1 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5 (fp_valid_at): Likewise.
7 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10 (qdf24xx_addrcost_table): Likewise.
11 (cortexa57_tunings): Update to use generic_branch_cost.
12 (cortexa72_tunings): Likewise.
13 (cortexa73_tunings): Likewise.
14 (qdf24xx_tunings): Likewise.
16 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19 (thunderx2t99_branch_cost): Likewise.
20 (cortexa35_tunings): Update to use generic_branch_cost.
21 (cortexa53_tunings): Likewise.
22 (cortexa57_tunings): Likewise.
23 (cortexa72_tunings): Likewise.
24 (cortexa73_tunings): Likewise.
25 (thunderx2t99_tunings): Likewise.
27 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
29 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
30 (sparc_option_override): Honour MASK_FSMULD.
31 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
32 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
33 * config/sparc/sparc.opt (mfsmuld): New option.
34 * doc/invoke.texi (mfsmuld): Document option.
36 2017-07-26 Marek Polacek <polacek@redhat.com>
39 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
41 2017-07-26 Richard Biener <rguenther@suse.de>
43 * gimple-match-head.c (do_valueize): Return OP if valueize
45 (get_def): New helper to get at the def stmt of a SSA name
47 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
48 do_valueize to get at the def stmt.
49 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
51 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
54 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
56 2017-07-26 Martin Liska <mliska@suse.cz>
59 * function.c (expand_function_start): Make expansion of
60 nonlocal_goto_save_area after parm_birth_insn.
62 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
64 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
65 from all CPU target flags enable members.
67 2017-07-26 Richard Biener <rguenther@suse.de>
69 * genmatch.c (dt_simplify::gen): Make iterator vars const.
70 (decision_tree::gen): Make 'type' const.
71 (write_predicate): Likewise.
73 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
75 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
76 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
77 (rs6000_option_override_internal): Likewise.
78 (rs6000_expand_vector_set): Likewise.
79 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
80 (TARGET_UPPER_REGS_SF): Likewise.
81 (TARGET_UPPER_REGS_DI): Likewise.
82 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
83 (TARGET_DIRECT_MOVE_64BIT): Likewise.
84 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
85 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
86 (Splitters for DI constants in Altivec registers): Likewise.
87 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
88 (vsx_set_v4sf_p9): Likewise.
89 (vsx_set_v4sf_p9_zero): Likewise.
90 (vsx_insert_extract_v4sf_p9): Likewise.
91 (vsx_insert_extract_v4sf_p9_2): Likewise.
93 2017-07-25 Carl Love <cel@us.ibm.com>
95 * doc/extend.texi: Update the built-in documentation file for the
96 existing built-in functions
97 vector signed char vec_cnttz (vector signed char);
98 vector unsigned char vec_cnttz (vector unsigned char);
99 vector signed short vec_cnttz (vector signed short);
100 vector unsigned short vec_cnttz (vector unsigned short);
101 vector signed int vec_cnttz (vector signed int);
102 vector unsigned int vec_cnttz (vector unsigned int);
103 vector signed long long vec_cnttz (vector signed long long);
104 vector unsigned long long vec_cnttz (vector unsigned long long);
106 2017-07-25 Andrew Pinski <apinski@cavium.com>
108 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
109 accesses where the use is for the first operand of a BIT_INSERT.
111 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
114 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
115 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
117 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
119 * config/i386/gstabs.h: Delete.
120 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
122 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
124 * config/i386/i386.c (ix86_decompose_address): Do not check for
125 register RTX when looking at index_reg or base_reg.
126 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
128 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
130 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
131 to update EH info here.
133 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
135 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
137 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
139 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
141 2017-07-25 Torsten Duwe <duwe@suse.de>
143 * common.opt: Introduce -fpatchable-function-entry
144 command line option, and its variables function_entry_patch_area_size
145 and function_entry_patch_area_start.
146 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
147 including a two-value parser.
148 * target.def (print_patchable_function_entry): New target hook.
149 * targhooks.h (default_print_patchable_function_entry): New function.
150 * targhooks.c (default_print_patchable_function_entry): Likewise.
151 * toplev.c (process_options): Switch off IPA-RA if
152 patchable function entries are being generated.
153 * varasm.c (assemble_start_function): Look at the
154 patchable-function-entry command line switch and current
155 function attributes and maybe generate NOP instructions by
156 calling the print_patchable_function_entry hook.
157 * doc/extend.texi: Document patchable_function_entry attribute.
158 * doc/invoke.texi: Document -fpatchable_function_entry
160 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
162 * doc/tm.texi: Re-generate.
164 2017-07-25 Jakub Jelinek <jakub@redhat.com>
167 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
168 TARGET_AVX512DQ rather than TARGET_AVX512BW.
170 2017-07-25 Tamar Christina <tamar.christina@arm.com>
172 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
174 2017-07-25 Richard Biener <rguenther@suse.de>
176 PR tree-optimization/81455
177 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
178 not walk in cycles when looking for guards.
180 2017-07-25 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/81529
183 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
184 when optimizing backedge uses.
186 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
188 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
190 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
191 to dl_section_ref. On AIX, append an expression to subtract
192 the size of the section length to dl_section_ref.
194 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
196 * configure.ac: If any of the config.* scripts fail, exit 1.
197 * configure: Regenerate.
199 2017-07-25 Richard Biener <rguenther@suse.de>
202 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
203 of immediate uses, be more verbose on errors.
205 2017-07-25 Richard Biener <rguenther@suse.de>
207 PR tree-optimization/81510
208 * tree-vect-loop.c (vect_is_simple_reduction): When the
209 reduction stmt is not inside the loop bail out.
211 2017-07-25 Richard Biener <rguenther@suse.de>
213 PR tree-optimization/81303
214 * tree-vect-loop-manip.c (vect_loop_versioning): Build
215 profitability check against LOOP_VINFO_NITERSM1.
217 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
219 * domwalk.c (cmp_bb_postorder): Simplify.
220 (sort_bbs_postorder): New function. Use it...
221 (dom_walker::walk): ...here to optimize common cases.
223 2017-07-25 Martin Liska <mliska@suse.cz>
226 * ipa-visibility.c (function_and_variable_visibility): Make the
227 redirection just on target that supports aliasing.
228 Fix GNU coding style.
230 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
233 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
235 * config/arm/rtems.h: Likewise.
236 * config/bfin/rtems.h: Likewise.
237 * config/i386/rtemself.h: Likewise.
238 * config/lm32/rtems.h: Likewise.
239 * config/m32c/rtems.h: Likewise.
240 * config/m68k/rtemself.h: Likewise.
241 * config/microblaze/rtems.h: Likewise.
242 * config/mips/rtems.h: Likewise.
243 * config/moxie/rtems.h: Likewise.
244 * config/nios2/rtems.h: Likewise.
245 * config/powerpcspe/rtems.h: Likewise.
246 * config/rs6000/rtems.h: Likewise.
247 * config/rtems.h: Likewise.
248 * config/sh/rtems.h: Likewise.
249 * config/sh/rtemself.h: Likewise.
250 * config/sparc/rtemself.h: Likewise.
252 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
255 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
256 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
257 * tree-ssa-structalias.c (alias_get_name): Same.
259 2017-07-25 Bin Cheng <bin.cheng@arm.com>
262 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
263 instructions if no du chain is found.
265 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
267 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
269 2017-07-25 Richard Biener <rguenther@suse.de>
272 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
275 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
277 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
279 (ISA_2_7_MASKS_SERVER): Likewise.
280 (ISA_3_0_MASKS_IEEE): Likewise.
281 (OTHER_P8_VECTOR_MASKS): Likewise.
282 (OTHER_VSX_VECTOR_MASKS): Likewise.
283 (POWERPC_MASKS): Likewise.
284 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
285 duplicate list of options.
286 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
287 explicit -mupper-regs options.
288 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
289 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
290 alias for -mupper-regs-df.
291 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
292 (rs6000_init_hard_regno_mode_ok): Likewise.
293 (rs6000_option_override_internal): Likewise.
294 (rs6000_opt_masks): Likewise.
295 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
296 options in terms of whether -mvsx or -mpower8-vector was used.
297 (TARGET_UPPER_REGS_DI): Likewise.
298 (TARGET_UPPER_REGS_SF): Likewise.
299 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
300 -mupper-regs-* options.
302 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
304 * passes.c (emergency_dump_function): Print some empty lines and a
305 header before the RTL dump.
307 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
309 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
311 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
314 * config/aarch64/aarch64.c (aarch64_classify_symbol):
315 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
317 2017-07-24 Carl Love <cel@us.ibm.com>
319 * config/rs6000/rs6000-c.c: Add support for built-in functions
320 vector float vec_extract_fp32_from_shorth (vector unsigned short);
321 vector float vec_extract_fp32_from_shortl (vector unsigned short);
322 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
323 vec_extract_fp_from_shortl): Add defines for the two builtins.
324 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
325 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
327 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
328 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
329 * doc/extend.texi: Update the built-in documentation file for the
330 new built-in function.
332 2017-07-24 Jakub Jelinek <jakub@redhat.com>
335 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
337 * doc/generic.texi: Likewise.
338 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
339 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
341 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
343 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
344 (aarch64_mls_elt_merge<mode>): Likewise.
346 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
348 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
351 2017-07-23 Michael Collison <michael.collison@arm.com>
353 * config/arm/arm.c (arm_option_override): Deprecate
354 use of -mstructure-size-boundary.
355 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
356 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
358 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
361 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
362 Reduce cost estimate for direct moves.
364 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
367 * config/i386/i386.c (ix86_option_override_internal): Disable
368 BMI, BMI2 and TBM instructions for -m16.
370 2017-07-21 Carl Love <cel@us.ibm.com>
372 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
373 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
374 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
375 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
376 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
377 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
378 VMULOSW): New enum "unspec" values.
379 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
380 altivec_vmulosw): New patterns.
381 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
382 VMULOSW): Add definitions.
384 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
386 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
388 * config/aarch64/aarch64-options-extensions.def (rdma); New.
389 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
390 (AARCH64_FL_V8_1): Renumber.
391 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
392 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
393 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
394 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
395 rdma to feature modifiers list.
397 2017-07-21 Yury Gribov <tetra2005@gmail.com>
400 * ipa-visibility (function_and_variable_visibility): Convert
401 recursive PLT call to direct call if appropriate.
403 2017-07-21 Andrew Pinski <apinski@cavium.com>
405 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
406 operand 1 to see if the types precision matches.
407 * fold-const.c (operand_equal_p): Likewise.
409 2017-07-21 Richard Biener <rguenther@suse.de>
411 PR tree-optimization/81303
412 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
413 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
414 (vect_peeling_hash_get_lowest_cost): Adjust.
415 (vect_enhance_data_refs_alignment): Likewise. Use
416 vect_get_peeling_costs_all_drs to compute the penalty for no
417 peeling to match up costs.
419 2017-07-21 Richard Biener <rguenther@suse.de>
421 PR tree-optimization/81500
422 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
423 we didn't identify a reduction path.
425 2017-07-21 Tom de Vries <tom@codesourcery.com>
426 Cesar Philippidis <cesar@codesourcery.com>
428 PR gcov-profile/81442
429 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
432 2017-07-21 Tom de Vries <tom@codesourcery.com>
435 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
437 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
438 nvptx_override_options_after_change.
440 2017-07-21 Ulrich Drepper <drepper@redhat.com>
442 * dwarf2out.c (output_file_names): Avoid double testing for
445 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
447 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
449 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
451 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
453 (try_crossjump_to_edge): Do not punt on partitioned functions.
455 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
457 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
458 Put all BBs reachable only via paths crossing cold region to cold
460 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
462 2016-07-21 Richard Biener <rguenther@suse.de>
464 PR tree-optimization/81303
465 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
466 into account prologue and epilogue iterations when raising
467 min_profitable_iters to sth at least covering one vector iteration.
469 2017-07-21 Tamar Christina <tamar.christina@arm.com>
471 * config/arm/arm.c (arm_test_cpu_arch_dat):
474 2017-07-20 Nathan Sidwell <nathan@acm.org>
477 * tree.h (TYPE_METHODS): Delete.
478 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
479 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
480 (dbxout_type_methods): Scan TYPE_FIELDS.
481 (dbxout_type): Don't check TYPE_METHODS here.
482 * function.c (use_register_for_decl): Always ignore register for
483 class types when not optimizing.
484 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
485 * tree.c (free_lang_data_in_type): Stitch out member functions and
486 templates from TYPE_FIELDS.
487 (build_distinct_type_copy, verify_type_variant,
488 verify_type): Member fns are on TYPE_FIELDS.
489 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
490 * tree-pretty-print.c (dump_generic_node): Likewise.
492 2017-07-20 Jakub Jelinek <jakub@redhat.com>
495 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
496 V2TImode and V4TImode.
497 (ix86_expand_vector_extract): Likewise.
498 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
499 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
500 (ssescalarmode): Handle V4TImode and V2TImode.
501 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
502 (*vec_extractv2ti, *vec_extractv4ti): New insns.
503 (VEXTRACTI128_MODE): New mode iterator.
504 (splitter for *vec_extractv?ti first element): New.
505 (VEC_INIT_MODE): New mode iterator.
506 (vec_init<mode>): Consolidate 3 expanders into one using
507 VEC_INIT_MODE mode iterator.
509 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
511 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
512 non_spilled_static_chain_regno_p.
514 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
516 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
518 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
520 * bb-reorder.c (connect_traces): Allow copying of blocks within
523 2017-07-20 Richard Biener <rguenther@suse.de>
525 * gimple.h (gimple_phi_result): Add gphi * overload.
526 (gimple_phi_result_ptr): Likewise.
527 (gimple_phi_arg): Likewise. Adjust index assert to only
528 allow actual argument accesses rather than all slots available
530 (gimple_phi_arg_def): Add gphi * overload.
531 * tree-phinodes.c (make_phi_node): Initialize only actual
533 (resize_phi_node): Clear memory not covered by old node,
534 do not initialize excess argument slots.
535 (reserve_phi_args_for_new_edge): Initialize new argument slot
538 2017-07-20 Bin Cheng <bin.cheng@arm.com>
540 PR tree-optimization/81388
542 2016-07-21 Bin Cheng <bin.cheng@arm.com>
544 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
545 by removing computation of may_be_zero.
547 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
548 Tom de Vries <tom@codesourcery.com>
551 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
552 when gimple level profile disagrees with what RTL expander did.
554 2017-07-20 Richard Biener <rguenther@suse.de>
556 PR tree-optimization/61171
557 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
558 (vect_analyze_stmt): Add slp instance parameter.
559 (vectorizable_reduction): Likewise.
560 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
561 (vect_is_simple_reduction): Deal with chains not detected
562 as SLP reduction chain, specifically not properly associated
563 chains containing a mix of plus/minus.
564 (get_reduction_op): Remove.
565 (get_initial_defs_for_reduction): Simplify, pass in whether
566 this is a reduction chain, pass in the SLP node for the PHIs.
567 (vect_create_epilog_for_reduction): Get the SLP instance as
569 (vectorizable_reduction): Get the SLP instance as arg.
570 During analysis remember the SLP node with the PHIs in the
571 instance. Simplify getting at the vectorized reduction PHIs.
572 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
573 through SLP instance.
574 (vect_slp_analyze_operations): Likewise.
575 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
576 (vect_transform_stmt): Likewise.
578 2017-07-20 Tom de Vries <tom@codesourcery.com>
580 PR tree-optimization/81489
581 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
582 read of phi arg location to before loop that modifies phi.
584 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
586 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
589 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
592 * except.c (execute): Fix ordering issue.
594 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
596 PR rtl-optimization/81423
597 * combine.c (make_compound_operation_int): Don't try to optimize
598 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
600 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
602 PR rtl-optimization/81423
603 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
604 with a constant that is -1 in the truncated to mode.
606 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
608 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
609 (determine_unlikely_bbs): ... here.
610 * predict.h (propagate_unlikely_bbs_forward): Declare.
611 * cfgexpand.c (pass_expand::execute): Use it.
612 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
614 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
615 propagate_unlikely_bbs_forward.
617 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
620 * except.c (maybe_add_nop_after_section_switch): New function.
623 2017-07-19 Tom de Vries <tom@codesourcery.com>
625 * gimple.h (gimple_phi_set_arg): Make assert more strict.
627 2017-07-19 Tom de Vries <tom@codesourcery.com>
629 * gimple.h (gimple_phi_arg): Make assert more strict.
631 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
633 * config.gcc (powerpc*-*-*): Add mmintrin.h.
634 * config/rs6000/mmintrin.h: New file.
635 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
637 2017-07-19 Jakub Jelinek <jakub@redhat.com>
639 PR tree-optimization/81346
640 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
642 2017-07-19 Tom de Vries <tom@codesourcery.com>
644 * config/nvptx/nvptx.md (VECIM): Add V2DI.
646 2017-07-19 Tom de Vries <tom@codesourcery.com>
648 * config/nvptx/nvptx-modes.def: Add V2DImode.
649 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
650 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
651 (nvptx_output_mov_insn): Handle lack of mov.b128.
652 (nvptx_print_operand): Handle 'H' and 'L' codes.
653 (nvptx_vector_mode_supported): Allow V2DImode.
654 (nvptx_preferred_simd_mode): New function.
655 (nvptx_data_alignment): New function.
656 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
657 nvptx_preferred_simd_mode.
658 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
660 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
662 2017-07-19 Tom de Vries <tom@codesourcery.com>
664 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
665 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
666 (nvptx_vector_mode_supported): New function. Allow V2SImode.
667 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
668 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
669 (mov<VECIM>_insn): New define_insn.
670 (define_expand "mov<VECIM>): New define_expand.
672 2017-07-19 Tom de Vries <tom@codesourcery.com>
674 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
676 2017-07-19 Jakub Jelinek <jakub@redhat.com>
678 PR tree-optimization/81346
679 * fold-const.h (fold_div_compare, range_check_type): Declare.
680 * fold-const.c (range_check_type): New function.
681 (build_range_check): Use range_check_type.
682 (fold_div_compare): No longer static, rewritten into
683 a match.pd helper function.
684 (fold_comparison): Don't call fold_div_compare here.
685 * match.pd (X / C1 op C2): New optimization using fold_div_compare
688 2017-07-19 Nathan Sidwell <nathan@acm.org>
690 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
691 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
692 * tree.c (find_decls_types_r, verify_type): Use
693 TYPE_{MIN,MAX}_VALUE_RAW.
694 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
695 (hash_tree): Likewise.
696 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
698 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
701 2017-07-18 Tom de Vries <tom@codesourcery.com>
704 * omp-expand.c (expand_omp_for_static_chunk): Handle
705 equal-argument loop exit phi.
707 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
710 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
711 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
713 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
715 (ror,rol -> rorx splitters): Use const_int_operand as
718 2017-06-18 Richard Biener <rguenther@suse.de>
720 PR tree-optimization/81410
721 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
722 the gap in the ! slp_perm SLP case after each group.
724 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
727 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
730 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
733 * predict.c (set_even_probabilities): Cleanup; do not affect
734 probabilities that are already known.
735 (combine_predictions_for_bb): Call even when count is set.
737 2017-07-18 Nathan Sidwell <nathan@acm.org>
739 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
742 2017-07-18 Bin Cheng <bin.cheng@arm.com>
745 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
746 optimization for loop niter analysis.
748 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
751 * config/avr/avr.c (avr_optimize_casesi): Don't use
752 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
754 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
756 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
757 body_cost_vec from _vect_peel_extended_info.
758 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
759 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
762 2017-07-18 Bin Cheng <bin.cheng@arm.com>
764 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
766 2017-07-18 Richard Biener <rguenther@suse.de>
768 PR tree-optimization/80620
769 PR tree-optimization/81403
770 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
771 info when re-using a VN table entry.
773 2017-07-18 Richard Biener <rguenther@suse.de>
775 PR tree-optimization/81418
776 * tree-vect-loop.c (vectorizable_reduction): Properly compute
777 vectype_in. Verify that with lane-reducing reduction operations
778 we have a single def-use cycle.
780 2017-07-17 Carl Love <cel@us.ibm.com>
782 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
784 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
785 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
786 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
787 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
788 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
789 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
790 VMULOSW): New enum "unspec" values.
791 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
792 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
793 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
794 altivec_vmulosw): New patterns.
795 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
796 VMULOSW): Add definitions.
797 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
799 * config/alpha/alpha.c: Include predict.h.
801 2017-07-17 Yury Gribov <tetra2005@gmail.com>
803 * tree-vrp.c (compare_assert_loc): Fix comparison function
804 to return predictable results.
806 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
808 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
811 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
812 * doc/invoke.texi (mexpand-adddi): Update text.
814 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
816 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
817 that also clobbers the CC register. The old expand code is moved
819 (*arc_clzsi2): ... here.
820 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
821 the CC register. The old expand code is moved to ...
822 (arc_ctzsi2): ... here.
824 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
826 * config/arc/arc.opt (mindexed-loads): Use initial value
827 TARGET_INDEXED_LOADS_DEFAULT.
828 (mauto-modify-reg): Use initial value
829 TARGET_AUTO_MODIFY_REG_DEFAULT.
830 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
831 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
832 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
833 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
835 2017-07-17 Martin Liska <mliska@suse.cz>
838 * opts.c (finish_options): Do not allow -fgnu-tm
839 w/ -fsanitize={kernel-,}address. Say sorry.
841 2017-07-17 Bin Cheng <bin.cheng@arm.com>
844 * tree-loop-distribution.c (classify_partition): Only assert on
846 (merge_dep_scc_partitions): Delete prameter. Update function call.
847 (distribute_loop): Remove code handling loop with unknown niters.
848 (pass_loop_distribution::execute): Skip loop with unknown niters.
850 2017-07-17 Bin Cheng <bin.cheng@arm.com>
853 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
854 function sort_partitions_by_post_order.
856 2017-07-17 Bin Cheng <bin.cheng@arm.com>
858 PR tree-optimization/81374
859 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
860 the max index of basic blocks, rather than number of basic blocks.
862 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
864 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
866 (arc_legitimate_pic_operand_p): Likewise.
867 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
869 (arc_needs_pcl_p): Likewise.
870 (arc_legitimate_pc_offset_p): Likewise.
871 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
872 function is also used in constrains.md.
873 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
874 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
875 PLUS. Only return true/false in known cases, otherwise assert.
876 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
877 is already called in arc_legitimate_constant_p.
878 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
880 (LEGITIMATE_PIC_OPERAND_P): Use
881 arc_raw_symbolic_reference_mentioned_p function.
882 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
887 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
888 Andrew Burgess <andrew.burgess@embecosm.com>
890 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
891 (arc_return_address_register): New function.
892 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
893 (arc_handle_fndecl_attribute): Add naked attribute.
894 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
895 (TARGET_WARN_FUNC_RETURN): Likewise.
896 (arc_allocate_stack_slots_for_args): New function.
897 (arc_warn_func_return): Likewise.
898 (machine_function): Change type fn_type.
899 (arc_compute_function_type): Consider new naked function type,
900 change function return type.
901 (arc_must_save_register): Adapt to handle new
902 arc_compute_function_type's return type.
903 (arc_expand_prologue): Likewise.
904 (arc_expand_epilogue): Likewise.
905 (arc_return_address_regs): Delete.
906 (arc_return_address_register): New function.
907 (arc_epilogue_uses): Use above function.
908 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
909 (arc_function_type): Change encoding, add naked type.
910 (ARC_INTERRUPT_P): Change to handle the new encoding.
911 (ARC_FAST_INTERRUPT_P): Likewise.
912 (ARC_NORMAL_P): Define.
913 (ARC_NAKED_P): Likewise.
914 (arc_compute_function_type): Delete prototype.
915 * config/arc/arc.md (in_ret_delay_slot): Use
916 arc_return_address_register function.
917 (simple_return): Likewise.
918 (p_return_i): Likewise.
920 2017-07-17 Jakub Jelinek <jakub@redhat.com>
922 PR tree-optimization/81428
923 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
924 can't be built for those types.
926 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
928 Remove stuff dead since r239246.
930 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
931 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
932 (avr_inform_devices): Remove dead stuff.
934 2017-07-17 Tamar Christina <tamar.christina@arm.com>
936 * config/arm/arm_neon.h: Fix softp typo.
938 2017-07-17 Jakub Jelinek <jakub@redhat.com>
940 PR tree-optimization/81365
941 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
942 aggregate moves onto bb predecessor edges, make sure there are no
943 loads that could alias the lhs in between the start of bb and the
946 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
949 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
950 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
951 [LSHIFTRT, outer_code = TRUNCATE]: Same.
953 2017-07-17 Jakub Jelinek <jakub@redhat.com>
955 PR tree-optimization/81396
956 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
957 (init_symbolic_number): Initialize it to 1.
958 (perform_symbolic_merge): Add n_ops from both operands into the new
960 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
961 without base_addr as useless if they need more than one operation.
962 (bswap_replace): Handle !bswap case for NULL base_addr.
964 2017-07-17 Tom de Vries <tom@codesourcery.com>
967 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
970 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
972 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
973 conditional builtin define __FIX_LEON3FT_B2BST.
975 2017-07-17 Daniel Cederman <cederman@gaisler.com>
977 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
978 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
981 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
983 PR rtl-optimization/81424
984 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
985 to remove potential trapping from operands if -fnon-call-exceptions.
987 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
989 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
990 profile_proability for scalling.
991 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
993 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
995 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
997 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
999 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1000 fixpoint arithmetics.
1002 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1004 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1005 fixpoint arithmetics.
1007 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1009 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1010 fixpoint arithmetics.
1012 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1014 * profile-count.h (profile_probability::from_reg_br_prob_note,
1015 profile_probability::to_reg_br_prob_note): New functions.
1016 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1017 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1018 * predict.c (probability_reliable_p): Update.
1019 (edge_probability_reliable_p): Update.
1020 (br_prob_note_reliable_p): Update.
1021 (invert_br_probabilities): Update.
1022 (add_reg_br_prob_note): New function.
1023 (combine_predictions_for_insn): Update.
1024 * asan.c (asan_clear_shadow): Update.
1025 * cfgbuild.c (compute_outgoing_frequencies): Update.
1026 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1027 (update_br_prob_note): Update.
1028 (rtl_verify_edges): Update.
1029 (purge_dead_edges): Update.
1030 (fixup_reorder_chain): Update.
1031 * emit-rtl.c (try_split): Update.
1032 * ifcvt.c (cond_exec_process_insns): Update.
1033 (cond_exec_process_if_block): Update.
1034 (dead_or_predicable): Update.
1035 * internal-fn.c (expand_addsub_overflow): Update.
1036 (expand_neg_overflow): Update.
1037 (expand_mul_overflow): Update.
1038 * loop-doloop.c (doloop_modify): Update.
1039 * loop-unroll.c (compare_and_jump_seq): Update.
1040 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1041 * predict.h: Update.
1042 * reorg.c (mostly_true_jump): Update.
1044 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1045 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1046 * config/arc/arc.c: (emit_unlikely_jump): Update.
1047 * config/arm/arm.c: (emit_unlikely_jump): Update.
1048 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1049 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1050 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1051 (ix86_print_operand): Update.
1052 (ix86_split_fp_branch): Update.
1053 (predict_jump): Update.
1054 * config/ia64/ia64.c (ia64_print_operand): Update.
1055 * config/mmix/mmix.c (mmix_print_operand): Update.
1056 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1057 (rs6000_expand_split_stack_prologue): Update.
1058 * config/rs6000/rs6000.c: Update.
1059 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1060 (s390_expand_vec_movstr): Update.
1061 (s390_expand_cs_tdsi): Update.
1062 (s390_expand_split_stack_prologue): Update.
1063 * config/sh/sh.c (sh_print_operand): Update.
1064 (expand_cbranchsi4): Update.
1065 (expand_cbranchdi4): Update.
1066 * config/sparc/sparc.c (output_v9branch): Update.
1067 * config/spu/spu.c (get_branch_target): Update.
1068 (ea_load_store_inline): Update.
1069 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1070 * config/tilepro/tilepro.c: Update.
1072 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1074 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1075 (gimplify_save_expr): Likewise.
1077 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1079 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1081 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1083 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1085 * timevar.def (TV_IPA_FNSUMMARY): Define.
1087 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1089 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1090 to back store errata sensitive sequence from being generated.
1091 (sqrtdf2_fix): Likewise.
1093 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1095 * tree-ssa-threadupdate.c (compute_path_counts,
1096 update_joiner_offpath_counts): Use profile_probability.
1098 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1101 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1103 * config/arm/arm-c.c (arm_cpu_builtins): Define
1104 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1106 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1108 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1109 array entries to represent __ieee128 versions of the
1110 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1111 scalar_extract_sig, and scalar_insert_exp built-in functions.
1112 (altivec_resolve_overloaded_builtin): Add special case handling
1113 for the __builtin_scalar_insert_exp function, as represented by
1114 the P9V_BUILTIN_VEC_VSIEDP constant.
1115 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1116 exponent support for __ieee128 argument.
1117 (VSESQP): Add scalar extract signature support for __ieee128
1119 (VSTDCNQP): Add scalar test negative support for __ieee128
1121 (VSIEQP): Add scalar insert exponent support for __int128 argument
1122 with __ieee128 result.
1123 (VSIEQPF): Add scalar insert exponent support for __ieee128
1124 argument with __ieee128 result.
1125 (VSTDCQP): Add scalar test data class support for __ieee128
1127 (VSTDCNQP): Add overload support for scalar test negative with
1129 (VSTDCQP): Add overload support for scalar test data class
1131 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1133 (UNSPEC_VSX_SIEXPQP): New constant.
1134 (xsxexpqp): New insn for VSX scalar extract exponent quad
1136 (xsxsigqp): New insn for VSX scalar extract significand quad
1138 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1139 precision with floating point argument.
1140 (xststdcqp): New expand for VSX scalar test data class quad
1142 (xststdcnegqp): New expand for VSX scalar test negative quad
1144 (xststdcqp): New insn to match expansions for VSX scalar test data
1145 class quad precision and VSX scalar test negative quad precision.
1146 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1147 special case operand checking to enforce that second operand of
1148 VSX scalar test data class with quad precision argument is a 7-bit
1150 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1151 prototypes and descriptions of __ieee128 versions of
1152 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1153 scalar_test_data_class, and scalar_test_neg built-in functions.
1155 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1157 PR tree-optimization/81162
1158 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1159 replace a negate with an add.
1161 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1163 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1165 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1167 * config/arm/arm-c.c (arm_cpu_builtins): Define
1168 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1170 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1172 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1173 (armv8-r): Set ARM Cortex-R52 as default CPU.
1174 * config/arm/arm-tables.opt: Regenerate.
1175 * config/arm/arm-tune.md: Regenerate.
1176 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1178 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1179 extension for -mcpu=cortex-r52.
1181 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1183 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1184 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1185 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1186 (fp-armv8): Define it as FP_ARMv8 only.
1187 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1188 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1190 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1191 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1192 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1193 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1194 than TARGET_FPU_ARMV8.
1195 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1196 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1197 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1199 * config/arm/neon.md (neon_vrint): Likewise.
1200 (neon_vcvt): Likewise.
1201 (neon_<fmaxmin_op><mode>): Likewise.
1202 (<fmaxmin><mode>3): Likewise.
1203 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1204 * config/arm/predicates.md (arm_cond_move_operator): Check against
1205 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1207 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1209 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1212 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1214 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1215 loop in comment with memset.
1217 2017-07-14 Martin Liska <mliska@suse.cz>
1219 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1220 * dwarf2out.c (is_java): Remove the function.
1221 (output_pubname): Remove usage of the function.
1222 (lower_bound_default): Remove usage of DW_LANG_Java.
1223 (gen_compile_unit_die): Likewise.
1224 * gcc.c: Remove compiler defaults for .java and .zip files.
1225 * gimple-expr.c (remove_suffix): Change as there's no longer
1226 extension than 4-letter one.
1227 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1228 (gimplify_save_expr): Likewise.
1229 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1230 as it's possible even for other languages than Java.
1231 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1232 * lto-opts.c (lto_write_options): Remove reference to Java.
1233 * opts.c (strip_off_ending): Update file extension handling.
1234 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1235 * tree-eh.c (lower_resx): Likewise.
1236 * tree.c (free_lang_data_in_type): Remove dead code.
1237 (find_decls_types_r): Likewise.
1238 (build_common_builtin_nodes): Remove Java from a comment.
1239 (verify_type): Remove dead code.
1240 * varasm.c (assemble_external): Remove Java from a comment.
1242 2017-07-14 Martin Liska <mliska@suse.cz>
1244 * opts.c (finish_options): Add quotes.
1245 (common_handle_option): Likewise.
1247 2017-07-14 Martin Liska <mliska@suse.cz>
1249 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1250 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1252 * dwarf2out.c (lower_bound_default): Do not handle
1254 (gen_compile_unit_die): Likewise.
1255 * gcc.c: Remove default extension binding for GNU Pascal.
1256 * stmt.c: Remove Pascal language from a comment.
1257 * xcoffout.c: Likewise.
1259 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1262 * diagnostic-show-locus.c (fixit_cmp): New function.
1263 (layout::layout): Sort m_fixit_hints.
1264 (column_range::column_range): Assert that the values are valid.
1265 (struct char_span): New struct.
1266 (correction::overwrite): New method.
1267 (struct source_line): New struct.
1268 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1269 calls in terms of classes source_line and char_span, and
1270 correction::overwrite.
1271 (selftest::test_overlapped_fixit_printing_2): New function.
1272 (selftest::diagnostic_show_locus_c_tests): Call it.
1274 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1276 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1277 early if there is no lhs.
1279 2017-07-13 Martin Liska <mliska@suse.cz>
1281 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1282 (gen_reference_type_die): Likewise.
1283 * stor-layout.c: Remove Pascal-related comment.
1285 2017-07-13 Martin Liska <mliska@suse.cz>
1287 * opts.c (finish_options): Add quotes to error messages.
1288 (parse_sanitizer_options): Likewise.
1290 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1292 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1294 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1296 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1298 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1300 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1302 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1304 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1307 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1308 provides the hardware capability bits, define the macro
1309 __BUILTIN_CPU_SUPPORTS__.
1310 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1311 if GLIBC does not provide the hardware capability bits. Add a
1312 gcc_unreachable call if the built-in cpu function is neither
1313 __builtin_cpu_is nor __builtin_cpu_supports.
1314 (rs6000_get_function_versions_dispatcher): Change the warning
1315 that an old GLIBC is used which does not export the capability
1316 bits to be an error.
1317 * doc/extend.texi (target_clones attribute): Document the
1318 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1319 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1320 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1321 the macros defined by GCC if the newer GLIBC is available.
1323 2017-07-12 Jeff Law <law@redhat.com>
1325 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1326 remaining includes slightly.
1327 * config/riscv/riscv-builtins.c: Include profile-count.h.
1329 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1332 * config/avr/avr.c (avr_set_current_function): In diagnostic
1333 messages: Quote keywords and (parts of) identifiers.
1334 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1337 2017-07-12 Carl Love <cel@us.ibm.com>
1339 * config/rs6000/rs6000-c.c: Add support for built-in functions
1340 vector bool char vec_revb (vector bool char);
1341 vector bool short vec_revb (vector short char);
1342 vector bool int vec_revb (vector bool int);
1343 vector bool long long vec_revb (vector bool long long);
1344 * doc/extend.texi: Update the built-in documentation file for the
1345 new built-in functions.
1347 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1349 * config/s390/s390.md: Remove movcc splitter.
1351 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1353 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1354 load/store on condition.
1356 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1359 * config/avr/avr.c (avr_encode_section_info)
1360 [progmem && !TREE_READONLY]: Error if progmem object needs
1363 2017-07-11 Michael Collison <michael.collison@arm.com>
1365 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1368 2017-07-11 Carl Love <cel@us.ibm.com>
1370 * config/rs6000/rs6000-c.c: Add support for builtins
1371 vector unsigned int vec_parity_lsbb (vector signed int);
1372 vector unsigned int vec_parity_lsbb (vector unsigned int);
1373 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1374 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1375 vector unsigned long long vec_parity_lsbb (vector signed long long);
1376 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1377 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1378 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1379 * doc/extend.texi: Update the built-in documentation file for the
1380 new built-in functions.
