_expand_builtin_args): Likewise.
1712 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
1713 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
1714 (ft32_setup_incoming_varargs): Likewise.
1715 (ft32_function_arg): Likewise.
1716 (ft32_function_arg_advance): Likewise.
1717 (ft32_pass_by_reference): Likewise.
1718 (ft32_arg_partial_bytes): Likewise.
1719 (ft32_valid_pointer_mode): Likewise.
1720 (ft32_addr_space_pointer_mode): Likewise.
1721 (ft32_addr_space_legitimate_address_p): Likewise.
1722 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
1724 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
1725 (ix86_emit_outlined_ms2sysv_restore): Likewise.
1726 (iamcu_alignment): Likewise.
1727 (canonicalize_vector_int_perm): Likewise.
1728 (ix86_noce_conversion_profitable_p): Likewise.
1729 (ix86_mpx_bound_mode): Likewise.
1730 (ix86_operands_ok_for_move_multiple): Likewise.
1731 * config/microblaze/microblaze-protos.h
1732 (microblaze_expand_conditional_branch_reg): Likewise.
1733 * config/microblaze/microblaze.c
1734 (microblaze_expand_conditional_branch_reg): Likewise.
1735 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1737 (rs6000_reassociation_width): Likewise.
1738 (rs6000_invalid_binary_op): Likewise.
1739 (fusion_p9_p): Likewise.
1740 (emit_fusion_p9_load): Likewise.
1741 (emit_fusion_p9_store): Likewise.
1742 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
1744 (riscv_hard_regno_mode_ok_p): Likewise.
1745 (riscv_address_insns): Likewise.
1746 (riscv_split_symbol): Likewise.
1747 (riscv_legitimize_move): Likewise.
1748 (riscv_function_value): Likewise.
1749 (riscv_hard_regno_nregs): Likewise.
1750 (riscv_expand_builtin): Likewise.
1751 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
1752 (riscv_build_integer): Likewise.
1753 (riscv_split_integer): Likewise.
1754 (riscv_legitimate_constant_p): Likewise.
1755 (riscv_cannot_force_const_mem): Likewise.
1756 (riscv_regno_mode_ok_for_base_p): Likewise.
1757 (riscv_valid_base_register_p): Likewise.
1758 (riscv_valid_offset_p): Likewise.
1759 (riscv_valid_lo_sum_p): Likewise.
1760 (riscv_classify_address): Likewise.
1761 (riscv_legitimate_address_p): Likewise.
1762 (riscv_address_insns): Likewise.
1763 (riscv_load_store_insns): Likewise.
1764 (riscv_force_binary): Likewise.
1765 (riscv_split_symbol): Likewise.
1766 (riscv_force_address): Likewise.
1767 (riscv_legitimize_address): Likewise.
1768 (riscv_move_integer): Likewise.
1769 (riscv_legitimize_const_move): Likewise.
1770 (riscv_legitimize_move): Likewise.
1771 (riscv_address_cost): Likewise.
1772 (riscv_subword): Likewise.
1773 (riscv_output_move): Likewise.
1774 (riscv_canonicalize_int_order_test): Likewise.
1775 (riscv_emit_int_order_test): Likewise.
1776 (riscv_function_arg_boundary): Likewise.
1777 (riscv_pass_mode_in_fpr_p): Likewise.
1778 (riscv_pass_fpr_single): Likewise.
1779 (riscv_pass_fpr_pair): Likewise.
1780 (riscv_get_arg_info): Likewise.
1781 (riscv_function_arg): Likewise.
1782 (riscv_function_arg_advance): Likewise.
1783 (riscv_arg_partial_bytes): Likewise.
1784 (riscv_function_value): Likewise.
1785 (riscv_pass_by_reference): Likewise.
1786 (riscv_setup_incoming_varargs): Likewise.
1787 (riscv_print_operand): Likewise.
1788 (riscv_elf_select_rtx_section): Likewise.
1789 (riscv_save_restore_reg): Likewise.
1790 (riscv_for_each_saved_reg): Likewise.
1791 (riscv_register_move_cost): Likewise.
1792 (riscv_hard_regno_mode_ok_p): Likewise.
1793 (riscv_hard_regno_nregs): Likewise.
1794 (riscv_class_max_nregs): Likewise.
1795 (riscv_memory_move_cost): Likewise.
1796 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
1797 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
1798 (rl78_addr_space_address_mode): Likewise.
1799 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1801 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1802 (rs6000_reassociation_width): Likewise.
1803 (rs6000_invalid_binary_op): Likewise.
1804 (fusion_p9_p): Likewise.
1805 (emit_fusion_p9_load): Likewise.
1806 (emit_fusion_p9_store): Likewise.
1807 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
1808 (ok_for_simple_move_operands): Likewise.
1809 (ok_for_simple_move_strict_operands): Likewise.
1810 (ok_for_simple_arith_logic_operands): Likewise.
1811 (visium_legitimize_reload_address): Likewise.
1812 (visium_select_cc_mode): Likewise.
1813 (output_cbranch): Likewise.
1814 (visium_split_double_move): Likewise.
1815 (visium_expand_copysign): Likewise.
1816 (visium_expand_int_cstore): Likewise.
1817 (visium_expand_fp_cstore): Likewise.
1818 * config/visium/visium.c (visium_pass_by_reference): Likewise.
1819 (visium_function_arg): Likewise.
1820 (visium_function_arg_advance): Likewise.
1821 (visium_libcall_value): Likewise.
1822 (visium_setup_incoming_varargs): Likewise.
1823 (visium_legitimate_constant_p): Likewise.
1824 (visium_legitimate_address_p): Likewise.
1825 (visium_legitimize_address): Likewise.
1826 (visium_secondary_reload): Likewise.
1827 (visium_register_move_cost): Likewise.
1828 (visium_memory_move_cost): Likewise.
1829 (prepare_move_operands): Likewise.
1830 (ok_for_simple_move_operands): Likewise.
1831 (ok_for_simple_move_strict_operands): Likewise.
1832 (ok_for_simple_arith_logic_operands): Likewise.
1833 (visium_function_value_1): Likewise.
1834 (rtx_ok_for_offset_p): Likewise.
1835 (visium_legitimize_reload_address): Likewise.
1836 (visium_split_double_move): Likewise.
1837 (visium_expand_copysign): Likewise.
1838 (visium_expand_int_cstore): Likewise.
1839 (visium_expand_fp_cstore): Likewise.
1840 (visium_split_cstore): Likewise.
1841 (visium_select_cc_mode): Likewise.
1842 (visium_split_cbranch): Likewise.
1843 (output_cbranch): Likewise.
1844 (visium_print_operand_address): Likewise.
1845 * expmed.c (flip_storage_order): Likewise.
1846 * expmed.h (emit_cstore): Likewise.
1847 (flip_storage_order): Likewise.
1848 * genrecog.c (validate_pattern): Likewise.
1849 * hsa-gen.c (gen_hsa_addr): Likewise.
1850 * internal-fn.c (expand_arith_overflow): Likewise.
1851 * ira-color.c (allocno_copy_cost_saving): Likewise.
1852 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1853 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
1854 (process_invariant_for_inheritance): Likewise.
1855 * lra-eliminations.c (move_plus_up): Likewise.
1856 * omp-low.c (lower_oacc_reductions): Likewise.
1857 * simplify-rtx.c (simplify_subreg): Likewise.
1858 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
1859 (TARGET_CHKP_BOUND_MODE): Likewise..
1860 * targhooks.c (default_chkp_bound_mode): Likewise.
1861 (default_setup_incoming_vararg_bounds): Likewise.
1862 * targhooks.h (default_chkp_bound_mode): Likewise.
1863 (default_setup_incoming_vararg_bounds): Likewise.
1864 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
1865 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
1866 (have_whole_vector_shift): Likewise.
1867 * tree-vect-stmts.c (vectorizable_load): Likewise.
1868 * doc/tm.texi: Regenerate.
1870 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
1872 Graceful degrade if Binutils PR21472 is not available.
1875 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
1876 .rodata in flash test fails.
1877 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
1878 * confgure: Regenerate.
1879 * config.in: Regenerate.
1880 * config/avr/avr.c (avr_asm_named_section)
1881 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
1882 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
1883 (avr_asm_init_sections): Same.
1885 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1887 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
1888 (fma<VH:mode>4_intrinsic): Likewise.
1889 (*fmsub<VCVTF:mode>4): Likewise.
1890 (*fmsub<VH:mode>4_intrinsic): Likewise.
1892 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
1895 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
1896 Don't depend on "optimize > 0".
1897 (out_movhi_r_mr, out_movqi_mr_r): Same.
1898 (out_movhi_mr_r, out_movqi_r_mr): Same.
1899 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
1900 io_address_operand on "optimize > 0".
1902 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1904 * tree-loop-distribution.c: Add general explanantion on the pass.
1905 (generate_loops_for_partition): Mark distributed loop.
1906 (pg_add_dependence_edges): New parameter. Handle alias data
1907 dependence specially and record it in the parameter if asked.
1908 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
1909 (init_partition_graph_vertices, add_partition_graph_edge): New.
1910 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
1911 (free_partition_graph_vdata, build_partition_graph): New.
1912 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
1913 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
1914 (data_ref_segment_size, latch_dominated_by_data_ref): New.
1915 (compute_alias_check_pairs, version_loop_by_alias_check): New.
1916 (version_for_distribution_p, finalize_partitions): New.
1917 (distribute_loop): Handle alias data dependence specially. Factor
1918 out loop fusion code as functions and call these functions.
1920 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1922 * tree-loop-distribution.c (classify_partition): New parameter and
1923 better handle reduction statement.
1924 (rdg_build_partitions): Revise comment.
1925 (distribute_loop): Compute statements in all partitions and pass it
1926 to classify_partition.
1928 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1930 * tree-loop-distribution.c (enum partition_type): New.
1931 (struct partition): New field type.
1932 (partition_merge_into): Add parameter. Update partition type.
1933 (data_dep_in_cycle_p, update_type_for_merge): New functions.
1934 (build_rdg_partition_for_vertex): Compute partition type.
1935 (rdg_build_partitions): Dump partition type.
1936 (distribute_loop): Update calls to partition_merge_into.
1938 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1940 * tree-loop-distribution.c (struct ddr_hasher): New.
1941 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
1943 (classify_partition): Call get_data_dependence.
1944 (pg_add_dependence_edges): Ditto.
1945 (distribute_loop): Release data dependence hash table.
1947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1949 * tree-loop-distribution.c (ref_base_address): Delete.
1950 (similar_memory_accesses): Rename ...
1951 (share_memory_accesses): ... to this. Check if partitions access
1952 the same memory reference.
1953 (distribute_loop): Call share_memory_accesses.
1955 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1957 * tree-loop-distribution.c (struct partition): New field recording
1959 (partition_alloc, partition_free): Init and release data refs.
1960 (partition_merge_into): Merge data refs.
1961 (build_rdg_partition_for_vertex): Collect data refs for partition.
1962 (pg_add_dependence_edges): Change parameters from vector to bitmap.
1964 (distribute_loop): Remve data refs from vertice data of partition
1967 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1969 * tree-loop-distribution.c (params.h): Include header file.
1970 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
1971 (datarefs_vec): New global var.
1972 (create_rdg_vertices): Use datarefs_vec directly.
1973 (free_rdg): Don't free data references.
1974 (build_rdg): Update use. Don't free data references.
1975 (distribute_loop): Compute global variable for data references.
1976 Bail out if there are too many data references.
1978 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1980 * tree-loop-distribution.c (loop_nest): New global var.
1981 (build_rdg): Use loop directly, rather than loop nest.
1982 (pg_add_dependence_edges): Remove loop nest parameter. Use global
1984 (distribute_loop): Compute global variable loop nest. Update use.
1986 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1988 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
1989 (partition_merge_into): New parameter. Dump reason for fusion.
1990 (distribute_loop): Update use of partition_merge_into.
1992 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1994 * tree-loop-distribution.c (bb_top_order_index): New.
1995 (bb_top_order_index_size, bb_top_order_cmp): New.
1996 (stmts_from_loop): Use topological order.
1997 (pass_loop_distribution::execute): Compute and release topological
1998 order for basic blocks.
2000 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2002 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2005 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2007 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2008 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2009 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2010 * internal-fn.def (LOOP_DIST_ALIAS): New.
2011 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2012 (fold_loop_internal_call): ... this.
2013 (vect_loop_dist_alias_call): New function.
2014 (set_uid_loop_bbs): Call fold_loop_internal_call.
2015 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2018 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2021 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2022 Require dead FLAGS_REG at the beginning of a peephole.
2024 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2027 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2028 arguments in the call to __builtin_ia32_sbb_u32.
2029 (_subborrow_u64): Swap _X and _Y arguments in the call to
2030 __builtin_ia32_sbb_u64.
2032 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2035 * tree-vrp.c (compare_assert_loc): Turn into a function template
2036 with stable template parameter. Only test if a->e is NULL,
2037 !a->e == !b->e has been verified already. Use e == NULL or
2038 e != NULL instead of e or ! e tests. If stable is true, don't use
2039 iterative_hash_expr, on the other side allow a or b or both NULL
2040 and sort the NULLs last.
2041 (process_assert_insertions): Sort using compare_assert_loc<false>
2042 instead of compare_assert_loc, later sort using
2043 compare_assert_loc<true> before calling process_assert_insertions_for
2044 in a loop. Use break instead of continue once seen NULL pointer.
2046 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2048 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2049 Cortex-R7 and Cortex-R8 processors.
2051 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2053 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2054 uninitialized while src is not.
2056 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2058 * common/config/arm/arm-common.c: Adjust include path for
2060 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2061 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2062 (arm-cpu-data.h): Likewise.
2063 (arm-cpu-cdata.h): Likewise.
2064 * config/arm/arm-cpu.h: Delete.
2065 * config/arm/arm-cpu-cdata.h: Delete.
2066 * config/arm/arm-cpu-data.h: Delete.
2068 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2070 * config/arm/arm-cpus.in (cortex-a55): New.
2071 (cortex-a75): Likewise.
2072 (cortex-a75.cortex-a55): Likewise.
2073 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2075 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2076 * config/arm/arm-cpu-cdata.h: Regenerate.
2077 * config/arm/arm-cpu-data.h: Regenerate.
2078 * config/arm/arm-cpu.h: Regenerate.
2079 * config/arm/arm-tables.opt: Regenerate.
2080 * config/arm/arm-tune.md: Regenerate.
2082 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2084 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2086 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2088 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2091 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2093 PR tree-optimization/81292
2094 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2095 full_string_p, also call adjust_related_strinfos if the adjustment
2096 is simple, otherwise invalidate related strinfos.
2098 2017-07-04 Martin Liska <mliska@suse.cz>
2101 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2102 newly created variable as DECL_IGNORED_P.
2104 2017-07-04 Martin Liska <mliska@suse.cz>
2107 * ipa-inline.c (inline_small_functions):
2108 Use xstrdup_for_dump.
2110 2017-07-04 Tom de Vries <tom@codesourcery.com>
2112 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2114 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2117 * config/darwin.c (darwin_function_switched_text_sections):
2120 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2122 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2124 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2126 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2128 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2130 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2131 min_profitable_iters, and th as inclusive lower bounds.
2132 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2133 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2134 for min_profitable_iters and min_profitable_estimate.
2135 (vect_transform_loop): Treat th as an inclusive lower bound.
2136 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2138 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2141 * config/darwin.c (darwin_function_switched_text_sections):
2142 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2143 in two pieces, and suppress the use of buf.
2145 2017-07-03 Nathan Sidwell <nathan@acm.org>
2147 * hash-table.h (hash_table_mod1): Fix indentation.
2149 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2152 * predict.c (force_edge_cold): Be more careful about propagation
2154 * profile-count.h (profile_probability::guessed,
2155 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2157 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2159 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2161 * doc/invoke.texi (rcpc architecture extension): Document it.
2163 2017-07-03 Richard Biener <rguenther@suse.de>
2165 PR tree-optimization/60510
2166 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2167 the scalar reduction PHI and use it.
2168 (vectorizable_reduction): Properly guard the single_defuse_cycle
2169 path for non-SLP reduction chains where we cannot use it.
2170 Rework reduc_def/index and vector type deduction. Rework
2171 vector operand gathering during reduction op code-gen.
2172 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2173 chains dissolve the chain and leave it to non-SLP reduction
2176 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2178 * tree-data-ref.h (dr_alignment): Declare.
2179 * tree-data-ref.c (dr_alignment): New function.
2180 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2181 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2183 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2185 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2187 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2188 and base_misalignment fields.
2189 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2190 * tree-data-ref.c: Include builtins.h.
2191 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2192 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2193 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2194 * tree-vect-data-refs.c: Include tree-cfg.h.
2195 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2196 fields instead of calculating an alignment here.
2197 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2198 innermost_loop_behavior fields.
2200 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2202 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2204 (DR_STEP_ALIGNMENT): New macro.
2205 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2206 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2207 (create_data_ref): Print it.
2208 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2209 to tell whether the step preserves vector (mis)alignment.
2210 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2211 Move the check for an integer step and generalise to all INTEGER_CST.
2212 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2213 Print the outer step alignment.
2215 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2217 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2218 with offset_alignment.
2219 (DR_ALIGNED_TO): Delete.
2220 (DR_OFFSET_ALIGNMENT): New macro.
2221 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2222 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2223 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2224 (create_data_ref): Likewise.
2225 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2226 (vect_analyze_data_refs): Likewise.
2227 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2228 creating dummy innermost behavior.
2230 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2232 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2233 with a "innermost_loop_behavior *" and refeence tree.
2234 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2235 (create_data_ref): Update call accordingly.
2236 * tree-predcom.c (find_looparound_phi): Likewise.
2238 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2240 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2241 fields with dr_wrt_vec_loop.
2242 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2243 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2244 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2245 (vect_dr_behavior): New function.
2246 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2247 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2248 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2249 track whether the step preserves the misalignment.
2250 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2251 Use vect_dr_behavior.
2252 (vect_setup_realignment): Update call accordingly.
2253 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2254 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2255 call to vect_create_addr_base_for_vector_ref.
2256 (vect_create_cond_for_align_checks): Likewise.
2257 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2258 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2259 (vect_recog_mask_conversion_pattern): Likewise.
2260 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2261 (new_stmt_vec_info): Remove redundant zeroing.
2263 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2265 * common/config/arm/arm-common.c (arm_be8_option): New function.
2266 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2267 (ISA_ARMv6): Add isa_bit_be8.
2268 * config/arm/arm.h (arm_be8_option): Add prototype.
2269 (BE8_SPEC_FUNCTION): New define.
2270 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2271 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2272 (mlittle-endian): Similarly.
2273 (mbe8, mbe32): New options.
2274 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2275 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2277 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2279 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2281 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2283 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2284 (cleanup_tree_cfg_bb): Use it.
2285 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2287 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2289 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2292 * loop-doloop.c (add_test): Update profile.
2294 2017-07-03 Martin Liska <mliska@suse.cz>
2297 * sanopt.c (rewrite_usage_of_param): New function.
2298 (sanitize_rewrite_addressable_params): Likewise.
2299 (pass_sanopt::execute): Call rewrite_usage_of_param.
2301 2017-07-03 Richard Biener <rguenther@suse.de>
2303 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2304 back to using VIEW_CONVERT_EXPR.
2306 2017-07-03 Martin Liska <mliska@suse.cz>
2309 * doc/extend.texi: Document when a resolver function is
2310 generated for target_clones.
2312 2017-07-03 Martin Liska <mliska@suse.cz>
2314 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2315 and shadow memory of auto variables which are subject of
2316 use-after-scope sanitization.
2317 (asan_expand_mark_ifn): Add do set only when is_poison.
2319 2016-07-03 Richard Biener <rguenther@suse.de>
2321 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2323 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2324 (vectorizable_reduction): Perform reduction PHI creation when
2325 visiting a reduction PHI and adjust and simplify code generation
2326 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2327 (vect_transform_loop): Visit reduction PHIs.
2328 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2329 defs into the SLP tree.
2330 (vect_build_slp_tree): Reduction defs terminate the recursion.
2331 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2333 (vect_get_vec_defs_for_stmt_copy): Export.
2334 (vect_get_vec_defs): Likewise.
2335 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2337 (vect_get_vec_defs_for_stmt_copy): Declare.
2338 (vect_get_vec_defs): Likewise.
2340 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2342 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2343 parameter with a "loop" parameter and use it instead of the
2344 loop containing DR_STMT. Don't check simple_iv when doing
2345 BB analysis. Describe the two analysis modes in the comment.
2347 2017-07-03 Tom de Vries <tom@codesourcery.com>
2349 PR tree-optimization/69468
2350 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2351 (find_same_succ_bb): Handle ignore_edge_flags.
2353 2017-07-03 Tom de Vries <tom@codesourcery.com>
2355 PR tree-optimization/81192
2356 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2358 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2360 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2362 2017-07-03 Tom de Vries <tom@codesourcery.com>
2364 PR tree-optimization/81192
2365 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2366 BB_SAME_SUCC (bb) == NULL.
2368 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2370 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2373 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2375 * dumpfile.c: Include profile-count.h
2376 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2378 (insert_cond_bb): Update profile.
2379 * tree-cfg.h (insert_cond_bb): Update prototype.
2380 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2381 * tree-dump.c: Do not include tree-cfg.
2383 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2385 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2387 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2389 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2394 * tree-complex.c (expand_complex_div_wide): update profile.
2396 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2397 Alan Hayward <alan.hayward@arm.com>
2398 David Sherwood <david.sherwood@arm.com>
2400 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2401 (CORETYPES_H): New define.
2402 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2403 (insn-modes-inline.h, s-modes-inline-h): New rules.
2404 (generated_files): Add insn-modes-inline.h.
2405 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2406 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2407 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2408 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2409 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2410 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2411 (build/gencodes.o, build/genconditions.o): Likewise.
2412 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
2413 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
2414 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
2415 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
2416 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
2417 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
2418 * coretypes.h: Include everything up to real.h for generators.
2419 Include insn-modes.h first. Include wide-int-print.h after
2420 wide-int.h. Include insn-modes-inline.h and then machmode.h.
2421 * machmode.h: Don't include insn-modes.h here.
2422 * function-tests.c: Remove includes of signop.h, machmode.h,
2423 double-int.h and wide-int.h.
2425 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
2427 * optc-save-gen.awk: Likewise.
2428 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
2429 * godump.c: Remove include of wide-int-print.h.
2430 * pretty-print.h: Likewise.
2431 * wide-int-print.cc: Likewise.
2432 * wide-int.cc: Likewise.
2433 * hash-map-tests.c: Remove include of signop.h.
2434 * hash-set-tests.c: Likewise.
2435 * rtl-tests.c: Likewise.
2436 * mkconfig.sh: Remove include of machmode.h.
2437 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
2439 (emit_insn_modes_inline_h): ...this new function. Emit the code
2440 into an insn-modes-inline.h header file, adding appropriate
2441 include guards and end comments.
2442 (emit_insn_modes_c_header): Remove include of machmode.h.
2443 (emit_min_insn_modes_c_header): Include coretypes.h rather than
2445 (main): Handle -i flag and call emit_insn_modes_inline_h when
2448 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2450 * tree-ssa-strlen.c (strinfo): Rename the length field to
2451 nonzero_chars. Add a full_string_p field.
2452 (compare_nonzero_chars, zero_length_string_p): New functions.
2453 (get_addr_stridx): Add an offset_out parameter.
2454 Use compare_nonzero_chars.
2455 (get_stridx): Update accordingly. Use compare_nonzero_chars.
2456 (new_strinfo): Update after above changes to strinfo.
2457 (set_endptr_and_length): Set full_string_p.
2458 (get_string_length): Update after above changes to strinfo.
2459 (unshare_strinfo): Update call to new_strinfo.
2460 (maybe_invalidate): Likewise.
2461 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
2462 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
2463 as a uhwi instead of an shwi. Update after above changes to
2464 strinfo and new_strinfo.
2465 (zero_length_string): Assert that chainsi contains full strings.
2466 Use zero_length_string_p. Update call to new_strinfo.
2467 (adjust_related_strinfos): Update after above changes to strinfo.
2468 Copy full_string_p from origsi.
2469 (adjust_last_stmt): Use zero_length_string_p.
2470 (handle_builtin_strlen): Update after above changes to strinfo and
2471 new_strinfo. Install the lhs as the string length if the previous
2472 entry didn't describe a full string.
2473 (handle_builtin_strchr): Update after above changes to strinfo
2475 (handle_builtin_strcpy): Likewise.
2476 (handle_builtin_strcat): Likewise.
2477 (handle_builtin_malloc): Likewise.
2478 (handle_pointer_plus): Likewise.
2479 (handle_builtin_memcpy): Likewise. Track nonzero characters
2480 that aren't necessarily followed by a nul terminator.
2481 (handle_char_store): Likewise.
2483 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2485 PR tree-optimization/80769
2486 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
2487 for malloc and calloc. Document the new invariant that all related
2488 strinfos have delayed lengths or none do.
2489 (verify_related_strinfos): Move earlier in file.
2490 (set_endptr_and_length): New function, split out from...
2491 (get_string_length): ...here. Also set the lengths of related
2493 (zero_length_string): Assert that chainsi has known (rather than
2495 (adjust_related_strinfos): Likewise.
2497 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2499 PR tree-optimization/81136
2500 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
2501 assert that two references with the same misalignment have the same
2502 compile-time misalignment if those compile-time misalignments
2505 2017-07-01 Andi Kleen <ak@linux.intel.com>
2507 * print-tree.c (print_node): Print all attributes.
2509 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2511 * cfg.c (scale_bbs_frequencies): New function.
2512 * cfg.h (scale_bbs_frequencies): Declare it.
2513 * cfgloopanal.c (single_likely_exit): Cleanup.
2514 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
2516 (scale_loop_profile): Likewise.
2517 (loop_version): Likewise.
2518 (create_empty_loop_on_edge): Update.
2519 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
2520 scale_loop_frequencies, scale_loop_profile, loopify,
2521 loop_version): Update prototypes.
2522 * modulo-sched.c (sms_schedule): Update.
2523 * predict.c (unlikely_executed_edge_p): Also check probability.
2524 (probably_never_executed_edge_p): Fix typo.
2525 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2526 * tree-parloops.c (gen_parallel_loop): Update.
2527 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
2528 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2529 * tree-ssa-loop-split.c (split_loop): Update.
2530 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2531 * tree-vect-loop-manip.c (vect_do_peeling): Update.
2532 (vect_loop_versioning): Update.
2533 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2535 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2537 * trans-mem.c (split_bb_make_tm_edge): Update profile.
2539 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2541 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
2542 to keep profile consistent.
2544 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2546 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
2547 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
2548 * profile-count.h (max_safe_multiplier): Make unsigned.
2549 (profile_count::guessed_zero): New.
2551 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2553 * bb-reorder.c (fix_up_crossing_landing_pad,
2554 fix_crossing_conditional_branches): Use make_single_succ_edge
2555 to keep profile consistent.
2557 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2559 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
2562 2017-07-01 Jakub Jelinek <jakub@redhat.com>
2565 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
2566 the right scopes, make sure cond_jump isn't preserved between multiple
2567 iterations. Search for fallthru edge whenever there are 3+ edges and
2568 use find_fallthru_edge for it.
2570 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2572 Patch by Alexander Monakov <amonakov@ispras.ru>
2573 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
2574 probabilities consistently.
2576 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2578 * pa.c (pa_expand_compare_and_swap_loop): Update call of
2579 emit_cmp_and_jump_insns.
2581 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2584 * tree-inline.c (expand_call_inline): Combine profile statuses.
2586 2017-06-30 Andrew Pinski <apinski@cavium.com>
2588 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
2589 fold_stmt returned true.
2591 2017-06-30 Nathan Sidwell <nathan@acm.org>
2593 * ggc.h (empty_string): Delete.
2594 * cfgexpand.c (expand_asm_stmt): Use plain "".
2595 * optabs.c (expand_asm_memory_barrier): Likewise.
2596 * stringpool.c (empty_string): Delete.
2597 (digit_vector, digit_string): Delete.
2598 (ggc_alloc_string): Use plain "", don't optimize single digit
2599 strings. Use ggc_alloc_atomic.
2601 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
2603 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
2604 comparison set and one other set, use the cost of the non-comparison
2607 2017-06-30 Nathan Sidwell <nathan@acm.org>
2609 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
2612 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
2614 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
2615 loops. Remove now unneeded calls to gimple_switch_set_label() that
2616 just set removed labels to NULL_TREE.
2618 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
2620 * tree-ssanames.c (set_range_info_raw): Abstract from ...
2621 (set_range_info): ...here. Only call set_range_info_raw if domain
2623 (set_nonzero_bits): Call set_range_info_raw.
2624 * tree-ssanames.h (set_range_info_raw): New.
2626 2017-06-30 Jakub Jelinek <jakub@redhat.com>
2629 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
2630 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
2631 of nonimmediate_operand and <store_mask_constraint> instead of m
2632 for the input operand. For V8FI iterator, always split if input
2633 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
2634 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
2635 <store_mask_predicate> instead of register_operand and
2636 <store_mask_constraint> instead of v for the input operand. Make
2637 sure both operands aren't MEMs for if not <mask_applied>.
2639 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
2641 * lto-wrapper.c (copy_file) Close both file descriptors before
2644 2017-06-30 Martin Liska <mliska@suse.cz>
2647 * multiple_target.c (create_dispatcher_calls): Make ifunc
2648 also for function that don't have calls or are not referenced.
2650 2017-06-30 Richard Biener <rguenther@suse.de>
2652 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
2653 analyze the first scalar stmt. Move vector type computation
2654 for the BB case here from ...
2655 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
2656 live operation processing in the SLP case properly.
2658 2017-06-30 Richard Biener <rguenther@suse.de>
2660 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
2662 2017-06-30 Martin Liska <mliska@suse.cz>
2665 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
2666 before BUILT_IN_UNWIND_RESUME when ASAN is used.
2668 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
2670 * doc/invoke.texi (AArch64): Add missing options and remove redundant
2673 2017-06-30 Richard Biener <rguenther@suse.de>
2675 PR tree-optimization/81249
2676 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
2677 condition reduction result to original scalar type.
2679 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2681 * profile-count.h (enum profile_quality): Fix typos and whitespace
2684 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2686 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
2687 type for branch probabilities.
2689 2017-06-29 Julian Brown <julian@codesourcery.com>
2690 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2692 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
2693 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
2694 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
2695 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
2697 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2699 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
2700 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
2701 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
2702 CC usage from generic code to here.
2703 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
2704 CC usage into the target macros.
2706 2017-06-29 Maya Rashish <coypu@sdf.org>
2708 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
2711 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2713 * arm/arm-builtins.c: Include profile-count.h
2714 * except.c (sjlj_emit_function_enter): Use
2715 profile_probability::unlikely.
2717 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2719 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
2720 and tocrel_offset be pointer args rather than implicitly using
2722 (legitimate_constant_pool_address_p, rs6000_emit_move,
2723 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
2724 tocrel_offset and use in toc_relative_expr_p call.
2725 (print_operand, print_operand_address): Use static tocrel_base_oac
2726 and tocrel_offset_oac.
2727 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
2730 2017-06-29 Maya Rashish <coypu@sdf.org>
2732 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
2734 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
2736 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
2737 objects, take into account only the alignment of 'op0' and 'mode1' if
2740 2017-06-29 Steve Ellcey <sellcey@cavium.com>
2742 * ccmp.c (ccmp_tree_comparison_p): New function.
2743 (ccmp_candidate_p): Update to use above function.
2744 (get_compare_parts): New function.
2745 (expand_ccmp_next): Update to use new functions.
2746 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
2748 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
2749 take mode as argument.
2750 * ccmp.h (expand_ccmp_expr): Add mode as argument.
2751 * expr.c (expand_expr_real_1): Pass mode as argument.
2753 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
2755 * combine.c (combine_instructions): Print insns to dump_file, together
2758 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2760 * asan.c (asan_emit_stack_protection): Update.
2761 (create_cond_insert_point): Update.
2762 * auto-profile.c (afdo_propagate_circuit): Update.
2763 * basic-block.h (struct edge_def): Turn probability to
2764 profile_probability.
2765 (EDGE_FREQUENCY): Update.
2766 * bb-reorder.c (find_traces_1_round): Update.
2767 (better_edge_p): Update.
2768 (sanitize_hot_paths): Update.
2769 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
2770 (make_single_succ_edge): Update.
2771 (check_bb_profile): Update.
2772 (dump_edge_info): Update.
2773 (update_bb_profile_for_threading): Update.
2774 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
2775 probabilitycount to 0.
2776 * cfgbuild.c (compute_outgoing_frequencies): Update.
2777 * cfgcleanup.c (try_forward_edges): Update.
2778 (outgoing_edges_match): Update.
2779 (try_crossjump_to_edge): Update.
2780 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
2781 (expand_gimple_tailcall): Update.
2782 (construct_init_block): Use make_single_succ_edge.
2783 (construct_exit_block): Use make_single_succ_edge.
2784 * cfghooks.c (verify_flow_info): Update.
2785 (redirect_edge_succ_nodup): Update.
2786 (split_edge): Update.
2787 (account_profile_record): Update.
2788 * cfgloopanal.c (single_likely_exit): Update.
2789 * cfgloopmanip.c (scale_loop_profile): Update.
2790 (set_zero_probability): Remove.
2791 (duplicate_loop_to_header_edge): Update.
2792 * cfgloopmanip.h (loop_version): Update prototype.
2793 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2794 (force_nonfallthru_and_redirect): Update.
2795 (update_br_prob_note): Update.
2796 (rtl_verify_edges): Update.
2797 (purge_dead_edges): Update.
2798 (rtl_lv_add_condition_to_bb): Update.
2799 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
2800 * cgraphunit.c (init_lowered_empty_function): Update.
2801 (cgraph_node::expand_thunk): Update.
2802 * cilk-common.c: Include profile-count.h
2803 * dojump.c (inv): Remove.
2804 (jumpifnot): Update.
2805 (jumpifnot_1): Update.
2806 (do_jump_1): Update.
2808 (do_jump_by_parts_greater_rtx): Update.
2809 (do_compare_rtx_and_jump): Update.
2810 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
2811 do_jump_1. do_compare_rtx_and_jump): Update prototype.
2812 * dwarf2cfi.c: Include profile-count.h
2813 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
2814 (sjlj_emit_dispatch_table): Likewise.
2815 * explow.c: Include profile-count.h
2816 * expmed.c (emit_store_flag_force): Update.
2817 (do_cmp_and_jump): Update.
2818 * expr.c (compare_by_pieces_d::generate): Update.
2819 (compare_by_pieces_d::finish_mode): Update.
2820 (emit_block_move_via_loop): Update.
2821 (store_expr_with_bounds): Update.
2822 (store_constructor): Update.
2823 (expand_expr_real_2): Update.
2824 (expand_expr_real_1): Update.
2825 * expr.h (try_casesi, try_tablejump): Update prototypes.
2826 * gimple-pretty-print.c (dump_probability): Update.
2827 (dump_profile): New.
2828 (dump_gimple_label): Update.
2829 (dump_gimple_bb_header): Update.
2830 * graph.c (draw_cfg_node_succ_edges): Update.
2831 * hsa-gen.c (convert_switch_statements): Update.
2832 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
2833 (find_if_case_1): Update.
2834 (find_if_case_2): Update.
2835 * internal-fn.c (expand_arith_overflow_result_store): Update.
2836 (expand_addsub_overflow): Update.
2837 (expand_neg_overflow): Update.
2838 (expand_mul_overflow): Update.
2839 (expand_vector_ubsan_overflow): Update.
2840 * ipa-cp.c (good_cloning_opportunity_p): Update.
2841 * ipa-split.c (split_function): Use make_single_succ_edge.
2842 * ipa-utils.c (ipa_merge_profiles): Update.
2843 * loop-doloop.c (add_test): Update.
2844 (doloop_modify): Update.
2845 * loop-unroll.c (compare_and_jump_seq): Update.
2846 (unroll_loop_runtime_iterations): Update.
2847 * lra-constraints.c (lra_inheritance): Update.
2848 * lto-streamer-in.c (input_cfg): Update.
2849 * lto-streamer-out.c (output_cfg): Update.
2850 * mcf.c (adjust_cfg_counts): Update.
2851 * modulo-sched.c (sms_schedule): Update.
2852 * omp-expand.c (expand_omp_for_init_counts): Update.
2853 (extract_omp_for_update_vars): Update.
2854 (expand_omp_ordered_sink): Update.
2855 (expand_omp_for_ordered_loops): Update.
2856 (expand_omp_for_generic): Update.
2857 (expand_omp_for_static_nochunk): Update.
2858 (expand_omp_for_static_chunk): Update.
2859 (expand_cilk_for): Update.
2860 (expand_omp_simd): Update.
2861 (expand_omp_taskloop_for_outer): Update.
2862 (expand_omp_taskloop_for_inner): Update.
2863 * omp-simd-clone.c (simd_clone_adjust): Update.
2864 * optabs.c (expand_doubleword_shift): Update.
2865 (expand_abs): Update.
2866 (emit_cmp_and_jump_insn_1): Update.
2867 (expand_compare_and_swap_loop): Update.
2868 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
2869 * predict.c (predictable_edge_p): Update.
2870 (edge_probability_reliable_p): Update.
2871 (set_even_probabilities): Update.
2872 (combine_predictions_for_insn): Update.
2873 (combine_predictions_for_bb): Update.
2874 (propagate_freq): Update.
2875 (estimate_bb_frequencies): Update.
2876 (force_edge_cold): Update.
2877 * profile-count.c (profile_count::dump): Add missing space into dump.
2878 (profile_count::debug): Add newline.
2879 (profile_count::differs_from_p): Explicitly convert to unsigned.
2880 (profile_count::stream_in): Update.
2881 (profile_probability::dump): New member function.
2882 (profile_probability::debug): New member function.
2883 (profile_probability::differs_from_p): New member function.
2884 (profile_probability::differs_lot_from_p): New member function.
2885 (profile_probability::stream_in): New member function.
2886 (profile_probability::stream_out): New member function.
2887 * profile-count.h (profile_count_quality): Rename to ...
2888 (profile_quality): ... this one.
2889 (profile_probability): New.
2890 (profile_count): Update.
2891 * profile.c (compute_branch_probabilities): Update.
2892 * recog.c (peep2_attempt): Update.
2893 * sched-ebb.c (schedule_ebbs): Update.
2894 * sched-rgn.c (find_single_block_region): Update.
2895 (compute_dom_prob_ps): Update.
2896 (schedule_region): Update.
2897 * sel-sched-ir.c (compute_succs_info): Update.
2898 * stmt.c (struct case_node): Update.
2899 (do_jump_if_equal): Update.
2900 (get_outgoing_edge_probs): Update.
2901 (conditional_probability): Update.
2902 (emit_case_dispatch_table): Update.
2903 (expand_case): Update.
2904 (expand_sjlj_dispatch_table): Update.
2905 (emit_case_nodes): Update.
2906 * targhooks.c: Update.
2907 * tracer.c (better_p): Update.
2908 (find_best_successor): Update.
2909 * trans-mem.c (expand_transaction): Update.
2910 * tree-call-cdce.c: Update.
2911 * tree-cfg.c (gimple_split_edge): Upate.
2912 (move_sese_region_to_fn): Upate.
2913 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
2914 * tree-eh.c (lower_resx): Upate.
2915 (cleanup_empty_eh_move_lp): Upate.
2916 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2917 * tree-inline.c (copy_edges_for_bb): Update.
2918 (copy_cfg_body): Update.
2919 * tree-parloops.c (gen_parallel_loop): Update.
2920 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
2921 (gimple_gen_time_profiler): Update.
2922 * tree-ssa-dce.c (remove_dead_stmt): Update.
2923 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2924 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2925 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2926 (unloop_loops): Update.
2927 (try_peel_loop): Update.
2928 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2929 * tree-ssa-loop-split.c (connect_loops): Update.
2930 (split_loop): Update.
2931 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2932 (hoist_guard): Update.
2933 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2934 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2935 (value_replacement): Update.
2936 * tree-ssa-reassoc.c (branch_fixup): Update.
2937 * tree-ssa-tail-merge.c (replace_block_by): Update.
2938 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2939 (create_edge_and_update_destination_phis): Update.
2940 (compute_path_counts): Update.
2941 (recompute_probabilities): Update.
2942 (update_joiner_offpath_counts): Update.
2943 (freqs_to_counts_path): Update.
2944 (duplicate_thread_path): Update.
2945 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2946 (struct switch_conv_info): Update.
2947 (gen_inbound_check): Update.
2948 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2949 (vect_do_peeling): Update.
2950 (vect_loop_versioning): Update.
2951 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2952 (optimize_mask_stores): Update.
2953 * ubsan.c (ubsan_expand_null_ifn): Update.
2954 * value-prof.c (gimple_divmod_fixed_value): Update.
2955 (gimple_divmod_fixed_value_transform): Update.
2956 (gimple_mod_pow2): Update.
2957 (gimple_mod_pow2_value_transform): Update.
2958 (gimple_mod_subtract): Update.
2959 (gimple_mod_subtract_transform): Update.
2960 (gimple_ic): Update.
2961 (gimple_stringop_fixed_value): Update.
2962 (gimple_stringops_transform): Update.
2963 * value-prof.h: Update.
2965 2017-06-29 Carl Love <cel@us.ibm.com>
2967 * config/rs6000/rs6000-c.c: Add support for built-in functions
2968 vector signed int vec_signed (vector float);
2969 vector signed long long vec_signed (vector double);
2970 vector signed int vec_signed2 (vector double, vector double);
2971 vector signed int vec_signede (vector double);
2972 vector signed int vec_signedo (vector double);
2973 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
2974 instruction generator.
2975 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
2976 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
2977 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
2979 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
2980 vunsignede_v2df): Add define_expands.
2981 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
2982 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
2983 VEC_UNSIGNEDO): Add definitions.
2984 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
2985 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
2986 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
2987 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
2988 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
2989 * config/rs6000/altivec.h (vec_signed, vec_signed2,
2990 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
2991 vec_unsignede, vec_unsignedo): Add builtin defines.
2992 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
2994 * doc/extend.texi: Update the built-in documentation file for the
2995 new built-in functions.
2997 2017-06-29 Richard Biener <rguenther@suse.de>
2999 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3000 reduction chains to LOOP_VINFO_REDUCTIONS.
3001 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3002 SLP reductions after processing reduction chains.
3004 2017-06-29 Nathan Sidwell <nathan@acm.org>
3006 * builtins.c (fold_builtin_FUNCTION): Use
3007 lang_hooks.decl_printable_name.
3009 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3012 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3013 with only one label.
3014 * stmt.c (expand_case): Assert NCASES is greater than one.
3016 2017-06-29 Richard Biener <rguenther@suse.de>
3018 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3020 (group_case_labels): Likewise.
3021 (find_taken_edge): Push sanity checking on val to workers...
3022 (find_taken_edge_cond_expr): ... here
3023 (find_taken_edge_switch_expr): ... and here, handle cases
3024 with just a default label.
3025 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3026 (group_case_labels): Likewise.
3027 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3028 group_case_labels does anything cleanup the CFG again.
3030 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3032 PR tree-optimization/81196
3033 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3034 exit condition comparing two IVs.
3036 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3038 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3039 profile to the dummy entry at the end of the list of architectures.
3040 * config/arm/arm-cpu-cdata.h: Regenerated.
3042 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3043 Michael Collison <michael.collison@arm.com>
3046 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3048 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3049 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3050 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3051 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3052 of shift when the shift amount is masked with constant equal to
3053 the size of the mode.
3054 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3057 2017-06-29 Martin Liska <mliska@suse.cz>
3059 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3061 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3064 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3067 (edge_growth_cache_entry::edge_growth_cache_entry): New
3069 (reset_edge_growth_cache): Update to use constructor.
3071 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3073 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3074 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3075 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3077 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3079 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3080 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3082 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3084 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3085 (*-linux-uclibc*): Add t-uclibc tmake_file.
3086 * config/t-musl: New.
3087 * config/t-uclibc: New.
3089 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3091 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3093 (gen_comm_data): Emit architectural setting of arch_prof.
3094 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3096 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3097 (armv8-m.base, armv8-m.main): Likewise.
3098 * arm-protos.h (arm_build_target): Add profile field.
3099 (arch_option): Likewise.
3100 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3102 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3103 arm_active_target.profile.
3105 2017-06-28 Richard Biener <rguenther@suse.de>
3108 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3109 TYPE_OVERFLOW_WRAPS.
3110 * match.pd (negate_expr_p): Likewise.
3111 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3112 fold_build2, not fold_binary.
3114 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3116 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3117 Convert memory address to Pmode.
3118 (aarch64_print_operand): Assert MEM operands are always Pmode.
3120 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3123 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3124 Remove redundant if.
3125 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3126 * config/arm/aarch-common-protos.h
3127 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3128 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3130 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3133 * multiple_target.c (create_dispatcher_calls): Set the default
3134 clone to be static, not public.
3136 2017-06-28 Richard Biener <rguenther@suse.de>
3138 * tree-vect-loop.c (vectorizable_reduction): Move special
3139 cond reduction IV var creation ...
3140 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3141 parameter. Use STMT_VINFO_VECTYPE.
3142 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3145 2017-06-28 Martin Liska <mliska@suse.cz>
3148 * ipa-visibility.c (non_local_p): Handle visibility.
3150 2017-06-28 Martin Liska <mliska@suse.cz>
3153 * common.opt: Add IntegerRange to various options.
3154 * opt-functions.awk (integer_range_info): New function.
3155 * optc-gen.awk: Add integer_range_info to cl_options struct.
3156 * opts-common.c (decode_cmdline_option): Handle
3157 CL_ERR_INT_RANGE_ARG.
3158 (cmdline_handle_error): Likewise.
3159 * opts.c (print_filtered_help): Show valid interval in
3160 when --help is provided.
3161 * opts.h (struct cl_option): Add range_min and range_max fields.
3162 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3164 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3166 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3167 (x * C EQ/NE y * C): New transformation.
3169 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3171 * genmultilib (combination_space): Accept '+' in option names.
3173 2017-06-28 Martin Liska <mliska@suse.cz>
3176 * asan.c (instrument_derefs): Bail out inner references
3177 that are hard register variables.
3179 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3182 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3183 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3185 2017-06-28 Richard Biener <rguenther@suse.de>
3187 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3188 (vect_get_slp_defs): Adjust.
3189 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3191 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3193 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3194 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3195 (vectorizable_reduction): Adjust.
3196 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3198 (vect_get_slp_defs): Likewise.
3199 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3200 (vectorizable_bswap): Adjust.
3201 (vectorizable_call): Likewise.
3202 (vectorizable_conversion): Likewise.
3203 (vectorizable_assignment): Likewise.
3204 (vectorizable_shift): Likewise.
3205 (vectorizable_operation): Likewise.
3206 (vectorizable_store): Likewise.
3207 (vectorizable_condition): Likewise.
3208 (vectorizable_comparison): Likewise.
3210 2017-06-28 Michael Collison <michael.collison@arm.com>
3213 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3215 (arm_gen_movmemqi): Removed unused variable 'i'.
3216 Convert 'for' loop into 'while' loop.
3217 (arm_expand_prologue): Remove last unnecessary set of insn.
3218 (thumb_pop): Remove unused variable 'pushed_words'.
3219 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3221 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3223 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3224 * config/s390/s390-protos.h: Add prototype of
3225 s390_rel_address_ok_p.
3226 * config/s390/s390.c (s390_got_symbol): New function.
3227 (s390_rel_address_ok_p): New function.
3228 (legitimize_pic_address): Use s390_rel_address_ok_p.
3229 (s390_load_got): Use s390_got_symbol.
3230 (s390_option_override): Issue error if
3231 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3232 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3234 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3236 2017-06-27 Andrew Pinski <apinski@cavium.com>
3238 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3239 (X * copysign (1.0, X)): New pattern.
3240 (X * copysign (1.0, -X)): New pattern.
3241 (copysign (-1.0, CST)): New pattern.
3243 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3245 * genmultilib (combination_space): Remove variable.
3246 Validate reuse rules against regular expression for any sequence
3247 of multilib options in any order.
3249 2017-06-27 Michael Collison <michael.collison@arm.com>
3251 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3252 call aarch64_split_simd_combine.
3253 * (aarch64_combine_internal<mode>): Delete pattern.
3254 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3255 Allow register and subreg operands.
3257 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3259 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3260 specific need, just fallback on defaults.
3261 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3263 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3264 Olivier Hainque <hainque@adacore.com>
3266 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3268 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3269 targets. Pick a default if no particular attempt applied.
3270 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3273 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3275 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3276 (x86_64-wrs-vxworks7): Likewise.
3278 2017-06-27 Marek Polacek <polacek@redhat.com>
3281 * ubsan.c (instrument_null): Check get_base_address's result for null.
3283 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3285 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3287 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3289 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3290 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3291 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3292 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3294 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3295 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3296 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3297 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3298 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3299 BUILT_IN_FEUPDATEENV): New builtins.
3300 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3301 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3302 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3303 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3305 (builtin_structptr_types): Adjust size.
3306 * tree.c (builtin_structptr_types): Add four entries.
3308 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3309 Olivier Hainque <hainque@adacore.com>
3311 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3312 (TLS_SYM): New local macro, forcing reference to __tls__ on
3313 link command lines for VxWorks 7 RTPs, triggering initialization
3315 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3316 OS features TLS support, true for RTPs on VxWorks 7.
3317 * config/vxworks.c (vxworks_override_options): Setup emutls
3320 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3322 * predict.c (test_prediction_value_range): Use -1U instead of -1
3323 to avoid narrowing conversion warning.
3324 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3325 to avoid narrowing conversion warning.
3326 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3328 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3330 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3332 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3333 64bit configurations.
3334 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3335 (SIZE_TYPE): Likewise.
3336 * config/vxworks.c (vxworks_emutls_var_fields): Use
3337 long_unsigned_type_node instead of unsigned_type_node as the offset
3338 field type, which is "pointer" mode in emutls.c.
3340 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3343 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3346 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3347 gimple_vuse copying separately from gimple_vdef copying.
3349 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3351 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3354 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3355 Olivier Hainque <hainque@adacore.com>
3357 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3358 for all vxworks7 targets.
3359 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3360 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3361 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3362 variations for VX6/VX7 and 32/64bits later on in ...
3363 (VXWORKS_LIB_SPEC): Leverage new macros.
3364 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3365 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3367 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3369 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3370 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3372 2017-06-26 Carl Love <cel@us.ibm.com>
3374 * config/rs6000/rs6000-c.c: Add support for built-in functions
3375 vector bool char vec_reve (vector bool char);
3376 vector signed char vec_reve (vector signed char);
3377 vector unsigned char vec_reve (vector unsigned char);
3378 vector bool int vec_reve (vector bool int);
3379 vector signed int vec_reve (vector signed int);
3380 vector unsigned int vec_reve (vector unsigned int);
3381 vector bool long long vec_reve (vector bool long long);
3382 vector signed long long vec_reve (vector signed long long);
3383 vector unsigned long long vec_reve (vector unsigned long long);
3384 vector bool short vec_reve (vector bool short);
3385 vector signed short vec_reve (vector signed short);
3386 vector double vec_reve (vector double);
3387 vector float vec_reve (vector float);
3388 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3389 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3390 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3391 (altivec_vreve): New pattern.
3392 * config/rs6000/altivec.h (vec_reve): New define.
3393 * doc/extend.texi (vec_rev): Update the built-in documentation file
3394 for the new built-in functions.
3396 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3398 PR tree-optimization/71815
3399 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3401 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3403 (slsr_process_phi): Likewise.
3404 (replace_uncond_cands_and_profitable_phis): Don't replace a
3405 multiply candidate with a stride of 1 (copy or cast).
3406 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3408 (lowest_cost_path): Likewise.
3409 (total_savings): Likewise.
3411 2017-06-26 Richard Biener <rguenther@suse.de>
3414 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
3415 Use def_builtin_pure for all gather builtins.
3417 2017-06-26 Richard Biener <rguenther@suse.de>
3419 PR tree-optimization/81203
3420 * tree-tailcall.c (find_tail_calls): Do not move stmts into
3423 2017-06-26 Marek Polacek <polacek@redhat.com>
3426 * doc/invoke.texi: Document -Wmultistatement-macros.
3428 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
3430 * doc/sourcebuild.texi (ARM-specific attributes): Document new
3431 arm_neon_ok_no_float_abi effective target.
3433 2017-06-26 Richard Biener <rguenther@suse.de>
3435 PR tree-optimization/80928
3436 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
3437 (copy_bbs): Set BB_DUPLICATED flag early.
3438 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
3440 (execute_on_shrinking_pred): Likewise.
3441 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
3442 BB_DUPLICATED blocks.
3443 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
3444 iterate over all PHIs considering removal of *gsi.
3446 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
3448 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
3451 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3453 * config/rs6000/rs6000-string.c: (expand_block_clear,
3454 do_load_for_compare, select_block_compare_mode,
3455 compute_current_alignment, expand_block_compare,
3456 expand_strncmp_align_check, expand_strn_compare,
3457 expand_block_move, rs6000_output_load_multiple)
3458 Move functions related to string/block move/compare
3460 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
3461 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
3462 for this function which is now used in two files.
3463 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
3464 * config.gcc: Add rs6000-string.o to extra_objs for
3465 targets powerpc*-*-* and rs6000*-*-*.
3467 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3470 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
3471 32-bit, since indexed is not valid for DImode.
3472 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
3473 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
3474 (define_peephole2 for Altivec d-form load): Add 32-bit support.
3475 (define_peephole2 for Altivec d-form store): Likewise.
3478 * multiple_target.c (create_dispatcher_calls): Only create the
3479 dispatcher call if the function is the default clone of a
3482 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
3485 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
3486 a call, force the call to not be a tail call.
3488 2017-06-23 Jeff Law <law@redhat.com>
3490 * doc/contrib.texi: Add entry for Steven Pemberton's work on
3493 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
3495 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
3496 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
3497 handling for early expansion of vector shifts (sl,sr,sra,rl).
3498 (builtin_function_type): Add vector shift right instructions
3499 to the unsigned argument list.
3501 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3503 rtl-optimizatoin/79286
3504 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
3505 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
3506 trap. PIC register plus a const unspec without offset can never trap.
3508 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
3510 * tree.h (builtin_structptr_type): New type.
3511 (builtin_structptr_types): Declare new array.
3512 * tree.c (builtin_structptr_types): New array.
3513 (free_lang_data, build_common_tree_nodes): Use it.
3515 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
3518 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
3521 2017-06-22 Matt Turner <mattst88@gmail.com>
3523 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
3524 Lake models to skylake case. Assume skylake for unknown
3525 models with clflushopt.
3527 2017-06-22 Jeff Law <law@redhat.com>
3529 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
3530 frame sizes that do not satisfy aarch64_uimm12_shift.
3532 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
3534 * profile-count.h (apply_probability,
3535 apply_scale, probability_in): Fix checks for zero.
3537 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3539 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
3540 * doc/cppdiropts.texi (-I @var{dir}): Document it.
3542 2016-06-22 Richard Biener <rguenther@suse.de>
3544 * tree-vect-loop.c (vect_model_reduction_cost): Handle
3545 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
3546 REDUC_MAX_EXPR support.
3547 (vectorizable_reduction): Likewise.
3548 (vect_create_epilog_for_reduction): Likewise.
3550 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
3552 * match.pd (A / (1 << B) -> A >> B): New.
3553 * generic-match-head.c: Include optabs-tree.h.
3554 * gimple-match-head.c: Likewise.
3555 * optabs-tree.h (target_supports_op_p): New.
3556 * optabs-tree.c (target_supports_op_p): New.
3558 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3560 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
3561 $gcc_cv_ld --help output.
3562 (gcc_cv_ld_demangle): Likewise.
3563 (gcc_cv_ld_eh_frame_hdr): Likewise.
3564 (gcc_cv_ld_pie): Likewise.
3565 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
3566 (gcc_cv_ld_buildid): Likewise.
3567 (gcc_cv_ld_sysroot): Likewise.
3568 (ld_bndplt_support): Likewise.
3569 (ld_pushpopstate_support): Likewise.
3570 * configure: Regenerate.
3571 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
3573 2017-06-21 Jakub Jelinek <jakub@redhat.com>
3576 * config/i386/sse.md (round<mode>2): Renumber match_dup and
3577 operands indexes to avoid gap between operands and match_dups.
3579 2017-06-21 Andrew Pinski <apinski@cavium.com>
3581 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
3582 Increment Arith_shift and Arith_shift_reg by 1.
3583 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
3585 * config/aarch64/aarch64.c (thunderx_tunings): Enable
3586 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
3587 (aarch64_strip_extend): Add new argument and test for it.
3588 (aarch64_cheap_mult_shift_p): New function.
3589 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
3590 add a cost if it is true.
3591 Update calls to aarch64_strip_extend.
3592 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
3594 2017-06-21 Andrew Pinski <apinski@cavium.com>
3596 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
3598 (thunderxt88): Likewise.
3599 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
3600 (thunderx_prefetch_tune): New variable.
3601 (thunderx2t99_prefetch_tune): Update for the correct values.
3602 (thunderxt88_tunings): New variable.
3603 (thunderx_tunings): Use thunderx_prefetch_tune instead of
3604 generic_prefetch_tune.
3605 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
3607 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3609 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
3610 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
3611 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
3612 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
3613 (aarch64_atomic_cas<mode>, GPI): Likewise.
3615 2017-06-21 Martin Liska <mliska@suse.cz>
3617 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
3618 statements on cold and hot labels.
3619 * predict.c (tree_estimate_probability_bb): Remove the
3620 prediction from this place.
3622 2017-06-21 Martin Liska <mliska@suse.cz>
3624 PR tree-optimization/79489
3625 * gimplify.c (maybe_add_early_return_predict_stmt): New
3627 (gimplify_return_expr): Call the function.
3628 * predict.c (tree_estimate_probability_bb): Remove handling
3630 * predict.def: Update comment about early return predictor.
3631 * gimple-predict.h (is_gimple_predict): New function.
3632 * predict.def: Change default value of early return to 66.
3633 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
3635 * passes.def: Put pass_strip_predict_hints to the beginning of
3638 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
3640 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
3641 FUNCTION_DECL declarations.
3642 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
3644 (dwaf2out_decl): Likewise.
3645 * godump.c (go_early_global_decl): Skip call to the real debug hook
3646 for FUNCTION_DECL declarations.
3647 * passes.c (rest_of_decl_compilation): Skip call to the
3648 early_global_decl debug hook for FUNCTION_DECL declarations, unless
3649 -fdump-go-spec is passed.
3651 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
3653 * config/i386/i386.c (struct builtin_isa): New field pure_p.
3654 Reorder for compactness.
3655 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
3656 (def_builtin_pure, def_builtin_pure2): New functions.
3657 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
3659 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
3661 * match.pd (nop_convert): New predicate.
3662 ((A +- CST1) +- CST2): Allow some NOP conversions.
3664 2017-06-21 Jakub Jelinek <jakub@redhat.com>
3667 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
3668 with ctors/dtors if GOVD_SHARED is set.
3670 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
3672 * config/aarch64/aarch64.md (movti_aarch64):
3673 Emit mov rather than orr.
3674 (movtf_aarch64): Likewise.
3675 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
3676 Emit mov rather than orr.
3678 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
3680 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
3681 Swap alternatives, make integer dup more expensive.
3683 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
3685 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3686 Return true for non-tls symbols.
3688 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
3690 * config/aarch64/aarch64-cores.def (cortex-a55): New.