1382 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1384 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1385 (layout::m_primary_loc): New field.
1386 (layout::layout): Initialize new field. Move location filtering
1387 logic from here to...
1388 (layout::maybe_add_location_range): ...this new method. Add
1389 support for filtering to just the lines already specified by other
1391 (layout::will_show_line_p): New method.
1392 (gcc_rich_location::add_location_if_nearby): New method.
1393 (selftest::test_add_location_if_nearby): New test function.
1394 (selftest::diagnostic_show_locus_c_tests): Call it.
1395 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1398 2017-07-11 Tom de Vries <tom@codesourcery.com>
1400 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1401 (bb_first_real_insn): New function.
1402 (nvptx_single): Add extra initialization of broadcasted condition
1405 2017-07-11 Nathan Sidwell <nathan@acm.org>
1407 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1409 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1411 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1412 Binutils doc as TEXI will mess them up.
1413 * doc/invoke.texi (AVR Options): Same here.
1415 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1417 * config/sparc/sparc.opt (mfix-ut700): New option.
1418 (mfix-gr712rc): Likewise.
1419 (sparc_fix_b2bst): New variable.
1420 * doc/invoke.texi (SPARC options): Document them.
1421 (ARM options): Fix warnings.
1422 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1423 instructions to prevent sequences that can trigger the store-store
1424 errata for certain LEON3FT processors.
1425 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1426 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1427 * config/sparc/sparc.md (fix_b2bst): New attribute.
1428 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1430 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1433 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1435 (*rsqrtsf2_sse): Ditto.
1437 (div<mode>3): Macroize insn from divdf3 and divsf3
1438 using MODEF mode iterator.
1440 2017-07-10 Martin Sebor <msebor@redhat.com>
1442 PR tree-optimization/80397
1443 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1444 instead of testing for equality to INTEGER_TYPE.
1446 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1448 * config.gcc: Remove uclibc from arc target spec.
1450 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1452 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1454 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1457 * lto-wrapper.c (remove_option): New function.
1458 (merge_and_complain): Merge PIC/PIE options more realistically.
1460 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1462 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1466 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1467 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1468 * config.in: Regenerate.
1469 * configure: Regenerate.
1470 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1471 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1472 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1473 (TARGET_GASISR_PROLOGUES): ...target mask.
1474 * common/config/avr/avr-common.c
1475 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1476 Set -mgas-isr-prologues.
1477 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1478 INSERT_PASS_BEFORE for it.
1479 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1480 * config/avr/avr.c (avr_option_override)
1481 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1482 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1483 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1484 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1485 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1487 (make_avr_pass_pre_proep): New function.
1488 (emit_push_sfr) <treg>: Add argument to function and use it
1490 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1491 and set machine->gasisr.yes.
1492 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1493 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1494 __gcc_isr.n_pushed to .L__stack_usage.
1495 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1496 (avr_asm_final_postscan_insn): ...this new static function.
1497 * config/avr/avr.h (machine_function)
1498 <is_no_gccisr, use_L__stack_usage>: New fields.
1499 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1500 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1501 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1502 (gasisr, *gasisr): New expander and insn.
1503 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1504 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1505 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1507 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1509 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1512 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1514 Move jump-tables out of .text again.
1517 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1518 (ASM_OUTPUT_ADDR_VEC): New function.
1519 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1520 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1521 INSN_ADDRESSes as asm comment.
1522 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1523 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1524 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1525 * config/avr/avr.md (*tablejump): Adjust comment.
1526 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1527 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1529 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1530 (avr_output_addr_vec): New proto.
1531 (avr_log_t) <insn_addresses>: New field.
1533 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1536 * config/i386/i386.c (ix86_function_arg_advance): Set
1537 outgoing_args_on_stack to true if there are outgoing arguments
1539 (ix86_function_arg): Likewise.
1540 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1541 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1542 * config/i386/i386.h (machine_function): Add
1543 outgoing_args_on_stack.
1545 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1547 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1549 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1551 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1553 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1554 (REAL_H): Remove $(MACHMODE_H).
1555 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1557 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1558 $(MACHMODE_H) and double-int.h.
1559 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1561 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1564 2017-07-07 Andrew Pinski <apinski@cavium.com>
1566 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1567 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1569 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1571 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1572 Add warning if GCC was not configured to link against a GLIBC that
1573 exports the hardware capability bits.
1574 (make_resolver_func): Make resolver function private and not a
1575 COMDAT function. Create the name with clone_function_name instead
1576 of make_unique_name.
1579 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1580 correct operand in doing the split.
1582 2017-07-07 Carl Love <cel@us.ibm.com>
1584 * config/rs6000/rs6000-c: Add support for built-in function
1585 vector unsigned short vec_pack_to_short_fp32 (vector float,
1587 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1588 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1589 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1590 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1591 (convert_4f32_8i16): Add define_expand.
1592 * doc/extend.texi: Update the built-in documentation file for the
1593 new built-in function.
1595 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1597 * config/sparc/m8.md: New file.
1598 * config/sparc/sparc.md: Include m8.md.
1600 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1602 * config/sparc/sparc.opt: New option -mvis4b.
1603 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1604 (sparc_option_override): Handle VIS4B.
1605 (enum sparc_builtins): Define
1606 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1607 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1608 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1609 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1610 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1611 (check_constant_argument): New function.
1612 (sparc_vis_init_builtins): Define builtins
1613 __builtin_vis_dictunpack{8,16,32},
1614 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1615 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1616 __builtin_vis_fpcmpde{8,16,32}shl and
1617 __builtin_vis_fpcmpur{8,16,32}shl.
1618 (sparc_expand_builtin): Check that the constant operands to
1619 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1620 constant and in range.
1621 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1623 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1624 (SPARC_IMM5_P): Likewise.
1625 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1626 (enabled): Handle vis4b.
1627 (UNSPEC_DICTUNPACK): New unspec.
1628 (UNSPEC_FPCMPSHL): Likewise.
1629 (UNSPEC_FPUCMPSHL): Likewise.
1630 (UNSPEC_FPCMPDESHL): Likewise.
1631 (UNSPEC_FPCMPURSHL): Likewise.
1632 (cpu_feature): New CPU feature `vis4b'.
1633 (dictunpack{8,16,32}): New insns.
1634 (FPCSMODE): New mode iterator.
1635 (fpcscond): New code iterator.
1636 (fpcsucond): Likewise.
1637 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1638 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1639 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1640 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1641 * config/sparc/constraints.md: Define constraints `q' for unsigned
1642 2-bit integer constants and `t' for unsigned 5-bit integer
1644 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1646 (imm5_operand_dictunpack16): Likewise.
1647 (imm5_operand_dictunpack32): Likewise.
1648 (imm2_operand): Likewise.
1649 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1650 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1651 ditunpack* and fpcmp*shl builtins.
1653 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1655 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1656 * config.in: Add HAVE_AS_SPARC6 define.
1657 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1659 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1661 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1662 (CPP_CPU_SPEC): Handle m8.
1663 (ASM_CPU_SPEC): Likewise.
1664 * config/sparc/sparc-opts.h (enum processor_type): Add
1666 * config/sparc/sparc.c (m8_costs): New struct.
1667 (sparc_option_override): Handle TARGET_CPU_m8.
1668 (sparc32_initialize_trampoline): Likewise.
1669 (sparc64_initialize_trampoline): Likewise.
1670 (sparc_issue_rate): Likewise.
1671 (sparc_register_move_cost): Likewise.
1672 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1673 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1674 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1675 (CPP_CPU_SPEC): Handle M8.
1676 (ASM_CPU_SPEC): Likewise.
1677 (AS_M8_FLAG): Define.
1678 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1679 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1680 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1682 * configure: Regenerate.
1683 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1686 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1688 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1690 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1691 ("*movdi_insn_sp32"): Do not set v3pipe.
1692 ("*movsi_insn"): Likewise.
1693 ("*movdi_insn_sp64"): Likewise.
1694 ("*movsf_insn"): Likewise.
1695 ("*movdf_insn_sp32"): Likewise.
1696 ("*movdf_insn_sp64"): Likewise.
1697 ("*zero_extendsidi2_insn_sp64"): Likewise.
1698 ("*sign_extendsidi2_insn"): Likewise.
1699 ("*mov<VM32:mode>_insn"): Likewise.
1700 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1701 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1702 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1703 ("<vlop:code><VL:mode>3"): Likewise.
1704 ("*not_<vlop:code><VL:mode>3"): Likewise.
1705 ("*nand<VL:mode>_vis"): Likewise.
1706 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1707 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1708 ("one_cmpl<VL:mode>2"): Likewise.
1709 ("faligndata<VM64:mode>_vis"): Likewise.
1710 ("alignaddrsi_vis"): Likewise.
1711 ("alignaddrdi_vis"): Likweise.
1712 ("alignaddrlsi_vis"): Likewise.
1713 ("alignaddrldi_vis"): Likewise.
1714 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1715 ("bmaskdi_vis"): Likewise.
1716 ("bmasksi_vis"): Likewise.
1717 ("bshuffle<VM64:mode>_vis"): Likewise.
1718 ("cmask8<P:mode>_vis"): Likewise.
1719 ("cmask16<P:mode>_vis"): Likewise.
1720 ("cmask32<P:mode>_vis"): Likewise.
1721 ("pdistn<P:mode>_vis"): Likewise.
1722 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1724 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1726 * config/sparc/sparc.md ("subtype"): New insn attribute.
1727 ("*wrgsr_sp64"): Set insn subtype.
1728 ("*rdgsr_sp64"): Likewise.
1729 ("alignaddrsi_vis"): Likewise.
1730 ("alignaddrdi_vis"): Likewise.
1731 ("alignaddrlsi_vis"): Likewise.
1732 ("alignaddrldi_vis"): Likewise.
1733 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1734 ("fexpand_vis"): Likewise.
1735 ("fpmerge_vis"): Likewise.
1736 ("faligndata<VM64:mode>_vis"): Likewise.
1737 ("bshuffle<VM64:mode>_vis"): Likewise.
1738 ("cmask8<P:mode>_vis"): Likewise.
1739 ("cmask16<P:mode>_vis"): Likewise.
1740 ("cmask32<P:mode>_vis"): Likewise.
1741 ("fchksm16_vis"): Likewise.
1742 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1743 ("fmean16_vis"): Likewise.
1744 ("fp<plusminus_insn>64_vis"): Likewise.
1745 ("<plusminus_insn>v8qi3"): Likewise.
1746 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1747 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1748 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1749 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1750 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1751 ("*movqi_insn"): Likewise.
1752 ("*movhi_insn"): Likewise.
1753 ("*movsi_insn"): Likewise.
1754 ("movsi_pic_gotdata_op"): Likewise.
1755 ("*movdi_insn_sp32"): Likewise.
1756 ("*movdi_insn_sp64"): Likewise.
1757 ("movdi_pic_gotdata_op"): Likewise.
1758 ("*movsf_insn"): Likewise.
1759 ("*movdf_insn_sp32"): Likewise.
1760 ("*movdf_insn_sp64"): Likewise.
1761 ("*zero_extendhisi2_insn"): Likewise.
1762 ("*zero_extendqihi2_insn"): Likewise.
1763 ("*zero_extendqisi2_insn"): Likewise.
1764 ("*zero_extendqidi2_insn"): Likewise.
1765 ("*zero_extendhidi2_insn"): Likewise.
1766 ("*zero_extendsidi2_insn_sp64"): Likewise.
1767 ("ldfsr"): Likewise.
1768 ("prefetch_64"): Likewise.
1769 ("prefetch_32"): Likewise.
1770 ("tie_ld32"): Likewise.
1771 ("tie_ld64"): Likewise.
1772 ("*tldo_ldub_sp32"): Likewise.
1773 ("*tldo_ldub1_sp32"): Likewise.
1774 ("*tldo_ldub2_sp32"): Likewise.
1775 ("*tldo_ldub_sp64"): Likewise.
1776 ("*tldo_ldub1_sp64"): Likewise.
1777 ("*tldo_ldub2_sp64"): Likewise.
1778 ("*tldo_ldub3_sp64"): Likewise.
1779 ("*tldo_lduh_sp32"): Likewise.
1780 ("*tldo_lduh1_sp32"): Likewise.
1781 ("*tldo_lduh_sp64"): Likewise.
1782 ("*tldo_lduh1_sp64"): Likewise.
1783 ("*tldo_lduh2_sp64"): Likewise.
1784 ("*tldo_lduw_sp32"): Likewise.
1785 ("*tldo_lduw_sp64"): Likewise.
1786 ("*tldo_lduw1_sp64"): Likewise.
1787 ("*tldo_ldx_sp64"): Likewise.
1788 ("*mov<VM32:mode>_insn"): Likewise.
1789 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1790 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1792 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1794 * config/sparc/sparc.md ("type"): New insn type viscmp.
1795 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1797 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1798 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1799 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1800 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1802 ("n7_vis_logical_11cycle"): Likewise.
1803 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1804 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1805 * config/sparc/niagara.md ("niag_vis"): Likewise.
1806 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1807 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1809 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1811 * config/sparc/sparc.md: New instruction type `bmask'.
1812 (bmaskdi_vis): Use the `bmask' type.
1813 (bmasksi_vis): Likewise.
1814 * config/sparc/ultra3.md (us3_array): Likewise.
1815 * config/sparc/niagara7.md (n7_array): Likewise.
1816 * config/sparc/niagara4.md (n4_array): Likewise.
1817 * config/sparc/niagara2.md (niag2_vis): Likewise.
1818 (niag3_vis): Likewise.
1819 * config/sparc/niagara.md (niag_vis): Likewise.
1821 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1823 * ipa-comdats.c: Remove optimize check from gate.
1824 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1825 for functions not optimized.
1826 (ipa_fn_summary_read): Skip optimize check.
1827 (ipa_fn_summary_write): Likewise.
1828 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1830 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1832 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1833 (check_callers): Give up on caller which is not optimized.
1834 (inline_small_functions): Likewise.
1835 (ipa_inline): Do not give up when not optimizing.
1836 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1837 away unoptimizes cdtors.
1838 (whole_program_function_and_variable_visibility): Do
1839 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1840 * ipa.c (process_references): Do not check optimize.
1841 (symbol_table::remove_unreachable_nodes): Update optimize check.
1842 (set_writeonly_bit): Update optimize check.
1843 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1844 (pass_ipa_single_use::gate): Remove.
1846 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1848 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1849 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1850 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1851 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1852 permute_load, permute_store, adjust_extract, adjust_splat,
1853 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1854 replace_swap_with_copy, dump_swap_insn_table,
1855 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1856 recombine_lvx_pattern, recombine_stvx_pattern,
1857 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1858 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1859 to file rs6000-p8swap.c.
1860 * config/rs6000/rs6000-p8swap.c: New file.
1861 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1862 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1863 and rs6000*-*-* targets.
1865 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1867 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1869 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1871 * lto-wrapper.c (merge_and_complain): Do not merge
1872 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
1873 fsigned_zeros, ftrapping_math, fwrapv.
1874 (append_compiler_options): Do not track these options.
1875 (append_linker_options): Likewie
1877 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1879 * cgraphunit.c (cgraph_node::finalize_function): When
1880 !flag_toplevel_reorde set no_reorder flag.
1881 (varpool_node::finalize_decl): Likewise.
1882 (symbol_table::compile): Drop no toplevel reorder path.
1884 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1886 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
1887 edges; zero probability is not better than uninitialized.
1889 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1891 * asan.h (asan_sanitize_allocas_p): Declare.
1892 * asan.c (asan_sanitize_allocas_p): New function.
1893 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
1894 (handle_builtin_alloca): Likewise.
1895 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
1896 if !asan_sanitize_allocas_p.
1897 * params.def (asan-instrument-allocas): Add new option.
1898 * params.h (ASAN_PROTECT_ALLOCAS): Define.
1899 * opts.c (common_handle_option): Disable allocas sanitization for
1902 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1904 * asan.c: Include gimple-fold.h.
1905 (get_last_alloca_addr): New function.
1906 (handle_builtin_stackrestore): Likewise.
1907 (handle_builtin_alloca): Likewise.
1908 (asan_emit_allocas_unpoison): Likewise.
1909 (get_mem_refs_of_builtin_call): Add new parameter, remove const
1910 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
1911 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
1912 (instrument_builtin_call): Pass gimple iterator to
1913 get_mem_refs_of_builtin_call.
1914 (last_alloca_addr): New global.
1915 * asan.h (asan_emit_allocas_unpoison): Declare.
1916 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
1917 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
1918 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
1919 if function calls alloca.
1920 * gimple-fold.c (replace_call_with_value): Remove static keyword.
1921 * gimple-fold.h (replace_call_with_value): Declare.
1922 * internal-fn.c: Include asan.h.
1923 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
1924 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
1926 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1928 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
1929 (C_SELFTEST_FLAGS): New.
1930 (CPP_SELFTEST_FLAGS): New.
1931 (SELFTEST_DEPS): New, from deps of s-selftest.
1932 (C_SELFTEST_DEPS): New, from deps of s-selftest.
1933 (CPP_SELFTEST_DEPS): New.
1934 (selftest): Add dependency on s-selftest-c++.
1935 (s-selftest): Rename to...
1936 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
1937 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
1938 than SELFTEST_FLAGS.
1939 (selftest-gdb): Rename to...
1940 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
1942 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
1943 (selftest-valgrind): Rename to...
1944 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
1946 (selftest-valgrind): Reintroduce as an alias for
1947 selftest-c-valgrind.
1948 (s-selftest-c++): New.
1949 (selftest-c++-gdb): New.
1950 (selftest-c++-valgrind): New.
1952 2017-07-06 Olivier Hainque <hainque@adacore.com>
1954 * gcc.c (process_command): When deciding if undefined variables
1955 should be ignored when processing specs, accept "gcc -v" as well.
1957 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1959 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
1960 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
1962 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1964 * config/arm/arm-cpus.in (armv8-r): Add new entry.
1965 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
1966 * config/arm/arm-tables.opt: Regenerate.
1967 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
1969 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
1971 2017-07-06 Carl Love <cel@us.ibm.com>
1973 * ChangeLog: Clean up from mid air collision
1975 2017-07-06 Carl Love <cel@us.ibm.com>
1977 * config/rs6000/rs6000-c.c: Add support for built-in functions
1978 vector signed int vec_subc (vector signed int, vector signed int);
1979 vector signed __int128 vec_subc (vector signed __int128,
1980 vector signed __int128);
1981 vector unsigned __int128 vec_subc (vector unsigned __int128,
1982 vector unsigned __int128);
1983 vector signed int vec_sube (vector signed int, vector signed int,
1985 vector unsigned int vec_sube (vector unsigned int,
1986 vector unsigned int,
1987 vector unsigned int);
1988 vector signed __int128 vec_sube (vector signed __int128,
1989 vector signed __int128,
1990 vector signed__int128);
1991 vector unsigned __int128 vec_sube (vector unsigned __int128,
1992 vector unsigned __int128,
1993 vector unsigned __int128);
1994 vector signed int vec_subec (vector signed int, vector signed int,
1996 vector unsigned int vec_subec (vector unsigned int,
1997 vector unsigned int,
1998 vector unsigned int);
1999 vector signed __int128 vec_subec (vector signed __int128,
2000 vector signed __int128,
2001 vector signed__int128);
2002 vector unsigned __int128 vec_subec (vector unsigned __int128,
2003 vector unsigned __int128,
2004 vector unsigned __int128);
2005 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2006 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2007 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2008 BU_ALTIVEC_OVERLOAD_X definitions.
2009 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2010 * doc/extend.texi: Update the built-in documentation file for the new
2013 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2016 * diagnostic-show-locus.c (layout::layout): Use start and finish
2017 spelling location for the start and finish of each range.
2018 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2019 Add unused aspect param.
2020 * input.c (expand_location_1): Add "aspect" param, and use it
2021 to access the correct part of the location.
2022 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2024 (expand_location_to_spelling_point): Likewise.
2025 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2026 param, and pass it to expand_location_1.
2028 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2030 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2031 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2032 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2033 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2034 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2035 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2036 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2037 _mm_maskz_getmant_ss): New intrinsics.
2038 (__builtin_ia32_getexpss128_mask): Changed to ...
2039 __builtin_ia32_getexpss128_round ... this.
2040 (__builtin_ia32_getexpsd128_mask): Changed to ...
2041 __builtin_ia32_getexpsd128_round ... this.
2042 * config/i386/i386-builtin-types.def
2043 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2044 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2045 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2046 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2047 __builtin_ia32_getmantss_mask_round): New builtins.
2048 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2049 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2050 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2051 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2052 * config/i386/sse.md
2053 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2054 avx512f_sgetexp<mode><mask_scalar_name>
2055 <round_saeonly_scalar_name> ... this.
2056 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2057 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2058 vgetexp<ssescalarmodesuffix>
2059 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2060 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2061 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2062 avx512f_vgetmant<mode><mask_scalar_name>
2063 <round_saeonly_scalar_name> ... this.
2064 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2065 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2066 vgetmant<ssescalarmodesuffix>
2067 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2068 %0<mask_scalar_operand4>, %1, %2
2069 <round_saeonly_scalar_mask_op4>, %3} ... this.
2070 * config/i386/subst.md (mask_scalar_operand4,
2071 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2072 round_saeonly_scalar_nimm_predicate): New subst attributes.
2074 2017-07-06 Julia Koval <julia.koval@intel.com>
2076 * config/i386/i386.c (ix86_erase_embedded_rounding):
2077 Remove code for old rounding pattern.
2079 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2081 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2083 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2085 * doc/sourcebuild.texi (Test Directives, Variants of
2086 dg-require-support): Add documentation for dg-require-stack-check.
2088 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2090 * config/i386/subst.md (mask_scalar, round_scalar,
2091 round_saeonly_scalar): New meta-templates.
2092 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2093 round_scalar_mask_operand3, round_scalar_mask_op3,
2094 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2095 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2096 round_saeonly_scalar_constraint,
2097 round_saeonly_scalar_prefix): New subst attribute.
2098 * config/i386/sse.md
2099 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2100 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2101 <round_scalar_name> ... this.
2102 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2103 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2104 <round_scalar_name> ... this.
2105 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2106 <sse>_vm<code><mode>3<mask_scalar_name>
2107 <round_saeonly_scalar_name> ... this.
2108 (v<plusminus_mnemonic><ssescalarmodesuffix>
2109 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2110 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2111 v<plusminus_mnemonic><ssescalarmodesuffix>
2112 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2113 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2114 (v<multdiv_mnemonic><ssescalarmodesuffix>
2115 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2116 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2117 v<multdiv_mnemonic><ssescalarmodesuffix>
2118 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2119 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2120 (v<maxmin_float><ssescalarmodesuffix>
2121 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2122 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2123 v<maxmin_float><ssescalarmodesuffix>
2124 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2125 %0<mask_scalar_operand3>, %1, %<iptr>2
2126 <round_saeonly_scalar_mask_op3>} ... this.
2128 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2130 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2131 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2133 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2134 Alan Hayward <alan.hayward@arm.com>
2135 David Sherwood <david.sherwood@arm.com>
2137 * combine.c (simplify_if_then_else): Remove "enum" before
2139 * compare-elim.c (can_eliminate_compare): Likewise.
2140 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2142 (aarch64_lookup_simd_builtin_type): Likewise.
2143 (aarch64_simd_builtin_type): Likewise.
2144 (aarch64_init_simd_builtin_types): Likewise.
2145 (aarch64_simd_expand_args): Likewise.
2146 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2148 (aarch64_reverse_mask): Likewise.
2149 (aarch64_simd_emit_reg_reg_move): Likewise.
2150 (aarch64_gen_adjusted_ldpstp): Likewise.
2151 (aarch64_ccmp_mode_to_code): Likewise.
2152 (aarch64_operands_ok_for_ldpstp): Likewise.
2153 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2154 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2156 (aarch64_min_divisions_for_recip_mul): Likewise.
2157 (aarch64_reassociation_width): Likewise.
2158 (aarch64_get_condition_code_1): Likewise.
2159 (aarch64_simd_emit_reg_reg_move): Likewise.
2160 (aarch64_simd_attr_length_rglist): Likewise.
2161 (aarch64_reverse_mask): Likewise.
2162 (aarch64_operands_ok_for_ldpstp): Likewise.
2163 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2164 (aarch64_gen_adjusted_ldpstp): Likewise.
2165 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2167 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2168 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2169 (arm_lookup_simd_builtin_type): Likewise.
2170 (arm_simd_builtin_type): Likewise.
2171 (arm_init_simd_builtin_types): Likewise.
2172 (arm_expand_builtin_args): Likewise.
2173 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2174 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2175 (ft32_setup_incoming_varargs): Likewise.
2176 (ft32_function_arg): Likewise.
2177 (ft32_function_arg_advance): Likewise.
2178 (ft32_pass_by_reference): Likewise.
2179 (ft32_arg_partial_bytes): Likewise.
2180 (ft32_valid_pointer_mode): Likewise.
2181 (ft32_addr_space_pointer_mode): Likewise.
2182 (ft32_addr_space_legitimate_address_p): Likewise.
2183 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2185 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2186 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2187 (iamcu_alignment): Likewise.
2188 (canonicalize_vector_int_perm): Likewise.
2189 (ix86_noce_conversion_profitable_p): Likewise.
2190 (ix86_mpx_bound_mode): Likewise.
2191 (ix86_operands_ok_for_move_multiple): Likewise.
2192 * config/microblaze/microblaze-protos.h
2193 (microblaze_expand_conditional_branch_reg): Likewise.
2194 * config/microblaze/microblaze.c
2195 (microblaze_expand_conditional_branch_reg): Likewise.
2196 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2198 (rs6000_reassociation_width): Likewise.
2199 (rs6000_invalid_binary_op): Likewise.
2200 (fusion_p9_p): Likewise.
2201 (emit_fusion_p9_load): Likewise.
2202 (emit_fusion_p9_store): Likewise.
2203 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2205 (riscv_hard_regno_mode_ok_p): Likewise.
2206 (riscv_address_insns): Likewise.
2207 (riscv_split_symbol): Likewise.
2208 (riscv_legitimize_move): Likewise.
2209 (riscv_function_value): Likewise.
2210 (riscv_hard_regno_nregs): Likewise.
2211 (riscv_expand_builtin): Likewise.
2212 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2213 (riscv_build_integer): Likewise.
2214 (riscv_split_integer): Likewise.
2215 (riscv_legitimate_constant_p): Likewise.
2216 (riscv_cannot_force_const_mem): Likewise.
2217 (riscv_regno_mode_ok_for_base_p): Likewise.
2218 (riscv_valid_base_register_p): Likewise.
2219 (riscv_valid_offset_p): Likewise.
2220 (riscv_valid_lo_sum_p): Likewise.
2221 (riscv_classify_address): Likewise.
2222 (riscv_legitimate_address_p): Likewise.
2223 (riscv_address_insns): Likewise.
2224 (riscv_load_store_insns): Likewise.
2225 (riscv_force_binary): Likewise.
2226 (riscv_split_symbol): Likewise.
2227 (riscv_force_address): Likewise.
2228 (riscv_legitimize_address): Likewise.
2229 (riscv_move_integer): Likewise.
2230 (riscv_legitimize_const_move): Likewise.
2231 (riscv_legitimize_move): Likewise.
2232 (riscv_address_cost): Likewise.
2233 (riscv_subword): Likewise.
2234 (riscv_output_move): Likewise.
2235 (riscv_canonicalize_int_order_test): Likewise.
2236 (riscv_emit_int_order_test): Likewise.
2237 (riscv_function_arg_boundary): Likewise.
2238 (riscv_pass_mode_in_fpr_p): Likewise.
2239 (riscv_pass_fpr_single): Likewise.
2240 (riscv_pass_fpr_pair): Likewise.
2241 (riscv_get_arg_info): Likewise.
2242 (riscv_function_arg): Likewise.
2243 (riscv_function_arg_advance): Likewise.
2244 (riscv_arg_partial_bytes): Likewise.
2245 (riscv_function_value): Likewise.
2246 (riscv_pass_by_reference): Likewise.
2247 (riscv_setup_incoming_varargs): Likewise.
2248 (riscv_print_operand): Likewise.
2249 (riscv_elf_select_rtx_section): Likewise.
2250 (riscv_save_restore_reg): Likewise.
2251 (riscv_for_each_saved_reg): Likewise.
2252 (riscv_register_move_cost): Likewise.
2253 (riscv_hard_regno_mode_ok_p): Likewise.
2254 (riscv_hard_regno_nregs): Likewise.
2255 (riscv_class_max_nregs): Likewise.
2256 (riscv_memory_move_cost): Likewise.
2257 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2258 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2259 (rl78_addr_space_address_mode): Likewise.
2260 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2262 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2263 (rs6000_reassociation_width): Likewise.
2264 (rs6000_invalid_binary_op): Likewise.
2265 (fusion_p9_p): Likewise.
2266 (emit_fusion_p9_load): Likewise.
2267 (emit_fusion_p9_store): Likewise.
2268 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2269 (ok_for_simple_move_operands): Likewise.
2270 (ok_for_simple_move_strict_operands): Likewise.
2271 (ok_for_simple_arith_logic_operands): Likewise.
2272 (visium_legitimize_reload_address): Likewise.
2273 (visium_select_cc_mode): Likewise.
2274 (output_cbranch): Likewise.
2275 (visium_split_double_move): Likewise.
2276 (visium_expand_copysign): Likewise.
2277 (visium_expand_int_cstore): Likewise.
2278 (visium_expand_fp_cstore): Likewise.
2279 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2280 (visium_function_arg): Likewise.
2281 (visium_function_arg_advance): Likewise.
2282 (visium_libcall_value): Likewise.
2283 (visium_setup_incoming_varargs): Likewise.
2284 (visium_legitimate_constant_p): Likewise.
2285 (visium_legitimate_address_p): Likewise.
2286 (visium_legitimize_address): Likewise.
2287 (visium_secondary_reload): Likewise.
2288 (visium_register_move_cost): Likewise.
2289 (visium_memory_move_cost): Likewise.
2290 (prepare_move_operands): Likewise.
2291 (ok_for_simple_move_operands): Likewise.
2292 (ok_for_simple_move_strict_operands): Likewise.
2293 (ok_for_simple_arith_logic_operands): Likewise.
2294 (visium_function_value_1): Likewise.
2295 (rtx_ok_for_offset_p): Likewise.
2296 (visium_legitimize_reload_address): Likewise.
2297 (visium_split_double_move): Likewise.
2298 (visium_expand_copysign): Likewise.
2299 (visium_expand_int_cstore): Likewise.
2300 (visium_expand_fp_cstore): Likewise.
2301 (visium_split_cstore): Likewise.
2302 (visium_select_cc_mode): Likewise.
2303 (visium_split_cbranch): Likewise.
2304 (output_cbranch): Likewise.
2305 (visium_print_operand_address): Likewise.
2306 * expmed.c (flip_storage_order): Likewise.
2307 * expmed.h (emit_cstore): Likewise.
2308 (flip_storage_order): Likewise.
2309 * genrecog.c (validate_pattern): Likewise.
2310 * hsa-gen.c (gen_hsa_addr): Likewise.
2311 * internal-fn.c (expand_arith_overflow): Likewise.
2312 * ira-color.c (allocno_copy_cost_saving): Likewise.
2313 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2314 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2315 (process_invariant_for_inheritance): Likewise.
2316 * lra-eliminations.c (move_plus_up): Likewise.
2317 * omp-low.c (lower_oacc_reductions): Likewise.
2318 * simplify-rtx.c (simplify_subreg): Likewise.
2319 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2320 (TARGET_CHKP_BOUND_MODE): Likewise..
2321 * targhooks.c (default_chkp_bound_mode): Likewise.
2322 (default_setup_incoming_vararg_bounds): Likewise.
2323 * targhooks.h (default_chkp_bound_mode): Likewise.
2324 (default_setup_incoming_vararg_bounds): Likewise.
2325 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2326 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2327 (have_whole_vector_shift): Likewise.
2328 * tree-vect-stmts.c (vectorizable_load): Likewise.
2329 * doc/tm.texi: Regenerate.
2331 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2333 Graceful degrade if Binutils PR21472 is not available.
2336 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2337 .rodata in flash test fails.
2338 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2339 * confgure: Regenerate.
2340 * config.in: Regenerate.
2341 * config/avr/avr.c (avr_asm_named_section)
2342 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2343 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2344 (avr_asm_init_sections): Same.
2346 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2348 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2349 (fma<VH:mode>4_intrinsic): Likewise.
2350 (*fmsub<VCVTF:mode>4): Likewise.
2351 (*fmsub<VH:mode>4_intrinsic): Likewise.
2353 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2356 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2357 Don't depend on "optimize > 0".
2358 (out_movhi_r_mr, out_movqi_mr_r): Same.
2359 (out_movhi_mr_r, out_movqi_r_mr): Same.
2360 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2361 io_address_operand on "optimize > 0".
2363 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2365 * tree-loop-distribution.c: Add general explanantion on the pass.
2366 (generate_loops_for_partition): Mark distributed loop.
2367 (pg_add_dependence_edges): New parameter. Handle alias data
2368 dependence specially and record it in the parameter if asked.
2369 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2370 (init_partition_graph_vertices, add_partition_graph_edge): New.
2371 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2372 (free_partition_graph_vdata, build_partition_graph): New.
2373 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2374 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2375 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2376 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2377 (version_for_distribution_p, finalize_partitions): New.
2378 (distribute_loop): Handle alias data dependence specially. Factor
2379 out loop fusion code as functions and call these functions.
2381 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2383 * tree-loop-distribution.c (classify_partition): New parameter and
2384 better handle reduction statement.
2385 (rdg_build_partitions): Revise comment.
2386 (distribute_loop): Compute statements in all partitions and pass it
2387 to classify_partition.
2389 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2391 * tree-loop-distribution.c (enum partition_type): New.
2392 (struct partition): New field type.
2393 (partition_merge_into): Add parameter. Update partition type.
2394 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2395 (build_rdg_partition_for_vertex): Compute partition type.
2396 (rdg_build_partitions): Dump partition type.
2397 (distribute_loop): Update calls to partition_merge_into.
2399 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2401 * tree-loop-distribution.c (struct ddr_hasher): New.
2402 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2404 (classify_partition): Call get_data_dependence.
2405 (pg_add_dependence_edges): Ditto.
2406 (distribute_loop): Release data dependence hash table.
2408 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2410 * tree-loop-distribution.c (ref_base_address): Delete.
2411 (similar_memory_accesses): Rename ...
2412 (share_memory_accesses): ... to this. Check if partitions access
2413 the same memory reference.
2414 (distribute_loop): Call share_memory_accesses.
2416 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2418 * tree-loop-distribution.c (struct partition): New field recording
2420 (partition_alloc, partition_free): Init and release data refs.
2421 (partition_merge_into): Merge data refs.
2422 (build_rdg_partition_for_vertex): Collect data refs for partition.
2423 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2425 (distribute_loop): Remve data refs from vertice data of partition
2428 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2430 * tree-loop-distribution.c (params.h): Include header file.
2431 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2432 (datarefs_vec): New global var.
2433 (create_rdg_vertices): Use datarefs_vec directly.
2434 (free_rdg): Don't free data references.
2435 (build_rdg): Update use. Don't free data references.
2436 (distribute_loop): Compute global variable for data references.
2437 Bail out if there are too many data references.
2439 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2441 * tree-loop-distribution.c (loop_nest): New global var.
2442 (build_rdg): Use loop directly, rather than loop nest.
2443 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2445 (distribute_loop): Compute global variable loop nest. Update use.
2447 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2449 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2450 (partition_merge_into): New parameter. Dump reason for fusion.
2451 (distribute_loop): Update use of partition_merge_into.
2453 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2455 * tree-loop-distribution.c (bb_top_order_index): New.
2456 (bb_top_order_index_size, bb_top_order_cmp): New.
2457 (stmts_from_loop): Use topological order.
2458 (pass_loop_distribution::execute): Compute and release topological
2459 order for basic blocks.
2461 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2463 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2466 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2468 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2469 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2470 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2471 * internal-fn.def (LOOP_DIST_ALIAS): New.
2472 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2473 (fold_loop_internal_call): ... this.
2474 (vect_loop_dist_alias_call): New function.
2475 (set_uid_loop_bbs): Call fold_loop_internal_call.
2476 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2479 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2482 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2483 Require dead FLAGS_REG at the beginning of a peephole.
2485 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2488 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2489 arguments in the call to __builtin_ia32_sbb_u32.
2490 (_subborrow_u64): Swap _X and _Y arguments in the call to
2491 __builtin_ia32_sbb_u64.
2493 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2496 * tree-vrp.c (compare_assert_loc): Turn into a function template
2497 with stable template parameter. Only test if a->e is NULL,
2498 !a->e == !b->e has been verified already. Use e == NULL or
2499 e != NULL instead of e or ! e tests. If stable is true, don't use
2500 iterative_hash_expr, on the other side allow a or b or both NULL
2501 and sort the NULLs last.
2502 (process_assert_insertions): Sort using compare_assert_loc<false>
2503 instead of compare_assert_loc, later sort using
2504 compare_assert_loc<true> before calling process_assert_insertions_for
2505 in a loop. Use break instead of continue once seen NULL pointer.
2507 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2509 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2510 Cortex-R7 and Cortex-R8 processors.
2512 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2514 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2515 uninitialized while src is not.
2517 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2519 * common/config/arm/arm-common.c: Adjust include path for
2521 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2522 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2523 (arm-cpu-data.h): Likewise.
2524 (arm-cpu-cdata.h): Likewise.
2525 * config/arm/arm-cpu.h: Delete.
2526 * config/arm/arm-cpu-cdata.h: Delete.
2527 * config/arm/arm-cpu-data.h: Delete.
2529 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2531 * config/arm/arm-cpus.in (cortex-a55): New.
2532 (cortex-a75): Likewise.
2533 (cortex-a75.cortex-a55): Likewise.
2534 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2536 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2537 * config/arm/arm-cpu-cdata.h: Regenerate.
2538 * config/arm/arm-cpu-data.h: Regenerate.
2539 * config/arm/arm-cpu.h: Regenerate.
2540 * config/arm/arm-tables.opt: Regenerate.
2541 * config/arm/arm-tune.md: Regenerate.
2543 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2545 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2547 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2549 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2552 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2554 PR tree-optimization/81292
2555 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2556 full_string_p, also call adjust_related_strinfos if the adjustment
2557 is simple, otherwise invalidate related strinfos.
2559 2017-07-04 Martin Liska <mliska@suse.cz>
2562 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2563 newly created variable as DECL_IGNORED_P.
2565 2017-07-04 Martin Liska <mliska@suse.cz>
2568 * ipa-inline.c (inline_small_functions):
2569 Use xstrdup_for_dump.
2571 2017-07-04 Tom de Vries <tom@codesourcery.com>
2573 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2575 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2578 * config/darwin.c (darwin_function_switched_text_sections):
2581 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2583 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2585 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2587 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2589 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2591 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2592 min_profitable_iters, and th as inclusive lower bounds.
2593 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2594 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2595 for min_profitable_iters and min_profitable_estimate.
2596 (vect_transform_loop): Treat th as an inclusive lower bound.
2597 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2599 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2602 * config/darwin.c (darwin_function_switched_text_sections):
2603 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2604 in two pieces, and suppress the use of buf.
2606 2017-07-03 Nathan Sidwell <nathan@acm.org>
2608 * hash-table.h (hash_table_mod1): Fix indentation.
2610 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2613 * predict.c (force_edge_cold): Be more careful about propagation
2615 * profile-count.h (profile_probability::guessed,
2616 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2618 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2620 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2622 * doc/invoke.texi (rcpc architecture extension): Document it.
2624 2017-07-03 Richard Biener <rguenther@suse.de>
2626 PR tree-optimization/60510
2627 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2628 the scalar reduction PHI and use it.
2629 (vectorizable_reduction): Properly guard the single_defuse_cycle
2630 path for non-SLP reduction chains where we cannot use it.
2631 Rework reduc_def/index and vector type deduction. Rework
2632 vector operand gathering during reduction op code-gen.
2633 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2634 chains dissolve the chain and leave it to non-SLP reduction
2637 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2639 * tree-data-ref.h (dr_alignment): Declare.
2640 * tree-data-ref.c (dr_alignment): New function.
2641 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2642 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2644 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2646 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2648 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2649 and base_misalignment fields.
2650 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2651 * tree-data-ref.c: Include builtins.h.
2652 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2653 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2654 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2655 * tree-vect-data-refs.c: Include tree-cfg.h.
2656 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2657 fields instead of calculating an alignment here.
2658 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2659 innermost_loop_behavior fields.
2661 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2663 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2665 (DR_STEP_ALIGNMENT): New macro.
2666 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2667 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2668 (create_data_ref): Print it.
2669 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2670 to tell whether the step preserves vector (mis)alignment.
2671 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2672 Move the check for an integer step and generalise to all INTEGER_CST.
2673 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2674 Print the outer step alignment.
2676 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2678 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2679 with offset_alignment.
2680 (DR_ALIGNED_TO): Delete.
2681 (DR_OFFSET_ALIGNMENT): New macro.
2682 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2683 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2684 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2685 (create_data_ref): Likewise.
2686 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2687 (vect_analyze_data_refs): Likewise.
2688 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2689 creating dummy innermost behavior.
2691 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2693 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2694 with a "innermost_loop_behavior *" and refeence tree.
2695 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2696 (create_data_ref): Update call accordingly.
2697 * tree-predcom.c (find_looparound_phi): Likewise.
2699 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2701 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2702 fields with dr_wrt_vec_loop.
2703 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2704 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2705 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2706 (vect_dr_behavior): New function.
2707 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2708 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2709 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2710 track whether the step preserves the misalignment.
2711 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2712 Use vect_dr_behavior.
2713 (vect_setup_realignment): Update call accordingly.
2714 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2715 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2716 call to vect_create_addr_base_for_vector_ref.
2717 (vect_create_cond_for_align_checks): Likewise.
2718 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2719 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2720 (vect_recog_mask_conversion_pattern): Likewise.
2721 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2722 (new_stmt_vec_info): Remove redundant zeroing.
2724 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2726 * common/config/arm/arm-common.c (arm_be8_option): New function.
2727 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2728 (ISA_ARMv6): Add isa_bit_be8.
2729 * config/arm/arm.h (arm_be8_option): Add prototype.
2730 (BE8_SPEC_FUNCTION): New define.
2731 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2732 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2733 (mlittle-endian): Similarly.
2734 (mbe8, mbe32): New options.
2735 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2736 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2738 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2740 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2742 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2744 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2745 (cleanup_tree_cfg_bb): Use it.
2746 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2748 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2750 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2753 * loop-doloop.c (add_test): Update profile.
2755 2017-07-03 Martin Liska <mliska@suse.cz>
2758 * sanopt.c (rewrite_usage_of_param): New function.
2759 (sanitize_rewrite_addressable_params): Likewise.
2760 (pass_sanopt::execute): Call rewrite_usage_of_param.
2762 2017-07-03 Richard Biener <rguenther@suse.de>
2764 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2765 back to using VIEW_CONVERT_EXPR.
2767 2017-07-03 Martin Liska <mliska@suse.cz>
2770 * doc/extend.texi: Document when a resolver function is
2771 generated for target_clones.
2773 2017-07-03 Martin Liska <mliska@suse.cz>
2775 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2776 and shadow memory of auto variables which are subject of
2777 use-after-scope sanitization.
2778 (asan_expand_mark_ifn): Add do set only when is_poison.
2780 2016-07-03 Richard Biener <rguenther@suse.de>
2782 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2784 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2785 (vectorizable_reduction): Perform reduction PHI creation when
2786 visiting a reduction PHI and adjust and simplify code generation
2787 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2788 (vect_transform_loop): Visit reduction PHIs.
2789 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2790 defs into the SLP tree.
2791 (vect_build_slp_tree): Reduction defs terminate the recursion.
2792 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2794 (vect_get_vec_defs_for_stmt_copy): Export.
2795 (vect_get_vec_defs): Likewise.
2796 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2798 (vect_get_vec_defs_for_stmt_copy): Declare.
2799 (vect_get_vec_defs): Likewise.
2801 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2803 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2804 parameter with a "loop" parameter and use it instead of the
2805 loop containing DR_STMT. Don't check simple_iv when doing
2806 BB analysis. Describe the two analysis modes in the comment.
2808 2017-07-03 Tom de Vries <tom@codesourcery.com>
2810 PR tree-optimization/69468
2811 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2812 (find_same_succ_bb): Handle ignore_edge_flags.
2814 2017-07-03 Tom de Vries <tom@codesourcery.com>
2816 PR tree-optimization/81192
2817 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2819 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2821 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2823 2017-07-03 Tom de Vries <tom@codesourcery.com>
2825 PR tree-optimization/81192
2826 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2827 BB_SAME_SUCC (bb) == NULL.
2829 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2831 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2834 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2836 * dumpfile.c: Include profile-count.h
2837 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2839 (insert_cond_bb): Update profile.
2840 * tree-cfg.h (insert_cond_bb): Update prototype.
2841 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2842 * tree-dump.c: Do not include tree-cfg.
2844 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2846 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2848 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2850 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2853 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2855 * tree-complex.c (expand_complex_div_wide): update profile.
2857 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2858 Alan Hayward <alan.hayward@arm.com>
2859 David Sherwood <david.sherwood@arm.com>
2861 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2862 (CORETYPES_H): New define.
2863 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2864 (insn-modes-inline.h, s-modes-inline-h): New rules.
2865 (generated_files): Add insn-modes-inline.h.
2866 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2867 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2868 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2869 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2870 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2871 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2872 (build/gencodes.o, build/genconditions.o): Likewise.
2873 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
2874 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
2875 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
2876 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
2877 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
2878 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
2879 * coretypes.h: Include everything up to real.h for generators.
2880 Include insn-modes.h first. Include wide-int-print.h after
2881 wide-int.h. Include insn-modes-inline.h and then machmode.h.
2882 * machmode.h: Don't include insn-modes.h here.
2883 * function-tests.c: Remove includes of signop.h, machmode.h,
2884 double-int.h and wide-int.h.
2886 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
2888 * optc-save-gen.awk: Likewise.
2889 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
2890 * godump.c: Remove include of wide-int-print.h.
2891 * pretty-print.h: Likewise.
2892 * wide-int-print.cc: Likewise.
2893 * wide-int.cc: Likewise.
2894 * hash-map-tests.c: Remove include of signop.h.
2895 * hash-set-tests.c: Likewise.
2896 * rtl-tests.c: Likewise.
2897 * mkconfig.sh: Remove include of machmode.h.
2898 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
2900 (emit_insn_modes_inline_h): ...this new function. Emit the code
2901 into an insn-modes-inline.h header file, adding appropriate
2902 include guards and end comments.
2903 (emit_insn_modes_c_header): Remove include of machmode.h.
2904 (emit_min_insn_modes_c_header): Include coretypes.h rather than
2906 (main): Handle -i flag and call emit_insn_modes_inline_h when
2909 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2911 * tree-ssa-strlen.c (strinfo): Rename the length field to
2912 nonzero_chars. Add a full_string_p field.
2913 (compare_nonzero_chars, zero_length_string_p): New functions.
2914 (get_addr_stridx): Add an offset_out parameter.
2915 Use compare_nonzero_chars.
2916 (get_stridx): Update accordingly. Use compare_nonzero_chars.
2917 (new_strinfo): Update after above changes to strinfo.
2918 (set_endptr_and_length): Set full_string_p.
2919 (get_string_length): Update after above changes to strinfo.
2920 (unshare_strinfo): Update call to new_strinfo.
2921 (maybe_invalidate): Likewise.
2922 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
2923 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
2924 as a uhwi instead of an shwi. Update after above changes to
2925 strinfo and new_strinfo.
2926 (zero_length_string): Assert that chainsi contains full strings.
2927 Use zero_length_string_p. Update call to new_strinfo.
2928 (adjust_related_strinfos): Update after above changes to strinfo.
2929 Copy full_string_p from origsi.
2930 (adjust_last_stmt): Use zero_length_string_p.
2931 (handle_builtin_strlen): Update after above changes to strinfo and
2932 new_strinfo. Install the lhs as the string length if the previous
2933 entry didn't describe a full string.
2934 (handle_builtin_strchr): Update after above changes to strinfo
2936 (handle_builtin_strcpy): Likewise.
2937 (handle_builtin_strcat): Likewise.
2938 (handle_builtin_malloc): Likewise.
2939 (handle_pointer_plus): Likewise.
2940 (handle_builtin_memcpy): Likewise. Track nonzero characters
2941 that aren't necessarily followed by a nul terminator.
2942 (handle_char_store): Likewise.
2944 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2946 PR tree-optimization/80769
2947 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
2948 for malloc and calloc. Document the new invariant that all related
2949 strinfos have delayed lengths or none do.
2950 (verify_related_strinfos): Move earlier in file.
2951 (set_endptr_and_length): New function, split out from...
2952 (get_string_length): ...here. Also set the lengths of related
2954 (zero_length_string): Assert that chainsi has known (rather than
2956 (adjust_related_strinfos): Likewise.
2958 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2960 PR tree-optimization/81136
2961 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
2962 assert that two references with the same misalignment have the same
2963 compile-time misalignment if those compile-time misalignments
2966 2017-07-01 Andi Kleen <ak@linux.intel.com>
2968 * print-tree.c (print_node): Print all attributes.
2970 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2972 * cfg.c (scale_bbs_frequencies): New function.
2973 * cfg.h (scale_bbs_frequencies): Declare it.
2974 * cfgloopanal.c (single_likely_exit): Cleanup.
2975 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
2977 (scale_loop_profile): Likewise.
2978 (loop_version): Likewise.
2979 (create_empty_loop_on_edge): Update.
2980 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
2981 scale_loop_frequencies, scale_loop_profile, loopify,
2982 loop_version): Update prototypes.
2983 * modulo-sched.c (sms_schedule): Update.
2984 * predict.c (unlikely_executed_edge_p): Also check probability.
2985 (probably_never_executed_edge_p): Fix typo.
2986 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2987 * tree-parloops.c (gen_parallel_loop): Update.
2988 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
2989 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2990 * tree-ssa-loop-split.c (split_loop): Update.
2991 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2992 * tree-vect-loop-manip.c (vect_do_peeling): Update.
2993 (vect_loop_versioning): Update.
2994 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2996 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2998 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3000 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3002 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3003 to keep profile consistent.
3005 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3007 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3008 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3009 * profile-count.h (max_safe_multiplier): Make unsigned.
3010 (profile_count::guessed_zero): New.
3012 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3014 * bb-reorder.c (fix_up_crossing_landing_pad,
3015 fix_crossing_conditional_branches): Use make_single_succ_edge
3016 to keep profile consistent.
3018 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3020 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3023 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3026 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3027 the right scopes, make sure cond_jump isn't preserved between multiple
3028 iterations. Search for fallthru edge whenever there are 3+ edges and
3029 use find_fallthru_edge for it.
3031 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3033 Patch by Alexander Monakov <amonakov@ispras.ru>
3034 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3035 probabilities consistently.
3037 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3039 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3040 emit_cmp_and_jump_insns.
3042 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3045 * tree-inline.c (expand_call_inline): Combine profile statuses.
3047 2017-06-30 Andrew Pinski <apinski@cavium.com>
3049 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3050 fold_stmt returned true.
3052 2017-06-30 Nathan Sidwell <nathan@acm.org>
3054 * ggc.h (empty_string): Delete.
3055 * cfgexpand.c (expand_asm_stmt): Use plain "".
3056 * optabs.c (expand_asm_memory_barrier): Likewise.
3057 * stringpool.c (empty_string): Delete.
3058 (digit_vector, digit_string): Delete.
3059 (ggc_alloc_string): Use plain "", don't optimize single digit
3060 strings. Use ggc_alloc_atomic.
3062 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3064 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3065 comparison set and one other set, use the cost of the non-comparison
3068 2017-06-30 Nathan Sidwell <nathan@acm.org>
3070 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3073 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3075 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3076 loops. Remove now unneeded calls to gimple_switch_set_label() that
3077 just set removed labels to NULL_TREE.
3079 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3081 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3082 (set_range_info): ...here. Only call set_range_info_raw if domain
3084 (set_nonzero_bits): Call set_range_info_raw.
3085 * tree-ssanames.h (set_range_info_raw): New.
3087 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3090 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3091 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3092 of nonimmediate_operand and <store_mask_constraint> instead of m
3093 for the input operand. For V8FI iterator, always split if input
3094 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3095 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3096 <store_mask_predicate> instead of register_operand and
3097 <store_mask_constraint> instead of v for the input operand. Make
3098 sure both operands aren't MEMs for if not <mask_applied>.
3100 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3102 * lto-wrapper.c (copy_file) Close both file descriptors before
3105 2017-06-30 Martin Liska <mliska@suse.cz>
3108 * multiple_target.c (create_dispatcher_calls): Make ifunc
3109 also for function that don't have calls or are not referenced.
3111 2017-06-30 Richard Biener <rguenther@suse.de>
3113 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3114 analyze the first scalar stmt. Move vector type computation
3115 for the BB case here from ...
3116 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3117 live operation processing in the SLP case properly.
3119 2017-06-30 Richard Biener <rguenther@suse.de>
3121 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3123 2017-06-30 Martin Liska <mliska@suse.cz>
3126 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3127 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3129 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3131 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3134 2017-06-30 Richard Biener <rguenther@suse.de>
3136 PR tree-optimization/81249
3137 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3138 condition reduction result to original scalar type.
3140 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3142 * profile-count.h (enum profile_quality): Fix typos and whitespace
3145 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3147 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3148 type for branch probabilities.
3150 2017-06-29 Julian Brown <julian@codesourcery.com>
3151 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3153 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3154 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3155 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3156 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3158 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3160 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3161 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3162 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3163 CC usage from generic code to here.
3164 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3165 CC usage into the target macros.
3167 2017-06-29 Maya Rashish <coypu@sdf.org>
3169 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3172 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3174 * arm/arm-builtins.c: Include profile-count.h
3175 * except.c (sjlj_emit_function_enter): Use
3176 profile_probability::unlikely.
3178 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3180 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3181 and tocrel_offset be pointer args rather than implicitly using
3183 (legitimate_constant_pool_address_p, rs6000_emit_move,
3184 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3185 tocrel_offset and use in toc_relative_expr_p call.
3186 (print_operand, print_operand_address): Use static tocrel_base_oac
3187 and tocrel_offset_oac.
3188 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3191 2017-06-29 Maya Rashish <coypu@sdf.org>
3193 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3195 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3197 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3198 objects, take into account only the alignment of 'op0' and 'mode1' if
3201 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3203 * ccmp.c (ccmp_tree_comparison_p): New function.
3204 (ccmp_candidate_p): Update to use above function.
3205 (get_compare_parts): New function.
3206 (expand_ccmp_next): Update to use new functions.
3207 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3209 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3210 take mode as argument.
3211 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3212 * expr.c (expand_expr_real_1): Pass mode as argument.
3214 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3216 * combine.c (combine_instructions): Print insns to dump_file, together
3219 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3221 * asan.c (asan_emit_stack_protection): Update.
3222 (create_cond_insert_point): Update.
3223 * auto-profile.c (afdo_propagate_circuit): Update.
3224 * basic-block.h (struct edge_def): Turn probability to
3225 profile_probability.
3226 (EDGE_FREQUENCY): Update.
3227 * bb-reorder.c (find_traces_1_round): Update.
3228 (better_edge_p): Update.
3229 (sanitize_hot_paths): Update.
3230 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3231 (make_single_succ_edge): Update.
3232 (check_bb_profile): Update.
3233 (dump_edge_info): Update.
3234 (update_bb_profile_for_threading): Update.
3235 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3236 probabilitycount to 0.
3237 * cfgbuild.c (compute_outgoing_frequencies): Update.
3238 * cfgcleanup.c (try_forward_edges): Update.
3239 (outgoing_edges_match): Update.
3240 (try_crossjump_to_edge): Update.
3241 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3242 (expand_gimple_tailcall): Update.
3243 (construct_init_block): Use make_single_succ_edge.
3244 (construct_exit_block): Use make_single_succ_edge.
3245 * cfghooks.c (verify_flow_info): Update.
3246 (redirect_edge_succ_nodup): Update.
3247 (split_edge): Update.
3248 (account_profile_record): Update.
3249 * cfgloopanal.c (single_likely_exit): Update.
3250 * cfgloopmanip.c (scale_loop_profile): Update.
3251 (set_zero_probability): Remove.
3252 (duplicate_loop_to_header_edge): Update.
3253 * cfgloopmanip.h (loop_version): Update prototype.
3254 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3255 (force_nonfallthru_and_redirect): Update.
3256 (update_br_prob_note): Update.
3257 (rtl_verify_edges): Update.
3258 (purge_dead_edges): Update.
3259 (rtl_lv_add_condition_to_bb): Update.
3260 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3261 * cgraphunit.c (init_lowered_empty_function): Update.
3262 (cgraph_node::expand_thunk): Update.
3263 * cilk-common.c: Include profile-count.h
3264 * dojump.c (inv): Remove.
3265 (jumpifnot): Update.
3266 (jumpifnot_1): Update.
3267 (do_jump_1): Update.
3269 (do_jump_by_parts_greater_rtx): Update.
3270 (do_compare_rtx_and_jump): Update.
3271 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3272 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3273 * dwarf2cfi.c: Include profile-count.h
3274 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3275 (sjlj_emit_dispatch_table): Likewise.
3276 * explow.c: Include profile-count.h
3277 * expmed.c (emit_store_flag_force): Update.
3278 (do_cmp_and_jump): Update.
3279 * expr.c (compare_by_pieces_d::generate): Update.
3280 (compare_by_pieces_d::finish_mode): Update.
3281 (emit_block_move_via_loop): Update.
3282 (store_expr_with_bounds): Update.
3283 (store_constructor): Update.
3284 (expand_expr_real_2): Update.
3285 (expand_expr_real_1): Update.
3286 * expr.h (try_casesi, try_tablejump): Update prototypes.
3287 * gimple-pretty-print.c (dump_probability): Update.
3288 (dump_profile): New.
3289 (dump_gimple_label): Update.
3290 (dump_gimple_bb_header): Update.
3291 * graph.c (draw_cfg_node_succ_edges): Update.
3292 * hsa-gen.c (convert_switch_statements): Update.
3293 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3294 (find_if_case_1): Update.
3295 (find_if_case_2): Update.
3296 * internal-fn.c (expand_arith_overflow_result_store): Update.
3297 (expand_addsub_overflow): Update.
3298 (expand_neg_overflow): Update.
3299 (expand_mul_overflow): Update.
3300 (expand_vector_ubsan_overflow): Update.
3301 * ipa-cp.c (good_cloning_opportunity_p): Update.
3302 * ipa-split.c (split_function): Use make_single_succ_edge.
3303 * ipa-utils.c (ipa_merge_profiles): Update.
3304 * loop-doloop.c (add_test): Update.
3305 (doloop_modify): Update.
3306 * loop-unroll.c (compare_and_jump_seq): Update.
3307 (unroll_loop_runtime_iterations): Update.
3308 * lra-constraints.c (lra_inheritance): Update.
3309 * lto-streamer-in.c (input_cfg): Update.
3310 * lto-streamer-out.c (output_cfg): Update.
3311 * mcf.c (adjust_cfg_counts): Update.
3312 * modulo-sched.c (sms_schedule): Update.
3313 * omp-expand.c (expand_omp_for_init_counts): Update.
3314 (extract_omp_for_update_vars): Update.
3315 (expand_omp_ordered_sink): Update.
3316 (expand_omp_for_ordered_loops): Update.
3317 (expand_omp_for_generic): Update.
3318 (expand_omp_for_static_nochunk): Update.
3319 (expand_omp_for_static_chunk): Update.
3320 (expand_cilk_for): Update.
3321 (expand_omp_simd): Update.
3322 (expand_omp_taskloop_for_outer): Update.
3323 (expand_omp_taskloop_for_inner): Update.
3324 * omp-simd-clone.c (simd_clone_adjust): Update.
3325 * optabs.c (expand_doubleword_shift): Update.
3326 (expand_abs): Update.
3327 (emit_cmp_and_jump_insn_1): Update.
3328 (expand_compare_and_swap_loop): Update.
3329 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3330 * predict.c (predictable_edge_p): Update.
3331 (edge_probability_reliable_p): Update.
3332 (set_even_probabilities): Update.
3333 (combine_predictions_for_insn): Update.
3334 (combine_predictions_for_bb): Update.
3335 (propagate_freq): Update.
3336 (estimate_bb_frequencies): Update.
3337 (force_edge_cold): Update.
3338 * profile-count.c (profile_count::dump): Add missing space into dump.
3339 (profile_count::debug): Add newline.
3340 (profile_count::differs_from_p): Explicitly convert to unsigned.
3341 (profile_count::stream_in): Update.
3342 (profile_probability::dump): New member function.
3343 (profile_probability::debug): New member function.
3344 (profile_probability::differs_from_p): New member function.
3345 (profile_probability::differs_lot_from_p): New member function.
3346 (profile_probability::stream_in): New member function.
3347 (profile_probability::stream_out): New member function.
3348 * profile-count.h (profile_count_quality): Rename to ...
3349 (profile_quality): ... this one.
3350 (profile_probability): New.
3351 (profile_count): Update.
3352 * profile.c (compute_branch_probabilities): Update.
3353 * recog.c (peep2_attempt): Update.
3354 * sched-ebb.c (schedule_ebbs): Update.
3355 * sched-rgn.c (find_single_block_region): Update.
3356 (compute_dom_prob_ps): Update.
3357 (schedule_region): Update.
3358 * sel-sched-ir.c (compute_succs_info): Update.
3359 * stmt.c (struct case_node): Update.
3360 (do_jump_if_equal): Update.
3361 (get_outgoing_edge_probs): Update.
3362 (conditional_probability): Update.
3363 (emit_case_dispatch_table): Update.
3364 (expand_case): Update.
3365 (expand_sjlj_dispatch_table): Update.
3366 (emit_case_nodes): Update.
3367 * targhooks.c: Update.
3368 * tracer.c (better_p): Update.
3369 (find_best_successor): Update.
3370 * trans-mem.c (expand_transaction): Update.
3371 * tree-call-cdce.c: Update.
3372 * tree-cfg.c (gimple_split_edge): Upate.
3373 (move_sese_region_to_fn): Upate.
3374 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3375 * tree-eh.c (lower_resx): Upate.
3376 (cleanup_empty_eh_move_lp): Upate.
3377 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3378 * tree-inline.c (copy_edges_for_bb): Update.
3379 (copy_cfg_body): Update.
3380 * tree-parloops.c (gen_parallel_loop): Update.
3381 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3382 (gimple_gen_time_profiler): Update.
3383 * tree-ssa-dce.c (remove_dead_stmt): Update.
3384 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3385 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3386 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3387 (unloop_loops): Update.
3388 (try_peel_loop): Update.
3389 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3390 * tree-ssa-loop-split.c (connect_loops): Update.
3391 (split_loop): Update.
3392 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3393 (hoist_guard): Update.
3394 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3395 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3396 (value_replacement): Update.
3397 * tree-ssa-reassoc.c (branch_fixup): Update.
3398 * tree-ssa-tail-merge.c (replace_block_by): Update.
3399 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3400 (create_edge_and_update_destination_phis): Update.
3401 (compute_path_counts): Update.
3402 (recompute_probabilities): Update.
3403 (update_joiner_offpath_counts): Update.
3404 (freqs_to_counts_path): Update.
3405 (duplicate_thread_path): Update.
3406 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3407 (struct switch_conv_info): Update.
3408 (gen_inbound_check): Update.
3409 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3410 (vect_do_peeling): Update.
3411 (vect_loop_versioning): Update.
3412 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3413 (optimize_mask_stores): Update.
3414 * ubsan.c (ubsan_expand_null_ifn): Update.
3415 * value-prof.c (gimple_divmod_fixed_value): Update.
3416 (gimple_divmod_fixed_value_transform): Update.
3417 (gimple_mod_pow2): Update.
3418 (gimple_mod_pow2_value_transform): Update.
3419 (gimple_mod_subtract): Update.
3420 (gimple_mod_subtract_transform): Update.
3421 (gimple_ic): Update.
3422 (gimple_stringop_fixed_value): Update.
3423 (gimple_stringops_transform): Update.
3424 * value-prof.h: Update.
3426 2017-06-29 Carl Love <cel@us.ibm.com>
3428 * config/rs6000/rs6000-c.c: Add support for built-in functions
3429 vector signed int vec_signed (vector float);
3430 vector signed long long vec_signed (vector double);
3431 vector signed int vec_signed2 (vector double, vector double);
3432 vector signed int vec_signede (vector double);
3433 vector signed int vec_signedo (vector double);
3434 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3435 instruction generator.
3436 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3437 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3438 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3440 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3441 vunsignede_v2df): Add define_expands.
3442 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3443 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3444 VEC_UNSIGNEDO): Add definitions.
3445 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3446 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3447 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3448 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3449 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3450 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3451 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3452 vec_unsignede, vec_unsignedo): Add builtin defines.
3453 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3455 * doc/extend.texi: Update the built-in documentation file for the
3456 new built-in functions.
3458 2017-06-29 Richard Biener <rguenther@suse.de>
3460 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3461 reduction chains to LOOP_VINFO_REDUCTIONS.
3462 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3463 SLP reductions after processing reduction chains.
3465 2017-06-29 Nathan Sidwell <nathan@acm.org>
3467 * builtins.c (fold_builtin_FUNCTION): Use
3468 lang_hooks.decl_printable_name.
3470 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3473 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3474 with only one label.
3475 * stmt.c (expand_case): Assert NCASES is greater than one.
3477 2017-06-29 Richard Biener <rguenther@suse.de>
3479 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3481 (group_case_labels): Likewise.
3482 (find_taken_edge): Push sanity checking on val to workers...
3483 (find_taken_edge_cond_expr): ... here
3484 (find_taken_edge_switch_expr): ... and here, handle cases
3485 with just a default label.
3486 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3487 (group_case_labels): Likewise.
3488 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3489 group_case_labels does anything cleanup the CFG again.
3491 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3493 PR tree-optimization/81196
3494 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3495 exit condition comparing two IVs.
3497 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3499 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3500 profile to the dummy entry at the end of the list of architectures.
3501 * config/arm/arm-cpu-cdata.h: Regenerated.
3503 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3504 Michael Collison <michael.collison@arm.com>
3507 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3509 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3510 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3511 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3512 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3513 of shift when the shift amount is masked with constant equal to
3514 the size of the mode.
3515 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3518 2017-06-29 Martin Liska <mliska@suse.cz>
3520 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3522 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3525 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3528 (edge_growth_cache_entry::edge_growth_cache_entry): New
3530 (reset_edge_growth_cache): Update to use constructor.
3532 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3534 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3535 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3536 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3538 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3540 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3541 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3543 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3545 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3546 (*-linux-uclibc*): Add t-uclibc tmake_file.
3547 * config/t-musl: New.
3548 * config/t-uclibc: New.
3550 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3552 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3554 (gen_comm_data): Emit architectural setting of arch_prof.
3555 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3557 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3558 (armv8-m.base, armv8-m.main): Likewise.
3559 * arm-protos.h (arm_build_target): Add profile field.
3560 (arch_option): Likewise.
3561 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3563 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3564 arm_active_target.profile.
3566 2017-06-28 Richard Biener <rguenther@suse.de>
3569 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3570 TYPE_OVERFLOW_WRAPS.
3571 * match.pd (negate_expr_p): Likewise.
3572 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3573 fold_build2, not fold_binary.
3575 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3577 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3578 Convert memory address to Pmode.
3579 (aarch64_print_operand): Assert MEM operands are always Pmode.
3581 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3584 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3585 Remove redundant if.
3586 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3587 * config/arm/aarch-common-protos.h
3588 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3589 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3591 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3594 * multiple_target.c (create_dispatcher_calls): Set the default
3595 clone to be static, not public.
3597 2017-06-28 Richard Biener <rguenther@suse.de>
3599 * tree-vect-loop.c (vectorizable_reduction): Move special
3600 cond reduction IV var creation ...
3601 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3602 parameter. Use STMT_VINFO_VECTYPE.
3603 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3606 2017-06-28 Martin Liska <mliska@suse.cz>
3609 * ipa-visibility.c (non_local_p): Handle visibility.
3611 2017-06-28 Martin Liska <mliska@suse.cz>
3614 * common.opt: Add IntegerRange to various options.
3615 * opt-functions.awk (integer_range_info): New function.
3616 * optc-gen.awk: Add integer_range_info to cl_options struct.
3617 * opts-common.c (decode_cmdline_option): Handle
3618 CL_ERR_INT_RANGE_ARG.
3619 (cmdline_handle_error): Likewise.
3620 * opts.c (print_filtered_help): Show valid interval in
3621 when --help is provided.
3622 * opts.h (struct cl_option): Add range_min and range_max fields.
3623 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3625 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3627 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3628 (x * C EQ/NE y * C): New transformation.
3630 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3632 * genmultilib (combination_space): Accept '+' in option names.
3634 2017-06-28 Martin Liska <mliska@suse.cz>
3637 * asan.c (instrument_derefs): Bail out inner references
3638 that are hard register variables.
3640 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3643 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3644 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3646 2017-06-28 Richard Biener <rguenther@suse.de>
3648 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3649 (vect_get_slp_defs): Adjust.
3650 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3652 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3654 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3655 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3656 (vectorizable_reduction): Adjust.
3657 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3659 (vect_get_slp_defs): Likewise.
3660 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3661 (vectorizable_bswap): Adjust.
3662 (vectorizable_call): Likewise.
3663 (vectorizable_conversion): Likewise.
3664 (vectorizable_assignment): Likewise.
3665 (vectorizable_shift): Likewise.
3666 (vectorizable_operation): Likewise.
3667 (vectorizable_store): Likewise.
3668 (vectorizable_condition): Likewise.
3669 (vectorizable_comparison): Likewise.
3671 2017-06-28 Michael Collison <michael.collison@arm.com>
3674 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3676 (arm_gen_movmemqi): Removed unused variable 'i'.
3677 Convert 'for' loop into 'while' loop.
3678 (arm_expand_prologue): Remove last unnecessary set of insn.
3679 (thumb_pop): Remove unused variable 'pushed_words'.
3680 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3682 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3684 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3685 * config/s390/s390-protos.h: Add prototype of
3686 s390_rel_address_ok_p.
3687 * config/s390/s390.c (s390_got_symbol): New function.
3688 (s390_rel_address_ok_p): New function.
3689 (legitimize_pic_address): Use s390_rel_address_ok_p.
3690 (s390_load_got): Use s390_got_symbol.
3691 (s390_option_override): Issue error if
3692 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3693 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3695 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3697 2017-06-27 Andrew Pinski <apinski@cavium.com>
3699 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3700 (X * copysign (1.0, X)): New pattern.
3701 (X * copysign (1.0, -X)): New pattern.
3702 (copysign (-1.0, CST)): New pattern.
3704 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3706 * genmultilib (combination_space): Remove variable.
3707 Validate reuse rules against regular expression for any sequence
3708 of multilib options in any order.
3710 2017-06-27 Michael Collison <michael.collison@arm.com>
3712 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3713 call aarch64_split_simd_combine.
3714 * (aarch64_combine_internal<mode>): Delete pattern.
3715 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3716 Allow register and subreg operands.
3718 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3720 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3721 specific need, just fallback on defaults.
3722 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3724 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3725 Olivier Hainque <hainque@adacore.com>
3727 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3729 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3730 targets. Pick a default if no particular attempt applied.
3731 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3734 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3736 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3737 (x86_64-wrs-vxworks7): Likewise.
3739 2017-06-27 Marek Polacek <polacek@redhat.com>
3742 * ubsan.c (instrument_null): Check get_base_address's result for null.
3744 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3746 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3748 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3750 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3751 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3752 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3753 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3755 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3756 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3757 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3758 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3759 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3760 BUILT_IN_FEUPDATEENV): New builtins.
3761 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3762 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3763 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3764 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3766 (builtin_structptr_types): Adjust size.