3691 (cortex-a75): Likewise.
3692 (cortex-a75.cortex-a55): Likewise.
3693 * config/aarch64/aarch64-tune.md: Regenerate.
3694 * doc/invoke.texi (-mtune): Document new values for -mtune.
3696 2017-06-21 Tom de Vries <tom@codesourcery.com>
3698 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
3700 (Effective-Target Keywords, Other attributes): Suggest using
3701 dg-add-options stack_size feature to get stack limit in stack_size
3702 effective target documentation.
3704 2017-06-21 Julian Brown <julian@codesourcery.com>
3705 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3707 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
3708 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
3709 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
3711 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
3712 attribute type list for neon_multiply.
3713 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
3714 attribute type list for neon_multiply.
3715 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
3716 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
3717 attribute type list for neon_multiply.
3718 * config/arm/types.md (crypto_pmull): Add.
3719 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
3720 attribute type list.
3722 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
3724 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
3727 2017-06-20 Jakub Jelinek <jakub@redhat.com>
3729 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
3730 to make sure not to dereference a NULL cost_classes_ptr pointer.
3732 2017-06-20 Carl Love <cel@us.ibm.com>
3734 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3735 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3736 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3737 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3738 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3739 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
3740 VMULOSW): New enum "unspec" values.
3741 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
3742 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
3743 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
3744 altivec_vmulosw): New patterns.
3745 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3746 VMULOSW): Add definitions.
3748 2017-06-20 Julia Koval <julia.koval@intel.com>
3750 * config/i386/i386.c: Fix rounding expand for new pattern.
3751 * config/i386/subst.md: Fix pattern (parallel -> unspec).
3753 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
3755 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
3756 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
3758 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
3760 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
3763 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
3765 * config/aarch64/aarch64-cores.def: Rearrange to sort by
3766 architecture, then by implementer ID.
3767 * config/aarch64/aarch64-tune.md: Regenerate.
3769 2017-06-20 Richard Biener <rguenther@suse.de>
3772 * fold-const.c (split_tree): Fold to type before negating.
3774 2017-06-20 David Malcolm <dmalcolm@redhat.com>
3776 * diagnostic-show-locus.c
3777 (selftest::test_fixit_deletion_affecting_newline): New function.
3778 (selftest::diagnostic_show_locus_c_tests): Call it.
3780 2017-06-20 Andreas Schwab <schwab@suse.de>
3783 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
3786 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
3788 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
3789 __ARM_FEATURE_COPROC according to support.
3791 2017-06-20 Jakub Jelinek <jakub@redhat.com>
3793 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
3794 Rewritten to avoid overflow for > 32-bit pointers.
3797 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
3798 by removing enum keyword.
3799 (ubsan_type_descriptor): Likewise. Formatting fix.
3802 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
3803 splitter): Require TARGET_SSE2 in the condition.
3805 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
3808 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
3809 for doing vector set of SFmode on ISA 3.0.
3810 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
3811 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
3813 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
3814 SFmode value into a V4SF variable that was extracted from another
3815 V4SF variable without converting the element to double precision
3816 and back to single precision vector format.
3817 (vsx_insert_extract_v4sf_p9_2): Likewise.
3819 2017-06-19 Jakub Jelinek <jakub@redhat.com>
3821 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
3822 in UWHI to avoid undefined overflow.
3825 * ubsan.h (enum ubsan_encode_value_phase): New.
3826 (ubsan_encode_value): Change second argument to
3827 enum ubsan_encode_value_phase with default value of
3828 UBSAN_ENCODE_VALUE_GENERIC.
3829 * ubsan.c (ubsan_encode_value): Change second argument to
3830 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
3831 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
3832 create_tmp_var_raw instead of create_tmp_var and use a
3834 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
3835 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
3836 ubsan_encode_value callers.
3839 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
3840 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
3841 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
3843 2017-06-19 Richard Biener <rguenther@suse.de>
3846 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
3847 estimates if we changed anything.
3849 2017-06-19 Richard Biener <rguenther@suse.de>
3851 PR tree-optimization/80887
3852 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
3853 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
3854 simplified lookups, then reset mprts_hook.
3855 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
3857 (try_to_simplify): Likewise.
3859 2017-06-19 Martin Liska <mliska@suse.cz>
3862 * gimplify.c (gimplify_switch_expr):
3863 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
3865 2017-06-19 Martin Liska <mliska@suse.cz>
3867 * doc/install.texi: Document that PGO runs in 4 stages.
3869 2017-06-19 Martin Liska <mliska@suse.cz>
3872 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
3873 to dispatcher function name.
3874 * multiple_target.c (replace_function_decl): New function.
3875 (create_dispatcher_calls): Redirect both edges and references.
3877 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
3879 * profile-count.c (profile_count::dump): Dump quality.
3880 (profile_count::differs_from_p): Update for unsigned val.
3881 * profile-count.h (profile_count_quality): New enum.
3882 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
3884 2017-06-19 Richard Biener <rguenther@suse.de>
3886 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
3887 struct function as arg.
3888 (estimate_numbers_of_iterations): Export overload with loop arg.
3889 (free_numbers_of_iterations_estimates_loop): Use an overload of
3890 free_numbers_of_iterations_estimates instead.
3891 * tree-cfg.c (remove_bb): Adjust.
3892 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
3893 * tree-parloops.c (gen_parallel_loop): Likewise.
3894 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
3896 (tree_unroll_loops_completely): Likewise.
3897 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
3898 Use an overload instead and export.
3899 (estimated_loop_iterations): Adjust.
3900 (max_loop_iterations): Likewise.
3901 (likely_max_loop_iterations): Likewise.
3902 (estimate_numbers_of_iterations): Take struct function as arg
3904 (loop_exits_before_overflow): Adjust.
3905 (free_numbers_of_iterations_estimates_loop): Use an overload.
3906 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
3907 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
3909 2017-06-19 Richard Biener <rguenther@suse.de>
3912 * ipa-prop.c (find_constructor_constant_at_offset): Handle
3913 RANGE_EXPR conservatively.
3915 2017-06-16 Carl Love <cel@us.ibm.com>
3917 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3918 definitions for vec_float, vec_float2, vec_floato,
3919 vec_floate built-ins.
3920 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
3921 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
3923 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
3924 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
3925 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
3926 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
3927 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
3928 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
3929 vec_floato): Add builtin defines.
3930 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
3931 Update the built-in documentation file for the new built-in
3934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3936 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
3937 (mthumb): Mark as the negative of -marm.
3939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3941 * doc/invoke.texi (ARM Options, -mcpu): Document supported
3943 (ARM Options, -mtune): Document that this accepts the same
3944 extension options as -mcpu.
3945 (ARM Options, -mfpu): Document addition of -mfpu=auto.
3947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3949 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
3950 permitted extensions.
3952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3954 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
3955 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
3956 (armv8-m.main): Add option +nodsp.
3957 * config/arm/arm-cpu-cdata.h: Regenerated.
3959 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3961 * config/arm/t-fuchsia: New file.
3962 * config.gcc (arm*-*-fuchsia*): Use it.
3964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3966 * config/arm/t-symbian: Rewrite for new option infrastructure.
3968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3970 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
3971 (MULTILIB_REQUIRED): Likewise.
3973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3975 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
3976 (MULTILIB_RESUE): Likewise.
3977 (MULTILIB_MATCHES): Likewise.
3978 (MULTLIB_REQUIRED): Likewise.
3980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3982 * config/arm/t-rtems: Rewrite for new option framework.
3984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3986 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
3987 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
3988 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
3989 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
3990 * config/arm/t-multilib: ... here.
3991 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
3992 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
3993 armv7-a and armv8*-a when A-profile libraries have not been built.
3994 * config/arm/t-rmprofile: Rewrite.
3996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3998 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
3999 with a backslash. Remove the backslash after substituting unescaped
4001 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4005 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4006 the driver to pass through the default setting of -mfloat-abi.
4007 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4009 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4010 (all_feat_combs): New rule.
4011 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4013 * config/arm/t-aprofile: Rewrite.
4015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4017 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4018 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4019 fpu is not specified by the user/command-line.
4020 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4021 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4022 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4023 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4024 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4025 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4029 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4030 * config/arm/t-arm-elf: Rewritten.
4032 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4034 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4035 have some floating-point instructions.
4036 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4037 (TARGET_MAYBE_HARD_FLOAT): New macro.
4038 * config/arm/arm-builtins.c (arm_init_builtins): Use
4039 TARGET_MAYBE_HARD_FLOAT.
4040 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4044 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4045 (configargs.h): Include it.
4046 (arm_print_hint_for_fpu_option): New function.
4047 (arm_parse_fpu_option): New function.
4048 (candidate_extension): New class.
4049 (arm_canon_for_multilib): New function.
4050 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4051 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4052 (ARCH_CANONICAL_SPECS): New macro.
4053 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4057 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4058 are set after handling multilib fragments. Set target_cpu_default2
4061 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4063 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4065 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4066 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4068 (check_arch): Likewise.
4069 * config/arm/arm.c (arm_configure_build_target): Handle
4070 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4071 options in the default.
4073 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4075 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4076 when an option is an alias of another.
4077 * config/arm/parsecpu.awk (optalias): New parser token.
4078 (gen_comm_data): Mark non-alias options as such. Emit entries
4079 for extension aliases.
4080 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4081 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4082 (armv6kz, armv6zk, armv6t2): Likewise.
4083 (armv7): Make vfpv3-d16 an alias.
4084 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4086 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4087 Sort in canonical order.
4088 (armv8-a): Sort in canonical order.
4089 (armv8.1-a, armv8.2-a): Likewise.
4090 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4092 (cortex-a9): Sort in canonical order.
4093 * config/arm/arm.c (selftests.h): Include it.
4094 (arm_test_cpu_arch_data): New function.
4095 (arm_run_self_tests): New function.
4096 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4097 (targetm): Move declaration to the end of the file.
4098 * arm-cpu-cdata.h: Regenerated.
4100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4102 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4103 call to target_mode_check describing the type of option passed.
4104 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4105 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4106 arm_parse_cpu_option_name to match parameters against list of
4108 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4109 arm_arch_core_flags data structure.
4110 * config/arm/arm-cpu_cdata.h: Regenerated.
4112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4114 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4116 (arm_print_hint_for_cpu_option): Likewise.
4117 (arm_print_hint_for_arch_option): Likewise.
4118 (arm_parse_cpu_option_name): Likewise.
4119 (arm_parse_arch_option_name): Likewise.
4120 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4121 of entries in the all_fpus list.
4122 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4123 (arm_parse_cpu_option_name): Declare.
4124 (arm_parse_arch_option_name): Declare.
4125 (arm_parse_option_features): Declare.
4126 (arm_intialize_isa): Declare.
4127 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4129 (gen_comm_data): ... here. Make definitions non-static.
4130 * config/arm/arm-cpu-data.h: Regenerated.
4131 * config/arm/arm-cpu-cdata.h: Regenerated.
4133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4135 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4136 (cpu_arch_extension): New structure.
4137 (cpu_arch_option, arch_option, cpu_option): New structures.
4138 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4140 (gen_data): Generate new format data tables.
4141 * config/arm/arm.c (cpu_tune): New structure.
4142 (cpu_option, processors): Delete.
4143 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4144 (arm_print_hint_for_cpu_option): ... this and ...
4145 (arm_print_hint_for_arch_option): ... this.
4146 (arm_parse_arch_cpu_name): Delete. Replace with ...
4147 (arm_parse_cpu_option_name): ... this and ...
4148 (arm_parse_arch_option_name): ... this.
4149 (arm_unrecognized_feature): Change type of target parameter to
4151 (arm_parse_arch_cpu_features): Delete. Replace with ...
4152 (arm_parse_option_features): ... this.
4153 (arm_configure_build_target): Rework to use new configuration data
4155 (arm_print_tune_info): Rework for new configuration data tables.
4156 * config/arm/arm-cpu-data.h: Regenerated.
4157 * config/arm/arm-cpu.h: Regenerated.
4159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4161 * Makefile.in (OBJS): Move sbitmap.o from here ...
4162 (OBJS-libcommon): ... to here.
4164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4166 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4167 (ISA_ALL_CRYPTO): New macro.
4168 (ISA_ALL_SIMD): New macro
4169 (ISA_ALL_FP): New macro.
4170 * config/arm/arm.c (fpu_bitlist): Update initializer.
4171 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4173 (arm9e): Add fpu. Add option for nofp
4174 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4175 (arm926ej-s, arm1026ej-s): Likewise.
4176 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4177 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4178 neon-fp16, neon-vfpv4, nofp and nosimd.
4179 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4180 (cortex-a8): Add fpu. Add option for nofp.
4181 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4182 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4183 (cortex-r4f): Add fpu.
4184 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4185 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4187 (cortex-r8): Likewise.
4188 (cortex-m4): Add fpu. Add option for nofp.
4189 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4190 (cortex-a17.cortex-a7): Likewise.
4191 (cortex-a32): Add fpu. Add options for crypto and nofp.
4192 (cortex-a35, cortex-a53): Likewise.
4193 (cortex-a57): Add fpu. Add option for crypto.
4194 (cortex-a72, cortex-a73): Likewise.
4195 (exynos-m1): Likewise.
4196 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4197 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4198 (cortex-m33): Add fpu. Add option for nofp.
4199 * config/arm/arm-cpu-cdata.h: Regenerated
4200 * config/arm/arm-cpu-data.h: Regenerated.
4202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4204 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4205 (armv5te, armv5tej): Likewise.
4206 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4207 (armv7): Add options fp and vfpv3-d16.
4208 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4209 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4211 (armv7ve): Likewise.
4212 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4213 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4214 (armv8-a): Add nocrypto option.
4215 (armv8.1-a, armv8.2-a): Likewise.
4216 (armv8-m.main): add options fp, fp.dp and nofp.
4218 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4220 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4222 (armv8-a+crc): Delete.
4223 (armv8.1-a): Add options simd, crypto and nofp.
4224 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4225 (armv8.2-a+fp16): Delete.
4226 (armv8-m.main): Add option dsp.
4227 (armv8-m.main+dsp): Delete.
4228 (cortex-a8): Add fpu. Add nofp option.
4229 (cortex-a9): Add fpu. Add nofp and nosimd options.
4230 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4231 link to main cpu and architecture data structures.
4232 (gen_comm_data): Only put isa attributes from the main architecture
4234 (option): New statement for architecture and CPU entries.
4235 * arm.c (struct cpu_option): New structure.
4236 (struct processors): Add entry for options.
4237 (arm_unrecognized_feature): New function.
4238 (arm_parse_arch_cpu_name): Ignore any characters after the first
4240 (arm_parse_arch_cpu_feature): New function.
4241 (arm_configure_build_target): Separate out any CPU and architecture
4242 features and parse separately. Don't error out if -mfpu=auto is
4243 used with only an architecture string.
4244 (arm_print_asm_arch_directives): New function.
4245 (arm_file_start): Call it.
4246 * config/arm/arm-cpu-cdata.h: Regenerated.
4247 * config/arm/arm-cpu-data.h: Likewise.
4248 * config/arm/arm-tables.opt: Likewise.
4250 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4252 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4253 assembler when it is not -mfpu=auto.
4255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4257 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4258 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4259 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4260 (ASM_CPU_SPEC): Rewrite.
4261 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4262 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4263 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4264 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4265 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4266 copied string is NUL-terminated. Also strip any characters prefixed
4268 (arm_rewrite_selected_arch): New function.
4269 (arm_rewrite_march): New function.
4271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4273 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4274 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4275 (march, mcpu, mtune): Convert to string-based options.
4276 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4277 (arm_parse_arch_cpu_name): New function.
4278 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4279 identify selected architecture or CPU.
4280 (arm_option_save): New function.
4281 (TARGET_OPTION_SAVE): Redefine.
4282 (arm_option_restore): Restore string options.
4283 (arm_option_print): Print string options.
4285 2017-06-16 Martin Sebor <msebor@redhat.com>
4287 PR tree-optimization/80933
4288 PR tree-optimization/80934
4289 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4290 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4291 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4292 (gimple_fold_builtin): Call them.
4294 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4296 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4297 as unlikely; update profile.
4299 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4301 * predict.c (force_edge_cold): Handle declaring edges impossible
4304 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4306 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4308 (try_unroll_loop_completely): Fix reporting.
4310 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4312 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4314 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4317 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4318 if given a non-constant argument for an intrinsic which requires a
4321 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4323 * profile.c (compare_freqs): New function.
4324 (branch_prob): Sort edge list.
4325 (find_spanning_tree): Assume that the list is priority sorted.
4327 2017-06-16 Richard Biener <rguenther@suse.de>
4329 PR tree-optimization/81090
4330 * passes.def (pass_record_bounds): Remove.
4331 * tree-pass.h (make_pass_record_bounds): Likewise.
4332 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4333 make_pass_record_bounds): Likewise.
4334 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4335 not free niter estimates at the beginning but at the end.
4336 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4338 2017-06-16 Richard Biener <rguenther@suse.de>
4340 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4341 initializer to workaround ICE in host GCC 4.8.
4343 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4345 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4347 (clone_inlined_nodes): Update.
4349 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4351 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4352 prefetch settings, and enable prefetching by default at -O3.
4354 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4356 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4357 Set flag_prefetch_loop_arrays according to tuning data.
4359 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4361 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4363 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4364 [Unrelated to main purpose of the patch] Place the pointer field last
4365 to enable type checking errors when tune structure are wrongly merged.
4366 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4367 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4368 (thunderx2t99_prefetch_tune): New tune constants.
4369 (tune_params *_tunings): Update all tunings (no functional change).
4370 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4371 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4372 from tunings structures.
4374 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4377 * ubsan.c (instrument_null): Add T argument, use it instead
4378 of computing it based on IS_LHS.
4379 (instrument_object_size): Likewise.
4380 (pass_ubsan::execute): Adjust instrument_null and
4381 instrument_object_size callers to pass gimple_get_lhs or
4382 gimple_assign_rhs1 result to it. Use instrument_null instead of
4383 calling get_base_address and instrument_mem_ref. Handle
4384 aggregate call arguments for object-size sanitization.
4386 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4388 PR tree-optimization/81089
4389 * tree-vrp.c (is_masked_range_test): Validate operands of
4392 2017-06-15 Martin Sebor <msebor@redhat.com>
4395 * dumpfile.c (dump_register): Avoid calling memset to initialize
4396 a class with a default ctor.
4397 * gcc.c (struct compiler): Remove const qualification.
4398 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4399 * hash-table.h: Ditto.
4400 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4402 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4403 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4405 * params.h (struct param_info): Make struct members non-const.
4406 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4407 with default initialization.
4408 * vec.h (vec_copy_construct, vec_default_construct): New helper
4410 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4411 with vec_copy_construct.
4412 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
4413 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
4414 * doc/invoke.texi (-Wclass-memaccess): Document.
4416 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4418 * emit-rtl.h (is_leaf): Update comment about local
4421 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4424 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
4425 for a variable to have a section before checking if the section has a
4427 Set section to.persistent if persistent attribute is set.
4428 Warn if .persistent attribute is used on an automatic variable.
4430 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
4432 PR rtl-optimization/80474
4433 * reorg.c (update_block): Do not ignore instructions in a delay slot.
4435 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
4437 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
4440 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
4442 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
4443 (casesi): Emit bounds checking as RTL.
4444 (casesi_internal_mips16_<mode>): Remove bounds checking.
4446 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
4448 * config/xtensa/xtensa.c (xtensa_option_override): Append
4449 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
4450 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
4451 xtensa_doloop_hooks): Define unconditionally.
4452 (xtensa_reorg_loops): Only call reorg_loops in the presence of
4454 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
4455 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
4456 for it in xtensa_option_override.
4457 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
4458 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
4460 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
4462 * doc/cppopts.texi: Document '-' special value to -MF.
4464 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
4466 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
4467 (cortex_a53_fconst): Likewise.
4468 (cortex_a53_fpmul): Likewise.
4469 (cortex_a53_f_load_64): Likewise.
4470 (cortex_a53_f_load_many): Likewise.
4471 (cortex_a53_advsimd_alu): Likewise.
4472 (cortex_a53_advsimd_alu_q): Likewise.
4473 (cortex_a53_advsimd_mul): Likewise.
4474 (cortex_a53_advsimd_mul_q): Likewise.
4475 (fpmac bypass): Add new bypass for fpmac-fpmac case.
4476 Add missing fmul, r2f_cvt and fconst cases.
4478 2017-06-14 Richard Biener <rguenther@suse.de>
4481 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
4483 (fold_binary_loc): When associating do not treat pre-existing
4484 TREE_OVERFLOW on literal constants as a reason to allow
4485 TREE_OVERFLOW on associated literal constants.
4487 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4489 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
4490 (MASK_FEATURES): New macro.
4491 * config/sparc/sparc.c (sparc_option_override): Remove the special
4492 handling of -mfpu and generalize it to all MASK_FEATURES switches.
4494 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4496 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
4497 a division of 0 if non-call exceptions are enabled.
4499 2017-06-14 Andrew Pinski <apinski@cavium.com>
4500 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4503 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
4504 Improve vector initialization code gen for only variable case.
4506 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4508 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
4510 2017-06-14 Richard Biener <rguenther@suse.de>
4512 PR tree-optimization/81083
4513 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
4516 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4518 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
4519 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
4520 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
4521 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
4522 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
4523 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
4525 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4527 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
4528 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
4530 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4532 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
4534 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4536 * config/rs6000/t-rtems: Don't handle SPE.
4538 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4540 * config/rs6000/t-linux: Don't handle SPE.
4542 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4544 * config/rs6000/eabispe.h: Delete file.
4546 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4548 * config/rs6000/t-spe: Delete file.
4550 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4552 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
4553 (rs6000_legitimate_offset_address_p): Return false for anything in
4554 V2SImode or V2SFmode.
4556 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4558 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
4559 except V2SF and V2SI. Rearrange the vector modes, and add comments.
4560 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
4562 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
4563 (rs6000_legitimate_offset_address_p): Ditto.
4564 (rs6000_emit_move): Ditto.
4565 (rs6000_init_builtins): Remove V4HI_type_node.
4567 2017-06-13 Martin Liska <mliska@suse.cz>
4570 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
4571 (gate_asan): Likewise.
4572 * asan.h (asan_no_sanitize_address_p): Remove the function.
4573 (sanitize_flags_p): New function.
4574 * builtins.def: Fix coding style.
4575 * common.opt: Use renamed enum value.
4576 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
4577 * doc/extend.texi: Document no_sanitize attribute.
4578 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
4579 to SANITIZE_UNDEFINED_NONDEFAULT.
4580 * gcc.c (sanitize_spec_function): Use the renamed enum value.
4581 * gimple-fold.c (optimize_atomic_compare_exchange_p):
4582 Use sanitize_flags_p.
4583 * gimplify.c (gimplify_function_tree): Likewise.
4584 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
4585 * opts.c (parse_no_sanitize_attribute): New function.
4586 (common_handle_option): Use renamed enum value.
4587 * opts.h (parse_no_sanitize_attribute): Declare.
4588 * tree.c (sanitize_flags_p): New function.
4589 * tree.h: Declared here.
4590 * tsan.c: Use sanitize_flags_p.
4591 * ubsan.c (ubsan_expand_null_ifn): Likewise.
4592 (instrument_mem_ref): Likewise.
4593 (instrument_bool_enum_load): Likewise.
4594 (do_ubsan_in_current_function): Remove the function.
4595 (pass_ubsan::execute): Use sanitize_flags_p.
4596 * ubsan.h: Remove do_ubsan_in_current_function
4597 * tree-cfg.c (print_no_sanitize_attr_value): New function.
4598 (dump_function_to_file): Use it here.
4600 2017-06-13 Martin Jambor <mjambor@suse.cz>
4602 PR tree-optimization/80803
4603 PR tree-optimization/81063
4604 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
4605 (propagate_subaccesses_across_link): Enqueue subtree whenever
4606 necessary instead of relying on the caller.
4608 2017-06-13 Martin Jambor <mjambor@suse.cz>
4610 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
4611 that have a first_link.
4612 (sort_and_splice_var_accesses): Do not check first_link before
4614 (subtree_mark_written_and_enqueue): Likewise.
4615 (propagate_all_subaccesses): Likewise and do not stop at first
4616 parent with a first_link.
4618 2017-06-13 Martin Jambor <mjambor@suse.cz>
4620 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
4623 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4625 * match.pd: New pattern.
4627 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4629 * tree-vrp.c (is_masked_range_test): New function.
4630 (register_edge_assert_for): Determine ranges for
4633 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4635 PR tree-optimization/67328
4636 * fold-const.c (maskable_range_p): New function.
4637 (build_range_check): Generate bittests if possible.
4639 2017-06-13 Martin Liska <mliska@suse.cz>
4641 * gimple-pretty-print.c (dump_probability): Add new argument.
4642 (dump_edge_probability): Dump both probability and count.
4643 (dump_gimple_label): Likewise.
4644 (dump_gimple_bb_header): Likewise.
4646 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
4649 * config/avr/avr-devices.c: Fix indentation.
4650 * config/avr/gen-avr-mmcu-specs.c: Dito.
4652 2017-06-13 Richard Biener <rguenther@suse.de>
4654 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
4655 instead get vector type from stmt_info.
4656 (vectorizable_reduction): Adjust. Remove dead code.
4658 2017-06-13 Richard Biener <rguenther@suse.de>
4661 * fold-const.c (extract_muldiv_1): Remove bogus distribution
4662 case of C * (x * C2 + C3).
4663 (fold_addr_of_array_ref_difference): Properly fold index difference.
4665 2017-06-12 David S. Miller <davem@davemloft.net>
4668 * config/sparc/sparc.md (return expander): Emit frame blockage if
4669 function uses alloca.
4671 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
4673 * combine.c (make_field_assignment): Check len rather than the mode
4674 precision when calling force_to_mode.
4676 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
4678 Support multilibs and devices that see flash in RAM address range.
4681 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
4682 (avr_mcu_t) <flash_pm_offset>: New field.
4683 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
4684 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
4685 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
4686 (AVR_TINY_PM_OFFSET): Remove macro.
4687 * config/avr/avr.opt (-mshort-calls): New option.
4688 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
4689 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
4690 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
4691 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
4692 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
4693 instead of avr_arch->have_jmp_call.
4694 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
4695 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
4696 avr_arch->flash_pm_offset to define.
4697 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
4698 new field flash_pm_offset. Add entry for avrxmega3.
4699 (avr_texinfo): Add entry for avrxmega3.
4700 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
4701 attiny212, attiny214,
4702 attiny412, attiny414, attiny416, attiny417,
4703 attiny814, attiny816, attiny817,
4704 attiny1614, attiny1616, attiny1617,
4705 attiny3214, attiny3216, attiny3217.
4706 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
4707 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
4708 (avr_print_operand_address) [AVR_TINY]: Same.
4709 (avr_asm_init_sections) <readonly_data_section>: Only patch
4710 callback if avr_arch->flash_pm_offset = 0.
4711 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
4712 for rodata if avr_arch->flash_pm_offset != 0.
4713 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
4714 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
4715 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
4716 (m_options): Append opt_rcall.
4717 (m_dirnames): Append dir_rcall.
4718 * config/avr/t-multilib: Regenerate.
4720 * configure.ac [target=avr]: Check whether avrxmega3 default
4721 linker description file works as needed.
4722 * configure: Regenerate.
4723 * doc/avr-mmcu.texi: Regenerate.
4724 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
4725 <__AVR_ARCH__>: Document avrxmega3 and 103.
4726 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
4727 <__AVR_SHORT_CALLS__>: Document it.
4728 <__AVR_PM_BASE_ADDRESS__>: Document it.
4729 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
4730 (AVR Variable Attributes) <progmem>: Document this is
4731 not needed for avrxmega3.
4732 (AVR Named Address Spaces) <__flash>: Dito.
4734 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
4736 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
4738 2017-06-12 Doug Rupp <rupp@adacore.com>
4740 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
4741 Append vxworks-stdint.h to the tm_file list.
4742 * config/vxworks-stdint.h: New file.
4744 2017-06-12 Martin Liska <mliska@suse.cz>
4746 PR tree-optimization/81041
4747 * tree-profile.c (gimple_gen_ic_func_profiler):
4748 Create an extra BB in profile-generate
4749 (gimple_gen_time_profiler): Likewise.
4751 2017-06-12 Jakub Jelinek <jakub@redhat.com>
4753 PR tree-optimization/81003
4754 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
4755 (update_range_test): Use it instead of force_gimple_operand_gsi.
4757 2017-06-12 Richard Biener <rguenther@suse.de>
4759 PR tree-optimization/81053
4760 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
4761 with backedge value not defined in loop. Simplify def stmt
4764 2017-06-11 Tom de Vries <tom@codesourcery.com>
4767 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
4769 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
4770 nvptx_cannot_force_const_mem.
4772 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
4774 * opts.c (finish_options): Move test for flag_split_stack after
4775 it has been initialized.
4777 2017-06-11 Jason Merrill <jason@redhat.com>
4779 * tree.h (id_equal): New.
4780 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
4781 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
4782 instead of strcmp of IDENTIFIER_POINTER.
4784 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
4786 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
4787 (mark_all_inlined_calls_cdtor): Fix formating.
4788 (inline_transform): Rescale profile before inlining.
4790 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
4792 * cgraph.h (cgraph_edge::clone): Update prototype.
4793 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
4794 (cgraph_node::create_clone): Update.
4795 (cgraph_node::create_version_clone): Update.
4796 * tree-inline.c (copy_bb): Update.
4797 (expand_call_inline): Update.
4799 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
4801 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
4802 factored out from ...
4803 (rs6000_emit_prologue): ... here.
4805 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
4807 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
4808 factored out from ...
4809 (rs6000_emit_prologue): ... here.
4811 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
4813 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
4815 (handle_missing_profiles): Fix computation of tp_first_run.
4816 (counts_to_freqs): Do not touch freqs when count is 0.
4818 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
4820 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
4823 2017-06-10 Tom de Vries <tom@codesourcery.com>
4825 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
4826 attributes): Document signal effective target.
4828 2017-06-10 Tom de Vries <tom@codesourcery.com>
4830 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
4831 Document effective target stack_size.
4833 2017-06-09 David Malcolm <dmalcolm@redhat.com>
4835 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
4836 to the edit_context if they can be auto-applied.
4838 2017-06-9 Ian Lance Taylor <iant@golang.org>
4840 * opts.c (finish_options): If -fsplit-stack, disable implicit
4841 -forder-blocks-and-partition.
4842 * doc/invoke.texi (Optimize Options): Document that when using
4843 -fsplit-stack -forder-blocks-and-partition is not implicitly
4846 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
4848 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
4849 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
4850 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
4851 * builtins.def (abort, trap, unreachable): Declare cold.
4852 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
4853 * tree-core.h (ECF_COLD): New.
4854 * tree.c (set_call_expr_flags): Handle ECF_COLD.
4855 (build_common_builtin_nodes): Mark unreachable and abort as cold.
4857 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
4859 * predict.c (unlikely_executed_stmt_p): Cleanup.
4861 2017-06-09 Richard Biener <rguenther@suse.de>
4863 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
4864 model if the ref is always written to.
4866 2017-06-09 Tamar Christina <tamar.christina@arm.com>
4868 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
4870 2017-06-09 Tamar Christina <tamar.christina@arm.com>
4872 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
4875 2017-06-09 Tom de Vries <tom@codesourcery.com>
4878 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
4879 "target cannot support label values" when encountering LABEL_REF.
4881 2017-06-09 Martin Liska <mliska@suse.cz>
4883 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
4884 (gimple_gen_ic_func_profiler): Emit direct comparison
4885 of __gcov_indirect_call_callee with NULL.
4886 (gimple_gen_time_profiler): Change probability from
4887 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
4889 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
4891 * profile.c (edge_gcov_counts): Turn to pointer.
4892 (compute_branch_probabilities, compute_branch_probabilities): Update.
4893 (branch_prob): Do not clear edge_gcov_count.
4894 * profile.h (edge_gcov_counts): Turn to pointer.
4895 (edge_gcov_count): Update.
4897 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
4899 * gimple.h (gimple_check_failed): Mark cold.
4901 2017-06-09 Richard Biener <rguenther@suse.de>
4903 PR tree-optimization/66623
4904 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
4905 refactor check_reduction into two parts, properly computing
4906 whether we have to check reduction validity for outer loop
4909 2017-06-09 Richard Biener <rguenther@suse.de>
4911 PR tree-optimization/79483
4912 * graphite-scop-detection.c (order): New global.
4913 (get_order): Compute bb to order mapping that satisfies code
4914 generation constraints.
4915 (cmp_pbbs): New helper.
4916 (build_scops): Start domwalk at entry block, sort generated
4919 2017-06-09 Richard Biener <rguenther@suse.de>
4922 * ipa-polymorphic-call.c
4923 (ipa_polymorphic_call_context::restrict_to_inner_class):
4924 Skip FIELD_DECLs with error_mark_node type.
4925 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
4928 2017-06-09 Martin Liska <mliska@suse.cz>
4930 * predict.c (struct branch_predictor): New struct.
4931 (test_prediction_value_range): New test.
4932 (predict_c_tests): New function.
4933 * selftest-run-tests.c (selftest::run_tests): Run the function.
4934 * selftest.h: Declare new tests.
4936 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
4939 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
4940 gen_add3_insn did not fail.
4941 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
4942 r0, construct that number in a temporary reg and add that reg to r0.
4943 If asked to put the result in r0 as well, fail.
4945 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
4947 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4948 for early expansion of vec_eqv.
4950 2017-06-08 Jakub Jelinek <jakub@redhat.com>
4953 * ubsan.c (instrument_null): Avoid pointless code temporary.
4954 (pass_ubsan::execute): Instrument aggregate arguments of calls.
4956 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
4960 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
4963 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
4966 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4968 * predict.c (unlikely_executed_edge_p): Move ahead.
4969 (probably_never_executed_edge_p): Use it.
4971 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
4974 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
4975 gimple_call_builtin_p call.
4977 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4979 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
4980 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
4981 rtl_check_failed_type2, rtl_check_failed_code1,
4982 rtl_check_failed_code2, rtl_check_failed_code_mode,
4983 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
4984 rtvec_check_failed_bounds, rtl_check_failed_flag,
4985 _fatal_insn_not_found, _fatal_insn): Likewise.
4986 * tree.h (tree_contains_struct_check_failed,
4987 tree_check_failed, tree_not_check_failed,
4988 tree_class_check_failed, tree_range_check_failed,
4989 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
4990 tree_vec_elt_check_failed, phi_node_elt_check_failed,
4991 tree_operand_check_failed, omp_clause_check_failed,
4992 omp_clause_operand_check_failed, omp_clause_range_check_failed):
4995 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4997 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
4998 flag_branch_probabilities.
4999 * ipa-inline.c (edge_badness): Likewise.
5000 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5001 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5002 * predict.c (maybe_hot_frequency_p): Likewise.
5003 (probably_never_executed): Likewise.
5004 * sched-ebb.c (schedule_ebbs): Likewise.
5005 * sched-rgn.c (find_single_block_region): Likewise.
5006 * tracer.c (tail_duplicate): Likewise.
5008 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5010 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5011 longer requires x_flag_profile_use.
5013 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5015 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5016 instead of flag_reorder_blocks_and_partition.
5017 * dbxout.c (dbxout_function_end): Likewise.
5018 * dwarf2out.c (gen_subprogram_die): Likewise.
5019 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5020 * hw-doloop.c (reorg_loops): Likewise.
5021 * varasm.c (assemble_start_function,
5022 assemble_end_function): Likewise.
5023 (decide_function_section): Do not check for
5024 flag_reorder_blocks_and_partition.
5026 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5028 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5030 (chkp_get_hard_register_fake_addr_expr): Ditto.
5031 (chkp_build_addr_expr): Add check for hard reg case.
5032 (chkp_parse_array_and_component_ref): Ditto.
5033 (chkp_find_bounds_1): Ditto.
5034 (chkp_process_stmt): Don't generate bounds store for
5037 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5039 * predict.c (maybe_hot_bb_p): Do not check profile status.
5040 (maybe_hot_edge_p): Likewise.
5041 (probably_never_executed): Check for zero counts even if profile
5043 (unlikely_executed_edge_p): New function.
5044 (unlikely_executed_stmt_p): New function.
5045 (unlikely_executed_bb_p): New function.
5046 (set_even_probabilities): Use unlikely predicates.
5047 (combine_predictions_for_bb): Likewise.
5048 (predict_paths_for_bb): Likewise.
5049 (predict_paths_leading_to_edge): Likewise.
5050 (determine_unlikely_bbs): New function.
5051 (estimate_bb_frequencies): Use it.
5052 (compute_function_frequency): Use zero counts even if profile is
5054 * profile-count.h: Fix typo.
5056 2017-08-08 Julia Koval <julia.koval@intel.com>
5058 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5059 _mm512_mask_cvtsepi16_storeu_epi8,
5060 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5061 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5062 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5063 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5064 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5065 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5066 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5067 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5068 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5069 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5070 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5071 __builtin_ia32_pmovuswb256mem_mask,
5072 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5073 __builtin_ia32_pmovwb512mem_mask): New builtins.
5075 2017-08-08 Julia Koval <julia.koval@intel.com>
5077 PR target/73350,80862
5078 * config/i386/subst.md (round): Fix round pattern.
5079 * config/i386/i386.c (ix86_erase_embedded_rounding):
5080 Fix erasing rounding for the fixed pattern.
5082 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5084 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5086 2017-06-08 Martin Liska <mliska@suse.cz>
5088 PR gcov-profile/80911
5089 * gcov.c (block_info::block_info): New constructor.
5091 2017-06-07 Carl Love <cel@us.ibm.com>
5093 * config/rs6000/rs6000-c: The return type of the following
5094 built-in functions was implemented as int not long long. Fix sign
5095 of return value for the unsigned version of vec_mulo and vec_mule.
5096 vector unsigned long long vec_bperm (vector unsigned long long,
5097 vector unsigned char)
5098 vector signed long long vec_mule (vector signed int,
5100 vector unsigned long long vec_mule (vector unsigned int,
5101 vector unsigned int)
5102 vector signed long long vec_mulo (vector signed int,
5104 vector unsigned long long vec_mulo (vector unsigned int,
5105 vector unsigned int)
5106 * doc/extend.texi: Fix the documentation for the built-in
5109 2017-06-07 Carl Love <cel@us.ibm.com>
5112 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5115 2017-06-07 Carl Love <cel@us.ibm.com>
5117 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5118 support, Generate doublehv for signed int/float for BE case only.
5120 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5122 * doc/invoke.texi (mcx16): Rewrite.
5124 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5126 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5127 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5128 *mov<mode>_softfloat, and an anonymous splitter): Use
5129 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5131 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5133 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5135 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5136 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5137 (rs6000_debug_reg_global): Adjust.
5138 (rs6000_init_hard_regno_mode_ok): Adjust.
5139 (rs6000_dbx_register_number): Adjust.
5140 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5141 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5142 Remove SPE_ACC and SPEFSCR.
5143 (REG_ALLOC_ORDER): Ditto.
5144 (FRAME_POINTER_REGNUM): Change to 111.
5145 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5146 (REG_CLASS_NAMES): Ditto.
5147 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5148 (REGISTER_NAMES): Ditto.
5149 (ADDITIONAL_REG_NAMES): Ditto.
5150 (rs6000_reg_names): Ditto.
5151 * config/rs6000/rs6000.md: Renumber some register number
5154 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5156 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5158 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5159 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5161 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5162 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5163 Delete the SPE high registers.
5164 (REG_ALLOC_ORDER): Ditto.
5165 (enum reg_class): Remove SPE_HIGH_REGS.
5166 (REG_CLASS_NAMES): Ditto.
5167 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5168 (REGISTER_NAMES): Ditto.
5169 (rs6000_reg_names): Ditto.
5170 * doc/tm.texi.in: Remove SPE as example.
5171 * doc/tm.texi: Regenerate.
5173 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5175 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5176 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5177 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5178 * config/rs6000/rs6000.md (type): Remove "brinc".
5180 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5182 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5183 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5184 * config/rs6000/linuxspe.h: Delete file.
5185 * config/rs6000/rs6000.md: Don't include spe.md.
5186 * config/rs6000/spe.h: Delete file.
5187 * config/rs6000/spe.md: Delete file.
5188 * config/rs6000/t-rs6000: Remove spe.md.
5190 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5192 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5193 (reg_or_none500mem_operand): Delete.
5194 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5195 instead of reg_or_none500mem_operand.
5197 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5199 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5200 handling of SPE flags.
5201 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5205 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5207 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5208 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5209 paired_divv2sf3): Similar.
5210 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5211 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5212 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5214 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5215 Rename the paired_* instruction patterns.
5216 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5218 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5219 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5220 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5221 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5222 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5224 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5225 (spe_func_has_64bit_regs_p): Delete.
5226 (spe_expand_predicate_builtin): Delete.
5227 (spe_expand_evsel_builtin): Delete.
5228 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5229 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5230 (invalid_e500_subreg): Delete.
5231 (rs6000_legitimize_address): Always force_reg op2 as well, for
5232 paired single memory accesses.
5233 (rs6000_member_type_forces_blk): Delete.
5234 (rs6000_spe_function_arg): Delete.
5235 (rs6000_expand_unop_builtin): Delete SPE handling.
5236 (rs6000_expand_binop_builtin): Ditto.
5237 (spe_expand_stv_builtin): Delete.
5238 (bdesc_2arg_spe): Delete.
5239 (spe_expand_builtin): Delete.
5240 (spe_expand_predicate_builtin): Delete.
5241 (spe_expand_evsel_builtin): Delete.
5242 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5243 (spe_init_builtins): Delete.
5244 (spe_func_has_64bit_regs_p): Delete.
5245 (savres_routine_name): Delete "info" parameter. Adjust callers.
5246 (rs6000_emit_stack_reset): Ditto.
5247 (rs6000_dwarf_register_span): Delete.
5248 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5249 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5250 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5252 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5254 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5255 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5256 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5257 mulv2sf3, divv2sf3): Delete expanders.
5259 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5261 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5263 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5265 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5266 * config/rs6000/rs6000.c: Ditto.
5268 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5270 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5271 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5272 comparison_operator.
5274 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5276 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5277 * config/rs6000/rs6000.opt: Ditto.
5278 * config/rs6000/t-rtems: Ditto.
5280 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5282 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5283 TARGET_E500_SINGLE by 0, simplify.
5284 * config/rs6000/rs6000.c: Ditto.
5285 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5286 (spe_build_register_parallel): Delete.
5287 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5288 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5289 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5290 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5291 (E500_CONVERT): Delete.
5292 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5294 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5296 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5297 * config/rs6000/dfp.md: Ditto.
5298 (negdd2, *negdd2_fpr): Merge.
5299 (absdd2, *absdd2_fpr): Merge.
5300 (negtd2, *negtd2_fpr): Merge.
5301 (abstd2, *abstd2_fpr): Merge.
5302 * config/rs6000/e500.h: Delete file.
5303 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5304 TARGET_FPRS by 1 and simplify.
5305 * config/rs6000/rs6000-c.c: Ditto.
5306 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5308 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5310 * config/rs6000/rs6000.md: Ditto.
5311 (floatdidf2, *floatdidf2_fpr): Merge.
5312 (move_from_CR_gt_bit): Delete.
5313 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5314 (E500_CR_IOR_COMPARE): Delete.
5315 (All patterns that require !TARGET_FPRS): Delete.
5316 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5318 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5320 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5322 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5324 * graphds.c (add_edge): Intitialize edge's attached data.
5325 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5326 pointer parameter. Call pointed function on each edge during
5327 graph traversing. Skip traversing the edge when the function
5329 (graphds_dfs, graphds_scc): Ditto.
5330 (for_each_edge): New parameter. Pass the new parameter to callback
5332 * graphds.h (skip_edge_callback): New function pointer type.
5333 (graphds_dfs, graphds_scc): New function pointer parameter.
5334 (graphds_edge_callback, for_each_edge): New parameter.
5336 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5338 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5339 out code checking if runtime alias check is possible to below ...
5340 Call the new function.
5341 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5342 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5344 2017-06-07 Marek Polacek <polacek@redhat.com>
5347 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5348 TYPE_OVERFLOW_WRAPS check.
5350 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5352 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5353 if versioning is required.
5354 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5355 peeling with the check for versioning.
5357 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5359 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5360 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5361 Set true to new parameter if new ssa variable is defined.
5362 (vect_gen_vector_loop_niters): Refactor. Set range information
5363 for the new vector loop bound variable.
5364 (vect_do_peeling): Ditto.
5366 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5368 * tree-affine.c (ssa.h): Include header file.
5369 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5370 has wrapping overflow behavior.
5372 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5374 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5376 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5378 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5379 (tree_to_aff_combination): ... here.
5381 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5383 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5384 reg_pressure model function.
5385 (ivopts_global_cost_for_size): Delete.
5386 (determine_set_costs, iv_ca_recount_cost): Call new model function
5387 ivopts_estimate_reg_pressure.
5389 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5391 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5392 expensive than udiv. Remove floating point cases from mod.
5394 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5396 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5399 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5401 * config/aarch64/aarch64.md
5402 (copysignsf3): Fix mask generation.
5404 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5406 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5408 (class dump_manager): Add register_dumps method.
5409 * dumpfile.c: Include langhooks.h.
5410 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5411 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
5412 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
5413 (dump_manager::dump_register): Start with 512 entries instead of 32.
5414 (dump_manager::register_dumps): New method.
5415 * toplev.c (general_init): Instead of invoking register_dumps
5416 langhook, invoke register_dumps method on the dump manager.
5417 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
5420 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
5422 * doc/md.texi: Clarify the restrictions on a define_insn condition.
5423 Say that # requires an associated define_split to exist, and that
5424 the define_split must be suitable for use after register allocation.
5426 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5428 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
5429 (compute_outgoing_frequencies): Also initialize zero counts.
5430 (find_many_sub_basic_blocks): Do not produce uninitialized profile
5431 around loops; preserve more of profile when nothing changes.
5433 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
5435 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
5437 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
5438 * config/arm/arm-cpu-cdata.h: Regenerate.
5439 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
5440 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
5441 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
5442 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
5443 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
5445 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
5447 * config/arm/t-rmprofile: Likewise.
5448 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
5450 2017-06-06 David S. Miller <davem@davemloft.net>
5453 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
5454 blockage if function uses alloca.
5456 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5458 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
5459 New "uid" fields to hold pretty-print IDs of group and ref.
5460 Memory references are now identified as <group_id>:<ref_id>
5461 instead of using [random] addresses.
5462 (dump_mem_details): Simplify, no functional change.
5463 (dump_mem_ref): Simplify and make output more concise.
5464 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
5465 (find_or_create_group): Initialize group uid.
5466 (record_ref): Initialize ref uid. Improve debug output.
5467 (prune_group_by_reuse, should_issue_prefetch_p,)
5468 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
5469 (mark_nontemporal_store, determine_loop_nest_reuse):
5470 Improve debug output.
5472 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5474 * dbgcnt.def (prefetch): New debug counter.
5475 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
5476 (schedule_prefetches): Stop issueing prefetches if debug counter
5479 2017-06-06 Tom de Vries <tom@codesourcery.com>
5481 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
5482 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
5484 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5486 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
5487 Use aarch64_reg_or_zero predicate for operand 4.
5488 (aarch64_compare_and_swap<mode> define_insn_and_split):
5489 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
5490 (aarch64_store_exclusive<mode>): Likewise for operand 2.
5492 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5494 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
5495 (arm_compute_save_core_reg_mask): This.
5496 (thumb1_compute_save_reg_mask): Rename into ...
5497 (thumb1_compute_save_core_reg_mask): This.
5498 (arm_compute_save_reg0_reg12_mask): Adapt comment.
5499 (arm_compute_frame_layout): Likewise.
5501 2017-06-06 Richard Biener <rguenther@suse.de>
5503 PR tree-optimization/80974
5504 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
5505 keep or clear leaders SSA info.
5507 2017-06-06 Tom de Vries <tom@codesourcery.com>
5509 * config/nvptx/nvptx.c (split_mode_p): New function.
5510 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
5512 2017-06-06 Tom de Vries <tom@codesourcery.com>
5514 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
5516 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5519 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
5522 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5524 * shrink-wrap.c (handle_simple_exit): Update profile.
5525 (try_shrink_wrapping): Upate profile.
5527 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5529 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
5530 (tree_guess_outgoing_edge_probabilities): New.
5531 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
5532 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
5534 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5536 * ipa-split.c (split_function): Initialize return bb profile.
5538 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5540 * profile.c (compute_branch_probabilities): Also initialize
5543 2017-06-06 Richard Biener <rguenther@suse.de>
5545 PR tree-optimization/80928
5546 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
5547 (vect_analyze_loop_operations): Properly guard analysis for
5549 (vect_transform_loop): Likewise.
5550 (vect_analyze_loop_2): Also reset SLP type on PHIs.
5551 (vect_model_induction_cost): Do not cost for pure SLP.
5552 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
5553 of induction in inner loop vectorization.
5554 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
5555 (vect_get_and_check_slp_defs): Handle vect_induction_def.
5556 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
5558 (vect_analyze_slp_cost_1): Cost induction.
5559 (vect_detect_hybrid_slp_stmts): Handle PHIs.
5560 (vect_get_slp_vect_defs): Likewise.
5561 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
5562 (vect_transform_stmt): Handle SLP reductions.
5563 * tree-vectorizer.h (vectorizable_induction): Adjust.
5565 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5567 * config/rs6000/rs6000.c (make_resolver_func): Update
5568 init_lowered_empty_function call.
5570 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
5572 * doc/invoke.texi: Document the -fprofile-abs-path option.
5573 * common.opt (fprofile-abs-path): New option.
5574 * gcov-io.h (gcov_write_filename): Declare.
5575 * gcov-io.c (gcov_write_filename): New function.
5576 * coverage.c (coverage_begin_function): Use gcov_write_filename.
5577 * profile.c (output_location): Likewise.
5579 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5581 * shring-wrap.c: Revert accidental commit.
5583 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
5585 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
5587 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5589 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
5591 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
5592 profile in callgraph edge.
5593 * profile-count.h (apply_probability): If THIS is 0, then result is 0
5594 (apply_scale): Likewise.
5595 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
5596 Also scale profile when inlining function with zero profile.
5597 (initialize_cfun): Update exit block profile even when it is zero.
5598 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
5599 when profile is read.
5601 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5603 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
5604 (CLONE_*): New constants to define the processors we can generate
5605 code for with the target_clone attribute.
5606 (rs6000_clone_map): New array to identify which clone processors
5607 the current program is running on.
5608 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
5609 target_clone attribute.
5610 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
5611 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
5612 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
5613 (cpu_expand_builtin): Add support for target_clone attribute.
5614 (rs6000_valid_attribute_p): Allow "default" attribute.
5615 (get_decl_name): New debug function to simplify printing the
5616 current function name in debugging statements.
5617 (rs6000_clone_priority): New functions to support the target_clone
5618 attribute, and be able to generate code to switch between ISA 2.05
5619 through ISA 3.0 (power6 through power9).
5620 (rs6000_compare_version_priority): Likewise.
5621 (rs6000_get_function_versions_dispatcher): Likewise.
5622 (make_resolver_func): Likewise.
5623 (add_condition_to_bb): Likewise.
5624 (dispatch_function_versions): Likewise.
5625 (rs6000_generate_version_dispatcher_body): Likewise.
5626 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
5627 (fusion_gpr_load_p): Fix a spacing issue.
5628 * doc/extend.texi (Common Function Attributes): Document that the
5629 PowerPC supports the target_clone attribute.
5631 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
5633 * config/arm/arm.h: explain F symbol found in description of ARM
5634 register allocation in its legend.
5636 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5638 * config/mips/frame-header-opt.c: Include profile-count.h.
5639 * config/riscv/riscv.c: Include profile-count.h
5641 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5643 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
5645 (sm_set_flag_if_changed): Add bbs field.
5646 (execute_sm_if_changed_flag_set): Pass BBS.
5647 (execute_sm): Update.
5649 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5651 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5654 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5656 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
5657 (peephole2): New peephole2 to emit the above.
5658 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
5660 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5662 * config/aarch64/aarch64.c (define_peephole2 above
5663 *sub_<shift>_<mode>): New peephole.
5665 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
5667 * config/i386/i386.c (make_resolver_func): Update.
5668 * Makefile.in: Add profile-count.h and profile-count.o
5669 * auto-profile.c (afdo_indirect_call): Update to new API.
5670 (afdo_set_bb_count): Update.
5671 (afdo_propagate_edge): Update.
5672 (afdo_propagate_circuit): Update.
5673 (afdo_calculate_branch_prob): Update.
5674 (afdo_annotate_cfg): Update.
5675 * basic-block.h: Include profile-count.h
5676 (struct edge_def): Turn count to profile_count.
5677 (struct basic_block_def): Likewie.
5678 (REG_BR_PROB_BASE): Move to profile-count.h
5679 (RDIV): Move to profile-count.h
5680 * bb-reorder.c (max_entry_count): Turn to profile_count.
5681 (find_traces): Update.
5682 (rotate_loop):Update.
5683 (connect_traces):Update.
5684 (sanitize_hot_paths):Update.
5685 * bt-load.c (migrate_btr_defs): Update.
5686 * cfg.c (RDIV): Remove.
5687 (init_flow): Use alloc_block.
5688 (alloc_block): Uninitialize count.
5689 (unchecked_make_edge): Uninitialize count.
5690 (check_bb_profile): Update.
5691 (dump_edge_info): Update.
5692 (dump_bb_info): Update.
5693 (update_bb_profile_for_threading): Update.
5694 (scale_bbs_frequencies_int): Update.
5695 (scale_bbs_frequencies_gcov_type): Update.
5696 (scale_bbs_frequencies_profile_count): New.
5697 * cfg.h (update_bb_profile_for_threading): Update.
5698 (scale_bbs_frequencies_profile_count): Declare.
5699 * cfgbuild.c (compute_outgoing_frequencies): Update.
5700 (find_many_sub_basic_blocks): Update.
5701 * cfgcleanup.c (try_forward_edges): Update.
5702 (try_crossjump_to_edge): Update.
5703 * cfgexpand.c (expand_gimple_tailcall): Update.
5704 (construct_exit_block): Update.
5705 * cfghooks.c (verify_flow_info): Update.
5706 (dump_bb_for_graph): Update.
5707 (split_edge): Update.
5708 (make_forwarder_block): Update.
5709 (duplicate_block): Update.
5710 (account_profile_record): Update.
5711 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5712 (get_estimated_loop_iterations): Update.
5713 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5714 (single_likely_exit): Update.
5715 * cfgloopmanip.c (scale_loop_profile): Update.
5717 (set_zero_probability): Update.
5718 (lv_adjust_loop_entry_edge): Update.
5719 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5720 (purge_dead_edges): Update.
5721 (rtl_account_profile_record): Update.
5722 * cgraph.c (cgraph_node::create): Uninitialize count.
5723 (symbol_table::create_edge): Uninitialize count.
5724 (cgraph_update_edges_for_call_stmt_node): Update.
5725 (cgraph_edge::dump_edge_flags): Update.
5726 (cgraph_node::dump): Update.
5727 (cgraph_edge::maybe_hot_p): Update.
5728 * cgraph.h: Include profile-count.h
5729 (create_clone), create_edge, create_indirect_edge): Update.
5730 (cgraph_node): Turn count to profile_count.
5731 (cgraph_edge0: Likewise.
5732 (make_speculative, clone): Update.
5733 (create_edge): Update.
5734 (init_lowered_empty_function): Update.
5735 * cgraphclones.c (cgraph_edge::clone): Update.
5736 (duplicate_thunk_for_node): Update.
5737 (cgraph_node::create_clone): Update.
5738 * cgraphunit.c (cgraph_node::analyze): Update.
5739 (cgraph_node::expand_thunk): Update.
5740 * final.c (dump_basic_block_info): Update.
5741 * gimple-streamer-in.c (input_bb): Update.
5742 * gimple-streamer-out.c (output_bb): Update.
5743 * graphite.c (print_global_statistics): Update.
5744 (print_graphite_scop_statistics): Update.
5745 * hsa-brig.c: Include basic-block.h.
5746 * hsa-dump.c: Include basic-block.h.
5747 * hsa-gen.c (T sum_slice): Update.
5748 (convert_switch_statements):Update.
5749 * hsa-regalloc.c: Include basic-block.h.
5750 * ipa-chkp.c (chkp_produce_thunks): Update.
5751 * ipa-cp.c (struct caller_statistics): Update.
5752 (init_caller_stats): Update.
5753 (gather_caller_stats): Update.
5754 (ipcp_cloning_candidate_p): Update.
5755 (good_cloning_opportunity_p): Update.
5756 (get_info_about_necessary_edges): Update.
5757 (dump_profile_updates): Update.
5758 (update_profiling_info): Update.
5759 (update_specialized_profile): Update.
5760 (perhaps_add_new_callers): Update.
5761 (decide_about_value): Update.
5762 (ipa_cp_c_finalize): Update.
5763 * ipa-devirt.c (struct odr_type_warn_count): Update.
5764 (struct decl_warn_count): Update.
5765 (struct final_warning_record): Update.
5766 (possible_polymorphic_call_targets): Update.
5767 (ipa_devirt): Update.
5768 * ipa-fnsummary.c (redirect_to_unreachable): Update.
5769 * ipa-icf.c (sem_function::merge): Update.
5770 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
5771 * ipa-inline.c (compute_uninlined_call_time): Update.
5772 (compute_inlined_call_time): Update.
5773 (want_inline_small_function_p): Update.
5774 (want_inline_self_recursive_call_p): Update.
5775 (edge_badness): Update.
5776 (lookup_recursive_calls): Update.
5777 (recursive_inlining): Update.
5778 (inline_small_functions): Update.
5779 (dump_overall_stats): Update.
5780 (dump_inline_stats): Update.
5781 * ipa-profile.c (ipa_profile_generate_summary): Update.
5782 (ipa_propagate_frequency): Update.
5783 (ipa_profile): Update.
5784 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
5785 * ipa-utils.c (ipa_merge_profiles): Update.
5786 * loop-doloop.c (doloop_modify): Update.
5787 * loop-unroll.c (report_unroll): Update.
5788 (unroll_loop_runtime_iterations): Update.
5789 * lto-cgraph.c (lto_output_edge): Update.
5790 (lto_output_node): Update.
5791 (input_node): Update.
5792 (input_edge): Update.
5793 (merge_profile_summaries): Update.
5794 * lto-streamer-in.c (input_cfg): Update.
5795 * lto-streamer-out.c (output_cfg): Update.
5796 * mcf.c (create_fixup_graph): Update.
5797 (adjust_cfg_counts): Update.
5798 (sum_edge_counts): Update.
5799 * modulo-sched.c (sms_schedule): Update.
5800 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5801 * predict.c (maybe_hot_count_p): Update.
5802 (probably_never_executed): Update.
5803 (dump_prediction): Update.
5804 (combine_predictions_for_bb): Update.
5805 (propagate_freq): Update.
5806 (handle_missing_profiles): Update.
5807 (counts_to_freqs): Update.
5808 (rebuild_frequencies): Update.
5809 (force_edge_cold): Update.
5810 * predict.h: Include profile-count.h
5811 (maybe_hot_count_p, counts_to_freqs): UPdate.
5812 * print-rtl-function.c: Do not include cfg.h
5813 * print-rtl.c: Include basic-block.h
5814 * profile-count.c: New file.
5815 * profile-count.h: New file.