3767 * tree.c (builtin_structptr_types): Add four entries.
3769 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3770 Olivier Hainque <hainque@adacore.com>
3772 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3773 (TLS_SYM): New local macro, forcing reference to __tls__ on
3774 link command lines for VxWorks 7 RTPs, triggering initialization
3776 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3777 OS features TLS support, true for RTPs on VxWorks 7.
3778 * config/vxworks.c (vxworks_override_options): Setup emutls
3781 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3783 * predict.c (test_prediction_value_range): Use -1U instead of -1
3784 to avoid narrowing conversion warning.
3785 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3786 to avoid narrowing conversion warning.
3787 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3789 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3791 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3793 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3794 64bit configurations.
3795 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3796 (SIZE_TYPE): Likewise.
3797 * config/vxworks.c (vxworks_emutls_var_fields): Use
3798 long_unsigned_type_node instead of unsigned_type_node as the offset
3799 field type, which is "pointer" mode in emutls.c.
3801 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3804 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3807 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3808 gimple_vuse copying separately from gimple_vdef copying.
3810 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3812 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3815 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3816 Olivier Hainque <hainque@adacore.com>
3818 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3819 for all vxworks7 targets.
3820 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3821 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3822 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3823 variations for VX6/VX7 and 32/64bits later on in ...
3824 (VXWORKS_LIB_SPEC): Leverage new macros.
3825 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3826 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3828 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3830 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3831 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3833 2017-06-26 Carl Love <cel@us.ibm.com>
3835 * config/rs6000/rs6000-c.c: Add support for built-in functions
3836 vector bool char vec_reve (vector bool char);
3837 vector signed char vec_reve (vector signed char);
3838 vector unsigned char vec_reve (vector unsigned char);
3839 vector bool int vec_reve (vector bool int);
3840 vector signed int vec_reve (vector signed int);
3841 vector unsigned int vec_reve (vector unsigned int);
3842 vector bool long long vec_reve (vector bool long long);
3843 vector signed long long vec_reve (vector signed long long);
3844 vector unsigned long long vec_reve (vector unsigned long long);
3845 vector bool short vec_reve (vector bool short);
3846 vector signed short vec_reve (vector signed short);
3847 vector double vec_reve (vector double);
3848 vector float vec_reve (vector float);
3849 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3850 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3851 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3852 (altivec_vreve): New pattern.
3853 * config/rs6000/altivec.h (vec_reve): New define.
3854 * doc/extend.texi (vec_rev): Update the built-in documentation file
3855 for the new built-in functions.
3857 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3859 PR tree-optimization/71815
3860 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3862 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3864 (slsr_process_phi): Likewise.
3865 (replace_uncond_cands_and_profitable_phis): Don't replace a
3866 multiply candidate with a stride of 1 (copy or cast).
3867 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3869 (lowest_cost_path): Likewise.
3870 (total_savings): Likewise.
3872 2017-06-26 Richard Biener <rguenther@suse.de>
3875 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
3876 Use def_builtin_pure for all gather builtins.
3878 2017-06-26 Richard Biener <rguenther@suse.de>
3880 PR tree-optimization/81203
3881 * tree-tailcall.c (find_tail_calls): Do not move stmts into
3884 2017-06-26 Marek Polacek <polacek@redhat.com>
3887 * doc/invoke.texi: Document -Wmultistatement-macros.
3889 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
3891 * doc/sourcebuild.texi (ARM-specific attributes): Document new
3892 arm_neon_ok_no_float_abi effective target.
3894 2017-06-26 Richard Biener <rguenther@suse.de>
3896 PR tree-optimization/80928
3897 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
3898 (copy_bbs): Set BB_DUPLICATED flag early.
3899 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
3901 (execute_on_shrinking_pred): Likewise.
3902 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
3903 BB_DUPLICATED blocks.
3904 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
3905 iterate over all PHIs considering removal of *gsi.
3907 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
3909 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
3912 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3914 * config/rs6000/rs6000-string.c: (expand_block_clear,
3915 do_load_for_compare, select_block_compare_mode,
3916 compute_current_alignment, expand_block_compare,
3917 expand_strncmp_align_check, expand_strn_compare,
3918 expand_block_move, rs6000_output_load_multiple)
3919 Move functions related to string/block move/compare
3921 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
3922 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
3923 for this function which is now used in two files.
3924 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
3925 * config.gcc: Add rs6000-string.o to extra_objs for
3926 targets powerpc*-*-* and rs6000*-*-*.
3928 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3931 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
3932 32-bit, since indexed is not valid for DImode.
3933 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
3934 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
3935 (define_peephole2 for Altivec d-form load): Add 32-bit support.
3936 (define_peephole2 for Altivec d-form store): Likewise.
3939 * multiple_target.c (create_dispatcher_calls): Only create the
3940 dispatcher call if the function is the default clone of a
3943 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
3946 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
3947 a call, force the call to not be a tail call.
3949 2017-06-23 Jeff Law <law@redhat.com>
3951 * doc/contrib.texi: Add entry for Steven Pemberton's work on
3954 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
3956 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
3957 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
3958 handling for early expansion of vector shifts (sl,sr,sra,rl).
3959 (builtin_function_type): Add vector shift right instructions
3960 to the unsigned argument list.
3962 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3964 rtl-optimizatoin/79286
3965 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
3966 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
3967 trap. PIC register plus a const unspec without offset can never trap.
3969 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
3971 * tree.h (builtin_structptr_type): New type.
3972 (builtin_structptr_types): Declare new array.
3973 * tree.c (builtin_structptr_types): New array.
3974 (free_lang_data, build_common_tree_nodes): Use it.
3976 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
3979 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
3982 2017-06-22 Matt Turner <mattst88@gmail.com>
3984 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
3985 Lake models to skylake case. Assume skylake for unknown
3986 models with clflushopt.
3988 2017-06-22 Jeff Law <law@redhat.com>
3990 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
3991 frame sizes that do not satisfy aarch64_uimm12_shift.
3993 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
3995 * profile-count.h (apply_probability,
3996 apply_scale, probability_in): Fix checks for zero.
3998 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4000 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4001 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4003 2016-06-22 Richard Biener <rguenther@suse.de>
4005 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4006 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4007 REDUC_MAX_EXPR support.
4008 (vectorizable_reduction): Likewise.
4009 (vect_create_epilog_for_reduction): Likewise.
4011 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4013 * match.pd (A / (1 << B) -> A >> B): New.
4014 * generic-match-head.c: Include optabs-tree.h.
4015 * gimple-match-head.c: Likewise.
4016 * optabs-tree.h (target_supports_op_p): New.
4017 * optabs-tree.c (target_supports_op_p): New.
4019 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4021 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4022 $gcc_cv_ld --help output.
4023 (gcc_cv_ld_demangle): Likewise.
4024 (gcc_cv_ld_eh_frame_hdr): Likewise.
4025 (gcc_cv_ld_pie): Likewise.
4026 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4027 (gcc_cv_ld_buildid): Likewise.
4028 (gcc_cv_ld_sysroot): Likewise.
4029 (ld_bndplt_support): Likewise.
4030 (ld_pushpopstate_support): Likewise.
4031 * configure: Regenerate.
4032 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4034 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4037 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4038 operands indexes to avoid gap between operands and match_dups.
4040 2017-06-21 Andrew Pinski <apinski@cavium.com>
4042 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4043 Increment Arith_shift and Arith_shift_reg by 1.
4044 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4046 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4047 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4048 (aarch64_strip_extend): Add new argument and test for it.
4049 (aarch64_cheap_mult_shift_p): New function.
4050 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4051 add a cost if it is true.
4052 Update calls to aarch64_strip_extend.
4053 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4055 2017-06-21 Andrew Pinski <apinski@cavium.com>
4057 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4059 (thunderxt88): Likewise.
4060 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4061 (thunderx_prefetch_tune): New variable.
4062 (thunderx2t99_prefetch_tune): Update for the correct values.
4063 (thunderxt88_tunings): New variable.
4064 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4065 generic_prefetch_tune.
4066 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4068 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4070 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4071 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4072 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4073 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4074 (aarch64_atomic_cas<mode>, GPI): Likewise.
4076 2017-06-21 Martin Liska <mliska@suse.cz>
4078 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4079 statements on cold and hot labels.
4080 * predict.c (tree_estimate_probability_bb): Remove the
4081 prediction from this place.
4083 2017-06-21 Martin Liska <mliska@suse.cz>
4085 PR tree-optimization/79489
4086 * gimplify.c (maybe_add_early_return_predict_stmt): New
4088 (gimplify_return_expr): Call the function.
4089 * predict.c (tree_estimate_probability_bb): Remove handling
4091 * predict.def: Update comment about early return predictor.
4092 * gimple-predict.h (is_gimple_predict): New function.
4093 * predict.def: Change default value of early return to 66.
4094 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4096 * passes.def: Put pass_strip_predict_hints to the beginning of
4099 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4101 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4102 FUNCTION_DECL declarations.
4103 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4105 (dwaf2out_decl): Likewise.
4106 * godump.c (go_early_global_decl): Skip call to the real debug hook
4107 for FUNCTION_DECL declarations.
4108 * passes.c (rest_of_decl_compilation): Skip call to the
4109 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4110 -fdump-go-spec is passed.
4112 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4114 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4115 Reorder for compactness.
4116 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4117 (def_builtin_pure, def_builtin_pure2): New functions.
4118 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4120 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4122 * match.pd (nop_convert): New predicate.
4123 ((A +- CST1) +- CST2): Allow some NOP conversions.
4125 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4128 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4129 with ctors/dtors if GOVD_SHARED is set.
4131 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4133 * config/aarch64/aarch64.md (movti_aarch64):
4134 Emit mov rather than orr.
4135 (movtf_aarch64): Likewise.
4136 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4137 Emit mov rather than orr.
4139 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4141 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4142 Swap alternatives, make integer dup more expensive.
4144 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4146 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4147 Return true for non-tls symbols.
4149 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4151 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4152 (cortex-a75): Likewise.
4153 (cortex-a75.cortex-a55): Likewise.
4154 * config/aarch64/aarch64-tune.md: Regenerate.
4155 * doc/invoke.texi (-mtune): Document new values for -mtune.
4157 2017-06-21 Tom de Vries <tom@codesourcery.com>
4159 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4161 (Effective-Target Keywords, Other attributes): Suggest using
4162 dg-add-options stack_size feature to get stack limit in stack_size
4163 effective target documentation.
4165 2017-06-21 Julian Brown <julian@codesourcery.com>
4166 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4168 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4169 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4170 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4172 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4173 attribute type list for neon_multiply.
4174 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4175 attribute type list for neon_multiply.
4176 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4177 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4178 attribute type list for neon_multiply.
4179 * config/arm/types.md (crypto_pmull): Add.
4180 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4181 attribute type list.
4183 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4185 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4188 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4190 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4191 to make sure not to dereference a NULL cost_classes_ptr pointer.
4193 2017-06-20 Carl Love <cel@us.ibm.com>
4195 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4196 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4197 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4198 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4199 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4200 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4201 VMULOSW): New enum "unspec" values.
4202 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4203 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4204 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4205 altivec_vmulosw): New patterns.
4206 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4207 VMULOSW): Add definitions.
4209 2017-06-20 Julia Koval <julia.koval@intel.com>
4211 * config/i386/i386.c: Fix rounding expand for new pattern.
4212 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4214 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4216 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4217 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4219 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4221 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4224 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4226 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4227 architecture, then by implementer ID.
4228 * config/aarch64/aarch64-tune.md: Regenerate.
4230 2017-06-20 Richard Biener <rguenther@suse.de>
4233 * fold-const.c (split_tree): Fold to type before negating.
4235 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4237 * diagnostic-show-locus.c
4238 (selftest::test_fixit_deletion_affecting_newline): New function.
4239 (selftest::diagnostic_show_locus_c_tests): Call it.
4241 2017-06-20 Andreas Schwab <schwab@suse.de>
4244 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4247 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4249 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4250 __ARM_FEATURE_COPROC according to support.
4252 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4254 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4255 Rewritten to avoid overflow for > 32-bit pointers.
4258 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4259 by removing enum keyword.
4260 (ubsan_type_descriptor): Likewise. Formatting fix.
4263 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4264 splitter): Require TARGET_SSE2 in the condition.
4266 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4269 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4270 for doing vector set of SFmode on ISA 3.0.
4271 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4272 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4274 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4275 SFmode value into a V4SF variable that was extracted from another
4276 V4SF variable without converting the element to double precision
4277 and back to single precision vector format.
4278 (vsx_insert_extract_v4sf_p9_2): Likewise.
4280 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4282 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4283 in UWHI to avoid undefined overflow.
4286 * ubsan.h (enum ubsan_encode_value_phase): New.
4287 (ubsan_encode_value): Change second argument to
4288 enum ubsan_encode_value_phase with default value of
4289 UBSAN_ENCODE_VALUE_GENERIC.
4290 * ubsan.c (ubsan_encode_value): Change second argument to
4291 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4292 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4293 create_tmp_var_raw instead of create_tmp_var and use a
4295 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4296 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4297 ubsan_encode_value callers.
4300 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4301 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4302 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4304 2017-06-19 Richard Biener <rguenther@suse.de>
4307 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4308 estimates if we changed anything.
4310 2017-06-19 Richard Biener <rguenther@suse.de>
4312 PR tree-optimization/80887
4313 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4314 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4315 simplified lookups, then reset mprts_hook.
4316 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4318 (try_to_simplify): Likewise.
4320 2017-06-19 Martin Liska <mliska@suse.cz>
4323 * gimplify.c (gimplify_switch_expr):
4324 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4326 2017-06-19 Martin Liska <mliska@suse.cz>
4328 * doc/install.texi: Document that PGO runs in 4 stages.
4330 2017-06-19 Martin Liska <mliska@suse.cz>
4333 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4334 to dispatcher function name.
4335 * multiple_target.c (replace_function_decl): New function.
4336 (create_dispatcher_calls): Redirect both edges and references.
4338 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4340 * profile-count.c (profile_count::dump): Dump quality.
4341 (profile_count::differs_from_p): Update for unsigned val.
4342 * profile-count.h (profile_count_quality): New enum.
4343 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4345 2017-06-19 Richard Biener <rguenther@suse.de>
4347 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4348 struct function as arg.
4349 (estimate_numbers_of_iterations): Export overload with loop arg.
4350 (free_numbers_of_iterations_estimates_loop): Use an overload of
4351 free_numbers_of_iterations_estimates instead.
4352 * tree-cfg.c (remove_bb): Adjust.
4353 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4354 * tree-parloops.c (gen_parallel_loop): Likewise.
4355 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4357 (tree_unroll_loops_completely): Likewise.
4358 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4359 Use an overload instead and export.
4360 (estimated_loop_iterations): Adjust.
4361 (max_loop_iterations): Likewise.
4362 (likely_max_loop_iterations): Likewise.
4363 (estimate_numbers_of_iterations): Take struct function as arg
4365 (loop_exits_before_overflow): Adjust.
4366 (free_numbers_of_iterations_estimates_loop): Use an overload.
4367 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4368 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4370 2017-06-19 Richard Biener <rguenther@suse.de>
4373 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4374 RANGE_EXPR conservatively.
4376 2017-06-16 Carl Love <cel@us.ibm.com>
4378 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4379 definitions for vec_float, vec_float2, vec_floato,
4380 vec_floate built-ins.
4381 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4382 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4384 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4385 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4386 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4387 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4388 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4389 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4390 vec_floato): Add builtin defines.
4391 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4392 Update the built-in documentation file for the new built-in
4395 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4397 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4398 (mthumb): Mark as the negative of -marm.
4400 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4402 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4404 (ARM Options, -mtune): Document that this accepts the same
4405 extension options as -mcpu.
4406 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4408 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4410 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4411 permitted extensions.
4413 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4415 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4416 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4417 (armv8-m.main): Add option +nodsp.
4418 * config/arm/arm-cpu-cdata.h: Regenerated.
4420 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4422 * config/arm/t-fuchsia: New file.
4423 * config.gcc (arm*-*-fuchsia*): Use it.
4425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4427 * config/arm/t-symbian: Rewrite for new option infrastructure.
4429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4431 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4432 (MULTILIB_REQUIRED): Likewise.
4434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4436 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4437 (MULTILIB_RESUE): Likewise.
4438 (MULTILIB_MATCHES): Likewise.
4439 (MULTLIB_REQUIRED): Likewise.
4441 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4443 * config/arm/t-rtems: Rewrite for new option framework.
4445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4447 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4448 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4449 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4450 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4451 * config/arm/t-multilib: ... here.
4452 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4453 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4454 armv7-a and armv8*-a when A-profile libraries have not been built.
4455 * config/arm/t-rmprofile: Rewrite.
4457 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4459 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4460 with a backslash. Remove the backslash after substituting unescaped
4462 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4466 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4467 the driver to pass through the default setting of -mfloat-abi.
4468 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4470 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4471 (all_feat_combs): New rule.
4472 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4474 * config/arm/t-aprofile: Rewrite.
4476 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4478 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4479 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4480 fpu is not specified by the user/command-line.
4481 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4482 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4483 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4484 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4485 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4486 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4490 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4491 * config/arm/t-arm-elf: Rewritten.
4493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4495 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4496 have some floating-point instructions.
4497 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4498 (TARGET_MAYBE_HARD_FLOAT): New macro.
4499 * config/arm/arm-builtins.c (arm_init_builtins): Use
4500 TARGET_MAYBE_HARD_FLOAT.
4501 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4503 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4505 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4506 (configargs.h): Include it.
4507 (arm_print_hint_for_fpu_option): New function.
4508 (arm_parse_fpu_option): New function.
4509 (candidate_extension): New class.
4510 (arm_canon_for_multilib): New function.
4511 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4512 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4513 (ARCH_CANONICAL_SPECS): New macro.
4514 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4516 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4518 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4519 are set after handling multilib fragments. Set target_cpu_default2
4522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4524 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4526 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4527 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4529 (check_arch): Likewise.
4530 * config/arm/arm.c (arm_configure_build_target): Handle
4531 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4532 options in the default.
4534 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4536 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4537 when an option is an alias of another.
4538 * config/arm/parsecpu.awk (optalias): New parser token.
4539 (gen_comm_data): Mark non-alias options as such. Emit entries
4540 for extension aliases.
4541 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4542 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4543 (armv6kz, armv6zk, armv6t2): Likewise.
4544 (armv7): Make vfpv3-d16 an alias.
4545 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4547 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4548 Sort in canonical order.
4549 (armv8-a): Sort in canonical order.
4550 (armv8.1-a, armv8.2-a): Likewise.
4551 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4553 (cortex-a9): Sort in canonical order.
4554 * config/arm/arm.c (selftests.h): Include it.
4555 (arm_test_cpu_arch_data): New function.
4556 (arm_run_self_tests): New function.
4557 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4558 (targetm): Move declaration to the end of the file.
4559 * arm-cpu-cdata.h: Regenerated.
4561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4563 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4564 call to target_mode_check describing the type of option passed.
4565 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4566 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4567 arm_parse_cpu_option_name to match parameters against list of
4569 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4570 arm_arch_core_flags data structure.
4571 * config/arm/arm-cpu_cdata.h: Regenerated.
4573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4575 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4577 (arm_print_hint_for_cpu_option): Likewise.
4578 (arm_print_hint_for_arch_option): Likewise.
4579 (arm_parse_cpu_option_name): Likewise.
4580 (arm_parse_arch_option_name): Likewise.
4581 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4582 of entries in the all_fpus list.
4583 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4584 (arm_parse_cpu_option_name): Declare.
4585 (arm_parse_arch_option_name): Declare.
4586 (arm_parse_option_features): Declare.
4587 (arm_intialize_isa): Declare.
4588 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4590 (gen_comm_data): ... here. Make definitions non-static.
4591 * config/arm/arm-cpu-data.h: Regenerated.
4592 * config/arm/arm-cpu-cdata.h: Regenerated.
4594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4596 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4597 (cpu_arch_extension): New structure.
4598 (cpu_arch_option, arch_option, cpu_option): New structures.
4599 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4601 (gen_data): Generate new format data tables.
4602 * config/arm/arm.c (cpu_tune): New structure.
4603 (cpu_option, processors): Delete.
4604 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4605 (arm_print_hint_for_cpu_option): ... this and ...
4606 (arm_print_hint_for_arch_option): ... this.
4607 (arm_parse_arch_cpu_name): Delete. Replace with ...
4608 (arm_parse_cpu_option_name): ... this and ...
4609 (arm_parse_arch_option_name): ... this.
4610 (arm_unrecognized_feature): Change type of target parameter to
4612 (arm_parse_arch_cpu_features): Delete. Replace with ...
4613 (arm_parse_option_features): ... this.
4614 (arm_configure_build_target): Rework to use new configuration data
4616 (arm_print_tune_info): Rework for new configuration data tables.
4617 * config/arm/arm-cpu-data.h: Regenerated.
4618 * config/arm/arm-cpu.h: Regenerated.
4620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4622 * Makefile.in (OBJS): Move sbitmap.o from here ...
4623 (OBJS-libcommon): ... to here.
4625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4627 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4628 (ISA_ALL_CRYPTO): New macro.
4629 (ISA_ALL_SIMD): New macro
4630 (ISA_ALL_FP): New macro.
4631 * config/arm/arm.c (fpu_bitlist): Update initializer.
4632 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4634 (arm9e): Add fpu. Add option for nofp
4635 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4636 (arm926ej-s, arm1026ej-s): Likewise.
4637 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4638 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4639 neon-fp16, neon-vfpv4, nofp and nosimd.
4640 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4641 (cortex-a8): Add fpu. Add option for nofp.
4642 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4643 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4644 (cortex-r4f): Add fpu.
4645 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4646 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4648 (cortex-r8): Likewise.
4649 (cortex-m4): Add fpu. Add option for nofp.
4650 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4651 (cortex-a17.cortex-a7): Likewise.
4652 (cortex-a32): Add fpu. Add options for crypto and nofp.
4653 (cortex-a35, cortex-a53): Likewise.
4654 (cortex-a57): Add fpu. Add option for crypto.
4655 (cortex-a72, cortex-a73): Likewise.
4656 (exynos-m1): Likewise.
4657 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4658 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4659 (cortex-m33): Add fpu. Add option for nofp.
4660 * config/arm/arm-cpu-cdata.h: Regenerated
4661 * config/arm/arm-cpu-data.h: Regenerated.
4663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4665 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4666 (armv5te, armv5tej): Likewise.
4667 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4668 (armv7): Add options fp and vfpv3-d16.
4669 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4670 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4672 (armv7ve): Likewise.
4673 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4674 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4675 (armv8-a): Add nocrypto option.
4676 (armv8.1-a, armv8.2-a): Likewise.
4677 (armv8-m.main): add options fp, fp.dp and nofp.
4679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4681 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4683 (armv8-a+crc): Delete.
4684 (armv8.1-a): Add options simd, crypto and nofp.
4685 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4686 (armv8.2-a+fp16): Delete.
4687 (armv8-m.main): Add option dsp.
4688 (armv8-m.main+dsp): Delete.
4689 (cortex-a8): Add fpu. Add nofp option.
4690 (cortex-a9): Add fpu. Add nofp and nosimd options.
4691 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4692 link to main cpu and architecture data structures.
4693 (gen_comm_data): Only put isa attributes from the main architecture
4695 (option): New statement for architecture and CPU entries.
4696 * arm.c (struct cpu_option): New structure.
4697 (struct processors): Add entry for options.
4698 (arm_unrecognized_feature): New function.
4699 (arm_parse_arch_cpu_name): Ignore any characters after the first
4701 (arm_parse_arch_cpu_feature): New function.
4702 (arm_configure_build_target): Separate out any CPU and architecture
4703 features and parse separately. Don't error out if -mfpu=auto is
4704 used with only an architecture string.
4705 (arm_print_asm_arch_directives): New function.
4706 (arm_file_start): Call it.
4707 * config/arm/arm-cpu-cdata.h: Regenerated.
4708 * config/arm/arm-cpu-data.h: Likewise.
4709 * config/arm/arm-tables.opt: Likewise.
4711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4713 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4714 assembler when it is not -mfpu=auto.
4716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4718 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4719 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4720 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4721 (ASM_CPU_SPEC): Rewrite.
4722 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4723 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4724 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4725 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4726 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4727 copied string is NUL-terminated. Also strip any characters prefixed
4729 (arm_rewrite_selected_arch): New function.
4730 (arm_rewrite_march): New function.
4732 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4734 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4735 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4736 (march, mcpu, mtune): Convert to string-based options.
4737 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4738 (arm_parse_arch_cpu_name): New function.
4739 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4740 identify selected architecture or CPU.
4741 (arm_option_save): New function.
4742 (TARGET_OPTION_SAVE): Redefine.
4743 (arm_option_restore): Restore string options.
4744 (arm_option_print): Print string options.
4746 2017-06-16 Martin Sebor <msebor@redhat.com>
4748 PR tree-optimization/80933
4749 PR tree-optimization/80934
4750 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4751 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4752 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4753 (gimple_fold_builtin): Call them.
4755 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4757 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4758 as unlikely; update profile.
4760 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4762 * predict.c (force_edge_cold): Handle declaring edges impossible
4765 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4767 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4769 (try_unroll_loop_completely): Fix reporting.
4771 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4773 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4775 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4778 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4779 if given a non-constant argument for an intrinsic which requires a
4782 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4784 * profile.c (compare_freqs): New function.
4785 (branch_prob): Sort edge list.
4786 (find_spanning_tree): Assume that the list is priority sorted.
4788 2017-06-16 Richard Biener <rguenther@suse.de>
4790 PR tree-optimization/81090
4791 * passes.def (pass_record_bounds): Remove.
4792 * tree-pass.h (make_pass_record_bounds): Likewise.
4793 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4794 make_pass_record_bounds): Likewise.
4795 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4796 not free niter estimates at the beginning but at the end.
4797 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4799 2017-06-16 Richard Biener <rguenther@suse.de>
4801 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4802 initializer to workaround ICE in host GCC 4.8.
4804 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4806 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4808 (clone_inlined_nodes): Update.
4810 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4812 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4813 prefetch settings, and enable prefetching by default at -O3.
4815 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4817 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4818 Set flag_prefetch_loop_arrays according to tuning data.
4820 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4822 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4824 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4825 [Unrelated to main purpose of the patch] Place the pointer field last
4826 to enable type checking errors when tune structure are wrongly merged.
4827 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4828 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4829 (thunderx2t99_prefetch_tune): New tune constants.
4830 (tune_params *_tunings): Update all tunings (no functional change).
4831 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4832 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4833 from tunings structures.
4835 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4838 * ubsan.c (instrument_null): Add T argument, use it instead
4839 of computing it based on IS_LHS.
4840 (instrument_object_size): Likewise.
4841 (pass_ubsan::execute): Adjust instrument_null and
4842 instrument_object_size callers to pass gimple_get_lhs or
4843 gimple_assign_rhs1 result to it. Use instrument_null instead of
4844 calling get_base_address and instrument_mem_ref. Handle
4845 aggregate call arguments for object-size sanitization.
4847 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4849 PR tree-optimization/81089
4850 * tree-vrp.c (is_masked_range_test): Validate operands of
4853 2017-06-15 Martin Sebor <msebor@redhat.com>
4856 * dumpfile.c (dump_register): Avoid calling memset to initialize
4857 a class with a default ctor.
4858 * gcc.c (struct compiler): Remove const qualification.
4859 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4860 * hash-table.h: Ditto.
4861 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4863 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4864 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4866 * params.h (struct param_info): Make struct members non-const.
4867 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4868 with default initialization.
4869 * vec.h (vec_copy_construct, vec_default_construct): New helper
4871 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4872 with vec_copy_construct.
4873 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
4874 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
4875 * doc/invoke.texi (-Wclass-memaccess): Document.
4877 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4879 * emit-rtl.h (is_leaf): Update comment about local
4882 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4885 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
4886 for a variable to have a section before checking if the section has a
4888 Set section to.persistent if persistent attribute is set.
4889 Warn if .persistent attribute is used on an automatic variable.
4891 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
4893 PR rtl-optimization/80474
4894 * reorg.c (update_block): Do not ignore instructions in a delay slot.
4896 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
4898 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
4901 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
4903 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
4904 (casesi): Emit bounds checking as RTL.
4905 (casesi_internal_mips16_<mode>): Remove bounds checking.
4907 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
4909 * config/xtensa/xtensa.c (xtensa_option_override): Append
4910 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
4911 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
4912 xtensa_doloop_hooks): Define unconditionally.
4913 (xtensa_reorg_loops): Only call reorg_loops in the presence of
4915 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
4916 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
4917 for it in xtensa_option_override.
4918 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
4919 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
4921 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
4923 * doc/cppopts.texi: Document '-' special value to -MF.
4925 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
4927 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
4928 (cortex_a53_fconst): Likewise.
4929 (cortex_a53_fpmul): Likewise.
4930 (cortex_a53_f_load_64): Likewise.
4931 (cortex_a53_f_load_many): Likewise.
4932 (cortex_a53_advsimd_alu): Likewise.
4933 (cortex_a53_advsimd_alu_q): Likewise.
4934 (cortex_a53_advsimd_mul): Likewise.
4935 (cortex_a53_advsimd_mul_q): Likewise.
4936 (fpmac bypass): Add new bypass for fpmac-fpmac case.
4937 Add missing fmul, r2f_cvt and fconst cases.
4939 2017-06-14 Richard Biener <rguenther@suse.de>
4942 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
4944 (fold_binary_loc): When associating do not treat pre-existing
4945 TREE_OVERFLOW on literal constants as a reason to allow
4946 TREE_OVERFLOW on associated literal constants.
4948 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4950 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
4951 (MASK_FEATURES): New macro.
4952 * config/sparc/sparc.c (sparc_option_override): Remove the special
4953 handling of -mfpu and generalize it to all MASK_FEATURES switches.
4955 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4957 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
4958 a division of 0 if non-call exceptions are enabled.
4960 2017-06-14 Andrew Pinski <apinski@cavium.com>
4961 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4964 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
4965 Improve vector initialization code gen for only variable case.
4967 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4969 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
4971 2017-06-14 Richard Biener <rguenther@suse.de>
4973 PR tree-optimization/81083
4974 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
4977 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4979 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
4980 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
4981 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
4982 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
4983 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
4984 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
4986 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4988 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
4989 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
4991 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4993 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
4995 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4997 * config/rs6000/t-rtems: Don't handle SPE.
4999 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5001 * config/rs6000/t-linux: Don't handle SPE.
5003 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5005 * config/rs6000/eabispe.h: Delete file.
5007 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5009 * config/rs6000/t-spe: Delete file.
5011 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5013 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5014 (rs6000_legitimate_offset_address_p): Return false for anything in
5015 V2SImode or V2SFmode.
5017 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5019 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5020 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5021 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5023 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5024 (rs6000_legitimate_offset_address_p): Ditto.
5025 (rs6000_emit_move): Ditto.
5026 (rs6000_init_builtins): Remove V4HI_type_node.
5028 2017-06-13 Martin Liska <mliska@suse.cz>
5031 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5032 (gate_asan): Likewise.
5033 * asan.h (asan_no_sanitize_address_p): Remove the function.
5034 (sanitize_flags_p): New function.
5035 * builtins.def: Fix coding style.
5036 * common.opt: Use renamed enum value.
5037 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5038 * doc/extend.texi: Document no_sanitize attribute.
5039 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5040 to SANITIZE_UNDEFINED_NONDEFAULT.
5041 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5042 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5043 Use sanitize_flags_p.
5044 * gimplify.c (gimplify_function_tree): Likewise.
5045 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5046 * opts.c (parse_no_sanitize_attribute): New function.
5047 (common_handle_option): Use renamed enum value.
5048 * opts.h (parse_no_sanitize_attribute): Declare.
5049 * tree.c (sanitize_flags_p): New function.
5050 * tree.h: Declared here.
5051 * tsan.c: Use sanitize_flags_p.
5052 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5053 (instrument_mem_ref): Likewise.
5054 (instrument_bool_enum_load): Likewise.
5055 (do_ubsan_in_current_function): Remove the function.
5056 (pass_ubsan::execute): Use sanitize_flags_p.
5057 * ubsan.h: Remove do_ubsan_in_current_function
5058 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5059 (dump_function_to_file): Use it here.
5061 2017-06-13 Martin Jambor <mjambor@suse.cz>
5063 PR tree-optimization/80803
5064 PR tree-optimization/81063
5065 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5066 (propagate_subaccesses_across_link): Enqueue subtree whenever
5067 necessary instead of relying on the caller.
5069 2017-06-13 Martin Jambor <mjambor@suse.cz>
5071 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5072 that have a first_link.
5073 (sort_and_splice_var_accesses): Do not check first_link before
5075 (subtree_mark_written_and_enqueue): Likewise.
5076 (propagate_all_subaccesses): Likewise and do not stop at first
5077 parent with a first_link.
5079 2017-06-13 Martin Jambor <mjambor@suse.cz>
5081 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5084 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5086 * match.pd: New pattern.
5088 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5090 * tree-vrp.c (is_masked_range_test): New function.
5091 (register_edge_assert_for): Determine ranges for
5094 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5096 PR tree-optimization/67328
5097 * fold-const.c (maskable_range_p): New function.
5098 (build_range_check): Generate bittests if possible.
5100 2017-06-13 Martin Liska <mliska@suse.cz>
5102 * gimple-pretty-print.c (dump_probability): Add new argument.
5103 (dump_edge_probability): Dump both probability and count.
5104 (dump_gimple_label): Likewise.
5105 (dump_gimple_bb_header): Likewise.
5107 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5110 * config/avr/avr-devices.c: Fix indentation.
5111 * config/avr/gen-avr-mmcu-specs.c: Dito.
5113 2017-06-13 Richard Biener <rguenther@suse.de>
5115 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5116 instead get vector type from stmt_info.
5117 (vectorizable_reduction): Adjust. Remove dead code.
5119 2017-06-13 Richard Biener <rguenther@suse.de>
5122 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5123 case of C * (x * C2 + C3).
5124 (fold_addr_of_array_ref_difference): Properly fold index difference.
5126 2017-06-12 David S. Miller <davem@davemloft.net>
5129 * config/sparc/sparc.md (return expander): Emit frame blockage if
5130 function uses alloca.
5132 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5134 * combine.c (make_field_assignment): Check len rather than the mode
5135 precision when calling force_to_mode.
5137 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5139 Support multilibs and devices that see flash in RAM address range.
5142 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5143 (avr_mcu_t) <flash_pm_offset>: New field.
5144 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5145 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5146 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5147 (AVR_TINY_PM_OFFSET): Remove macro.
5148 * config/avr/avr.opt (-mshort-calls): New option.
5149 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5150 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5151 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5152 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5153 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5154 instead of avr_arch->have_jmp_call.
5155 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5156 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5157 avr_arch->flash_pm_offset to define.
5158 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5159 new field flash_pm_offset. Add entry for avrxmega3.
5160 (avr_texinfo): Add entry for avrxmega3.
5161 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5162 attiny212, attiny214,
5163 attiny412, attiny414, attiny416, attiny417,
5164 attiny814, attiny816, attiny817,
5165 attiny1614, attiny1616, attiny1617,
5166 attiny3214, attiny3216, attiny3217.
5167 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5168 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5169 (avr_print_operand_address) [AVR_TINY]: Same.
5170 (avr_asm_init_sections) <readonly_data_section>: Only patch
5171 callback if avr_arch->flash_pm_offset = 0.
5172 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5173 for rodata if avr_arch->flash_pm_offset != 0.
5174 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5175 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5176 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5177 (m_options): Append opt_rcall.
5178 (m_dirnames): Append dir_rcall.
5179 * config/avr/t-multilib: Regenerate.
5181 * configure.ac [target=avr]: Check whether avrxmega3 default
5182 linker description file works as needed.
5183 * configure: Regenerate.
5184 * doc/avr-mmcu.texi: Regenerate.
5185 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5186 <__AVR_ARCH__>: Document avrxmega3 and 103.
5187 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5188 <__AVR_SHORT_CALLS__>: Document it.
5189 <__AVR_PM_BASE_ADDRESS__>: Document it.
5190 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5191 (AVR Variable Attributes) <progmem>: Document this is
5192 not needed for avrxmega3.
5193 (AVR Named Address Spaces) <__flash>: Dito.
5195 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5197 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5199 2017-06-12 Doug Rupp <rupp@adacore.com>
5201 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5202 Append vxworks-stdint.h to the tm_file list.
5203 * config/vxworks-stdint.h: New file.
5205 2017-06-12 Martin Liska <mliska@suse.cz>
5207 PR tree-optimization/81041
5208 * tree-profile.c (gimple_gen_ic_func_profiler):
5209 Create an extra BB in profile-generate
5210 (gimple_gen_time_profiler): Likewise.
5212 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5214 PR tree-optimization/81003
5215 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5216 (update_range_test): Use it instead of force_gimple_operand_gsi.
5218 2017-06-12 Richard Biener <rguenther@suse.de>
5220 PR tree-optimization/81053
5221 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5222 with backedge value not defined in loop. Simplify def stmt
5225 2017-06-11 Tom de Vries <tom@codesourcery.com>
5228 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5230 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5231 nvptx_cannot_force_const_mem.
5233 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5235 * opts.c (finish_options): Move test for flag_split_stack after
5236 it has been initialized.
5238 2017-06-11 Jason Merrill <jason@redhat.com>
5240 * tree.h (id_equal): New.
5241 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5242 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5243 instead of strcmp of IDENTIFIER_POINTER.
5245 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5247 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5248 (mark_all_inlined_calls_cdtor): Fix formating.
5249 (inline_transform): Rescale profile before inlining.
5251 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5253 * cgraph.h (cgraph_edge::clone): Update prototype.
5254 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5255 (cgraph_node::create_clone): Update.
5256 (cgraph_node::create_version_clone): Update.
5257 * tree-inline.c (copy_bb): Update.
5258 (expand_call_inline): Update.
5260 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5262 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5263 factored out from ...
5264 (rs6000_emit_prologue): ... here.
5266 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5268 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5269 factored out from ...
5270 (rs6000_emit_prologue): ... here.
5272 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5274 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5276 (handle_missing_profiles): Fix computation of tp_first_run.
5277 (counts_to_freqs): Do not touch freqs when count is 0.
5279 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5281 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5284 2017-06-10 Tom de Vries <tom@codesourcery.com>
5286 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5287 attributes): Document signal effective target.
5289 2017-06-10 Tom de Vries <tom@codesourcery.com>
5291 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5292 Document effective target stack_size.
5294 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5296 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5297 to the edit_context if they can be auto-applied.
5299 2017-06-9 Ian Lance Taylor <iant@golang.org>
5301 * opts.c (finish_options): If -fsplit-stack, disable implicit
5302 -forder-blocks-and-partition.
5303 * doc/invoke.texi (Optimize Options): Document that when using
5304 -fsplit-stack -forder-blocks-and-partition is not implicitly
5307 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5309 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5310 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5311 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5312 * builtins.def (abort, trap, unreachable): Declare cold.
5313 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5314 * tree-core.h (ECF_COLD): New.
5315 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5316 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5318 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5320 * predict.c (unlikely_executed_stmt_p): Cleanup.
5322 2017-06-09 Richard Biener <rguenther@suse.de>
5324 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5325 model if the ref is always written to.
5327 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5329 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5331 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5333 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5336 2017-06-09 Tom de Vries <tom@codesourcery.com>
5339 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5340 "target cannot support label values" when encountering LABEL_REF.
5342 2017-06-09 Martin Liska <mliska@suse.cz>
5344 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5345 (gimple_gen_ic_func_profiler): Emit direct comparison
5346 of __gcov_indirect_call_callee with NULL.
5347 (gimple_gen_time_profiler): Change probability from
5348 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5350 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5352 * profile.c (edge_gcov_counts): Turn to pointer.
5353 (compute_branch_probabilities, compute_branch_probabilities): Update.
5354 (branch_prob): Do not clear edge_gcov_count.
5355 * profile.h (edge_gcov_counts): Turn to pointer.
5356 (edge_gcov_count): Update.
5358 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5360 * gimple.h (gimple_check_failed): Mark cold.
5362 2017-06-09 Richard Biener <rguenther@suse.de>
5364 PR tree-optimization/66623
5365 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5366 refactor check_reduction into two parts, properly computing
5367 whether we have to check reduction validity for outer loop
5370 2017-06-09 Richard Biener <rguenther@suse.de>
5372 PR tree-optimization/79483
5373 * graphite-scop-detection.c (order): New global.
5374 (get_order): Compute bb to order mapping that satisfies code
5375 generation constraints.
5376 (cmp_pbbs): New helper.
5377 (build_scops): Start domwalk at entry block, sort generated
5380 2017-06-09 Richard Biener <rguenther@suse.de>
5383 * ipa-polymorphic-call.c
5384 (ipa_polymorphic_call_context::restrict_to_inner_class):
5385 Skip FIELD_DECLs with error_mark_node type.
5386 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5389 2017-06-09 Martin Liska <mliska@suse.cz>
5391 * predict.c (struct branch_predictor): New struct.
5392 (test_prediction_value_range): New test.
5393 (predict_c_tests): New function.
5394 * selftest-run-tests.c (selftest::run_tests): Run the function.
5395 * selftest.h: Declare new tests.
5397 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5400 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5401 gen_add3_insn did not fail.
5402 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5403 r0, construct that number in a temporary reg and add that reg to r0.
5404 If asked to put the result in r0 as well, fail.
5406 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5409 for early expansion of vec_eqv.
5411 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5414 * ubsan.c (instrument_null): Avoid pointless code temporary.
5415 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5417 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5421 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5424 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5427 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5429 * predict.c (unlikely_executed_edge_p): Move ahead.
5430 (probably_never_executed_edge_p): Use it.
5432 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5435 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5436 gimple_call_builtin_p call.
5438 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5440 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5441 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5442 rtl_check_failed_type2, rtl_check_failed_code1,
5443 rtl_check_failed_code2, rtl_check_failed_code_mode,
5444 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5445 rtvec_check_failed_bounds, rtl_check_failed_flag,
5446 _fatal_insn_not_found, _fatal_insn): Likewise.
5447 * tree.h (tree_contains_struct_check_failed,
5448 tree_check_failed, tree_not_check_failed,
5449 tree_class_check_failed, tree_range_check_failed,
5450 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5451 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5452 tree_operand_check_failed, omp_clause_check_failed,
5453 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5456 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5458 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5459 flag_branch_probabilities.
5460 * ipa-inline.c (edge_badness): Likewise.
5461 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5462 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5463 * predict.c (maybe_hot_frequency_p): Likewise.
5464 (probably_never_executed): Likewise.
5465 * sched-ebb.c (schedule_ebbs): Likewise.
5466 * sched-rgn.c (find_single_block_region): Likewise.
5467 * tracer.c (tail_duplicate): Likewise.
5469 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5471 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5472 longer requires x_flag_profile_use.
5474 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5476 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5477 instead of flag_reorder_blocks_and_partition.
5478 * dbxout.c (dbxout_function_end): Likewise.
5479 * dwarf2out.c (gen_subprogram_die): Likewise.
5480 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5481 * hw-doloop.c (reorg_loops): Likewise.
5482 * varasm.c (assemble_start_function,
5483 assemble_end_function): Likewise.
5484 (decide_function_section): Do not check for
5485 flag_reorder_blocks_and_partition.
5487 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5489 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5491 (chkp_get_hard_register_fake_addr_expr): Ditto.
5492 (chkp_build_addr_expr): Add check for hard reg case.
5493 (chkp_parse_array_and_component_ref): Ditto.
5494 (chkp_find_bounds_1): Ditto.
5495 (chkp_process_stmt): Don't generate bounds store for
5498 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5500 * predict.c (maybe_hot_bb_p): Do not check profile status.
5501 (maybe_hot_edge_p): Likewise.
5502 (probably_never_executed): Check for zero counts even if profile
5504 (unlikely_executed_edge_p): New function.
5505 (unlikely_executed_stmt_p): New function.
5506 (unlikely_executed_bb_p): New function.
5507 (set_even_probabilities): Use unlikely predicates.
5508 (combine_predictions_for_bb): Likewise.
5509 (predict_paths_for_bb): Likewise.
5510 (predict_paths_leading_to_edge): Likewise.
5511 (determine_unlikely_bbs): New function.
5512 (estimate_bb_frequencies): Use it.
5513 (compute_function_frequency): Use zero counts even if profile is
5515 * profile-count.h: Fix typo.
5517 2017-08-08 Julia Koval <julia.koval@intel.com>
5519 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5520 _mm512_mask_cvtsepi16_storeu_epi8,
5521 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5522 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5523 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5524 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5525 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5526 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5527 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5528 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5529 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5530 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5531 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5532 __builtin_ia32_pmovuswb256mem_mask,
5533 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5534 __builtin_ia32_pmovwb512mem_mask): New builtins.
5536 2017-08-08 Julia Koval <julia.koval@intel.com>
5538 PR target/73350,80862
5539 * config/i386/subst.md (round): Fix round pattern.
5540 * config/i386/i386.c (ix86_erase_embedded_rounding):
5541 Fix erasing rounding for the fixed pattern.
5543 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5545 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5547 2017-06-08 Martin Liska <mliska@suse.cz>
5549 PR gcov-profile/80911
5550 * gcov.c (block_info::block_info): New constructor.
5552 2017-06-07 Carl Love <cel@us.ibm.com>
5554 * config/rs6000/rs6000-c: The return type of the following
5555 built-in functions was implemented as int not long long. Fix sign
5556 of return value for the unsigned version of vec_mulo and vec_mule.
5557 vector unsigned long long vec_bperm (vector unsigned long long,
5558 vector unsigned char)
5559 vector signed long long vec_mule (vector signed int,
5561 vector unsigned long long vec_mule (vector unsigned int,
5562 vector unsigned int)
5563 vector signed long long vec_mulo (vector signed int,
5565 vector unsigned long long vec_mulo (vector unsigned int,
5566 vector unsigned int)
5567 * doc/extend.texi: Fix the documentation for the built-in
5570 2017-06-07 Carl Love <cel@us.ibm.com>
5573 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5576 2017-06-07 Carl Love <cel@us.ibm.com>
5578 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5579 support, Generate doublehv for signed int/float for BE case only.
5581 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5583 * doc/invoke.texi (mcx16): Rewrite.
5585 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5587 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5588 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5589 *mov<mode>_softfloat, and an anonymous splitter): Use
5590 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5592 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5594 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5596 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5597 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5598 (rs6000_debug_reg_global): Adjust.
5599 (rs6000_init_hard_regno_mode_ok): Adjust.
5600 (rs6000_dbx_register_number): Adjust.
5601 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5602 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5603 Remove SPE_ACC and SPEFSCR.
5604 (REG_ALLOC_ORDER): Ditto.
5605 (FRAME_POINTER_REGNUM): Change to 111.
5606 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5607 (REG_CLASS_NAMES): Ditto.
5608 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5609 (REGISTER_NAMES): Ditto.
5610 (ADDITIONAL_REG_NAMES): Ditto.
5611 (rs6000_reg_names): Ditto.
5612 * config/rs6000/rs6000.md: Renumber some register number
5615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5617 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5619 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5620 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5622 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5623 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5624 Delete the SPE high registers.
5625 (REG_ALLOC_ORDER): Ditto.
5626 (enum reg_class): Remove SPE_HIGH_REGS.
5627 (REG_CLASS_NAMES): Ditto.
5628 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5629 (REGISTER_NAMES): Ditto.
5630 (rs6000_reg_names): Ditto.
5631 * doc/tm.texi.in: Remove SPE as example.
5632 * doc/tm.texi: Regenerate.
5634 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5636 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5637 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5638 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5639 * config/rs6000/rs6000.md (type): Remove "brinc".
5641 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5643 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5644 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5645 * config/rs6000/linuxspe.h: Delete file.
5646 * config/rs6000/rs6000.md: Don't include spe.md.
5647 * config/rs6000/spe.h: Delete file.
5648 * config/rs6000/spe.md: Delete file.
5649 * config/rs6000/t-rs6000: Remove spe.md.
5651 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5653 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5654 (reg_or_none500mem_operand): Delete.
5655 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5656 instead of reg_or_none500mem_operand.
5658 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5660 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5661 handling of SPE flags.
5662 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5666 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5668 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5669 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5670 paired_divv2sf3): Similar.
5671 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5672 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5673 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5675 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5676 Rename the paired_* instruction patterns.
5677 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5679 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5680 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5681 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5682 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5683 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5685 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5686 (spe_func_has_64bit_regs_p): Delete.
5687 (spe_expand_predicate_builtin): Delete.
5688 (spe_expand_evsel_builtin): Delete.
5689 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5690 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5691 (invalid_e500_subreg): Delete.
5692 (rs6000_legitimize_address): Always force_reg op2 as well, for
5693 paired single memory accesses.
5694 (rs6000_member_type_forces_blk): Delete.
5695 (rs6000_spe_function_arg): Delete.
5696 (rs6000_expand_unop_builtin): Delete SPE handling.
5697 (rs6000_expand_binop_builtin): Ditto.
5698 (spe_expand_stv_builtin): Delete.
5699 (bdesc_2arg_spe): Delete.
5700 (spe_expand_builtin): Delete.
5701 (spe_expand_predicate_builtin): Delete.
5702 (spe_expand_evsel_builtin): Delete.
5703 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5704 (spe_init_builtins): Delete.
5705 (spe_func_has_64bit_regs_p): Delete.
5706 (savres_routine_name): Delete "info" parameter. Adjust callers.
5707 (rs6000_emit_stack_reset): Ditto.
5708 (rs6000_dwarf_register_span): Delete.
5709 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5710 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5711 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5713 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5715 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5716 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5717 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5718 mulv2sf3, divv2sf3): Delete expanders.
5720 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5722 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5724 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5726 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5727 * config/rs6000/rs6000.c: Ditto.
5729 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5731 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5732 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5733 comparison_operator.
5735 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5737 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5738 * config/rs6000/rs6000.opt: Ditto.
5739 * config/rs6000/t-rtems: Ditto.
5741 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5743 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5744 TARGET_E500_SINGLE by 0, simplify.
5745 * config/rs6000/rs6000.c: Ditto.
5746 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5747 (spe_build_register_parallel): Delete.
5748 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5749 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5750 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5751 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5752 (E500_CONVERT): Delete.
5753 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5755 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5757 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5758 * config/rs6000/dfp.md: Ditto.
5759 (negdd2, *negdd2_fpr): Merge.
5760 (absdd2, *absdd2_fpr): Merge.
5761 (negtd2, *negtd2_fpr): Merge.
5762 (abstd2, *abstd2_fpr): Merge.
5763 * config/rs6000/e500.h: Delete file.
5764 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5765 TARGET_FPRS by 1 and simplify.
5766 * config/rs6000/rs6000-c.c: Ditto.
5767 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5769 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5771 * config/rs6000/rs6000.md: Ditto.
5772 (floatdidf2, *floatdidf2_fpr): Merge.
5773 (move_from_CR_gt_bit): Delete.
5774 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5775 (E500_CR_IOR_COMPARE): Delete.
5776 (All patterns that require !TARGET_FPRS): Delete.
5777 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5779 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5781 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5783 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5785 * graphds.c (add_edge): Intitialize edge's attached data.
5786 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5787 pointer parameter. Call pointed function on each edge during
5788 graph traversing. Skip traversing the edge when the function
5790 (graphds_dfs, graphds_scc): Ditto.
5791 (for_each_edge): New parameter. Pass the new parameter to callback
5793 * graphds.h (skip_edge_callback): New function pointer type.
5794 (graphds_dfs, graphds_scc): New function pointer parameter.
5795 (graphds_edge_callback, for_each_edge): New parameter.
5797 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5799 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5800 out code checking if runtime alias check is possible to below ...
5801 Call the new function.
5802 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5803 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5805 2017-06-07 Marek Polacek <polacek@redhat.com>
5808 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5809 TYPE_OVERFLOW_WRAPS check.
5811 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5813 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5814 if versioning is required.
5815 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5816 peeling with the check for versioning.
5818 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5820 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5821 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5822 Set true to new parameter if new ssa variable is defined.
5823 (vect_gen_vector_loop_niters): Refactor. Set range information
5824 for the new vector loop bound variable.
5825 (vect_do_peeling): Ditto.
5827 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5829 * tree-affine.c (ssa.h): Include header file.
5830 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5831 has wrapping overflow behavior.
5833 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5835 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5837 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5839 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5840 (tree_to_aff_combination): ... here.
5842 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5844 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5845 reg_pressure model function.
5846 (ivopts_global_cost_for_size): Delete.
5847 (determine_set_costs, iv_ca_recount_cost): Call new model function
5848 ivopts_estimate_reg_pressure.
5850 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5852 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5853 expensive than udiv. Remove floating point cases from mod.
5855 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5857 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5860 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5862 * config/aarch64/aarch64.md
5863 (copysignsf3): Fix mask generation.
5865 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5867 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5869 (class dump_manager): Add register_dumps method.
5870 * dumpfile.c: Include langhooks.h.
5871 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5872 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
5873 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
5874 (dump_manager::dump_register): Start with 512 entries instead of 32.
5875 (dump_manager::register_dumps): New method.
5876 * toplev.c (general_init): Instead of invoking register_dumps
5877 langhook, invoke register_dumps method on the dump manager.
5878 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
5881 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
5883 * doc/md.texi: Clarify the restrictions on a define_insn condition.
5884 Say that # requires an associated define_split to exist, and that
5885 the define_split must be suitable for use after register allocation.
5887 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5889 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
5890 (compute_outgoing_frequencies): Also initialize zero counts.
5891 (find_many_sub_basic_blocks): Do not produce uninitialized profile
5892 around loops; preserve more of profile when nothing changes.
5894 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
5896 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
5898 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
5899 * config/arm/arm-cpu-cdata.h: Regenerate.
5900 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
5901 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
5902 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
5903 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
5904 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
5906 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
5908 * config/arm/t-rmprofile: Likewise.
5909 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
5911 2017-06-06 David S. Miller <davem@davemloft.net>
5914 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
5915 blockage if function uses alloca.
5917 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5919 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
5920 New "uid" fields to hold pretty-print IDs of group and ref.
5921 Memory references are now identified as <group_id>:<ref_id>
5922 instead of using [random] addresses.
5923 (dump_mem_details): Simplify, no functional change.
5924 (dump_mem_ref): Simplify and make output more concise.
5925 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
5926 (find_or_create_group): Initialize group uid.
5927 (record_ref): Initialize ref uid. Improve debug output.
5928 (prune_group_by_reuse, should_issue_prefetch_p,)
5929 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
5930 (mark_nontemporal_store, determine_loop_nest_reuse):
5931 Improve debug output.
5933 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5935 * dbgcnt.def (prefetch): New debug counter.
5936 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
5937 (schedule_prefetches): Stop issueing prefetches if debug counter
5940 2017-06-06 Tom de Vries <tom@codesourcery.com>
5942 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
5943 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
5945 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5947 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
5948 Use aarch64_reg_or_zero predicate for operand 4.
5949 (aarch64_compare_and_swap<mode> define_insn_and_split):
5950 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
5951 (aarch64_store_exclusive<mode>): Likewise for operand 2.
5953 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5955 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
5956 (arm_compute_save_core_reg_mask): This.
5957 (thumb1_compute_save_reg_mask): Rename into ...
5958 (thumb1_compute_save_core_reg_mask): This.
5959 (arm_compute_save_reg0_reg12_mask): Adapt comment.
5960 (arm_compute_frame_layout): Likewise.
5962 2017-06-06 Richard Biener <rguenther@suse.de>
5964 PR tree-optimization/80974
5965 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
5966 keep or clear leaders SSA info.
5968 2017-06-06 Tom de Vries <tom@codesourcery.com>
5970 * config/nvptx/nvptx.c (split_mode_p): New function.
5971 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
5973 2017-06-06 Tom de Vries <tom@codesourcery.com>
5975 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
5977 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5980 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
5983 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5985 * shrink-wrap.c (handle_simple_exit): Update profile.
5986 (try_shrink_wrapping): Upate profile.
5988 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5990 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
5991 (tree_guess_outgoing_edge_probabilities): New.
5992 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
5993 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
5995 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5997 * ipa-split.c (split_function): Initialize return bb profile.
5999 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6001 * profile.c (compute_branch_probabilities): Also initialize
6004 2017-06-06 Richard Biener <rguenther@suse.de>
6006 PR tree-optimization/80928
6007 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6008 (vect_analyze_loop_operations): Properly guard analysis for
6010 (vect_transform_loop): Likewise.
6011 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6012 (vect_model_induction_cost): Do not cost for pure SLP.
6013 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6014 of induction in inner loop vectorization.
6015 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6016 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6017 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6019 (vect_analyze_slp_cost_1): Cost induction.
6020 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6021 (vect_get_slp_vect_defs): Likewise.
6022 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6023 (vect_transform_stmt): Handle SLP reductions.
6024 * tree-vectorizer.h (vectorizable_induction): Adjust.
6026 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6028 * config/rs6000/rs6000.c (make_resolver_func): Update
6029 init_lowered_empty_function call.
6031 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6033 * doc/invoke.texi: Document the -fprofile-abs-path option.
6034 * common.opt (fprofile-abs-path): New option.
6035 * gcov-io.h (gcov_write_filename): Declare.
6036 * gcov-io.c (gcov_write_filename): New function.
6037 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6038 * profile.c (output_location): Likewise.
6040 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6042 * shring-wrap.c: Revert accidental commit.
6044 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6046 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6048 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6050 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6052 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6053 profile in callgraph edge.
6054 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6055 (apply_scale): Likewise.
6056 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6057 Also scale profile when inlining function with zero profile.
6058 (initialize_cfun): Update exit block profile even when it is zero.
6059 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6060 when profile is read.
6062 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6064 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6065 (CLONE_*): New constants to define the processors we can generate
6066 code for with the target_clone attribute.
6067 (rs6000_clone_map): New array to identify which clone processors
6068 the current program is running on.
6069 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6070 target_clone attribute.
6071 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6072 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6073 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6074 (cpu_expand_builtin): Add support for target_clone attribute.
6075 (rs6000_valid_attribute_p): Allow "default" attribute.
6076 (get_decl_name): New debug function to simplify printing the
6077 current function name in debugging statements.
6078 (rs6000_clone_priority): New functions to support the target_clone
6079 attribute, and be able to generate code to switch between ISA 2.05
6080 through ISA 3.0 (power6 through power9).
6081 (rs6000_compare_version_priority): Likewise.
6082 (rs6000_get_function_versions_dispatcher): Likewise.
6083 (make_resolver_func): Likewise.
6084 (add_condition_to_bb): Likewise.
6085 (dispatch_function_versions): Likewise.
6086 (rs6000_generate_version_dispatcher_body): Likewise.
6087 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6088 (fusion_gpr_load_p): Fix a spacing issue.
6089 * doc/extend.texi (Common Function Attributes): Document that the
6090 PowerPC supports the target_clone attribute.
6092 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6094 * config/arm/arm.h: explain F symbol found in description of ARM
6095 register allocation in its legend.
6097 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6099 * config/mips/frame-header-opt.c: Include profile-count.h.
6100 * config/riscv/riscv.c: Include profile-count.h
6102 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6104 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6106 (sm_set_flag_if_changed): Add bbs field.
6107 (execute_sm_if_changed_flag_set): Pass BBS.
6108 (execute_sm): Update.
6110 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6112 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6115 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6117 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6118 (peephole2): New peephole2 to emit the above.
6119 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6121 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6123 * config/aarch64/aarch64.c (define_peephole2 above
6124 *sub_<shift>_<mode>): New peephole.
6126 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6128 * config/i386/i386.c (make_resolver_func): Update.
6129 * Makefile.in: Add profile-count.h and profile-count.o
6130 * auto-profile.c (afdo_indirect_call): Update to new API.
6131 (afdo_set_bb_count): Update.
6132 (afdo_propagate_edge): Update.
6133 (afdo_propagate_circuit): Update.
6134 (afdo_calculate_branch_prob): Update.
6135 (afdo_annotate_cfg): Update.
6136 * basic-block.h: Include profile-count.h
6137 (struct edge_def): Turn count to profile_count.
6138 (struct basic_block_def): Likewie.
6139 (REG_BR_PROB_BASE): Move to profile-count.h
6140 (RDIV): Move to profile-count.h
6141 * bb-reorder.c (max_entry_count): Turn to profile_count.
6142 (find_traces): Update.
6143 (rotate_loop):Update.
6144 (connect_traces):Update.
6145 (sanitize_hot_paths):Update.
6146 * bt-load.c (migrate_btr_defs): Update.
6147 * cfg.c (RDIV): Remove.
6148 (init_flow): Use alloc_block.
6149 (alloc_block): Uninitialize count.
6150 (unchecked_make_edge): Uninitialize count.
6151 (check_bb_profile): Update.
6152 (dump_edge_info): Update.
6153 (dump_bb_info): Update.
6154 (update_bb_profile_for_threading): Update.
6155 (scale_bbs_frequencies_int): Update.
6156 (scale_bbs_frequencies_gcov_type): Update.
6157 (scale_bbs_frequencies_profile_count): New.
6158 * cfg.h (update_bb_profile_for_threading): Update.
6159 (scale_bbs_frequencies_profile_count): Declare.
6160 * cfgbuild.c (compute_outgoing_frequencies): Update.
6161 (find_many_sub_basic_blocks): Update.
6162 * cfgcleanup.c (try_forward_edges): Update.
6163 (try_crossjump_to_edge): Update.
6164 * cfgexpand.c (expand_gimple_tailcall): Update.
6165 (construct_exit_block): Update.
6166 * cfghooks.c (verify_flow_info): Update.
6167 (dump_bb_for_graph): Update.
6168 (split_edge): Update.
6169 (make_forwarder_block): Update.
6170 (duplicate_block): Update.
6171 (account_profile_record): Update.
6172 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6173 (get_estimated_loop_iterations): Update.
6174 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6175 (single_likely_exit): Update.
6176 * cfgloopmanip.c (scale_loop_profile): Update.
6178 (set_zero_probability): Update.
6179 (lv_adjust_loop_entry_edge): Update.
6180 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6181 (purge_dead_edges): Update.
6182 (rtl_account_profile_record): Update.
6183 * cgraph.c (cgraph_node::create): Uninitialize count.
6184 (symbol_table::create_edge): Uninitialize count.
6185 (cgraph_update_edges_for_call_stmt_node): Update.
6186 (cgraph_edge::dump_edge_flags): Update.
6187 (cgraph_node::dump): Update.
6188 (cgraph_edge::maybe_hot_p): Update.
6189 * cgraph.h: Include profile-count.h
6190 (create_clone), create_edge, create_indirect_edge): Update.
6191 (cgraph_node): Turn count to profile_count.
6192 (cgraph_edge0: Likewise.
6193 (make_speculative, clone): Update.
6194 (create_edge): Update.
6195 (init_lowered_empty_function): Update.
6196 * cgraphclones.c (cgraph_edge::clone): Update.
6197 (duplicate_thunk_for_node): Update.
6198 (cgraph_node::create_clone): Update.
6199 * cgraphunit.c (cgraph_node::analyze): Update.
6200 (cgraph_node::expand_thunk): Update.
6201 * final.c (dump_basic_block_info): Update.
6202 * gimple-streamer-in.c (input_bb): Update.
6203 * gimple-streamer-out.c (output_bb): Update.
6204 * graphite.c (print_global_statistics): Update.
6205 (print_graphite_scop_statistics): Update.
6206 * hsa-brig.c: Include basic-block.h.
6207 * hsa-dump.c: Include basic-block.h.
6208 * hsa-gen.c (T sum_slice): Update.
6209 (convert_switch_statements):Update.
6210 * hsa-regalloc.c: Include basic-block.h.
6211 * ipa-chkp.c (chkp_produce_thunks): Update.
6212 * ipa-cp.c (struct caller_statistics): Update.
6213 (init_caller_stats): Update.
6214 (gather_caller_stats): Update.
6215 (ipcp_cloning_candidate_p): Update.
6216 (good_cloning_opportunity_p): Update.
6217 (get_info_about_necessary_edges): Update.
6218 (dump_profile_updates): Update.
6219 (update_profiling_info): Update.
6220 (update_specialized_profile): Update.
6221 (perhaps_add_new_callers): Update.
6222 (decide_about_value): Update.
6223 (ipa_cp_c_finalize): Update.
6224 * ipa-devirt.c (struct odr_type_warn_count): Update.
6225 (struct decl_warn_count): Update.
6226 (struct final_warning_record): Update.
6227 (possible_polymorphic_call_targets): Update.
6228 (ipa_devirt): Update.
6229 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6230 * ipa-icf.c (sem_function::merge): Update.
6231 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6232 * ipa-inline.c (compute_uninlined_call_time): Update.
6233 (compute_inlined_call_time): Update.
6234 (want_inline_small_function_p): Update.
6235 (want_inline_self_recursive_call_p): Update.
6236 (edge_badness): Update.
6237 (lookup_recursive_calls): Update.
6238 (recursive_inlining): Update.
6239 (inline_small_functions): Update.
6240 (dump_overall_stats): Update.
6241 (dump_inline_stats): Update.
6242 * ipa-profile.c (ipa_profile_generate_summary): Update.
6243 (ipa_propagate_frequency): Update.
6244 (ipa_profile): Update.
6245 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6246 * ipa-utils.c (ipa_merge_profiles): Update.
6247 * loop-doloop.c (doloop_modify): Update.
6248 * loop-unroll.c (report_unroll): Update.
6249 (unroll_loop_runtime_iterations): Update.
6250 * lto-cgraph.c (lto_output_edge): Update.
6251 (lto_output_node): Update.
6252 (input_node): Update.
6253 (input_edge): Update.
6254 (merge_profile_summaries): Update.
6255 * lto-streamer-in.c (input_cfg): Update.
6256 * lto-streamer-out.c (output_cfg): Update.
6257 * mcf.c (create_fixup_graph): Update.
6258 (adjust_cfg_counts): Update.
6259 (sum_edge_counts): Update.
6260 * modulo-sched.c (sms_schedule): Update.
6261 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6262 * predict.c (maybe_hot_count_p): Update.
6263 (probably_never_executed): Update.
6264 (dump_prediction): Update.
6265 (combine_predictions_for_bb): Update.
6266 (propagate_freq): Update.
6267 (handle_missing_profiles): Update.
6268 (counts_to_freqs): Update.
6269 (rebuild_frequencies): Update.
6270 (force_edge_cold): Update.
6271 * predict.h: Include profile-count.h
6272 (maybe_hot_count_p, counts_to_freqs): UPdate.
6273 * print-rtl-function.c: Do not include cfg.h
6274 * print-rtl.c: Include basic-block.h
6275 * profile-count.c: New file.
6276 * profile-count.h: New file.
6277 * profile.c (is_edge_inconsistent): Update.
6278 (correct_negative_edge_counts): Update.
6279 (is_inconsistent): Update.
6280 (set_bb_counts): Update.
6281 (read_profile_edge_counts): Update.
6282 (compute_frequency_overlap): Update.
6283 (compute_branch_probabilities): Update; Initialize and deinitialize
6285 (branch_prob): Update.
6286 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6287 (edge_gcov_count): New.
6288 (bb_gcov_count): New.
6289 * shrink-wrap.c (try_shrink_wrapping): Update.
6290 * tracer.c (better_p): Update.
6291 * trans-mem.c (expand_transaction): Update.
6292 (ipa_tm_insert_irr_call): Update.
6293 (ipa_tm_insert_gettmclone_call): Update.
6294 * tree-call-cdce.c: Update.
6295 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6296 (gimple_duplicate_sese_tail): Update.
6297 (gimple_account_profile_record): Update.
6298 (execute_fixup_cfg): Update.
6299 * tree-inline.c (copy_bb): Update.
6300 (copy_edges_for_bb): Update.
6301 (initialize_cfun): Update.
6302 (freqs_to_counts): Update.
6303 (copy_cfg_body): Update.
6304 (expand_call_inline): Update.
6305 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6306 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6307 (try_unroll_loop_completely): Update.
6308 (try_peel_loop): Update.
6309 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6310 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6311 * tree-ssa-loop-split.c (connect_loops): Update.
6312 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6313 * tree-ssa-reassoc.c (branch_fixup): Update.
6314 * tree-ssa-tail-merge.c (replace_block_by): Update.
6315 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6316 (compute_path_counts): Update.
6317 (update_profile): Update.
6318 (recompute_probabilities): Update.
6319 (update_joiner_offpath_counts): Update.
6320 (estimated_freqs_path): Update.
6321 (freqs_to_counts_path): Update.
6322 (clear_counts_path): Update.
6323 (ssa_fix_duplicate_block_edges): Update.
6324 (duplicate_thread_path): Update.
6325 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6326 (struct switch_conv_info): Update.
6327 * tree-tailcall.c (decrease_profile): Update.
6328 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6329 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6330 * value-prof.c (check_counter): Update.
6331 (gimple_divmod_fixed_value): Update.
6332 (gimple_mod_pow2): Update.
6333 (gimple_mod_subtract): Update.
6334 (gimple_ic_transform): Update.
6335 (gimple_stringop_fixed_value): Update.
6336 * value-prof.h (gimple_ic): Update.
6338 2017-06-02 Carl Love <cel@us.ibm.com>
6340 * config/rs6000/rs6000-c: Add support for built-in functions
6341 vector double vec_doublee (vector signed int);
6342 vector double vec_doublee (vector unsigned int);
6343 vector double vec_doublee (vector float);
6344 vector double vec_doubleh (vector signed int);
6345 vector double vec_doubleh (vector unsigned int);
6346 vector double vec_doubleh (vector float);
6347 vector double vec_doublel (vector signed int);
6348 vector double vec_doublel (vector unsigned int);
6349 vector double vec_doublel (vector float);
6350 vector double vec_doubleo (vector signed int);
6351 vector double vec_doubleo (vector unsigned int);
6352 vector double vec_doubleo (vector float);.
6353 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6354 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6356 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6357 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6358 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6360 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6361 vec_doublel, vec_doubleh.
6362 * doc/extend.texi: Update the built-in documentation file for the
6363 new built-in functions.
6365 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6368 * ipa-inline-analysis.c (free_growth_caches): Set
6369 edge_removal_hook_holder to NULL after removing it.
6371 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6373 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6374 comparision with zero.
6376 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6377 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6378 for early expansion of vec_min and vec_max builtins.
6379 (builtin_function_type): Add min/max unsigned variants to those
6380 identified as having unsigned arguments.
6382 2017-06-02 Olivier Hainque <hainque@adacore.com>
6384 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6386 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6388 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6389 Use VALL_F16 iterator rather than VALL.
6391 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6393 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6394 Emit CBNZ inside loop when doing a strong exchange and comparing
6395 against zero. Generate the CC flags after the loop.
6397 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6399 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6400 (dl_section_ref): New.
6401 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6402 On AIX, append an expression to subtract the size of the
6403 section length to dl_section_ref.
6405 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6408 for early expansion of vector absolute builtins.
6410 2017-06-02 Richard Biener <rguenther@suse.de>
6412 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6413 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6415 2017-06-02 Richard Biener <rguenther@suse.de>
6417 PR tree-optimization/80948
6418 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6421 2017-06-02 Richard Biener <rguenther@suse.de>
6423 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6425 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6426 for inductions as relevant.
6428 2017-06-02 Richard Biener <rguenther@suse.de>
6430 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6431 (vectorizable_induction): ... this. Remove dead code.
6433 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6435 * builtins. (expand_builtin_alloca): Remove second parameter and
6436 infer its value from the first parameter instead.
6437 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6439 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6441 PR rtl-optimization/80903
6442 * loop-doloop.c (add_test): Unshare sequence.
6444 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6446 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6448 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6450 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6452 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6453 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6454 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6455 (xlogue_layout::get_stub_rtx): Make static.
6456 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6457 (xlogue_layout::compute_stub_managed_regs): Rename to...
6458 (xlogue_layout::count_stub_managed_regs): ...this.
6459 (xlogue_layout::is_stub_managed_reg): New function.
6460 (xlogue_layout::m_stub_names): Rename to...