5816 * profile.c (is_edge_inconsistent): Update.
5817 (correct_negative_edge_counts): Update.
5818 (is_inconsistent): Update.
5819 (set_bb_counts): Update.
5820 (read_profile_edge_counts): Update.
5821 (compute_frequency_overlap): Update.
5822 (compute_branch_probabilities): Update; Initialize and deinitialize
5824 (branch_prob): Update.
5825 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
5826 (edge_gcov_count): New.
5827 (bb_gcov_count): New.
5828 * shrink-wrap.c (try_shrink_wrapping): Update.
5829 * tracer.c (better_p): Update.
5830 * trans-mem.c (expand_transaction): Update.
5831 (ipa_tm_insert_irr_call): Update.
5832 (ipa_tm_insert_gettmclone_call): Update.
5833 * tree-call-cdce.c: Update.
5834 * tree-cfg.c (gimple_duplicate_sese_region): Update.
5835 (gimple_duplicate_sese_tail): Update.
5836 (gimple_account_profile_record): Update.
5837 (execute_fixup_cfg): Update.
5838 * tree-inline.c (copy_bb): Update.
5839 (copy_edges_for_bb): Update.
5840 (initialize_cfun): Update.
5841 (freqs_to_counts): Update.
5842 (copy_cfg_body): Update.
5843 (expand_call_inline): Update.
5844 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5845 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
5846 (try_unroll_loop_completely): Update.
5847 (try_peel_loop): Update.
5848 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5849 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
5850 * tree-ssa-loop-split.c (connect_loops): Update.
5851 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5852 * tree-ssa-reassoc.c (branch_fixup): Update.
5853 * tree-ssa-tail-merge.c (replace_block_by): Update.
5854 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
5855 (compute_path_counts): Update.
5856 (update_profile): Update.
5857 (recompute_probabilities): Update.
5858 (update_joiner_offpath_counts): Update.
5859 (estimated_freqs_path): Update.
5860 (freqs_to_counts_path): Update.
5861 (clear_counts_path): Update.
5862 (ssa_fix_duplicate_block_edges): Update.
5863 (duplicate_thread_path): Update.
5864 * tree-switch-conversion.c (case_bit_test_cmp): Update.
5865 (struct switch_conv_info): Update.
5866 * tree-tailcall.c (decrease_profile): Update.
5867 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5868 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5869 * value-prof.c (check_counter): Update.
5870 (gimple_divmod_fixed_value): Update.
5871 (gimple_mod_pow2): Update.
5872 (gimple_mod_subtract): Update.
5873 (gimple_ic_transform): Update.
5874 (gimple_stringop_fixed_value): Update.
5875 * value-prof.h (gimple_ic): Update.
5877 2017-06-02 Carl Love <cel@us.ibm.com>
5879 * config/rs6000/rs6000-c: Add support for built-in functions
5880 vector double vec_doublee (vector signed int);
5881 vector double vec_doublee (vector unsigned int);
5882 vector double vec_doublee (vector float);
5883 vector double vec_doubleh (vector signed int);
5884 vector double vec_doubleh (vector unsigned int);
5885 vector double vec_doubleh (vector float);
5886 vector double vec_doublel (vector signed int);
5887 vector double vec_doublel (vector unsigned int);
5888 vector double vec_doublel (vector float);
5889 vector double vec_doubleo (vector signed int);
5890 vector double vec_doubleo (vector unsigned int);
5891 vector double vec_doubleo (vector float);.
5892 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
5893 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
5895 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
5896 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
5897 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
5899 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
5900 vec_doublel, vec_doubleh.
5901 * doc/extend.texi: Update the built-in documentation file for the
5902 new built-in functions.
5904 2017-06-02 David Malcolm <dmalcolm@redhat.com>
5907 * ipa-inline-analysis.c (free_growth_caches): Set
5908 edge_removal_hook_holder to NULL after removing it.
5910 2017-06-02 Sudakshina Das <sudi.das@arm.com>
5912 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5913 comparision with zero.
5915 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
5916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5917 for early expansion of vec_min and vec_max builtins.
5918 (builtin_function_type): Add min/max unsigned variants to those
5919 identified as having unsigned arguments.
5921 2017-06-02 Olivier Hainque <hainque@adacore.com>
5923 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
5925 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5927 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
5928 Use VALL_F16 iterator rather than VALL.
5930 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5932 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
5933 Emit CBNZ inside loop when doing a strong exchange and comparing
5934 against zero. Generate the CC flags after the loop.
5936 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
5938 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
5939 (dl_section_ref): New.
5940 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
5941 On AIX, append an expression to subtract the size of the
5942 section length to dl_section_ref.
5944 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
5946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5947 for early expansion of vector absolute builtins.
5949 2017-06-02 Richard Biener <rguenther@suse.de>
5951 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
5952 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
5954 2017-06-02 Richard Biener <rguenther@suse.de>
5956 PR tree-optimization/80948
5957 * tree-tailcall.c (find_tail_calls): Track stmts to move in
5960 2017-06-02 Richard Biener <rguenther@suse.de>
5962 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
5964 * tree-vect-stmts.c (process_use): Do not mark backedge defs
5965 for inductions as relevant.
5967 2017-06-02 Richard Biener <rguenther@suse.de>
5969 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
5970 (vectorizable_induction): ... this. Remove dead code.
5972 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
5974 * builtins. (expand_builtin_alloca): Remove second parameter and
5975 infer its value from the first parameter instead.
5976 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
5978 2017-06-02 Jakub Jelinek <jakub@redhat.com>
5980 PR rtl-optimization/80903
5981 * loop-doloop.c (add_test): Unshare sequence.
5983 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
5985 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
5987 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
5989 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
5991 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
5992 xlogue_layout::get_instance, logue_layout::xlogue_layout,
5993 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
5994 (xlogue_layout::get_stub_rtx): Make static.
5995 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
5996 (xlogue_layout::compute_stub_managed_regs): Rename to...
5997 (xlogue_layout::count_stub_managed_regs): ...this.
5998 (xlogue_layout::is_stub_managed_reg): New function.
5999 (xlogue_layout::m_stub_names): Rename to...
6000 (xlogue_layout::s_stub_names): ...this, make static.
6001 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6002 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6003 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6004 xlogue_layout::s_stub_names): Instantiate statics.
6005 (stub_managed_regs): Remove.
6006 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6007 (disable_call_ms2sysv_xlogues): Rename to...
6008 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6009 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6011 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6012 change after reload_completed.
6013 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6015 (ix86_expand_prologue): Likewise.
6016 (ix86_expand_epilogue): Likewise.
6017 (ix86_expand_split_stack_prologue): Likewise.
6018 (ix86_compute_frame_layout): Remove frame parameter ...
6019 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6020 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6022 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6023 (ix86_frame): Move from here ...
6024 * config/i386/i386.h (ix86_frame): ... to here.
6025 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6026 complete ix86_frame data structure instead. Remove some_ld_name.
6028 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6030 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6031 symbols that hold a DECL_VALUE_EXPR.
6033 2017-06-01 Martin Jambor <mjambor@suse.cz>
6035 PR tree-optimization/80898
6036 * tree-sra.c (process_subtree_disqualification): Removed.
6037 (disqualify_candidate): Do not acll
6038 process_subtree_disqualification.
6039 (subtree_mark_written_and_enqueue): New function.
6040 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6041 RHS has been disqualified and re-queue LHS if necessary. Apart
6042 from that, ignore disqualified RHS.
6044 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6046 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6047 address fetch for z10 or later.
6049 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6051 * config/arc/arc.md (tst_movb): Add guard when splitting.
6053 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6055 * config/arc/arc.c (arc_can_eliminate): Test against
6056 arc_frame_pointer_needed.
6058 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6060 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6061 to prevent store reordering.
6062 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6063 (type): Add block type.
6064 (stack_tie): Define special instruction to be used in
6067 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6069 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6070 constraint. It is not valid for the pattern.
6071 (noncommutative_binary_comparison): Likewise.
6073 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6075 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6078 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6080 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6081 be used by the reg-alloc.
6083 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6085 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6086 reg-alloc when having mul64 or mul32x16 instructions.
6087 (mulsidi3): Likewise.
6088 (umulsidi3): Likewise.
6089 (mulsi32x16): New pattern.
6090 (mulsi64): Likewise.
6091 (mulsidi64): Likewise.
6092 (umulsidi64): Likewise.
6093 (MUL32x16_REG): Define.
6094 (mul64_600): Use MUL32x16_REG.
6095 (mac64_600): Likewise.
6096 (umul64_600): Likewise.
6097 (umac64_600): Likewise.
6099 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6101 * config/arc/arc.md (mulsi3_700): Make it commutative.
6103 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6105 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6107 (*sign_extendsidi2_insn): Likewise for movstosw.
6109 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6111 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6112 the type of the input discriminant value. Convert the
6113 discriminant value of signedness vary.
6115 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6117 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6118 Add to -Wall section.
6120 2017-06-01 Richard Biener <rguenther@suse.de>
6123 * fold-const.c (fold_plusminus_mult_expr): If the factored
6124 factor may be zero use a wrapping type for the inner operation.
6125 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6126 and handle moved defs.
6127 (process_assignment): Properly guard the unary op case. Return a
6128 tri-state indicating that moving the stmt before the call may allow
6129 to continue. Pass through to_move.
6130 (find_tail_calls): Handle moving unrelated defs before
6133 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6136 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6137 splitter result in the canonical way.
6139 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6141 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6142 also for 32bit target. Update insn attributes.
6143 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6145 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6147 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6148 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6149 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6150 (_mm_maskz_min_ss): New intrinsics.
6152 2017-05-31 Martin Liska <mliska@suse.cz>
6154 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6155 Change comment style to one we normally use.
6156 (vectorizable_reduction): Likewise.
6157 (vectorizable_induction): Likewise.
6158 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6159 (vectorizable_call): Likewise.
6160 (vectorizable_simd_clone_call): Likewise.
6161 (vectorizable_conversion): Likewise.
6162 (vectorizable_assignment): Likewise.
6163 (vectorizable_shift): Likewise.
6164 (vectorizable_operation): Likewise.
6165 (vectorizable_store): Likewise.
6166 (vectorizable_load): Likewise.
6167 * tree-vectorizer.h: Likewise.
6169 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6171 * passes.c (emergency_dump_function): New.
6172 * tree-pass.h (emergency_dump_function): Declare.
6173 * plugin.c (plugins_internal_error_function): Remove.
6174 * plugin.h (plugins_internal_error_function): Remove declaration.
6175 * toplev.c (internal_error_function): New static function. Use it...
6176 (general_init): ...here.
6178 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6180 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6181 (arc_rtx_costs): Add costs for new patterns.
6182 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6183 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6185 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6187 * tree-ssa-strlen.c (get_next_strinfo): New function.
6188 (get_stridx_plus_constant): Use it.
6189 (zero_length_string): Likewise.
6190 (adjust_related_strinfos): Likewise.
6191 (adjust_last_stmt): Likewise.
6193 2017-05-31 Richard Biener <rguenther@suse.de>
6196 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6197 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6199 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6201 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6202 loop_vinfo argument and use of dependence distance vectors.
6203 Check instead whether the two references differ only in their
6204 initial value and assume that they have the same alignment if the
6205 difference is a multiple of the vector alignment.
6206 (vect_analyze_data_refs_alignment): Update call accordingly.
6208 2017-05-31 Martin Liska <mliska@suse.cz>
6211 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6213 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6215 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6216 (create_intersect_range_checks): Move from ...
6217 * tree-data-ref.c (create_intersect_range_checks_index)
6218 (create_intersect_range_checks): ... to here.
6219 (create_runtime_alias_checks): New function factored from ...
6220 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6221 here. Call above function.
6222 * tree-data-ref.h (create_runtime_alias_checks): New function.
6224 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6226 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6227 segment length for dr_b and compute it in wide_int.
6229 2017-05-31 Richard Biener <rguenther@suse.de>
6231 PR tree-optimization/80906
6232 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6233 and pass through iv_map.
6234 (copy_bb_and_scalar_dependences): Adjust.
6235 (translate_pending_phi_nodes): Likewise.
6236 (copy_loop_close_phi_args): Handle code-generating IVs instead
6239 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6241 * diagnostic-color.c (color_dict): Add "type-diff".
6242 (parse_gcc_colors): Update comment.
6243 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6244 -fdiagnostics-show-template-tree and -fno-elide-type.
6245 (GCC_COLORS): Add type-diff to example.
6247 (-fdiagnostics-show-template-tree): New.
6248 (-fno-elide-type): New.
6249 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6250 the pp_format_decoder callback. Call any m_format_postprocessor's
6252 (pretty_printer::pretty_printer): Initialize
6253 m_format_postprocessor.
6254 (pretty_printer::~pretty_printer): Delete any
6255 m_format_postprocessor.
6256 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6257 (class format_postprocessor): New class.
6258 (struct pretty_printer::format_decoder): Document the new parameters.
6259 (struct pretty_printer::m_format_postprocessor): New field.
6260 * tree-diagnostic.c (default_tree_printer): Update for new
6261 bool and const char ** params.
6262 * tree-diagnostic.h (default_tree_printer): Likewise.
6264 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6266 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6267 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6268 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6269 rs6000_gen_cell_microcode code.
6270 (rs6000_final_prescan_insn): Delete.
6271 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6272 "warn-cell-microcode" entries.
6273 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6274 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6275 throughout. Change cc_reg_not_micro_cr0_operand to
6276 cc_reg_not_cr0_operand throughout.
6277 (*extendhi<mode>2_noload): Delete.
6278 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6279 (mwarn-cell-microcode): Delete.
6280 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6281 -mgen-cell-microcode and -mwarn-cell-microcode.
6283 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6286 * config/i386/constraints.md (Yd): New constraint.
6288 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6289 and (?Yd, r) alternatives. Update insn attributes.
6290 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6291 and (?*Yd, r) alternatives. Update insn attributes.
6292 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6294 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6296 * gimplify.c (gimplify_modify_expr): Don't create a
6297 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6300 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6302 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6304 2017-05-30 Richard Biener <rguenther@suse.de>
6306 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6307 and reduc_def fields.
6308 (STMT_VINFO_REDUC_TYPE): New define.
6309 (STMT_VINFO_REDUC_DEF): Likewise.
6310 (vect_force_simple_reduction): Adjust prototype.
6311 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6312 (vect_is_simple_reduction): Remove check_reduction argument.
6313 (vect_force_simple_reduction): Adjust and set
6314 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6315 (vectorizable_reduction): Do not re-do reduction analysis
6316 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6317 * tree-parloops.c (gather_scalar_reductions): Adjust.
6319 2017-05-30 Richard Biener <rguenther@suse.de>
6322 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6325 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6327 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6328 Introduce unknown_misalignment parameter and remove vf.
6329 (vect_peeling_hash_get_lowest_cost):
6330 Pass unknown_misalignment parameter.
6331 (vect_enhance_data_refs_alignment):
6332 Fix unsupportable data ref treatment.
6334 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6336 * tree-vect-data-refs.c (vect_get_data_access_cost):
6337 Workaround for SLP handling.
6338 (vect_enhance_data_refs_alignment):
6339 Compute costs for doing no peeling at all, compare to the best
6340 peeling costs so far and avoid peeling if cheaper.
6342 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6344 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6345 Return peeling info and set costs to zero for unlimited cost
6347 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6348 with unknown misalignment. Compute and costs for unknown
6349 misalignment, compare them to the costs for known misalignment
6350 and choose the cheapest for peeling.
6352 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6354 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6355 (vect_get_peeling_costs_all_drs): Create function.
6356 (vect_peeling_hash_get_lowest_cost):
6357 Use vect_get_peeling_costs_all_drs.
6358 (vect_peeling_supportable): Create function.
6359 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6361 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6363 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6364 DR_HAS_NEGATIVE_STEP.
6365 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6366 (vect_enhance_data_refs_alignment): Use.
6367 (vect_duplicate_ssa_name_ptr_info): Use.
6368 * tree-vectorizer.h (dr_misalignment): Use.
6369 (known_alignment_for_access_p): Use.
6371 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6374 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6376 (has_section_name): New function.
6378 2017-05-30 Martin Liska <mliska@suse.cz>
6381 * auto-profile.c (get_function_decl_from_block): Fix
6384 2017-05-30 Richard Biener <rguenther@suse.de>
6387 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6389 2017-05-30 Martin Liska <mliska@suse.cz>
6391 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6392 * dumpfile.h (struct dump_file_info): Remove ctors.
6394 2017-05-30 Martin Liska <mliska@suse.cz>
6396 * predict.def: Fix GNU coding style.
6398 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6400 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6401 Mark 'to' argument with ATTRIBUTE_UNUSED.
6403 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6405 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6406 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6407 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6410 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
6412 * doc/install.texi (Options specification): Restore entry of
6413 --enable-sjlj-exceptions.
6415 2017-05-27 Michael Eager <eager@eagercon.com>
6418 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
6420 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
6422 * config/microblaze/microblaze.h
6423 (FIXED_REGISTERS): Update in macro.
6424 (CALL_USED_REGISTERS): Update in macro.
6426 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
6428 * doc/install.texi: Add links to macOS binary distributions.
6430 2017-05-27 Jakub Jelinek <jakub@redhat.com>
6434 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6436 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6438 2017-05-26 Martin Liska <mliska@suse.cz>
6440 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
6442 2017-05-26 Martin Liska <mliska@suse.cz>
6444 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
6445 always leading ';; '.
6446 (dump_bb_info): Likewise.
6447 (brief_dump_cfg): Likewise.
6448 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
6449 * dumpfile.c: Remove usage of TDF_VERBOSE.
6450 * dumpfile.h (enum dump_kind): Likewise.
6451 (dump_gimple_bb_header): Do not use TDF_COMMENT.
6452 * print-tree.c (debug_verbose): Remove.
6453 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
6454 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
6455 * tree-diagnostic.c (default_tree_printer): Replace
6456 TDF_DIAGNOSTIC with TDF_SLIM.
6458 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6460 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
6461 in parameter loop, rather than loop_vinfo.
6462 (create_intersect_range_checks): Ditto.
6463 (vect_create_cond_for_alias_checks): Update call to above functions.
6465 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6467 PR tree-optimization/80815
6468 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
6469 for merging runtime alias checks. Handle negative DR_STEPs.
6471 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6473 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
6475 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
6476 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
6477 out code pruning runtime alias checks.
6478 * tree-data-ref.c (prune_runtime_alias_test_list): New function
6479 factored out from above.
6480 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6482 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6484 (prune_runtime_alias_test_list): New decalaration.
6486 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6488 * tree-vect-data-refs.c (compare_tree): Rename and move ...
6489 * tree-data-ref.c (data_ref_compare_tree): ... to here.
6490 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
6491 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
6492 (operator==, comp_dr_with_seg_len_pair): Ditto.
6493 (vect_prune_runtime_alias_test_list): Ditto.
6495 2017-05-26 Martin Liska <mliska@suse.cz>
6498 * params.def: Bound partial-inlining-entry-probability param.
6500 2017-05-26 Marek Polacek <polacek@redhat.com>
6503 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
6506 2017-05-26 Richard Biener <rguenther@suse.de>
6508 PR tree-optimization/80842
6509 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
6512 2017-05-26 Richard Biener <rguenther@suse.de>
6514 PR tree-optimization/80844
6515 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
6517 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
6519 * doc/md.texi (Machine Constraints): Update x86 family
6520 machine constraints section to match 'config/i386/constraints.md'.
6522 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
6524 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
6526 2017-05-25 Nathan Sidwell <nathan@acm.org>
6528 * doc/invoke.texi (--enable-languages): Update documentation.
6530 2017-05-25 Martin Liska <mliska@suse.cz>
6532 * dumpfile.c: Add TDF_FOLDING.
6533 * dumpfile.h (enum dump_kind): Likewise.
6534 * genmatch.c (dt_simplify::gen_1): Use it.
6536 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6538 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
6540 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6542 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6543 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
6545 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6547 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
6548 * match.pd (X == C): Rewrite it here.
6549 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
6550 with_certain_nonzero_bits2): New predicates.
6551 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
6553 2017-05-24 Nathan Sidwell <nathan@acm.org>
6555 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
6558 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
6561 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6563 * config/powerpcspe: New port. Files are copied from the rs6000
6564 port, with "rs6000" in filenames replaced by "powerpcspe".
6566 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
6568 PR rtl-optimization/80754
6569 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
6571 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
6573 * config/sparc/sparc.md (length): Return the correct value for -mflat
6574 sibcalls to match output_sibcall.
6576 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6580 * config/rs6000/rs6000.c (struct machine_function): Add new field
6582 (rs6000_get_separate_components): Init that field, use it.
6583 (rs6000_components_for_bb): Use the field.
6585 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
6587 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
6589 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
6592 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
6594 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6597 * config/s390/s390.c (s390_check_qrst_address): Check incoming
6598 address against address_operand predicate.
6599 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
6601 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
6603 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
6604 parameters passed indirectly.
6606 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
6608 * config/i386/i386.md (*movdi_internal): Remove SSE4
6609 alternative 18 (?r, *v). Update insn attributes.
6610 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
6611 Update insn attributes.
6612 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
6613 Update insn attributes.
6614 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
6615 alternative 1 (r, v). Remove isa attribute.
6616 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
6617 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
6618 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
6620 2017-05-23 Tom de Vries <tom@codesourcery.com>
6622 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
6625 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6627 * cgraphunit.c (symbol_table::process_new_functions): Update.
6628 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
6629 (inline_generate_summary): Rename to ...
6630 (ipa_fn_summary_generate): ... this one.
6631 (inline_read_summary): Rename to ...
6632 (ipa_fn_summary_read): ... this one.
6633 (inline_write_summary): Rename to ...
6634 (ipa_fn_summary_write): ... this one.
6635 (inline_free_summary): Rename to ...
6636 (ipa_free_fn_summary): ... this one.
6637 (pass_data_local_fn_summary, pass_local_fn_summary,
6638 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
6639 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
6640 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
6641 make_pass_ipa_fn_summary): New.
6642 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
6643 inline_write_summary, inline_free_summary): Remove.
6644 (ipa_free_fn_summary) : New.
6645 * ipa-inline.c (ipa_inline): Update.
6646 (pass_ipa_inline): Do not generate summaries.
6647 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
6649 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
6650 and add pass_ipa_fn_summary.
6651 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
6653 (make_pass_inline_parameters): Remove.
6655 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
6657 * omp-low.c (struct omp_context): Remove "default_kind" member.
6660 * omp-offload.c (execute_oacc_device_lower): Remove the
6661 parallelism dimensions function attributes for unparallelized
6662 OpenACC kernels constructs.
6664 2017-05-23 Martin Liska <mliska@suse.cz>
6666 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
6668 (cgraph_edge::make_speculative): Likewise.
6669 (cgraph_edge::resolve_speculation): Likewise.
6670 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
6671 (cgraph_node::dump): Likewise.
6672 * cgraph.h: Likewise.
6673 * cgraphunit.c (analyze_functions): Likewise.
6674 (symbol_table::compile): Likewise.
6675 * ipa-cp.c (print_all_lattices): Likewise.
6676 (determine_versionability): Likewise.
6677 (initialize_node_lattices): Likewise.
6678 (ipcp_verify_propagated_values): Likewise.
6679 (estimate_local_effects): Likewise.
6680 (update_profiling_info): Likewise.
6681 (create_specialized_node): Likewise.
6682 (perhaps_add_new_callers): Likewise.
6683 (decide_about_value): Likewise.
6684 (decide_whether_version_node): Likewise.
6685 (identify_dead_nodes): Likewise.
6686 (ipcp_store_bits_results): Likewise.
6687 * ipa-devirt.c (dump_targets): Likewise.
6688 (ipa_devirt): Likewise.
6689 * ipa-icf.c (sem_item::dump): Likewise.
6690 (sem_function::equals): Likewise.
6691 (sem_variable::equals): Likewise.
6692 (sem_item_optimizer::read_section): Likewise.
6693 (sem_item_optimizer::execute): Likewise.
6694 (congruence_class::dump): Likewise.
6695 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
6696 (dump_inline_summary): Likewise.
6697 (estimate_node_size_and_time): Likewise.
6698 (inline_analyze_function): Likewise.
6699 * ipa-inline-transform.c (inline_call): Likewise.
6700 * ipa-inline.c (report_inline_failed_reason): Likewise.
6701 (want_early_inline_function_p): Likewise.
6702 (edge_badness): Likewise.
6703 (update_edge_key): Likewise.
6704 (inline_small_functions): Likewise.
6705 * ipa-profile.c (ipa_profile): Likewise.
6706 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
6707 (ipa_make_edge_direct_to_target): Likewise.
6708 (remove_described_reference): Likewise.
6709 (ipa_impossible_devirt_target): Likewise.
6710 (propagate_controlled_uses): Likewise.
6711 (ipa_print_node_params): Likewise.
6712 (ipcp_transform_function): Likewise.
6713 * ipa-pure-const.c (pure_const_read_summary): Likewise.
6714 (propagate_pure_const): Likewise.
6715 * ipa-reference.c (generate_summary): Likewise.
6716 (read_write_all_from_decl): Likewise.
6717 (propagate): Likewise.
6718 (ipa_reference_read_optimization_summary): Likewise.
6719 * ipa-utils.c (ipa_merge_profiles): Likewise.
6720 * ipa.c (walk_polymorphic_call_targets): Likewise.
6721 (symbol_table::remove_unreachable_nodes): Likewise.
6722 (ipa_single_use): Likewise.
6723 * passes.c (execute_todo): Likewise.
6724 * predict.c (drop_profile): Likewise.
6725 * symtab.c (symtab_node::get_dump_name): New function.
6726 (symtab_node::dump_name): Likewise.
6727 (symtab_node::dump_asm_name): Likewise.
6728 (symtab_node::dump_references): Likewise.
6729 (symtab_node::dump_referring): Likewise.
6730 (symtab_node::dump_base): Likewise.
6731 (symtab_node::debug_symtab): Likewise.
6732 * tree-sra.c (convert_callers_for_node): Likewise.
6733 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
6734 * value-prof.c (init_node_map): Likewise.
6736 2017-05-23 Martin Liska <mliska@suse.cz>
6738 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
6739 and symtab_node::debug_symtab to symbol_table::debug.
6740 * cgraphunit.c (analyze_functions): Use the renamed function.
6741 (symbol_table::compile): Likewise.
6742 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
6743 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
6744 * passes.c (execute_todo): Likewise.
6745 * symtab.c (symbol_table::dump): New function.
6746 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
6748 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6750 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
6751 that nonconst implies exec.
6753 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6755 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
6756 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
6757 (inline_edge_summary_vec): Turn into ...
6758 (ipa_call_summaries): ... this one.
6759 (redirect_to_unreachable, edge_set_predicate,
6760 evaluate_properties_for_edge, inline_summary_alloc,
6761 reset_ipa_call_summary, reset_inline_summary,
6762 inline_summary_t::duplicate): Update.
6763 (inline_edge_duplication_hook): Turn to ...
6764 (ipa_call_summary_t::duplicate): ... this one.
6765 (inline_edge_removal_hook): Turn to ...
6766 (ipa_call_summary_t::remove): ... this one.
6767 (dump_inline_edge_summary): Turn to ...
6768 (dump_ipa_call_summary): ... this one.
6769 (estimate_function_body_sizes): Update.
6770 (inline_update_callee_summaries): Update.
6771 (remap_edge_change_prob): Update.
6772 (remap_edge_summaries): Update.
6773 (inline_merge_summary): Update.
6774 (do_estimate_edge_time): Update.
6775 (inline_generate_summary): Update.
6776 (inline_read_section): Update.
6777 (inline_read_summary): Update.
6778 (inline_free_summary): Update.
6779 * ipa-inline.c (can_inline_edge_p): Update.
6780 (compute_inlined_call_time): Update.
6781 (want_inline_small_function_p): Update.
6782 (edge_badness): Update.
6783 (early_inliner): Update.
6784 * ipa-inline.h (inline_edge_summary): Turn to ...
6785 (ipa_call_summary): ... this one.
6786 (ipa_call_summary_t): New class.
6787 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
6788 (ipa_call_summaries): New.
6789 (inline_edge_summary): Remove.
6790 (estimate_edge_growth): Update.
6791 * ipa-profile.c (ipa_propagate_frequency_1): Update.
6792 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6793 * ipa-split.c (execute_split_functions): Update.
6794 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
6796 2017-05-23 Tom de Vries <tom@codesourcery.com>
6798 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
6799 attributes): Document rdrand effective target.
6801 2017-05-23 Tom de Vries <tom@codesourcery.com>
6803 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
6804 attributes): Sort alphabetically.
6806 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
6808 * config/avr/genmultilib.awk: Use gsub instead of gensub.
6810 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
6813 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
6814 V2DF/V2DI splat into two separate patterns, one that handles
6815 registers, and the other that only handles memory. Drop support
6816 for splatting from a GPR on ISA 2.07 and then splitting the
6817 splat into direct move and splat.
6818 (vsx_splat_<mode>_reg): Likewise.
6819 (vsx_splat_<mode>_mem): Likewise.
6821 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
6823 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
6825 2017-05-22 Jakub Jelinek <jakub@redhat.com>
6828 * omp-low.c (finish_taskreg_remap): New function.
6829 (finish_taskreg_scan): If unit size of ctx->record_type
6830 is non-constant, unshare the size expression and replace
6831 decls in it with possible outer var refs.
6834 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
6835 GOVD_SHARED rather than GOVD_PRIVATE with it.
6836 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
6837 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
6840 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
6841 as last argument to build_outer_var_ref for pointer bases of array
6844 2017-05-19 Martin Sebor <msebor@redhat.com>
6846 * print-tree.c (print_node): Print DECL_READ_P flag.
6848 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6850 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
6851 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
6852 * cgraph.c: Likewise.
6853 * cgraphunit.c: Likewise.
6854 * gengtype.c: Likewise.
6855 * ipa-cp.c: Likewise.
6856 * ipa-devirt.c: Likewise.
6857 * ipa-icf.c: Likewise.
6858 * ipa-predicate.c: Likewise.
6859 * ipa-profile.c: Likewise.
6860 * ipa-prop.c: Likewise.
6861 * ipa-split.c: Likewise.
6863 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
6864 edge_predicate_pool, dump_inline_hints,
6865 inline_summary::account_size_time, redirect_to_unreachable,
6866 edge_set_predicate, set_hint_predicate,
6867 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
6868 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
6869 inline_summary_t::remove, remap_hint_predicate_after_duplication,
6870 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
6871 ipa_call_summary_t::remove, initialize_growth_caches,
6872 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
6873 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
6874 mark_modified, unmodified_parm_1, unmodified_parm,
6875 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
6876 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
6877 compute_bb_predicates, will_be_nonconstant_expr_predicate,
6878 will_be_nonconstant_predicate, record_modified_bb_info,
6879 get_minimal_bb, record_modified, param_change_prob,
6880 phi_result_unknown_predicate, predicate_for_phi_result,
6881 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
6882 estimate_function_body_sizes, compute_inline_parameters,
6883 compute_inline_parameters_for_curren, pass_data_inline_parameters,
6884 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
6885 inline_update_callee_summaries, remap_edge_change_prob,
6886 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
6887 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
6888 inline_analyze_function, inline_summary_t::insert,
6889 inline_generate_summary, read_ipa_call_summary, inline_read_section,
6890 inline_read_summary, write_ipa_call_summary, inline_write_summary,
6891 inline_free_summary): Move to ipa-fnsummary.h
6892 (predicate_t): Remove.
6893 * ipa-fnsummary.c: New file.
6894 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
6895 (enum inline_hints_vals, inline_hints, agg_position_info,
6896 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
6897 inline_summaries, ipa_call_summary, ipa_call_summary_t,
6898 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
6899 dump_inline_summary, dump_inline_hints, inline_generate_summary,
6900 inline_read_summary, inline_write_summary, inline_free_summary,
6901 inline_analyze_function, initialize_inline_failed,
6902 inline_merge_summary, inline_update_overall_summary,
6903 compute_inline_parameters): Move to ipa-fnsummary.h
6904 * ipa-fnsummary.h: New file.
6905 * ipa-inline-transform.h: Include ipa-inline.h.
6906 * ipa-inline.c: LIkewise.
6908 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6910 * ipa-inline.c (edge_badness): Use inlined_time instead of
6911 inline_summaries->get.
6913 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6915 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
6917 2017-05-22 Nathan Sidwell <nathan@acm.org>
6919 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
6920 (fdump-lang): Document 'raw' option.
6921 * dumpfile.h (TDI_tu): Delete.
6922 * dumpfile.c (dump_files): Remove translation-unit.
6923 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
6925 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
6927 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
6928 command option from $(AWK) call.
6929 * config/avr/genmultilib.awk: Simplify and rewrite so that it
6930 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
6931 [FORMAT]: Remove handling of variable.
6932 * config/avr/t-multilib: Regenerate.
6934 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6936 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
6938 (dump_inline_summary): Do not print self_time.
6939 (estimate_function_body_sizes): Do not set self_time.
6940 (compute_inline_parameters): Likewise.
6941 (inline_read_section, inline_write_summary): Do not stream self_time.
6942 * ipa-inline.h (inline_summary): Drop self_time.
6944 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6946 * ipa-inline-analysis.c (account_size_time): Rename to ...
6947 (inline_summary::account_size_time): ... this one.
6948 (reset_ipa_call_summary): Turn to ...
6949 (ipa_call_summary::reset): ... this one.
6950 (reset_inline_summary): Turn to ...
6951 (inline_summary::reset): ... this one.
6952 (inline_summary_t::remove): Update.
6953 (inline_summary_t::duplicate): Update.
6954 (ipa_call_summary_t::remove): Update.
6955 (dump_inline_summary): Update.
6956 (estimate_function_body_sizes): Update.
6957 (compute_inline_parameters): Update.
6958 (estimate_node_size_and_time): Update.
6959 (inline_merge_summary): Update.
6960 (inline_update_overall_summary): Update.
6961 (inline_read_section): Update.
6962 (inline_write_summary): Update.
6963 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
6964 add account_size_time and reset member functions.
6965 (ipa_call_summary): Add reset function.
6966 * ipa-predicate.h (predicate::operator &): Constify.
6968 2017-05-22 Richard Biener <rguenther@suse.de>
6970 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
6972 2017-05-19 Jason Merrill <jason@redhat.com>
6974 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
6976 2017-05-19 Marek Polacek <polacek@redhat.com>
6979 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
6980 TYPE_OVERFLOW_WRAPS checks.
6982 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
6984 * tree-core.h (enum omp_clause_default_kind): Add
6985 "OMP_CLAUSE_DEFAULT_PRESENT".
6986 * tree-pretty-print.c (dump_omp_clause): Handle it.
6987 * gimplify.c (enum gimplify_omp_var_data): Add
6988 "GOVD_MAP_FORCE_PRESENT".
6989 (gimplify_adjust_omp_clauses_1): Map it to
6990 "GOMP_MAP_FORCE_PRESENT".
6991 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
6993 * gimplify.c (oacc_default_clause): Clarify.
6995 2017-05-19 Nathan Sidwell <nathan@acm.org>
6997 LANG_HOOK_REGISTER_DUMPS
6998 * toplev.c (general_init): Call register dump lang hook.
6999 * doc/invoke.texi: Document -fdump-lang option family.
7000 * dumpfile.c (dump_files): Remove class dump here.
7001 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7002 * dumpfile.h (tree_dump_index): Remove TDI_class.
7003 * langhooks-def.h (lhd_register_dumps): Declare.
7004 (LANG_HOOKS_REGISTER_DUMPS): Define.
7005 (LANG_HOOKS_INITIALIZER): Add it.
7006 * langhooks.c (lhd_register_dumps): Define.
7007 * langhooks.h (struct lang_hooks): Add register_dumps.
7009 2017-05-19 Nathan Sidwell <nathan@acm.org>
7011 * context.h (context::set_passes): New.
7012 * context.c (context::context): Do not create pass manager.
7013 * toplev.c (general_init): Create pass manager here.
7015 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7017 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7018 use this splitter if two add or or instructions would also work for
7019 the constant we want to generate.
7021 2017-05-19 Richard Biener <rguenther@suse.de>
7024 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7025 predicate evaluation.
7027 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7029 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7031 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7034 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7036 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7037 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7038 (inline_edge_summary_vec): Turn into ...
7039 (ipa_call_summaries): ... this one.
7040 (redirect_to_unreachable, edge_set_predicate,
7041 evaluate_properties_for_edge, inline_summary_alloc,
7042 reset_ipa_call_summary, reset_inline_summary,
7043 inline_summary_t::duplicate): Update.
7044 (inline_edge_duplication_hook): Turn to ...
7045 (ipa_call_summary_t::duplicate): ... this one.
7046 (inline_edge_removal_hook): Turn to ...
7047 (ipa_call_summary_t::remove): ... this one.
7048 (dump_inline_edge_summary): Turn to ...
7049 (dump_ipa_call_summary): ... this one.
7050 (estimate_function_body_sizes): Update.
7051 (inline_update_callee_summaries): Update.
7052 (remap_edge_change_prob): Update.
7053 (remap_edge_summaries): Update.
7054 (inline_merge_summary): Update.
7055 (do_estimate_edge_time): Update.
7056 (inline_generate_summary): Update.
7057 (inline_read_section): Update.
7058 (inline_read_summary): Update.
7059 (inline_free_summary): Update.
7060 * ipa-inline.c (can_inline_edge_p): Update.
7061 (compute_inlined_call_time): Update.
7062 (want_inline_small_function_p): Update.
7063 (edge_badness): Update.
7064 (early_inliner): Update.
7065 * ipa-inline.h (inline_edge_summary): Turn to ...
7066 (ipa_call_summary): ... this one.
7067 (ipa_call_summary_t): New class.
7068 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7069 (ipa_call_summaries): New.
7070 (inline_edge_summary): Remove.
7071 (estimate_edge_growth): Update.
7072 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7073 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7074 * ipa-split.c (execute_split_functions): Update.
7075 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7077 2017-05-19 Richard Biener <rguenther@suse.de>
7080 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7082 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7084 * config/rs6000/rs6000.c (struct machine_function): Add field
7085 fpr_is_wrapped_separately.
7086 (rs6000_get_separate_components): Use 64 components. Handle the
7088 (rs6000_components_for_bb): Handle the FPR components.
7089 (rs6000_emit_prologue_components): Handle the FPR components.
7090 (rs6000_emit_epilogue_components): Handle the FPR components.
7091 (rs6000_set_handled_components): Handle the FPR components.
7092 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7093 that are already separately shrink-wrapped.
7094 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7095 that are already separately shrink-wrapped.
7097 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7100 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7103 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7104 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7105 cases where the register allocator uses a move and an offsettable
7106 memory operation to/from a FPR register on ISA 2.06/2.07.
7107 (define_peephole2 for Altivec d-form store): Likewise.
7109 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7112 * config/i386/mmx.md (*mov<mode>_internal): Enable
7113 alternatives 11, 12, 13 and 14 also for 32bit targets.
7114 Remove alternatives 15, 16, 17 and 18.
7115 * config/i386/sse.md (vec_concatv2di): Change
7116 alternative (!x, *y) to (x, ?!*Yn).
7118 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7120 * dumpfile.h (enum dump_kind): Remove stray comma.
7122 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7124 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7125 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7126 predicate::num_conditions
7127 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7128 (CHANGED): turn into predicate::changed.
7129 (agg_position_info): Move to ipa-predicate.h
7130 (add_condition, predicate::add_clause, predicate::operator &=,
7131 predicate::or_with, predicate::evaluate, predicate::probability,
7132 dump_condition, dump_clause, predicate::dump,
7133 predicate::remap_after_duplication, predicate::remap_after_inlining,
7134 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7135 (evaluate_conditions_for_known_args): Update.
7136 (set_cond_stmt_execution_predicate): Update.
7137 * ipa-inline.h: Include ipa-predicate.h
7138 (condition, inline_param_summary, conditions, agg_position_info,
7139 predicate): Move to ipa-predicate.h
7140 * ipa-predicate.c: New file.
7141 * ipa-predicate.h: New file.
7143 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7145 * final.c (leaf_function_p): Check we are not in a sequence.
7147 2017-05-18 Martin Liska <mliska@suse.cz>
7149 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7150 * dumpfile.c (dump_register): Use new enum dump_kind.
7151 (get_dump_file_name): Likewise.
7152 (dump_enable_all): Likewise.
7153 (dump_switch_p_1): Likewise.
7154 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7155 * dumpfile.h (enum dump_kind): New enum type.
7156 (struct dump_file_info): Create constructor and
7157 format fields and comments.
7158 * passes.c (pass_manager::register_one_dump_file):
7160 * statistics.c (statistics_early_init): Likewise.
7161 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7162 TDF_TREE with TDF_SLIM.
7163 (gather_memory_references_ref): Likewise.
7165 2017-05-18 Martin Liska <mliska@suse.cz>
7167 * vec.h (struct vnull): Use it.
7169 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7171 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7172 (true_predicate, false_predicate, true_predicate_p,
7173 false_predicate_p): Remove.
7174 (single_cond_predicate, not_inlined_predicate): Turn to member function
7176 (add_condition): Update.
7177 (add_clause): Turn to...
7178 (predicate::add_clause): ... this one; update; allow passing NULL
7180 (and_predicates): Turn to ...
7181 (predicate::operator &=): ... this one.
7182 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7183 (or_predicates): Turn to ...
7184 (predicate::or_with): ... this one.
7185 (evaluate_predicate): Turn to ...
7186 (predicate::evaluate): ... this one.
7187 (predicate_probability): Turn to ...
7188 (predicate::probability): ... this one.
7189 (dump_condition): Update.
7190 (dump_predicate): Turn to ...
7191 (predicate::dump): ... this one.
7192 (account_size_time): Update.
7193 (edge_set_predicate): Update.
7194 (set_hint_predicate): UPdate.
7195 (evaluate_conditions_for_known_args): Update.
7196 (evaluate_properties_for_edge): Update.
7197 (remap_predicate_after_duplication): Turn to...
7198 (predicate::remap_after_duplication): ... this one.
7199 (remap_hint_predicate_after_duplication): Update.
7200 (inline_summary_t::duplicate): UPdate.
7201 (dump_inline_edge_summary): Update.
7202 (dump_inline_summary): Update.
7203 (set_cond_stmt_execution_predicate): Update.
7204 (set_switch_stmt_execution_predicate): Update.
7205 (compute_bb_predicates): Update.
7206 (will_be_nonconstant_expr_predicate): Update.
7207 (will_be_nonconstant_predicate): Update.
7208 (phi_result_unknown_predicate): Update.
7209 (predicate_for_phi_result): Update.
7210 (array_index_predicate): Update.
7211 (estimate_function_body_sizes): Update.
7212 (estimate_node_size_and_time): Update.
7213 (estimate_ipcp_clone_size_and_time): Update.
7214 (remap_predicate): Rename to ...
7215 (predicate::remap_after_inlining): ... this one.
7216 (remap_hint_predicate): Update.
7217 (inline_merge_summary): Update.
7218 (inline_update_overall_summary): Update.
7219 (estimate_size_after_inlining): Update.
7220 (read_predicate): Rename to ...
7221 (predicate::stream_in): ... this one.
7222 (read_inline_edge_summary): Update.
7223 (write_predicate): Rename to ...
7224 (predicate::stream_out): ... this one.
7225 (write_inline_edge_summary): Update.
7226 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7227 (clause_t): Turn to uint32_t
7228 (predicate): Turn to class; implement constructor and operators
7230 (size_time_entry): Update.
7231 (inline_summary): Update.
7232 (inline_edge_summary): Update.
7234 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7236 * fold-const.c (fold_binary_loc): Move transformation...
7237 * match.pd (C - X CMP X): ... here.
7239 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7241 * config/sparc/sparc.c (sparc_option_override): Set function
7242 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7243 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7245 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7247 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7249 2017-05-18 Marek Polacek <polacek@redhat.com>
7252 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7253 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7254 gimple_assign_load_p.
7256 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7259 * real.c (do_compare): Give decimal_do_compare preference over
7260 comparing just the signs.
7262 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7264 * doc/md.texi (Canonicalization of Instructions): Describe the
7265 canonical form of instructions that inherently set a condition
7268 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7271 * tree-cfg.c: Move deletion of unreachable case statements to after
7272 the merging of consecutive case labels.
7274 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7276 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7277 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7278 restoring of callee-saved registers.
7280 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7282 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7283 * config/visium/visium.c (single_set_and_flags): Likewise.
7284 * config/visium/visium.md (Substitutions): Likewise.
7286 2017-05-17 Martin Liska <mliska@suse.cz>
7288 * cfg.c: Introduce dump_flags_t type and
7289 use it instead of int type.
7291 * cfghooks.c: Likewise.
7292 * cfghooks.h (struct cfg_hooks): Likewise.
7293 * cfgrtl.c: Likewise.
7294 * cfgrtl.h: Likewise.
7295 * cgraph.c (cgraph_node::get_body): Likewise.
7296 * coretypes.h: Likewise.
7297 * domwalk.c: Likewise.
7298 * domwalk.h: Likewise.
7299 * dumpfile.c (struct dump_option_value_info): Likewise.
7300 (dump_enable_all): Likewise.
7301 (dump_switch_p_1): Likewise.
7302 (opt_info_switch_p): Likewise.
7303 * dumpfile.h (enum tree_dump_index): Likewise.
7304 (struct dump_file_info): Likewise.
7305 * genemit.c: Likewise.
7306 * generic-match-head.c: Likewise.
7307 * gengtype.c (open_base_files): Likewise.
7308 * gimple-pretty-print.c: Likewise.
7309 * gimple-pretty-print.h: Likewise.
7310 * graph.c (print_graph_cfg): Likewise.
7311 * graphite-scop-detection.c (dot_all_sese): Likewise.
7312 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7313 * loop-unroll.c (report_unroll): Likewise.
7314 * passes.c (pass_manager::register_one_dump_file): Likewise.
7315 * print-tree.c: Likewise.
7316 * statistics.c: Likewise.
7317 * tree-cfg.c: Likewise.
7318 * tree-cfg.h: Likewise.
7319 * tree-dfa.c: Likewise.
7320 * tree-dfa.h: Likewise.
7321 * tree-dump.c (dump_function): Likewise.
7322 * tree-dump.h (struct dump_info): Likewise.
7323 * tree-pretty-print.c: Likewise.
7324 * tree-pretty-print.h: Likewise.
7325 * tree-ssa-live.c: Likewise.
7326 * tree-ssa-live.h: Likewise.
7327 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7328 * tree-vect-loop.c: Likewise.
7329 * tree-vect-slp.c: Likewise.
7331 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7332 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7334 PR tree-optimization/80457
7335 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7336 of all arguments to a statement as scalar_to_vec operations.
7337 (vectorizable_call): Adjust call to vect_model_simple_cost for
7339 (vectorizable_conversion): Likewise.
7340 (vectorizable_assignment): Likewise.
7341 (vectorizable_shift): Likewise.
7342 (vectorizable_operation): Likewise.
7343 (vectorizable_comparison): Likewise.
7344 (vect_is_simple_cond): Record the def types for operands.
7345 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7346 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7347 for statement argument count.
7349 2017-05-16 Carl Love <cel@us.ibm.com>
7351 * config/rs6000/rs6000-c: Add support for built-in functions
7352 vector unsigned long long vec_bperm (vector unsigned long long,
7353 vector unsigned char)
7354 vector signed long long vec_mule (vector signed int,
7356 vector unsigned long long vec_mule (vector unsigned int,
7357 vector unsigned int)
7358 vector signed long long vec_mulo (vector signed int,
7360 vector unsigned long long vec_mulo (vector unsigned int,
7361 vector unsigned int)
7362 vector signed char vec_sldw (vector signed char,
7365 vector unsigned char vec_sldw (vector unsigned char,
7366 vector unsigned char,
7368 vector signed short vec_sldw (vector signed short,
7369 vector signed short,
7371 vector unsigned short vec_sldw (vector unsigned short,
7372 vector unsigned short,
7374 vector signed int vec_sldw (vector signed int,
7377 vector unsigned int vec_sldw (vector unsigned int,
7378 vector unsigned int,
7380 vector signed long long vec_sldw (vector signed long long,
7381 vector signed long long,
7383 vector unsigned long long vec_sldw (vector unsigned long long,
7384 vector unsigned long long,
7386 * config/rs6000/rs6000-c: Add support for built-in functions
7387 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7388 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7389 * doc/extend.texi: Update the built-in documentation for the
7390 new built-in functions.
7392 2017-05-16 Marek Polacek <polacek@redhat.com>
7396 * tree.c (save_expr): Don't fold the expression.
7398 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7400 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7401 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7402 and (?*y,m). Update insn attributes.
7404 2017-05-16 Martin Liska <mliska@suse.cz>
7406 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7407 flags argument of print_gimple_stmt, print_gimple_expr,
7408 print_generic_stmt and print_generic_expr.
7409 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7410 * coretypes.h: Likewise.
7411 * except.c (dump_eh_tree): Likewise.
7412 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7413 * gimple-pretty-print.h: Likewise.
7414 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
7415 (backprop::push_to_worklist): Likewise.
7416 (backprop::pop_from_worklist): Likewise.
7417 (backprop::process_use): Likewise.
7418 (backprop::intersect_uses): Likewise.
7419 (note_replacement): Likewise.
7420 * gimple-ssa-store-merging.c
7421 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
7422 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7423 (pass_store_merging::execute): Likewise.
7424 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
7425 (ssa_base_cand_dump_callback): Likewise.
7426 (dump_incr_vec): Likewise.
7427 (replace_refs): Likewise.
7428 (replace_mult_candidate): Likewise.
7429 (create_add_on_incoming_edge): Likewise.
7430 (create_phi_basis): Likewise.
7431 (insert_initializers): Likewise.
7432 (all_phi_incrs_profitable): Likewise.
7433 (introduce_cast_before_cand): Likewise.
7434 (replace_one_candidate): Likewise.
7435 * gimplify.c (gimplify_expr): Likewise.
7436 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
7437 (set_rename): Likewise.
7438 (rename_uses): Likewise.
7439 (copy_loop_phi_nodes): Likewise.
7440 (add_close_phis_to_merge_points): Likewise.
7441 (copy_loop_close_phi_args): Likewise.
7442 (copy_cond_phi_args): Likewise.
7443 (graphite_copy_stmts_from_block): Likewise.
7444 (translate_pending_phi_nodes): Likewise.
7445 * graphite-poly.c (print_pdr): Likewise.
7446 (dump_gbb_cases): Likewise.
7447 (dump_gbb_conditions): Likewise.
7448 (print_scop_params): Likewise.
7449 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
7450 (build_cross_bb_scalars_use): Likewise.
7451 (gather_bbs::before_dom_children): Likewise.
7452 * hsa-dump.c (dump_hsa_immed): Likewise.
7453 * ipa-cp.c (print_ipcp_constant_value): Likewise.
7454 (get_replacement_map): Likewise.
7455 * ipa-inline-analysis.c (dump_condition): Likewise.
7456 (estimate_function_body_sizes): Likewise.
7457 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
7458 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
7459 * ipa-prop.c (ipa_dump_param): Likewise.
7460 (ipa_print_node_jump_functions_for_edge): Likewise.
7461 (ipa_modify_call_arguments): Likewise.
7462 (ipa_modify_expr): Likewise.
7463 (ipa_dump_param_adjustments): Likewise.
7464 (ipa_dump_agg_replacement_values): Likewise.
7465 (ipcp_modif_dom_walker::before_dom_children): Likewise.
7466 * ipa-pure-const.c (check_stmt): Likewise.
7467 (pass_nothrow::execute): Likewise.
7468 * ipa-split.c (execute_split_functions): Likewise.
7469 * omp-offload.c (dump_oacc_loop_part): Likewise.
7470 (dump_oacc_loop): Likewise.
7471 * trans-mem.c (tm_log_emit): Likewise.
7472 (tm_memopt_accumulate_memops): Likewise.
7473 (dump_tm_memopt_set): Likewise.
7474 (dump_tm_memopt_transform): Likewise.
7475 * tree-cfg.c (gimple_verify_flow_info): Likewise.
7476 (print_loop): Likewise.
7477 * tree-chkp-opt.c (chkp_print_addr): Likewise.
7478 (chkp_gather_checks_info): Likewise.
7479 (chkp_get_check_result): Likewise.
7480 (chkp_remove_check_if_pass): Likewise.
7481 (chkp_use_outer_bounds_if_possible): Likewise.
7482 (chkp_reduce_bounds_lifetime): Likewise.
7483 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
7484 (chkp_mark_completed_bounds): Likewise.
7485 (chkp_register_incomplete_bounds): Likewise.
7486 (chkp_mark_invalid_bounds): Likewise.
7487 (chkp_maybe_copy_and_register_bounds): Likewise.
7488 (chkp_build_returned_bound): Likewise.
7489 (chkp_get_bound_for_parm): Likewise.
7490 (chkp_build_bndldx): Likewise.
7491 (chkp_get_bounds_by_definition): Likewise.
7492 (chkp_generate_extern_var_bounds): Likewise.
7493 (chkp_get_bounds_for_decl_addr): Likewise.
7494 * tree-chrec.c (chrec_apply): Likewise.
7495 * tree-data-ref.c (dump_data_reference): Likewise.
7496 (dump_subscript): Likewise.
7497 (dump_data_dependence_relation): Likewise.
7498 (analyze_overlapping_iterations): Likewise.
7499 * tree-inline.c (expand_call_inline): Likewise.
7500 (tree_function_versioning): Likewise.
7501 * tree-into-ssa.c (dump_defs_stack): Likewise.
7502 (dump_currdefs): Likewise.
7503 (dump_names_replaced_by): Likewise.
7504 (dump_update_ssa): Likewise.
7505 (update_ssa): Likewise.
7506 * tree-object-size.c (pass_object_sizes::execute): Likewise.
7507 * tree-parloops.c (build_new_reduction): Likewise.
7508 (try_create_reduction_list): Likewise.
7509 (ref_conflicts_with_region): Likewise.
7510 (oacc_entry_exit_ok_1): Likewise.
7511 (oacc_entry_exit_single_gang): Likewise.
7512 * tree-pretty-print.h: Likewise.
7513 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
7514 (get_scalar_evolution): Likewise.
7515 (add_to_evolution): Likewise.
7516 (get_loop_exit_condition): Likewise.
7517 (analyze_evolution_in_loop): Likewise.
7518 (analyze_initial_condition): Likewise.
7519 (analyze_scalar_evolution): Likewise.
7520 (instantiate_scev): Likewise.
7521 (number_of_latch_executions): Likewise.
7522 (gather_chrec_stats): Likewise.
7523 (final_value_replacement_loop): Likewise.
7524 (scev_const_prop): Likewise.
7525 * tree-sra.c (dump_access): Likewise.
7526 (disqualify_candidate): Likewise.
7527 (create_access): Likewise.
7529 (maybe_add_sra_candidate): Likewise.
7530 (create_access_replacement): Likewise.
7531 (analyze_access_subtree): Likewise.
7532 (analyze_all_variable_accesses): Likewise.
7533 (sra_modify_assign): Likewise.
7534 (initialize_constant_pool_replacements): Likewise.
7535 (find_param_candidates): Likewise.
7536 (decide_one_param_reduction): Likewise.
7537 (replace_removed_params_ssa_names): Likewise.
7538 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
7539 * tree-ssa-copy.c (dump_copy_of): Likewise.
7540 (copy_prop_visit_cond_stmt): Likewise.
7541 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
7542 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
7543 (record_equivalences_from_stmt): Likewise.
7544 * tree-ssa-dse.c (compute_trims): Likewise.
7545 (delete_dead_call): Likewise.
7546 (delete_dead_assignment): Likewise.
7547 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
7548 (forward_propagate_into_cond): Likewise.
7549 (pass_forwprop::execute): Likewise.
7550 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7551 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
7553 (move_computations_worker): Likewise.
7554 (execute_sm): Likewise.
7555 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
7556 (remove_exits_and_undefined_stmts): Likewise.
7557 (remove_redundant_iv_tests): Likewise.
7558 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
7559 (adjust_iv_update_pos): Likewise.
7560 * tree-ssa-math-opts.c (bswap_replace): Likewise.
7561 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
7562 (value_replacement): Likewise.
7563 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
7564 * tree-ssa-pre.c (print_pre_expr): Likewise.
7565 (get_representative_for): Likewise.
7566 (create_expression_by_pieces): Likewise.
7567 (insert_into_preds_of_block): Likewise.
7568 (eliminate_insert): Likewise.
7569 (eliminate_dom_walker::before_dom_children): Likewise.
7570 (eliminate): Likewise.
7571 (remove_dead_inserted_code): Likewise.
7572 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
7573 * tree-ssa-reassoc.c (get_rank): Likewise.
7574 (eliminate_duplicate_pair): Likewise.
7575 (eliminate_plus_minus_pair): Likewise.
7576 (eliminate_not_pairs): Likewise.
7577 (undistribute_ops_list): Likewise.
7578 (eliminate_redundant_comparison): Likewise.
7579 (update_range_test): Likewise.
7580 (optimize_range_tests_var_bound): Likewise.
7581 (optimize_vec_cond_expr): Likewise.
7582 (rewrite_expr_tree): Likewise.
7583 (rewrite_expr_tree_parallel): Likewise.
7584 (linearize_expr): Likewise.
7585 (break_up_subtract): Likewise.
7586 (linearize_expr_tree): Likewise.
7587 (attempt_builtin_powi): Likewise.
7588 (attempt_builtin_copysign): Likewise.
7589 (transform_stmt_to_copy): Likewise.
7590 (transform_stmt_to_multiply): Likewise.
7591 (dump_ops_vector): Likewise.
7592 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
7593 (print_scc): Likewise.
7594 (set_ssa_val_to): Likewise.
7595 (visit_reference_op_store): Likewise.
7596 (visit_use): Likewise.
7597 (sccvn_dom_walker::before_dom_children): Likewise.
7598 (run_scc_vn): Likewise.
7599 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7601 (expr_hash_elt::print): Likewise.
7602 (const_and_copies::pop_to_marker): Likewise.
7603 (const_and_copies::record_const_or_copy_raw): Likewise.
7604 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
7605 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
7606 (dump_predicates): Likewise.
7607 (find_uninit_use): Likewise.
7608 (warn_uninitialized_phi): Likewise.
7609 (pass_late_warn_uninitialized::execute): Likewise.
7610 * tree-ssa.c (verify_vssa): Likewise.
7611 (verify_ssa): Likewise.
7612 (maybe_optimize_var): Likewise.
7613 * tree-vrp.c (dump_value_range): Likewise.
7614 (dump_all_value_ranges): Likewise.
7615 (dump_asserts_for): Likewise.
7616 (register_edge_assert_for_2): Likewise.
7617 (vrp_visit_cond_stmt): Likewise.
7618 (vrp_visit_switch_stmt): Likewise.
7619 (vrp_visit_stmt): Likewise.
7620 (vrp_visit_phi_node): Likewise.
7621 (simplify_cond_using_ranges_1): Likewise.
7622 (fold_predicate_in): Likewise.
7623 (evrp_dom_walker::before_dom_children): Likewise.
7624 (evrp_dom_walker::push_value_range): Likewise.
7625 (evrp_dom_walker::pop_value_range): Likewise.
7626 (execute_early_vrp): Likewise.
7628 2017-05-16 Richard Biener <rguenther@suse.de>
7630 * dwarf2out.c (loc_list_from_tree_1): Do not create
7631 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
7633 2017-05-16 Richard Biener <rguenther@suse.de>
7635 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
7637 (note_variable_value_in_expr): If we resolved the decl ref
7638 do not push to the stack.
7640 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
7642 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
7643 operations in fast-math mode.
7644 (vaddq_f16): Likewise.
7645 (vmul_f16): Likewise.
7646 (vmulq_f16): Likewise.
7647 (vsub_f16): Likewise.
7648 (vsubq_f16): Likewise.
7649 * config/arm/neon.md (add<mode>3): New.
7651 (fma:<VH:mode>3): New. Also remove outdated comment.
7654 2017-05-16 Martin Liska <mliska@suse.cz>
7658 * ipa-devirt.c (warn_types_mismatch): Fix typo.
7659 (odr_types_equivalent_p): Likewise.
7661 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
7663 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
7665 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
7668 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
7669 non-interunit SSE move alternatives with '?'.
7670 (zero-extendsidi peephole2): New peephole to skip intermediate
7671 general register in SSE zero-extend sequence.
7673 2017-05-15 Jeff Law <law@redhat.com>
7675 * reorg.c (relax_delay_slots): Create a new variable to hold
7676 the temporary target rather than clobbering TARGET_LABEL.
7678 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
7679 missing argument to extract_bit_field call.
7680 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
7682 2017-05-15 Martin Liska <mliska@suse.cz>
7685 * gcc.c (process_command): Do not allow empty argument of -o option.
7687 2017-05-15 Renlin Li <renlin.li@arm.com>
7689 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
7690 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
7691 * config/aarch64/constraints.md (Usf): Add long call check.
7692 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
7693 (call_value): Likewise.
7694 (sibcall): Likewise.
7695 (sibcall_value): Likewise.
7697 (call_value_insn): New.
7698 (sibcall_insn): Update rtx pattern.
7699 (sibcall_value_insn): Likewise.
7700 (call_internal): Remove.
7701 (call_value_internal): Likewise.
7702 (sibcall_internal): Likewise.
7703 (sibcall_value_internal): Likewise.
7704 (call_reg): Likewise.
7705 (call_symbol): Likewise.
7706 (call_value_reg): Likewise.
7707 (call_value_symbol): Likewise.
7709 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
7712 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
7714 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
7716 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
7717 compatible with CCGOCmode and with CCZmode.
7719 2017-05-14 Martin Sebor <msebor@redhat.com>
7722 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
7723 (gimple_fold_builtin_snprintf): Same.
7724 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
7725 (gimple_fold_builtin_snprintf): Same.
7726 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
7728 (is_call_safe): New function.
7729 (try_substitute_return_value): Call it.
7730 (try_simplify_call): New function.
7731 (pass_sprintf_length::handle_gimple_call): Call it.
7733 2017-05-14 Martin Sebor <msebor@redhat.com>
7736 * builtins.c (expand_builtin_stpncpy): Simplify.
7738 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
7740 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
7741 * config/i386/i386.h
7742 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
7743 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
7744 (struct machine_function): Add new members call_ms2sysv,
7745 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
7746 (struct machine_frame_state): New fields sp_realigned and
7747 sp_realigned_offset.
7748 * config/i386/i386.c
7749 (enum xlogue_stub): New enum.
7750 (enum xlogue_stub_sets): New enum.
7751 (class xlogue_layout): New class.
7752 (struct ix86_frame): New fields stack_realign_allocate_offset,
7753 stack_realign_offset and outlined_save_offset. Modify comments to
7754 detail stack layout when using out-of-line stubs.
7755 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
7756 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
7757 -mcall-ms2sysv-xlogues.
7758 (stub_managed_regs): New static variable.
7759 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
7760 registers managed by out-of-line stub.
7761 (disable_call_ms2sysv_xlogues): New function.
7762 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
7763 m->call_ms2sysv when appropriate and compute frame layout for
7765 (sp_valid_at, fp_valid_at): New inline functions.
7766 (choose_basereg): New function.
7767 (choose_baseaddr): Add align parameter, use choose_basereg and modify
7769 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
7770 Use align parameter of choose_baseaddr to generated aligned SSE movs
7772 (pro_epilogue_adjust_stack): Modify to track
7773 machine_frame_state::sp_realigned.
7774 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
7775 (ix86_nsaved_sseregs): Likewise.
7776 (ix86_emit_save_regs): Likewise.
7777 (ix86_emit_save_regs_using_mov): Likewise.
7778 (ix86_emit_save_sse_regs_using_mov): Likewise.
7779 (get_scratch_register_on_entry): Likewise.
7780 (gen_frame_set): New function.
7781 (gen_frame_load): Likewise.
7782 (gen_frame_store): Likewise.
7783 (emit_outlined_ms2sysv_save): Likewise.
7784 (emit_outlined_ms2sysv_restore): Likewise.
7785 (ix86_expand_prologue): Modify stack re-alignment code and call
7786 emit_outlined_ms2sysv_save when appropriate.
7787 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
7788 parameter rtx_insn *insn, which allows the function to be used to only
7790 (ix86_expand_epilogue): Modify validity checks of frame and stack
7791 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
7792 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
7793 * config/i386/predicates.md
7794 (save_multiple): New predicate.
7795 (restore_multiple): Likewise.
7796 * config/i386/sse.md
7797 (save_multiple<mode>): New pattern.
7798 (save_multiple_realign<mode>): Likewise.
7799 (restore_multiple<mode>): Likewise.
7800 (restore_multiple_and_return<mode>): Likewise.
7801 (restore_multiple_leave_return<mode>): Likewise.
7802 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
7804 2017-05-14 Julia Koval <julia.koval@intel.com>
7806 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
7807 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
7808 (__builtin_ia32_xsetbv): New builtins.
7809 * config/i386/i386.c (ix86_expand_special_args_builtin):
7811 (ix86_expand_builtin): Special expand for new intrinsics.
7812 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
7813 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
7814 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
7816 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7818 * cfganal.c (inverted_post_order_compute): Change argument type
7820 * cfganal.h (inverted_post_order_compute): Adjust prototype.
7821 * df-core.c (rest_of_handle_df_initialize): Adjust.
7822 (rest_of_handle_df_finish): Likewise.
7823 (df_analyze_1): Likewise.
7824 (df_analyze): Likewise.
7825 (loop_inverted_post_order_compute): Change argument to be a vec *.
7826 (df_analyze_loop): Adjust.
7827 (df_get_n_blocks): Likewise.
7828 (df_get_postorder): Likewise.
7829 * df.h (struct df_d): Change field to be a vec.
7830 * lcm.c (compute_laterin): Adjust.
7831 (compute_available): Likewise.
7832 * lra-lives.c (lra_create_live_ranges_1): Likewise.
7833 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
7834 * tree-ssa-pre.c (compute_antic): Likewise.
7836 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7838 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
7839 (depth_first_search::depth_first_search): Change structure init
7840 function to this constructor.
7841 (depth_first_search::add_bb): Rename function to this member.
7842 (depth_first_search::execute): Likewise.
7843 (flow_dfs_compute_reverse_finish): Adjust.
7845 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7847 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
7848 (longest_simple_path): Likewise.
7849 * shrink-wrap.c (spread_components): Likewise.
7850 (disqualify_problematic_components): Likewise.
7851 (emit_common_heads_for_components): Likewise.
7852 (emit_common_tails_for_components): Likewise.
7853 (insert_prologue_epilogue_for_components): Likewise.
7855 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7857 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
7860 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7862 * df-core.c (df_set_blocks): Start using auto_bitmap.
7863 (df_compact_blocks): Likewise.
7864 * df-problems.c (df_rd_confluence_n): Likewise.
7865 * df-scan.c (df_insn_rescan_all): Likewise.
7866 (df_process_deferred_rescans): Likewise.
7867 (df_update_entry_block_defs): Likewise.
7868 (df_update_exit_block_uses): Likewise.
7869 (df_entry_block_bitmap_verify): Likewise.
7870 (df_exit_block_bitmap_verify): Likewise.
7871 (df_scan_verify): Likewise.
7872 * lra-constraints.c (lra_constraints): Likewise.
7873 (undo_optional_reloads): Likewise.
7874 (lra_undo_inheritance): Likewise.
7875 * lra-remat.c (calculate_gen_cands): Likewise.
7876 (do_remat): Likewise.
7877 * lra-spills.c (assign_spill_hard_regs): Likewise.
7878 (spill_pseudos): Likewise.
7879 * tree-ssa-pre.c (bitmap_set_and): Likewise.
7880 (bitmap_set_subtract_values): Likewise.
7882 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7884 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
7885 management with auto_bitmap.
7886 (fix_inter_tick): Likewise.
7887 (fix_recovery_deps): Likewise.
7888 * ira.c (add_store_equivs): Likewise.
7889 (find_moveable_pseudos): Likewise.
7890 (split_live_ranges_for_shrink_wrap): Likewise.
7891 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
7892 (rtx_reuse_manager::seen_def_p): Likewise.
7893 (rtx_reuse_manager::set_seen_def): Likewise.
7894 * print-rtl.h (class rtx_reuse_manager): Likewise.
7896 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7898 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
7900 (migrate_btr_def): Likewise.
7901 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
7902 * df-core.c (loop_post_order_compute): Likewise.
7903 (loop_inverted_post_order_compute): Likewise.
7904 * hsa-common.h: Likewise.
7905 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
7906 * init-regs.c (initialize_uninitialized_regs): Likewise.
7907 * ipa-inline.c (resolve_noninline_speculation): Likewise.
7908 (inline_small_functions): Likewise.
7909 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
7910 * ira.c (combine_and_move_insns): Likewise.
7911 (build_insn_chain): Likewise.
7912 * loop-invariant.c (find_invariants): Likewise.
7913 * lower-subreg.c (propagate_pseudo_copies): Likewise.
7914 * predict.c (tree_predict_by_opcode): Likewise.
7915 (predict_paths_leading_to): Likewise.
7916 (predict_paths_leading_to_edge): Likewise.
7917 (estimate_loops_at_level): Likewise.
7918 (estimate_loops): Likewise.
7919 * shrink-wrap.c (try_shrink_wrapping): Likewise.
7920 (spread_components): Likewise.
7921 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
7922 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
7923 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
7924 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
7925 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
7926 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
7927 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
7928 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
7929 (mark_threaded_blocks): Likewise.
7930 (thread_through_all_blocks): Likewise.
7931 * tree-ssa.c (verify_ssa): Likewise.
7932 (execute_update_addresses_taken): Likewise.
7933 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
7935 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7937 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
7939 (post_order_compute): Likewise.
7940 (inverted_post_order_compute): Likewise.
7941 (pre_and_rev_post_order_compute_fn): Likewise.
7943 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7945 * genrecog.c (int_set::int_set): Explicitly construct our
7946 auto_vec base class.
7947 * vec.h (auto_vec::auto_vec): New constructor.
7949 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7951 * bitmap.h (class auto_bitmap): New constructor taking
7952 bitmap_obstack * argument.
7954 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7956 * bitmap.h (class auto_bitmap): Change type of m_bits to
7957 bitmap_head, and adjust ctor / dtor and member operators.
7959 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
7961 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
7962 when returned register mode doesn't match original mode.
7964 2017-05-12 Jeff Law <law@redhat.com>
7965 Jakub Jelinek <jakub@redhat.com>
7967 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
7968 we look for cc setter after the compare-elim changes.
7969 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
7970 within the vector to match what compare-elim now expects.
7971 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
7972 (xorsi3_flags, one_cmplsi2_flags): Likewise.
7974 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
7975 after the compare-elim changes.
7976 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
7977 the vector to match what compare-elim now expects.
7978 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
7979 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
7980 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
7981 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
7982 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
7984 * config/visium/visium.c (single_set_and_flags): Fix where
7985 we look for cc setter after the compare-elim changes.
7986 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
7987 with the vector to match what compare-elim now expects.
7988 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
7989 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
7990 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
7991 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
7992 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
7993 (neg<mode>2_insn_set_overflow): Likewise.
7995 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
7998 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
7999 maybe_expand_insn call, set ops[0].target. If still set after call,
8000 set alt_rtl. Add extra arg to recursive calls.
8001 (extract_bit_field): Add alt_rtl argument. Pass to
8003 * expmed.h (extract_bit_field): Fix prototype.
8004 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8005 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8006 to extract_bit_field_calls.
8007 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8008 Pass alt_rtl to extract_bit_field calls.
8009 * calls.c (store_unaligned_arguments_into_psuedos)
8010 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8011 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8013 * optabs.h (struct expand_operand): Add target bitfield.
8015 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8017 * compare-elim.c (try_eliminate_compare): Canonicalize
8018 operation with embedded compare to
8019 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8020 (set (reg) (operation)].
8022 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8024 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8027 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8028 cost of adding a carry flag for ADC instruction.
8029 [case MINUS]: Ignore the cost of subtracting a carry flag
8030 for SBB instruction.
8032 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8034 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8036 * config/rs6000/bmiintrin.h: New file.
8037 * config/rs6000/bmi2intrin.h: New file.
8038 * config/rs6000/x86intrin.h: New file.
8040 2017-05-12 Jeff Law <law@redhat.com>
8042 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8045 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8048 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8050 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8052 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8053 early expansion of vector divide builtins.
8054 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8055 builtins identified as having unsigned arguments.
8057 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8059 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8060 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8061 expansion of vector logical operations (and, andc, or, xor,
8064 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8066 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8067 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8069 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8071 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8072 early GIMPLE expansion of vector multiplies.
8074 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8076 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8077 TARGET_HAVE_MOVT conditional.
8078 (movt splitter): Likewise.
8080 2017-05-12 Richard Biener <rguenther@suse.de>
8082 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8083 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8084 Fold all stmts not inplace.
8086 2017-05-12 Richard Biener <rguenther@suse.de>
8088 PR tree-optimization/80713
8089 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8090 inserted_exprs bit for not removed stmts.
8092 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8095 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8096 parallelized" attribute for parallelized OpenACC kernels.
8097 * omp-offload.c (execute_oacc_device_lower): Use it.
8099 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8100 Set "oacc kernels" attribute.
8101 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8102 parameter. Adjust all users.
8103 (oacc_fn_attrib_kernels_p): Remove function.
8104 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8105 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8106 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8107 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8108 assert "oacc kernels" attribute is set.
8110 2017-05-11 Carl Love <cel@us.ibm.com>
8112 * config/rs6000/rs6000-c: Add support for built-in functions
8113 vector unsigned char vec_popcnt (vector signed char)
8114 vector unsigned char vec_popcnt (vector unsigned char)
8115 vector unsigned short vec_popcnt (vector signed short)
8116 vector unsigned short vec_popcnt (vector unsigned short)
8117 vector unsigned int vec_popcnt (vector signed int)
8118 vector unsigned int vec_popcnt (vector unsigned int)
8119 vector unsigned long long vec_popcnt (vector signed long long)
8120 vector unsigned long long vec_popcnt (vector unsigned long long)
8121 vector signed long long vec_slo (vector signed long long,
8123 vector signed long long vec_slo (vector signed long long,
8124 vector unsigned char)
8125 vector unsigned long long vec_slo (vector unsigned long long,
8127 vector unsigned long long vec_slo (vector unsigned long long,
8128 vector unsigned char)
8129 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8130 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8131 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8132 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8133 * doc/extend.texi: Update the built-in documentation file for the
8134 new built-in functions.
8136 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8138 * attribs.h (sorted_attr_string): Move machine independent
8139 functions for target clone support from the i386 port to common
8140 code. Rename ix86_function_versions to common_function_versions.
8141 Rename make_name to make_unique_name.
8142 (common_function_versions): Likewise.
8143 (make_unique_name): Likewise.
8144 (make_dispatcher_decl): Likewise.
8145 (is_function_default_version): Likewise.
8146 * attribs.c (attr_strcmp): Likewise.
8147 (sorted_attr_string): Likewise.
8148 (common_function_versions): Likewise.
8149 (make_unique_name): Likewise.
8150 (make_dispatcher_decl): Likewise.
8151 (is_function_default_version): Likewise.
8152 * config/i386/i386.c (attr_strcmp): Likewise.
8153 (sorted_attr_string): Likewise.
8154 (ix86_function_versions): Likewise.
8155 (make_name): Likewise.
8156 (make_dispatcher_decl): Likewise.
8157 (is_function_default_version): Likewise.
8158 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8160 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8163 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8164 Account for direct move costs for vec_construct of integer
8167 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8170 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8171 (UNSPEC_STX_ATOMIC): Ditto.
8172 (loaddi_via_sse): New insn.
8173 (storedi_via_sse): Ditto.
8174 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8175 Update corresponding peephole2 patterns.
8176 (atomic_storedi_fpu): Ditto.
8178 2017-05-11 Julia Koval <julia.koval@intel.com>
8180 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8181 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8183 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8184 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8185 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8187 2017-05-11 Nathan Sidwell <nathan@acm.org>
8189 * graphite-poly.c: Include dumpfile.h.
8191 * dumpfle.h (dump_function): Declare here ...
8192 * tree-dump.h (dump_function): ... not here.
8193 * dumpfile.c: #include tree-cfg.h.
8194 (dump_function): Move here from ...
8195 * tree-dump.c (dump_function): ... here.
8196 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8197 * graphite-poly.c: Don't include tree-dump.h.
8198 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8199 * print-tree.c: Likewise.
8200 * stor-layout.c: Likewise.
8201 * tree-nested.c: Likewise.
8203 * dumpfile.c (dump_start): Use TDF_FLAGS.
8204 (dump_enable_all): Fix TDF_KIND check thinko.
8206 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8208 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8209 array entries to represent two legal parameterizations of the
8210 overloaded __builtin_cmpb function, as represented by the
8211 P6_OV_BUILTIN_CMPB constant.
8212 (altivec_resolve_overloaded_builtin): Add special case handling
8213 for the __builtin_cmpb function, as represented by the
8214 P6_OV_BUILTIN_CMPB constant.
8215 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8216 (BU_P6_64BIT_2): New macro.
8217 (BU_P6_OVERLOAD_2): New macro
8218 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8219 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8220 (CMPB): Add overload support to represent both 32-bit and 64-bit
8221 compare-bytes function.
8222 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8223 support for TARGET_CMPB.
8224 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8225 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8226 documentation of the __builtin_cmpb overloaded built-in function.
8228 2017-05-11 Richard Biener <rguenther@suse.de>
8230 PR tree-optimization/80705
8231 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8232 bases are not vectorizable.
8234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8236 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8237 when counting register pressure.
8239 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8241 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8242 (iv_ca_more_deps): Renamed to ...
8243 (iv_ca_compare_deps): ... this.
8244 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8246 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8248 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8250 (determine_group_iv_costs): ... here.
8251 (find_inv_vars_cb): Record inv var if it's not recorded before.
8253 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8255 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8256 (get_shiftadd_cost): Ditto.
8258 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8260 * tree-ssa-address.c: Include header file.
8261 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8263 (add_to_parts): Refactor.
8264 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8265 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8268 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8270 PR tree-optimization/53090
8271 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8273 (extract_cond_operands): Detect condition with IV on both sides
8274 and return COMP_IV_EXPR_2.
8275 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8276 (rewrite_use_compare): Simplify by removing call to function
8277 extract_cond_operands.
8279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8281 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8282 (extract_cond_operands): Detect condition comparing against non-
8283 invariant bound and return appropriate enum value.
8284 (find_interesting_uses_cond): Update use of extract_cond_operands.
8285 Handle its return value accordingly.
8286 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8288 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8290 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8291 nonlinear iv_use computation in loop invariant sensitive way.
8293 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8295 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8296 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8300 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8301 (dump_cand): Support iv_cand.inv_exprs.
8302 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8304 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8309 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8311 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8312 as local function. Include necessary header files.
8313 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8315 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8317 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8319 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8321 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8322 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8323 RSHIFT_EXPR and BIT_NOT_EXPR.
8325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8327 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8328 (adjust_setup_cost): New parameter supporting round up adjustment.
8329 (struct address_cost_data): Delete.
8330 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8331 (split_address_cost, ptr_difference_cost): Delete.
8332 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8333 (struct ainc_cost_data): New struct.
8334 (get_address_cost_ainc): New function.
8335 (get_address_cost, get_computation_cost): Reimplement.
8336 (determine_group_iv_cost_address): Record inv_expr for all uses of
8338 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8339 (iv_ca_has_deps): Reimplemented to ...
8340 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8342 (iv_ca_extend): Call iv_ca_more_deps.
8344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8346 * tree-ssa-address.c (struct mem_address): Move to header file.
8347 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8348 * tree-ssa-address.h (struct mem_address): Move from C file.
8349 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8351 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8353 * tree-affine.h (aff_combination_type): New interface.
8354 (aff_combination_zero_p): Remove static.
8355 (aff_combination_const_p): New interface.
8356 (aff_combination_singleton_var_p): New interfaces.
8358 2017-05-11 Richard Biener <rguenther@suse.de>
8360 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8361 Skip unreachable blocks and destinations.
8362 (eliminate): Move stmt removal and fixup ...
8363 (fini_eliminate): ... here. Skip inserted exprs.
8364 (pass_pre::execute): Move fini_pre after fini_eliminate.
8365 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8366 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8367 PRE to get rid of dead code that has invalid SSA form and
8368 split critical edges again.
8370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8372 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8374 2017-05-11 Richard Biener <rguenther@suse.de>
8376 * passes.c (execute_function_todo): Verify loops if they are
8377 said to be up-to-date.
8378 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8379 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8381 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8384 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8385 handle calling assemble_external ourself.
8388 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8389 modes with zero size. Enhance comment.
8391 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8393 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8394 built-ins for vec_xl and vec_xst with short and char pointer
8397 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8399 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8400 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8401 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8402 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8403 (_mm_maskz_min_round_ss): New intrinsics.
8404 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8405 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8406 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8407 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8408 (__builtin_ia32_minss_mask_round): New builtins.
8409 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8410 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8411 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
8413 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
8414 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
8416 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
8419 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8421 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
8422 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
8423 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
8424 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
8425 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
8426 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
8427 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
8428 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8429 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8430 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
8431 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
8432 (__builtin_ia32_mulss_mask_round): New builtins.
8433 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8434 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8435 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
8437 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
8438 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8440 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8443 2017-05-10 Julia Koval <julia.koval@intel.com>
8445 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
8446 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
8447 (_mm256_setr_m128i): New intrinsics.
8449 2017-05-10 Julia Koval <julia.koval@intel.com>
8451 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
8452 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
8453 (_mm_maskz_rcp14_ss): New intrinsics.
8454 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
8455 (__builtin_ia32_rcp14ss_mask): New builtins.
8456 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
8458 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
8460 PR tree-optimization/51513
8461 * tree-cfg.c (gimple_seq_unreachable_p): New function.
8462 (assert_unreachable_fallthru_edge_p): Use it.
8463 (group_case_labels_stmt): Likewise.
8464 * tree-cfg.h: Prototype it.
8465 * stmt.c: Include cfghooks.h and tree-cfg.h.
8466 (emit_case_dispatch_table) <gap_label>: New local variable.
8467 Use it to fill dispatch table gaps.
8468 Test for default_label before updating probabilities.
8469 (expand_case) <default_label>: Remove unneeded initialization.
8470 Test for unreachable default case statement and remove its edge.
8471 Set default_label accordingly.
8472 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
8474 2017-05-10 Carl Love <cel@us.ibm.com>
8476 * config/rs6000/rs6000-c: Add support for built-in functions
8477 vector signed char vec_neg (vector signed char)
8478 vector signed short int vec_neg (vector short int)
8479 vector signed int vec_neg (vector signed int)
8480 vector signed long long vec_neg (vector signed long long)
8481 vector float vec_neg (vector float)
8482 vector double vec_neg (vector double)
8483 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
8485 * config/rs6000/altivec.h: Add define for vec_neg
8486 * doc/extend.texi: Update the built-in documentation for the
8487 new built-in functions.
8489 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8491 PR tree-optimization/77644
8492 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
8494 2017-05-10 Nathan Sidwell <nathan@acm.org>
8496 * dumpfile.h (TDI_lang_all): New.
8497 (TDF_KIND): New. Renumber others
8498 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
8500 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
8502 (get_dump_file_name): Adjust suffix generation.
8503 (dump_enable_all): Use TDF_KIND.
8504 * doc/invoke.texi (-fdump-lang-all): Document.
8506 * dumpfile.h: Tabify.
8508 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
8511 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
8512 Move member access before delete.
8514 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
8516 * tree-inline.c (expand_call_inline): Split block at stmt
8519 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8522 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
8523 are now unused after splitting mov{sf,sd}_hardfloat.
8524 (f32_lr2): Likewise.
8526 (f32_lm2): Likewise.
8528 (f32_li2): Likewise.
8531 (f32_sr2): Likewise.
8533 (f32_sm2): Likewise.
8535 (f32_si2): Likewise.
8538 (f32_vsx): Likewise.
8540 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
8541 For movsf, order stores so the VSX stores occur before the GPR
8542 store which encourages the register allocator to use a traditional
8543 FPR instead of a GPR. For movsd, order the stores so that the GPR
8544 store comes before the VSX stores to allow the power6 to work.
8545 This is due to the power6 not having a 32-bit integer store
8546 instruction from a FPR.
8547 (movsf_hardfloat): Likewise.
8548 (movsd_hardfloat): Likewise.
8550 2017-05-09 Martin Sebor <msebor@redhat.com>
8552 PR translation/80280
8553 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
8556 PR translation/80280
8557 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
8558 data member added in r247778.
8559 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
8561 2017-05-09 Nathan Sidwell <nathan@acm.org>
8563 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
8565 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
8568 2017-05-09 Marek Polacek <polacek@redhat.com>
8570 * doc/invoke.texi: Fix typo.
8572 2017-05-09 Richard Biener <rguenther@suse.de>
8574 * tree-vrp.c (vrp_val_is_max): Adjust comment.
8575 (vrp_val_is_min): Likewise.
8576 (set_value_range_to_value): Likewise.
8577 (set_value_range_to_nonnegative): Likewise.
8578 (gimple_assign_nonzero_p): Likewise.
8579 (gimple_stmt_nonzero_p): Likewise.
8580 (vrp_int_const_binop): Likewise. Remove unreachable case.
8581 (adjust_range_with_scev): Adjust comments.
8582 (compare_range_with_value): Likewise.
8583 (extract_range_from_phi_node): Likewise.
8584 (test_for_singularity): Likewise.
8586 2017-05-09 Richard Biener <rguenther@suse.de>
8588 * tree-vrp.c (get_single_symbol): Add assert that we don't
8589 get overflowed constants as invariant part.
8590 (compare_values_warnv): Add comment before the TREE_NO_WARNING
8591 checks. Use wi::cmp instead of recursing for integer constants.
8592 (compare_values): Just ignore whether we assumed undefined
8593 overflow instead of failing the compare.
8594 (extract_range_for_var_from_comparison_expr): Add comment before the
8595 TREE_NO_WARNING sets.
8596 (test_for_singularity): Likewise.
8597 (extract_range_from_comparison): Do not disable optimization
8598 when we assumed undefined overflow.
8599 (extract_range_basic): Remove init of unused var.
8601 2017-05-09 Richard Biener <rguenther@suse.de>
8603 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
8604 (extract_range_from_multiplicative_op_1): Adjust.
8605 (extract_range_from_binary_expr_1): Use int_const_binop.
8607 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
8610 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
8611 rs6000_store_data_bypass_p in seven define_bypass directives and
8612 in several comments.
8613 * config/rs6000/rs6000-protos.h: Add prototype for
8614 rs6000_store_data_bypass_p function.
8615 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
8616 function implements slightly different (rs6000-specific) semantics
8617 than store_data_bypass_p, returning false rather than aborting
8618 with assertion error when arguments do not satisfy the
8619 requirements of store data bypass.
8620 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
8621 rs6000_store_data_bypass_p.
8623 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
8625 * config/xtensa/xtensa-protos.h
8626 (xtensa_initial_elimination_offset): New declaration.
8627 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
8628 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
8629 macro definition, add case for FRAME_POINTER_REGNUM when
8630 FRAME_GROWS_DOWNWARD.
8631 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
8632 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
8633 xtensa_initial_elimination_offset.
8635 2017-05-08 Nathan Sidwell <nathan@acm.org>
8637 * doc/invoke.texi: Alphabetize -fdump options.
8639 2017-05-08 Martin Sebor <msebor@redhat.com>
8641 PR translation/80280
8642 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
8644 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
8646 * target.def (compute_frame_layout): New optional target hook.
8647 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
8648 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
8649 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
8651 * reload1.c (verify_initial_elim_offsets): Likewise.
8652 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
8653 (use_simple_return_p): Call arm_compute_frame_layout if needed.
8654 (arm_get_frame_offsets): Split up into this ...
8655 (arm_compute_frame_layout): ... and this function.
8657 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
8659 * config/aarch64/constraints.md (Usa): New constraint.
8660 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
8662 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
8664 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
8665 with_multilib_list after it has been checked.
8667 2017-05-08 Richard Biener <rguenther@suse.de>
8669 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
8670 (bitmap_set_subtract_values): Likewise.
8672 2017-05-08 Richard Biener <rguenther@suse.de>
8674 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
8675 (gimple_assign_nonzero): ... this and remove strict_overflow_p
8677 (gimple_stmt_nonzero_warnv_p): Rename to ...
8678 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
8680 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
8681 (extract_range_basic): Adjust, do not disable propagation on
8682 strict overflow sensitive simplification.
8683 (vrp_visit_cond_stmt): Likewise.
8685 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
8687 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
8688 body size unconditionally.
8690 2017-05-07 Jeff Law <law@redhat.com>
8693 2017-05-06 Jeff Law <law@redhat.com>
8694 PR tree-optimization/78496
8695 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
8698 PR tree-optimization/78496
8699 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
8700 (simplify_stmt_using_ranges): Call it.
8701 (vrp_dom_walker::before_dom_children): Extract equivalences
8702 from an ASSERT_EXPR with an equality comparison against a
8705 2017-05-06 Jeff Law <law@redhat.com>
8707 PR tree-optimization/78496
8708 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
8711 PR tree-optimization/78496
8712 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
8713 (simplify_stmt_using_ranges): Call it.
8714 (vrp_dom_walker::before_dom_children): Extract equivalences
8715 from an ASSERT_EXPR with an equality comparison against a
8718 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
8720 * lra-constraints.c (lra_copy_reg_equiv): New function.
8721 (split_reg): Use it to copy equivalence information from the
8722 original register to the spill register.
8724 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
8726 PR rtl-optimization/75964
8727 * simplify-rtx.c (simplify_const_relational_operation): Remove
8728 invalid handling of comparisons of integer ABS.
8730 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
8732 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
8734 (init_regs): Remove declaration.
8735 (function_arg_advance_32): Initialize error_p as boolean variable.
8737 2017-05-05 Nathan Sidwell <nathan@acm.org>
8739 * store-motion.c (remove_reachable_equiv_notes): Reformat long
8740 lines. Use for (;;).
8742 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8744 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
8745 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
8746 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
8747 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
8748 VF=2 that require versioning.
8750 2017-05-05 David Malcolm <dmalcolm@redhat.com>
8752 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
8755 2017-05-05 David Malcolm <dmalcolm@redhat.com>
8757 * diagnostic.h (diagnostic_override_option_index): Convert from
8758 macro to inline function.
8760 2017-05-05 David Malcolm <dmalcolm@redhat.com>
8762 * diagnostic.c (last_module_changed_p): New function.
8763 (set_last_module): New function.
8764 (diagnostic_report_current_module): Convert macro usage to
8765 the above functions.
8766 * diagnostic.h (diagnostic_context::last_module): Strengthen
8767 from const line_map * to const line_map_ordinary *.
8768 (diagnostic_last_module_changed): Delete macro.
8769 (diagnostic_set_last_module): Delete macro.
8771 2017-05-05 David Malcolm <dmalcolm@redhat.com>
8773 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
8774 with diagnostic_report_diagnostic.
8775 (diagnostic_n_impl_richloc): Likewise.
8776 * diagnostic.h (report_diagnostic): Delete macro.
8777 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
8778 with diagnostic_report_diagnostic.
8779 * substring-locations.c (format_warning_va): Likewise.
8781 2017-05-05 David Malcolm <dmalcolm@redhat.com>
8783 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
8784 save/restor of format_spec. Move option-printing code to...
8785 (print_option_information): ...this new function, and
8786 reimplement by simply printing to the pretty_printer,
8787 rather than appending to the format string.
8789 2017-05-05 David Malcolm <dmalcolm@redhat.com>
8791 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
8792 handling logic into...
8793 (update_effective_level_from_pragmas): ...this new function.
8795 2017-05-04 Andrew Waterman <andrew@sifive.com>
8797 * config/riscv/riscv.opt (mstrict-align): New option.
8798 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
8799 (SLOW_UNALIGNED_ACCESS): Define.
8800 (riscv_slow_unaligned_access): Declare.
8801 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
8803 (riscv_slow_unaligned_access): New variable.
8804 (rocket_tune_info): Set slow_unaligned_access to true.
8805 (optimize_size_tune_info): Set slow_unaligned_access to false.
8806 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
8807 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
8808 (riscv_option_override): Set riscv_slow_unaligned_access.
8809 * doc/invoke.texi: Add -mstrict-align to RISC-V.
8811 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
8813 * config/riscv/riscv.md: Unify indentation.
8815 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
8820 * config/rs6000/rs6000.md (u code attribute): Add FIX and
8822 (extendsi<mode>2): Add support for doing sign extension via
8823 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
8824 don't have ISA 3.0 instructions.
8825 (extendsi<mode>2 splitter): Likewise.
8826 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
8827 generate the normal insns since SImode can now go in vector
8828 registers. Disallow the special UNSPECs needed for previous
8829 machines to hide SImode being used. Add new insns
8830 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
8831 (fix_trunc<mode>si2_stfiwx): Likewise.
8832 (fix_trunc<mode>si2_internal): Likewise.
8833 (fixuns_trunc<mode>si2): Likewise.
8834 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8835 (fctiw<u>z_<mode>_smallint): Likewise.
8836 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
8837 of floating point to 32-bit integer from doing a direct move to
8838 the GPR registers to do a store.
8839 (fctiwz_<mode>): Break long line.
8841 2017-05-05 Bin Cheng <bin.cheng@arm.com>
8843 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
8844 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
8845 (addr_list, addr_offset_valid_p): New.
8846 (split_address_groups): Check offset validity with above function.
8847 (gt-tree-ssa-loop-ivopts.h): Include header file.
8849 2017-05-05 Nathan Sidwell <nathan@acm.org>
8851 * config.gcc (arm*-*-*): Add missing 'fi'.
8853 2017-05-05 Steve Ellcey <sellcey@cavium.com>
8855 * doc/invoke.texi (-fopt-info): Explicitly say order of options
8856 included in -fopt-info does not matter.
8857 * doc/optinfo.texi (-fopt-info): Fix description of default
8858 behavour. Explicitly say order of options included in -fopt-info
8861 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
8863 * config.gcc: Allow combinations of aprofile and rmprofile values for
8864 --with-multilib-list.
8865 * config/arm/t-multilib: New file.
8866 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
8867 variables. Remove setting of ISA and floating-point ABI in
8868 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
8869 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
8870 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
8871 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
8873 * config/arm/t-rmprofile: Likewise except for the matches changes.
8874 * doc/install.texi (--with-multilib-list): Document the combination of
8875 aprofile and rmprofile values and warn about pitfalls in doing that.
8877 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
8879 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
8880 (movdi_aarch64): Likewise.
8882 2017-05-05 Jakub Jelinek <jakub@redhat.com>
8884 PR tree-optimization/80632
8885 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
8887 (build_arrays): Initialize it for virtual phis.
8888 (fix_phi_nodes): Use it for virtual phis.
8890 PR tree-optimization/80558
8891 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
8892 [x, y] op z into [x op, y op z] for op & or | if conditions
8895 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
8896 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8899 * config/arm/arm.md (use_literal_pool): Remove.
8900 (64-bit immediate split): No longer takes cost into consideration
8901 if arm_disable_literal_pool is enabled.
8902 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
8903 used when arm_disable_literal_pool is enabled.
8904 (arm_max_const_double_inline_cost): Remove use of
8905 arm_disable_literal_pool.
8906 (push_minipool_fix): Add assert.
8907 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
8908 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
8909 (no_literal_pool_sf_immediate): New.
8911 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8913 PR tree-optimization/80613
8914 * tree-ssa-dce.c (propagate_necessity): Remove cases for
8915 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
8917 2017-05-05 Richard Biener <rguenther@suse.de>
8919 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
8921 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
8923 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
8924 of this flag from insn conditions due to removal from r247495.
8926 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
8928 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
8930 (arm_early_store_addr_dep_ptr): Likewise.
8931 * config/arm/aarch-common-protos.h
8932 (arm_early_load_addr_dep_ptr): Add prototype.
8933 (arm_early_store_addr_dep_ptr): Likewise.
8934 * config/arm/cortex-a53.md: Add new bypasses.
8936 2017-05-05 Jakub Jelinek <jakub@redhat.com>
8938 * tree.c (next_type_uid): Change type to unsigned.
8939 (type_hash_canon): Decrement back next_type_uid if
8940 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
8941 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
8944 2017-05-04 Martin Sebor <msebor@redhat.com>
8946 * builtins.c: Fix a trivial typo in a comment.
8949 * builtins.c (check_sizes): Adjust to handle reading past the end.
8950 Avoid printing excessive upper bound of ranges. Use %E to print
8951 tree nodes instead of converting them to %wu.
8952 (expand_builtin_memchr): New function.
8953 (compute_dest_size): Rename...
8954 (compute_objsize): ...to this.
8955 (expand_builtin_memcpy): Adjust.
8956 (expand_builtin_mempcpy): Adjust.
8957 (expand_builtin_strcat): Adjust.
8958 (expand_builtin_strcpy): Adjust.
8959 (check_strncat_sizes): Adjust.
8960 (expand_builtin_strncat): Adjust.
8961 (expand_builtin_strncpy): Adjust and simplify.
8962 (expand_builtin_memset): Adjust.
8963 (expand_builtin_bzero): Adjust.
8964 (expand_builtin_memcmp): Adjust.
8965 (expand_builtin): Handle memcmp.
8966 (maybe_emit_chk_warning): Check strncat just once.
8968 2017-05-04 Martin Sebor <msebor@redhat.com>
8970 PR preprocessor/79214
8973 * builtins.c (check_sizes): Add inlining context and issue
8974 warnings even when -Wno-system-headers is set.
8975 (check_strncat_sizes): Same.
8976 (expand_builtin_strncat): Same.
8977 (expand_builtin_memmove): New function.
8978 (expand_builtin_stpncpy): Same.
8979 (expand_builtin): Handle memmove and stpncpy.
8981 2017-05-04 Bin Cheng <bin.cheng@arm.com>
8983 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
8984 which is not used any more.
8986 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
8988 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
8990 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
8992 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
8993 (cortexa53_tunings): Likewise.
8994 (cortexa57_tunings): Likewise.
8995 (cortexa72_tunings): Likewise.
8996 (cortexa73_tunings): Likewise.
8998 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9000 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9001 Set loop alignment to 8.
9003 2017-05-04 Martin Sebor <msebor@redhat.com>
9005 PR translation/80280
9006 * builtins.c (expand_builtin_object_size): Add missing quoting to
9007 %D and like directives.
9008 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9009 (hsa_type_for_tree_type): Same.
9010 (verify_function_arguments): Same.
9011 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9012 * varasm.c (get_section): Same.
9015 2017-05-04 Martin Sebor <msebor@redhat.com>
9017 PR translation/80280
9018 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9020 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9022 * config/aarch64/aarch64.c (generic_addrcost_table):
9023 Change HI/TI mode setting.
9025 2017-05-04 Martin Jambor <mjambor@suse.cz>
9027 PR tree-optimization/80622
9028 * tree-sra.c (comes_initialized_p): New function.
9029 (build_accesses_from_assign): Only set write lazily when
9030 comes_initialized_p is false.
9031 (analyze_access_subtree): Use comes_initialized_p.
9032 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9033 instead of testing for PARM_DECL.
9035 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9037 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9038 constraint on operand 0 to allow more general addressing modes.
9039 Adjust output template.
9040 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9042 * config/aarch64/aarch64-protos.h
9043 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9044 * config/aarch64/constraints.md (Dp): New address constraint.
9045 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9048 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9050 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9051 update use of estimate_ipcp_clone_size_and_time.
9052 (estimate_local_effects): Update use of
9053 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9054 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9055 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9056 Return nonspecialized time.
9058 2017-05-04 Richard Biener <rguenther@suse.de>
9060 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9061 for the last VUSE which def dominates the PHI. Directly call
9063 (get_continuation_for_phi_1): Remove.
9065 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9067 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9068 to explain the use of truncating division. Cap the number of
9069 iterations to the maximum given by nb_iterations_upper_bound,
9072 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9074 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9075 * configure: Regenerate.
9076 * config.in: Regenerate.
9077 * config/i386/driver-mingw32.c: new file.
9078 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9079 * config.host: Link driver-mingw32.o on MinGW host.
9080 * doc/install.texi: Document new --enable-mingw-wildcard configure
9083 2017-05-04 Marek Polacek <polacek@redhat.com>
9085 PR tree-optimization/80612
9086 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9088 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9089 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9091 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9092 (movt splitter): Likewise.
9093 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9094 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9095 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9096 block for Thumb-1 with MOVT.
9097 (thumb2_legitimate_address_p): Move code block ...
9098 (can_avoid_literal_pool_for_label_p): ... into this new function.
9099 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9101 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9102 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9103 "M-profile targets with the MOVT instruction".
9105 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9107 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9108 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9109 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9111 2017-05-04 Martin Liska <mliska@suse.cz>
9113 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9116 2017-05-04 Richard Biener <rguenther@suse.de>
9118 * tree.c (array_at_struct_end_p): Handle arrays at struct
9119 end with flexarrays more conservatively. Refactor and treat
9120 arrays of arrays or aggregates more strict. Fix
9121 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9122 * tree.c (array_at_struct_end_p): Adjust prototype.
9123 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9124 * gimple-fold.c (get_range_strlen): Likewise.
9125 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9127 2017-05-04 Richard Biener <rguenther@suse.de>
9129 PR tree-optimization/31130
9130 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9132 (supports_overflow_infinity): Likewise.
9133 (is_negative_overflow_infinity): Likewise.
9134 (is_positive_overflow_infinity): Likewise.
9135 (is_overflow_infinity): Likewise.
9136 (stmt_overflow_infinity): Likewise.
9137 (overflow_infinity_range_p): Likewise.
9138 (usable_range_p): Remove as always returning true.
9139 (make_overflow_infinity): Remove.
9140 (negative_overflow_infinity): Likewise.
9141 (positive_overflow_infinity): Likewise.
9142 (avoid_overflow_infinity): Likewise.
9143 (set_value_range): Adjust accordingly.
9144 (set_value_range_to_nonnegative): Likewise, remove now unused
9145 overflow_infinity arg.
9146 (vrp_operand_equal_p): Adjust.
9147 (update_value_range): Likewise.
9148 (range_int_cst_singleton_p): Likewise.
9149 (operand_less_p): Likewise.
9150 (compare_values_warnv): Likewise.
9151 (extract_range_for_var_from_comparison_expr): Likewise.
9152 (vrp_int_const_binop): Likewise.
9153 (zero_nonzero_bits_from_vr): Likewise.
9154 (extract_range_from_multiplicative_op_1): Likewise.
9155 (extract_range_from_binary_expr_1): Likewise.
9156 (extract_range_from_unary_expr): Likewise.
9157 (extract_range_from_comparison): Likewise.
9158 (extract_range_basic): Likewise.
9159 (adjust_range_with_scev): Likewise.
9160 (compare_ranges): Likewise.
9161 (compare_range_with_value): Likewise.
9162 (dump_value_range): Likewise.
9163 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9165 (simplify_cond_using_ranges): Adjust.
9167 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9169 * brig-builtins.def: Added a builtin for class_f64.
9170 * builtin-types.def: Added a builtin type needed by class_f64.
9172 2017-05-03 Jason Merrill <jason@redhat.com>
9174 * timevar.def: Add TV_CONSTEXPR.
9176 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9178 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9180 2017-05-03 Martin Jambor <mjambor@suse.cz>
9182 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9183 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9184 * ipa-cp.c (ipcp_propagate_stage): Do not call
9185 ipa_update_after_lto_read.
9186 * ipa-inline.c (ipa_inline): Likewise.
9188 2017-05-03 Martin Jambor <mjambor@suse.cz>
9190 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9191 tag. Added a default constructor and a destructor.
9192 (ipa_edge_args_sum_t): New class;
9193 (ipa_edge_args_sum): Declare.
9194 (ipa_edge_args_vector): Remove declaration.
9195 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9196 (ipa_free_edge_args_substructures): Remove declaration.
9197 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9198 (ipa_edge_args_info_available_for_edge_p): Likewise.
9199 * ipa-prop.c (ipa_edge_args_vector): Removed.
9200 (edge_removal_hook_holder): Likewise.
9201 (edge_duplication_hook_holder): Likewise.
9202 (ipa_edge_args_sum): New variable.
9203 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9204 ipa_edge_args_vector.
9205 (ipa_free_edge_args_substructures): Likewise.
9206 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9207 ipa_edge_args_vector.
9208 (ipa_edge_removal_hook): Turned into method
9209 ipa_edge_args_sum_t::remove.
9210 (ipa_edge_duplication_hook): Turned into method
9211 ipa_edge_args_sum_t::duplicate.
9212 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9213 registering edge hooks.
9214 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9215 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9216 ipa_edge_args_sum instead of ipa_edge_args_vector.
9217 * ipa-profile.c (ipa_profile): Likewise.
9219 2017-05-03 Martin Jambor <mjambor@suse.cz>
9221 * symbol-summary.h (function_summary): New method exists.
9222 (function_summary::symtab_removal): Deallocate through release.
9223 (call_summary): New class.
9224 (gt_ggc_mx): New overload.
9225 (gt_pch_nx): Likewise.
9226 (gt_pch_nx): Likewise.
9228 2017-05-03 Jeff Law <law@redhat.com>
9230 PR tree-optimization/78496
9231 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9232 from simplify_cond_using_ranges. Split off code to walk
9233 backwards through casts into ...
9234 (simplify_cond_using_ranges_2): New function.
9235 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9236 (execute_vrp): After identifying jump threads, call
9237 simplify_cond_using_ranges_2.
9239 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9242 * ipa-inline.h (inline_summary): Add ctor.
9243 (create_ggc): Do not use ggc_cleared_alloc.
9245 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9246 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9248 * gcc.c (handle_braces): Support escaping in switch matching
9250 * doc/invoke.texi (Spec Files): Document it.
9251 Remove superfluous @code markup in items.
9253 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9255 * diagnostic-show-locus.c (struct column_range): New struct.
9256 (get_affected_columns): New function.
9257 (get_printed_columns): New function.
9258 (struct correction): New struct.
9259 (correction::ensure_capacity): New function.
9260 (correction::ensure_terminated): New function.
9261 (struct line_corrections): New struct.
9262 (line_corrections::~line_corrections): New dtor.
9263 (line_corrections::add_hint): New function.
9264 (layout::print_trailing_fixits): Reimplement in terms of the new
9266 (selftest::test_overlapped_fixit_printing): New function.
9267 (selftest::diagnostic_show_locus_c_tests): Call it.
9269 2017-05-03 Nathan Sidwell <nathan@acm.org>
9271 Canonicalize canonical type hashing
9272 * tree.h (type_hash_canon_hash): Declare.
9273 * tree.c (type_hash_list, attribute_hash_list): Move into
9274 type_hash_canon_hash.
9275 (build_type_attribute_qual_variant): Break out hash code calc into
9276 type_hash_canon_hash.
9277 (type_hash_canon_hash): New. Generic type hash computation.
9278 (build_range_type_1, build_array_type_1, build_function_type,
9279 build_method_type_directly, build_offset_type, build_complex_type,
9280 make_vector_type): Call it.
9282 2017-05-03 Richard Biener <rguenther@suse.de>
9284 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9285 When all DRs have unknown misaligned do not always peel
9286 when there is a store but apply the same costing model as if
9287 there were only loads.
9289 2017-05-03 Richard Biener <rguenther@suse.de>
9292 PR tree-optimization/80492
9293 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9294 compare_base_decls returning dont-know properly.
9296 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9298 * config/arm/iterators.md (CCSI): New mode iterator.
9299 (arch): New mode attribute.
9300 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9301 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9302 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9303 code iterator for success result mode.
9304 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9305 the corresponding new insn generators.
9307 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9310 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9311 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9313 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9315 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9316 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9317 (DDR_A): Wrap DDR argument in brackets.
9318 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9319 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9320 (DDR_REVERSED_P): Likewise.
9322 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9324 PR tree-optimization/79472
9325 * tree-switch-conversion.c (struct switch_conv_info): Add
9326 contiguous_range and default_case_nonstandard fields.
9327 (collect_switch_conv_info): Compute contiguous_range and
9328 default_case_nonstandard fields, don't clear final_bb if
9329 contiguous_range and only the default case doesn't have the required
9331 (check_all_empty_except_final): Set default_case_nonstandard instead
9332 of failing if contiguous_range and the default case doesn't have empty
9334 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9335 and only the default case doesn't have the required constants. Skip
9337 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9338 if default_case_nonstandard.
9339 (build_constructors): Build constant 1 just once. Assert that default
9340 values aren't inserted in between cases if contiguous_range. Skip
9342 (build_arrays): Skip virtual phis.
9343 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9344 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9345 Handle virtual phis.
9346 (gen_inbound_check): Handle default_case_nonstandard case.
9347 (process_switch): Adjust check_final_bb caller. Call
9348 gather_default_values with the first non-default case instead of
9349 default case if default_case_nonstandard.
9351 2017-05-02 Nathan Sidwell <nathan@acm.org>
9353 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9354 check. Fix formatting.
9356 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9358 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9359 errors when comparing specialized and unspecialized times.
9361 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9363 * diagnostic-show-locus.c
9364 (layout::should_print_annotation_line_p): Make private.
9365 (layout::print_annotation_line): Make private.
9366 (layout::annotation_line_showed_range_p): Make private.
9367 (layout::show_ruler): Make private.
9368 (layout::print_source_line): Make private. Pass in line and
9369 line_width, rather than calling location_get_source_line. Drop
9371 (layout::print_leading_fixits): New method.
9372 (layout::print_any_fixits): Rename to...
9373 (layout::print_trailing_fixits): ...this, and make private.
9374 Don't print newline fixits.
9375 (diagnostic_show_locus): Move logic for printing one row into...
9376 (layout::print_line): ...this new function. Move the
9377 location_get_source_line call and error-handling from
9378 print_source_line to here. Call print_leading_fixits, and rename
9379 print_any_fixits to print_trailing_fixits.
9380 (selftest::test_fixit_insert_containing_newline): Update now that
9381 newlines are partially supported.
9382 (selftest::test_fixit_insert_containing_newline_2): New test.
9383 (selftest::test_fixit_replace_containing_newline): Update comments.
9384 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9385 * edit-context.c (class added_line): New class.
9386 (class edited_line): Describe newline handling in comment.
9387 (edited_line::actually_edited_p): New method.
9388 (edited_line::print_content): Delete redundant decl.
9389 (edited_line::m_predecessors): New field.
9390 (edited_file::print_content): Call edited_line::print_content.
9391 (edited_file::print_diff): Update to support newlines.
9392 (edited_file::print_diff_hunk): Likewise.
9393 (edited_file::print_run_of_changed_lines): New function.
9394 (edited_file::print_diff_line): Convert to...
9395 (print_diff_line): ...this.
9396 (edited_file::get_effective_line_count): New function.
9397 (edited_line::edited_line): Initialize new field m_predecessors.
9398 (edited_line::~edited_line): Clean up m_predecessors.
9399 (edited_line::apply_fixit): Handle newlines.
9400 (edited_line::get_effective_line_count): New function.
9401 (edited_line::print_content): New function.
9402 (edited_line::print_diff_lines): New function.
9403 (selftest::test_applying_fixits_insert_containing_newline): New
9405 (selftest::test_applying_fixits_replace_containing_newline): New
9407 (selftest::insert_line): New function.
9408 (selftest::test_applying_fixits_multiple_lines): Add example of
9410 (selftest::edit_context_c_tests): Call the new tests.
9412 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9414 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
9415 parameter cand. Update dump information.
9416 (get_computation_cost): Update uses.
9418 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9420 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
9421 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
9422 (get_computation_at, rewrite_use_address): Update use of
9423 get_computation_aff.
9425 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9427 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
9428 (get_computation): Delete.
9429 (get_computation_cost): Implement like get_computation_cost_at.
9430 Use get_computation_at.
9431 (get_computation_cost_at): Delete.
9432 (rewrite_use_nonlinear_expr): Use get_computation_at.
9433 (rewrite_use_compare, remove_unused_ivs): Ditto.
9435 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9437 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
9439 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9441 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
9442 (ivopts_global_cost_for_size): Rename parameter and update uses.
9443 (iv_ca_recount_cost): Update uses.
9444 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
9445 candidates seperately in n_invs and n_cands.
9446 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
9448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9450 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
9451 (find_inv_vars_cb): New.
9452 (find_depends): Renamed to ...
9453 (find_inv_vars): ... this.
9454 (add_candidate_1, force_var_cost): Call find_inv_vars.
9455 (split_address_cost, determine_group_iv_cost_cond): Ditto.
9457 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9459 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
9460 inv_vars. Add inv_exprs.
9461 (struct iv_cand): Rename depends_on to inv_vars.
9462 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
9463 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
9464 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
9465 (dump_cand): Dump inv_vars.
9466 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
9467 (record_invariant, find_depends, add_candidate_1): Ditto.
9468 (set_group_iv_cost, force_var_cost): Ditto.
9469 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
9470 (get_computation_cost_at, get_computation_cost): Ditto.
9471 (determine_group_iv_cost_generic): Ditto.
9472 (determine_group_iv_cost_address): Ditto.
9473 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
9474 (determine_group_iv_costs): Ditto.
9475 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
9476 (iv_ca_set_remove_invariants): Renamed to ...
9477 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
9478 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
9479 (iv_ca_set_add_invariants): Renamed to ...
9480 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
9481 (iv_ca_set_cp): Use iv_ca_set_add_invs.
9482 (iv_ca_has_deps): Support inv_vars and inv_exprs.
9483 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
9484 (create_new_ivs): Remove useless dump.
9486 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9488 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
9490 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
9491 (iv_ca_set_no_cp, create_new_iv): Ditto.
9493 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9495 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9497 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9499 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
9500 function tree_check2.
9502 2017-05-02 Martin Liska <mliska@suse.cz>
9504 * doc/gcov.texi: Add missing preposition.
9505 * gcov.c (function_info::function_info): Properly fill up
9506 all member variables.
9508 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9510 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
9512 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9514 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
9516 2017-05-02 Martin Liska <mliska@suse.cz>
9519 * lto-streamer-in.c (lto_read_tree_1): Remove
9521 * lto-streamer.c (struct tree_hash_entry): Likewise.
9522 (struct tree_entry_hasher): Likewise.
9523 (tree_entry_hasher::hash): Likewise.
9524 (tree_entry_hasher::equal): Likewise.
9525 (lto_streamer_init): Likewise.
9526 (lto_orig_address_map): Likewise.
9527 (lto_orig_address_get): Likewise.
9528 (lto_orig_address_remove): Likewise.
9529 * lto-streamer.h: Likewise.
9530 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
9531 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
9533 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
9535 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
9536 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
9537 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
9538 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
9539 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
9540 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
9541 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
9542 (mm_maskz_sub_ss): New intrinsics.
9543 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9544 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9545 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
9546 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
9547 (__builtin_ia32_subss_mask_round): New builtins.
9548 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9549 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9550 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
9552 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
9553 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9555 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9558 2017-05-02 Martin Jambor <mjambor@suse.cz>
9560 PR tree-optimization/78687
9561 * tree-sra.c (access): New field parent.
9562 (process_subtree_disqualification): New function.
9563 (disqualify_candidate): Call it.
9564 (build_accesses_from_assign): Reset write flag if creating an
9566 (build_access_subtree): Fill in parent field and also prpagate
9567 down grp_write flag.
9568 (create_artificial_child_access): New parameter set_grp_write, set
9569 grp_write to its value.
9570 (propagate_subaccesses_across_link): Also propagate grp_write flag
9572 (propagate_all_subaccesses): Push the closest parent back to work
9573 queue if add_access_to_work_queue returned true.
9575 2017-05-02 Richard Biener <rguenther@suse.de>
9577 * common.opt (fstrict-overflow): Alias negative to fwrapv.
9578 * doc/invoke.texi (fstrict-overflow): Remove all traces of
9579 -fstrict-overflow documentation.
9580 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
9581 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
9582 flag_strict_overflow.
9583 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
9584 * lto-opts.c (lto_write_options): Do not stream it.
9585 * lto-wrapper.c (merge_and_complain): Do not handle it.
9586 * opts.c (default_options_table): Do not set -fstrict-overflow.
9587 (finish_options): Likewise do not clear it when sanitizing.
9588 * simplify-rtx.c (simplify_const_relational_operation): Do not
9589 test flag_strict_overflow.
9591 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9593 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
9594 using enabled attribute.
9595 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
9596 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
9597 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
9598 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
9599 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
9600 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
9601 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
9602 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
9603 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
9604 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
9606 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9608 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
9610 2017-05-02 Richard Biener <rguenther@suse.de>
9612 PR tree-optimization/80591
9614 2017-04-10 Richard Biener <rguenther@suse.de>
9616 * tree-ssa-structalias.c (find_func_aliases): Properly handle
9619 2017-05-02 Richard Biener <rguenther@suse.de>
9621 PR tree-optimization/80549
9622 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
9623 (cleanup_tree_cfg_noloop): Create forwarders to known loop
9624 headers if they do not have a preheader.
9626 2017-05-02 Martin Liska <mliska@suse.cz>
9629 * common.opt: Fix typo.
9630 * doc/invoke.texi: Likewise.
9632 2017-05-01 Jan Beulich <jbeulich@suse.com>
9634 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
9635 swapping, add (x,x,m,x,n) alternative.
9637 2017-05-01 Nathan Sidwell <nathan@acm.org>
9639 * calls.c (combine_pending_stack_adjustment_and_call): Remove
9640 unnecessary unadjusted_alignment check.
9642 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9645 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
9647 * gimplify.c (gimplify_cilk_detach): New function.
9648 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
9649 * tree-core.h: Document EXPR_CILK_SPAWN.
9650 * tree.h (EXPR_CILK_SPAWN): Define.
9652 2017-05-01 David Malcolm <dmalcolm@redhat.com>
9654 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
9655 to use new fixit_hint representation, using the "replace" logic.
9656 (get_line_span_for_fixit_hint): Likewise.
9657 (layout::print_any_fixits): Likewise.
9658 (selftest::test_one_liner_many_fixits): Rename to...
9659 (selftest::test_one_liner_many_fixits_1): ...this, and update
9660 comment and expected output to reflect that the multiple fix-it
9661 hints are now consolidated into one insertion.
9662 (selftest::test_one_liner_many_fixits_2): New test.
9663 (selftest::test_diagnostic_show_locus_one_liner): Update for
9665 (selftest::test_fixit_consolidation): Update for fix-it API
9667 * diagnostic.c (print_parseable_fixits): Likewise.
9668 * edit-context.c (edited_line::m_line_events): Convert from
9669 auto_vec <line_event *> to auto_vec <line_event>.
9670 (class line_event): Convert from abstract base class to a concrete
9671 class, taking over the role of replace_event.
9672 (class insert_event): Delete.