6461 (xlogue_layout::s_stub_names): ...this, make static.
6462 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6463 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6464 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6465 xlogue_layout::s_stub_names): Instantiate statics.
6466 (stub_managed_regs): Remove.
6467 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6468 (disable_call_ms2sysv_xlogues): Rename to...
6469 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6470 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6472 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6473 change after reload_completed.
6474 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6476 (ix86_expand_prologue): Likewise.
6477 (ix86_expand_epilogue): Likewise.
6478 (ix86_expand_split_stack_prologue): Likewise.
6479 (ix86_compute_frame_layout): Remove frame parameter ...
6480 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6481 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6483 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6484 (ix86_frame): Move from here ...
6485 * config/i386/i386.h (ix86_frame): ... to here.
6486 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6487 complete ix86_frame data structure instead. Remove some_ld_name.
6489 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6491 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6492 symbols that hold a DECL_VALUE_EXPR.
6494 2017-06-01 Martin Jambor <mjambor@suse.cz>
6496 PR tree-optimization/80898
6497 * tree-sra.c (process_subtree_disqualification): Removed.
6498 (disqualify_candidate): Do not acll
6499 process_subtree_disqualification.
6500 (subtree_mark_written_and_enqueue): New function.
6501 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6502 RHS has been disqualified and re-queue LHS if necessary. Apart
6503 from that, ignore disqualified RHS.
6505 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6507 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6508 address fetch for z10 or later.
6510 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6512 * config/arc/arc.md (tst_movb): Add guard when splitting.
6514 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6516 * config/arc/arc.c (arc_can_eliminate): Test against
6517 arc_frame_pointer_needed.
6519 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6521 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6522 to prevent store reordering.
6523 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6524 (type): Add block type.
6525 (stack_tie): Define special instruction to be used in
6528 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6530 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6531 constraint. It is not valid for the pattern.
6532 (noncommutative_binary_comparison): Likewise.
6534 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6536 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6539 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6541 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6542 be used by the reg-alloc.
6544 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6546 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6547 reg-alloc when having mul64 or mul32x16 instructions.
6548 (mulsidi3): Likewise.
6549 (umulsidi3): Likewise.
6550 (mulsi32x16): New pattern.
6551 (mulsi64): Likewise.
6552 (mulsidi64): Likewise.
6553 (umulsidi64): Likewise.
6554 (MUL32x16_REG): Define.
6555 (mul64_600): Use MUL32x16_REG.
6556 (mac64_600): Likewise.
6557 (umul64_600): Likewise.
6558 (umac64_600): Likewise.
6560 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6562 * config/arc/arc.md (mulsi3_700): Make it commutative.
6564 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6566 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6568 (*sign_extendsidi2_insn): Likewise for movstosw.
6570 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6572 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6573 the type of the input discriminant value. Convert the
6574 discriminant value of signedness vary.
6576 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6578 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6579 Add to -Wall section.
6581 2017-06-01 Richard Biener <rguenther@suse.de>
6584 * fold-const.c (fold_plusminus_mult_expr): If the factored
6585 factor may be zero use a wrapping type for the inner operation.
6586 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6587 and handle moved defs.
6588 (process_assignment): Properly guard the unary op case. Return a
6589 tri-state indicating that moving the stmt before the call may allow
6590 to continue. Pass through to_move.
6591 (find_tail_calls): Handle moving unrelated defs before
6594 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6597 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6598 splitter result in the canonical way.
6600 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6602 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6603 also for 32bit target. Update insn attributes.
6604 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6606 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6608 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6609 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6610 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6611 (_mm_maskz_min_ss): New intrinsics.
6613 2017-05-31 Martin Liska <mliska@suse.cz>
6615 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6616 Change comment style to one we normally use.
6617 (vectorizable_reduction): Likewise.
6618 (vectorizable_induction): Likewise.
6619 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6620 (vectorizable_call): Likewise.
6621 (vectorizable_simd_clone_call): Likewise.
6622 (vectorizable_conversion): Likewise.
6623 (vectorizable_assignment): Likewise.
6624 (vectorizable_shift): Likewise.
6625 (vectorizable_operation): Likewise.
6626 (vectorizable_store): Likewise.
6627 (vectorizable_load): Likewise.
6628 * tree-vectorizer.h: Likewise.
6630 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6632 * passes.c (emergency_dump_function): New.
6633 * tree-pass.h (emergency_dump_function): Declare.
6634 * plugin.c (plugins_internal_error_function): Remove.
6635 * plugin.h (plugins_internal_error_function): Remove declaration.
6636 * toplev.c (internal_error_function): New static function. Use it...
6637 (general_init): ...here.
6639 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6641 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6642 (arc_rtx_costs): Add costs for new patterns.
6643 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6644 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6646 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6648 * tree-ssa-strlen.c (get_next_strinfo): New function.
6649 (get_stridx_plus_constant): Use it.
6650 (zero_length_string): Likewise.
6651 (adjust_related_strinfos): Likewise.
6652 (adjust_last_stmt): Likewise.
6654 2017-05-31 Richard Biener <rguenther@suse.de>
6657 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6658 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6660 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6662 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6663 loop_vinfo argument and use of dependence distance vectors.
6664 Check instead whether the two references differ only in their
6665 initial value and assume that they have the same alignment if the
6666 difference is a multiple of the vector alignment.
6667 (vect_analyze_data_refs_alignment): Update call accordingly.
6669 2017-05-31 Martin Liska <mliska@suse.cz>
6672 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6674 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6676 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6677 (create_intersect_range_checks): Move from ...
6678 * tree-data-ref.c (create_intersect_range_checks_index)
6679 (create_intersect_range_checks): ... to here.
6680 (create_runtime_alias_checks): New function factored from ...
6681 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6682 here. Call above function.
6683 * tree-data-ref.h (create_runtime_alias_checks): New function.
6685 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6687 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6688 segment length for dr_b and compute it in wide_int.
6690 2017-05-31 Richard Biener <rguenther@suse.de>
6692 PR tree-optimization/80906
6693 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6694 and pass through iv_map.
6695 (copy_bb_and_scalar_dependences): Adjust.
6696 (translate_pending_phi_nodes): Likewise.
6697 (copy_loop_close_phi_args): Handle code-generating IVs instead
6700 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6702 * diagnostic-color.c (color_dict): Add "type-diff".
6703 (parse_gcc_colors): Update comment.
6704 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6705 -fdiagnostics-show-template-tree and -fno-elide-type.
6706 (GCC_COLORS): Add type-diff to example.
6708 (-fdiagnostics-show-template-tree): New.
6709 (-fno-elide-type): New.
6710 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6711 the pp_format_decoder callback. Call any m_format_postprocessor's
6713 (pretty_printer::pretty_printer): Initialize
6714 m_format_postprocessor.
6715 (pretty_printer::~pretty_printer): Delete any
6716 m_format_postprocessor.
6717 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6718 (class format_postprocessor): New class.
6719 (struct pretty_printer::format_decoder): Document the new parameters.
6720 (struct pretty_printer::m_format_postprocessor): New field.
6721 * tree-diagnostic.c (default_tree_printer): Update for new
6722 bool and const char ** params.
6723 * tree-diagnostic.h (default_tree_printer): Likewise.
6725 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6727 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6728 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6730 rs6000_gen_cell_microcode code.
6731 (rs6000_final_prescan_insn): Delete.
6732 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6733 "warn-cell-microcode" entries.
6734 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6735 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6736 throughout. Change cc_reg_not_micro_cr0_operand to
6737 cc_reg_not_cr0_operand throughout.
6738 (*extendhi<mode>2_noload): Delete.
6739 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6740 (mwarn-cell-microcode): Delete.
6741 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6742 -mgen-cell-microcode and -mwarn-cell-microcode.
6744 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6747 * config/i386/constraints.md (Yd): New constraint.
6749 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6750 and (?Yd, r) alternatives. Update insn attributes.
6751 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6752 and (?*Yd, r) alternatives. Update insn attributes.
6753 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6755 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6757 * gimplify.c (gimplify_modify_expr): Don't create a
6758 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6761 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6763 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6765 2017-05-30 Richard Biener <rguenther@suse.de>
6767 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6768 and reduc_def fields.
6769 (STMT_VINFO_REDUC_TYPE): New define.
6770 (STMT_VINFO_REDUC_DEF): Likewise.
6771 (vect_force_simple_reduction): Adjust prototype.
6772 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6773 (vect_is_simple_reduction): Remove check_reduction argument.
6774 (vect_force_simple_reduction): Adjust and set
6775 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6776 (vectorizable_reduction): Do not re-do reduction analysis
6777 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6778 * tree-parloops.c (gather_scalar_reductions): Adjust.
6780 2017-05-30 Richard Biener <rguenther@suse.de>
6783 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6786 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6788 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6789 Introduce unknown_misalignment parameter and remove vf.
6790 (vect_peeling_hash_get_lowest_cost):
6791 Pass unknown_misalignment parameter.
6792 (vect_enhance_data_refs_alignment):
6793 Fix unsupportable data ref treatment.
6795 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6797 * tree-vect-data-refs.c (vect_get_data_access_cost):
6798 Workaround for SLP handling.
6799 (vect_enhance_data_refs_alignment):
6800 Compute costs for doing no peeling at all, compare to the best
6801 peeling costs so far and avoid peeling if cheaper.
6803 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6805 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6806 Return peeling info and set costs to zero for unlimited cost
6808 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6809 with unknown misalignment. Compute and costs for unknown
6810 misalignment, compare them to the costs for known misalignment
6811 and choose the cheapest for peeling.
6813 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6815 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6816 (vect_get_peeling_costs_all_drs): Create function.
6817 (vect_peeling_hash_get_lowest_cost):
6818 Use vect_get_peeling_costs_all_drs.
6819 (vect_peeling_supportable): Create function.
6820 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6822 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6824 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6825 DR_HAS_NEGATIVE_STEP.
6826 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6827 (vect_enhance_data_refs_alignment): Use.
6828 (vect_duplicate_ssa_name_ptr_info): Use.
6829 * tree-vectorizer.h (dr_misalignment): Use.
6830 (known_alignment_for_access_p): Use.
6832 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6835 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6837 (has_section_name): New function.
6839 2017-05-30 Martin Liska <mliska@suse.cz>
6842 * auto-profile.c (get_function_decl_from_block): Fix
6845 2017-05-30 Richard Biener <rguenther@suse.de>
6848 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6850 2017-05-30 Martin Liska <mliska@suse.cz>
6852 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6853 * dumpfile.h (struct dump_file_info): Remove ctors.
6855 2017-05-30 Martin Liska <mliska@suse.cz>
6857 * predict.def: Fix GNU coding style.
6859 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6861 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6862 Mark 'to' argument with ATTRIBUTE_UNUSED.
6864 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6866 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6867 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6868 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6871 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
6873 * doc/install.texi (Options specification): Restore entry of
6874 --enable-sjlj-exceptions.
6876 2017-05-27 Michael Eager <eager@eagercon.com>
6879 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
6881 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
6883 * config/microblaze/microblaze.h
6884 (FIXED_REGISTERS): Update in macro.
6885 (CALL_USED_REGISTERS): Update in macro.
6887 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
6889 * doc/install.texi: Add links to macOS binary distributions.
6891 2017-05-27 Jakub Jelinek <jakub@redhat.com>
6895 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6897 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6899 2017-05-26 Martin Liska <mliska@suse.cz>
6901 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
6903 2017-05-26 Martin Liska <mliska@suse.cz>
6905 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
6906 always leading ';; '.
6907 (dump_bb_info): Likewise.
6908 (brief_dump_cfg): Likewise.
6909 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
6910 * dumpfile.c: Remove usage of TDF_VERBOSE.
6911 * dumpfile.h (enum dump_kind): Likewise.
6912 (dump_gimple_bb_header): Do not use TDF_COMMENT.
6913 * print-tree.c (debug_verbose): Remove.
6914 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
6915 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
6916 * tree-diagnostic.c (default_tree_printer): Replace
6917 TDF_DIAGNOSTIC with TDF_SLIM.
6919 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6921 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
6922 in parameter loop, rather than loop_vinfo.
6923 (create_intersect_range_checks): Ditto.
6924 (vect_create_cond_for_alias_checks): Update call to above functions.
6926 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6928 PR tree-optimization/80815
6929 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
6930 for merging runtime alias checks. Handle negative DR_STEPs.
6932 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6934 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
6936 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
6937 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
6938 out code pruning runtime alias checks.
6939 * tree-data-ref.c (prune_runtime_alias_test_list): New function
6940 factored out from above.
6941 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6943 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6945 (prune_runtime_alias_test_list): New decalaration.
6947 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6949 * tree-vect-data-refs.c (compare_tree): Rename and move ...
6950 * tree-data-ref.c (data_ref_compare_tree): ... to here.
6951 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
6952 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
6953 (operator==, comp_dr_with_seg_len_pair): Ditto.
6954 (vect_prune_runtime_alias_test_list): Ditto.
6956 2017-05-26 Martin Liska <mliska@suse.cz>
6959 * params.def: Bound partial-inlining-entry-probability param.
6961 2017-05-26 Marek Polacek <polacek@redhat.com>
6964 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
6967 2017-05-26 Richard Biener <rguenther@suse.de>
6969 PR tree-optimization/80842
6970 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
6973 2017-05-26 Richard Biener <rguenther@suse.de>
6975 PR tree-optimization/80844
6976 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
6978 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
6980 * doc/md.texi (Machine Constraints): Update x86 family
6981 machine constraints section to match 'config/i386/constraints.md'.
6983 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
6985 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
6987 2017-05-25 Nathan Sidwell <nathan@acm.org>
6989 * doc/invoke.texi (--enable-languages): Update documentation.
6991 2017-05-25 Martin Liska <mliska@suse.cz>
6993 * dumpfile.c: Add TDF_FOLDING.
6994 * dumpfile.h (enum dump_kind): Likewise.
6995 * genmatch.c (dt_simplify::gen_1): Use it.
6997 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6999 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7001 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7003 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7004 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7006 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7008 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7009 * match.pd (X == C): Rewrite it here.
7010 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7011 with_certain_nonzero_bits2): New predicates.
7012 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7014 2017-05-24 Nathan Sidwell <nathan@acm.org>
7016 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7019 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7022 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7024 * config/powerpcspe: New port. Files are copied from the rs6000
7025 port, with "rs6000" in filenames replaced by "powerpcspe".
7027 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7029 PR rtl-optimization/80754
7030 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7032 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7034 * config/sparc/sparc.md (length): Return the correct value for -mflat
7035 sibcalls to match output_sibcall.
7037 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7041 * config/rs6000/rs6000.c (struct machine_function): Add new field
7043 (rs6000_get_separate_components): Init that field, use it.
7044 (rs6000_components_for_bb): Use the field.
7046 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7048 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7050 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7053 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7055 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7058 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7059 address against address_operand predicate.
7060 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7062 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7064 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7065 parameters passed indirectly.
7067 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7069 * config/i386/i386.md (*movdi_internal): Remove SSE4
7070 alternative 18 (?r, *v). Update insn attributes.
7071 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7072 Update insn attributes.
7073 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7074 Update insn attributes.
7075 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7076 alternative 1 (r, v). Remove isa attribute.
7077 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7078 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7079 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7081 2017-05-23 Tom de Vries <tom@codesourcery.com>
7083 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7086 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7088 * cgraphunit.c (symbol_table::process_new_functions): Update.
7089 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7090 (inline_generate_summary): Rename to ...
7091 (ipa_fn_summary_generate): ... this one.
7092 (inline_read_summary): Rename to ...
7093 (ipa_fn_summary_read): ... this one.
7094 (inline_write_summary): Rename to ...
7095 (ipa_fn_summary_write): ... this one.
7096 (inline_free_summary): Rename to ...
7097 (ipa_free_fn_summary): ... this one.
7098 (pass_data_local_fn_summary, pass_local_fn_summary,
7099 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7100 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7101 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7102 make_pass_ipa_fn_summary): New.
7103 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7104 inline_write_summary, inline_free_summary): Remove.
7105 (ipa_free_fn_summary) : New.
7106 * ipa-inline.c (ipa_inline): Update.
7107 (pass_ipa_inline): Do not generate summaries.
7108 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7110 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7111 and add pass_ipa_fn_summary.
7112 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7114 (make_pass_inline_parameters): Remove.
7116 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7118 * omp-low.c (struct omp_context): Remove "default_kind" member.
7121 * omp-offload.c (execute_oacc_device_lower): Remove the
7122 parallelism dimensions function attributes for unparallelized
7123 OpenACC kernels constructs.
7125 2017-05-23 Martin Liska <mliska@suse.cz>
7127 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7129 (cgraph_edge::make_speculative): Likewise.
7130 (cgraph_edge::resolve_speculation): Likewise.
7131 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7132 (cgraph_node::dump): Likewise.
7133 * cgraph.h: Likewise.
7134 * cgraphunit.c (analyze_functions): Likewise.
7135 (symbol_table::compile): Likewise.
7136 * ipa-cp.c (print_all_lattices): Likewise.
7137 (determine_versionability): Likewise.
7138 (initialize_node_lattices): Likewise.
7139 (ipcp_verify_propagated_values): Likewise.
7140 (estimate_local_effects): Likewise.
7141 (update_profiling_info): Likewise.
7142 (create_specialized_node): Likewise.
7143 (perhaps_add_new_callers): Likewise.
7144 (decide_about_value): Likewise.
7145 (decide_whether_version_node): Likewise.
7146 (identify_dead_nodes): Likewise.
7147 (ipcp_store_bits_results): Likewise.
7148 * ipa-devirt.c (dump_targets): Likewise.
7149 (ipa_devirt): Likewise.
7150 * ipa-icf.c (sem_item::dump): Likewise.
7151 (sem_function::equals): Likewise.
7152 (sem_variable::equals): Likewise.
7153 (sem_item_optimizer::read_section): Likewise.
7154 (sem_item_optimizer::execute): Likewise.
7155 (congruence_class::dump): Likewise.
7156 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7157 (dump_inline_summary): Likewise.
7158 (estimate_node_size_and_time): Likewise.
7159 (inline_analyze_function): Likewise.
7160 * ipa-inline-transform.c (inline_call): Likewise.
7161 * ipa-inline.c (report_inline_failed_reason): Likewise.
7162 (want_early_inline_function_p): Likewise.
7163 (edge_badness): Likewise.
7164 (update_edge_key): Likewise.
7165 (inline_small_functions): Likewise.
7166 * ipa-profile.c (ipa_profile): Likewise.
7167 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7168 (ipa_make_edge_direct_to_target): Likewise.
7169 (remove_described_reference): Likewise.
7170 (ipa_impossible_devirt_target): Likewise.
7171 (propagate_controlled_uses): Likewise.
7172 (ipa_print_node_params): Likewise.
7173 (ipcp_transform_function): Likewise.
7174 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7175 (propagate_pure_const): Likewise.
7176 * ipa-reference.c (generate_summary): Likewise.
7177 (read_write_all_from_decl): Likewise.
7178 (propagate): Likewise.
7179 (ipa_reference_read_optimization_summary): Likewise.
7180 * ipa-utils.c (ipa_merge_profiles): Likewise.
7181 * ipa.c (walk_polymorphic_call_targets): Likewise.
7182 (symbol_table::remove_unreachable_nodes): Likewise.
7183 (ipa_single_use): Likewise.
7184 * passes.c (execute_todo): Likewise.
7185 * predict.c (drop_profile): Likewise.
7186 * symtab.c (symtab_node::get_dump_name): New function.
7187 (symtab_node::dump_name): Likewise.
7188 (symtab_node::dump_asm_name): Likewise.
7189 (symtab_node::dump_references): Likewise.
7190 (symtab_node::dump_referring): Likewise.
7191 (symtab_node::dump_base): Likewise.
7192 (symtab_node::debug_symtab): Likewise.
7193 * tree-sra.c (convert_callers_for_node): Likewise.
7194 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7195 * value-prof.c (init_node_map): Likewise.
7197 2017-05-23 Martin Liska <mliska@suse.cz>
7199 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7200 and symtab_node::debug_symtab to symbol_table::debug.
7201 * cgraphunit.c (analyze_functions): Use the renamed function.
7202 (symbol_table::compile): Likewise.
7203 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7204 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7205 * passes.c (execute_todo): Likewise.
7206 * symtab.c (symbol_table::dump): New function.
7207 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7209 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7211 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7212 that nonconst implies exec.
7214 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7216 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7217 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7218 (inline_edge_summary_vec): Turn into ...
7219 (ipa_call_summaries): ... this one.
7220 (redirect_to_unreachable, edge_set_predicate,
7221 evaluate_properties_for_edge, inline_summary_alloc,
7222 reset_ipa_call_summary, reset_inline_summary,
7223 inline_summary_t::duplicate): Update.
7224 (inline_edge_duplication_hook): Turn to ...
7225 (ipa_call_summary_t::duplicate): ... this one.
7226 (inline_edge_removal_hook): Turn to ...
7227 (ipa_call_summary_t::remove): ... this one.
7228 (dump_inline_edge_summary): Turn to ...
7229 (dump_ipa_call_summary): ... this one.
7230 (estimate_function_body_sizes): Update.
7231 (inline_update_callee_summaries): Update.
7232 (remap_edge_change_prob): Update.
7233 (remap_edge_summaries): Update.
7234 (inline_merge_summary): Update.
7235 (do_estimate_edge_time): Update.
7236 (inline_generate_summary): Update.
7237 (inline_read_section): Update.
7238 (inline_read_summary): Update.
7239 (inline_free_summary): Update.
7240 * ipa-inline.c (can_inline_edge_p): Update.
7241 (compute_inlined_call_time): Update.
7242 (want_inline_small_function_p): Update.
7243 (edge_badness): Update.
7244 (early_inliner): Update.
7245 * ipa-inline.h (inline_edge_summary): Turn to ...
7246 (ipa_call_summary): ... this one.
7247 (ipa_call_summary_t): New class.
7248 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7249 (ipa_call_summaries): New.
7250 (inline_edge_summary): Remove.
7251 (estimate_edge_growth): Update.
7252 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7253 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7254 * ipa-split.c (execute_split_functions): Update.
7255 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7257 2017-05-23 Tom de Vries <tom@codesourcery.com>
7259 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7260 attributes): Document rdrand effective target.
7262 2017-05-23 Tom de Vries <tom@codesourcery.com>
7264 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7265 attributes): Sort alphabetically.
7267 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7269 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7271 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7274 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7275 V2DF/V2DI splat into two separate patterns, one that handles
7276 registers, and the other that only handles memory. Drop support
7277 for splatting from a GPR on ISA 2.07 and then splitting the
7278 splat into direct move and splat.
7279 (vsx_splat_<mode>_reg): Likewise.
7280 (vsx_splat_<mode>_mem): Likewise.
7282 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7284 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7286 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7289 * omp-low.c (finish_taskreg_remap): New function.
7290 (finish_taskreg_scan): If unit size of ctx->record_type
7291 is non-constant, unshare the size expression and replace
7292 decls in it with possible outer var refs.
7295 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7296 GOVD_SHARED rather than GOVD_PRIVATE with it.
7297 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7298 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7301 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7302 as last argument to build_outer_var_ref for pointer bases of array
7305 2017-05-19 Martin Sebor <msebor@redhat.com>
7307 * print-tree.c (print_node): Print DECL_READ_P flag.
7309 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7311 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7312 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7313 * cgraph.c: Likewise.
7314 * cgraphunit.c: Likewise.
7315 * gengtype.c: Likewise.
7316 * ipa-cp.c: Likewise.
7317 * ipa-devirt.c: Likewise.
7318 * ipa-icf.c: Likewise.
7319 * ipa-predicate.c: Likewise.
7320 * ipa-profile.c: Likewise.
7321 * ipa-prop.c: Likewise.
7322 * ipa-split.c: Likewise.
7324 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7325 edge_predicate_pool, dump_inline_hints,
7326 inline_summary::account_size_time, redirect_to_unreachable,
7327 edge_set_predicate, set_hint_predicate,
7328 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7329 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7330 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7331 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7332 ipa_call_summary_t::remove, initialize_growth_caches,
7333 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7334 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7335 mark_modified, unmodified_parm_1, unmodified_parm,
7336 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7337 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7338 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7339 will_be_nonconstant_predicate, record_modified_bb_info,
7340 get_minimal_bb, record_modified, param_change_prob,
7341 phi_result_unknown_predicate, predicate_for_phi_result,
7342 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7343 estimate_function_body_sizes, compute_inline_parameters,
7344 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7345 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7346 inline_update_callee_summaries, remap_edge_change_prob,
7347 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7348 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7349 inline_analyze_function, inline_summary_t::insert,
7350 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7351 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7352 inline_free_summary): Move to ipa-fnsummary.h
7353 (predicate_t): Remove.
7354 * ipa-fnsummary.c: New file.
7355 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7356 (enum inline_hints_vals, inline_hints, agg_position_info,
7357 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7358 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7359 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7360 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7361 inline_read_summary, inline_write_summary, inline_free_summary,
7362 inline_analyze_function, initialize_inline_failed,
7363 inline_merge_summary, inline_update_overall_summary,
7364 compute_inline_parameters): Move to ipa-fnsummary.h
7365 * ipa-fnsummary.h: New file.
7366 * ipa-inline-transform.h: Include ipa-inline.h.
7367 * ipa-inline.c: LIkewise.
7369 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7371 * ipa-inline.c (edge_badness): Use inlined_time instead of
7372 inline_summaries->get.
7374 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7376 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7378 2017-05-22 Nathan Sidwell <nathan@acm.org>
7380 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7381 (fdump-lang): Document 'raw' option.
7382 * dumpfile.h (TDI_tu): Delete.
7383 * dumpfile.c (dump_files): Remove translation-unit.
7384 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7386 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7388 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7389 command option from $(AWK) call.
7390 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7391 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7392 [FORMAT]: Remove handling of variable.
7393 * config/avr/t-multilib: Regenerate.
7395 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7397 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7399 (dump_inline_summary): Do not print self_time.
7400 (estimate_function_body_sizes): Do not set self_time.
7401 (compute_inline_parameters): Likewise.
7402 (inline_read_section, inline_write_summary): Do not stream self_time.
7403 * ipa-inline.h (inline_summary): Drop self_time.
7405 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7407 * ipa-inline-analysis.c (account_size_time): Rename to ...
7408 (inline_summary::account_size_time): ... this one.
7409 (reset_ipa_call_summary): Turn to ...
7410 (ipa_call_summary::reset): ... this one.
7411 (reset_inline_summary): Turn to ...
7412 (inline_summary::reset): ... this one.
7413 (inline_summary_t::remove): Update.
7414 (inline_summary_t::duplicate): Update.
7415 (ipa_call_summary_t::remove): Update.
7416 (dump_inline_summary): Update.
7417 (estimate_function_body_sizes): Update.
7418 (compute_inline_parameters): Update.
7419 (estimate_node_size_and_time): Update.
7420 (inline_merge_summary): Update.
7421 (inline_update_overall_summary): Update.
7422 (inline_read_section): Update.
7423 (inline_write_summary): Update.
7424 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7425 add account_size_time and reset member functions.
7426 (ipa_call_summary): Add reset function.
7427 * ipa-predicate.h (predicate::operator &): Constify.
7429 2017-05-22 Richard Biener <rguenther@suse.de>
7431 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7433 2017-05-19 Jason Merrill <jason@redhat.com>
7435 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7437 2017-05-19 Marek Polacek <polacek@redhat.com>
7440 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7441 TYPE_OVERFLOW_WRAPS checks.
7443 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7445 * tree-core.h (enum omp_clause_default_kind): Add
7446 "OMP_CLAUSE_DEFAULT_PRESENT".
7447 * tree-pretty-print.c (dump_omp_clause): Handle it.
7448 * gimplify.c (enum gimplify_omp_var_data): Add
7449 "GOVD_MAP_FORCE_PRESENT".
7450 (gimplify_adjust_omp_clauses_1): Map it to
7451 "GOMP_MAP_FORCE_PRESENT".
7452 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7454 * gimplify.c (oacc_default_clause): Clarify.
7456 2017-05-19 Nathan Sidwell <nathan@acm.org>
7458 LANG_HOOK_REGISTER_DUMPS
7459 * toplev.c (general_init): Call register dump lang hook.
7460 * doc/invoke.texi: Document -fdump-lang option family.
7461 * dumpfile.c (dump_files): Remove class dump here.
7462 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7463 * dumpfile.h (tree_dump_index): Remove TDI_class.
7464 * langhooks-def.h (lhd_register_dumps): Declare.
7465 (LANG_HOOKS_REGISTER_DUMPS): Define.
7466 (LANG_HOOKS_INITIALIZER): Add it.
7467 * langhooks.c (lhd_register_dumps): Define.
7468 * langhooks.h (struct lang_hooks): Add register_dumps.
7470 2017-05-19 Nathan Sidwell <nathan@acm.org>
7472 * context.h (context::set_passes): New.
7473 * context.c (context::context): Do not create pass manager.
7474 * toplev.c (general_init): Create pass manager here.
7476 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7478 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7479 use this splitter if two add or or instructions would also work for
7480 the constant we want to generate.
7482 2017-05-19 Richard Biener <rguenther@suse.de>
7485 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7486 predicate evaluation.
7488 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7490 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7492 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7495 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7497 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7498 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7499 (inline_edge_summary_vec): Turn into ...
7500 (ipa_call_summaries): ... this one.
7501 (redirect_to_unreachable, edge_set_predicate,
7502 evaluate_properties_for_edge, inline_summary_alloc,
7503 reset_ipa_call_summary, reset_inline_summary,
7504 inline_summary_t::duplicate): Update.
7505 (inline_edge_duplication_hook): Turn to ...
7506 (ipa_call_summary_t::duplicate): ... this one.
7507 (inline_edge_removal_hook): Turn to ...
7508 (ipa_call_summary_t::remove): ... this one.
7509 (dump_inline_edge_summary): Turn to ...
7510 (dump_ipa_call_summary): ... this one.
7511 (estimate_function_body_sizes): Update.
7512 (inline_update_callee_summaries): Update.
7513 (remap_edge_change_prob): Update.
7514 (remap_edge_summaries): Update.
7515 (inline_merge_summary): Update.
7516 (do_estimate_edge_time): Update.
7517 (inline_generate_summary): Update.
7518 (inline_read_section): Update.
7519 (inline_read_summary): Update.
7520 (inline_free_summary): Update.
7521 * ipa-inline.c (can_inline_edge_p): Update.
7522 (compute_inlined_call_time): Update.
7523 (want_inline_small_function_p): Update.
7524 (edge_badness): Update.
7525 (early_inliner): Update.
7526 * ipa-inline.h (inline_edge_summary): Turn to ...
7527 (ipa_call_summary): ... this one.
7528 (ipa_call_summary_t): New class.
7529 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7530 (ipa_call_summaries): New.
7531 (inline_edge_summary): Remove.
7532 (estimate_edge_growth): Update.
7533 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7534 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7535 * ipa-split.c (execute_split_functions): Update.
7536 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7538 2017-05-19 Richard Biener <rguenther@suse.de>
7541 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7543 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7545 * config/rs6000/rs6000.c (struct machine_function): Add field
7546 fpr_is_wrapped_separately.
7547 (rs6000_get_separate_components): Use 64 components. Handle the
7549 (rs6000_components_for_bb): Handle the FPR components.
7550 (rs6000_emit_prologue_components): Handle the FPR components.
7551 (rs6000_emit_epilogue_components): Handle the FPR components.
7552 (rs6000_set_handled_components): Handle the FPR components.
7553 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7554 that are already separately shrink-wrapped.
7555 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7556 that are already separately shrink-wrapped.
7558 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7561 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7564 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7565 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7566 cases where the register allocator uses a move and an offsettable
7567 memory operation to/from a FPR register on ISA 2.06/2.07.
7568 (define_peephole2 for Altivec d-form store): Likewise.
7570 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7573 * config/i386/mmx.md (*mov<mode>_internal): Enable
7574 alternatives 11, 12, 13 and 14 also for 32bit targets.
7575 Remove alternatives 15, 16, 17 and 18.
7576 * config/i386/sse.md (vec_concatv2di): Change
7577 alternative (!x, *y) to (x, ?!*Yn).
7579 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7581 * dumpfile.h (enum dump_kind): Remove stray comma.
7583 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7585 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7586 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7587 predicate::num_conditions
7588 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7589 (CHANGED): turn into predicate::changed.
7590 (agg_position_info): Move to ipa-predicate.h
7591 (add_condition, predicate::add_clause, predicate::operator &=,
7592 predicate::or_with, predicate::evaluate, predicate::probability,
7593 dump_condition, dump_clause, predicate::dump,
7594 predicate::remap_after_duplication, predicate::remap_after_inlining,
7595 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7596 (evaluate_conditions_for_known_args): Update.
7597 (set_cond_stmt_execution_predicate): Update.
7598 * ipa-inline.h: Include ipa-predicate.h
7599 (condition, inline_param_summary, conditions, agg_position_info,
7600 predicate): Move to ipa-predicate.h
7601 * ipa-predicate.c: New file.
7602 * ipa-predicate.h: New file.
7604 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7606 * final.c (leaf_function_p): Check we are not in a sequence.
7608 2017-05-18 Martin Liska <mliska@suse.cz>
7610 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7611 * dumpfile.c (dump_register): Use new enum dump_kind.
7612 (get_dump_file_name): Likewise.
7613 (dump_enable_all): Likewise.
7614 (dump_switch_p_1): Likewise.
7615 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7616 * dumpfile.h (enum dump_kind): New enum type.
7617 (struct dump_file_info): Create constructor and
7618 format fields and comments.
7619 * passes.c (pass_manager::register_one_dump_file):
7621 * statistics.c (statistics_early_init): Likewise.
7622 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7623 TDF_TREE with TDF_SLIM.
7624 (gather_memory_references_ref): Likewise.
7626 2017-05-18 Martin Liska <mliska@suse.cz>
7628 * vec.h (struct vnull): Use it.
7630 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7632 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7633 (true_predicate, false_predicate, true_predicate_p,
7634 false_predicate_p): Remove.
7635 (single_cond_predicate, not_inlined_predicate): Turn to member function
7637 (add_condition): Update.
7638 (add_clause): Turn to...
7639 (predicate::add_clause): ... this one; update; allow passing NULL
7641 (and_predicates): Turn to ...
7642 (predicate::operator &=): ... this one.
7643 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7644 (or_predicates): Turn to ...
7645 (predicate::or_with): ... this one.
7646 (evaluate_predicate): Turn to ...
7647 (predicate::evaluate): ... this one.
7648 (predicate_probability): Turn to ...
7649 (predicate::probability): ... this one.
7650 (dump_condition): Update.
7651 (dump_predicate): Turn to ...
7652 (predicate::dump): ... this one.
7653 (account_size_time): Update.
7654 (edge_set_predicate): Update.
7655 (set_hint_predicate): UPdate.
7656 (evaluate_conditions_for_known_args): Update.
7657 (evaluate_properties_for_edge): Update.
7658 (remap_predicate_after_duplication): Turn to...
7659 (predicate::remap_after_duplication): ... this one.
7660 (remap_hint_predicate_after_duplication): Update.
7661 (inline_summary_t::duplicate): UPdate.
7662 (dump_inline_edge_summary): Update.
7663 (dump_inline_summary): Update.
7664 (set_cond_stmt_execution_predicate): Update.
7665 (set_switch_stmt_execution_predicate): Update.
7666 (compute_bb_predicates): Update.
7667 (will_be_nonconstant_expr_predicate): Update.
7668 (will_be_nonconstant_predicate): Update.
7669 (phi_result_unknown_predicate): Update.
7670 (predicate_for_phi_result): Update.
7671 (array_index_predicate): Update.
7672 (estimate_function_body_sizes): Update.
7673 (estimate_node_size_and_time): Update.
7674 (estimate_ipcp_clone_size_and_time): Update.
7675 (remap_predicate): Rename to ...
7676 (predicate::remap_after_inlining): ... this one.
7677 (remap_hint_predicate): Update.
7678 (inline_merge_summary): Update.
7679 (inline_update_overall_summary): Update.
7680 (estimate_size_after_inlining): Update.
7681 (read_predicate): Rename to ...