9673 (class replace_event): Rename to class line_event. Convert to
9675 (edit_context::add_fixits): Reimplement.
9676 (edit_context::apply_insert): Delete.
9677 (edit_context::apply_replace): Rename to...
9678 (edit_context::apply_fixit): ...this. Convert to half-open range.
9679 (edited_file::apply_insert): Delete.
9680 (edited_file::apply_replace): Rename to...
9681 (edited_file::apply_fixit): ...this.
9682 (edited_line::~edited_line): Drop deletion of events.
9683 (edited_line::apply_insert): Delete.
9684 (edited_line::apply_replace): Rename to...
9685 (edited_line::apply_fixit): ...this. Convert to half-open range.
9686 Update for change to type of m_line_events.
9687 * edit-context.h (edit_context::apply_insert): Delete.
9688 (edit_context::apply_replace): Rename to...
9689 (edit_context::apply_fixit): ...this.
9691 2017-05-01 Martin Sebor <msebor@redhat.com>
9693 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
9696 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
9699 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
9700 __get_cpuid_max returns 0.
9701 (__get_cpuid_count): Ditto.
9703 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
9705 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
9706 replacement expression is another instance of one of its arguments.
9708 2017-05-01 Jakub Jelinek <jakub@redhat.com>
9711 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
9712 check for stack push/pop autoinc.
9713 * config/i386/i386.c (ix86_agi_dependent): Return false
9714 if the only reason why modified_in_p returned true is that
9715 addr is SP based and set_insn is a push or pop.
9717 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
9719 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
9722 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
9725 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
9726 (account_size_time): Use two predicates - exec_pred and
9728 (evaluate_conditions_for_known_args): Compute both clause and
9730 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
9731 (inline_summary_t::duplicate): Update.
9732 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
9734 (compute_inline_parameters): Likewise.
9735 (estimate_edge_size_and_time): Update caluclation of time.
9736 (estimate_node_size_and_time): Compute both time and nonspecialized
9738 (estimate_ipcp_clone_size_and_time): Update.
9739 (inline_merge_summary): Update.
9740 (do_estimate_edge_time): Update.
9741 (do_estimate_edge_size): Update.
9742 (do_estimate_edge_hints): Update.
9743 (inline_read_section, inline_write_summary): Stream both new predicates.
9744 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
9746 (compute_inlined_call_time): Cleanup.
9747 (big_speedup_p): Update.
9748 (edge_badness): Update.
9749 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
9750 (size_time_entry): Replace predicate by exec_predicate and
9752 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
9753 (estimate_edge_time): Return also nonspec_time.
9754 (reset_edge_growth_cache): Update.
9756 2017-04-29 Jakub Jelinek <jakub@redhat.com>
9758 PR rtl-optimization/80491
9759 * ifcvt.c (noce_process_if_block): When looking for x setter
9760 with missing else_bb, don't check only the insn right before
9761 cond_earliest, but look for the last insn that x is modified in
9764 PR rtl-optimization/80491
9765 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
9767 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
9769 PR tree-optimization/80487
9770 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
9772 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9774 PR tree-optimization/79697
9775 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
9776 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
9777 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
9779 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
9780 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
9782 2017-04-28 Martin Sebor <msebor@redhat.com>
9784 PR tree-optimization/80523
9785 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
9786 (init_target_to_host_charmap, target_to_host, target_strtol10): New
9788 (maybe_warn, format_directive, parse_directive): Use new functions.
9789 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
9791 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
9793 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
9795 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
9797 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
9798 target_header_dir): Set correctly.
9799 * configure: Regenerated.
9800 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
9801 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
9802 instead of SYSTEM_HEADER_DIR.
9804 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
9806 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
9807 (estimate_local_effects): Likewise.
9808 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
9809 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
9810 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
9811 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
9812 do_estimate_edge_time, estimate_edge_time): Likewise.
9813 * ipa-inline-analysis.c (estimate_node_size_and_time,
9814 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
9815 (estimate_time_after_inlining): Remove.
9817 2017-04-28 Martin Liska <mliska@suse.cz>
9819 * doc/gcov.texi: Enhance documentation of gcov.
9821 2017-04-28 Martin Liska <mliska@suse.cz>
9823 * doc/gcov.texi: Sort options in alphabetic order.
9824 * doc/gcov-dump.texi: Likewise.
9825 * doc/gcov-tool.texi: Likewise.
9826 * gcov.c (print_usage): Likewise.
9827 * gcov-dump.c (print_usage): Likewise.
9828 * gcov-tool.c (print_merge_usage_message): Likewise.
9829 (print_rewrite_usage_message): Likewise.
9830 (print_overlap_usage_message): Likewise.
9832 2017-04-28 Martin Liska <mliska@suse.cz>
9834 PR gcov-profile/53915
9835 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
9837 2017-04-28 Martin Liska <mliska@suse.cz>
9839 PR gcov-profile/79891
9840 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
9841 is marked by compiler as living on a line.
9842 (get_cycles_count): Remove usage of the union.
9843 (output_intermediate_file): Likewise.
9844 (find_source): Fix GNU coding style.
9845 (accumulate_line_counts): Remove old non-all block mode.
9846 (output_lines): Remove usage of the union.
9847 * profile.c (output_location): Include all BBs, even if
9848 belonging to a same line (and file) as a previous BB.
9850 2017-04-28 Martin Liska <mliska@suse.cz>
9852 * gcov.c (process_args): Handle new argument 'w'.
9853 (read_graph_file): Assign ID to BBs.
9854 (output_branch_count): Display BB # if verbose flag is set.
9855 (output_lines): Likewise for arcs.
9856 (print_usage): Add '--verbose' option help.
9857 * doc/gcov.texi: Document --verbose (-w) option.
9859 2017-04-28 Martin Liska <mliska@suse.cz>
9861 * gcov.c (struct block_location_info): New struct.
9862 (process_file): Fill up the new structure.
9863 (read_graph_file): Replace usage of encoding by the newly added
9865 (add_line_counts): Likewise.
9866 (accumulate_line_counts): Remove usage of the union.
9867 (function_info::function_info): New function.
9868 (function_info::~function_info): Likewise.
9869 (process_file): Call delete instead of release_function.
9870 (release_function): Release the function.
9871 (release_structures): Call delete instead of release_function.
9872 (solve_flow_graph): Replace usage of num_blocks.
9873 (find_exception_blocks): Likewise.
9874 (output_lines): Fix GNU coding style.
9876 2017-04-28 Martin Liska <mliska@suse.cz>
9879 * coverage.c (coverage_remove_note_file): New function.
9880 * coverage.h: Declare the function.
9881 * toplev.c (finalize): Clean if an error has been seen.
9883 2017-04-28 Martin Liska <mliska@suse.cz>
9885 PR gcov-profile/80031
9886 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
9887 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
9888 * gcov.c (read_graph_file): Read just number of blocks.
9889 * profile.c (branch_prob): Do not stream 0 flags per a basic
9892 2017-04-28 Martin Liska <mliska@suse.cz>
9894 * gcov-dump.c (tag_*): Add new argument to declarations.
9895 (dump_gcov_file): Likewise.
9896 (tag_blocks): Add and use new argument depth.
9897 (tag_arcs): Likewise.
9898 (tag_lines): Likewise.
9899 (tag_counters): Likewise.
9900 (tag_summary): Likewise.
9901 (dump_working_sets): Use depth to do a proper indentation.
9903 2017-04-28 Jakub Jelinek <jakub@redhat.com>
9906 * cgraph.h (symtab_node::debug_symtab): No longer inline.
9907 * symtab.c (symtab_node::debug_symtab): Move definition here.
9909 2017-04-28 Richard Biener <rguenther@suse.de>
9911 * lto-streamer.h (LTO_major_version): Bump to 7.
9913 2017-04-28 Richard Biener <rguenther@suse.de>
9915 * tree-vrp.c (assert_info): New struct.
9916 (add_assert_info): New helper.
9917 (register_edge_assert_for_2): Refactor to add asserts to a vector
9919 (register_edge_assert_for_1): Likewise.
9920 (register_edge_assert_for): Likewise.
9921 (finish_register_edge_assert_for): New helper actually registering
9922 asserts where live on edge.
9923 (find_conditional_asserts): Adjust.
9924 (find_switch_asserts): Likewise.
9925 (evrp_dom_walker::try_find_new_range): Generalize.
9926 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
9928 2017-04-27 Marek Polacek <polacek@redhat.com>
9931 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
9932 arg10 and arg11 to itype.
9934 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
9936 * doc/extend.texi (Object Size Checking): Improve grammar.
9938 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
9941 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
9942 that the logic for permitting reciprocal estimates matches that
9945 2017-04-27 Jakub Jelinek <jakub@redhat.com>
9948 * tree.c (type_cache_hasher::equal): Only compare
9949 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
9950 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
9951 non-aggregate element types.
9952 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
9953 about the flag on ARRAY_TYPEs in the comment, formatting fix.
9955 2017-04-27 Richard Biener <rguenther@suse.de>
9958 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
9959 stripping ARRAY_REFs from MEM_EXPR make sure we're not
9960 keeping a reference to a trailing array.
9962 2017-04-27 Richard Biener <rguenther@suse.de>
9965 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
9966 being in loop-closed SSA form conservatively.
9967 (chrec_fold_multiply_poly_poly): Likewise.
9969 2017-04-27 Tamar Christina <tamar.christina@arm.com>
9972 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
9973 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
9975 2017-04-27 Jakub Jelinek <jakub@redhat.com>
9978 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
9979 (aarch64_function_arg_alignment): Return unsigned int again, but still
9980 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
9981 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
9982 Don't emit -Wpsabi note.
9983 (aarch64_function_arg_boundary): Likewise.
9984 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
9987 2017-04-26 Nathan Sidwell <nathan@acm.org>
9989 * tree.h (crc32_unsigned_n): Declare.
9990 (crc32_unsigned, crc32_unsigned): Make inline.
9991 * tree.c (crc32_unsigned_bits): Replace with ...
9992 (crc32_unsigned_n): ... this.
9993 (crc32_unsigned, crc32_byte): Remove.
9994 (crc32_string): Remove unnecessary braces.
9996 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
9998 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
9999 * ipa-inline-analysis.c (MAX_TIME): Remove.
10000 (account_size_time): Use sreal for time.
10001 (dump_inline_summary): Update.
10002 (estimate_function_body_sizes): Update.
10003 (estimate_edge_size_and_time): Update.
10004 (estimate_calls_size_and_time): Update.
10005 (estimate_node_size_and_time): Update.
10006 (inline_merge_summary): Update.
10007 (inline_update_overall_summary): Update.
10008 (estimate_time_after_inlining): Update.
10009 (inline_read_section): Update.
10010 (inline_write_summary): Update.
10011 * ipa-inline.c (compute_uninlined_call_time): Update.
10012 (compute_inlined_call_time): Update.
10013 (recursive_inlining): Update.
10014 (inline_small_functions): Update.
10015 (dump_overall_stats): Update.
10016 * ipa-inline.h: Include sreal.h.
10017 (size_time_entry): Turn time to sreal.
10018 (inline_summary): Turn self_time nad time to sreal.
10020 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10022 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10024 (sreal::stream_out, sreal::stream_in): New.
10025 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10027 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10029 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10032 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10035 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10037 (dimode_scalar_chain::compute_convert_gain): Ditto.
10038 (dimode_scalar_chain::make_vector_copies): Ditto.
10039 (dimode_scalar_chain::convert_reg): Ditto.
10040 (dimode_scalar_chain::convert_insn): Ditto.
10041 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10042 (VI248_AVX512BW_1): New mode iterator.
10043 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10044 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10047 2017-04-25 Martin Sebor <msebor@redhat.com>
10049 PR tree-optimization/80497
10050 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10051 constants are representable in HOST_WIDE_INT.
10052 (parse_directive): Ditto.
10054 2017-04-25 Martin Sebor <msebor@redhat.com>
10057 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10058 (new_zero_array): Adjust signature.
10059 (dom_info::dom_init): Used unsigned rather that size_t.
10060 (dom_info::dom_info): Same.
10062 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10063 Jakub Jelinek <jakub@redhat.com>
10066 * config/arm/arm.c: Include gimple.h.
10067 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10068 returns negative, increment ncrn only if it returned positive.
10069 (arm_needs_doubleword_align): Return int instead of bool,
10070 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10071 members, but if there is any such non-FIELD_DECL
10072 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10073 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10074 returns negative, increment nregs only if it returned positive.
10075 (arm_setup_incoming_varargs): Likewise.
10076 (arm_function_arg_boundary): Emit -Wpsabi note if
10077 arm_needs_doubleword_align returns negative, return
10078 DOUBLEWORD_ALIGNMENT only if it returned positive.
10080 2017-04-25 Marek Polacek <polacek@redhat.com>
10083 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10084 first argument to type.
10086 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10089 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10090 type checks to test for compatibility instead of equality.
10092 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10093 Jakub Jelinek <jakub@redhat.com>
10096 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10098 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10099 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10100 the alignment computation, but return their maximum in warn_alignment.
10101 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10102 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10104 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10105 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10108 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10110 * config/arc/simdext.md (dmpyh): Fix typo.
10112 2017-04-25 Richard Biener <rguenther@suse.de>
10114 PR tree-optimization/80492
10115 * alias.c (compare_base_decls): Handle registers with asm
10116 specification conservatively.
10117 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10118 compare_base_decls returning dont-know properly.
10120 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10122 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10123 (legitimate_offset_address_p): New function.
10124 (arc_legitimate_address_p): Use above function.
10126 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10128 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10130 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10132 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10133 ACCH registers whenever they are available.
10135 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10137 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10138 double regs fix when not used.
10140 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10142 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10144 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10145 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10147 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10149 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10150 when emitting short ADD instructions.
10152 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10154 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10156 (cmpsi_cc_c_insn): Likewise.
10157 (cbranchsi4_scratch): Compute proper instruction length using
10158 compact_hreg_operand.
10159 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10161 2017-04-25 Richard Biener <rguenther@suse.de>
10163 PR middle-end/80509
10164 * passes.c (pass_manager::pass_manager): Initialize
10165 m_name_to_pass_map.
10167 2017-04-25 Richard Biener <rguenther@suse.de>
10169 PR tree-optimization/79201
10170 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10172 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10175 * config/s390/vector.md: Split MEM->GPR vector moves for
10176 non-s_operand addresses.
10178 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10181 * config/s390/predicates.md (reload_const_wide_int_operand): New
10183 * config/s390/s390.md ("movti"): Remove d/P alternative.
10184 ("movti_bigconst"): New pattern definition.
10186 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10189 * s390-protos.h (s390_expand_cs_hqi): Removed.
10190 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10191 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10192 modes as well as CCZ1mode and CCZmode.
10193 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10194 signature of s390_emit_compare_and_swap.
10195 (s390_expand_cs_hqi): Likewise, make static.
10196 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10197 compare-and-swap, in some cases.
10198 (s390_expand_cs): Wrapper function.
10199 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10201 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10202 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10203 patterns for small and large integers. Forbid symref memory operands.
10204 Move expander to s390.c. Require cc register.
10205 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10206 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10207 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10208 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10209 symref memory operands. Remove CC mode and call s390_match_ccmode
10211 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10213 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10215 * config/s390/s390.md (define_peephole2): New peephole to help
10216 combining the load-and-test pattern with volatile memory.
10218 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10220 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10221 with CCZmode for TARGET_Z196.
10223 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10225 PR rtl-optimization/80501
10226 * combine.c (make_compound_operation_int): Set subreg_code to SET
10227 even for AND with mask of the sign bit of mode.
10229 PR rtl-optimization/80500
10230 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10231 sum's initial value.
10233 2017-04-25 Julian Brown <julian@codesourcery.com>
10234 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10236 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10238 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10240 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10242 2017-04-25 Julian Brown <julian@codesourcery.com>
10243 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10245 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10246 (thunderx2t99_sha): New Reservation.
10248 2017-04-25 Julian Brown <julian@codesourcery.com>
10249 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10251 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10252 type for 1-element load.
10254 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10256 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10258 2017-04-24 Martin Jambor <mjambor@suse.cz>
10260 PR tree-optimization/80293
10261 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10262 char arrays not totally scalarizable if it is false.
10263 (analyze_all_variable_accesses): Pass correct value in the new
10264 parameter. Add a statistics counter.
10266 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10268 PR middle-end/79931
10269 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10271 2017-04-24 Richard Biener <rguenther@suse.de>
10273 PR tree-optimization/80494
10274 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10275 out for complex types.
10277 2017-04-24 Richard Biener <rguenther@suse.de>
10279 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10280 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10281 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10282 (DFS): Adjust and never fail.
10283 (sccvn_dom_walker::fail): Remove.
10284 (sccvn_dom_walker::before_dom_children): Adjust.
10285 (run_scc_vn): Likewise and never fail.
10286 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10287 (pass_fre::execute): Likewise.
10289 2017-04-24 Richard Biener <rguenther@suse.de>
10291 PR tree-optimization/79725
10292 * tree-ssa-sink.c (statement_sink_location): Return whether
10293 failure reason was zero uses. Move that check later.
10294 (sink_code_in_bb): Deal with zero uses by removing the stmt
10297 2017-04-24 Richard Biener <rguenther@suse.de>
10300 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10301 pointer-based references.
10303 2017-04-24 Richard Biener <rguenther@suse.de>
10306 * pass_manager.h (pass_manager::operator new): Remove.
10307 (pass_manager::operator delete): Likewise.
10308 * passes.c (pass_manager::operator new): Remove.
10309 (pass_manager::operator delete): Likewise.
10310 (pass_manager::pass_manager): Zero individual pass members.
10312 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10315 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10316 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10317 Check "XEXP (src, 1)" operand here.
10318 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10319 Check "XEXP (src, 1)" operand here.
10320 (dimode_scalar_chain::make_vector_copies): Detect count register
10321 of a shift instruction. Zero extend count register from QImode
10322 to DImode to satisfy vector shift pattern count operand predicate.
10323 Substitute vector shift count operand with a DImode copy.
10324 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10327 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10329 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10330 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10331 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10332 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10333 (UNSPEC_NOREX_MEM): Remove definition.
10335 2017-04-21 Richard Biener <rguenther@suse.de>
10337 PR tree-optimization/79547
10338 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10339 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10340 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10341 without any constraints.
10343 2017-04-21 Richard Biener <rguenther@suse.de>
10345 PR tree-optimization/78847
10346 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10348 2017-04-21 Richard Biener <rguenther@suse.de>
10350 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10351 (build_distinct_type_copy): Likewise.
10352 (build_variant_type_copy): Likewise.
10353 * tree.c (build_qualified_type): Pass down mem-stat info.
10354 (build_distinct_type_copy): Likewise.
10355 (build_variant_type_copy): Likewise.
10357 2017-04-21 Richard Biener <rguenther@suse.de>
10359 PR tree-optimization/80237
10360 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10362 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10363 for a simplified result.
10365 2016-04-21 Richard Biener <rguenther@suse.de>
10367 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10368 sth as strict as a simple_iv but a chrec without symbols and an
10369 operand defined in the loop we are peeling (and not some subloop).
10370 (propagate_constants_for_unrolling): Propagate all constants.
10372 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10375 * config/i386/i386.c (print_reg): Remove assert for disalowed
10376 regno values, call output_operand_lossage instead.
10378 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10381 * config/i386/constraints.md (Yc): New register constraint.
10382 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10383 Use Yc constraint for alternative 2 of operand 0. Remove
10384 preferred_for_speed attribute.
10386 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10388 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10389 lastprivate clauses in SIMT case.
10391 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10393 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10395 2017-04-20 Richard Biener <rguenther@suse.de>
10397 PR tree-optimization/57796
10398 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10399 as N scalar stores.
10400 (vect_model_load_cost): Cost gathers as N scalar loads.
10402 2017-04-20 Richard Biener <rguenther@suse.de>
10404 * ggc-page.c (ggc_allocated_p): Rename to ...
10405 (safe_lookup_page_table_entry): ... this and return the lookup
10407 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10409 2017-04-20 Richard Biener <rguenther@suse.de>
10411 PR tree-optimization/80453
10412 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
10413 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
10414 from the conditions.
10415 (vn_phi_eq): Pass them down.
10416 (vn_phi_lookup): Record them.
10417 (vn_phi_insert): Likewise.
10419 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
10421 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
10422 uninitialized variable warning to avoid buffer overrun.
10424 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10427 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
10428 is suppressed for '{ 0 }' in C.
10430 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10432 * BASE-VER: Set to 8.0.0.
10434 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
10436 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
10437 priority .init_array and .fini_array section with SECTION_NOTYPE
10440 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10442 PR middle-end/80423
10443 * tree.h (build_array_type): Add typeless_storage default argument.
10444 * tree.c (type_cache_hasher::equal): Also compare
10445 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
10446 (build_array_type): Add typeless_storage argument, set
10447 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
10449 (build_nonshared_array_type): Adjust build_array_type_1 caller.
10450 (build_array_type): Likewise. Add typeless_storage argument.
10452 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10453 Jakub Jelinek <jakub@redhat.com>
10455 PR tree-optimization/80426
10456 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
10457 operation on symbolic operands, also compute the overflow for the
10458 invariant part when the operation degenerates into a negation.
10460 2017-04-19 Jakub Jelinek <jakub@redhat.com>
10463 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
10464 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
10467 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
10469 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
10472 * config/avr/avr.c (tree.h): Include it.
10473 (cgraph.h): Include it.
10474 (avr_encode_section_info): Don't warn for uninitialized progmem
10475 variable if it's just an alias.
10477 2017-04-19 Richard Biener <rguenther@suse.de>
10480 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
10481 when needed by AutoPGO.
10483 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
10486 * doc/lto.texi: Remove an extra 'that'.
10488 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
10490 PR rtl-optimization/80429
10491 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
10492 are only used in debug insns.
10494 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10495 Vladimir Makarov <vmakarov@redhat.com>
10497 * config/sparc/predicates.md (input_operand): Add comment. Return
10498 true for any memory operand when LRA is in progress.
10499 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
10501 2017-04-18 Jeff Law <law@redhat.com>
10504 * mips.md ({return,simple_return}_internal): Do not overwrite
10507 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10509 PR tree-optimization/80443
10510 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
10511 instead of adding 1, subtract -1 and similarly instead of subtracting
10514 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
10516 PR rtl-optimization/80357
10517 * haifa-sched.c (tmp_bitmap): New variable.
10518 (model_recompute): Handle duplicate use records.
10519 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
10520 (free_global_sched_pressure_data): Free it.
10522 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
10525 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10526 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10527 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10528 instead of SYSTEM_HEADER_DIR.
10530 2017-04-18 Jeff Law <law@redhat.com>
10532 PR middle-end/80422
10533 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
10534 predecessors after walking up the insn chain.
10536 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10539 * dwarf2out.c (modified_type_die): Try harder not to emit internal
10540 sizetype type into debug info.
10542 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
10545 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
10546 unneeded test for TARGET_UPPER_REGS_SF.
10547 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
10549 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10552 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
10553 instead of gsi_after_labels.
10555 2017-04-18 Jeff Law <law@redhat.com>
10557 * regcprop.c (maybe_mode_change): Avoid creating copies of the
10561 2017-04-13 Jeff Law <law@redhat.com>
10562 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10563 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10565 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
10568 * config/avr/avr.c (intl.h): Include it.
10569 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
10571 2017-04-18 Martin Liska <mliska@suse.cz>
10573 PR gcov-profile/78783
10574 * gcov-tool.c (gcov_output_files): Validate that destination
10575 file is either removed by the tool or by a user.
10577 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10578 Guy Benyei <guybe@mellanox.com>
10580 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
10581 block, and do not negate it, the stored id is already negative.
10583 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10585 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
10587 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10590 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
10591 masks of options that should be turned off if the VSX vector
10592 options are turned off.
10593 (OTHER_P8_VECTOR_MASKS): Likewise.
10594 (OTHER_VSX_VECTOR_MASKS): Likewise.
10595 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
10596 rs6000_disable_incompatible_switches to validate no type switches
10598 (rs6000_incompatible_switch): New function to disallow turning on
10599 other vector options if -mno-vsx, -mno-power8-vector, or
10600 -mno-power9-vector are specified.
10602 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10604 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
10606 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10608 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
10609 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
10610 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
10611 (ARG_POINTER_CFA_OFFSET): Likewise.
10613 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10615 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
10616 conditions to take advantage of various optimizations.
10618 2017-04-13 Jeff Law <law@redhat.com>
10620 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10621 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10622 (zero_extendsidi2_dext): Likewise.
10624 2017-04-13 Jakub Jelinek <jakub@redhat.com>
10627 * fold-const.c (fold_ternary_loc): Revert
10628 use op0 instead of fold_convert_loc (loc, type, arg0) part of
10631 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
10633 PR rtl-optimization/80343
10634 * lra-remat.c (update_scratch_ops): Assign original hard reg to
10635 new scratch pseudo.
10637 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
10640 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
10641 to ubsan_encode_value.
10643 2017-04-13 Jeff Law <law@redhat.com>
10645 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
10646 appearing in DEBUG_INSNs.
10648 2017-04-13 Martin Liska <mliska@suse.cz>
10650 PR gcov-profile/80413
10651 * gcov-io.c (gcov_write_string): Copy to buffer just when
10652 allocated size is greater than zero.
10654 2017-04-13 Jakub Jelinek <jakub@redhat.com>
10657 * dwarf2out.c (decls_for_scope): Ignore declarations of
10658 current_function_decl in BLOCK_NONLOCALIZED_VARS.
10660 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
10663 * ipa-visibility.c (non_local_p): Fix typos.
10664 (localize_node): When localizing symbol in same comdat group,
10665 dissolve the group only when we know external symbols are going
10667 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
10669 2017-04-12 Jakub Jelinek <jakub@redhat.com>
10671 PR tree-optimization/79390
10672 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
10673 order does not result in usable sequence, retry with reversed operand
10679 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
10680 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
10681 op0 instead of fold_convert_loc (loc, type, arg0).
10683 2017-04-12 Jeff Law <law@redhat.com>
10685 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
10686 has a delay slot in the generated code.
10688 * config/cris/cris.md (cris_preferred_reload_class): Return
10689 GENNONACR_REGS rather than GENERAL_REGS.
10691 2017-04-12 Jakub Jelinek <jakub@redhat.com>
10694 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
10695 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
10696 signedness of the result type.
10698 2017-04-12 Richard Biener <rguenther@suse.de>
10699 Jeff Law <law@redhat.com>
10701 PR tree-optimization/80359
10702 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
10703 trim stores to TARGET_MEM_REFs.
10705 2017-04-12 Richard Biener <rguenther@suse.de>
10707 PR tree-optimization/79390
10708 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
10709 threading case even more.
10711 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
10714 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
10715 for quad_address_p for TImode, instead of just not indexed_address.
10717 2017-04-12 Richard Biener <rguenther@suse.de>
10718 Bernd Edlinger <bernd.edlinger@hotmail.de>
10720 PR middle-end/79671
10721 * alias.c (component_uses_parent_alias_set_from): Handle
10722 TYPE_TYPELESS_STORAGE.
10723 (get_alias_set): Likewise.
10724 * tree-core.h (tree_type_common): Add typeless_storage flag.
10725 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
10726 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
10727 for types containing members with TYPE_TYPELESS_STORAGE.
10728 (place_field): Likewise.
10729 (layout_type): Likewise for ARRAY_TYPE.
10730 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
10731 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
10732 TYPE_TYPELESS_STORAGE.
10733 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
10735 2017-04-12 Jakub Jelinek <jakub@redhat.com>
10738 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10739 first argument to type.
10741 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10745 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
10746 CONST0_RTX (mode) rather than const0_rtx where appropriate.
10747 (rs6000_expand_binop_builtin): Likewise.
10748 (rs6000_expand_ternop_builtin): Likewise; also add missing
10749 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
10750 vshasigma built-ins.
10751 * doc/extend.texi: Document that vec_xxpermdi's third argument
10752 must be a constant.
10754 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
10756 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
10757 Use shift_const cost parameter when calculating gain of STV shifts.
10759 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
10761 PR rtl-optimization/70478
10762 * lra-constraints.c (process_alt_operands): Check memory for
10763 disfavoring memory insn operand.
10765 2017-04-11 Jakub Jelinek <jakub@redhat.com>
10767 PR middle-end/80100
10768 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
10769 left shift in unsigned HOST_WIDE_INT type.
10771 PR rtl-optimization/80385
10772 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
10773 (not (neg X)) into (plus X -1) for complex or non-integral modes.
10776 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
10777 if they have any depend clauses.
10779 2017-04-11 Martin Liska <mliska@suse.cz>
10782 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
10783 * ipa-split.c (split_function): Create a local comdat symbol
10784 if caller is in a comdat group.
10786 2017-04-11 Martin Liska <mliska@suse.cz>
10789 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
10792 2017-04-11 Martin Sebor <msebor@redhat.com>
10794 PR middle-end/80364
10795 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
10796 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
10798 (directive::set_width, directive::set_precision, format_character):
10800 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
10803 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
10806 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
10807 conflict, set target->arch_name instead of target->cpu_name.
10809 2017-04-11 Richard Biener <rguenther@suse.de>
10811 PR tree-optimization/80374
10812 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
10813 build_zero_cst, remove fold_convertible_p check again.
10815 2017-04-11 Martin Liska <mliska@suse.cz>
10818 * ubsan.c (instrument_object_size): Do not instrument register
10821 2017-04-11 Jakub Jelinek <jakub@redhat.com>
10824 * config/i386/i386-builtin-types.def
10825 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
10826 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
10827 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
10828 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
10829 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
10830 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
10831 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
10832 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
10833 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
10834 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
10835 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
10836 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
10837 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
10838 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
10839 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
10840 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
10841 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
10842 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
10843 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
10844 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
10845 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
10846 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
10847 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
10848 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
10849 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
10850 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
10851 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
10852 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
10853 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
10854 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
10855 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
10856 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
10857 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
10858 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
10859 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
10860 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
10861 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
10862 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
10863 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
10864 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
10865 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
10866 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
10867 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
10868 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
10869 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
10870 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
10872 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
10873 flag to second_arg_count, handle 4 argument function type _COUNT
10874 aliases, handle second_arg_count on second argument rather than last.
10876 2017-04-10 Jeff Law <law@redhat.com>
10878 PR tree-optimization/80374
10879 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
10880 record anything if we can not convert integer_zero_node to the
10883 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
10886 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10887 Enhance special handling given to the TARGET_P9_MINMAX option in
10888 relation to certain other options.
10890 2017-04-10 Bin Cheng <bin.cheng@arm.com>
10892 PR tree-optimization/80153
10893 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
10894 remove POINTER_PLUS_EXPR's base part directly, rather than through
10897 2017-04-10 Richard Biener <rguenther@suse.de>
10898 Bin Cheng <bin.cheng@arm.com>
10900 PR tree-optimization/80153
10901 * tree-affine.c (aff_combination_to_tree): Get base pointer from
10902 the first element of pointer type aff_tree. Build result expr in
10904 (add_elt_to_tree): Convert to type unconditionally. Remove other
10905 fold_convert calls.
10906 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
10907 (rewrite_use_nonlinear_expr): Check invariant using iv information.
10909 2017-04-10 Richard Biener <rguenther@suse.de>
10911 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10914 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
10916 PR rtl-optimization/70478
10917 * lra-constraints.c (curr_small_class_check): New.
10918 (update_and_check_small_class_inputs): New.
10919 (process_alt_operands): Update curr_small_class_check. Disfavor
10920 alternative insn memory operands. Check available regs for small
10923 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
10926 * config/mips/mips.opt (-mvirt): Update description.
10927 * doc/invoke.texi (-mvirt): Likewise.
10929 2017-04-10 Richard Biener <rguenther@suse.de>
10931 PR middle-end/80362
10932 * fold-const.c (fold_binary_loc): Look at unstripped ops when
10933 looking for NEGATE_EXPR in -A / -B to A / B folding.
10935 2017-04-10 Martin Liska <mliska@suse.cz>
10937 PR gcov-profile/80224
10938 * gcov.c (print_usage): Fix usage string.
10939 (get_gcov_intermediate_filename): Remove.
10940 (output_gcov_file): Use both for normal and intermediate format.
10941 (generate_results): Do not initialize special file for
10942 intermediate format.
10944 2017-04-10 Richard Biener <rguenther@suse.de>
10946 PR tree-optimization/80304
10947 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
10950 2017-04-10 Nathan Sidwell <nathan@acm.org>
10953 * config/rs6000/rs6000.c (rs6000_vector_type): New.
10954 (rs6000_init_builtins): Use it.
10956 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10958 * config/arm/arm.md (<mrc>): Add mode to SET source.
10959 (<mrrc>): Likewise.
10961 2017-04-10 Richard Biener <rguenther@suse.de>
10963 PR middle-end/80344
10964 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
10966 2017-04-10 Jakub Jelinek <jakub@redhat.com>
10969 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
10970 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
10971 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
10972 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
10973 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
10974 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
10975 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
10976 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
10977 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
10978 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
10979 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
10980 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
10981 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
10982 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
10983 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
10984 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
10985 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
10986 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
10987 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
10988 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
10989 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
10990 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
10991 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
10993 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
10995 PR rtl-optimization/70478
10996 * lra-constraints.c: Reverse the last patch.
10998 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11000 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11001 Add comment for WCHAR_T.
11003 2017-04-08 Martin Liska <mliska@suse.cz>
11006 2017-04-07 Martin Liska <mliska@suse.cz>
11009 * ipa-split.c (split_function): Add function part to a same comdat
11012 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11015 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11017 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11019 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11020 (vec_pairing): ...to this.
11021 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11022 (rs6000_sched_init): Adjust for name change.
11023 (struct rs6000_sched_context): Likewise.
11024 (rs6000_init_sched_context): Likewise.
11025 (rs6000_set_sched_context): Likewise.
11027 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11033 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11035 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11036 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11037 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11039 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11041 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11043 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11045 PR rtl-optimization/70703
11046 * ira-color.c (update_conflict_hard_regno_costs): Use
11047 int64_t instead of HOST_WIDE_INT.
11049 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11051 PR rtl-optimization/70478
11052 * lra-constraints.c (process_alt_operands): Disfavor alternative
11053 insn memory operands.
11055 2017-04-07 Jeff Law <law@redhat.com>
11057 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11058 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11060 2017-04-07 Martin Liska <mliska@suse.cz>
11063 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11064 Show error message instead of an ICE.
11066 2017-04-07 Martin Liska <mliska@suse.cz>
11069 * ipa-split.c (split_function): Add function part to a same comdat
11072 2017-04-07 Richard Biener <rguenther@suse.de>
11074 PR middle-end/80341
11075 * tree.c (get_unwidened): Also handle ! for_type case for
11077 * convert.c (do_narrow): Split out from ...
11078 (convert_to_integer_1): ... here. Do not pass final truncation
11079 type to get_unwidened for TRUNC_DIV_EXPR.
11081 2017-04-07 Richard Biener <rguenther@suse.de>
11083 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11085 (aff_combination_const): Adjust.
11086 (aff_combination_scale): Likewise.
11087 (aff_combination_add_elt): Likewise.
11088 (aff_combination_add_cst): Likewise.
11089 (aff_combination_convert): Likewise.
11090 (add_elt_to_tree): Likewise. Remove unused argument.
11091 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11093 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11095 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11097 * config/arm/arm.c (arm_default_short_enums): Use
11098 ARM_DEFAULT_SHORT_ENUMS.
11099 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11101 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11104 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11105 members with redundant out-of-class redeclaration.
11107 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11110 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11111 * config/i386/i386.md (*zero_extendsidi2):
11112 Add (?*x,*x) and (?*v,*v) alternatives.
11114 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11117 * config/i386/i386.c (ix86_expand_builtin)
11118 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11119 mode from insn data. Convert operands to insn operand mode.
11120 Copy operands that don't satisfy insn predicate to a register.
11122 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11124 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11127 2017-04-06 Richard Biener <rguenther@suse.de>
11129 PR tree-optimization/80334
11130 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11131 preserve alignment of accesses.
11133 2017-04-06 Richard Biener <rguenther@suse.de>
11135 PR tree-optimization/80262
11136 * tree-sra.c (build_ref_for_offset): Preserve address-space
11138 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11139 Drop useless address-space information on MEM_REF offsets.
11141 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11143 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11145 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11147 PR rtl-optimization/70703
11148 * ira-color.c (update_conflict_hard_regno_costs): Use
11149 HOST_WIDE_INT instead of long.
11151 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11154 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11155 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11157 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11158 for x86_64 target. Handle -m3dnowa option.
11160 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11162 PR rtl-optimization/70703
11163 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11164 (update_conflict_hard_regno_costs): Use long instead of unsigned
11165 arithmetic for cost calculation.
11167 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11168 Bernd Edlinger <bernd.edlinger@hotmail.de>
11171 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11174 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11177 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11178 ptr_mode with Pmode throughout.
11179 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11180 into probe_stack_range and use DImode.
11182 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11185 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11186 call_eh_return is true.
11188 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11190 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11191 Initialize last_match_fntype_index.
11193 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11196 * tree-nvr.c: Include internal-fn.h.
11197 (pass_return_slot::execute): Ignore internal calls without
11200 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11201 Richard Biener <rguenther@suse.de>
11204 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11205 captures used multiple times, except for the last use.
11206 * generic-match-head.c: Include gimplify.h.
11208 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11210 PR tree-optimization/79390
11211 * target.h (struct noce_if_info): Declare.
11212 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11213 * target.def (noce_conversion_profitable_p): New target hook.
11214 * ifcvt.h (struct noce_if_info): New type, moved from ...
11215 * ifcvt.c (struct noce_if_info): ... here.
11216 (noce_conversion_profitable_p): Renamed to ...
11217 (default_noce_conversion_profitable_p): ... this. No longer
11219 (noce_try_store_flag_constants, noce_try_addcc,
11220 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11221 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11222 instead of noce_conversion_profitable_p.
11223 * config/i386/i386.c: Include ifcvt.h.
11224 (ix86_option_override_internal): Don't override
11225 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11226 (ix86_noce_conversion_profitable_p): New function.
11227 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11228 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11229 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11230 * doc/tm.texi: Regenerated.
11232 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11234 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11237 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11240 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11241 instructions for small multiply cores.
11243 2017-04-04 Jeff Law <law@redhat.com>
11245 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11247 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11248 that are not set by the loop over the elements.
11250 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11253 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11254 int mode, convert_modes it to mode as unsigned, otherwise use
11255 lowpart_subreg to mode rather than SImode.
11256 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11257 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11258 Use DImode instead of SImode for the shift count operand.
11259 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11262 2017-04-04 Richard Biener <rguenther@suse.de>
11264 PR middle-end/80281
11265 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11266 arithmetic done for the negate or the plus. Simplify.
11267 (A - (-B) -> A + B): Likewise.
11268 * fold-const.c (split_tree): Make sure to not negate pointers.
11270 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11272 PR rtl-optimization/60818
11273 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11274 a compare of comparisons with the thing compared if this results
11275 in a different machine mode.
11277 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11279 * alias.c (base_alias_check): Fix typo in comment.
11280 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11281 * cgraphunit.c (symbol_table::compile): Likewise.
11282 * collect2.c (maybe_run_lto_and_relink): Likewise.
11283 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11284 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11285 * config/avr/avr.c (avr_map_op_t): Likewise.
11286 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11287 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11288 * config/epiphany/epiphany.md (movcc): Likewise.
11289 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11290 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11292 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11293 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11294 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11295 * config/sh/sh.c (sh_rtx_costs): Likewise.
11296 * fold-const.c (fold_truth_andor): Likewise.
11297 * genautomata.c (collapse_flag): Likewise.
11298 * gengtype.h (struct type::u::s): Likewise.
11299 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11300 * input.c (FORMAT_AMOUNT): Likewise.
11301 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11302 (known_aggs_to_agg_replacement_list): Likewise.
11303 * ipa-inline-analysis.c: Likewise.
11304 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11305 * ipa-polymorphic-call.c
11306 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11307 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11308 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11310 * modulo-sched.c (apply_reg_moves): Likewise.
11311 * omp-expand.c (build_omp_regions_1): Likewise.
11312 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11313 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11314 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11315 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11316 * value-prof.c: Likewise.
11317 * var-tracking.c (val_reset): Likewise.
11319 2017-04-03 Richard Biener <rguenther@suse.de>
11321 PR tree-optimization/80275
11322 * fold-const.c (split_address_to_core_and_offset): Handle
11325 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11327 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11328 descriptors is at least equal to that of functions.
11330 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11332 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11334 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11337 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11338 (mov<IMOD4:mode>): New expander.
11339 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11341 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11343 PR rtl-optimization/79405
11344 * fwprop.c (propagations_left): New variable.
11345 (forward_propagate_into): Decrement it.
11346 (fwprop_init): Initialize it.
11347 (fw_prop): If the variable has reached zero, stop propagating.
11348 (fwprop_addr): Ditto.
11350 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11353 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11354 a FUNCTION_DECL, pass it as decl instead of origin to
11357 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11359 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11362 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11365 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11366 TARGET_VSX_SMALL_INTEGER.
11368 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11370 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11371 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11373 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11375 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11376 extraction from odd-numbered MSA register.
11378 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11380 PR middle-end/80173
11381 * expmed.c (store_bit_field_1): Don't attempt to create
11382 a word subreg out of hard registers wider than word if they
11383 have HARD_REGNO_NREGS of 1 for their mode.
11385 PR middle-end/80163
11386 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11387 conversions to integer types wider than word and pointer.
11390 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11391 (rtx_equal_for_cselib_p): Pass 0 to it.
11392 * cselib.c (cselib_hasher::equal): Likewise.
11393 (rtx_equal_for_cselib_1): Add depth argument. If depth
11394 is 128, don't look up VALUE locs and punt. Increment
11395 depth in recursive calls when walking VALUE locs.
11397 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11399 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11400 (make_gcov_file_name): Use the canonical path name for generating
11402 (read_line): Fix handling of files with ascii null bytes.
11404 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11406 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11407 to initialise a vector register instead
11408 of using a const_int.
11410 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11412 PR translation/80189
11413 * gimplify.c (omp_default_clause): Use %qs instead of %s in
11414 diagnostic messages.
11416 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
11419 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
11420 (dfp_diex_<mode>): Update mode of operand 1.
11421 * doc/extend.texi (dxex, dxexq): Document change to return type.
11422 (diex, diexq): Document change to argument type.
11424 2017-03-30 Martin Jambor <mjambor@suse.cz>
11427 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
11428 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
11429 it reflects the signature changes performed at the callee side.
11430 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
11431 to cgraph_build_function_type_skip_args.
11432 (build_function_decl_skip_args): Adjust call to the above function.
11434 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11437 * config/i386/sse.md
11438 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
11439 register as dest whenever it is a MEM not rtx_equal_p to the
11440 corresponding dup operand, and when forcing into reg move the
11441 reg into the memory afterwards.
11442 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
11443 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
11444 for the force_reg mode.
11445 (avx512vl_vextractf128<mode>): Use register as dest either
11446 always when a MEM, or when it is a MEM not rtx_equal_p to the
11447 corresponding dup operand, or even not when it is a CONST_VECTOR
11448 depending on the mode and lo vs. hi.
11449 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
11451 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
11452 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
11453 Likewise. Require that operands[2] is even.
11454 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
11455 Remove extraneous parens. Require that operands[2] is a multiple
11457 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
11458 operands[0] is a MEM if <mask_applied>, the predicates/constraints
11459 disallow memory then.
11461 2017-03-30 Richard Biener <rguenther@suse.de>
11463 PR tree-optimization/77498
11464 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
11465 to non-constants over backedges.
11467 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
11469 PR rtl-optimization/80233
11470 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
11471 as last_combined_insn. Do not test for BARRIER_P separately.
11473 2017-03-29 Andreas Schwab <schwab@suse.de>
11476 * calls.c (prepare_call_address): Convert funexp to Pmode before
11477 copying to temp reg.
11479 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11481 PR tree-optimization/80158
11482 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
11483 Handle possible future case of more than one alternate
11485 (replace_rhs_if_not_dup): Likewise.
11486 (replace_one_candidate): Likewise.
11488 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
11490 PR rtl-optimization/80193
11491 * ira.c (ira): Do not check allocation for LRA.
11493 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
11495 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
11496 (nvptx_output_simt_exit): Declare.
11497 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
11498 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
11499 (init_softstack_frame): Move initialization of crtl->is_leaf to...
11500 (nvptx_declare_function_name): ...here. Emit declaration of local
11501 memory space buffer for omp_simt_enter insn.
11502 (nvptx_output_unisimt_switch): New.
11503 (nvptx_output_softstack_switch): New.
11504 (nvptx_output_simt_enter): New.
11505 (nvptx_output_simt_exit): New.
11506 * config/nvptx/nvptx.h (struct machine_function): New fields
11507 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
11508 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
11509 (UNSPECV_SIMT_EXIT): Ditto.
11510 (omp_simt_enter_insn): New insn.
11511 (omp_simt_enter): New expansion.
11512 (omp_simt_exit): New insn.
11513 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
11515 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
11516 (expand_GOMP_SIMT_ENTER_ALLOC): New.
11517 (expand_GOMP_SIMT_EXIT): New.
11518 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
11519 (GOMP_SIMT_ENTER_ALLOC): Ditto.
11520 (GOMP_SIMT_EXIT): Ditto.
11521 * target-insns.def (omp_simt_enter): New insn.
11522 (omp_simt_exit): Ditto.
11523 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
11525 (lower_rec_simd_input_clauses): Implement SIMT privatization.
11526 (lower_rec_input_clauses): Likewise.
11527 (lower_lastprivate_clauses): Handle SIMT privatization.
11529 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
11530 (ompdevlow_adjust_simt_enter): New.
11531 (find_simtpriv_var_op): New.
11532 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
11533 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
11535 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
11536 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
11537 (copy_decl_for_dup_finish): Ditto.
11539 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
11541 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
11544 * config/i386/i386.c (ix86_option_override_internal): Always
11545 allow -mpreferred-stack-boundary=3 for 64-bit targets.
11547 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11549 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
11551 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11553 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
11554 mark new edge's irreducible flag accordign to it.
11555 (vect_do_peeling): Check loop preheader edge's irreducible flag
11556 and pass it to function slpeel_add_loop_guard.
11558 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
11560 PR tree-optimization/80218
11561 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
11562 Update block frequencies and counts.
11564 2017-03-28 Richard Biener <rguenther@suse.de>
11566 PR tree-optimization/78644
11567 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
11568 of a simplification result we may not use it at all.
11570 2017-03-28 Richard Biener <rguenther@suse.de>
11573 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
11574 without arguments, generate default definition of a SSA name.
11576 2017-03-28 Richard Biener <rguenther@suse.de>
11578 PR middle-end/80222
11579 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
11580 TYPE_REF_CAN_ALIAS_ALL references.
11581 * fold-const.c (fold_indirect_ref_1): Likewise.
11583 2017-03-28 Martin Liska <mliska@suse.cz>
11586 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
11587 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
11589 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11590 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
11592 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
11593 (EXTRA_SPECS): Define.
11594 (SUBTARGET_EXTRA_SPECS): Likewise.
11595 (SUBTARGET_CPP_SPEC): Likewise.
11596 * config/arc/elf.h (EXTRA_SPECS): Renamed to
11597 SUBTARGET_EXTRA_SPECS.
11598 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
11600 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11602 * config/arc/simdext.md (vst64_insn): Update pattern.
11603 (vld32wh_insn): Likewise.
11604 (vld32wl_insn): Likewise.
11605 (vld64_insn): Likewise.
11606 (vld32_insn): Likewise.
11608 2017-03-28 Marek Polacek <polacek@redhat.com>
11611 * fold-const.c (fold_comparison): Use protected_set_expr_location
11612 instead of SET_EXPR_LOCATION.
11614 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
11616 * tree.c (add_expr): Avoid name lookup warning.
11618 2017-03-27 Jeff Law <law@redhat.com>
11620 PR tree-optimization/80216
11621 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
11622 function name. Limit recursion depth.
11623 (record_temporary_equivalences): Corresponding changes.
11625 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
11627 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
11630 2017-03-27 Jakub Jelinek <jakub@redhat.com>
11633 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
11635 * cfgcleanup.c (reg_note_cfa_p): New array.
11636 (insns_have_identical_cfa_notes): New function.
11637 (old_insns_match_p): Don't cross-jump in between /f
11638 and non-/f instructions. If both i1 and i2 are frame related,
11639 verify all CFA notes, their order and content.
11641 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
11644 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
11645 HImode and SImode with zero extend to DImode to one insn.
11646 (bswap<mode>2_extenddi): Likewise.
11647 (bswapsi2_extenddi): Likewise.
11648 (bswaphi2_extendsi): Likewise.
11649 (bswaphi2): Combine bswap HImode and SImode into one insn.
11650 Separate memory insns from swapping register.
11651 (bswapsi2): Likewise.
11652 (bswap<mode>2): Likewise.
11653 (bswaphi2_internal): Delete, no longer used.
11654 (bswapsi2_internal): Likewise.
11655 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
11656 store, and gpr<-gpr swap insns.
11657 (bswap<mode>2_store): Likewise.
11658 (bswaphi2_reg): Register only splitter, combine with the splitter.
11659 (bswaphi2 splitter): Likewise.
11660 (bswapsi2_reg): Likewise.
11661 (bswapsi2 splitter): Likewise.
11662 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
11663 the insns into load, store, and register/register insns.
11664 (bswapdi2_ldbrx): Likewise.
11665 (bswapdi2_load): Likewise.
11666 (bswapdi2_store): Likewise.
11667 (bswapdi2_reg): Likewise.
11669 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
11671 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
11672 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
11674 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11677 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
11679 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
11680 special handling for target option conflicts between dform
11681 options (-mpower9-dform, -mpower9-dform-vector,
11682 -mpower9-dform-scalar) and -mno-direct-move.
11684 2017-03-27 Richard Biener <rguenther@suse.de>
11686 PR tree-optimization/80181
11687 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
11689 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
11691 * config/arc/predicates.md (move_double_src_operand): Replace the
11692 call to move_double_src_operand with a call to address_operand.
11694 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
11696 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
11697 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
11698 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
11700 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
11702 * config/arc/predicates.md (long_immediate_loadstore_operand):
11703 Consider scaled addresses cases.
11705 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
11707 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
11708 restored when in interrupt.
11709 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
11710 doesn't have delay slot.
11712 2017-03-27 Richard Biener <rguenther@suse.de>
11715 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
11716 inlined thunk clones.
11718 2017-03-27 Jakub Jelinek <jakub@redhat.com>
11721 * asan.c (instrument_derefs): Copy over last operand from
11722 original COMPONENT_REF to the new COMPONENT_REF with
11723 DECL_BIT_FIELD_REPRESENTATIVE.
11724 * ubsan.c (instrument_object_size): Likewise.
11726 2017-03-27 Richard Biener <rguenther@suse.de>
11728 PR tree-optimization/80170
11729 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
11730 sure DR/SCEV didnt fold in constants we do not see when looking
11731 at the reference base alignment.
11733 2017-03-27 Richard Biener <rguenther@suse.de>
11735 PR middle-end/80171
11736 * gimple-fold.c (fold_ctor_reference): Properly guard against
11737 NULL return value from canonicalize_constructor_val.
11739 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
11742 * config/i386/i386.c (ix86_expand_builtin)
11743 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
11744 flags reg setting and flags reg using instructions.
11745 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
11746 clobbering instructions to zero extend op2.
11748 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
11750 * doc/install.texi (Configuration) <--with-aix-soname>:
11751 Update link to AIX ld.
11753 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
11755 PR rtl-optimization/80160
11756 PR rtl-optimization/80159
11757 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
11758 reg_alternate_class into account.
11760 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
11763 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
11764 to consider in curr_insn_transform.
11766 2017-03-24 Jakub Jelinek <jakub@redhat.com>
11768 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
11769 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
11770 and emit_mode_inner.
11772 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11774 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
11775 argument to the overloaded builtin variants. Use the new flag to
11776 deprecate certain builtin variants.
11777 * config/s390/s390-builtin-types.def: Add new builtin types.
11778 * config/s390/s390-builtins.h: Support new flags field for
11779 overloaded builtins.
11780 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
11781 (s390_macro_to_expand): Enable vector float data type.
11782 (s390_cpu_cpp_builtins_internal): Indicate support of the new
11783 builtins by incrementing the __VEC__ version number.
11784 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
11786 (s390_resolve_overloaded_builtin): Emit error messages depending
11787 on the builtin flags.
11788 * config/s390/s390.c (s390_expand_builtin): Support additional
11789 flags argument. Change error message to match the messages
11790 emitted in s390-c.c.
11791 * config/s390/s390.md: New UNSPEC_* constants.
11792 (op_type): Add new instruction types.
11793 * config/s390/vecintrin.h: Add new builtins and test data class
11795 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
11796 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
11797 (VEC_INEXACT, VEC_NOINEXACT): New constants.
11798 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
11799 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
11800 ("vec_mergel<mode>"): V_HW -> VEC_HW.
11802 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
11803 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
11804 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
11805 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
11807 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
11808 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
11809 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
11810 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
11812 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
11813 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
11814 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
11815 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
11816 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
11817 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
11818 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
11820 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
11821 ("vec_scatter_element<V_HW_4:mode>_DI")
11822 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
11823 ("vec_fpint<mode>", "vflls")
11824 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
11825 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
11826 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
11827 ("*vec_cmphe<mode>_cc"): ... these.
11829 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
11830 mode constant instead of magic value.
11832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11834 * config/s390/s390.c (s390_expand_vec_compare): Support other
11835 vector floating point modes than just V2DF.
11836 (s390_expand_vcond): Likewise.
11837 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
11838 (s390_cannot_change_mode_class): Prevent mode changes between TF
11839 and V1TF in vector registers.
11840 * config/s390/s390.md (DF, SF): New mode attributes.
11841 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
11842 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
11843 SFmode support for VRs.
11844 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
11846 (VFT, VF_HW): New mode iterators.
11847 (vw, sdx): New mode attributes.
11848 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
11849 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
11850 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
11851 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
11852 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
11853 also the new vector floating point modes. Renaming to ...
11855 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
11856 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
11857 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
11858 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
11859 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
11860 ("vec_unordered<mode>"): ... these.
11862 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
11863 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
11864 ("*vec_extendv2df"): New insn definitions.
11866 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11868 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
11869 ("mulditi3_2", "*muldi3_sign"): New patterns.
11870 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
11871 rename the pattern definition.
11873 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11875 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
11877 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
11879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11881 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
11882 instruction if possible.
11883 * config/s390/vector.md (vec_halfnumelts): New mode
11885 ("*vec_vllezlf<mode>"): New pattern.
11887 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11889 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
11890 ("popcountv4si2", "popcountv2di2"): Rename to ...
11891 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
11892 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
11894 ("popcount<mode>2_vxe"): New pattern.
11896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11898 * common/config/s390/s390-common.c (processor_flags_table): Add
11900 * config.gcc: Add arch12.
11901 * config/s390/driver-native.c (s390_host_detect_local_cpu):
11902 Default to arch12 for unknown CPU model numbers.
11903 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
11904 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
11905 PROCESSOR_max sanity check.
11906 * config/s390/s390-opts.h (enum processor_type): Add
11908 * config/s390/s390.c (processor_table): Add arch12.
11909 (s390_expand_builtin): Add check for B_VXE flag.
11910 (s390_issue_rate): Add PROCESSOR_ARCH12.
11911 (s390_get_sched_attrmask): Likewise.
11912 (s390_get_unit_mask): Likewise.
11913 (s390_sched_score): Enable z13 scheduling for arch12.
11914 (s390_sched_reorder): Likewise.
11915 (s390_sched_variable_issue): Likewise.
11916 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
11918 (s390_tune_attr): Use z13 scheduling also for arch12.
11919 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
11920 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
11921 (TARGET_VXE_P): New macros.
11922 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
11923 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
11924 * config/s390/s390.opt: Add arch12 as processor_type.
11926 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11928 * config/s390/s390.md
11929 ("fixuns_truncdddi2", "fixuns_trunctddi2")
11930 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
11931 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
11933 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
11934 Rename expanders to ...
11936 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
11937 ("fixuns_truncdddi2_emu"): ... these.
11939 ("fixuns_trunc<mode>si2_emu"): New expander.
11941 ("*fixuns_truncdfdi2_z13"): Rename to ...
11942 ("*fixuns_truncdfdi2_vx"): ... this.
11944 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11946 * config/s390/2964.md: Remove the single element vector compare
11947 instructions which are no longer used.
11948 * config/s390/s390.c (s390_select_ccmode): Remove handling of
11950 (s390_canonicalize_comparison): Remove handling of DFmode
11952 (s390_expand_vec_compare_scalar): Remove function.
11953 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
11954 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
11956 ("*cmp<mode>_ccs"): Add wfcdb instruction.
11958 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11960 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
11962 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
11963 will anyway by matched by mov<mode>_64dfp.
11965 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11967 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
11968 vlef/vstef. Add missing operand to vleif.
11970 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11972 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
11973 pair for all vector types with 64 bit elements.
11974 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
11975 * config/s390/vector.md (V_HW_64): ... here.
11976 (V_128_NOSINGLE): New mode iterator.
11977 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
11978 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
11979 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
11980 ("*vec_load_pairv2di"): Change to ...
11981 ("*vec_load_pair<mode>"): ... this one.
11983 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11985 * config/s390/constraints.md: Add comments.
11986 (jKK): Reject element sizes > 8 bytes.
11987 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
11989 * config/s390/s390.md: Add the s_operand checks formerly in
11990 s390_split_ok_p to various splitters where they are still
11992 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
11993 for 128 bit vectors. Plus two splitters.
11995 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11997 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12000 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12003 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12004 error if the boundary argument is not constant.
12006 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12008 PR rtl-optimization/80112
12009 * loop-doloop.c (doloop_condition_get): Don't check condition
12010 if cmp isn't SET with IF_THEN_ELSE src.
12012 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12014 PR tree-optimization/80158
12015 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12016 replacing a candidate statement, also replace it for the
12017 candidate's alternate interpretation.
12018 (replace_rhs_if_not_dup): Likewise.
12019 (replace_one_candidate): Likewise.
12021 2017-03-24 Richard Biener <rguenther@suse.de>
12023 PR tree-optimization/80167
12024 * graphite-isl-ast-to-gimple.c
12025 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12027 (translate_isl_ast_to_gimple::get_rename): Likewise.
12029 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12031 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12032 handling of certain combinations of target options, including the
12033 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12034 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12036 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12039 * config/arm/arm.md (*load_multiple): Add reload_completed to
12040 matching condition.
12042 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12043 Richard Biener <rguenth@suse.de>
12045 PR tree-optimization/79908
12046 PR tree-optimization/80136
12047 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12048 been cast away, gimplify_and_add suffices.
12050 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12052 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12054 2017-03-23 Richard Biener <rguenther@suse.de>
12056 PR tree-optimization/80032
12057 * gimplify.c (gimple_push_cleanup): Forced unconditional
12058 cleanups still have to go to the conditional_cleanups
12061 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12063 PR tree-optimization/80072
12064 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12066 (next_operand_entry_id): Change type to unsigned int.
12067 (sort_by_operand_rank): Make sure to return the right return value
12068 even if unsigned fields are bigger than INT_MAX.
12069 (struct oecount): Change cnt and id type to unsigned int.
12070 (oecount_hasher::equal): Formatting fix.
12071 (oecount_cmp): Make sure to return the right return value
12072 even if unsigned fields are bigger than INT_MAX.
12073 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12076 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12077 TREE_READONLY on result if writing it more than once.
12080 * doc/invoke.texi (-fsanitize=thread): Document that with
12081 -fnon-call-exceptions atomics are not able to throw
12085 * tsan.c: Include tree-eh.h.
12086 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12087 maybe_clean_or_replace_eh_stmt where needed.
12088 (instrument_memory_accesses): Add cfg_changed argument.
12089 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12090 if it returned true.
12091 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12093 PR rtl-optimization/63191
12094 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12095 wrapper function, moved the whole old content into ...
12096 (ix86_delegitimize_address_1): ... this. New inline function.
12097 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12098 true as last argument instead of ix86_delegitimize_address.
12100 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12102 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12103 cortexa57_branch_cost.
12105 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12107 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12109 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12112 * doc/md.texi (Constraints): Document wA constraint.
12113 * config/rs6000/constraints.md (wA): New.
12114 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12115 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12116 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12117 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12119 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12122 * gimplify.c (is_oacc_declared): New function.
12123 (oacc_default_clause): Use it to set default flags for acc declared
12124 variables inside parallel regions.
12125 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12126 declared variables.
12127 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12128 declare attribute to any decl as necessary.
12130 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12133 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12134 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12135 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12136 (arm_arch_lpae): This.
12137 * config/arm/arm.c (arm_arch7ve): Rename into ...
12138 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12139 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12142 2017-03-22 Martin Liska <mliska@suse.cz>
12145 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12146 error message instead of an ICE.
12148 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12150 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12152 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12154 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12156 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12159 2017-03-21 Martin Sebor <msebor@redhat.com>
12161 * doc/extend.texi: Use "cannot" instead of "can't."
12162 * doc/hostconfig.texi: Same.
12163 * doc/install.texi: Same.
12164 * doc/invoke.texi: Same.
12165 * doc/loop.texi: Same.
12166 * doc/md.texi: Same.
12167 * doc/objc.texi: Same.
12168 * doc/rtl.texi: Same.
12169 * doc/tm.texi: Same.
12170 * doc/tm.texi.in: Same.
12171 * doc/trouble.texi: Same.
12173 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12176 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12177 (collect_checksum_attributes): Set it.
12178 (die_checksum_ordered): Use it.
12180 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12182 PR tree-optimization/79908
12183 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12184 change: For a VA_ARG whose LHS has been cast away, use
12185 force_gimple_operand to construct the side effects.
12187 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12189 PR translation/80001
12190 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12191 more amenable to translation.
12192 (oacc_loop_auto_partitions): Likewise.
12194 2017-03-21 Marek Polacek <polacek@redhat.com>
12195 Martin Sebor <msebor@redhat.com>
12197 PR tree-optimization/80109
12198 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12199 on INTEGRAL_TYPE_P.
12201 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12202 Segher Boessenkool <segher@kernel.crashing.org>
12205 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12206 check reg_used_between_p between insn and one of succ or succ2
12207 depending on if succ is artificial insn not inserted into insn
12210 2017-03-21 Martin Liska <mliska@suse.cz>
12212 PR gcov-profile/80081
12213 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12214 * doc/gcc.texi: Include gcov-dump stuff.
12215 * doc/gcov-dump.texi: New file.
12217 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12219 PR rtl-optimization/79150
12220 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12221 conditional jump, if the jump is the last insn of the loop.
12223 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12224 Richard Biener <rguenth@suse.de>
12226 PR tree-optimization/79908
12227 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12228 been cast away, use force_gimple_operand to construct the side
12231 2017-03-21 Martin Liska <mliska@suse.cz>
12233 PR libfortran/79956
12234 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12237 2017-03-21 Brad Spengler <spender@grsecurity.net>
12240 * plugin.c (htab_hash_plugin): New function.
12241 (add_new_plugin): Use it and adjust.
12242 (parse_plugin_arg_opt): Adjust.
12243 (init_one_plugin): Likewise.
12245 2017-03-21 Richard Biener <rguenther@suse.de>
12247 PR tree-optimization/80032
12248 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12249 if set force the cleanup to happen unconditionally.
12250 (gimplify_target_expr): Push inserted clobbers with force_uncond
12251 to avoid them being removed by control-dependent DCE.
12253 2017-03-21 Richard Biener <rguenther@suse.de>
12255 PR tree-optimization/80122
12256 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12257 va_arg_pack_len when the inlined call stmt requires pack
12259 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12261 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12264 * tsan.c (instrument_builtin_call): If the memory model argument
12265 is not a constant, assume it is valid.
12268 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12271 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12273 PR rtl-optimization/79910
12274 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12275 if its dest is used by an insn before I2 (other than the combined
12276 insns themselves, which are properly handled already).
12278 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12281 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12283 * combine.c (record_used_regs): New static function.
12284 (try_combine): Handle situations where there is an additional
12285 instruction between I2 and I3 which needs to have a LOG_LINK
12289 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12291 * combine.c (try_combine): Delete redundant i1 test. Call
12292 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12294 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12297 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12298 alternatives 13/14.
12300 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12302 PR tree-optimization/80054
12303 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12304 the optimization if a PHI or any of its arguments is not dominated
12305 by the candidate's basis. Use gphi* rather than gimple* as
12307 (replace_profitable_candidates): Clean up a gimple* variable that
12308 should be a gphi* variable.
12310 2017-03-20 Martin Sebor <msebor@redhat.com>
12313 * doc/extend.texi (attribute constructor): Document present limitation.
12315 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12318 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12319 __POWER9_VECTOR__ #ifdef control, change template definition to
12320 use Power9-specific built-in function.
12321 (vec_any_eq): Likewise.
12322 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12323 to control outcomes from this test.
12324 (vector_ae_<mode>p): For VEC_F modes, likewise.
12326 2017-03-20 Ian Lance Taylor <iant@google.com>
12328 * config/i386/i386.c (ix86_function_regparm): Save an extra
12329 register for -fsplit-stack with DECL_STATIC_CHAIN.
12331 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12334 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12335 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12337 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12339 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12341 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12344 2017-03-20 Marek Polacek <polacek@redhat.com>
12347 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12349 2017-03-20 Richard Biener <rguenther@suse.de>
12351 PR tree-optimization/80113
12352 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12353 allocate extra SSA name for PHI def.
12354 (add_close_phis_to_outer_loops): Likewise.
12355 (add_close_phis_to_merge_points): Likewise.
12356 (copy_loop_close_phi_args): Likewise.
12357 (copy_cond_phi_nodes): Likewise.
12359 2017-03-20 Martin Liska <mliska@suse.cz>
12361 PR middle-end/79753
12362 * tree-chkp.c (chkp_build_returned_bound): Do not build
12363 returned bounds for a LHS that's not a BOUNDED_P type.
12365 2017-03-20 Martin Liska <mliska@suse.cz>
12369 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12370 COMPLEX_CST and VECTOR_CST.
12372 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12375 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12376 target operand. A new splitter adds the clobber statement in case
12377 the target operand is dead anyway.
12379 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12381 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12382 to age-old versions of binutils and glibc.
12384 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12386 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12388 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12390 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12392 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12394 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12395 requirement for binutils 2.13.
12397 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12399 * combine.c (try_combine): Delete redundant i1 test. Call
12400 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12402 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12404 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12405 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12407 <riscv64-*-elf>: Re-arrange section
12408 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12409 <riscv32-*-linux>: Likewise.
12410 <riscv64-*-elf>: Likewise
12411 <riscv64-*-linux>: Likewise.
12413 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
12416 * aarch64.opt(verbose-cost-dump): Fix typo.
12418 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
12421 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
12422 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
12424 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12426 * reload.c (find_reloads): When reloading a nonoffsettable address,
12427 use RELOAD_OTHER for it and its address reloads.
12429 PR rtl-optimization/79910
12430 * combine.c (record_used_regs): New static function.
12431 (try_combine): Handle situations where there is an additional
12432 instruction between I2 and I3 which needs to have a LOG_LINK
12435 2017-03-17 Jeff Law <law@redhat.com>
12437 PR tree-optimization/71437
12438 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
12439 conditional in the hash table first.
12440 (vrp_dom_walker::before_dom_children): Extract condition from
12441 ASSERT_EXPR. Record condition, its inverion and any implied
12442 conditions as well.
12444 2017-03-17 Marek Polacek <polacek@redhat.com>
12445 Markus Trippelsdorf <markus@trippelsdorf.de>
12447 PR tree-optimization/80079
12448 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
12451 2017-03-17 Richard Biener <rguenther@suse.de>
12453 PR middle-end/80075
12454 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
12455 Properly verify the LHS before the RHS possibly claims to be
12457 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
12460 2017-03-17 Martin Jambor <mjambor@suse.cz>
12462 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
12463 (List of -O2 options): Likewise.
12464 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
12467 2017-03-17 Tom de Vries <tom@codesourcery.com>
12469 * gcov-dump.c (print_usage): Print bug_report_url.
12471 2017-03-17 Richard Biener <rguenther@suse.de>
12473 PR middle-end/80050
12474 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
12475 (parser::peek): Likewise.
12477 2017-03-17 Richard Biener <rguenther@suse.de>
12479 PR tree-optimization/80048
12480 * sese.c (free_sese_info): Properly release rename_map and
12481 copied_bb_map elements.
12483 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
12485 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
12486 Add linked-list forward and backlinks. Insert on
12487 construction, remove on destruction.
12488 (class pass_store_merging): Add m_stores_head field.
12489 (pass_store_merging::terminate_and_process_all_chains):
12490 Iterate over m_stores_head list.
12491 (pass_store_merging::terminate_all_aliasing_chains):
12493 (pass_store_merging::execute): Check for debug stmts first.
12494 Push new chains onto the m_stores_head stack.
12496 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
12499 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
12500 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
12501 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
12503 2017-03-16 Jeff Law <law@redhat.com>
12505 PR tree-optimization/71437
12506 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
12507 member function. Implementation moved into after_dom_children
12508 member function and into the threader's thread_outgoing_edges
12510 (dom_opt_dom_walker::after_dom_children): Simplify by moving
12511 some code into new thread_outgoing_edges.
12512 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
12513 definition. Simplify marker handling (do it here). Assume we always
12514 have the available expression and the const/copies tables.
12515 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
12517 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
12518 * tree-vrp.c (equiv_stack): No longer file scoped.
12519 (vrp_dom_walker): New class.
12520 (vrp_dom_walker::before_dom_children): New member function.
12521 (vrp_dom_walker::after_dom_children): Likewise.
12522 (identify_jump_threads): Setup domwalker. Use it rather than
12523 walking edges in a random order by hand. Simplify setup/finalization.
12524 (finalize_jump_threads): Remove.
12525 (vrp_finalize): Do not call identify_jump_threads here.
12526 (execute_vrp): Do it here instead and call thread_through_all_blocks
12529 PR tree-optimization/71437
12530 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
12532 (simplify_stmt_for_jump_threading): Add basic_block argument. All
12534 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
12535 (dom_opt_dom_walker::thread_across_edge): Remove
12536 handle_dominating_asserts argument. All callers changed.
12537 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
12538 changes. Remove calls to lhs_of_dominating_assert. Other
12539 uses of handle_dominating_asserts turn into unconditional code
12540 (simplify_control_stmt_condition_1): Likewise.
12541 (simplify_control_stmt_condition): Likewise.
12542 (thread_through_normal_block, thread_across_edge): Likewise.
12543 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
12544 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
12545 object if it is not an SSA_NAME.
12546 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
12547 before calling into the VRP specific simplifiers.
12548 (identify_jump_threads): Remove handle_dominating_asserts
12551 2017-03-16 Jakub Jelinek <jakub@redhat.com>
12554 * tree-diagnostic.c (default_tree_printer): No longer static.
12555 * tree-diagnostic.h (default_tree_printer): New prototype.
12557 2017-03-16 Tamar Christina <tamar.christina@arm.com>
12559 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
12560 Change ins into fmov.
12562 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12564 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
12565 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
12566 Use h_con constraint for operand 1.
12567 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
12568 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
12570 2017-03-15 Jeff Law <law@redhat.com>
12572 PR tree-optimization/71437
12573 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
12574 (record_temporary_equivalences): Use it.
12576 PR tree-optimization/71437
12577 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
12578 tree-ssa-scopedtables.
12579 (lookup_avail_expr, build_and_record_new_cond): Likewise.
12580 (record_conditions, record_cond, vuse_eq): Likewise.
12581 (record_edge_info): Adjust to API tweak of record_conditions.
12582 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
12583 (record_temporary_equivalences, optimize_stmt): Likewise.
12584 (eliminate_redundant_computations): Likewise.
12585 (record_equivalences_from_stmt): Likewise.
12586 * tree-ssa-scopedtables.c: Include options.h and params.h.
12587 (vuse_eq): New function, moved from tree-ssa-dom.c
12588 (build_and_record_new_cond): Likewise.
12589 (record_conditions): Likewise. Accept vector of conditions rather
12590 than edge_equivalence structure for first argument.
12591 for the first argument.
12592 (avail_exprs_stack::lookup_avail_expr): New member function, moved
12593 from tree-ssa-dom.c.
12594 (avail_exprs_stack::record_cond): Likewise.
12595 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
12596 from tree-ssa-dom.c.
12597 (avail_exprs_stack): Add new member functions lookup_avail_expr
12599 (record_conditions): Declare.
12601 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
12604 * lra-constraints.c (process_alt_operands): Increase reject for
12605 reloading an input/output operand.
12607 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
12610 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
12611 insns to convert from signed/unsigned char/short to IEEE 128-bit
12613 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
12615 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
12618 * config/i386/i386.c (ix86_vector_duplicate_value): Create
12619 subreg of inner mode for values already in registers.
12621 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
12623 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
12624 iteration reg is used after the loop.
12626 2017-03-14 Martin Sebor <msebor@redhat.com>
12628 PR tree-optimization/79800
12629 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
12630 precision in negative-positive range.
12631 (format_floating): Call non-const overload with adjusted precision.
12633 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12636 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
12639 2017-03-14 Martin Sebor <msebor@redhat.com>
12641 PR middle-end/80020
12642 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
12643 * builtins.def (aligned_alloc): Use it.
12646 * Makefile.in (GTFILES): Add calls.c.
12647 * calls.c: Include "gt-calls.h".
12649 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
12651 PR rtl-optimization/79728
12652 * regs.h (struct target_regs): New field
12653 x_contains_allocatable_regs_of_mode.
12654 (contains_allocatable_regs_of_mode): New macro.
12655 * reginfo.c (init_reg_sets_1): Initialize it, and change
12656 contains_reg_of_mode so it includes global regs as well.
12657 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
12658 rather than contains_regs_of_mode.
12660 2017-03-14 Martin Liska <mliska@suse.cz>
12662 * doc/invoke.texi: Document options that can't be combined with
12663 -fcheck-pointer-bounds.
12665 2017-03-14 Martin Liska <mliska@suse.cz>
12667 PR middle-end/79831
12668 * doc/invoke.texi (-Wchkp): Document the option.
12670 2017-03-14 Martin Liska <mliska@suse.cz>
12672 * Makefile.in: Install gcov-dump.
12674 2017-03-14 Martin Liska <mliska@suse.cz>
12676 * multiple_target.c (expand_target_clones): Bail out for
12677 an invalid attribute.
12679 2017-03-14 Richard Biener <rguenther@suse.de>
12681 * alias.c (struct alias_set_entry): Pack properly.
12682 * cfgloop.h (struct loop): Likewise.
12683 * cse.c (struct set): Likewise.
12684 * ipa-utils.c (struct searchc_env): Likewise.
12685 * loop-invariant.c (struct invariant): Likewise.
12686 * lra-remat.c (struct cand): Likewise.
12687 * recog.c (struct change_t): Likewise.
12688 * rtl.h (struct address_info): Likewise.
12689 * symbol-summary.h (function_summary): Likewise.
12690 * tree-loop-distribution.c (struct partition): Likewise.
12691 * tree-object-size.c (struct object_size_info): Likewise.
12692 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
12693 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
12694 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
12695 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
12696 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
12697 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
12698 (struct _stmt_vec_info): Likewise.
12700 2017-03-14 Martin Liska <mliska@suse.cz>
12703 * multiple_target.c (create_dispatcher_calls): Check that
12704 a target can create a function dispatcher.
12706 2017-03-14 Martin Liska <mliska@suse.cz>
12709 * multiple_target.c (expand_target_clones): Drop local.local
12710 flag for default implementation.
12712 2017-03-14 Richard Biener <rguenther@suse.de>
12714 PR tree-optimization/80030
12715 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
12717 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
12719 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
12720 gcc_fallthrough() instead of __attribute__((fallthrough));
12722 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
12724 * doc/gcc.texi: Remove "up" link to (DIR).
12725 * doc/gccint.texi: Ditto.
12727 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
12729 * doc/install.texi (Specific) <avr>: Remove reference to
12732 2017-03-13 Jeff Law <law@redhat.com>
12734 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
12735 attribute rather than comments.
12737 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
12738 match_scratch operand is highest.
12740 2017-03-13 Martin Liska <mliska@suse.cz>
12742 PR middle-end/78339
12743 * ipa-pure-const.c (warn_function_noreturn): If the declarations
12744 is a CHKP clone, use original declaration.
12746 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
12748 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
12749 (arc_conditional_register_usage): Use a different allocation order
12750 when optimizing for size.
12751 * common/config/arc/arc-common.c (arc_option_optimization_table):
12752 Section anchors default on when optimizing for size.
12754 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
12756 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
12758 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
12760 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
12761 * config/arc/arc.md (cpu_facility): Add cd variant.
12762 (*movqi_insn): Add code density variant.
12763 (*movhi_insn): Likewise.
12764 (*movqi_insn): Likewise.
12765 (*addsi3_mixed): Likewise.
12766 (subsi3_insn): Likewise.
12768 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
12770 * config/arc/arc.md (movsi_cond_exec): Update constraint.
12772 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
12774 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
12775 expressions with MINUS and UNARY ops.
12777 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12780 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
12782 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
12783 between vec_select and vector argument.
12784 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
12785 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
12786 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
12787 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
12788 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
12789 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
12791 2017-03-13 Richard Biener <rguenther@suse.de>
12794 * params.def (vect-max-peeling-for-alignment): Fix typo.
12796 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
12798 * doc/install.texi (Specific) <mips-*-*>: Remove description of
12799 issue that only occurred with binutils below 2.18.
12801 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
12803 * doc/install.texi (Specific) <cris-axis-elf>: No longer
12804 refer to binutils 2.11/2.12 minimum.
12806 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
12808 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
12809 ftp.kernel.org and simplify binutils requirement.
12811 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
12813 * doc/invoke.texi (Warning Options): Fix spelling of link-time
12815 (Optimize Options): Ditto. Also remove redundancy.
12817 2017-03-10 David Malcolm <dmalcolm@redhat.com>
12819 PR translation/79848
12820 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
12822 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
12823 to G_ to avoid double translation.
12825 2017-03-10 David Malcolm <dmalcolm@redhat.com>
12827 PR translation/79923
12828 * auto-profile.c (get_combined_location): Convert leading
12829 character of diagnostics to lower case and remove trailing period.
12830 (read_profile): Likewise for various diagnostics.
12831 * config/arm/arm.c (arm_option_override): Remove trailing period
12832 from various diagnostics.
12833 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
12834 (msp430_expand_delay_cycles): Likewise.
12836 2017-03-10 David Malcolm <dmalcolm@redhat.com>
12839 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
12840 full command-line argument, rather than just "str".
12841 (aarch64_validate_march): Likewise.
12842 (aarch64_validate_mtune): Likewise.
12844 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
12846 PR rtl-optimization/78911
12847 * lra-assigns.c (must_not_spill_p): New function.
12848 (spill_for): Use it.
12850 2017-03-10 Jakub Jelinek <jakub@redhat.com>
12852 PR tree-optimization/79981
12853 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
12854 ATOMIC_COMPARE_EXCHANGE ifn result.
12855 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
12856 IFN_ATOMIC_COMPARE_EXCHANGE.
12858 2017-03-10 David Malcolm <dmalcolm@redhat.com>
12861 * opts.c (parse_sanitizer_options): Add missing question mark to
12862 "did you mean" message.
12864 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12866 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
12868 (VMULEUH_UNS): Likewise.
12869 (VMULOUB_UNS): Likewise.
12870 (VMULOUH_UNS): Likewise.
12871 * config/rs6000/rs6000.c (builtin_function_type): Remove
12872 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
12874 2017-03-10 David Malcolm <dmalcolm@redhat.com>
12877 * read-rtl-function.c (function_reader::read_rtx_operand): Update
12878 x with result of extra_parsing_for_operand_code_0.
12879 (function_reader::extra_parsing_for_operand_code_0): Convert
12880 return type from void to rtx, returning x. When reading
12881 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
12882 larger size containing struct block_symbol.
12884 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
12886 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
12887 -mfloat128-hardware without -m64.
12889 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
12892 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
12893 entries to the case statement that marks unsigned arguments to
12894 overloaded functions.
12896 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
12898 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
12899 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
12901 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
12904 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
12905 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
12907 2017-03-10 Martin Liska <mliska@suse.cz>
12911 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
12912 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
12915 2017-03-10 Olivier Hainque <hainque@adacore.com>
12917 * tree-switch-conversion (array_value_type): Start by resetting
12918 candidate type to it's main variant.
12920 2017-03-10 Jakub Jelinek <jakub@redhat.com>
12922 PR rtl-optimization/79909
12923 * combine.c (try_combine): Use simplify_replace_rtx on individual
12924 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
12925 of the whole CALL_INSN_FUNCTION_USAGE.
12927 PR tree-optimization/79972
12928 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
12929 get_range_info on SSA_NAMEs. Formatting fixes.
12931 2017-03-10 Richard Biener <rguenther@suse.de>
12932 Jakub Jelinek <jakub@redhat.com>
12934 PR tree-optimization/77975
12935 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
12936 edge to be constant.
12937 (get_val_for): For constant x return it. Formatting fix.
12938 (loop_niter_by_eval): Avoid pointless looping if the next iteration
12939 would use the same bases as the current one.
12941 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12943 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
12944 instead of vec_select for V1TImode.
12945 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
12947 (VSX_LE_128): Add V1TI to this mode iterator.
12948 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
12949 (*vsx_le_perm_store_<mode>): Likewise.
12950 (pre-reload splitter for VSX stores): Likewise.
12951 (post-reload splitter for VSX stores): Likewise.
12952 (*vsx_xxpermdi2_le_<mode>): Likewise.
12953 (*vsx_lxvd2x2_le_<mode>): Likewise.
12954 (*vsx_stxvd2x2_le_<mode>): Likewise.
12956 2017-03-09 Michael Eager <eager@eagercon.com>
12958 Correct failures with --enable-checking=yes,rtl.
12960 * config/microblaze/microblaze.c (microblaze_expand_shift):
12961 Replace GET_CODE test with CONST_INT_P and INTVAL test with
12962 test for const0_rtx.
12963 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
12964 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
12966 2017-03-09 Richard Biener <rguenther@suse.de>
12968 PR tree-optimization/79977
12969 * graphite-scop-detection.c (scop_detection::merge_sese):
12970 Handle the case of extra exits to blocks dominating the entry.
12972 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
12974 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12977 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
12979 PR rtl-optimization/79949
12980 * lra-constraints.c (process_alt_operands): Check memory when
12981 trying to predict a cycle. Print about the overall increase.
12983 2017-03-09 Richard Biener <rguenther@suse.de>
12985 PR middle-end/79971
12986 * gimple-expr.c (useless_type_conversion_p): Preserve
12987 TYPE_SATURATING for fixed-point types.
12989 2017-03-09 Richard Biener <rguenther@suse.de>
12992 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
12993 alignment of BLKmode params.
12995 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12998 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
12999 (VALL_NO_V2Q): Likewise.
13001 * config/aarch64/aarch64-simd.md
13002 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13004 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13005 VALL_NO_V2Q mode iterator.
13006 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13008 2017-03-09 Martin Liska <mliska@suse.cz>
13010 PR tree-optimization/79631
13011 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13012 tree_int_cst_sign_bit just for INTEGER constants.
13014 2017-03-09 Martin Liska <mliska@suse.cz>
13018 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13021 2017-03-09 Marek Polacek <polacek@redhat.com>
13024 * tree.c (inchash::add_expr): Handle TREE_VEC.
13026 2017-03-09 Martin Liska <mliska@suse.cz>
13029 (chkp_narrow_size_and_offset): New function.
13030 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13031 (void chkp_parse_bit_field_ref): New function.
13032 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13033 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13035 2017-03-09 Martin Liska <mliska@suse.cz>
13038 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13039 (chkp_find_bounds_1): Remove gcc_unreachable.
13041 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13044 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13045 BUILT_IN_SYNC*, determine the access type from the size suffix and
13046 always build a MEM_REF with that type. Handle forgotten
13047 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13050 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13051 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13052 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13053 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13054 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13055 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13056 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13057 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13058 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13059 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13060 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13061 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13062 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13063 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13064 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13065 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13066 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13067 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13068 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13069 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13070 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13071 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13072 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13073 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13074 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13075 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13076 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13077 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13078 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13079 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13080 definitions outside of __OPTIMIZE__ guarded section.
13083 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13084 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13085 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13086 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13089 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13091 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13092 ("vfenez<mode>"): Add missing constraints.
13094 2017-03-08 Martin Sebor <msebor@redhat.com>
13097 * config/nds32/nds32.c (nds32_option_override):
13098 Fix misspelled diagnostic.
13100 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13103 * gimplify.c (gimplify_omp_for): Replace index var in outer
13104 taskloop statement with an artificial variable and add
13105 OMP_CLAUSE_PRIVATE clause for it.
13107 2017-03-08 Richard Biener <rguenther@suse.de>
13109 PR tree-optimization/79955
13110 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13111 for accesses that are completely outside of the variable.
13113 2017-03-08 Andrew Haley <aph@redhat.com>
13115 PR tree-optimization/79943
13116 * tree-ssa-loop-split.c (compute_new_first_bound): When
13117 calculating the new upper bound, (END-BEG) should be added, not
13120 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13122 * config/avr/avr.md (setmemhi): Make sure match_dup
13123 operand number comes before match_scratch.
13125 2017-03-08 Richard Biener <rguenther@suse.de>
13127 PR tree-optimization/79920
13128 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13129 with ncopies == 1 to ...
13130 (vect_transform_slp_perm_load): ... here. Properly compute
13131 all element loads by iterating VF times over the group. Do
13132 not handle ncopies (computed in a broken way) in
13133 vect_create_mask_and_perm.
13135 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13138 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13139 is a uniform vector, use uniform_vector_p return value instead of
13140 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13142 2017-03-07 Marek Polacek <polacek@redhat.com>
13144 PR middle-end/79809
13145 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13146 (alloca_call_type): Likewise.
13148 2017-03-07 Martin Liska <mliska@suse.cz>
13150 * gcov.c (process_args): Put comment to correct location.
13152 2017-03-07 Martin Liska <mliska@suse.cz>
13154 PR middle-end/68270
13155 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13156 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13157 (chkp_narrow_bounds_for_field): Likewise.
13158 (chkp_parse_array_and_component_ref): Pass one more argument to
13161 2017-03-07 Richard Biener <rguenther@suse.de>
13163 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13166 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13168 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13169 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13171 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13174 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13175 to end of description.
13176 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13178 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13180 PR rtl-optimization/79901
13181 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13183 (*avx512f_<code><mode>3<mask_name>): ... this.
13184 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13185 iterator instead of VI8_AVX2_AVX512BW.
13187 PR rtl-optimization/79901
13188 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13189 min/max expander, expand it using expand_vec_cond_expr.
13192 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13195 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13198 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13199 to void * for PCH reasons.
13200 * dwarf2out.c (output_loc_operands, output_die): Cast
13201 v.val_vec.array to unsigned char *.
13203 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13206 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13209 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13211 PR rtl-optimization/79571
13212 * lra-constraints.c (process_alt_operands): Calculate static
13213 reject and subtract it from overall when only addresses will be
13216 2017-03-06 Julia Koval <julia.koval@intel.com>
13219 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13220 incoming stack boundary to 128 for 64-bit targets.
13222 2017-03-06 Richard Biener <rguenther@suse.de>
13224 PR tree-optimization/79894
13225 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13226 to NULL after folding it.
13228 2017-03-06 Richard Biener <rguenther@suse.de>
13230 PR tree-optimization/79824
13231 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13232 check disabling peeling for gaps.
13234 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13236 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13237 attributes): Document gettimeofday.
13239 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13241 * config/s390/s390.c (s390_option_override_internal): Set
13242 PARAM_MIN_VECT_LOOP_BOUND
13244 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13246 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13247 * config/s390/s390.md: Likewise.
13249 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13252 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13253 (<avx2_avx512>_perm<mode>): Rename to ...
13254 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13256 (<avx512>_perm<mode>_mask): Rename to ...
13257 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13259 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13260 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13261 instead of VI8F_256_512.
13262 (avx512f_perm<mode>): New define_expand.
13263 (avx512f_perm<mode>_mask): Likewise.
13264 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13265 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13267 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13269 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13270 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13272 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13274 2017-03-06 Martin Liska <mliska@suse.cz>
13277 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13278 when having a SSA NAME w/o VAR_DECL assigned to it.
13280 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13282 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13283 msa_dpsub_<su>_d): Fix MODE for vec_select.
13285 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13287 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13289 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13291 2017-03-06 Richard Biener <rguenther@suse.de>
13293 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13294 * plugin.c (register_plugin_info): Likewise.
13295 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13297 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13299 * config/i386/sse.md (sse_storehps, sse_storelps,
13300 avx_<castmode><avxsizesuffix>_<castmode>,
13301 avx512f_<castmode><avxsizesuffix>_<castmode>,
13302 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13303 in condition that at least one operand is not a MEM.
13305 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13307 PR middle-end/79805
13308 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13309 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13311 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13312 gimple_call_nothrow_p flag based on whether original builtin can throw.
13313 If it can, emit following stmts on the fallthrough edge.
13314 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13315 don't create new bb if inserting just debug stmts on the edge, try to
13316 insert them on the fallthru bb or just reset debug stmts.
13318 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13321 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13322 restore recog_data (including the operand rtxes inside it) around
13323 the call to get_insn_template.
13325 2017-03-03 Martin Sebor <msebor@redhat.com>
13327 PR tree-optimization/79699
13328 * context.c (context::~context): Free MPFR caches to avoid
13329 a memory leak on program exit.
13331 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13333 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13334 Use wide_int::ulow () instead of .elt (0).
13336 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13338 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13339 (*pushxf): Limit oF constraint to 32bit targets and add oC
13340 constraint for 64bit targets.
13341 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13342 (*pushdf): Change rmF constraint to rmC.
13344 2017-03-03 Martin Liska <mliska@suse.cz>
13346 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13347 Remove unused variable.
13349 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13352 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13353 is a memory operand, increase num_memory.
13354 (ix86_expand_args_builtin): Likewise.
13356 2017-03-03 Jan Hubicka <jh@suse.cz>
13359 * ipa-devirt.c (maybe_record_node): Properly handle
13360 __cxa_pure_virtual visibility.
13362 2017-03-03 Martin Liska <mliska@suse.cz>
13364 PR tree-optimization/79803
13365 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13367 (pass_loop_prefetch::execute): Disabled optimization if an
13368 assumption about L1 cache size is not met.
13370 2017-03-03 Martin Liska <mliska@suse.cz>
13372 PR rtl-optimization/79574
13373 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13374 (hash_scan_set): Likewise.
13375 (dump_hash_table): Likewise.
13376 (hoist_code): Likewise.
13378 2017-03-03 Richard Biener <rguenther@suse.de>
13380 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13381 in place of uhigh ().
13382 (fixed_convert_from_real): Likewise.
13384 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13387 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13389 2017-03-03 Richard Biener <rguenther@suse.de>
13391 PR middle-end/79818
13392 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13393 TYPE_OVERFLOW_UNDEFINED check.
13395 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13397 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13399 (vector_ae_<mode>_p): Likewise.
13400 (vector_nez_<mode>_p): Likewise.
13401 (vector_ne_v2di_p): Likewise.
13402 (vector_ae_v2di_p): Likewise.
13403 (vector_ne_<mode>_p): Likewise.
13404 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13406 (vsx_tsqrt<mode>2_fe): Likewise.
13408 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13411 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
13413 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13415 PR rtl-optimization/79780
13416 * cprop.c (one_cprop_pass): When second and further conditional trap
13417 in a single basic block is turned into an unconditional trap, turn it
13418 into a deleted note to avoid RTL verification failures.
13420 2017-03-02 Richard Biener <rguenther@suse.de>
13422 * fold-const.c (const_binop): Use ulow () instead of elt (0).
13424 2017-03-02 Richard Biener <rguenther@suse.de>
13426 PR tree-optimization/79345
13428 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
13429 param and abort the walk, returning -1 if it is hit.
13430 (walk_aliased_vdefs): Take a limit param and pass it on.
13431 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
13432 defaulting to 0 and return a signed int.
13433 * tree-ssa-uninit.c (struct check_defs_data): New struct.
13434 (check_defs): New helper.
13435 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
13436 about uninitialized memory.
13437 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
13438 bogus uninitialized warning.
13439 (fixed_convert_from_real): Likewise.
13441 2017-03-02 Bin Cheng <bin.cheng@arm.com>
13443 PR tree-optimization/66768
13444 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
13445 iv_use if base object can't be determined.
13447 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13449 PR tree-optimization/79345
13450 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
13451 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
13452 (get_pattern_stats): Initialize it.
13453 * genemit.c (gen_expand): Verify match_scratch numbers come after
13454 match_operand/match_dup numbers.
13455 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
13456 match_scratch numbers.
13457 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
13459 * config/s390/s390.md (trunctdsd2): Likewise.
13461 2017-03-02 Richard Biener <rguenther@suse.de>
13463 * wide-int.h (wide_int_storage::operator=): Implement in terms
13466 2017-03-02 Richard Biener <rguenther@suse.de>
13468 PR tree-optimization/79777
13469 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
13470 the to insert expression to sth existing.
13472 2017-03-01 Martin Sebor <msebor@redhat.com>
13474 PR middle-end/79692
13475 * gimple-ssa-sprintf.c
13476 (directive::known_width_and_precision): New function.
13477 (format_integer): Use it.
13478 (get_mpfr_format_length): Consider the full range of precision
13479 when computing %g output with the # flag. Set the likely byte
13480 count to 3 rather than 1 when precision is indeterminate.
13481 (format_floating): Correct the lower bound of precision.
13483 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13485 * doc/invoke.texi: Document default code model for 64-bit Linux.
13487 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13490 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
13491 udiv rather than div since input pattern is unsigned.
13493 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
13495 * config/i386/i386.c (print_reg): Warn for values of
13496 unsupported size in integer register.
13498 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
13501 * config/rs6000/predicates.md (current_file_function_operand): Do
13502 not allow self calls to be local if the function is replaceable.
13504 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
13507 * config/rs6000/altivec.h (vec_ctz and others): Change the
13508 preprocessor macro that controls conditional compilation from
13509 _ARCH_PWR9 to __POWER9_VECTOR__.
13510 (vec_all_ne): Change parameterization of __altivec_scalar_pred
13511 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
13512 control (instead of _ARCH_PWR9 control) so that template
13513 definition uses power9-specific function.
13514 (vec_any_eq): Likewise.
13515 (vec_all_ne): Change macro definition to use a power9-specific
13516 expansion under #ifdef __POWER9_VECTOR__ control (instead of
13517 _ARCH_PWR9 control).
13518 (vec_any_eq) Likewise.
13519 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
13520 expansion for CMPNEF to remove support for xvcmpnesp instruction.
13521 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
13522 support for xvcmpnedp instruction.
13523 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
13524 macro expansion so that Power9 implementation of vec_all_ne does
13525 not use the AltiVec predicate framework.
13526 (VCMPNEH_P): Likewise.
13527 (VCMPNEW_P): Likewise.
13528 (VCMPNED_P): Likewise.
13529 (VCMPNEFP_P): Likewise.
13530 (VCMPNEDP_P): Likewise.
13531 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
13532 implementation of vec_any_eq to not use AltiVec predicate
13534 (VCMPAEH_P): Likewise.
13535 (VCMPAEW_P): Likewise.
13536 (VCMPAED_P): Likewise.
13537 (VCMPAEFP_P): Likewise.
13538 (VCMPAEDP_P): Likewise.
13539 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
13540 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
13541 not use the AltiVec predicate framework.
13542 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
13543 of vec_any_eq to not use AltiVec predicate framework.
13544 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13545 support for predefined __POWER9_VECTOR__ macro to indicate that
13546 Power9 instruction selection is enabled.
13547 (altivec_overloaded_builtins): Remove extraneous
13548 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
13549 function argument types RS6000_BTI_bool_V16QI and
13550 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
13551 entry for overloaded function argument types RS6000_BTI_bool_V4SI
13552 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
13553 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
13554 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
13555 Power9 for implementations of vec_cmpne. Change the signature for
13556 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
13557 (representing vec_all_ne) to remove the previously described first
13558 argument of type RS6000_BTI_INTSI, as this was an artifact of
13559 reliance on the AltiVec predicate framework, which is no longer
13560 used in the implementation of these functions. Add
13561 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
13562 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
13563 since, unlike the AltiVec predicate framework implementation, we
13564 do not share function descriptors between vec_alle and vec_anyeq.
13565 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
13566 set of modes that receive special treatment even when
13567 TARGET_P9_VECTOR is true. The special treatment emits code that
13568 does not depend on Power9 instructions.
13569 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
13570 define_expand to not rely on AltiVec predicate framework.
13571 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13573 (vector_ne_v2di_p): Change this define_expand to not rely on
13574 AltiVec predicate framework.
13575 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
13577 (vector_ne_<mode>_p): Change this define_expand to not rely on
13578 AltiVec predicate framework.
13579 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13581 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
13582 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
13583 define_insn pattern.
13584 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
13585 define_insn pattern because the xvcmpne<VSs>. instruction is not
13587 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
13588 instruction is not supported.
13590 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13592 * config/nvptx/nvptx.c: Include intl.h.
13594 2017-03-01 Martin Jambor <mjambor@suse.cz>
13597 * ipa-prop.h (ipa_bits): Removed field known.
13598 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
13599 to pointers. Adjusted their comments to warn about their sharing.
13600 (ipcp_transformation_summary): Change bits to a vector of pointers.
13601 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
13602 (ipa_get_ipa_bits_for_value): Declare.
13603 * tree-vrp.h (value_range): Mark as GTY((for_user)).
13604 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
13605 (ipa_bits_hash_table): Likewise.
13606 (ipa_vr_ggc_hash_traits): Likewise.
13607 (ipa_vr_hash_table): Likewise.
13608 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
13609 being pointers and vr_known being removed.
13610 (ipa_set_jf_unknown): Likewise.
13611 (ipa_get_ipa_bits_for_value): New function.
13612 (ipa_set_jfunc_bits): Likewise.
13613 (ipa_get_value_range): New overloaded functions.
13614 (ipa_set_jfunc_vr): Likewise.
13615 (ipa_compute_jump_functions_for_edge): Use the above functions to
13616 construct bits and vr parts of jump functions.
13617 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
13618 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
13620 (ipcp_grow_transformations_if_necessary): Also allocate
13621 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
13623 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
13624 them. Fix too long lines.
13625 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
13626 vr_known being removed.
13627 (ipa_read_jump_function): Use new setter functions to construct bits
13628 and vr parts of jump functions or set them to NULL.
13629 (write_ipcp_transformation_info): Adjust for bits being pointers.
13630 (read_ipcp_transformation_info): Likewise.
13631 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
13633 Include gt-ipa-prop.h.
13634 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
13636 (ipcp_store_bits_results): Likewise.
13637 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
13638 Do not write to existing jump functions but use a temporary instead.
13640 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13643 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
13644 attempt to use its first operand as BIT_FIELD_REF base.
13646 2017-03-01 Richard Biener <rguenther@suse.de>
13648 PR middle-end/79721
13649 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
13650 interpolating formula in wrapping arithmetic.
13651 (chrec_apply): Convert chrec_evaluate return value to wanted type.
13653 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13655 PR tree-optimization/79734
13656 * tree-vect-generic.c (expand_vector_condition): Optimize
13657 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
13658 Handle VEC_COND_EXPR where comparison has different inner width from
13659 type's inner width.
13661 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
13663 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
13664 markup, and similar issues. Remove @opindex entries for things
13665 that aren't options. Add missing -mmpy-option entries.
13667 2017-02-28 Jakub Jelinek <jakub@redhat.com>
13669 PR tree-optimization/79737
13670 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
13671 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
13672 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
13673 instead of byte_size. Formatting fix.
13674 (shift_bytes_in_array_right): Formatting fix.
13676 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
13679 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
13680 condition on optimize for the leaf function test.
13682 2017-02-28 Martin Liska <mliska@suse.cz>
13685 * read-rtl-function.c (function_reader::handle_unknown_directive):
13686 Bail out when one uses -flto.
13688 2017-02-28 Martin Liska <mliska@suse.cz>
13690 * common.opt: Replace space with tabular for options of <number>
13692 * config/i386/i386.opt: Show <number> value for
13693 -mlarge-data-threshold.
13694 * opts.c (print_filtered_help): Do not display number in hexadecimal
13697 2017-02-28 Martin Liska <mliska@suse.cz>
13699 * common.opt: Fix --help=option -Q for options which are of
13702 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
13704 * config/i386/i386.c (print_reg): Error out for values
13705 of 8-bit size in invalid integer register.
13707 2017-02-28 Martin Sebor <msebor@redhat.com>
13709 PR tree-optimization/79691
13710 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
13712 2017-02-28 Jakub Jelinek <jakub@redhat.com>
13715 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
13716 gcc_unreachable with output_operand_lossage.
13718 2017-02-28 Richard Biener <rguenther@suse.de>
13720 PR tree-optimization/79740
13721 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
13723 (visit_nary_op): Insert the nary into the hashtable if we
13724 pattern-matched sth.
13725 * tree-ssa-pre.c (eliminate_insert): Robustify.
13727 2017-02-28 Richard Biener <rguenther@suse.de>
13729 PR middle-end/79731
13730 * fold-const.c (decode_field_reference): Reject out-of-bound
13733 2017-02-28 Jakub Jelinek <jakub@redhat.com>
13735 * config/i386/i386.c: Include intl.h.
13736 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
13737 instead of just cond ? "..." : "...".
13738 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
13739 * coverage.c (read_counts_file): Likewise.
13740 * omp-offload.c: Include intl.h.
13741 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
13742 of just cond ? "..." : "...".
13743 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
13744 of just cond ? "..." : "...".
13746 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
13749 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
13751 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
13752 'tune for' CPU name.
13753 * config/arm/arm-cpu-data.h: Regenerated.
13755 2017-02-28 Richard Biener <rguenther@suse.de>
13757 PR tree-optimization/79732
13758 * tree-inline.c (expand_call_inline): Do not shadow var.
13760 2017-02-28 Richard Biener <rguenther@suse.de>
13762 PR tree-optimization/79723
13763 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
13764 address-space properly.
13766 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
13768 * doc/optinfo.texi (Optimization groups): Fix option used for
13770 * doc/invoke.texi (-fopt-info): Document "omp".
13771 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
13772 (OPTGROUP_ALL): Add OPTGROUP_OMP.
13773 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
13774 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
13775 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
13777 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
13779 * dumpfile.c (optgroup_options): Instead of "openmp", associate
13780 OPTGROUP_OMP with "omp".
13782 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
13785 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
13786 for arithmetic shift of unsigned V2DI.
13788 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
13790 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
13791 arc/linux.h headers.
13792 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
13793 (LINK_SPEC): Likewise.
13794 (ARC_TLS_EXTRA_START_SPEC): Likewise.
13795 (EXTRA_SPECS): Likewise.
13796 (STARTFILE_SPEC): Likewise.
13797 (ENDFILE_SPEC): Likewise.
13798 (LIB_SPEC): Likewise.
13799 (TARGET_SDATA_DEFAULT): Likewise.
13800 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
13801 (MULTILIB_DEFAULTS): Likewise.
13802 (DWARF2_UNWIND_INFO): Likewise.
13803 * config/arc/big.h: New file.
13804 * config/arc/elf.h: Likewise.
13805 * config/arc/linux.h: Likewise.
13806 * config/arc/t-uClibc: Remove.
13808 2017-02-27 Bin Cheng <bin.cheng@arm.com>
13810 PR tree-optimization/77536
13811 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
13812 (tree_transform_and_unroll_loop): Use above function to compute the
13813 estimated niter of unrolled loop and use it when scaling profile.
13814 Also use count info rather than frequency if it's non-zero.
13815 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
13816 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
13817 (vect_transform_loop): Call above function.
13819 2017-02-27 Richard Biener <rguenther@suse.de>
13821 PR tree-optimization/45397
13822 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
13823 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
13824 (visit_nary_op): Add pattern matching for CSEing sign-changed
13825 or truncated operations with wider ones.
13827 2017-02-27 Richard Biener <rguenther@suse.de>
13829 PR tree-optimization/79690
13830 * tree-vect-stmts.c (vectorizable_store): Use vector type
13831 built from the DR with address-space.
13833 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
13835 * doc/invoke.texi (Optimize Options): Refine the description
13836 of asan-use-after-return.
13838 2017-02-25 Alan Modra <amodra@gmail.com>
13840 PR rtl-optimization/79584
13841 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
13842 base, not ad->base_term, the reg within base. Remove assertion
13843 that ad->base == ad->base_term. Replace gen_int_mode using
13844 bogus mode with const0_rtx.
13846 2017-02-25 Jakub Jelinek <jakub@redhat.com>
13848 PR middle-end/79396
13849 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
13850 FMA_EXPR like tcc_binary or tcc_unary.
13852 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
13855 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
13857 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
13858 (output_loc_operands): Handle DW_OP_call_ref and
13859 DW_OP_GNU_variable_value.
13860 (struct variable_value_struct): New type.
13861 (struct variable_value_hasher): Likewise.
13862 (variable_value_hash): New variable.
13863 (string_types): Remove.
13864 (copy_loc_descr): New function.
13865 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
13866 (prepend_loc_descr_to_each): New function.
13867 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
13868 instead of add_loc_descr_to_each if the first argument is single
13869 location list and the second has multiple.
13870 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
13871 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
13872 when looking for variable value which doesn't have other location info.
13873 (loc_list_from_tree): Formatting fix.
13874 (gen_array_type_die): Simplify DW_AT_string_length handling.
13875 (adjust_string_types): Remove.
13876 (gen_subprogram_die): Don't call adjust_string_types nor test/set
13877 string_types. Call resolve_variable_values.
13878 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
13879 (resolve_addr_in_expr): Likewise. Add A argument.
13880 (copy_deref_exprloc): Remove deref argument. Adjust for the
13881 original expression being DW_OP_GNU_variable_value with optionally
13882 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
13883 optionally after it.
13884 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
13885 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
13886 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
13887 (variable_value_hasher::hash, variable_value_hasher::equal): New
13889 (resolve_variable_value_in_expr, resolve_variable_value,
13890 resolve_variable_values, note_variable_value_in_expr,
13891 note_variable_value): New functions.
13892 (dwarf2out_early_finish): Call note_variable_value on all toplevel
13895 2017-02-24 Jakub Jelinek <jakub@redhat.com>
13898 * opts.h (handle_generated_option): Add GENERATED_P argument.
13899 * opts-common.c (handle_option): Adjust function comment.
13900 (handle_generated_option): Add GENERATED_P argument, pass it to
13902 (control_warning_option): Pass false to handle_generated_option
13904 * opts.c (maybe_default_option): Pass true to handle_generated_option
13906 * optc-gen.awk: Likewise.
13908 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13910 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
13911 a REG, look at the REG it is a SUBREG of.
13912 (splitter for cmpeqsi_t): Ditto.
13914 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13916 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
13917 the special USEs with the pattern of the insn, not the insn itself.
13919 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
13922 * doc/invoke.texi: Document -mload-store-pairs.
13924 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13925 Sandra Loosemore <sandra@codesourcery.com>
13927 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
13928 argument isn't a CONST_INT.
13929 (nios2_alternate_compare_const): Assert op is a CONST_INT.
13930 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
13931 (nios2_validate_compare): Bypass alternate compare logic if *op2
13932 is not a CONST_INT.
13933 (ldstwm_operation_p): Return false if first_base is not a REG or
13934 if first_offset is not a CONST_INT.
13936 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13938 * config/cris/cris.md: Use correct operand in a define_peephole2.
13940 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13942 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
13944 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13946 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
13947 this_insn if it is an INSN or JUMP_INSN.
13948 (force_offsettable): Look at base, not at addr.
13949 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
13950 on things that aren't necessarily CONST_INTs.
13952 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
13954 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
13955 -mfpmath=sse is the default also for x86-32 targets with SSE2
13956 instruction set when @option{-ffast-math} is enabled
13958 2017-02-24 Jeff Law <law@redhat.com>
13960 PR rtl-optimizatoin/79286
13961 * ira.c (update_equiv_regs): Drop may_trap_p exception to
13964 2017-02-24 Richard Biener <rguenther@suse.de>
13966 PR tree-optimization/79389
13967 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
13970 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
13972 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
13973 function comment to reflect reality.
13974 (loop_exits_before_overflow): Fix typo in function description.
13976 2017-02-24 Richard Biener <rguenther@suse.de>
13978 PR tree-optimization/79389
13979 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
13980 properly that a threading opportunity exists. Detect conditional
13981 copy/constant propagation opportunities.
13983 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
13985 * config/visium/visium.md (type): Add trap.
13986 (b): New mode attribute.
13987 (*btst): Rename into...
13988 (*btst<mode>): ...this and adjust.
13989 (*cbranchsi4_btst_insn): Rename into...
13990 (*cbranch<mode>4_btst_insn): ...this and adjust.
13991 (trap): New define_insn.
13993 2017-02-23 Jakub Jelinek <jakub@redhat.com>
13995 PR tree-optimization/79389
13996 * ifcvt.c (struct noce_if_info): Add rev_cond field.
13997 (noce_reversed_cond_code): New function.
13998 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
13999 reversed_comparison_code. Formatting fix.
14000 (noce_try_store_flag): Test rev_cond != NULL in addition to
14001 reversed_comparison_code.
14002 (noce_try_store_flag_constants): Likewise.
14003 (noce_try_store_flag_mask): Likewise.
14004 (noce_try_addcc): Use rev_cond if non-NULL instead of
14005 reversed_comparison_code.
14006 (noce_try_cmove_arith): Likewise. Formatting fixes.
14007 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14008 (noce_find_if_block): Initialize rev_cond.
14009 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14010 instead of false as last argument never attempt to reverse it
14013 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14015 PR tree-optimization/79663
14016 * tree-predcom.c (combine_chains): Process refs in reverse order
14017 only for ZERO length chains, and add explaining comment.
14019 2017-02-23 Jeff Law <law@redhat.com>
14021 PR tree-optimization/79578
14022 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14023 in call to operand_equal_p.
14025 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14028 * config/i386/cpuid.h: Fix another undefined behavior.
14030 2017-02-23 Richard Biener <rguenther@suse.de>
14032 PR tree-optimization/79683
14033 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14034 vector types for data-refs.
14036 2017-02-23 Martin Liska <mliska@suse.cz>
14038 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14040 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14042 PR middle-end/79665
14043 * internal-fn.c (get_range_pos_neg): Moved to ...
14044 * tree.c (get_range_pos_neg): ... here. No longer static.
14045 * tree.h (get_range_pos_neg): New prototype.
14046 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14047 are known to be in between 0 and signed maximum inclusive, try to
14048 expand both unsigned and signed divmod and use the cheaper one from
14051 2017-02-22 Jeff Law <law@redhat.com>
14053 PR tree-optimization/79578
14054 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14055 to compare base operands.
14057 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14060 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14061 gpc_reg_operand instead of fpr_reg_operand.
14063 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14065 * config/mips/mips.c (mips_return_in_memory): Force FP
14066 vector types to be returned in memory for o32 ABI.
14068 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14070 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14071 instead of DW_TAG_member for static data member declarations and don't
14072 set no_linkage_name for static inline data members.
14073 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14076 2017-02-22 Martin Liska <mliska@suse.cz>
14078 * doc/invoke.texi: Replace inequality signs with square brackets
14081 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14083 PR tree-optimization/68644
14084 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14086 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14089 * lra-constraints.c (simplify_operand_subreg): Handle
14090 WORD_REGISTER_OPERATIONS targets.
14092 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14095 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14096 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14097 elimination by swapping fld*.
14099 2017-02-22 Richard Biener <rguenther@suse.de>
14101 PR tree-optimization/79673
14102 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14103 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14104 irrelevant address-space qualifiers and avoiding a
14105 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14107 2017-02-22 Richard Biener <rguenther@suse.de>
14109 PR tree-optimization/79666
14110 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14111 to not symbolically negate if that may introduce undefined
14114 2017-02-22 Martin Liska <mliska@suse.cz>
14117 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14118 * data-streamer-out.c (streamer_write_gcov_count_stream):
14120 * value-prof.c (stream_out_histogram_value): Make assert more
14121 precise based on type of counter.
14123 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14126 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14127 Use nonimmediate_operand instead of memory_operand for operand 1.
14128 (float-extend standard_x87sse_constant_load splitter): Ditto.
14130 2017-02-21 Jeff Law <law@redhat.com>
14132 PR tree-optimization/79621
14133 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14134 blocks with edges to themselves.
14136 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14139 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14140 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14141 Use gimple_call_builtin_p.
14144 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14145 on temporarily removed DEBUG_INSNs.
14147 PR tree-optimization/79649
14148 * tree-loop-distribution.c (classify_partition): Give up on
14149 non-generic address space loads/stores.
14151 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14153 * doc/loop.texi (Loop manipulation): Remove nonexistent
14154 tree_ssa_loop_version from the documentation.
14155 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14157 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14160 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14161 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14162 * config/rs6000/rs6000.c: Include except.h.
14163 (rs6000_expand_split_stack_prologue): Call
14164 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14166 2017-02-21 Martin Jambor <mjambor@suse.cz>
14169 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14170 have been analyzed.
14172 2017-02-21 Martin Jambor <mjambor@suse.cz>
14174 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14175 for backward compatibility only.
14176 * doc/invoke.texi (Option Summary): Remove all references to
14177 -fipa-cp-alignment.
14179 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14183 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14185 * lra-constraints.c (curr_insn_transform): Handle
14186 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14188 2017-02-21 Martin Liska <mliska@suse.cz>
14190 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14192 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14194 PR translation/79638
14195 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14197 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14200 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14201 (arm_function_ok_for_sibcall): Return false for an indirect call by
14202 descriptor if all the argument registers are used.
14203 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14204 alignment of the function.
14206 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14208 PR tree-optimization/61441
14209 * simplify-rtx.c (simplify_const_unary_operation): For
14210 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14211 the sNaN unmodified.
14213 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14215 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14216 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14217 instead of SYSTEM_HEADER_DIR.
14219 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14220 Martin Liška <mliska@suse.cz>
14222 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14223 Fix typos and grammar, use active voice, and clarify.
14225 2017-02-20 Marek Polacek <polacek@redhat.com>
14227 PR middle-end/79537
14228 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14231 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14233 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14236 * config/i386/i386.c (ix86_expand_builtin): Handle
14237 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14238 ix86_builtins_isa[fcode].isa as a requirement of those
14239 flags and any other flag in the bitmask.
14240 (ix86_init_mmx_sse_builtins): Use 0 instead of
14241 ~OPTION_MASK_ISA_64BIT as mask.
14242 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14243 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14244 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14245 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14247 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14250 * lra-constraints.c (split_reg): Check requested split mode
14251 is supported by the register.
14253 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14255 * lra-constraints.c (simplify_operand_subreg): Remove early
14258 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14261 * lra-constraints.c (curr_insn_transform): Tighten condition
14262 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14264 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14267 * lra-constraints.c (curr_insn_transform): Handle
14268 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14270 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14273 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14275 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14277 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14280 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14283 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14285 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14288 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14290 * final.c (last_columnnum, override_columnnum): New variables.
14291 (final_start_function): Set last_columnnum, pass it to begin_prologue
14292 hook and pass 0 to dwarf2out_begin_prologue.
14293 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14294 to source_line debug hook.
14295 (notice_source_line): Compute last_columnnum and for debug_column_info
14296 return true on column changes.
14297 * debug.h (struct gcc_debug_hooks): Add column argument to
14298 source_line and begin_prologue hooks.
14299 (debug_nothing_int_charstar_int_bool): Remove prototype.
14300 (debug_nothing_int_int_charstar,
14301 debug_nothing_int_int_charstar_int_bool): New prototypes.
14302 (dwarf2out_begin_prologue): Add column argument.
14303 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14304 begin_prologue hooks.
14305 (debug_nothing_int_charstar_int_bool): Remove.
14306 (debug_nothing_int_int_charstar,
14307 debug_nothing_int_int_charstar_int_bool): New functions.
14308 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14309 through to dwarf2out_source_line.
14310 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14311 (dwarf2out_source_line): Add column argument, emit it if requested.
14312 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14314 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14315 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14316 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14317 through to dwarf2out_begin_prologue.
14318 (vmsdbgout_source_line): Add column argument, pass it through to
14319 dwarf2out_source_line.
14320 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14321 dbxout_source_line caller.
14322 (dbxout_source_line): Add column argument.
14324 * common.opt (gno-column-info, gcolumn-info): New options.
14325 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14326 (check_die): Also test for multiple DW_AT_decl_column attributes.
14327 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14328 DW_AT_decl_column if requested.
14329 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14331 (gen_variable_die): Likewise.
14332 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14333 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14336 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14337 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14338 (ix86_handle_option): Handle OPT_m3dnowa.
14339 * doc/invoke.texi (-m3dnowa): Document.
14340 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14341 -m3dnowa instead of -m3dnow -march=athlon.
14344 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14345 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14347 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14350 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14351 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14352 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14353 generator for vsx_xxpermdi_<mode>_be.
14354 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14355 force big-endian semantics.
14356 (vsx_xxpermdi_<mode>_be): New define_expand with same
14357 implementation as previous version of vsx_xxpermdi_<mode>.
14359 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14361 PR tree-optimization/79327
14362 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14363 variable, its initialization and use.
14365 2017-02-17 Julia Koval <julia.koval@intel.com>
14367 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14368 (OPTION_MASK_ISA_PKU_UNSET): New.
14369 (ix86_handle_option): Handle -mrdpid.
14370 * config/i386/cpuid.h (bit_RDPID): New.
14371 * config/i386/driver-i386.c (host_detect_local_cpu):
14372 Detect RDPID feature.
14373 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14374 * config/i386/i386-c.c (ix86_target_macros_internal):
14376 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14377 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14378 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14379 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14380 * config/i386/i386.md (define_insn "rdpid"): New.
14381 * config/i386/i386.opt Add -mrdpid.
14382 * config/i386/immintrin.h (_rdpid_u32): New.
14384 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14386 PR rtl-optimization/79541
14387 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14388 instead of transforming it into USE.
14390 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14392 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14393 If HONOR_SNANS (SFmode) force the input to a register.
14394 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14395 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14396 an frsp or similar insn.
14398 2017-02-17 Martin Liska <mliska@suse.cz>
14400 PR rtl-optimization/79577
14401 * params.def (selsched-max-sched-times): Increase minimum to 1.
14403 2017-02-17 Martin Liska <mliska@suse.cz>
14405 PR rtl-optimization/79574
14406 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14408 2017-02-17 Martin Liska <mliska@suse.cz>
14410 PR tree-optimization/79529
14411 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
14412 ssa_defined_default_def_p to handle cases which are implicitly
14414 * tree-ssa.c (ssa_defined_default_def_p): New function.