7682 (predicate::stream_in): ... this one.
7683 (read_inline_edge_summary): Update.
7684 (write_predicate): Rename to ...
7685 (predicate::stream_out): ... this one.
7686 (write_inline_edge_summary): Update.
7687 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7688 (clause_t): Turn to uint32_t
7689 (predicate): Turn to class; implement constructor and operators
7691 (size_time_entry): Update.
7692 (inline_summary): Update.
7693 (inline_edge_summary): Update.
7695 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7697 * fold-const.c (fold_binary_loc): Move transformation...
7698 * match.pd (C - X CMP X): ... here.
7700 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7702 * config/sparc/sparc.c (sparc_option_override): Set function
7703 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7704 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7706 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7708 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7710 2017-05-18 Marek Polacek <polacek@redhat.com>
7713 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7714 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7715 gimple_assign_load_p.
7717 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7720 * real.c (do_compare): Give decimal_do_compare preference over
7721 comparing just the signs.
7723 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7725 * doc/md.texi (Canonicalization of Instructions): Describe the
7726 canonical form of instructions that inherently set a condition
7729 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7732 * tree-cfg.c: Move deletion of unreachable case statements to after
7733 the merging of consecutive case labels.
7735 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7737 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7738 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7739 restoring of callee-saved registers.
7741 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7743 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7744 * config/visium/visium.c (single_set_and_flags): Likewise.
7745 * config/visium/visium.md (Substitutions): Likewise.
7747 2017-05-17 Martin Liska <mliska@suse.cz>
7749 * cfg.c: Introduce dump_flags_t type and
7750 use it instead of int type.
7752 * cfghooks.c: Likewise.
7753 * cfghooks.h (struct cfg_hooks): Likewise.
7754 * cfgrtl.c: Likewise.
7755 * cfgrtl.h: Likewise.
7756 * cgraph.c (cgraph_node::get_body): Likewise.
7757 * coretypes.h: Likewise.
7758 * domwalk.c: Likewise.
7759 * domwalk.h: Likewise.
7760 * dumpfile.c (struct dump_option_value_info): Likewise.
7761 (dump_enable_all): Likewise.
7762 (dump_switch_p_1): Likewise.
7763 (opt_info_switch_p): Likewise.
7764 * dumpfile.h (enum tree_dump_index): Likewise.
7765 (struct dump_file_info): Likewise.
7766 * genemit.c: Likewise.
7767 * generic-match-head.c: Likewise.
7768 * gengtype.c (open_base_files): Likewise.
7769 * gimple-pretty-print.c: Likewise.
7770 * gimple-pretty-print.h: Likewise.
7771 * graph.c (print_graph_cfg): Likewise.
7772 * graphite-scop-detection.c (dot_all_sese): Likewise.
7773 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7774 * loop-unroll.c (report_unroll): Likewise.
7775 * passes.c (pass_manager::register_one_dump_file): Likewise.
7776 * print-tree.c: Likewise.
7777 * statistics.c: Likewise.
7778 * tree-cfg.c: Likewise.
7779 * tree-cfg.h: Likewise.
7780 * tree-dfa.c: Likewise.
7781 * tree-dfa.h: Likewise.
7782 * tree-dump.c (dump_function): Likewise.
7783 * tree-dump.h (struct dump_info): Likewise.
7784 * tree-pretty-print.c: Likewise.
7785 * tree-pretty-print.h: Likewise.
7786 * tree-ssa-live.c: Likewise.
7787 * tree-ssa-live.h: Likewise.
7788 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7789 * tree-vect-loop.c: Likewise.
7790 * tree-vect-slp.c: Likewise.
7792 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7793 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7795 PR tree-optimization/80457
7796 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7797 of all arguments to a statement as scalar_to_vec operations.
7798 (vectorizable_call): Adjust call to vect_model_simple_cost for
7800 (vectorizable_conversion): Likewise.
7801 (vectorizable_assignment): Likewise.
7802 (vectorizable_shift): Likewise.
7803 (vectorizable_operation): Likewise.
7804 (vectorizable_comparison): Likewise.
7805 (vect_is_simple_cond): Record the def types for operands.
7806 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7807 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7808 for statement argument count.
7810 2017-05-16 Carl Love <cel@us.ibm.com>
7812 * config/rs6000/rs6000-c: Add support for built-in functions
7813 vector unsigned long long vec_bperm (vector unsigned long long,
7814 vector unsigned char)
7815 vector signed long long vec_mule (vector signed int,
7817 vector unsigned long long vec_mule (vector unsigned int,
7818 vector unsigned int)
7819 vector signed long long vec_mulo (vector signed int,
7821 vector unsigned long long vec_mulo (vector unsigned int,
7822 vector unsigned int)
7823 vector signed char vec_sldw (vector signed char,
7826 vector unsigned char vec_sldw (vector unsigned char,
7827 vector unsigned char,
7829 vector signed short vec_sldw (vector signed short,
7830 vector signed short,
7832 vector unsigned short vec_sldw (vector unsigned short,
7833 vector unsigned short,
7835 vector signed int vec_sldw (vector signed int,
7838 vector unsigned int vec_sldw (vector unsigned int,
7839 vector unsigned int,
7841 vector signed long long vec_sldw (vector signed long long,
7842 vector signed long long,
7844 vector unsigned long long vec_sldw (vector unsigned long long,
7845 vector unsigned long long,
7847 * config/rs6000/rs6000-c: Add support for built-in functions
7848 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7849 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7850 * doc/extend.texi: Update the built-in documentation for the
7851 new built-in functions.
7853 2017-05-16 Marek Polacek <polacek@redhat.com>
7857 * tree.c (save_expr): Don't fold the expression.
7859 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7861 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7862 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7863 and (?*y,m). Update insn attributes.
7865 2017-05-16 Martin Liska <mliska@suse.cz>
7867 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7868 flags argument of print_gimple_stmt, print_gimple_expr,
7869 print_generic_stmt and print_generic_expr.
7870 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7871 * coretypes.h: Likewise.
7872 * except.c (dump_eh_tree): Likewise.
7873 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7874 * gimple-pretty-print.h: Likewise.
7875 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
7876 (backprop::push_to_worklist): Likewise.
7877 (backprop::pop_from_worklist): Likewise.
7878 (backprop::process_use): Likewise.
7879 (backprop::intersect_uses): Likewise.
7880 (note_replacement): Likewise.
7881 * gimple-ssa-store-merging.c
7882 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
7883 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7884 (pass_store_merging::execute): Likewise.
7885 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
7886 (ssa_base_cand_dump_callback): Likewise.
7887 (dump_incr_vec): Likewise.
7888 (replace_refs): Likewise.
7889 (replace_mult_candidate): Likewise.
7890 (create_add_on_incoming_edge): Likewise.
7891 (create_phi_basis): Likewise.
7892 (insert_initializers): Likewise.
7893 (all_phi_incrs_profitable): Likewise.
7894 (introduce_cast_before_cand): Likewise.
7895 (replace_one_candidate): Likewise.
7896 * gimplify.c (gimplify_expr): Likewise.
7897 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
7898 (set_rename): Likewise.
7899 (rename_uses): Likewise.
7900 (copy_loop_phi_nodes): Likewise.
7901 (add_close_phis_to_merge_points): Likewise.
7902 (copy_loop_close_phi_args): Likewise.
7903 (copy_cond_phi_args): Likewise.
7904 (graphite_copy_stmts_from_block): Likewise.
7905 (translate_pending_phi_nodes): Likewise.
7906 * graphite-poly.c (print_pdr): Likewise.
7907 (dump_gbb_cases): Likewise.
7908 (dump_gbb_conditions): Likewise.
7909 (print_scop_params): Likewise.
7910 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
7911 (build_cross_bb_scalars_use): Likewise.
7912 (gather_bbs::before_dom_children): Likewise.
7913 * hsa-dump.c (dump_hsa_immed): Likewise.
7914 * ipa-cp.c (print_ipcp_constant_value): Likewise.
7915 (get_replacement_map): Likewise.
7916 * ipa-inline-analysis.c (dump_condition): Likewise.
7917 (estimate_function_body_sizes): Likewise.
7918 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
7919 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
7920 * ipa-prop.c (ipa_dump_param): Likewise.
7921 (ipa_print_node_jump_functions_for_edge): Likewise.
7922 (ipa_modify_call_arguments): Likewise.
7923 (ipa_modify_expr): Likewise.
7924 (ipa_dump_param_adjustments): Likewise.
7925 (ipa_dump_agg_replacement_values): Likewise.
7926 (ipcp_modif_dom_walker::before_dom_children): Likewise.
7927 * ipa-pure-const.c (check_stmt): Likewise.
7928 (pass_nothrow::execute): Likewise.
7929 * ipa-split.c (execute_split_functions): Likewise.
7930 * omp-offload.c (dump_oacc_loop_part): Likewise.
7931 (dump_oacc_loop): Likewise.
7932 * trans-mem.c (tm_log_emit): Likewise.
7933 (tm_memopt_accumulate_memops): Likewise.
7934 (dump_tm_memopt_set): Likewise.
7935 (dump_tm_memopt_transform): Likewise.
7936 * tree-cfg.c (gimple_verify_flow_info): Likewise.
7937 (print_loop): Likewise.
7938 * tree-chkp-opt.c (chkp_print_addr): Likewise.
7939 (chkp_gather_checks_info): Likewise.
7940 (chkp_get_check_result): Likewise.
7941 (chkp_remove_check_if_pass): Likewise.
7942 (chkp_use_outer_bounds_if_possible): Likewise.
7943 (chkp_reduce_bounds_lifetime): Likewise.
7944 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
7945 (chkp_mark_completed_bounds): Likewise.
7946 (chkp_register_incomplete_bounds): Likewise.
7947 (chkp_mark_invalid_bounds): Likewise.
7948 (chkp_maybe_copy_and_register_bounds): Likewise.
7949 (chkp_build_returned_bound): Likewise.
7950 (chkp_get_bound_for_parm): Likewise.
7951 (chkp_build_bndldx): Likewise.
7952 (chkp_get_bounds_by_definition): Likewise.
7953 (chkp_generate_extern_var_bounds): Likewise.
7954 (chkp_get_bounds_for_decl_addr): Likewise.
7955 * tree-chrec.c (chrec_apply): Likewise.
7956 * tree-data-ref.c (dump_data_reference): Likewise.
7957 (dump_subscript): Likewise.
7958 (dump_data_dependence_relation): Likewise.
7959 (analyze_overlapping_iterations): Likewise.
7960 * tree-inline.c (expand_call_inline): Likewise.
7961 (tree_function_versioning): Likewise.
7962 * tree-into-ssa.c (dump_defs_stack): Likewise.
7963 (dump_currdefs): Likewise.
7964 (dump_names_replaced_by): Likewise.
7965 (dump_update_ssa): Likewise.
7966 (update_ssa): Likewise.
7967 * tree-object-size.c (pass_object_sizes::execute): Likewise.
7968 * tree-parloops.c (build_new_reduction): Likewise.
7969 (try_create_reduction_list): Likewise.
7970 (ref_conflicts_with_region): Likewise.
7971 (oacc_entry_exit_ok_1): Likewise.
7972 (oacc_entry_exit_single_gang): Likewise.
7973 * tree-pretty-print.h: Likewise.
7974 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
7975 (get_scalar_evolution): Likewise.
7976 (add_to_evolution): Likewise.
7977 (get_loop_exit_condition): Likewise.
7978 (analyze_evolution_in_loop): Likewise.
7979 (analyze_initial_condition): Likewise.
7980 (analyze_scalar_evolution): Likewise.
7981 (instantiate_scev): Likewise.
7982 (number_of_latch_executions): Likewise.
7983 (gather_chrec_stats): Likewise.
7984 (final_value_replacement_loop): Likewise.
7985 (scev_const_prop): Likewise.
7986 * tree-sra.c (dump_access): Likewise.
7987 (disqualify_candidate): Likewise.
7988 (create_access): Likewise.
7990 (maybe_add_sra_candidate): Likewise.
7991 (create_access_replacement): Likewise.
7992 (analyze_access_subtree): Likewise.
7993 (analyze_all_variable_accesses): Likewise.
7994 (sra_modify_assign): Likewise.
7995 (initialize_constant_pool_replacements): Likewise.
7996 (find_param_candidates): Likewise.
7997 (decide_one_param_reduction): Likewise.
7998 (replace_removed_params_ssa_names): Likewise.
7999 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8000 * tree-ssa-copy.c (dump_copy_of): Likewise.
8001 (copy_prop_visit_cond_stmt): Likewise.
8002 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8003 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8004 (record_equivalences_from_stmt): Likewise.
8005 * tree-ssa-dse.c (compute_trims): Likewise.
8006 (delete_dead_call): Likewise.
8007 (delete_dead_assignment): Likewise.
8008 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8009 (forward_propagate_into_cond): Likewise.
8010 (pass_forwprop::execute): Likewise.
8011 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8012 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8014 (move_computations_worker): Likewise.
8015 (execute_sm): Likewise.
8016 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8017 (remove_exits_and_undefined_stmts): Likewise.
8018 (remove_redundant_iv_tests): Likewise.
8019 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8020 (adjust_iv_update_pos): Likewise.
8021 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8022 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8023 (value_replacement): Likewise.
8024 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8025 * tree-ssa-pre.c (print_pre_expr): Likewise.
8026 (get_representative_for): Likewise.
8027 (create_expression_by_pieces): Likewise.
8028 (insert_into_preds_of_block): Likewise.
8029 (eliminate_insert): Likewise.
8030 (eliminate_dom_walker::before_dom_children): Likewise.
8031 (eliminate): Likewise.
8032 (remove_dead_inserted_code): Likewise.
8033 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8034 * tree-ssa-reassoc.c (get_rank): Likewise.
8035 (eliminate_duplicate_pair): Likewise.
8036 (eliminate_plus_minus_pair): Likewise.
8037 (eliminate_not_pairs): Likewise.
8038 (undistribute_ops_list): Likewise.
8039 (eliminate_redundant_comparison): Likewise.
8040 (update_range_test): Likewise.
8041 (optimize_range_tests_var_bound): Likewise.
8042 (optimize_vec_cond_expr): Likewise.
8043 (rewrite_expr_tree): Likewise.
8044 (rewrite_expr_tree_parallel): Likewise.
8045 (linearize_expr): Likewise.
8046 (break_up_subtract): Likewise.
8047 (linearize_expr_tree): Likewise.
8048 (attempt_builtin_powi): Likewise.
8049 (attempt_builtin_copysign): Likewise.
8050 (transform_stmt_to_copy): Likewise.
8051 (transform_stmt_to_multiply): Likewise.
8052 (dump_ops_vector): Likewise.
8053 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8054 (print_scc): Likewise.
8055 (set_ssa_val_to): Likewise.
8056 (visit_reference_op_store): Likewise.
8057 (visit_use): Likewise.
8058 (sccvn_dom_walker::before_dom_children): Likewise.
8059 (run_scc_vn): Likewise.
8060 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8062 (expr_hash_elt::print): Likewise.
8063 (const_and_copies::pop_to_marker): Likewise.
8064 (const_and_copies::record_const_or_copy_raw): Likewise.
8065 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8066 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8067 (dump_predicates): Likewise.
8068 (find_uninit_use): Likewise.
8069 (warn_uninitialized_phi): Likewise.
8070 (pass_late_warn_uninitialized::execute): Likewise.
8071 * tree-ssa.c (verify_vssa): Likewise.
8072 (verify_ssa): Likewise.
8073 (maybe_optimize_var): Likewise.
8074 * tree-vrp.c (dump_value_range): Likewise.
8075 (dump_all_value_ranges): Likewise.
8076 (dump_asserts_for): Likewise.
8077 (register_edge_assert_for_2): Likewise.
8078 (vrp_visit_cond_stmt): Likewise.
8079 (vrp_visit_switch_stmt): Likewise.
8080 (vrp_visit_stmt): Likewise.
8081 (vrp_visit_phi_node): Likewise.
8082 (simplify_cond_using_ranges_1): Likewise.
8083 (fold_predicate_in): Likewise.
8084 (evrp_dom_walker::before_dom_children): Likewise.
8085 (evrp_dom_walker::push_value_range): Likewise.
8086 (evrp_dom_walker::pop_value_range): Likewise.
8087 (execute_early_vrp): Likewise.
8089 2017-05-16 Richard Biener <rguenther@suse.de>
8091 * dwarf2out.c (loc_list_from_tree_1): Do not create
8092 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8094 2017-05-16 Richard Biener <rguenther@suse.de>
8096 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8098 (note_variable_value_in_expr): If we resolved the decl ref
8099 do not push to the stack.
8101 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8103 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8104 operations in fast-math mode.
8105 (vaddq_f16): Likewise.
8106 (vmul_f16): Likewise.
8107 (vmulq_f16): Likewise.
8108 (vsub_f16): Likewise.
8109 (vsubq_f16): Likewise.
8110 * config/arm/neon.md (add<mode>3): New.
8112 (fma:<VH:mode>3): New. Also remove outdated comment.
8115 2017-05-16 Martin Liska <mliska@suse.cz>
8119 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8120 (odr_types_equivalent_p): Likewise.
8122 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8124 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8126 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8129 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8130 non-interunit SSE move alternatives with '?'.
8131 (zero-extendsidi peephole2): New peephole to skip intermediate
8132 general register in SSE zero-extend sequence.
8134 2017-05-15 Jeff Law <law@redhat.com>
8136 * reorg.c (relax_delay_slots): Create a new variable to hold
8137 the temporary target rather than clobbering TARGET_LABEL.
8139 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8140 missing argument to extract_bit_field call.
8141 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8143 2017-05-15 Martin Liska <mliska@suse.cz>
8146 * gcc.c (process_command): Do not allow empty argument of -o option.
8148 2017-05-15 Renlin Li <renlin.li@arm.com>
8150 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8151 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8152 * config/aarch64/constraints.md (Usf): Add long call check.
8153 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8154 (call_value): Likewise.
8155 (sibcall): Likewise.
8156 (sibcall_value): Likewise.
8158 (call_value_insn): New.
8159 (sibcall_insn): Update rtx pattern.
8160 (sibcall_value_insn): Likewise.
8161 (call_internal): Remove.
8162 (call_value_internal): Likewise.
8163 (sibcall_internal): Likewise.
8164 (sibcall_value_internal): Likewise.
8165 (call_reg): Likewise.
8166 (call_symbol): Likewise.
8167 (call_value_reg): Likewise.
8168 (call_value_symbol): Likewise.
8170 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8173 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8175 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8177 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8178 compatible with CCGOCmode and with CCZmode.
8180 2017-05-14 Martin Sebor <msebor@redhat.com>
8183 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8184 (gimple_fold_builtin_snprintf): Same.
8185 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8186 (gimple_fold_builtin_snprintf): Same.
8187 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8189 (is_call_safe): New function.
8190 (try_substitute_return_value): Call it.
8191 (try_simplify_call): New function.
8192 (pass_sprintf_length::handle_gimple_call): Call it.
8194 2017-05-14 Martin Sebor <msebor@redhat.com>
8197 * builtins.c (expand_builtin_stpncpy): Simplify.
8199 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8201 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8202 * config/i386/i386.h
8203 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8204 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8205 (struct machine_function): Add new members call_ms2sysv,
8206 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8207 (struct machine_frame_state): New fields sp_realigned and
8208 sp_realigned_offset.
8209 * config/i386/i386.c
8210 (enum xlogue_stub): New enum.
8211 (enum xlogue_stub_sets): New enum.
8212 (class xlogue_layout): New class.
8213 (struct ix86_frame): New fields stack_realign_allocate_offset,
8214 stack_realign_offset and outlined_save_offset. Modify comments to
8215 detail stack layout when using out-of-line stubs.
8216 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8217 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8218 -mcall-ms2sysv-xlogues.
8219 (stub_managed_regs): New static variable.
8220 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8221 registers managed by out-of-line stub.
8222 (disable_call_ms2sysv_xlogues): New function.
8223 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8224 m->call_ms2sysv when appropriate and compute frame layout for
8226 (sp_valid_at, fp_valid_at): New inline functions.
8227 (choose_basereg): New function.
8228 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8230 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8231 Use align parameter of choose_baseaddr to generated aligned SSE movs
8233 (pro_epilogue_adjust_stack): Modify to track
8234 machine_frame_state::sp_realigned.
8235 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8236 (ix86_nsaved_sseregs): Likewise.
8237 (ix86_emit_save_regs): Likewise.
8238 (ix86_emit_save_regs_using_mov): Likewise.
8239 (ix86_emit_save_sse_regs_using_mov): Likewise.
8240 (get_scratch_register_on_entry): Likewise.
8241 (gen_frame_set): New function.
8242 (gen_frame_load): Likewise.
8243 (gen_frame_store): Likewise.
8244 (emit_outlined_ms2sysv_save): Likewise.
8245 (emit_outlined_ms2sysv_restore): Likewise.
8246 (ix86_expand_prologue): Modify stack re-alignment code and call
8247 emit_outlined_ms2sysv_save when appropriate.
8248 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8249 parameter rtx_insn *insn, which allows the function to be used to only
8251 (ix86_expand_epilogue): Modify validity checks of frame and stack
8252 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8253 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8254 * config/i386/predicates.md
8255 (save_multiple): New predicate.
8256 (restore_multiple): Likewise.
8257 * config/i386/sse.md
8258 (save_multiple<mode>): New pattern.
8259 (save_multiple_realign<mode>): Likewise.
8260 (restore_multiple<mode>): Likewise.
8261 (restore_multiple_and_return<mode>): Likewise.
8262 (restore_multiple_leave_return<mode>): Likewise.
8263 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8265 2017-05-14 Julia Koval <julia.koval@intel.com>
8267 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8268 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8269 (__builtin_ia32_xsetbv): New builtins.
8270 * config/i386/i386.c (ix86_expand_special_args_builtin):
8272 (ix86_expand_builtin): Special expand for new intrinsics.
8273 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8274 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8275 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8277 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8279 * cfganal.c (inverted_post_order_compute): Change argument type
8281 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8282 * df-core.c (rest_of_handle_df_initialize): Adjust.
8283 (rest_of_handle_df_finish): Likewise.
8284 (df_analyze_1): Likewise.
8285 (df_analyze): Likewise.
8286 (loop_inverted_post_order_compute): Change argument to be a vec *.
8287 (df_analyze_loop): Adjust.
8288 (df_get_n_blocks): Likewise.
8289 (df_get_postorder): Likewise.
8290 * df.h (struct df_d): Change field to be a vec.
8291 * lcm.c (compute_laterin): Adjust.
8292 (compute_available): Likewise.
8293 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8294 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8295 * tree-ssa-pre.c (compute_antic): Likewise.
8297 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8299 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8300 (depth_first_search::depth_first_search): Change structure init
8301 function to this constructor.
8302 (depth_first_search::add_bb): Rename function to this member.
8303 (depth_first_search::execute): Likewise.
8304 (flow_dfs_compute_reverse_finish): Adjust.
8306 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8308 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8309 (longest_simple_path): Likewise.
8310 * shrink-wrap.c (spread_components): Likewise.
8311 (disqualify_problematic_components): Likewise.
8312 (emit_common_heads_for_components): Likewise.
8313 (emit_common_tails_for_components): Likewise.
8314 (insert_prologue_epilogue_for_components): Likewise.
8316 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8318 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8321 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8323 * df-core.c (df_set_blocks): Start using auto_bitmap.
8324 (df_compact_blocks): Likewise.
8325 * df-problems.c (df_rd_confluence_n): Likewise.
8326 * df-scan.c (df_insn_rescan_all): Likewise.
8327 (df_process_deferred_rescans): Likewise.
8328 (df_update_entry_block_defs): Likewise.
8329 (df_update_exit_block_uses): Likewise.
8330 (df_entry_block_bitmap_verify): Likewise.
8331 (df_exit_block_bitmap_verify): Likewise.
8332 (df_scan_verify): Likewise.
8333 * lra-constraints.c (lra_constraints): Likewise.
8334 (undo_optional_reloads): Likewise.
8335 (lra_undo_inheritance): Likewise.
8336 * lra-remat.c (calculate_gen_cands): Likewise.
8337 (do_remat): Likewise.
8338 * lra-spills.c (assign_spill_hard_regs): Likewise.
8339 (spill_pseudos): Likewise.
8340 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8341 (bitmap_set_subtract_values): Likewise.
8343 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8345 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8346 management with auto_bitmap.
8347 (fix_inter_tick): Likewise.
8348 (fix_recovery_deps): Likewise.
8349 * ira.c (add_store_equivs): Likewise.
8350 (find_moveable_pseudos): Likewise.
8351 (split_live_ranges_for_shrink_wrap): Likewise.
8352 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8353 (rtx_reuse_manager::seen_def_p): Likewise.
8354 (rtx_reuse_manager::set_seen_def): Likewise.
8355 * print-rtl.h (class rtx_reuse_manager): Likewise.
8357 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8359 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8361 (migrate_btr_def): Likewise.
8362 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8363 * df-core.c (loop_post_order_compute): Likewise.
8364 (loop_inverted_post_order_compute): Likewise.
8365 * hsa-common.h: Likewise.
8366 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8367 * init-regs.c (initialize_uninitialized_regs): Likewise.
8368 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8369 (inline_small_functions): Likewise.
8370 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8371 * ira.c (combine_and_move_insns): Likewise.
8372 (build_insn_chain): Likewise.
8373 * loop-invariant.c (find_invariants): Likewise.
8374 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8375 * predict.c (tree_predict_by_opcode): Likewise.
8376 (predict_paths_leading_to): Likewise.
8377 (predict_paths_leading_to_edge): Likewise.
8378 (estimate_loops_at_level): Likewise.
8379 (estimate_loops): Likewise.
8380 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8381 (spread_components): Likewise.
8382 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8383 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8384 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8385 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8386 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8387 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8388 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8389 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8390 (mark_threaded_blocks): Likewise.
8391 (thread_through_all_blocks): Likewise.
8392 * tree-ssa.c (verify_ssa): Likewise.
8393 (execute_update_addresses_taken): Likewise.
8394 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8396 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8398 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8400 (post_order_compute): Likewise.
8401 (inverted_post_order_compute): Likewise.
8402 (pre_and_rev_post_order_compute_fn): Likewise.
8404 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8406 * genrecog.c (int_set::int_set): Explicitly construct our
8407 auto_vec base class.
8408 * vec.h (auto_vec::auto_vec): New constructor.
8410 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8412 * bitmap.h (class auto_bitmap): New constructor taking
8413 bitmap_obstack * argument.
8415 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8417 * bitmap.h (class auto_bitmap): Change type of m_bits to
8418 bitmap_head, and adjust ctor / dtor and member operators.
8420 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8422 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8423 when returned register mode doesn't match original mode.
8425 2017-05-12 Jeff Law <law@redhat.com>
8426 Jakub Jelinek <jakub@redhat.com>
8428 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8429 we look for cc setter after the compare-elim changes.
8430 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8431 within the vector to match what compare-elim now expects.
8432 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8433 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8435 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8436 after the compare-elim changes.
8437 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8438 the vector to match what compare-elim now expects.
8439 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8440 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8441 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8442 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8443 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8445 * config/visium/visium.c (single_set_and_flags): Fix where
8446 we look for cc setter after the compare-elim changes.
8447 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8448 with the vector to match what compare-elim now expects.
8449 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8450 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8451 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8452 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8453 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8454 (neg<mode>2_insn_set_overflow): Likewise.
8456 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8459 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8460 maybe_expand_insn call, set ops[0].target. If still set after call,
8461 set alt_rtl. Add extra arg to recursive calls.
8462 (extract_bit_field): Add alt_rtl argument. Pass to
8464 * expmed.h (extract_bit_field): Fix prototype.
8465 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8466 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8467 to extract_bit_field_calls.
8468 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8469 Pass alt_rtl to extract_bit_field calls.
8470 * calls.c (store_unaligned_arguments_into_psuedos)
8471 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8472 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8474 * optabs.h (struct expand_operand): Add target bitfield.
8476 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8478 * compare-elim.c (try_eliminate_compare): Canonicalize
8479 operation with embedded compare to
8480 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8481 (set (reg) (operation)].
8483 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8485 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8488 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8489 cost of adding a carry flag for ADC instruction.
8490 [case MINUS]: Ignore the cost of subtracting a carry flag
8491 for SBB instruction.
8493 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8495 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8497 * config/rs6000/bmiintrin.h: New file.
8498 * config/rs6000/bmi2intrin.h: New file.
8499 * config/rs6000/x86intrin.h: New file.
8501 2017-05-12 Jeff Law <law@redhat.com>
8503 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8506 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8509 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8511 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8514 early expansion of vector divide builtins.
8515 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8516 builtins identified as having unsigned arguments.
8518 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8520 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8521 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8522 expansion of vector logical operations (and, andc, or, xor,
8525 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8527 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8528 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8530 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8532 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8533 early GIMPLE expansion of vector multiplies.
8535 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8537 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8538 TARGET_HAVE_MOVT conditional.
8539 (movt splitter): Likewise.
8541 2017-05-12 Richard Biener <rguenther@suse.de>
8543 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8544 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8545 Fold all stmts not inplace.
8547 2017-05-12 Richard Biener <rguenther@suse.de>
8549 PR tree-optimization/80713
8550 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8551 inserted_exprs bit for not removed stmts.
8553 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8556 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8557 parallelized" attribute for parallelized OpenACC kernels.
8558 * omp-offload.c (execute_oacc_device_lower): Use it.
8560 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8561 Set "oacc kernels" attribute.
8562 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8563 parameter. Adjust all users.
8564 (oacc_fn_attrib_kernels_p): Remove function.
8565 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8566 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8567 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8568 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8569 assert "oacc kernels" attribute is set.
8571 2017-05-11 Carl Love <cel@us.ibm.com>
8573 * config/rs6000/rs6000-c: Add support for built-in functions
8574 vector unsigned char vec_popcnt (vector signed char)
8575 vector unsigned char vec_popcnt (vector unsigned char)
8576 vector unsigned short vec_popcnt (vector signed short)
8577 vector unsigned short vec_popcnt (vector unsigned short)
8578 vector unsigned int vec_popcnt (vector signed int)
8579 vector unsigned int vec_popcnt (vector unsigned int)
8580 vector unsigned long long vec_popcnt (vector signed long long)
8581 vector unsigned long long vec_popcnt (vector unsigned long long)
8582 vector signed long long vec_slo (vector signed long long,
8584 vector signed long long vec_slo (vector signed long long,
8585 vector unsigned char)
8586 vector unsigned long long vec_slo (vector unsigned long long,
8588 vector unsigned long long vec_slo (vector unsigned long long,
8589 vector unsigned char)
8590 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8591 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8592 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8593 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8594 * doc/extend.texi: Update the built-in documentation file for the
8595 new built-in functions.
8597 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8599 * attribs.h (sorted_attr_string): Move machine independent
8600 functions for target clone support from the i386 port to common
8601 code. Rename ix86_function_versions to common_function_versions.
8602 Rename make_name to make_unique_name.
8603 (common_function_versions): Likewise.
8604 (make_unique_name): Likewise.
8605 (make_dispatcher_decl): Likewise.
8606 (is_function_default_version): Likewise.
8607 * attribs.c (attr_strcmp): Likewise.
8608 (sorted_attr_string): Likewise.
8609 (common_function_versions): Likewise.
8610 (make_unique_name): Likewise.
8611 (make_dispatcher_decl): Likewise.
8612 (is_function_default_version): Likewise.
8613 * config/i386/i386.c (attr_strcmp): Likewise.
8614 (sorted_attr_string): Likewise.
8615 (ix86_function_versions): Likewise.
8616 (make_name): Likewise.
8617 (make_dispatcher_decl): Likewise.
8618 (is_function_default_version): Likewise.
8619 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8621 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8624 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8625 Account for direct move costs for vec_construct of integer
8628 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8631 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8632 (UNSPEC_STX_ATOMIC): Ditto.
8633 (loaddi_via_sse): New insn.
8634 (storedi_via_sse): Ditto.
8635 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8636 Update corresponding peephole2 patterns.
8637 (atomic_storedi_fpu): Ditto.
8639 2017-05-11 Julia Koval <julia.koval@intel.com>
8641 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8642 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8644 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8645 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8646 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8648 2017-05-11 Nathan Sidwell <nathan@acm.org>
8650 * graphite-poly.c: Include dumpfile.h.
8652 * dumpfle.h (dump_function): Declare here ...
8653 * tree-dump.h (dump_function): ... not here.
8654 * dumpfile.c: #include tree-cfg.h.
8655 (dump_function): Move here from ...
8656 * tree-dump.c (dump_function): ... here.
8657 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8658 * graphite-poly.c: Don't include tree-dump.h.
8659 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8660 * print-tree.c: Likewise.
8661 * stor-layout.c: Likewise.
8662 * tree-nested.c: Likewise.
8664 * dumpfile.c (dump_start): Use TDF_FLAGS.
8665 (dump_enable_all): Fix TDF_KIND check thinko.
8667 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8669 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8670 array entries to represent two legal parameterizations of the
8671 overloaded __builtin_cmpb function, as represented by the
8672 P6_OV_BUILTIN_CMPB constant.
8673 (altivec_resolve_overloaded_builtin): Add special case handling
8674 for the __builtin_cmpb function, as represented by the
8675 P6_OV_BUILTIN_CMPB constant.
8676 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8677 (BU_P6_64BIT_2): New macro.
8678 (BU_P6_OVERLOAD_2): New macro
8679 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8680 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8681 (CMPB): Add overload support to represent both 32-bit and 64-bit
8682 compare-bytes function.
8683 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8684 support for TARGET_CMPB.
8685 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8686 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8687 documentation of the __builtin_cmpb overloaded built-in function.
8689 2017-05-11 Richard Biener <rguenther@suse.de>
8691 PR tree-optimization/80705
8692 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8693 bases are not vectorizable.
8695 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8697 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8698 when counting register pressure.
8700 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8702 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8703 (iv_ca_more_deps): Renamed to ...
8704 (iv_ca_compare_deps): ... this.
8705 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8707 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8709 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8711 (determine_group_iv_costs): ... here.
8712 (find_inv_vars_cb): Record inv var if it's not recorded before.
8714 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8716 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8717 (get_shiftadd_cost): Ditto.
8719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8721 * tree-ssa-address.c: Include header file.
8722 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8724 (add_to_parts): Refactor.
8725 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8726 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8729 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8731 PR tree-optimization/53090
8732 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8734 (extract_cond_operands): Detect condition with IV on both sides
8735 and return COMP_IV_EXPR_2.
8736 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8737 (rewrite_use_compare): Simplify by removing call to function
8738 extract_cond_operands.
8740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8742 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8743 (extract_cond_operands): Detect condition comparing against non-
8744 invariant bound and return appropriate enum value.
8745 (find_interesting_uses_cond): Update use of extract_cond_operands.
8746 Handle its return value accordingly.
8747 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8751 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8752 nonlinear iv_use computation in loop invariant sensitive way.
8754 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8756 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8757 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8761 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8762 (dump_cand): Support iv_cand.inv_exprs.
8763 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8765 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8768 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8770 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8772 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8773 as local function. Include necessary header files.
8774 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8776 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8778 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8782 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8783 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8784 RSHIFT_EXPR and BIT_NOT_EXPR.
8786 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8788 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8789 (adjust_setup_cost): New parameter supporting round up adjustment.
8790 (struct address_cost_data): Delete.
8791 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8792 (split_address_cost, ptr_difference_cost): Delete.
8793 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8794 (struct ainc_cost_data): New struct.
8795 (get_address_cost_ainc): New function.
8796 (get_address_cost, get_computation_cost): Reimplement.
8797 (determine_group_iv_cost_address): Record inv_expr for all uses of
8799 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8800 (iv_ca_has_deps): Reimplemented to ...
8801 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8803 (iv_ca_extend): Call iv_ca_more_deps.
8805 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8807 * tree-ssa-address.c (struct mem_address): Move to header file.
8808 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8809 * tree-ssa-address.h (struct mem_address): Move from C file.
8810 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8814 * tree-affine.h (aff_combination_type): New interface.
8815 (aff_combination_zero_p): Remove static.
8816 (aff_combination_const_p): New interface.