14415 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
14416 which are implicitly defined.
14417 * tree-ssa.h (ssa_defined_default_def_p): Declare.
14419 2017-02-17 Richard Biener <rguenther@suse.de>
14421 PR middle-end/79576
14422 * params.def (max-ssa-name-query-depth): Limit to 10.
14424 2017-02-17 Richard Biener <rguenther@suse.de>
14426 PR tree-optimization/79552
14427 * tree-ssa-structalias.c (visit_loadstore): Properly verify
14430 2017-02-17 Richard Biener <rguenther@suse.de>
14433 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
14435 2017-02-17 Marek Polacek <polacek@redhat.com>
14437 PR middle-end/79536
14438 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
14439 (fold_negate_expr): New wrapper.
14441 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
14443 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
14444 Correct terminology and de-emphasize pre-standard behavior.
14446 2017-02-16 Alan Modra <amodra@gmail.com>
14448 PR rtl-optimization/79286
14449 * ira.c (def_dominates_uses): New function.
14450 (update_equiv_regs): Don't create an equivalence for insns that
14451 may trap where the register def does not dominate the use.
14453 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
14455 PR rtl-optimization/78127
14456 * lra.c (lra): Call lra_eliminate before finish the loop after
14459 2017-02-16 Richard Biener <rguenther@suse.de>
14461 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
14463 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
14464 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
14465 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
14466 (isl_val_int_from_wi): New function.
14467 (extract_affine_gmp): Rename to ...
14468 (extract_affine_wi): ... this, take a widest_int.
14469 (extract_affine_int): Just wrap extract_affine_wi.
14470 (add_param_constraints): Use isl_val_int_from_wi.
14471 (add_loop_constraints): Likewise, and extract_affine_wi.
14473 2017-02-15 Jeff Law <law@redhat.com>
14475 PR middle-end/79521
14476 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
14477 ira_init_register_move_cost_if_necessary.
14479 2017-02-15 Martin Sebor <msebor@redhat.com>
14481 PR middle-end/32003
14482 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
14483 removed in a prior commit.
14485 2017-02-15 Bin Cheng <bin.cheng@arm.com>
14487 PR tree-optimization/79347
14488 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
14489 counters during peeling.
14491 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
14493 * Makefile.in (site.exp): Remove "set ISLVER".
14495 2017-02-15 Jakub Jelinek <jakub@redhat.com>
14498 * real.c (real_from_integer): Call real_convert even for decimal.
14500 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
14503 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
14505 2017-02-14 Andrew Pinski <apinski@cavium.com>
14507 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
14508 cores and change the partno/implementer to be correct.
14509 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
14510 the 'B" as the implementer.
14511 * config/aarch64/aarch64-tune.md: Regenerate.
14513 2017-02-14 Carl Love <cel@us.ibm.com>
14515 * config/rs6000/rs6000.c: Add case statement entry to make the
14516 xvcvuxdsp built-in argument unsigned.
14517 * config/rs6000/vsx.md: Fix the source and return operand types so they
14518 match the instruction definitions from the ISA document. Fix typo
14519 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
14522 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
14525 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
14526 member early_clobber_alts.
14527 * lra-lives.c (reg_early_clobber_p): New.
14528 (process_bb_lives): Use it.
14529 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
14530 (debug_operand_data): Initialize early_clobber_alts.
14531 (setup_operand_alternative): Set up early_clobber_alts.
14532 (collect_non_operand_hard_regs): Ditto. Pass early clobber
14533 alternatives to new_insn_reg.
14534 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
14536 (lra_update_insn_regno_info): Pass the new arg.
14538 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14540 PR middle-end/79505
14541 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
14542 (new_oacc_loop_raw): Don't clear already cleared fields.
14545 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
14546 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
14547 _mm512_prefetch_i64gather_ps): New inline functions and macros.
14549 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
14552 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
14554 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
14557 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
14558 the extra instruction to the right place to store 128-bit constant
14561 2017-02-14 Martin Sebor <msebor@redhat.com>
14563 PR middle-end/79448
14564 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
14565 warning for strings of unknown length.
14567 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
14569 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
14571 2017-02-14 Jeff Law <law@redhat.com>
14574 * ira-costs.c (scan_one_insn): Initialize register move costs
14575 for pseudos seen in USE/CLOBBER insns.
14577 PR tree-optimization/79095
14578 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
14579 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
14580 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
14581 if the operands are known to be not equal, then the resulting range
14583 (intersect_ranges): If the new range is ~[0,0] and the old range is
14584 wide, then prefer ~[0,0].
14585 * tree-vrp.c (overflow_comparison_p_1): New function.
14586 (overflow_comparison_p): New function.
14587 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
14588 if NAME is used in an overflow test.
14589 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
14590 overflow check that can be expressed as an equality test, then adjust
14591 ops to be that equality test.
14593 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14595 * config/s390/s390-builtin-types.def: Remove flags argument.
14596 * config/s390/s390.c (s390_init_builtins): Likewise.
14598 2017-02-14 Martin Liska <mliska@suse.cz>
14600 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
14601 vector. Fix trailing white spaces.
14603 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
14605 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
14608 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14610 PR rtl-optimization/68664
14611 * config/arm/arm.c (arm_sched_can_speculate_insn):
14612 New function. Declare prototype.
14613 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
14615 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14617 PR rtl-optimization/68664
14618 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
14620 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
14622 2017-02-14 Amit Pawar <amit.pawar@amd.com>
14624 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
14625 max skip bytes for function, loop and jump.
14627 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14629 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
14630 ABS_EXPR for gimple dump.
14632 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14635 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
14637 PR tree-optimization/79408
14638 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
14639 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
14640 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
14641 also if rhs1 is INTEGER_CST.
14643 2017-02-14 Richard Biener <rguenther@suse.de>
14645 PR middle-end/79432
14646 * tree-into-ssa.c (insert_phi_nodes): When the function can
14647 have abnormal edges rewrite SSA names with broken use-def
14648 dominance out of SSA and register them for PHI insertion.
14650 2017-02-13 Martin Sebor <msebor@redhat.com>
14652 PR middle-end/79496
14653 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
14654 clearing info.nowrite flag when snprintf size argument is a range.
14656 2017-02-13 Jakub Jelinek <jakub@redhat.com>
14658 * cprop.c (cprop_jump): Add missing space in string literal.
14659 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
14660 (get_constraint_for_component_ref): Likewise.
14661 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
14662 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
14663 * lra-constraints.c (process_alt_operands): Likewise.
14664 * ipa-inline.c (inline_small_functions): Likewise.
14665 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
14666 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14667 * trans-mem.c (diagnose_tm_1_op): Likewise.
14668 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
14669 (grid_parallel_clauses_gridifiable): Likewise.
14671 * config/nvptx/mkoffload.c (process): Add space in between
14674 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
14675 "MOD4_SSE_REGS" and "ALL_REGS".
14677 * spellcheck.c (test_data): Add , in between "foo" and "food".
14679 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14682 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
14683 boundary crossing check and subsequent code generation agree.
14685 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14687 * config/aarch64/aarch64.c (has_memory_op): Delete.
14688 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
14691 2017-02-13 Jakub Jelinek <jakub@redhat.com>
14693 PR rtl-optimization/79388
14694 PR rtl-optimization/79450
14695 * combine.c (distribute_notes): When removing TEM_INSN for which
14696 corresponding dest has last value recorded, invalidate that last
14699 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14701 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
14702 of explicit '@'. Add missing assembly comment marker on branch costs
14705 2017-02-13 Nathan Sidwell <nathan@acm.org>
14707 * gengtype-lex.l (<in_struct>): Add '/'.
14709 2017-02-13 Martin Liska <mliska@suse.cz>
14712 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
14714 2017-02-13 Richard Biener <rguenther@suse.de>
14716 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
14718 * configure: Re-generate.
14719 * config.in: Likewise.
14720 * graphite-dependences.c: Simplify as if
14721 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
14722 * graphite-isl-ast-to-gimple.c: Likewise.
14723 * graphite-optimize-isl.c: Likewise.
14724 * graphite-poly.c: Likewise.
14725 * graphite-sese-to-poly.c: Likewise.
14726 * graphite.h: Likewise.
14727 * toplev.c: Include isl/version.h and use isl_version () for
14728 printing the ISL version.
14729 * doc/install.texi: Update ISL requirement.
14731 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
14733 * doc/standards.texi (Standards): Update reference to
14736 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
14738 * doc/extend.texi (Named Address Spaces): sourceware.org now
14740 * doc/install.texi (Binaries): Ditto.
14743 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
14745 * doc/cpp.texi: Replace "stringify"/"stringification" with C
14746 standard terminology "stringize"/"stringizing" throughout.
14747 * doc/cppinternals.texi: Likewise.
14749 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
14751 * doc/extend.texi: Fix some spelling mistakes and typos.
14752 * doc/invoke.texi: Likewise.
14754 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
14757 * params.def (inline-min-speedup) Change from 10 to 8.
14759 2017-02-11 Jakub Jelinek <jakub@redhat.com>
14761 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
14764 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
14767 * ipa-inline-analysis.c (get_minimal_bb): New function.
14768 (record_modified): Use it.
14769 (remap_edge_change_prob): Handle also ancestor functions.
14771 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
14773 * doc/contrib.texi (Contributors): Remove broken link into
14774 the Mauve CVS repository.
14776 2017-02-11 Jakub Jelinek <jakub@redhat.com>
14778 PR middle-end/79454
14779 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
14780 result computation whenever lhs doesn't have vector mode, not
14781 just when it has BLKmode.
14783 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
14785 * doc/makefile.texi (profiledbootstrap): Refer to the
14786 installation instructions only in textual form.
14788 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14791 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
14793 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
14795 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
14796 (Specific): Update mingw-w64 reference.
14798 (Specific): Remove broken link to Renesas RX processor.
14800 2017-02-10 Richard Biener <rguenther@suse.de>
14802 * toplev.c (process_options): Do not mention obsolete graphite
14803 options when printing sorry message about missing graphite support.
14804 Mention -floop-nest-optimize.
14806 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
14808 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
14809 (vtst_p16): Likewise.
14810 (vtstq_p8): Likewise.
14811 (vtstq_p16): Likewise.
14813 (vtstq_p64): Likewise.
14814 * config/arm/arm_neon.h (vgetq_lane_p64): New.
14815 (vset_lane_p64): New.
14816 (vsetq_lane_p64): New.
14818 2017-02-10 Jakub Jelinek <jakub@redhat.com>
14820 PR tree-optimization/79411
14821 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
14822 stmt operands are SSA_NAMEs used in abnormal phis.
14823 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
14826 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
14829 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
14832 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
14834 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
14836 2017-02-09 Jakub Jelinek <jakub@redhat.com>
14838 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
14842 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
14843 not arbitrary TREE_CONSTANT.
14846 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
14847 "omp declare target link" attribute unless is_global_var.
14848 * omp-offload.c (find_link_var_op): Likewise.
14850 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
14851 Chung-Lin Tang <cltang@codesourcery.com>
14853 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
14855 (gimplify_adjust_omp_clauses): Don't delete TILE.
14856 (gimplify_omp_for): Deal with TILE.
14857 * internal-fn.c (expand_GOACC_TILE): New function.
14858 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
14860 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
14861 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
14863 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
14864 avoid DIV for outermost collapse var.
14865 (expand_oacc_for): Insert tile element loop as needed. Adjust.
14866 Remove out of date comments, fix whitespace.
14867 * omp-general.c (omp_extract_for_data): Deal with tiling.
14868 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
14869 adjust OLF_DIM_BASE value.
14870 (struct omp_for_data): Add tiling field.
14871 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
14872 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
14873 for auto loops. Remove default auto determining, moved to
14874 oacc_loop_fixed_partitions.
14875 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
14876 stmts, add e_mask field.
14877 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
14878 (oacc_thread_numbers): Use oacc_dim_call.
14879 (oacc_xform_tile): New.
14880 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
14881 (finish_oacc_loop): Adjust for ifns vector.
14882 (oacc_loop_discover_walk): Append loop abstraction sites to list,
14883 add case for GOACC_TILE fns.
14884 (oacc_loop_xform_loop): Delete.
14885 (oacc_loop_process): Iterate over call list directly, and add
14886 handling for GOACC_TILE fns.
14887 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
14889 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
14890 vector partitioning to outer loops. Assign 2 partitions to loops
14891 when available. Add TILE handling.
14892 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
14893 (execite_oacc_device_lower): Process GOACC_TILE fns,
14894 ignore unknown specs.
14895 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
14896 * tree.c (omp_clause_num_ops): Adjust TILE ops.
14897 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
14899 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
14901 * configure.ac (ACX_BUGURL): Update.
14902 * configure: Regenerate.
14904 2017-02-09 Richard Biener <rguenther@suse.de>
14906 PR tree-optimization/69823
14907 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
14908 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
14910 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
14912 * config/arc/arc-c.def: Add __NPS400__ definition.
14913 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
14914 (TARGET_NPS400): Define.
14916 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
14918 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
14920 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
14921 pointer, arch_info.
14922 (arc_cpu_types): Fill the arch_info field with a pointer into the
14923 arc_arch_types table.
14924 (arc_selected_cpu): Declare.
14925 * config/arc/arc.c (arc_selected_cpu): Make global.
14926 (arc_selected_arch): Delete.
14927 (arc_base_cpu): Delete.
14928 (arc_override_options): Remove references to deleted variables,
14929 update access to arch information.
14930 (ARC_OPT): Update access to arch information.
14931 (ARC_OPTX): Likewise.
14932 * config/arc/arc.h (arc_base_cpu): Remove declaration.
14933 (TARGET_ARC600): Update access to arch information.
14934 (TARGET_ARC601): Likewise.
14935 (TARGET_ARC700): Likewise.
14936 (TARGET_EM): Likewise.
14937 (TARGET_HS): Likewise.
14938 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
14941 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
14944 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
14945 condition/operands for integer GE/LE/GEU/LEU operations.
14947 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
14949 PR translation/79397
14950 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
14953 2017-02-08 Martin Jambor <mjambor@suse.cz>
14956 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
14957 whether allocation happened.
14958 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
14959 nothing was allocated.
14961 2017-02-08 Jakub Jelinek <jakub@redhat.com>
14963 PR tree-optimization/79408
14964 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
14965 constant, but SSA_NAME with a known integer range, use the minimum
14966 of that range instead of op1 to determine if modulo can be replaced
14967 with its first operand.
14969 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14971 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
14973 2017-02-08 Richard Biener <rguenther@suse.de>
14975 PR tree-optimization/71824
14976 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
14977 Check all loops contained in the merged region.
14979 2017-02-07 Andrew Pinski <apinski@cavium.com>
14981 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
14983 2017-02-07 Andrew Pinski <apinski@cavium.com>
14985 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
14986 (thunderxt88): Likewise.
14987 (thunderxt81): Disable LSE and change v8.1 to v8.
14988 (thunderxt83): Likewise.
14990 2017-02-07 Jakub Jelinek <jakub@redhat.com>
14991 Richard Biener <rguenther@suse.de>
14993 PR middle-end/79399
14994 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
14995 type from int to size_t.
14996 * ira-costs.c (struct_costs_size): Change type from int to size_t.
14998 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15000 PR rtl-optimization/79386
15001 * cprop.c (bypass_conditional_jumps): Initialize
15002 bypass_last_basic_block already before splitting bbs after
15003 unconditional traps...
15004 (bypass_conditional_jumps): ... rather than here.
15007 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15008 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15009 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15010 fix -masm=intel patterns.
15012 2017-02-07 Richard Biener <rguenther@suse.de>
15014 PR tree-optimization/79256
15015 PR middle-end/79278
15016 * builtins.c (get_object_alignment_2): Use min_align_of_type
15017 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15018 and ADJUST_FIELD_ALIGN.
15020 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15022 * doc/tm.texi: Regenerate.
15023 * stor-layout.c (layout_decl): Adjust.
15024 (update_alignment_for_field): Likewise.
15025 (place_field): Likewise.
15026 (min_align_of_type): Likewise.
15027 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15028 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15029 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15030 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15031 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15032 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15033 * config/i386/i386.c (x86_field_alignment): Likewise.
15034 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15035 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15036 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15037 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15038 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15039 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15043 2017-01-30 Richard Biener <rguenther@suse.de>
15045 PR tree-optimization/79256
15046 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15047 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15050 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15052 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15053 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15054 builtins to SImode and emit a zero-extend, if necessary.
15056 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15058 * docs/invoke.texi (RISC-V Options): Alphabetize.
15060 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15062 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15065 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15067 * config/riscv/riscv.c: New file.
15068 * gcc/common/config/riscv/riscv-common.c: Likewise.
15069 * config.gcc: Likewise.
15070 * config/riscv/constraints.md: Likewise.
15071 * config/riscv/elf.h: Likewise.
15072 * config/riscv/generic.md: Likewise.
15073 * config/riscv/linux.h: Likewise.
15074 * config/riscv/multilib-generator: Likewise.
15075 * config/riscv/peephole.md: Likewise.
15076 * config/riscv/pic.md: Likewise.
15077 * config/riscv/predicates.md: Likewise.
15078 * config/riscv/riscv-builtins.c: Likewise.
15079 * config/riscv/riscv-c.c: Likewise.
15080 * config/riscv/riscv-ftypes.def: Likewise.
15081 * config/riscv/riscv-modes.def: Likewise.
15082 * config/riscv/riscv-opts.h: Likewise.
15083 * config/riscv/riscv-protos.h: Likewise.
15084 * config/riscv/riscv.h: Likewise.
15085 * config/riscv/riscv.md: Likewise.
15086 * config/riscv/riscv.opt: Likewise.
15087 * config/riscv/sync.md: Likewise.
15088 * config/riscv/t-elf-multilib: Likewise.
15089 * config/riscv/t-linux: Likewise.
15090 * config/riscv/t-linux-multilib: Likewise.
15091 * config/riscv/t-riscv: Likewise.
15092 * configure.ac: Likewise.
15093 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15094 Waterman as RISC-V maintainers.
15095 * doc/install.texi: Add RISC-V entries.
15096 * doc/invoke.texi: Add RISC-V options section.
15097 * doc/md.texi: Add RISC-V constraints section.
15098 * configure: Regenerated.
15100 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15103 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15104 false values to be constant vectors with all 0 or all 1 bits set.
15105 (vcondu<mode><mode>): Likewise.
15106 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15108 (fpmask_comparison_operator): Update comment.
15109 (vecint_comparison_operator): New predicate.
15110 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15111 vector conditionals when the true and false values are constant
15112 vectors with all 0 bits or all 1 bits set.
15114 2017-02-06 Martin Sebor <msebor@redhat.com>
15116 PR tree-optimization/79376
15117 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15119 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15121 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15122 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15123 to simplify split condition.
15125 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15127 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15128 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15131 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15133 PR rtl-optimization/68664
15134 * target.def (can_speculate_insn): New hook.
15135 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15136 * doc/tm.texi: Regenerate.
15137 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15138 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15139 (rs6000_sched_can_speculate_insn): New function.
15141 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15143 PR tree-optimization/79284
15144 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15145 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15146 vectorizable_mask_load_store, vectorizable_operation,
15147 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15148 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15149 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15150 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15151 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15152 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15153 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15154 is_gimple_assign (stmt). Replace another such test with
15155 is_gimple_assign (stmt).
15157 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15160 * config/avr/avr.c (rtl-iter.h): Include it.
15161 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15162 (avr_legitimate_combined_insn): ...and implementation.
15164 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15166 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15167 * config/s390/s390.c (s390_const_operand_ok)
15168 (s390_canonicalize_comparison, s390_extract_part)
15169 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15170 (s390_contiguous_bitmask_p, s390_rtx_costs)
15171 (legitimize_pic_address): Likewise.
15172 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15173 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15174 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15175 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15176 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15178 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15180 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15181 REGNO($0) == REGNO($1).
15183 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15185 * config/s390/linux.h(SIZE_TYPE): Add comment.
15187 2017-02-06 Julian Brown <julian@codesourcery.com>
15188 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15189 Virendra Pathak <virendra.pathak@broadcom.com>
15191 * config/aarch64/aarch64-cores.def: Change the scheduler
15193 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15194 * config/aarch64/thunderx2t99.md: New file.
15196 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15198 * doc/standards.texi (Go Language): Update link to language
15201 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15203 * tree-eh.c (lower_resx): Sanitize profile.
15204 (cleanup_empty_eh_move_lp): Likewise.
15206 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15209 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15211 * cfgloopmanip.h (loop_version): Update prototype.
15212 * modulo-sched.c (sms_schedule): Update call of loop_version.
15213 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15214 * tree-parloops.c (gen_parallel_loop): Likewise.
15215 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15216 * tree-ssa-loop-split.c (split_loop): Likewise.
15217 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15218 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15220 2017-02-05 Martin Liska <mliska@suse.cz>
15223 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15225 (print_operand_address): Initialize a struct to zero.
15227 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15229 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15230 garbage collector only in textual form.
15232 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15234 * doc/extend.texi (x86 specific memory model extensions for
15235 transactional memory): Simplify a phrase.
15237 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15240 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15241 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15242 (atomic_storedi_1): Likewise.
15244 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15246 PR tree-optimization/79338
15247 * tree-parloops.c (gather_scalar_reductions): Don't call
15248 vect_analyze_loop_form for loop->inner before destroying loop's
15251 2017-02-03 Martin Sebor <msebor@redhat.com>
15253 PR tree-optimization/79327
15254 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15255 when precision has resulted in leading zeros.
15256 (format_integer): Adjust the likely counter to assume an unknown
15257 argument that may be zero is non-zero.
15259 2017-02-03 Jason Merrill <jason@redhat.com>
15262 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15263 avoid copying non-taken branch.
15265 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15267 PR tree-optimization/79340
15268 * tree-vect-loop.c (vectorizable_reduction): Release
15269 vec_defs elements after safe_splicing them into other vectors.
15272 PR tree-optimization/79327
15273 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15274 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15276 (format_integer): Use wide_int_to_tree instead of build_int_cst
15277 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15278 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15279 of shortest and longest sequence.
15281 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15283 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15284 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15286 2017-02-03 Walter Lee <walt@tilera.com>
15289 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15290 after initial stackframe link reg save.
15291 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15293 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15296 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15297 wu for stxssp alternative.
15299 2017-02-03 Martin Sebor <msebor@redhat.com>
15301 PR tree-optimization/79352
15302 * gimple-fold.c (get_range_strlen): Add argument.
15303 (get_range_strlen): Change return type to bool.
15304 (get_maxval_strlen): Pass in a dummy argument.
15305 * gimple-fold.h (get_range_strlen): Change return type to bool.
15306 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15307 * tree.h (array_at_struct_end_p): Add argument.
15308 * tree.c (array_at_struct_end_p): Handle it.
15310 2017-02-03 Martin Liska <mliska@suse.cz>
15313 * multiple_target.c (create_dispatcher_calls): Redirect edge
15314 from a caller of a dispatcher.
15315 (expand_target_clones): Make the clones local.
15316 (ipa_target_clone): Do both target clones and resolvers.
15317 (ipa_dispatcher_calls): Remove the pass.
15318 (pass_dispatcher_calls::gate): Likewise.
15319 (make_pass_dispatcher_calls): Likewise.
15320 * passes.def (pass_target_clone): Put as very first IPA early
15323 2017-02-03 Martin Liska <mliska@suse.cz>
15325 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15326 in case of a function with ifunc attribute.
15328 2017-02-03 Martin Liska <mliska@suse.cz>
15330 * cgraph.c (cgraph_node::dump): Dump function version info.
15331 * symtab.c (symtab_node::dump_base): Add missing new line.
15333 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15335 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15336 (ifcombine_ifandif): Use it.
15338 2017-02-03 Martin Liska <mliska@suse.cz>
15340 * doc/invoke.texi: Document default value for
15341 use-after-scope-direct-emission-threshold.
15343 2017-02-03 Martin Liska <mliska@suse.cz>
15345 PR tree-optimization/79339
15346 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15347 (format_floating): Likewise.
15349 2017-02-03 Martin Liska <mliska@suse.cz>
15352 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15354 (ipa_node_params_t::remove): Likewise.
15355 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15356 initialization from removed ipa_node_params_t::insert.
15357 (ipa_node_params::~ipa_node_params): Move from removed
15358 ipa_node_params_t::release.
15359 * symbol-summary.h (symbol_summary::m_released): New member.
15360 Do not release a summary twice. Do not allow to call finalizer
15361 for types of a summary that live in GGC memory.
15363 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15365 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15368 2017-02-02 Martin Sebor <msebor@redhat.com>
15370 PR middle-end/79275
15371 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15372 (format_string): Tighten up the range of output for non-constant
15373 strings and correct the expected range for wide non-constant strings.
15375 2017-02-02 Martin Sebor <msebor@redhat.com>
15377 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15379 PR middle-end/32003
15380 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15382 (-fdump-tree-@var): Add to index and document how to come up
15383 with pass-specific option and dump file names.
15384 (-fdump-passes): Clarify where to look for output.
15386 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15388 PR middle-end/77445
15389 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15390 statistics of the analyzed path; allow threading for speed when
15391 any of BBs along the path are optimized for speed.
15393 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15395 PR middle-end/78468
15396 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15397 settings of the virtual registers.
15400 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15402 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15403 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15406 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15408 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15409 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15411 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15413 * config/s390/s390.md: Add missing comments with the expanded
15415 * config/s390/vector.md: Likewise.
15416 * config/s390/vx-builtins.md: Likewise.
15418 2017-02-02 Jakub Jelinek <jakub@redhat.com>
15421 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
15422 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
15423 conditions on a single line.
15425 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15427 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15428 __S390_VX__ to __VX__.
15430 2017-02-01 Andrew Pinski <apinski@cavium.com>
15432 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
15433 stmt_info to record_stmt_cost.
15434 (vect_get_known_peeling_cost): Pass stmt_info if known to
15436 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
15437 cpu_vector_cost field into
15438 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
15439 field into vec_int_stmt_cost and vec_fp_stmt_cost.
15440 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
15441 splitting of scalar_stmt_cost and vec_stmt_cost.
15442 (thunderx_vector_cost): Likewise.
15443 (cortexa57_vector_cost): LIkewise.
15444 (exynosm1_vector_cost): Likewise.
15445 (xgene1_vector_cost): Likewise.
15446 (thunderx2t99_vector_cost): Improve after the splitting of the two
15448 (aarch64_builtin_vectorization_cost): Update for the splitting of
15449 scalar_stmt_cost and vec_stmt_cost.
15451 2017-02-01 Torvald Riegel <triegel@redhat.com>
15452 Richard Henderson <rth@redhat.com>
15454 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
15455 conditional on existance of a fast atomic load.
15456 * optabs-query.c (can_atomic_load_p): New function.
15457 * optabs-query.h (can_atomic_load_p): Declare it.
15458 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
15459 no fast atomic load is available for the particular size of access.
15460 (expand_atomic_compare_and_swap): Likewise.
15461 (expand_atomic_load): Likewise.
15462 (expand_atomic_store): Likewise.
15463 (expand_atomic_fetch_op): Likewise.
15464 * testsuite/lib/target-supports.exp
15465 (check_effective_target_sync_int_128): Remove x86 because it provides
15466 no fast atomic load.
15467 (check_effective_target_sync_int_128_runtime): Likewise.
15469 2017-02-01 Richard Biener <rguenther@suse.de>
15471 * graphite.c: Include tree-vectorizer.h for find_loop_location.
15472 (graphite_transform_loops): Provide opt-info for optimized nests.
15473 * tree-parloop.c (parallelize_loops): Provide opt-info for
15474 parallelized loops.
15476 2017-02-01 Richard Biener <rguenther@suse.de>
15478 PR middle-end/79315
15479 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
15480 was not set before.
15482 2017-02-01 Richard Biener <rguenther@suse.de>
15484 PR tree-optimization/71824
15485 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15486 Verify the loops are valid in the merged SESE region.
15487 (scop_detection::can_represent_loop_1): Check analyzing the
15488 evolution of the number of iterations in the region succeeds.
15490 2017-01-31 Ian Lance Taylor <iant@golang.org>
15492 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
15493 REG_ARGS_SIZE note to 32-bit push insns and call insn.
15495 2017-01-31 David Malcolm <dmalcolm@redhat.com>
15497 PR preprocessor/79210
15498 * input.c (get_substring_ranges_for_loc): Replace line_width
15499 assertion with error-handling.
15501 2017-01-31 Richard Biener <rguenther@suse.de>
15503 PR tree-optimization/77318
15504 * graphite-sese-to-poly.c (extract_affine): Fix assert.
15505 (create_pw_aff_from_tree): Take loop parameter.
15506 (add_condition_to_pbb): Pass loop of the condition to
15507 create_pw_aff_from_tree.
15509 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15511 * config/s390/s390.c (s390_asan_shadow_offset): New function.
15512 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
15514 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
15518 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
15520 (convert_int_to_float128): Likewise.
15521 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
15522 (convert_int_to_float128): Likewise.
15523 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
15524 (UNSPEC_IEEE128_CONVERT): Likewise.
15525 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
15526 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
15527 Use local variables for IBM extended format.
15528 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
15529 (fix_trunc<mode>si2_fprs): Likewise.
15530 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
15531 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
15532 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
15533 to know that we can now have integers of all sizes in vector
15535 (fix<uns>_<mode>di2_hw): Likewise.
15536 (float<uns>_<mode>si2_hw): Likewise.
15537 (fix_<mode>si2_hw): Likewise.
15538 (fixuns_<mode>si2_hw): Likewise.
15539 (float<uns>_<mode>di2_hw): Likewise.
15540 (float_<mode>di2_hw): Likewise.
15541 (float_<mode>si2_hw): Likewise.
15542 (floatuns_<mode>di2_hw): Likewise.
15543 (floatuns_<mode>si2_hw): Likewise.
15544 (xscvqp<su>wz_<mode>): Delete, no longer used.
15545 (xscvqp<su>dz_<mode>): Likewise.
15546 (xscv<su>dqp_<mode>): Likewise.
15547 (ieee128_mfvsrd_64bit): Likewise.
15548 (ieee128_mfvsrd_32bit): Likewise.
15549 (ieee128_mfvsrwz): Likewise.
15550 (ieee128_mtvsrw): Likewise.
15551 (ieee128_mtvsrd_64bit): Likewise.
15552 (ieee128_mtvsrd_32bit): Likewise.
15554 2017-01-31 Martin Liska <mliska@suse.cz>
15557 * ipa-prop.c (ipa_free_all_node_params): Call release method
15558 instead of ~sumbol_summary to not to trigger double times
15561 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
15563 PR tree-optimization/71691
15564 * bitmap.h (class auto_bitmap): New.
15565 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
15566 is_maybe_undefined instead of ssa_undefined_value_p.
15568 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15570 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15571 __S390_ARCH_LEVEL__ to __ARCH__.
15573 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15575 PR tree-optimization/79267
15576 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
15577 if should_remove_lhs_p is true.
15579 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
15582 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
15583 (add_alignment_attribute): New.
15584 (base_type_die): Add alignment attribute.
15585 (subrange_type_die): Likewise.
15586 (modified_type_die): Likewise.
15587 (gen_array_type_die): Likewise.
15588 (gen_descr_array_type_die: Likewise.
15589 (gen_enumeration_type_die): Likewise.
15590 (gen_subprogram_die): Likewise.
15591 (gen_variable_die): Likewise.
15592 (gen_field_die): Likewise.
15593 (gen_ptr_to_mbr_type_die): Likewise.
15594 (gen_struct_or_union_type_die): Likewise.
15595 (gen_subroutine_type_die): Likewise.
15596 (gen_typedef_die): Likewise.
15597 (base_type_cmp): Compare alignment attribute.
15599 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15602 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
15603 (setb_unsigned) New pattern for setb with CCUNS.
15604 * config/rs6000/rs6000.c (expand_block_compare): Use a different
15605 subfc./subfe sequence to avoid overflow problems. Generate a
15606 shorter sequence with cmpld/setb for power9.
15607 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
15608 for generating subfc. instruction.
15609 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
15610 now uses this instruction.
15612 2017-01-30 Ian Lance Taylor <iant@google.com>
15615 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
15616 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
15618 2017-01-30 Martin Sebor <msebor@redhat.com>
15620 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
15621 Move constant to the right of a relational operator.
15622 (get_mpfr_format_length, format_character, format_string): Ditto.
15623 (should_warn_p, maybe_warn): Same.
15625 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
15627 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
15630 * asan.c (get_translation_unit_decl): Remove function.
15631 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
15633 2017-01-30 Martin Liska <mliska@suse.cz>
15635 PR gcov-profile/79259
15636 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
15637 -fprofile-generate.
15639 2017-01-30 Martin Liska <mliska@suse.cz>
15642 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
15643 Initialize variables with NULL value.
15645 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
15648 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
15650 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
15652 2017-01-30 Richard Biener <rguenther@suse.de>
15654 PR tree-optimization/79276
15655 * tree-vrp.c (process_assert_insertions): Properly adjust common
15656 when removing a duplicate.
15658 * gcc.dg/torture/pr79276.c: New testcase.
15660 2017-01-30 Richard Biener <rguenther@suse.de>
15662 PR tree-optimization/79256
15663 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15664 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15666 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
15668 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
15671 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
15672 ("*r<noxa>sbg_<mode>_sll_bitmask")
15673 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
15674 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
15675 Use contiguous_bitmask_nowrap_operand.
15677 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15680 * config/rs6000/altivec.h (vec_xl): Revise #define.
15681 (vec_xst): Likewise.
15683 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
15685 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
15687 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
15689 PR rtl-optimization/79194
15690 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
15691 traps before call to bypass_conditional_jumps.
15693 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
15695 PR tree-optimization/71374
15696 * lra-constraints.c (check_conflict_input_operands): New.
15697 (match_reload): Use it.
15699 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
15702 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
15703 account to calculate conflict_set.
15705 2017-01-27 Bin Cheng <bin.cheng@arm.com>
15707 PR rtl-optimization/78559
15708 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
15709 other_insn in combine.
15711 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15713 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
15714 uint16_type_node for BT_UINT16.
15716 2017-01-27 David Malcolm <dmalcolm@redhat.com>
15718 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
15719 "RTL Tests" to menu.
15720 (GIMPLE Tests): New node.
15721 (RTL Tests): New node.
15723 2017-01-27 Richard Biener <rguenther@suse.de>
15725 PR tree-optimization/79245
15726 * tree-loop-distribution.c (distribute_loop): Apply cost
15727 modeling also to detected patterns.
15729 2017-01-27 Richard Biener <rguenther@suse.de>
15731 PR tree-optimization/71433
15732 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
15733 (compare_assert_loc): New function.
15734 (process_assert_insertions): Sort and optimize assert locations
15735 to remove duplicates and push down identical assertions on
15736 edges to their destination block.
15738 2017-01-27 Richard Biener <rguenther@suse.de>
15740 PR tree-optimization/79244
15741 * tree-vrp.c (remove_range_assertions): Forcefully propagate
15742 out SSA names even if abnormal.
15744 2017-01-27 Jakub Jelinek <jakub@redhat.com>
15746 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
15747 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
15748 instead of MPFR_RNDN.
15750 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
15753 * arm.c (arm_option_override): Don't call build_target_option_node
15754 until after doing all option overrides.
15755 (arm_valid_target_attribute_tree): Likewise.
15757 2017-01-27 Martin Liska <mliska@suse.cz>
15759 * doc/invoke.texi (-fprofile-arcs): Document profiling support
15760 for {cd}tors and C++ {cd}tors.
15762 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
15764 * config/s390/s390.md ("*setmem_long_and")
15765 ("*setmem_long_and_31z"): Use zero_extend instead of and.
15767 2017-01-26 Martin Sebor <msebor@redhat.com>
15769 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
15772 2017-01-26 Martin Sebor <msebor@redhat.com>
15774 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
15775 HAVE_DFmode before using XFmode or DFmode.
15776 (parse_directive): Avoid using the z length modifier to avoid
15777 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
15779 PR middle-end/78703
15780 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
15781 to accept adjustment as an array.
15782 (get_int_range): New function.
15783 (struct directive): Make width and prec arrays.
15784 (directive::set_width, directive::set_precision): Call get_int_range.
15785 (format_integer, format_floating): Handle width and precision ranges.
15786 (format_string, parse_directive): Same.
15788 2017-01-26 Jakub Jelinek <jakub@redhat.com>
15791 * dwarf2out.c (generate_skeleton_bottom_up): For children with
15792 comdat_type_p set, just clone them, but keep the children in the
15796 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
15797 which have direct callers with -fvar-tracking-assignments enabled
15799 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
15800 inside of type units.
15802 2017-01-26 Martin Sebor <msebor@redhat.com>
15804 PR middle-end/78703
15805 * gimple-ssa-sprintf.c (struct result_range): Add likely and
15807 (struct format_result): Replace number_chars, number_chars_min,
15808 and number_chars_max with a single member of struct result_range.
15810 (format_result::operator+=): Adjust.
15811 (struct fmtresult): Remove bounded. Handle likely and unlikely
15813 (fmtresult::adjust_for_width_or_precision): New function.
15814 (fmtresult:type_max_digits): New function.
15815 (bytes_remaining): Handle likely and unlikely counters.
15816 (min_bytes_remaining): Remove.
15817 (format_percent): Simplify.
15818 (format_integer, format_floating): Set likely and unlikely counters.
15819 (get_string_length, format_character, format_string): Same.
15820 (format_plain, should_warn_p): New function.
15821 (maybe_warn): Call should_warn_p. Update diagnostic messages
15822 and handle those for all directives, including plain strings.
15823 (format_directive): Handle likely and unlikely counters.
15824 Remove unnecessary quoting from diagnostics. Add an informational
15826 (add_bytes): Remove.
15827 (pass_sprintf_length::compute_format_length): Simplify.
15828 (try_substitute_return_value): Handle likely and unlikely counters.
15830 2017-01-26 Carl Love <cel@us.ibm.com>
15832 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
15833 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
15835 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
15838 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
15839 endianess for subregs into account.
15840 * lra-constraints.c (lra_constraints): Do risky transformations
15841 always on the first iteration.
15842 * lra-lives.c (check_pseudos_live_through_calls): Add arg
15843 last_call_used_reg_set.
15844 (process_bb_lives): Define and use last_call_used_reg_set.
15845 * lra.c (lra): Always continue after lra_constraints on the first
15848 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
15850 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
15852 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
15854 2017-01-26 Jakub Jelinek <jakub@redhat.com>
15856 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
15857 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
15858 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
15859 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
15860 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
15861 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
15862 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
15863 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
15864 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
15866 2017-01-26 Marek Polacek <polacek@redhat.com>
15869 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
15870 for the third operand.
15872 2017-01-26 Jakub Jelinek <jakub@redhat.com>
15874 PR middle-end/79236
15875 * omp-low.c (struct omp_context): Add simt_stmt field.
15876 (scan_omp_for): Return omp_context *.
15877 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
15878 context to the _simt_ SIMD stmt.
15879 (lower_omp_for): For combined SIMD with sibling _simt_
15880 SIMD, make sure to use the same decls in _looptemp_
15881 clauses as in the sibling.
15883 2017-01-26 David Sherwood <david.sherwood@arm.com>
15885 PR middle-end/79212
15886 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
15889 2017-01-26 Jakub Jelinek <jakub@redhat.com>
15892 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
15893 emit fld b; fld a; if possible.
15895 * brig-builtins.def: Update copyright years.
15896 * config/arm/arm_acle_builtins.def: Update copyright years.
15898 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15901 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
15902 constraint instead of o for the stxsd instruction.
15904 2017-01-25 Carl Love <cel@us.ibm.com>
15906 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
15907 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
15909 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
15911 * doc/invoke.texi (C++ Dialect Options): Fix typo.
15913 2017-01-25 Richard Biener <rguenther@suse.de>
15915 PR tree-optimization/69264
15916 * target.def (vector_alignment_reachable): Improve documentation.
15917 * doc/tm.texi: Regenerate.
15918 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
15920 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
15921 earlier changes with respect to TYPE_USER_ALIGN.
15922 (vector_alignment_reachable_p): Likewise. Improve dumping.
15924 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15927 * config/arm/arm.md (xordi3): Force constant operand into a register
15930 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15932 * doc/invoke.texi (-fstore-merging): Correct default optimization
15933 levels at which it is enabled.
15934 (-O): Move -fstore-merging from list to...
15937 2017-01-25 Richard Biener <rguenther@suse.de>
15940 * omp-expand.c: Include debug.h.
15941 (expand_omp_taskreg): Make sure to generate early debug before
15942 outlining anything from a function.
15943 (expand_omp_target): Likewise.
15944 (grid_expand_target_grid_body): Likewise.
15946 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
15949 * asan.c (get_translation_unit_decl): New function.
15950 (asan_add_global): Extract modules file name from globals
15951 TRANSLATION_UNIT_DECL name.
15953 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
15956 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
15957 for long calls with APCS frame and VFP.
15959 2017-01-24 David Malcolm <dmalcolm@redhat.com>
15961 * cfg.c (original_copy_tables_initialized_p): New function.
15962 * cfg.h (original_copy_tables_initialized_p): New decl.
15963 * cfgrtl.c (relink_block_chain): Guard the call to
15964 free_original_copy_tables with a call to
15965 original_copy_tables_initialized_p.
15966 * cgraph.h (symtab_node::native_rtl_p): New decl.
15967 * cgraphunit.c (symtab_node::native_rtl_p): New function.
15968 (symtab_node::needed_p): Don't assert for early assembly output
15969 for __RTL functions.
15970 (cgraph_node::finalize_function): Set "force_output" for __RTL
15972 (cgraph_node::analyze): Bail out early for __RTL functions.
15973 (analyze_functions): Update assertion to support __RTL functions.
15974 (cgraph_node::expand): Bail out early for __RTL functions.
15975 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
15977 * function.h (struct function): Update comment for field
15979 * gimple-expr.c: Include "tree-pass.h".
15980 (gimple_has_body_p): Return false for __RTL functions.
15981 * Makefile.in (OBJS): Add run-rtl-passes.o.
15982 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
15984 (gcc::pass_manager::get_clean_slate): New accessor.
15985 * passes.c: Include "insn-addr.h".
15986 (should_skip_pass_p): Add logging. Update logic for running
15987 "expand" to be compatible with both __GIMPLE and __RTL. Guard
15988 property-provider override so it is only done for gimple passes.
15990 (skip_pass): New function.
15991 (execute_one_pass): Call skip_pass when skipping passes.
15992 * read-md.c (md_reader::read_char): Support filtering
15993 the input to a subset of line numbers.
15994 (md_reader::md_reader): Initialize fields
15995 m_first_line and m_last_line.
15996 (md_reader::read_file_fragment): New function.
15997 * read-md.h (md_reader::read_file_fragment): New decl.
15998 (md_reader::m_first_line): New field.
15999 (md_reader::m_last_line): New field.
16000 * read-rtl-function.c (function_reader::create_function): Only
16001 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16002 curr_properties. Set DECL_INITIAL to a dummy block.
16003 (read_rtl_function_body_from_file_range): New function.
16004 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16006 * run-rtl-passes.c: New file.
16007 * run-rtl-passes.h: New file.
16009 2017-01-24 Jeff Law <law@redhat.com>
16011 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16014 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16016 PR tree-optimization/79159
16017 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16018 (record_nonwrapping_iv): Improve boundary using above function if no
16019 value range information.
16021 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16022 Martin Jambor <mjambor@suse.cz>
16024 * brig-builtins.def: New file.
16025 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16026 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16027 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16028 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16029 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16030 * builtin-types.def (BT_INT8): New.
16031 (BT_INT16): Likewise.
16032 (BT_UINT8): Likewise.
16033 (BT_UINT16): Likewise.
16034 (BT_FN_ULONG): Likewise.
16035 (BT_FN_UINT_INT): Likewise.
16036 (BT_FN_UINT_ULONG): Likewise.
16037 (BT_FN_UINT_LONG): Likewise.
16038 (BT_FN_UINT_PTR): Likewise.
16039 (BT_FN_ULONG_PTR): Likewise.
16040 (BT_FN_INT8_FLOAT): Likewise.
16041 (BT_FN_INT16_FLOAT): Likewise.
16042 (BT_FN_UINT32_FLOAT): Likewise.
16043 (BT_FN_UINT16_FLOAT): Likewise.
16044 (BT_FN_UINT8_FLOAT): Likewise.
16045 (BT_FN_UINT64_FLOAT): Likewise.
16046 (BT_FN_UINT16_UINT32): Likewise.
16047 (BT_FN_UINT32_UINT16): Likewise.
16048 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16049 (BT_FN_INT_PTR_INT): Likewise.
16050 (BT_FN_UINT_PTR_UINT): Likewise.
16051 (BT_FN_LONG_PTR_LONG): Likewise.
16052 (BT_FN_ULONG_PTR_ULONG): Likewise.
16053 (BT_FN_VOID_UINT64_UINT64): Likewise.
16054 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16055 (BT_FN_INT8_INT8_INT8): Likewise.
16056 (BT_FN_INT16_INT16_INT16): Likewise.
16057 (BT_FN_INT_INT_INT): Likewise.
16058 (BT_FN_UINT_FLOAT_UINT): Likewise.
16059 (BT_FN_FLOAT_UINT_UINT): Likewise.
16060 (BT_FN_ULONG_UINT_UINT): Likewise.
16061 (BT_FN_ULONG_UINT_PTR): Likewise.
16062 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16063 (BT_FN_UINT_UINT_UINT): Likewise.
16064 (BT_FN_VOID_UINT_PTR): Likewise.
16065 (BT_FN_UINT_UINT_PTR: Likewise.
16066 (BT_FN_UINT32_UINT64_PTR): Likewise.
16067 (BT_FN_INT_INT_UINT_UINT): Likewise.
16068 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16069 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16070 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16071 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16072 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16073 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16074 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16075 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16076 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16077 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16078 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16079 * doc/frontends.texi: List BRIG FE.
16080 * doc/install.texi (Testing): Add BRIG tesring requirements.
16081 * doc/invoke.texi (Overall Options): Mention BRIG.
16082 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16084 2017-01-24 Richard Biener <rguenther@suse.de>
16086 PR translation/79208
16087 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16089 2017-01-24 Martin Jambor <mjambor@suse.cz>
16092 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16093 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16094 and known_contexts.
16096 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16098 PR middle-end/79123
16099 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16100 casts from signed to unsigned really don't have a range.
16102 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16104 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16105 GMP_RNDx for compatiblity.
16107 2017-01-24 Martin Liska <mliska@suse.cz>
16110 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16111 that would prevent us to call alloca with -1 as argument.
16113 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16115 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16116 Avoid -Wformat-security warning.
16118 2017-01-23 Andrew Pinski <apinski@cavium.com>
16120 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16123 2017-01-23 Martin Sebor <msebor@redhat.com>
16125 PR middle-end/78703
16126 * gimple-ssa-sprintf.c (warn_level): New global.
16127 (format_integer): Use it here and throughout the rest of the file.
16128 Use the same switch to compute sign as base.
16129 (maybe_warn): New function.
16130 (format_directive): Factor out warnings into maybe_warn.
16131 Add debugging output. Use warn_level.
16132 (add_bytes): Use warn_level.
16133 (pass_sprintf_length::compute_format_length): Add debugging output.
16134 (try_substitute_return_value): Same.
16135 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16137 PR middle-end/78703
16138 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16139 (struct fmtresult, format_integer, format_floating): Adjust.
16140 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16141 (get_string_length, format_string,format_directive): Same.
16142 (pass_sprintf_length::compute_format_length): Same.
16143 (try_substitute_return_value): Simplify slightly.
16145 PR middle-end/78703
16146 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16147 (fmtresult::operator+=): Outlined.
16148 (struct fmtresult): Add ctors.
16149 (struct conversion_spec): Rename...
16150 (struct directive): ...to this. Add and remove data members.
16151 (directive::set_width, directive::set_precision): New functions.
16152 (format_percent): Use fmtresult ctor.
16153 (get_width_and_precision): Remove.
16154 (format_integer): Make naming changes. Avoid computing width and
16156 (format_floating): Same. Adjust indentation.
16157 (format_character, format_none): New functions.
16158 (format_string): Moved character handling to format_character.
16159 (format_directive): Remove arguments, change return type.
16160 (parse_directive): New function.
16161 (pass_sprintf_length::compute_format_length): Move directive
16162 parsing to parse_directive.
16164 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16166 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16167 (assign_assembler_name_if_needed): ... this.
16168 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16169 (assign_assembler_name_if_needed): ... this.
16170 (free_lang_data_in_cgraph): Adjust callers.
16171 * cgraphunit.c (cgraph_node::analyze): Likewise.
16172 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16175 2017-01-23 Richard Biener <rguenther@suse.de>
16177 PR tree-optimization/79088
16178 PR tree-optimization/79188
16179 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16180 resetting loop bounds after last path deletion. Reset loop
16181 bounds of the target loop, make code match the comments.
16182 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16183 Make sure loops need no fixups.
16185 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16187 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16188 exponent support with double type for first argument.
16189 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16190 type returned by __builtin_vec_extract_sig,
16191 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16192 functions from "vector int" to "vector unsigned int" or from
16193 "vector long long int" to "vector unsigned long long int".
16194 Changed type returned by __builtin_vec_extract_exp,
16195 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16196 functions from "vector int" to "vector unsigned int" or from
16197 "vector long long int" to "vector unsigned long long int".
16198 Changed return type of __builtin_vec_test_data_class,
16199 __builtin_vec_test_data_class_sp, and
16200 __builtin_vec_test_data_class_dp from "vector int" to
16201 "vector bool int" or from "vector long long int" to "vector bool
16202 long long int" and changed second argument type from "unsigned
16203 int" to "int". Added new overloaded function forms "vector float
16204 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16205 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16206 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16207 double, vector unsigned long long int)" and "vector double
16208 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16209 long int)". Changed return type of
16210 __builtin_scalar_test_data_class and
16211 __builtin_scalar_test_data_class_sp and
16212 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16213 int" and changed second argument from "unsigned int" to "int".
16214 Changed type returned by __builtin_scalar_test_neg,
16215 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16216 from "int" to "bool int". Added new overloaded function form
16217 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16218 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16219 exponent double-precision with floating point first argument.
16220 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16221 documentation of scalar_test_data_class, scalar_test_neg,
16222 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16223 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16224 vec_test_data_class built-in functions to reflect refinements in
16225 their type signatures.
16227 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16229 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16231 (aarch64_elf_asm_destructor): Likewise.
16233 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16235 PR rtl-optimization/78634
16236 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16237 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16238 * ifcvt.c (noce_try_cmove): Add missing cost check.
16240 PR rtl-optimization/71724
16241 * combine.c (if_then_else_cond): Look for situations where it is
16242 beneficial to undo the work of one of the recursive calls.
16244 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16246 PR tree-optimization/70754
16247 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16248 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16249 combined stmt before it if not NULL.
16250 (combine_chains): Process refs reversely and compute dominance point
16253 2017-01-23 Martin Liska <mliska@suse.cz>
16255 PR tree-optimization/79196
16256 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16257 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16259 (strlen_optimize_stmt): Call the renamed function.
16261 2017-01-23 Michael Matz <matz@suse.de>
16263 PR tree-optimization/78384
16264 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16266 2017-01-23 Richard Biener <rguenther@suse.de>
16268 PR tree-optimization/79186
16269 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16270 both incoming edges before moving an assert.
16272 2017-01-23 Martin Jambor <mjambor@suse.cz>
16274 * ipa-prop.c (load_from_param_1): Removed.
16275 (load_from_unmodified_param): Bits from load_from_param_1 put back
16277 (load_from_param): Removed.
16278 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16279 with stmt. Reverted back to use of load_from_unmodified_param.
16281 2017-01-23 Martin Jambor <mjambor@suse.cz>
16284 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16285 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16286 field a pointer to garbage collected vector, mark lattices and
16287 ipcp_orig_node with GTY((skip)).
16288 (ipa_get_param_count): Adjust to descriptors being a pointer.
16289 (ipa_get_param): Likewise.
16290 (ipa_get_type): Likewise.
16291 (ipa_get_param_move_cost): Likewise.
16292 (ipa_set_param_used): Likewise.
16293 (ipa_get_controlled_uses): Likewise.
16294 (ipa_set_controlled_uses): Likewise.
16295 (ipa_is_param_used): Likewise.
16296 (ipa_node_params_t): Move into garbage collector. New methods insert
16298 (ipa_node_params_sum): Annotate wth GTY(()).
16299 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16301 (ipa_load_from_parm_agg): Adjust declaration.
16302 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16303 * ipa-profile.c (ipa_profile): Likewise.
16304 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16305 (ipa_populate_param_decls): Make descriptors parameter garbage
16307 (ipa_dump_param): Adjust to descriptors being a pointer.
16308 (ipa_alloc_node_params): Likewise.
16309 (ipa_initialize_node_params): Likewise.
16310 (load_from_param_1): Make descriptors parameter garbage collected.
16311 (load_from_unmodified_param): Likewise.
16312 (load_from_param): Likewise.
16313 (ipa_load_from_parm_agg): Likewise.
16314 (ipa_node_params::~ipa_node_params): Removed.
16315 (ipa_free_all_node_params): Remove call to delete operator.
16316 (ipa_node_params_t::insert): New.
16317 (ipa_node_params_t::remove): Likewise.
16318 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16319 copy known_csts and known_contexts vectors.
16320 (ipa_read_node_info): Adjust to descriptors being a pointer.
16321 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16323 (ipcp_transform_function): Make descriptors variable garbage
16326 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16328 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16329 * config/i386/avx512dqintrin.h: Ditto.
16330 * config/i386/avx512fintrin.h: Ditto.
16331 * gcc/config/i386/i386.c: Handle new builtins.
16332 * config/i386/i386-builtin.def: Add new builtins.
16333 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16334 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16336 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16337 Martin Liska <mliska@suse.cz>
16339 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16340 * asan.c (asan_expand_poison_ifn): Support stores and use
16341 appropriate ASAN report function.
16342 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16343 * internal-fn.def (ASAN_POISON_USE): Declare.
16344 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16345 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16346 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16347 ASAN_POISON calls w/o LHS.
16348 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16349 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16350 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16351 * gimplify.c (asan_poison_variables): Add attribute
16352 use_after_scope_memory to variables that really needs to live
16354 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16355 having the attribute.
16357 2017-01-23 Martin Liska <mliska@suse.cz>
16359 * asan.c (create_asan_shadow_var): New function.
16360 (asan_expand_poison_ifn): Likewise.
16361 * asan.h (asan_expand_poison_ifn): New declaration.
16362 * internal-fn.c (expand_ASAN_POISON): Likewise.
16363 * internal-fn.def (ASAN_POISON): New builtin.
16364 * sanopt.c (pass_sanopt::execute): Expand
16365 asan_expand_poison_ifn.
16366 * tree-inline.c (copy_decl_for_dup_finish): Make function
16368 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16369 * tree-ssa.c (is_asan_mark_p): New function.
16370 (execute_update_addresses_taken): Rewrite local variables
16371 (identified just by use-after-scope as addressable) into SSA.
16373 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16375 * doc/install.texi (Specific): opensource.apple.com uses https
16376 now. Remove trailing slash.
16378 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16380 * README.Portability: Remove note on an Irix compatibility issue.
16382 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16384 * gcov.c (INCLUDE_ALGORITHM): Define.
16385 (INCLUDE_VECTOR): Define.
16386 No longer include <vector> and <algorithm> directly.
16388 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16390 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16392 * doc/invoke.texi (Code Gen Options): Ditto.
16394 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16397 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16399 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16401 rtl-optimization/79125
16402 * cprop.c (local_cprop_pass): Handle cases where we make an
16403 unconditional trap.
16405 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16409 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16410 read from, for big endian.
16412 2017-01-20 Jiong Wang <jiong.wang@arm.com>
16414 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
16415 register pauth builtins for LP64 only.
16417 2017-01-20 Marek Polacek <polacek@redhat.com>
16420 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
16423 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16425 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
16427 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
16428 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
16429 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
16431 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16434 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
16435 in big-endian mode when they are not a single duplicated value.
16437 2017-01-20 Richard Biener <rguenther@suse.de>
16439 * BASE-VER: Bump to 7.0.1.
16441 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16443 * omp-low.c (omplow_simd_context): New struct. Use it...
16444 (lower_rec_simd_input_clauses): ...here and...
16445 (lower_rec_input_clauses): ...here to hold common data. Adjust all
16446 references to idx, lane, max_vf, is_simt.
16448 2017-01-20 Graham Markall <graham.markall@embecosm.com>
16450 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
16453 2017-01-20 Martin Jambor <mjambor@suse.cz>
16455 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
16456 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
16458 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
16459 (GTFILES): Rename hsa.c to hsa-common.c.
16460 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
16461 * hsa-dump.c: Likewise.
16462 * hsa-gen.c: Likewise.
16463 * hsa-regalloc.c: Likewise.
16464 * ipa-hsa.c: Likewise.
16465 * omp-expand.c: Likewise.
16466 * omp-low.c: Likewise.
16467 * toplev.c: Likewise.
16469 2017-01-20 Marek Polacek <polacek@redhat.com>
16472 * doc/invoke.texi: Document -Wduplicated-branches.
16473 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
16474 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
16475 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
16476 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
16477 return 0 only when not OEP_LEXICOGRAPHIC.
16478 (fold_build_cleanup_point_expr): Use the expression
16479 location when building CLEANUP_POINT_EXPR.
16480 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
16481 * tree.c (add_expr): Handle error_mark_node.
16483 2017-01-20 Martin Liska <mliska@suse.cz>
16486 * tree-profile.c (init_ic_make_global_vars): Do not call
16488 (gimple_init_gcov_profiler): Likewise.
16490 2017-01-20 Martin Liska <mliska@suse.cz>
16493 * cgraph.h (maybe_create_reference): Remove argument and
16495 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
16497 * ipa-cp.c (create_specialized_node): Likewise.
16498 * symtab.c (symtab_node::maybe_create_reference): Handle
16499 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
16501 2017-01-20 Martin Liska <mliska@suse.cz>
16503 * read-rtl-function.c (function_reader::create_function): Use
16504 build_decl instread of build_decl_stat.
16506 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
16508 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
16509 * config/i386/avx512dqintrin.h: Ditto.
16510 * config/i386/avx512fintrin.h: Ditto.
16511 * config/i386/i386-builtin-types.def: Add new types.
16512 * gcc/config/i386/i386.c: Handle new types.
16513 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
16514 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
16515 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
16516 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
16517 (__builtin_ia32_kshiftridi): New.
16518 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
16520 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
16524 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
16525 define to rs6000_init_stack_protect_guard.
16526 (rs6000_init_stack_protect_guard): New function.
16528 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16529 Yunqiang Su <yunqiang.su@imgtec.com>
16531 * config.gcc (supported_defaults): Add madd4.
16532 (with_madd4): Add validation.
16533 (all_defaults): Add madd4.
16534 * config/mips/mips.opt (mmadd4): New option.
16535 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16537 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
16539 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
16540 (ISA_HAS_FUSED_MADD4): Likewise.
16541 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
16542 * gcc/doc/install.texi (--with-madd4): Document the new option.
16544 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16546 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
16547 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
16548 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
16549 (aarch64_init_pauth_hint_builtins): New.
16550 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
16551 (aarch64_expand_builtin): Expand new builtins.
16553 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16555 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
16556 * combine-stack-adj.c (no_unhandled_cfa): Handle
16557 REG_CFA_TOGGLE_RA_MANGLE.
16558 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
16559 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
16560 info for return address signing.
16561 (aarch64_expand_epilogue): Likewise.
16563 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16565 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
16566 * config/aarch64/aarch64-protos.h
16567 (aarch64_return_address_signing_enabled): New declaration.
16568 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
16570 (aarch64_expand_prologue): Sign return address before it's pushed onto
16572 (aarch64_expand_epilogue): Authenticate return address fetched from
16574 (aarch64_override_options): Sanity check for ILP32 and ISA level.
16575 (aarch64_attributes): New function attributes for "sign-return-address".
16576 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
16577 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
16578 ("*do_return"): Generate combined instructions according to key index.
16579 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
16580 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
16582 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
16583 * config/aarch64/aarch64.opt (msign-return-address=): New.
16584 * doc/extend.texi (AArch64 Function Attributes): Documents
16585 "sign-return-address=".
16586 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
16588 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16590 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
16591 overall option summary.
16593 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16595 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
16596 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
16597 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
16598 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
16600 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
16602 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
16603 -mpower9-minmax by default for -mcpu=power9.
16604 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
16605 128-bit floating point.
16607 2017-01-20 Alan Modra <amodra@gmail.com>
16609 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
16610 optimizing for size.
16612 2017-01-20 Alan Modra <amodra@gmail.com>
16615 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
16616 for strcmp and strncmp from corresponding builtin decl.
16618 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
16620 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
16621 instead of i386/rtems-64.h.
16622 * config/i386/rtems-64.h: Remove.
16624 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
16628 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
16630 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
16632 2017-01-19 Tamar Christina <tamar.christina@arm.com>
16634 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
16635 Change int to HOST_WIDE_INT.
16636 * config/aarch64/aarch64-protos.h
16637 (aarch64_simd_gen_const_vector_dup): Likewise.
16638 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
16640 2017-01-19 David Malcolm <dmalcolm@redhat.com>
16642 * langhooks-def.h (lhd_type_for_size): New decl.
16643 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
16644 * langhooks.c (lhd_type_for_size): New function, taken from
16647 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
16649 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
16650 define_bypass for CR latency.
16651 (power9-cracked-alu): Update bypass latency and remove power9-branch.
16652 (power9-alu2): Add define_bypass for CR latency.
16654 (power9-mul): Update insn latency.
16655 (power9-mul-compare): Update insn latency, bypass latency and remove
16658 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16660 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
16662 * config/aarch64/aarch64.md
16663 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
16664 aarch64_nopcrelative_literal_loads.
16665 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
16667 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
16669 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
16670 TARGET_LOONGSON_3A.
16671 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
16673 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
16676 * config.gcc (supported_defaults): Add lxc1-sxc1.
16677 (with_lxc1_sxc1): Add validation.
16678 (all_defaults): Add lxc1-sxc1.
16679 * config/mips/mips.opt (mlxc1-sxc1): New option.
16680 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16682 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
16683 __mips_no_lxc1_sxc1.
16684 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
16685 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
16686 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
16688 2017-01-19 Richard Biener <rguenther@suse.de>
16690 PR tree-optimization/72488
16691 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
16692 sure to restore SSA info.
16693 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
16695 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
16697 PR rtl-optimization/79121
16698 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
16699 of the inner type when shifting an extended value.
16701 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
16704 * symtab.c (symtab_node::equal_address_to): Fix comparing of
16705 interposable aliases.
16707 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
16710 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
16711 Use the evmergelohi instruction.
16712 (mov_si<mode>_e500_subreg4_2_le): Likewise.
16713 (mov_sitf_e500_subreg8_2_be): Likewise.
16714 (mov_sitf_e500_subreg12_2_le): Likewise.
16715 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
16716 (mov_si<mode>_e500_subreg4_2_be): Likewise.
16717 (mov_sitf_e500_subreg8_2_le): Likewise.
16718 (mov_sitf_e500_subreg12_2_be): Likewise.
16720 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16722 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
16723 attribute from vecsimple to vecperm.
16724 (altivec_vbpermq2): Likewise.
16726 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16729 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
16731 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16732 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
16733 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
16734 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
16735 case where N arg is SIZE_MAX.
16736 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
16737 (cmpstrsi): Add pattern.
16739 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16742 __builtin_vec_revb builtins.
16743 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
16744 built-in functions to support generation of the ISA 3.0 XXBR<x>
16745 vector byte reverse instructions.
16746 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
16747 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
16748 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
16749 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
16750 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
16751 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
16752 (P9V_BUILTIN_VEC_REVB): Likewise.
16753 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
16754 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
16755 (p9_xxbrq_v16qi): Likewise.
16756 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
16757 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
16758 (p9_xxbrh_v8hi): Likewise.
16759 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
16760 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
16761 vec_revb built-in functions.
16763 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
16765 PR rtl-optimization/78952
16766 * config/i386/i386.md (any_extract): New code iterator.
16767 (*insvqi_2): Use any_extract for source operand.
16768 (*insvqi_3): Use any_shiftrt for source operand.
16770 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
16772 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
16774 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
16776 2017-01-18 Matthias Klose <doko@ubuntu.com>
16778 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
16780 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16782 * config/rs6000/altivec.h (vec_bperm): Change #define.
16783 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
16784 (altivec_vbpermq2): New define_insn.
16785 (altivec_vbpermd): Likewise.
16786 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
16787 function interface.
16788 (VBPERMD): Likewise.
16789 (VBPERM): New polymorphic function interface.
16790 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
16791 Add entries for P9V_BUILTIN_VEC_VBPERM.
16792 * doc/extend.texi: Add interfaces for vec_bperm.
16794 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16796 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
16797 first letter of error messages.
16798 (s390_resolve_overloaded_builtin): Likewise.
16799 * config/s390/s390.c (s390_expand_builtin): Likewise.
16800 (s390_invalid_arg_for_unprototyped_fn): Likewise.
16801 (s390_valid_target_attribute_inner_p): Likewise.
16802 * config/s390/s390.md ("tabort"): Likewise.
16804 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
16806 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
16807 (ISA_AVOID_DIV_HILO): New macro.
16808 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
16809 (ISA_HAS_DDIV): Likewise.
16811 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
16813 * doc/invoke.texi (fabi-version): Correct number of occurrences.
16815 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
16817 * doc/invoke.texi (fabi-version): Spelling fix.
16819 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
16822 * doc/invoke.texi (fabi-version): Mention mangling fix for
16825 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
16828 * doc/invoke.texi (fabi-version): Document discriminator mangling.
16830 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
16833 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
16834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
16836 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
16838 (stack_protect_test): Ditto.
16839 * config/rs6000/rs6000.opt (mstack-protector-guard=,
16840 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
16842 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
16843 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
16844 -mstack-protector-guard-offset=.
16845 (RS/6000 and PowerPC Options): Ditto.
16847 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
16849 * config/i386/i386.h (MASK_CLASS_P): New define.
16850 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
16851 there are no registers from different register sets also when
16852 mask registers are used. Update function comment.
16853 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
16854 to (*k/*r) and (*k/*km) alternatives.
16856 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
16858 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
16859 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
16860 (EH_RETURN_HANDLER_RTX): New define.
16861 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
16862 Force frame pointer in EH return functions.
16863 (aarch64_expand_epilogue): Add barrier for eh_return.
16864 (aarch64_final_eh_return_addr): Remove.
16865 (aarch64_eh_return_handler_rtx): New function.
16866 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
16868 (aarch64_eh_return_handler_rtx): New prototype.
16870 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16872 * config/rs6000/altivec.h (vec_rlmi): New #define.
16873 (vec_vrlnm): Likewise.
16874 (vec_rlnm): Likewise.
16875 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
16876 (UNSPEC_VRLNM): Likewise.
16877 (VIlong): New mode iterator.
16878 (altivec_vrl<VI_char>mi): New define_insn.
16879 (altivec_vrl<VI_char>nm): Likewise.
16880 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
16882 (VRLDNM): Likewise.
16883 (RLNM): New polymorphic function entry.
16884 (VRLWMI): New monomorphic function entry.
16885 (VRLDMI): Likewise.
16886 (RLMI): New polymorphic function entry.
16887 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
16888 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
16889 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
16892 2017-01-17 Jakub Jelinek <jakub@redhat.com>
16895 * dwarf2out.c (field_byte_offset): Restore the
16896 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
16897 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
16898 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
16901 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
16904 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
16906 2017-01-17 Jakub Jelinek <jakub@redhat.com>
16909 * dwarf2out.c (add_data_member_location_attribute): For constant
16910 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
16911 instead of DW_AT_data_member_location, DW_AT_bit_offset and
16912 DW_AT_byte_size attributes.
16914 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
16916 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
16917 after forcing to constant memory when the code model is medium.
16919 2017-01-17 Julia Koval <julia.koval@intel.com>
16922 * config/i386/avx512fintrin.h
16923 (_mm512_i32gather_ps): Change __addr type to void const*.
16924 (_mm512_mask_i32gather_ps): Ditto.
16925 (_mm512_i32gather_pd): Ditto.
16926 (_mm512_mask_i32gather_pd): Ditto.
16927 (_mm512_i64gather_ps): Ditto.
16928 (_mm512_mask_i64gather_ps): Ditto.
16929 (_mm512_i64gather_pd): Ditto.
16930 (_mm512_mask_i64gather_pd): Ditto.
16931 (_mm512_i32gather_epi32): Ditto.
16932 (_mm512_mask_i32gather_epi32): Ditto.
16933 (_mm512_i32gather_epi64): Ditto.
16934 (_mm512_mask_i32gather_epi64): Ditto.
16935 (_mm512_i64gather_epi32): Ditto.
16936 (_mm512_mask_i64gather_epi32): Ditto.
16937 (_mm512_i64gather_epi64): Ditto.
16938 (_mm512_mask_i64gather_epi64): Ditto.
16939 (_mm512_i32scatter_ps): Change __addr type to void*.
16940 (_mm512_mask_i32scatter_ps): Ditto.
16941 (_mm512_i32scatter_pd): Ditto.
16942 (_mm512_mask_i32scatter_pd): Ditto.
16943 (_mm512_i64scatter_ps): Ditto.
16944 (_mm512_mask_i64scatter_ps): Ditto.
16945 (_mm512_i64scatter_pd): Ditto.
16946 (_mm512_mask_i64scatter_pd): Ditto.
16947 (_mm512_i32scatter_epi32): Ditto.
16948 (_mm512_mask_i32scatter_epi32): Ditto.
16949 (_mm512_i32scatter_epi64): Ditto.
16950 (_mm512_mask_i32scatter_epi64): Ditto.
16951 (_mm512_i64scatter_epi32): Ditto.
16952 (_mm512_mask_i64scatter_epi32): Ditto.
16953 (_mm512_i64scatter_epi64): Ditto.
16954 (_mm512_mask_i64scatter_epi64): Ditto.
16955 * config/i386/avx512pfintrin.h
16956 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
16957 (_mm512_mask_prefetch_i32gather_ps): Ditto.
16958 (_mm512_mask_prefetch_i64gather_pd): Ditto.
16959 (_mm512_mask_prefetch_i64gather_ps): Ditto.
16960 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
16961 (_mm512_prefetch_i32scatter_ps): Ditto.
16962 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
16963 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
16964 (_mm512_prefetch_i64scatter_pd): Ditto.
16965 (_mm512_prefetch_i64scatter_ps): Ditto.
16966 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
16967 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
16968 * config/i386/avx512vlintrin.h
16969 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
16970 (_mm_mmask_i32gather_ps): Ditto.
16971 (_mm256_mmask_i32gather_pd): Ditto.
16972 (_mm_mmask_i32gather_pd): Ditto.
16973 (_mm256_mmask_i64gather_ps): Ditto.
16974 (_mm_mmask_i64gather_ps): Ditto.
16975 (_mm256_mmask_i64gather_pd): Ditto.
16976 (_mm_mmask_i64gather_pd): Ditto.
16977 (_mm256_mmask_i32gather_epi32): Ditto.
16978 (_mm_mmask_i32gather_epi32): Ditto.
16979 (_mm256_mmask_i32gather_epi64): Ditto.
16980 (_mm_mmask_i32gather_epi64): Ditto.
16981 (_mm256_mmask_i64gather_epi32): Ditto.
16982 (_mm_mmask_i64gather_epi32): Ditto.
16983 (_mm256_mmask_i64gather_epi64): Ditto.
16984 (_mm_mmask_i64gather_epi64): Ditto.
16985 (_mm256_i32scatter_ps): Change __addr type to void*.
16986 (_mm256_mask_i32scatter_ps): Ditto.
16987 (_mm_i32scatter_ps): Ditto.
16988 (_mm_mask_i32scatter_ps): Ditto.
16989 (_mm256_i32scatter_pd): Ditto.
16990 (_mm256_mask_i32scatter_pd): Ditto.
16991 (_mm_i32scatter_pd): Ditto.
16992 (_mm_mask_i32scatter_pd): Ditto.
16993 (_mm256_i64scatter_ps): Ditto.
16994 (_mm256_mask_i64scatter_ps): Ditto.
16995 (_mm_i64scatter_ps): Ditto.
16996 (_mm_mask_i64scatter_ps): Ditto.
16997 (_mm256_i64scatter_pd): Ditto.
16998 (_mm256_mask_i64scatter_pd): Ditto.
16999 (_mm_i64scatter_pd): Ditto.
17000 (_mm_mask_i64scatter_pd): Ditto.
17001 (_mm256_i32scatter_epi32): Ditto.
17002 (_mm256_mask_i32scatter_epi32): Ditto.
17003 (_mm_i32scatter_epi32): Ditto.
17004 (_mm_mask_i32scatter_epi32): Ditto.
17005 (_mm256_i32scatter_epi64): Ditto.
17006 (_mm256_mask_i32scatter_epi64): Ditto.
17007 (_mm_i32scatter_epi64): Ditto.
17008 (_mm_mask_i32scatter_epi64): Ditto.
17009 (_mm256_i64scatter_epi32): Ditto.
17010 (_mm256_mask_i64scatter_epi32): Ditto.
17011 (_mm_i64scatter_epi32): Ditto.
17012 (_mm_mask_i64scatter_epi32): Ditto.
17013 (_mm256_i64scatter_epi64): Ditto.
17014 (_mm256_mask_i64scatter_epi64): Ditto.
17015 (_mm_i64scatter_epi64): Ditto.
17016 (_mm_mask_i64scatter_epi64): Ditto.
17017 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17018 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17019 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17020 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17021 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17022 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17023 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17024 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17025 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17026 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17027 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17028 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17029 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17030 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17031 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17032 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17033 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17034 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17035 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17036 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17037 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17038 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17039 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17040 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17041 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17042 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17043 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17044 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17045 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17046 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17047 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17048 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17049 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17050 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17051 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17052 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17053 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17054 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17055 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17056 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17057 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17058 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17059 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17060 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17061 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17062 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17063 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17064 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17065 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17066 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17067 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17068 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17069 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17070 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17071 definitions accordingly.
17073 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17074 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17077 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17080 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17083 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17084 hard regs for inner regno.
17086 2017-01-17 Martin Liska <mliska@suse.cz>
17089 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17090 assumption and add comment.
17092 2017-01-17 Nathan Sidwell <nathan@acm.org>
17094 * ipa-visibility.c (localize_node): New function, broken out of ...
17095 (function_and_variable_visibility): ... here. Call it.
17097 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17099 PR middle-end/77445
17100 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17101 correctly set frequency of oudgoing edge.
17102 (duplicate_thread_path): Fix profile updating.
17104 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17107 * configure.ac: Add GCC_BASE_VER.
17108 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17109 version from BASE-VER file.
17110 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17111 (gcc.o): Depend on $(BASEVER).
17112 * common.opt (dumpfullversion): New option.
17113 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17114 * doc/invoke.texi: Document -dumpfullversion.
17115 * doc/install.texi: Document --with-gcc-major-version-only.
17116 * configure: Regenerated.
17118 2017-01-17 Richard Biener <rguenther@suse.de>
17120 PR tree-optimization/71433
17121 * tree-vrp.c (register_new_assert_for): Merge same asserts
17122 on all incoming edges.
17123 (process_assert_insertions_for): Handle insertions at the
17126 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17128 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17129 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17131 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17134 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17137 2017-01-17 Alan Modra <amodra@gmail.com>
17140 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17141 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17142 symbolic stack limit when pic.
17144 2017-01-16 Martin Sebor <msebor@redhat.com>
17146 PR tree-optimization/78608
17147 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17149 2017-01-16 Jeff Law <law@redhat.com>
17152 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17153 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17154 for several include directories that may be relative to sysroot.
17155 * config/i386/x-mingw32 (gplus_includedir): Define.
17156 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17157 (native_system_includedir): Likewise.
17158 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17159 override if TARGET_SYSTEM_ROOT is defined.
17160 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17162 PR tree-optimization/79090
17163 PR tree-optimization/33562
17164 PR tree-optimization/61912
17165 PR tree-optimization/77485
17166 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17167 and computed trims into the dump file.
17169 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17171 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17173 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17176 * gimplify.c (gimplify_init_constructor): If want_value and
17177 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17181 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17182 sequence. Formatting fixes.
17183 (doloop_optimize): Formatting fixes.
17186 * gcc.c (debug_level_greater_than_spec_func): New function.
17187 (static_spec_functions): Add debug-level-gt spec function.
17188 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17190 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17191 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17192 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17193 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17194 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17195 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17197 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17199 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17200 QImode fixups to general and mask registers only.
17202 2017-01-16 Carl Love <cel@us.ibm.com>
17204 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17205 for built-in functions
17206 vector signed char vec_nabs (vector signed char)
17207 vector signed short vec_nabs (vector signed short)
17208 vector signed int vec_nabs (vector signed int)
17209 vector signed long long vec_nabs (vector signed long long)
17210 vector float vec_nabs (vector float)
17211 vector double vec_nabs (vector double)
17212 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17214 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17215 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17216 * doc/extend.texi: Update the documentation file for the new built-in
17219 2017-01-16 Martin Sebor <msebor@redhat.com>
17221 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17224 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17226 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17227 UNSPEC_VSX__XXSPLTD to require special splat handling.
17229 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17232 * system.h: Poison strndup.
17234 2017-01-16 Alan Modra <amodra@gmail.com>
17237 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17240 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17242 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17244 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17246 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17247 call recog here. Assert that INSN_CODE (insn) is non-negative.
17249 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17252 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17254 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17255 in the currently scheduled RTL fragment.
17257 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17259 PR rtl-optimization/78751
17260 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17263 2017-01-14 Jeff Law <law@redhat.com>
17265 PR tree-optimization/79090
17266 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17267 variable length stores.
17268 (compute_trims): Delete dead assignment to *trim_tail.
17269 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17272 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17274 PR rtl-optimization/78626
17275 PR rtl-optimization/78727
17276 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17277 of a block, and split such blocks after everything else is finished.
17279 2017-01-14 Alan Modra <amodra@gmail.com>
17282 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17283 target legitimate_combined_insn.
17284 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17285 (rs6000_legitimate_combined_insn): New function.
17286 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17288 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17289 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17290 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17292 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17294 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17296 2017-01-13 Jeff Law <law@redhat.com>
17298 PR tree-optimization/33562
17299 PR tree-optimization/61912
17300 PR tree-optimization/77485
17301 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17303 (delete_dead_assignment): Likewise.
17304 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17305 statement to delete_dead_call and delete_dead_assignment.
17307 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17310 * substring-locations.c (format_warning_va): Strengthen case 1 so
17311 that both endpoints of the substring must be within the format
17312 range for just the substring to be printed.
17314 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17316 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17317 * config/i386/i386.c (ix86_target_string): Add missing options
17318 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17319 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17320 flags_other and ix86_target_other to flags2_other. Display unknown
17322 (ix86_valid_target_attribute_inner_p): Add missing options and
17323 reorder options by implied ISAs, as in ix86_target_string.
17325 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17327 * hash-table.h (hash_table::too_empty_p): New function.
17328 (hash_table::expand): Use it.
17329 (hash_table::traverse): Likewise.
17330 (hash_table::empty_slot): Use sizeof (value_type) instead of
17331 sizeof (PTR) to convert bytes to elements. Shrink the table
17332 if the current size is excessive for the current number of
17335 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17337 * ira-costs.c (record_reg_classes): Break from the inner loop
17338 early once alt_fail is known to be true. Update outer loop
17339 handling accordingly.
17341 2017-01-13 Jeff Law <law@redhat.com>
17343 * tree-ssa-dse.c (decrement_count): New function.
17344 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17345 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17346 when we know the partially dead statement is a mem* function.
17348 PR tree-optimization/61912
17349 PR tree-optimization/77485
17350 * tree-ssa-dse.c: Include expr.h.
17351 (maybe_trim_constructor_store): New function.
17352 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17354 PR tree-optimization/33562
17355 PR tree-optimization/61912
17356 PR tree-optimization/77485
17357 * doc/invoke.texi: Document new dse-max-object-size param.
17358 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17359 * tree-ssa-dse.c: Include params.h.
17360 (dse_store_status): New enum.
17361 (initialize_ao_ref_for_dse): New, partially extracted from
17363 (valid_ao_ref_for_dse, normalize_ref): New.
17364 (setup_live_bytes_from_ref, compute_trims): Likewise.
17365 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17366 (maybe_trim_partially_dead_store): Likewise.
17367 (maybe_trim_complex_store): Likewise.
17368 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17369 Track what bytes live from the original store. Return tri-state
17370 for dead, partially dead or live.
17371 (dse_dom_walker): Add constructor, destructor and new private members.
17372 (delete_dead_call, delete_dead_assignment): New extracted from
17374 (dse_optimize_stmt): Make a member of dse_dom_walker.
17375 Use initialize_ao_ref_for_dse.
17377 PR tree-optimization/33562
17378 PR tree-optimization/61912
17379 PR tree-optimization/77485
17380 * sbitmap.h (bitmap_count_bits): Prototype.
17381 (bitmap_clear_range, bitmap_set_range): Likewise.
17382 * sbitmap.c (bitmap_clear_range): New function.
17383 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17385 2017-01-13 Martin Liska <mliska@suse.cz>
17388 * function.c (set_cfun): Add new argument force.
17389 * function.h (set_cfun): Likewise.
17390 * ipa-inline-transform.c (inline_call): Use the function when
17391 strict alising from is dropped for function we inline to.
17393 2017-01-13 Richard Biener <rguenther@suse.de>
17395 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17396 for dumping GIMPLE INTEGER_CSTs.
17398 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17400 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17401 to 201112L since C++17.
17403 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17406 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17407 if -fsanitize=kernel-address is present.
17409 2017-01-13 Richard Biener <rguenther@suse.de>
17411 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
17412 as _Literal ( type ) number in case usual suffixes do not
17413 preserve all information.
17415 2017-01-13 Richard Biener <rguenther@suse.de>
17417 PR tree-optimization/77283
17418 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
17419 and ssa-iterators.h.
17420 (is_feasible_trace): Implement a cost model based on joiner
17423 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17426 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
17427 char or short to __float128/_Float128 directly.
17429 2017-01-12 Martin Sebor <msebor@redhat.com>
17431 to -Wformat-overflow.
17432 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
17433 (min_bytes_remaining): Same.
17434 (get_string_length): Same.
17435 (format_string): Same.
17436 (format_directive): Same.
17438 (pass_sprintf_length::handle_gimple_call): Same.
17440 2017-01-12 Jakub Jelinek <jakub@redhat.com>
17442 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
17443 info.nowrite calls with no lhs that can't throw. Return bool
17444 whether gsi_remove has been called or not.
17445 (pass_sprintf_length::handle_gimple_call): Return bool whether
17446 try_substitute_return_value called gsi_remove. Formatting fix.
17447 (pass_sprintf_length::execute): Don't use gsi_remove if
17448 handle_gimple_call returned true.
17451 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
17452 be removed due to side-effects, don't remove following barrier nor
17453 turn the successor edge into fallthru edge.
17455 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17458 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
17459 element-reversing loads and stores as not swappable.
17461 2017-01-12 Nathan Sidwell <nathan@acm.org>
17462 Nicolai Stange <nicstange@gmail.com>
17464 * combine.c (try_combine): Don't ignore result of overlap checking
17465 loop. Combine overlap & asm check into single loop.
17467 2017-01-12 Richard Biener <rguenther@suse.de>
17469 * tree-pretty-print.c (dump_generic_node): Provide -gimple
17470 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
17472 2017-01-12 Richard Biener <rguenther@suse.de>
17474 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
17475 and TS_TARGET_OPTION directly derive from TS_BASE.
17476 * tree-core.h (tree_optimization_option): Derive from tree_base.
17477 (tree_target_option): Likewise.
17479 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
17481 * config/i386/i386.c (memory_address_length): Increase len
17482 only when rip_relative_addr_p returns false.
17484 2017-01-11 Julia Koval <julia.koval@intel.com>
17486 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
17487 (OPTION_MASK_ISA_SGX_SET): New.
17488 (ix86_handle_option): Handle OPT_msgx.
17489 * config.gcc: Added sgxintrin.h.
17490 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
17491 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
17492 * config/i386/i386.c (ix86_target_string): Add -msgx.
17494 (ix86_option_override_internal): Handle new options.
17495 (ix86_valid_target_attribute_inner_p): Add sgx.
17496 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
17497 * config/i386/i386.opt: Add msgx.
17498 * config/i386/sgxintrin.h: New file.
17499 * config/i386/x86intrin.h: Add sgxintrin.h.
17501 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17504 * fold-const.c (maybe_nonzero_address): Return 1 for function
17506 (tree_single_nonzero_warnv_p): Don't handle function local objects
17510 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
17513 2017-01-11 David Malcolm <dmalcolm@redhat.com>
17516 * opts.c: Include "spellcheck.h"
17517 (struct string_fragment): New struct.
17518 (struct edit_distance_traits<const string_fragment &>): New
17520 (get_closest_sanitizer_option): New function.
17521 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
17523 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17525 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
17527 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
17528 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
17529 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
17530 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
17531 for initial die_offset if dwarf_split_debug_info.
17532 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
17533 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
17535 (output_skeleton_debug_sections): Formatting fix. Use
17536 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
17537 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
17539 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
17541 * config/arm/cortex-a53.md: Add bypasses for
17542 cortex_a53_r2f_cvt.
17543 (cortex_a53_r2f): Only use for transfers.
17544 (cortex_a53_f2r): Likewise.
17545 (cortex_a53_r2f_cvt): Add reservation for conversions.
17546 (cortex_a53_f2r_cvt): Likewise.
17548 2017-01-11 Tamar Christina <tamar.christina@arm.com>
17550 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
17551 to all inlined functions, change static to extern.
17553 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
17556 * config/arm/arm.c (legitimize_pic_address): Handle reference to
17558 (arm_assemble_integer): Likewise.
17560 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
17562 * config.gcc: Use new awk script to check CPU, FPU and architecture
17563 parameters for --with-... options.
17564 * config/arm/parsecpu.awk: New file
17565 * config/arm/arm-cpus.in: New file.
17566 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
17568 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
17570 * config/arm/t-arm: Update dependency rules.
17571 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
17572 of processing .def files.
17573 * config/arm/genopt.sh: Deleted.
17574 * config/arm/gentune.sh: Deleted.
17575 * config/arm/arm-cores.def: Deleted.
17576 * config/arm/arm-arches.def: Deleted.
17577 * config/arm/arm-fpus.def: Deleted.
17578 * config/arm/arm-tune.md: Regenerated.
17579 * config/arm/arm-tables.opt: Regenerated.
17580 * config/arm/arm-cpu.h: New generated file.
17581 * config/arm/arm-cpu-data.h: New generated file.
17582 * config/arm/arm-cpu-cdata.h: New generated file.
17584 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
17587 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
17589 (input_varpool_node): Unpack dynamically_initialized bit.
17591 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
17593 PR rtl-optimization/79032
17594 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
17595 the alignment of the adjusted memory reference against that of MODE,
17596 instead of the alignment of the original memory reference.
17598 2017-01-11 Martin Jambor <mjambor@suse.cz>
17600 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
17602 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
17603 decorated functions.
17605 2017-01-11 Richard Biener <rguenther@suse.de>
17607 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
17608 set range/nonnull info for PHI results. Do not set it on
17609 stmts marked for removal.
17611 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
17613 * expr.c (store_field): In the bitfield case, fetch the return value
17614 from the registers before applying a single big-endian adjustment.
17615 Always do a final load for a BLKmode value not larger than a word.
17617 2017-01-10 David Malcolm <dmalcolm@redhat.com>
17620 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
17621 that we correctly handle column numbers greater than
17622 LINE_MAP_MAX_COLUMN_NUMBER.
17624 2017-01-10 Martin Sebor <msebor@redhat.com>
17626 PR middle-end/78245
17627 * gimple-ssa-sprintf.c (get_destination_size): Call
17628 {init,fini}object_sizes.
17629 * tree-object-size.c (addr_object_size): Adjust.
17630 (pass_through_call): Adjust.
17631 (pass_object_sizes::execute): Adjust.
17632 * tree-object-size.h (fini_object_sizes): Declare.
17634 2017-01-10 Martin Sebor <msebor@redhat.com>
17636 PR tree-optimization/78775
17637 * builtins.c (get_size_range): Move...
17638 * calls.c: ...to here.
17639 (alloc_max_size): Accept zero argument.
17640 (operand_signed_p): Remove.
17641 (maybe_warn_alloc_args_overflow): Call get_size_range.
17642 * calls.h (get_size_range): Declare.
17644 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
17646 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
17647 from TI's devices.csv file as of September 2016.
17648 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
17650 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
17652 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
17653 * doc/invoke.texi: Likewise.
17654 * doc/md.texi: Likewise.
17655 * doc/objc.texi: Likewise.
17657 2017-01-10 Joshua Conner <joshconner@google.com>
17659 * config/arm/fuchsia-elf.h: New file.
17660 * config/fuchsia.h: New file.
17661 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
17662 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
17664 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
17666 2016-01-10 Richard Biener <rguenther@suse.de>
17668 PR tree-optimization/79034
17669 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17670 Propagate out degenerate PHIs in the joiner.
17672 2017-01-10 Martin Liska <mliska@suse.cz>
17674 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
17675 (sort_congruence_classes_by_decl_uid): Likewise.
17676 (sort_congruence_class_groups_by_decl_uid): Likewise.
17677 (sem_item_optimizer::merge_classes): Sort class, groups in these
17678 classes and members in the groups by DECL_UID of declarations.
17679 This would make merge operations stable.
17681 2017-01-10 Martin Liska <mliska@suse.cz>
17683 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
17684 usage of m_classes_vec.
17685 (sem_item_optimizer::~sem_item_optimizer): Likewise.
17686 (sem_item_optimizer::get_group_by_hash): Likewise.
17687 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
17688 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
17689 (sem_item_optimizer::verify_classes): Likewise.
17690 (sem_item_optimizer::process_cong_reduction): Likewise.
17691 (sem_item_optimizer::dump_cong_classes): Likewise.
17692 (sem_item_optimizer::merge_classes): Likewise.
17693 * ipa-icf.h (congruence_class_hash): Rename from
17694 congruence_class_group_hash. Remove declaration of m_classes_vec.
17696 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
17698 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
17699 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
17700 * config.gcc: Add avx512vpopcntdqintrin.h.
17701 * config/i386/avx512vpopcntdqintrin.h: New.
17702 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
17703 * config/i386/i386-builtin-types.def: Add new types.
17704 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
17705 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
17706 __builtin_ia32_vpopcountq_v8di_mask): New.
17707 * config/i386/i386-c.c (ix86_target_macros_internal): Define
17708 __AVX512VPOPCNTDQ__.
17709 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
17710 (PTA_AVX512VPOPCNTDQ): Define.
17711 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
17712 TARGET_AVX512VPOPCNTDQ_P): Define.
17713 * config/i386/i386.opt: Add mavx512vpopcntdq.
17714 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
17715 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
17717 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
17719 PR middle-end/77484
17720 * predict.def (PRED_CALL): Set to 67.
17722 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
17724 * expr.c (store_field): In the bitfield case, if the value comes from
17725 a function call and is of an aggregate type returned in registers, do
17726 not modify the field mode; extract the value in all cases if the mode
17727 is BLKmode and the size is not larger than a word.
17729 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
17732 * config/i386/cpuid.h: Fix undefined behavior.
17734 2017-01-04 Jeff Law <law@redhat.com>
17736 PR tree-optimization/79007
17737 PR tree-optimization/67955
17738 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
17739 conservative for pt.null when flag_non_call_exceptions is on.
17741 2017-01-09 Jakub Jelinek <jakub@redhat.com>
17743 PR translation/79019
17744 PR translation/79020
17745 * params.def (PARAM_INLINE_MIN_SPEEDUP,
17746 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
17747 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
17749 * config/avr/avr.opt (maccumulate-args): Likewise.
17750 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
17751 * common.opt (freport-bug): Likewise.
17752 * cif-code.def (CIF_FINAL_ERROR): Likewise.
17753 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
17754 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
17755 translatable string.
17756 * config/i386/i386.c (function_value_32): Likewise.
17757 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
17758 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
17760 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
17761 * common/config/msp430/msp430-common.c (msp430_handle_option):
17763 * symtab.c (symtab_node::verify_base): Likewise.
17764 * opts.c (set_debug_level): Likewise.
17765 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
17766 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
17767 missing whitespace to translatable strings.
17768 * config/avr/avr.md (bswapsi2): Fix typo in comment.
17769 * config/sh/superh.h: Likewise.
17770 * config/i386/xopintrin.h: Likewise.
17771 * config/i386/znver1.md: Likewise.
17772 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
17773 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
17774 * double-int.h (struct double_int): Likewise.
17775 * double-int.c (div_and_round_double): Likewise.
17776 * wide-int.cc: Likewise.
17777 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
17778 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
17779 * cfgcleanup.c (crossjumps_occured): Renamed to ...
17780 (crossjumps_occurred): ... this.
17781 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
17784 PR tree-optimization/78899
17785 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
17786 returning bool return struct loop *, NULL for failure and the new
17788 (versionable_outer_loop_p): Don't version outer loop if it has
17789 dont_vectorized bit set.
17790 (tree_if_conversion): When versioning outer loop, ensure
17791 tree_if_conversion is performed also on the inner loop of the
17792 non-vectorizable outer loop copy.
17793 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
17794 LOOP_VECTORIZED in inner loop of the scalar outer loop and
17795 prevent vectorization of it.
17796 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
17797 the outer loop vectorization of the non-scalar version is attempted
17798 before vectorization of the inner loop in scalar version. If
17799 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
17800 vectorization of its inner loop.
17801 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
17802 has 2 inner loops, rename also on edges from bb whose single pred
17803 is outer_loop->header. Fix typo in function comment.
17805 2017-01-09 Martin Sebor <msebor@redhat.com>
17808 * asan.c (asan_emit_stack_protection): Increase local buffer size
17809 to avoid snprintf truncation warning.
17811 2017-01-09 Andrew Pinski <apinski@cavium.com>
17813 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
17814 to reference thunderx2t99 for the tuning structure
17815 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
17817 (thunderx2t99_extra_costs): This.
17818 * config/aarch64/aarch64-tune.md: Regenerate.
17819 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
17820 (vulcan_addrcost_table): This.
17821 (vulcan_regmove_cost): Rename to ...
17822 (thunderx2t99_regmove_cost): This.
17823 (vulcan_vector_cost): Rename to ...
17824 (thunderx2t99_vector_cost): this.
17825 (vulcan_branch_cost): Rename to ...
17826 (thunderx2t99_branch_cost): This.
17827 (vulcan_tunings): Rename to ...
17828 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
17829 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
17831 2017-01-09 Martin Jambor <mjambor@suse.cz>
17835 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
17836 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
17837 (propagate_vr_accross_jump_function): Use the above function for all
17838 value range computations for pass-through jump functions and type
17839 converasion from explicit value range values.
17840 (ipcp_propagate_stage): Do not attempt to deduce types of formal
17841 parameters from TYPE_ARG_TYPES.
17842 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
17843 (ipa_write_node_info): Stream type of the actual argument.
17844 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
17846 2017-01-09 Martin Liska <mliska@suse.cz>
17849 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
17850 (lookup_compiler): Do not show error message with have_E.
17852 2017-01-09 Jakub Jelinek <jakub@redhat.com>
17854 PR tree-optimization/78938
17855 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
17856 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
17857 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
17858 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
17861 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17863 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
17866 2017-01-09 Richard Biener <rguenther@suse.de>
17868 PR tree-optimization/78997
17869 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
17870 name condition properly.
17872 2017-01-09 Richard Biener <rguenther@suse.de>
17875 * dwarf2out.c (is_cxx): New overload with context.
17876 (is_naming_typedef_decl): Use it.
17878 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
17880 * invoke.texi (Option Summary): Correct spacing in option lists
17881 and add line breaks to fix over-long lines.
17883 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
17885 PR middle-end/17660
17887 * extend.texi (Common Variable Attributes): Add xref to GCC
17888 Internals manual to explain mode attribute keywords.
17890 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
17893 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
17894 and Preprocessor Options.
17895 (Options for Linking): Document -pthread here....
17896 (RS/6000 and PowerPC Options): ...not here.
17897 (Solaris 2 Options): ...or here.
17898 * doc/cppopts.texi: Document -pthread.
17900 2017-01-08 Martin Sebor <msebor@redhat.com>
17902 PR middle-end/77708
17903 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
17904 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
17905 New member functions.
17906 (format_directive): Used them.
17908 (pass_sprintf_length::handle_gimple_call): Same.
17909 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
17910 to avoid truncation for any argument.
17911 (extract_affine_mul): Same.
17912 * tree.c (get_file_function_name): Same.
17914 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
17916 PR middle-end/77484
17917 * predict.def (PRED_INDIR_CALL): Set to 86.
17919 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
17921 PR preprocessor/54124
17922 * doc/cppopts.texi: Reformat -d subtable to list the full name
17923 of the options. Add cross-reference to the docs for the general
17924 compiler -d options.
17925 * doc/invoke.texi (Developer Options): Add cross-reference to the
17926 preprocessor-specific -d option documentation.
17928 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
17930 PR preprocessor/13498
17931 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
17932 redudant material, and reflect new command-line options.
17933 (System Headers): Likewise.
17935 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
17937 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
17938 -isystem, and -idirafter. Copy-edit.
17939 * doc/cppopts.texi: Copy-edit. Remove contradiction about
17940 default for -ftrack-macro-expansion. Delete obsolete and
17941 badly-formatted implementation details about -fdebug-cpp output.
17942 * doc/cppwarnopts.texi: Copy-edit.
17944 2017-01-07 David Malcolm <dmalcolm@redhat.com>
17947 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
17948 that the transition from a max line width >= 1<<10 to narrower
17949 lines works correctly.
17951 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
17953 * doc/options.texi (PerFunction): New.
17954 * opt-functions.awk (switch_flags): Map both Optimization and
17955 PerFunction to CL_OPTIMIZATION.
17956 * opth-gen.awk: Test for PerFunction flag along with
17958 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
17959 it only when the latter is present. Skip those that don't in
17960 the hash function generator.
17961 * common.opt (fvar-tracking): Mark as PerFunction instead of
17963 (fvar-tracking-assignments): Likewise.
17964 (fvar-tracking-assignments-toggle): Likewise.
17965 (fvar-tracking-uninit): Likewise.
17967 2017-01-07 Jakub Jelinek <jakub@redhat.com>
17969 PR translation/79018
17970 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
17973 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
17976 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
17977 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
17978 TARGET_LONG_JUMP_TABLE_OFFSETS.
17979 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
17980 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
17981 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
17982 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
17983 * config/m68k/m68k.md (tablejump expander): Likewise.
17984 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
17985 TARGET_LONG_JUMP_TABLE_OFFSETS.
17986 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
17987 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
17989 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
17990 David Holsgrove <david.holsgrove@xilinx.com>
17992 * common/config/microblaze/microblaze-common.c
17993 (TARGET_EXCEPT_UNWIND_INFO): Remove.
17994 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
17996 * config/microblaze/microblaze.c (microblaze_must_save_register)
17997 (microblaze_expand_epilogue, microblaze_return_addr): Handle
17999 (microblaze_eh_return): New function.
18000 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18001 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18002 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18003 * config/microblaze/microblaze.md (eh_return): New pattern.
18005 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18007 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18008 GCC_DIAGNOSTIC_STRINGIFY): Define.
18010 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18012 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18014 * config/arm/arm.md (<mcrr>): New.
18016 * config/arm/arm.c (arm_arch5te): New.
18017 (arm_option_override): Set arm_arch5te.
18018 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18020 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18021 (arm_mcrr_qualifiers): ... this. New.
18022 (MRRC_QUALIFIERS): Define to...
18023 (arm_mrrc_qualifiers): ... this. New.
18024 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18026 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18027 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18028 (MRRCI, mrrc, MRRC): New.
18029 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18030 VUNSPEC_MRRC2): New.
18032 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18034 * config/arm/arm.md (<mcr>): New.
18036 * config/arm/arm.c (arm_coproc_builtin_available): Add
18037 support for mcr, mrc, mcr2 and mrc2.
18038 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18039 (arm_mcr_qualifiers): ... this. New.
18040 (MRC_QUALIFIERS): Define to ...
18041 (arm_mrc_qualifiers): ... this. New.
18042 (MCR_QUALIFIERS): Define to ...
18043 (arm_mcr_qualifiers): ... this. New.
18044 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18046 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18047 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18048 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18049 VUNSPEC_MRC2): New.
18051 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18053 * config/arm/arm.md (*ldc): New.
18057 * config/arm/arm.c (arm_coproc_builtin_available): Add
18058 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18059 (arm_coproc_ldc_stc_legitimate_address): New.
18060 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18061 'qualifier_const_pointer'.
18062 (LDC_QUALIFIERS): Define to...
18063 (arm_ldc_qualifiers): ... this. New.
18064 (STC_QUALIFIERS): Define to...
18065 (arm_stc_qualifiers): ... this. New.
18066 * config/arm/arm-protos.h
18067 (arm_coproc_ldc_stc_legitimate_address): New.
18068 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18069 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18070 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18071 stc2, stcl, stc2l): New.
18072 * config/arm/constraints.md (Uz): New.
18073 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18074 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18075 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18076 VUNSPEC_STC2L): New.
18078 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18080 * config/arm/arm.md (<cdp>): New.
18081 * config/arm/arm.c (neon_const_bounds): Rename this ...
18082 (arm_const_bounds): ... this.
18083 (arm_coproc_builtin_available): New.
18084 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18085 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18086 (CDP_QUALIFIERS): Define to...
18087 (arm_cdp_qualifiers): ... this. New.
18089 (arm_expand_builtin_args): Add case for 6 arguments.
18090 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18091 (arm_const_bounds): ... this.
18092 (arm_coproc_builtin_available): New.
18093 * config/arm/arm_acle.h (__arm_cdp): New.
18095 * config/arm/arm_acle_builtins.def (cdp): New.
18097 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18098 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18099 'arm_const_bounds'.
18100 * config/arm/types.md (coproc): New.
18101 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18102 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18103 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18104 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18106 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18108 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18109 (UBINOP_QUALIFIERS): New.
18111 (acle_builtin_data): New. Change comment.
18112 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18113 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18114 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18115 arm_acle_builtins.def.
18116 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18117 (arm_init_acle_builtins): New.
18118 (CRC32_BUILTIN): Remove.
18119 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18120 crc32cb, crc32ch and crc32cw.
18121 (arm_init_crc32_builtins): Remove.
18122 (arm_init_builtins): Use arm_init_acle_builtins rather
18123 than arm_init_crc32_builtins.
18124 (arm_expand_acle_builtin): New.
18125 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18126 * config/arm/arm_acle_builtins.def: New.
18128 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18130 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18131 (arm_builtin_datum): ... this.
18132 (arm_init_neon_builtin): Rename to ...
18133 (arm_init_builtin): ... this. Add a new parameters PREFIX
18134 and USE_SIG_IN_NAME.
18135 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18136 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18137 'arm_builtin_datum'.
18138 (arm_init_vfp_builtins): Likewise.
18139 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18140 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18141 (arm_expand_neon_args): Rename to ...
18142 (arm_expand_builtin_args): ... this. Rename builtin_arg
18143 enum values and differentiate between ARG_BUILTIN_MEMORY
18144 and ARG_BUILTIN_NEON_MEMORY.
18145 (arm_expand_neon_builtin_1): Rename to ...
18146 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18147 values, arm_expand_builtin_args and add bool parameter NEON.
18148 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18149 (arm_expand_vfp_builtin): Likewise.
18150 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18152 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18154 PR middle-end/77484
18155 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18156 * predict.c (tree_estimate_probability_bb): Reverse direction of
18157 polymorphic call predictor.
18159 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18161 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18162 (determine_pass_name_match): ...this new function and...
18163 (should_skip_pass_p): ...this new function.
18165 2017-01-06 Nathan Sidwell <nathan@acm.org>
18167 * ipa-visibility.c (function_and_variable_visibility): Reformat
18168 comments and long lines. Remove extrneous if.
18169 * symtab.c (symtab_node::make_decl_local): Fix code format.
18170 (symtab_node::set_section_for_node): Fix comment typo.
18172 2017-01-06 Martin Liska <mliska@suse.cz>
18175 * lra-constraints.c: Rename invariant to lra_invariant.
18176 * predict.c (set_even_probabilities): Initialize e to NULL.
18178 2017-01-05 Martin Sebor <msebor@redhat.com>
18180 PR tree-optimization/78910
18181 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18182 (format_integer): Correct off-by-one error in the handling
18183 of precision with negative numbers in signed conversions..
18185 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18187 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18189 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18191 PR tree-optimization/71016
18192 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18193 factor_out_conditional_conversion. Formatting fix.
18194 (factor_out_conditional_conversion): Add cond_stmt argument.
18195 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18196 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18199 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18201 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18202 read-rtl-function.o, and selftest-rtl.o.
18203 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18204 (selftest::aarch64_test_loading_full_dump): New function.
18205 (selftest::aarch64_run_selftests): New function.
18206 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18207 selftest::aarch64_run_selftests.
18208 * config/i386/i386.c
18209 (selftest::ix86_test_loading_dump_fragment_1): New function.
18210 (selftest::ix86_test_loading_call_insn): New function.
18211 (selftest::ix86_test_loading_full_dump): New function.
18212 (selftest::ix86_test_loading_unspec): New function.
18213 (selftest::ix86_run_selftests): Call the new functions.
18214 * emit-rtl.c (maybe_set_max_label_num): New function.
18215 * emit-rtl.h (maybe_set_max_label_num): New decl.
18216 * function.c (instantiate_decls): Guard call to
18217 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18218 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18220 * gensupport.c (gen_reader::gen_reader): Pass "false"
18221 for new "compact" param of rtx_reader.
18222 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18223 rather than an empty string for NULL strings.
18224 * read-md.c: Potentially include config.h rather than bconfig.h.
18225 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18226 (have_error): New global, copied from errors.c.
18227 (md_reader::read_name): Rename to...
18228 (md_reader::read_name_1): ...this, adding "out_loc" param,
18229 and converting "missing name or number" to returning false, rather
18231 (md_reader::read_name): Reimplement in terms of read_name_1.
18232 (md_reader::read_name_or_nil): New function.
18233 (md_reader::read_string): Handle "(nil)" by returning NULL.
18234 (md_reader::md_reader): Add new param "compact".
18235 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18236 (md_reader::read_file): New method.
18237 * read-md.h (md_reader::md_reader): Add new param "compact".
18238 (md_reader::read_file): New method.
18239 (md_reader::is_compact): New accessor.
18240 (md_reader::read_name): Convert return type from void to file_location.
18241 (md_reader::read_name_or_nil): New decl.
18242 (md_reader::read_name_1): New decl.
18243 (md_reader::m_compact): New field.
18244 (noop_reader::noop_reader): Pass "false" for new "compact" param
18246 (rtx_reader::rtx_reader): Add new "compact" param.
18247 (rtx_reader::read_rtx_operand): Make virtual and convert return
18248 type from void to rtx.
18249 (rtx_reader::read_until): New decl.
18250 (rtx_reader::handle_any_trailing_information): New virtual function.
18251 (rtx_reader::postprocess): New virtual function.
18252 (rtx_reader::finalize_string): New virtual function.
18253 (rtx_reader::m_in_call_function_usage): New field.
18254 (rtx_reader::m_reuse_rtx_by_id): New field.
18255 * read-rtl-function.c: New file.
18256 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18257 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18258 (selftest::verify_three_block_rtl_cfg): New decl.
18259 * read-rtl-function.h: New file.
18260 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18261 For host, include function.h, memmodel.h, and emit-rtl.h.
18262 (one_time_initialization): New function.
18263 (struct compact_insn_name): New struct.
18264 (compact_insn_names): New array.
18265 (find_code): Handle insn codes in compact dumps.
18266 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18267 (bind_subst_iter_and_attr): Likewise.
18268 (add_condition_to_string): Likewise.
18269 (add_condition_to_rtx): Likewise.
18270 (apply_attribute_uses): Likewise.
18271 (add_current_iterators): Likewise.
18272 (apply_iterators): Likewise.
18273 (initialize_iterators): Guard usage of apply_subst_iterator with
18274 #ifdef GENERATOR_FILE.
18275 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18276 (md_reader::read_mapping): Likewise.
18277 (add_define_attr_for_define_subst): Likewise.
18278 (add_define_subst_attr): Likewise.
18279 (read_subst_mapping): Likewise.
18280 (check_code_iterator): Likewise.
18281 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18283 (one_time_initialization): New function.
18284 (rtx_reader::read_until): New method.
18285 (read_flags): New function.
18286 (parse_reg_note_name): New function.
18287 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18288 Handle reuse_rtx ids.
18289 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18290 Add parsing support for RTL dumps, mirroring the special-cases in
18291 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18292 values, and calling handle_any_trailing_information.
18293 (rtx_reader::read_rtx_operand): Convert return type from void
18294 to rtx, returning return_rtx. Handle case 'e'. Call
18295 finalize_string on XSTR and XTMPL fields.
18296 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18297 "(nil)" values were omitted. Call the postprocess vfunc on the
18299 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18300 class ctor. Initialize m_in_call_function_usage. Call
18301 one_time_initialization.
18302 * rtl-tests.c (selftest::test_uncond_jump): Call
18303 set_new_first_and_last_insn.
18304 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18305 * selftest-rtl.c: New file.
18306 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18307 (selftest::get_insn_by_uid): New decl.
18308 * selftest-run-tests.c (selftest::run_tests): Call
18309 read_rtl_function_c_tests.
18310 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18311 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18314 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18316 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18317 operands in a special way. Assert that pos+len <= mode precision.
18319 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18321 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18322 3 argument Alias with unlimited for the negative form.
18323 (fno-vect-cost-model): Removed.
18325 2017-01-05 Martin Liska <mliska@suse.cz>
18327 * hsa-gen.c (gen_hsa_divmod): New function.
18328 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18330 2017-01-05 Martin Liska <mliska@suse.cz>
18333 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18336 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18338 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18339 small constant length operands.
18341 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18343 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18344 between loop iterations.
18346 2017-01-05 Martin Liska <mliska@suse.cz>
18349 * gimplify.c (gimplify_decl_expr): Compare to
18350 asan_poisoned_variables instread of checking flags.
18351 (gimplify_target_expr): Likewise.
18352 (gimplify_expr): Likewise.
18353 (gimplify_function_tree): Conditionally initialize
18354 asan_poisoned_variables.
18356 2017-01-04 Jeff Law <law@redhat.com>
18358 PR tree-optimizatin/78812
18359 * rtl.h (contains_mem_rtx_p): Prototype.
18360 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18361 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18362 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18363 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18364 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18366 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18368 * input.c (assert_char_at_range): Default-initialize actual_range.
18370 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18372 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18373 to match the caller.
18375 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18377 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18378 insns after final jump in test to emit dummy move.
18380 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18382 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18383 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18385 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18387 * multiple_target.c (create_dispatcher_calls): Init e_next.
18388 * tree-ssa-loop-split.c (split_loop): Init border.
18389 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18392 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18397 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18398 (altivec_register_operand): Do not return true if the operand
18399 contains a SUBREG mixing SImode and SFmode.
18400 (vsx_register_operand): Likewise.
18401 (vsx_reg_sfsubreg_ok): New predicate.
18402 (vfloat_operand): Do not return true if the operand contains a
18403 SUBREG mixing SImode and SFmode.
18404 (vint_operand): Likewise.
18405 (vlogical_operand): Likewise.
18406 (gpc_reg_operand): Likewise.
18407 (int_reg_operand): Likewise.
18408 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18409 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18410 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
18412 (rs6000_emit_move_si_sf_subreg): New helper function.
18413 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
18414 fixup SUBREGs involving SImode and SFmode.
18415 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
18416 numbers for the new peephole2 optimization.
18417 (peephole2 for SFmode unions): New peephole2 to optimize cases in
18418 the GLIBC math library that do AND/IOR/XOR operations on single
18419 precision floating point.
18420 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
18421 target macros to say whether we need to avoid SUBREGs mixing
18423 (TARGET_ALLOW_SF_SUBREG): Likewise.
18424 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
18425 (UNSPEC_SI_FROM_SF): Likewise.
18426 (iorxor): Change spacing.
18427 (and_ior_xor): New iterator for AND, IOR, and XOR.
18428 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
18429 (movdi_from_sf_zero_ext): Likewise.
18430 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
18431 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
18432 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
18433 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
18434 (fms<mode>4): Likewise.
18435 (fnma<mode>4): Likewise.
18436 (fnms<mode>4): Likewise.
18437 (nfma<mode>4): Likewise.
18438 (nfms<mode>4): Likewise.
18440 2017-01-04 Marek Polacek <polacek@redhat.com>
18443 * doc/invoke.texi: Document -Wpointer-compare.
18445 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18447 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
18450 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
18451 descriptions for -gdwarf-5 and emit them as uleb128 instead of
18454 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
18457 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
18458 documentation of the powerpc_popcntb_ok attribute.
18459 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18460 code to issue warning messages if a requested CPU configuration is
18461 not supported by the binary (assembler and loader) toolchain.
18462 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
18463 made to define a built-in function that has been disabled.
18464 (paired_init_builtins): Add assertion to prevent ICE if attempt is
18465 made to define a built-in function that has been disabled.
18466 (altivec_init_builtins): Add comment explaining why definition
18467 of the DST built-in functions is not preceded by an assertion
18468 check. Add assertions to prevent ICE if attempts are made to
18469 define an altivec predicate or an abs* built-in function that has
18471 (htm_init_builtins): Add comment explaining why definition of the
18472 htm built-in functions is not preceded by an assertion check.
18474 2017-01-04 Jeff Law <law@redhat.com>
18476 PR tree-optimizatin/67955
18477 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
18478 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
18479 the points-to solution does not include pt_null. Use DECL_PT_UID
18482 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
18484 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
18485 Use gen_int_mode instead of gen_lopwart for const_int operands.
18487 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18489 PR tree-optimization/71563
18490 * match.pd: Simplify X << Y into X if Y is known to be 0 or
18491 out of range value - has low bits known to be zero.
18493 2017-01-04 Alan Modra <amodra@gmail.com>
18495 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
18496 * configure: Regenerate.
18497 * config.in: Regenerate.
18499 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18502 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
18503 a substring of the message, but strcmp with the whole message. Ifdef
18504 ENABLE_NLS, translate the message first using dgettext.
18506 2017-01-03 Jeff Law <law@redhat.com>
18508 PR tree-optimizatin/78856
18509 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
18510 (mark_threaded_blocks): Remove code to truncate thread paths that
18511 cross multiple loop headers. Instead invalidate the cached loop
18512 iteration information and handle case of a thread path walking
18513 into an irreducible region.
18515 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
18518 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
18519 assertions. Add support for doing the signbit if the IEEE 128-bit
18520 floating point value is in a GPR.
18521 * config/rs6000/rs6000.md (Fsignbit): Delete.
18522 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
18523 Update the length attribute if the value is in a GPR.
18524 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
18525 the sign or zero extension instruction, since the value is always 0/1.
18526 (signbit<mode>2_dm2): Delete using <Fsignbit>.
18529 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
18530 extracting SImode to a GPR register so that we can generate a
18531 store, limit the vector to be in a traditional Altivec register
18532 for the vextuwrx instruction.
18534 2017-01-03 Ian Lance Taylor <iant@google.com>
18536 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
18538 2017-01-03 Martin Sebor <msebor@redhat.com>
18540 PR tree-optimization/78696
18541 * gimple-ssa-sprintf.c (format_floating): Correct handling of
18542 precision. Use MPFR for %f for greater fidelity. Correct handling
18544 (pass_sprintf_length::compute_format_length): Set width and precision
18545 specified by asrerisk to void_node for vararg functions.
18546 (try_substitute_return_value): Adjust dump output.
18548 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
18550 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
18552 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
18554 * doc/invoke.texi (SPARC options): Document -mlra as the default.
18555 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
18556 -mlra/-mno-lra was passed to the compiler.
18558 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
18560 PR rtl-optimization/65618
18561 * emit-rtl.c (try_split): Move initialization of "before" and
18562 "after" to just before the call to emit_insn_after_setloc.
18564 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
18566 * doc/md.texi (Standard Names): Remove reference to Java frontend.
18568 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
18570 * dwarf2out.c (gen_enumeration_type_die): When
18571 -gno-strict-dwarf, add a DW_AT_encoding attribute.
18573 2017-01-03 Jakub Jelinek <jakub@redhat.com>
18575 PR tree-optimization/78965
18576 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
18577 Change first argument from const call_info & to call_info &. For %n
18578 set info.nowrite to false.
18580 PR middle-end/78901
18581 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
18582 possibly throwing calls.
18584 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
18585 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
18586 and fns handling, rather than in a separate case SSA_NAME.
18588 2017-01-02 Jeff Law <law@redhat.com>
18590 * config/darwin-driver.c (darwin_driver_init): Const-correctness
18591 fixes for first_period and second_period variables.
18593 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
18596 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
18597 (*insvqi_1): New insn pattern.
18598 (*insvqi_1_mem_rex64): Ditto.
18599 (*insvqi_2): Ditto.
18600 (*insvqi_3): Rename from *insvqi.
18602 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
18604 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
18606 * doc/cfg.texi (Edges): Remove reference to Java.
18607 (Maintaining the CFG): Ditto.
18609 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18611 PR middle-end/77674
18612 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
18613 transparent aliases.
18615 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18617 PR middle-end/77484
18618 * predict.def (PRED_CALL): Update hitrate.
18619 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
18620 * predict.c (tree_estimate_probability_bb): Split CALL predictor
18621 into direct/indirect/polymorphic variants.
18623 2017-01-01 Jakub Jelinek <jakub@redhat.com>
18625 Update copyright years.
18627 * gcc.c (process_command): Update copyright notice dates.
18628 * gcov-dump.c (print_version): Ditto.
18629 * gcov.c (print_version): Ditto.
18630 * gcov-tool.c (print_version): Ditto.
18631 * gengtype.c (create_file): Ditto.
18632 * doc/cpp.texi: Bump @copying's copyright year.
18633 * doc/cppinternals.texi: Ditto.
18634 * doc/gcc.texi: Ditto.
18635 * doc/gccint.texi: Ditto.
18636 * doc/gcov.texi: Ditto.
18637 * doc/install.texi: Ditto.
18638 * doc/invoke.texi: Ditto.
18640 Copyright (C) 2017 Free Software Foundation, Inc.
18642 Copying and distribution of this file, with or without modification,
18643 are permitted in any medium without royalty provided the copyright
18644 notice and this notice are preserved.