8817 (aff_combination_singleton_var_p): New interfaces.
8819 2017-05-11 Richard Biener <rguenther@suse.de>
8821 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8822 Skip unreachable blocks and destinations.
8823 (eliminate): Move stmt removal and fixup ...
8824 (fini_eliminate): ... here. Skip inserted exprs.
8825 (pass_pre::execute): Move fini_pre after fini_eliminate.
8826 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8827 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8828 PRE to get rid of dead code that has invalid SSA form and
8829 split critical edges again.
8831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8833 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8835 2017-05-11 Richard Biener <rguenther@suse.de>
8837 * passes.c (execute_function_todo): Verify loops if they are
8838 said to be up-to-date.
8839 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8840 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8842 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8845 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8846 handle calling assemble_external ourself.
8849 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8850 modes with zero size. Enhance comment.
8852 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8854 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8855 built-ins for vec_xl and vec_xst with short and char pointer
8858 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8860 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8861 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8862 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8863 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8864 (_mm_maskz_min_round_ss): New intrinsics.
8865 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8866 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8867 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8868 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8869 (__builtin_ia32_minss_mask_round): New builtins.
8870 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8871 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8872 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
8874 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
8875 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
8877 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
8880 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8882 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
8883 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
8884 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
8885 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
8886 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
8887 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
8888 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
8889 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8890 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8891 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
8892 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
8893 (__builtin_ia32_mulss_mask_round): New builtins.
8894 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8895 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8896 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
8898 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
8899 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8901 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8904 2017-05-10 Julia Koval <julia.koval@intel.com>
8906 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
8907 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
8908 (_mm256_setr_m128i): New intrinsics.
8910 2017-05-10 Julia Koval <julia.koval@intel.com>
8912 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
8913 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
8914 (_mm_maskz_rcp14_ss): New intrinsics.
8915 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
8916 (__builtin_ia32_rcp14ss_mask): New builtins.
8917 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
8919 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
8921 PR tree-optimization/51513
8922 * tree-cfg.c (gimple_seq_unreachable_p): New function.
8923 (assert_unreachable_fallthru_edge_p): Use it.
8924 (group_case_labels_stmt): Likewise.
8925 * tree-cfg.h: Prototype it.
8926 * stmt.c: Include cfghooks.h and tree-cfg.h.
8927 (emit_case_dispatch_table) <gap_label>: New local variable.
8928 Use it to fill dispatch table gaps.
8929 Test for default_label before updating probabilities.
8930 (expand_case) <default_label>: Remove unneeded initialization.
8931 Test for unreachable default case statement and remove its edge.
8932 Set default_label accordingly.
8933 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
8935 2017-05-10 Carl Love <cel@us.ibm.com>
8937 * config/rs6000/rs6000-c: Add support for built-in functions
8938 vector signed char vec_neg (vector signed char)
8939 vector signed short int vec_neg (vector short int)
8940 vector signed int vec_neg (vector signed int)
8941 vector signed long long vec_neg (vector signed long long)
8942 vector float vec_neg (vector float)
8943 vector double vec_neg (vector double)
8944 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
8946 * config/rs6000/altivec.h: Add define for vec_neg
8947 * doc/extend.texi: Update the built-in documentation for the
8948 new built-in functions.
8950 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8952 PR tree-optimization/77644
8953 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
8955 2017-05-10 Nathan Sidwell <nathan@acm.org>
8957 * dumpfile.h (TDI_lang_all): New.
8958 (TDF_KIND): New. Renumber others
8959 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
8961 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
8963 (get_dump_file_name): Adjust suffix generation.
8964 (dump_enable_all): Use TDF_KIND.
8965 * doc/invoke.texi (-fdump-lang-all): Document.
8967 * dumpfile.h: Tabify.
8969 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
8972 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
8973 Move member access before delete.
8975 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
8977 * tree-inline.c (expand_call_inline): Split block at stmt
8980 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8983 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
8984 are now unused after splitting mov{sf,sd}_hardfloat.
8985 (f32_lr2): Likewise.
8987 (f32_lm2): Likewise.
8989 (f32_li2): Likewise.
8992 (f32_sr2): Likewise.
8994 (f32_sm2): Likewise.
8996 (f32_si2): Likewise.
8999 (f32_vsx): Likewise.
9001 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9002 For movsf, order stores so the VSX stores occur before the GPR
9003 store which encourages the register allocator to use a traditional
9004 FPR instead of a GPR. For movsd, order the stores so that the GPR
9005 store comes before the VSX stores to allow the power6 to work.
9006 This is due to the power6 not having a 32-bit integer store
9007 instruction from a FPR.
9008 (movsf_hardfloat): Likewise.
9009 (movsd_hardfloat): Likewise.
9011 2017-05-09 Martin Sebor <msebor@redhat.com>
9013 PR translation/80280
9014 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9017 PR translation/80280
9018 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9019 data member added in r247778.
9020 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9022 2017-05-09 Nathan Sidwell <nathan@acm.org>
9024 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9026 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9029 2017-05-09 Marek Polacek <polacek@redhat.com>
9031 * doc/invoke.texi: Fix typo.
9033 2017-05-09 Richard Biener <rguenther@suse.de>
9035 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9036 (vrp_val_is_min): Likewise.
9037 (set_value_range_to_value): Likewise.
9038 (set_value_range_to_nonnegative): Likewise.
9039 (gimple_assign_nonzero_p): Likewise.
9040 (gimple_stmt_nonzero_p): Likewise.
9041 (vrp_int_const_binop): Likewise. Remove unreachable case.
9042 (adjust_range_with_scev): Adjust comments.
9043 (compare_range_with_value): Likewise.
9044 (extract_range_from_phi_node): Likewise.
9045 (test_for_singularity): Likewise.
9047 2017-05-09 Richard Biener <rguenther@suse.de>
9049 * tree-vrp.c (get_single_symbol): Add assert that we don't
9050 get overflowed constants as invariant part.
9051 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9052 checks. Use wi::cmp instead of recursing for integer constants.
9053 (compare_values): Just ignore whether we assumed undefined
9054 overflow instead of failing the compare.
9055 (extract_range_for_var_from_comparison_expr): Add comment before the
9056 TREE_NO_WARNING sets.
9057 (test_for_singularity): Likewise.
9058 (extract_range_from_comparison): Do not disable optimization
9059 when we assumed undefined overflow.
9060 (extract_range_basic): Remove init of unused var.
9062 2017-05-09 Richard Biener <rguenther@suse.de>
9064 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9065 (extract_range_from_multiplicative_op_1): Adjust.
9066 (extract_range_from_binary_expr_1): Use int_const_binop.
9068 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9071 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9072 rs6000_store_data_bypass_p in seven define_bypass directives and
9073 in several comments.
9074 * config/rs6000/rs6000-protos.h: Add prototype for
9075 rs6000_store_data_bypass_p function.
9076 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9077 function implements slightly different (rs6000-specific) semantics
9078 than store_data_bypass_p, returning false rather than aborting
9079 with assertion error when arguments do not satisfy the
9080 requirements of store data bypass.
9081 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9082 rs6000_store_data_bypass_p.
9084 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9086 * config/xtensa/xtensa-protos.h
9087 (xtensa_initial_elimination_offset): New declaration.
9088 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9089 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9090 macro definition, add case for FRAME_POINTER_REGNUM when
9091 FRAME_GROWS_DOWNWARD.
9092 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9093 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9094 xtensa_initial_elimination_offset.
9096 2017-05-08 Nathan Sidwell <nathan@acm.org>
9098 * doc/invoke.texi: Alphabetize -fdump options.
9100 2017-05-08 Martin Sebor <msebor@redhat.com>
9102 PR translation/80280
9103 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9105 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9107 * target.def (compute_frame_layout): New optional target hook.
9108 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9109 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9110 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9112 * reload1.c (verify_initial_elim_offsets): Likewise.
9113 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9114 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9115 (arm_get_frame_offsets): Split up into this ...
9116 (arm_compute_frame_layout): ... and this function.
9118 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9120 * config/aarch64/constraints.md (Usa): New constraint.
9121 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9123 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9125 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9126 with_multilib_list after it has been checked.
9128 2017-05-08 Richard Biener <rguenther@suse.de>
9130 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9131 (bitmap_set_subtract_values): Likewise.
9133 2017-05-08 Richard Biener <rguenther@suse.de>
9135 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9136 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9138 (gimple_stmt_nonzero_warnv_p): Rename to ...
9139 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9141 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9142 (extract_range_basic): Adjust, do not disable propagation on
9143 strict overflow sensitive simplification.
9144 (vrp_visit_cond_stmt): Likewise.
9146 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9148 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9149 body size unconditionally.
9151 2017-05-07 Jeff Law <law@redhat.com>
9154 2017-05-06 Jeff Law <law@redhat.com>
9155 PR tree-optimization/78496
9156 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9159 PR tree-optimization/78496
9160 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9161 (simplify_stmt_using_ranges): Call it.
9162 (vrp_dom_walker::before_dom_children): Extract equivalences
9163 from an ASSERT_EXPR with an equality comparison against a
9166 2017-05-06 Jeff Law <law@redhat.com>
9168 PR tree-optimization/78496
9169 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9172 PR tree-optimization/78496
9173 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9174 (simplify_stmt_using_ranges): Call it.
9175 (vrp_dom_walker::before_dom_children): Extract equivalences
9176 from an ASSERT_EXPR with an equality comparison against a
9179 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9181 * lra-constraints.c (lra_copy_reg_equiv): New function.
9182 (split_reg): Use it to copy equivalence information from the
9183 original register to the spill register.
9185 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9187 PR rtl-optimization/75964
9188 * simplify-rtx.c (simplify_const_relational_operation): Remove
9189 invalid handling of comparisons of integer ABS.
9191 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9193 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9195 (init_regs): Remove declaration.
9196 (function_arg_advance_32): Initialize error_p as boolean variable.
9198 2017-05-05 Nathan Sidwell <nathan@acm.org>
9200 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9201 lines. Use for (;;).
9203 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9205 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9206 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9207 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9208 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9209 VF=2 that require versioning.
9211 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9213 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9216 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9218 * diagnostic.h (diagnostic_override_option_index): Convert from
9219 macro to inline function.
9221 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9223 * diagnostic.c (last_module_changed_p): New function.
9224 (set_last_module): New function.
9225 (diagnostic_report_current_module): Convert macro usage to
9226 the above functions.
9227 * diagnostic.h (diagnostic_context::last_module): Strengthen
9228 from const line_map * to const line_map_ordinary *.
9229 (diagnostic_last_module_changed): Delete macro.
9230 (diagnostic_set_last_module): Delete macro.
9232 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9234 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9235 with diagnostic_report_diagnostic.
9236 (diagnostic_n_impl_richloc): Likewise.
9237 * diagnostic.h (report_diagnostic): Delete macro.
9238 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9239 with diagnostic_report_diagnostic.
9240 * substring-locations.c (format_warning_va): Likewise.
9242 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9244 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9245 save/restor of format_spec. Move option-printing code to...
9246 (print_option_information): ...this new function, and
9247 reimplement by simply printing to the pretty_printer,
9248 rather than appending to the format string.
9250 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9252 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9253 handling logic into...
9254 (update_effective_level_from_pragmas): ...this new function.
9256 2017-05-04 Andrew Waterman <andrew@sifive.com>
9258 * config/riscv/riscv.opt (mstrict-align): New option.
9259 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9260 (SLOW_UNALIGNED_ACCESS): Define.
9261 (riscv_slow_unaligned_access): Declare.
9262 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9264 (riscv_slow_unaligned_access): New variable.
9265 (rocket_tune_info): Set slow_unaligned_access to true.
9266 (optimize_size_tune_info): Set slow_unaligned_access to false.
9267 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9268 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9269 (riscv_option_override): Set riscv_slow_unaligned_access.
9270 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9272 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9274 * config/riscv/riscv.md: Unify indentation.
9276 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9281 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9283 (extendsi<mode>2): Add support for doing sign extension via
9284 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9285 don't have ISA 3.0 instructions.
9286 (extendsi<mode>2 splitter): Likewise.
9287 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9288 generate the normal insns since SImode can now go in vector
9289 registers. Disallow the special UNSPECs needed for previous
9290 machines to hide SImode being used. Add new insns
9291 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9292 (fix_trunc<mode>si2_stfiwx): Likewise.
9293 (fix_trunc<mode>si2_internal): Likewise.
9294 (fixuns_trunc<mode>si2): Likewise.
9295 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9296 (fctiw<u>z_<mode>_smallint): Likewise.
9297 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9298 of floating point to 32-bit integer from doing a direct move to
9299 the GPR registers to do a store.
9300 (fctiwz_<mode>): Break long line.
9302 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9304 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9305 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9306 (addr_list, addr_offset_valid_p): New.
9307 (split_address_groups): Check offset validity with above function.
9308 (gt-tree-ssa-loop-ivopts.h): Include header file.
9310 2017-05-05 Nathan Sidwell <nathan@acm.org>
9312 * config.gcc (arm*-*-*): Add missing 'fi'.
9314 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9316 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9317 included in -fopt-info does not matter.
9318 * doc/optinfo.texi (-fopt-info): Fix description of default
9319 behavour. Explicitly say order of options included in -fopt-info
9322 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9324 * config.gcc: Allow combinations of aprofile and rmprofile values for
9325 --with-multilib-list.
9326 * config/arm/t-multilib: New file.
9327 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9328 variables. Remove setting of ISA and floating-point ABI in
9329 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9330 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9331 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9332 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9334 * config/arm/t-rmprofile: Likewise except for the matches changes.
9335 * doc/install.texi (--with-multilib-list): Document the combination of
9336 aprofile and rmprofile values and warn about pitfalls in doing that.
9338 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9340 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9341 (movdi_aarch64): Likewise.
9343 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9345 PR tree-optimization/80632
9346 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9348 (build_arrays): Initialize it for virtual phis.
9349 (fix_phi_nodes): Use it for virtual phis.
9351 PR tree-optimization/80558
9352 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9353 [x, y] op z into [x op, y op z] for op & or | if conditions
9356 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9357 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9360 * config/arm/arm.md (use_literal_pool): Remove.
9361 (64-bit immediate split): No longer takes cost into consideration
9362 if arm_disable_literal_pool is enabled.
9363 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9364 used when arm_disable_literal_pool is enabled.
9365 (arm_max_const_double_inline_cost): Remove use of
9366 arm_disable_literal_pool.
9367 (push_minipool_fix): Add assert.
9368 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9369 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9370 (no_literal_pool_sf_immediate): New.
9372 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9374 PR tree-optimization/80613
9375 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9376 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9378 2017-05-05 Richard Biener <rguenther@suse.de>
9380 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9382 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9384 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9385 of this flag from insn conditions due to removal from r247495.
9387 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9389 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9391 (arm_early_store_addr_dep_ptr): Likewise.
9392 * config/arm/aarch-common-protos.h
9393 (arm_early_load_addr_dep_ptr): Add prototype.
9394 (arm_early_store_addr_dep_ptr): Likewise.
9395 * config/arm/cortex-a53.md: Add new bypasses.
9397 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9399 * tree.c (next_type_uid): Change type to unsigned.
9400 (type_hash_canon): Decrement back next_type_uid if
9401 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9402 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9405 2017-05-04 Martin Sebor <msebor@redhat.com>
9407 * builtins.c: Fix a trivial typo in a comment.
9410 * builtins.c (check_sizes): Adjust to handle reading past the end.
9411 Avoid printing excessive upper bound of ranges. Use %E to print
9412 tree nodes instead of converting them to %wu.
9413 (expand_builtin_memchr): New function.
9414 (compute_dest_size): Rename...
9415 (compute_objsize): ...to this.
9416 (expand_builtin_memcpy): Adjust.
9417 (expand_builtin_mempcpy): Adjust.
9418 (expand_builtin_strcat): Adjust.
9419 (expand_builtin_strcpy): Adjust.
9420 (check_strncat_sizes): Adjust.
9421 (expand_builtin_strncat): Adjust.
9422 (expand_builtin_strncpy): Adjust and simplify.
9423 (expand_builtin_memset): Adjust.
9424 (expand_builtin_bzero): Adjust.
9425 (expand_builtin_memcmp): Adjust.
9426 (expand_builtin): Handle memcmp.
9427 (maybe_emit_chk_warning): Check strncat just once.
9429 2017-05-04 Martin Sebor <msebor@redhat.com>
9431 PR preprocessor/79214
9434 * builtins.c (check_sizes): Add inlining context and issue
9435 warnings even when -Wno-system-headers is set.
9436 (check_strncat_sizes): Same.
9437 (expand_builtin_strncat): Same.
9438 (expand_builtin_memmove): New function.
9439 (expand_builtin_stpncpy): Same.
9440 (expand_builtin): Handle memmove and stpncpy.
9442 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9444 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9445 which is not used any more.
9447 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9449 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9451 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9453 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9454 (cortexa53_tunings): Likewise.
9455 (cortexa57_tunings): Likewise.
9456 (cortexa72_tunings): Likewise.
9457 (cortexa73_tunings): Likewise.
9459 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9461 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9462 Set loop alignment to 8.
9464 2017-05-04 Martin Sebor <msebor@redhat.com>
9466 PR translation/80280
9467 * builtins.c (expand_builtin_object_size): Add missing quoting to
9468 %D and like directives.
9469 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9470 (hsa_type_for_tree_type): Same.
9471 (verify_function_arguments): Same.
9472 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9473 * varasm.c (get_section): Same.
9476 2017-05-04 Martin Sebor <msebor@redhat.com>
9478 PR translation/80280
9479 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9481 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9483 * config/aarch64/aarch64.c (generic_addrcost_table):
9484 Change HI/TI mode setting.
9486 2017-05-04 Martin Jambor <mjambor@suse.cz>
9488 PR tree-optimization/80622
9489 * tree-sra.c (comes_initialized_p): New function.
9490 (build_accesses_from_assign): Only set write lazily when
9491 comes_initialized_p is false.
9492 (analyze_access_subtree): Use comes_initialized_p.
9493 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9494 instead of testing for PARM_DECL.
9496 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9498 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9499 constraint on operand 0 to allow more general addressing modes.
9500 Adjust output template.
9501 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9503 * config/aarch64/aarch64-protos.h
9504 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9505 * config/aarch64/constraints.md (Dp): New address constraint.
9506 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9509 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9511 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9512 update use of estimate_ipcp_clone_size_and_time.
9513 (estimate_local_effects): Update use of
9514 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9515 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9516 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9517 Return nonspecialized time.
9519 2017-05-04 Richard Biener <rguenther@suse.de>
9521 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9522 for the last VUSE which def dominates the PHI. Directly call
9524 (get_continuation_for_phi_1): Remove.
9526 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9528 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9529 to explain the use of truncating division. Cap the number of
9530 iterations to the maximum given by nb_iterations_upper_bound,
9533 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9535 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9536 * configure: Regenerate.
9537 * config.in: Regenerate.
9538 * config/i386/driver-mingw32.c: new file.
9539 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9540 * config.host: Link driver-mingw32.o on MinGW host.
9541 * doc/install.texi: Document new --enable-mingw-wildcard configure
9544 2017-05-04 Marek Polacek <polacek@redhat.com>
9546 PR tree-optimization/80612
9547 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9549 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9550 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9552 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9553 (movt splitter): Likewise.
9554 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9555 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9556 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9557 block for Thumb-1 with MOVT.
9558 (thumb2_legitimate_address_p): Move code block ...
9559 (can_avoid_literal_pool_for_label_p): ... into this new function.
9560 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9562 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9563 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9564 "M-profile targets with the MOVT instruction".
9566 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9568 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9569 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9570 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9572 2017-05-04 Martin Liska <mliska@suse.cz>
9574 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9577 2017-05-04 Richard Biener <rguenther@suse.de>
9579 * tree.c (array_at_struct_end_p): Handle arrays at struct
9580 end with flexarrays more conservatively. Refactor and treat
9581 arrays of arrays or aggregates more strict. Fix
9582 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9583 * tree.c (array_at_struct_end_p): Adjust prototype.
9584 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9585 * gimple-fold.c (get_range_strlen): Likewise.
9586 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9588 2017-05-04 Richard Biener <rguenther@suse.de>
9590 PR tree-optimization/31130
9591 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9593 (supports_overflow_infinity): Likewise.
9594 (is_negative_overflow_infinity): Likewise.
9595 (is_positive_overflow_infinity): Likewise.
9596 (is_overflow_infinity): Likewise.
9597 (stmt_overflow_infinity): Likewise.
9598 (overflow_infinity_range_p): Likewise.
9599 (usable_range_p): Remove as always returning true.
9600 (make_overflow_infinity): Remove.
9601 (negative_overflow_infinity): Likewise.
9602 (positive_overflow_infinity): Likewise.
9603 (avoid_overflow_infinity): Likewise.
9604 (set_value_range): Adjust accordingly.
9605 (set_value_range_to_nonnegative): Likewise, remove now unused
9606 overflow_infinity arg.
9607 (vrp_operand_equal_p): Adjust.
9608 (update_value_range): Likewise.
9609 (range_int_cst_singleton_p): Likewise.
9610 (operand_less_p): Likewise.
9611 (compare_values_warnv): Likewise.
9612 (extract_range_for_var_from_comparison_expr): Likewise.
9613 (vrp_int_const_binop): Likewise.
9614 (zero_nonzero_bits_from_vr): Likewise.
9615 (extract_range_from_multiplicative_op_1): Likewise.
9616 (extract_range_from_binary_expr_1): Likewise.
9617 (extract_range_from_unary_expr): Likewise.
9618 (extract_range_from_comparison): Likewise.
9619 (extract_range_basic): Likewise.
9620 (adjust_range_with_scev): Likewise.
9621 (compare_ranges): Likewise.
9622 (compare_range_with_value): Likewise.
9623 (dump_value_range): Likewise.
9624 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9626 (simplify_cond_using_ranges): Adjust.
9628 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9630 * brig-builtins.def: Added a builtin for class_f64.
9631 * builtin-types.def: Added a builtin type needed by class_f64.
9633 2017-05-03 Jason Merrill <jason@redhat.com>
9635 * timevar.def: Add TV_CONSTEXPR.
9637 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9639 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9641 2017-05-03 Martin Jambor <mjambor@suse.cz>
9643 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9644 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9645 * ipa-cp.c (ipcp_propagate_stage): Do not call
9646 ipa_update_after_lto_read.
9647 * ipa-inline.c (ipa_inline): Likewise.
9649 2017-05-03 Martin Jambor <mjambor@suse.cz>
9651 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9652 tag. Added a default constructor and a destructor.
9653 (ipa_edge_args_sum_t): New class;
9654 (ipa_edge_args_sum): Declare.
9655 (ipa_edge_args_vector): Remove declaration.
9656 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9657 (ipa_free_edge_args_substructures): Remove declaration.
9658 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9659 (ipa_edge_args_info_available_for_edge_p): Likewise.
9660 * ipa-prop.c (ipa_edge_args_vector): Removed.
9661 (edge_removal_hook_holder): Likewise.
9662 (edge_duplication_hook_holder): Likewise.
9663 (ipa_edge_args_sum): New variable.
9664 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9665 ipa_edge_args_vector.
9666 (ipa_free_edge_args_substructures): Likewise.
9667 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9668 ipa_edge_args_vector.
9669 (ipa_edge_removal_hook): Turned into method
9670 ipa_edge_args_sum_t::remove.
9671 (ipa_edge_duplication_hook): Turned into method
9672 ipa_edge_args_sum_t::duplicate.
9673 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9674 registering edge hooks.
9675 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9676 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9677 ipa_edge_args_sum instead of ipa_edge_args_vector.
9678 * ipa-profile.c (ipa_profile): Likewise.
9680 2017-05-03 Martin Jambor <mjambor@suse.cz>
9682 * symbol-summary.h (function_summary): New method exists.
9683 (function_summary::symtab_removal): Deallocate through release.
9684 (call_summary): New class.
9685 (gt_ggc_mx): New overload.
9686 (gt_pch_nx): Likewise.
9687 (gt_pch_nx): Likewise.
9689 2017-05-03 Jeff Law <law@redhat.com>
9691 PR tree-optimization/78496
9692 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9693 from simplify_cond_using_ranges. Split off code to walk
9694 backwards through casts into ...
9695 (simplify_cond_using_ranges_2): New function.
9696 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9697 (execute_vrp): After identifying jump threads, call
9698 simplify_cond_using_ranges_2.
9700 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9703 * ipa-inline.h (inline_summary): Add ctor.
9704 (create_ggc): Do not use ggc_cleared_alloc.
9706 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9707 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9709 * gcc.c (handle_braces): Support escaping in switch matching
9711 * doc/invoke.texi (Spec Files): Document it.
9712 Remove superfluous @code markup in items.
9714 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9716 * diagnostic-show-locus.c (struct column_range): New struct.
9717 (get_affected_columns): New function.
9718 (get_printed_columns): New function.
9719 (struct correction): New struct.
9720 (correction::ensure_capacity): New function.
9721 (correction::ensure_terminated): New function.
9722 (struct line_corrections): New struct.
9723 (line_corrections::~line_corrections): New dtor.
9724 (line_corrections::add_hint): New function.
9725 (layout::print_trailing_fixits): Reimplement in terms of the new
9727 (selftest::test_overlapped_fixit_printing): New function.
9728 (selftest::diagnostic_show_locus_c_tests): Call it.
9730 2017-05-03 Nathan Sidwell <nathan@acm.org>
9732 Canonicalize canonical type hashing
9733 * tree.h (type_hash_canon_hash): Declare.
9734 * tree.c (type_hash_list, attribute_hash_list): Move into
9735 type_hash_canon_hash.
9736 (build_type_attribute_qual_variant): Break out hash code calc into
9737 type_hash_canon_hash.
9738 (type_hash_canon_hash): New. Generic type hash computation.
9739 (build_range_type_1, build_array_type_1, build_function_type,
9740 build_method_type_directly, build_offset_type, build_complex_type,
9741 make_vector_type): Call it.
9743 2017-05-03 Richard Biener <rguenther@suse.de>
9745 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9746 When all DRs have unknown misaligned do not always peel
9747 when there is a store but apply the same costing model as if
9748 there were only loads.
9750 2017-05-03 Richard Biener <rguenther@suse.de>
9753 PR tree-optimization/80492
9754 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9755 compare_base_decls returning dont-know properly.
9757 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9759 * config/arm/iterators.md (CCSI): New mode iterator.
9760 (arch): New mode attribute.
9761 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9762 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9763 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9764 code iterator for success result mode.
9765 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9766 the corresponding new insn generators.
9768 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9771 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9772 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9774 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9776 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9777 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9778 (DDR_A): Wrap DDR argument in brackets.
9779 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9780 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9781 (DDR_REVERSED_P): Likewise.
9783 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9785 PR tree-optimization/79472
9786 * tree-switch-conversion.c (struct switch_conv_info): Add
9787 contiguous_range and default_case_nonstandard fields.
9788 (collect_switch_conv_info): Compute contiguous_range and
9789 default_case_nonstandard fields, don't clear final_bb if
9790 contiguous_range and only the default case doesn't have the required
9792 (check_all_empty_except_final): Set default_case_nonstandard instead
9793 of failing if contiguous_range and the default case doesn't have empty
9795 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9796 and only the default case doesn't have the required constants. Skip
9798 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9799 if default_case_nonstandard.
9800 (build_constructors): Build constant 1 just once. Assert that default
9801 values aren't inserted in between cases if contiguous_range. Skip
9803 (build_arrays): Skip virtual phis.
9804 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9805 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9806 Handle virtual phis.
9807 (gen_inbound_check): Handle default_case_nonstandard case.
9808 (process_switch): Adjust check_final_bb caller. Call
9809 gather_default_values with the first non-default case instead of
9810 default case if default_case_nonstandard.
9812 2017-05-02 Nathan Sidwell <nathan@acm.org>
9814 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9815 check. Fix formatting.
9817 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9819 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9820 errors when comparing specialized and unspecialized times.
9822 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9824 * diagnostic-show-locus.c
9825 (layout::should_print_annotation_line_p): Make private.
9826 (layout::print_annotation_line): Make private.
9827 (layout::annotation_line_showed_range_p): Make private.
9828 (layout::show_ruler): Make private.
9829 (layout::print_source_line): Make private. Pass in line and
9830 line_width, rather than calling location_get_source_line. Drop
9832 (layout::print_leading_fixits): New method.
9833 (layout::print_any_fixits): Rename to...
9834 (layout::print_trailing_fixits): ...this, and make private.
9835 Don't print newline fixits.
9836 (diagnostic_show_locus): Move logic for printing one row into...
9837 (layout::print_line): ...this new function. Move the
9838 location_get_source_line call and error-handling from
9839 print_source_line to here. Call print_leading_fixits, and rename
9840 print_any_fixits to print_trailing_fixits.
9841 (selftest::test_fixit_insert_containing_newline): Update now that
9842 newlines are partially supported.
9843 (selftest::test_fixit_insert_containing_newline_2): New test.
9844 (selftest::test_fixit_replace_containing_newline): Update comments.
9845 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9846 * edit-context.c (class added_line): New class.
9847 (class edited_line): Describe newline handling in comment.
9848 (edited_line::actually_edited_p): New method.
9849 (edited_line::print_content): Delete redundant decl.
9850 (edited_line::m_predecessors): New field.
9851 (edited_file::print_content): Call edited_line::print_content.
9852 (edited_file::print_diff): Update to support newlines.
9853 (edited_file::print_diff_hunk): Likewise.
9854 (edited_file::print_run_of_changed_lines): New function.
9855 (edited_file::print_diff_line): Convert to...
9856 (print_diff_line): ...this.
9857 (edited_file::get_effective_line_count): New function.
9858 (edited_line::edited_line): Initialize new field m_predecessors.
9859 (edited_line::~edited_line): Clean up m_predecessors.
9860 (edited_line::apply_fixit): Handle newlines.
9861 (edited_line::get_effective_line_count): New function.
9862 (edited_line::print_content): New function.
9863 (edited_line::print_diff_lines): New function.
9864 (selftest::test_applying_fixits_insert_containing_newline): New
9866 (selftest::test_applying_fixits_replace_containing_newline): New
9868 (selftest::insert_line): New function.
9869 (selftest::test_applying_fixits_multiple_lines): Add example of
9871 (selftest::edit_context_c_tests): Call the new tests.
9873 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9875 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
9876 parameter cand. Update dump information.
9877 (get_computation_cost): Update uses.
9879 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9881 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
9882 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
9883 (get_computation_at, rewrite_use_address): Update use of
9884 get_computation_aff.
9886 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9888 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
9889 (get_computation): Delete.
9890 (get_computation_cost): Implement like get_computation_cost_at.
9891 Use get_computation_at.
9892 (get_computation_cost_at): Delete.
9893 (rewrite_use_nonlinear_expr): Use get_computation_at.
9894 (rewrite_use_compare, remove_unused_ivs): Ditto.
9896 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9898 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
9900 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9902 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
9903 (ivopts_global_cost_for_size): Rename parameter and update uses.
9904 (iv_ca_recount_cost): Update uses.
9905 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
9906 candidates seperately in n_invs and n_cands.
9907 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
9909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9911 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
9912 (find_inv_vars_cb): New.
9913 (find_depends): Renamed to ...
9914 (find_inv_vars): ... this.
9915 (add_candidate_1, force_var_cost): Call find_inv_vars.
9916 (split_address_cost, determine_group_iv_cost_cond): Ditto.
9918 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9920 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
9921 inv_vars. Add inv_exprs.
9922 (struct iv_cand): Rename depends_on to inv_vars.
9923 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
9924 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
9925 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
9926 (dump_cand): Dump inv_vars.
9927 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
9928 (record_invariant, find_depends, add_candidate_1): Ditto.
9929 (set_group_iv_cost, force_var_cost): Ditto.
9930 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
9931 (get_computation_cost_at, get_computation_cost): Ditto.
9932 (determine_group_iv_cost_generic): Ditto.
9933 (determine_group_iv_cost_address): Ditto.
9934 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
9935 (determine_group_iv_costs): Ditto.
9936 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
9937 (iv_ca_set_remove_invariants): Renamed to ...
9938 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
9939 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
9940 (iv_ca_set_add_invariants): Renamed to ...
9941 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
9942 (iv_ca_set_cp): Use iv_ca_set_add_invs.
9943 (iv_ca_has_deps): Support inv_vars and inv_exprs.
9944 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
9945 (create_new_ivs): Remove useless dump.
9947 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9949 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
9951 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
9952 (iv_ca_set_no_cp, create_new_iv): Ditto.
9954 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9956 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9958 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9960 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
9961 function tree_check2.
9963 2017-05-02 Martin Liska <mliska@suse.cz>
9965 * doc/gcov.texi: Add missing preposition.
9966 * gcov.c (function_info::function_info): Properly fill up
9967 all member variables.
9969 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9971 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
9973 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9975 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
9977 2017-05-02 Martin Liska <mliska@suse.cz>
9980 * lto-streamer-in.c (lto_read_tree_1): Remove
9982 * lto-streamer.c (struct tree_hash_entry): Likewise.
9983 (struct tree_entry_hasher): Likewise.
9984 (tree_entry_hasher::hash): Likewise.
9985 (tree_entry_hasher::equal): Likewise.
9986 (lto_streamer_init): Likewise.
9987 (lto_orig_address_map): Likewise.
9988 (lto_orig_address_get): Likewise.
9989 (lto_orig_address_remove): Likewise.
9990 * lto-streamer.h: Likewise.
9991 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
9992 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
9994 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
9996 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
9997 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
9998 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
9999 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10000 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10001 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10002 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10003 (mm_maskz_sub_ss): New intrinsics.
10004 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10005 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10006 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10007 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10008 (__builtin_ia32_subss_mask_round): New builtins.
10009 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10010 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10011 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10013 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10014 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10016 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10019 2017-05-02 Martin Jambor <mjambor@suse.cz>
10021 PR tree-optimization/78687
10022 * tree-sra.c (access): New field parent.
10023 (process_subtree_disqualification): New function.
10024 (disqualify_candidate): Call it.
10025 (build_accesses_from_assign): Reset write flag if creating an
10027 (build_access_subtree): Fill in parent field and also prpagate
10028 down grp_write flag.
10029 (create_artificial_child_access): New parameter set_grp_write, set
10030 grp_write to its value.
10031 (propagate_subaccesses_across_link): Also propagate grp_write flag
10033 (propagate_all_subaccesses): Push the closest parent back to work
10034 queue if add_access_to_work_queue returned true.
10036 2017-05-02 Richard Biener <rguenther@suse.de>
10038 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10039 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10040 -fstrict-overflow documentation.
10041 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10042 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10043 flag_strict_overflow.
10044 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10045 * lto-opts.c (lto_write_options): Do not stream it.
10046 * lto-wrapper.c (merge_and_complain): Do not handle it.
10047 * opts.c (default_options_table): Do not set -fstrict-overflow.
10048 (finish_options): Likewise do not clear it when sanitizing.
10049 * simplify-rtx.c (simplify_const_relational_operation): Do not
10050 test flag_strict_overflow.
10052 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10054 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10055 using enabled attribute.
10056 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10057 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10058 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10059 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10060 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10061 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10062 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10063 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10064 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10065 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10067 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10069 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10071 2017-05-02 Richard Biener <rguenther@suse.de>
10073 PR tree-optimization/80591
10075 2017-04-10 Richard Biener <rguenther@suse.de>
10077 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10080 2017-05-02 Richard Biener <rguenther@suse.de>
10082 PR tree-optimization/80549
10083 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10084 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10085 headers if they do not have a preheader.
10087 2017-05-02 Martin Liska <mliska@suse.cz>
10090 * common.opt: Fix typo.
10091 * doc/invoke.texi: Likewise.
10093 2017-05-01 Jan Beulich <jbeulich@suse.com>
10095 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10096 swapping, add (x,x,m,x,n) alternative.
10098 2017-05-01 Nathan Sidwell <nathan@acm.org>
10100 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10101 unnecessary unadjusted_alignment check.
10103 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10106 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10108 * gimplify.c (gimplify_cilk_detach): New function.
10109 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10110 * tree-core.h: Document EXPR_CILK_SPAWN.
10111 * tree.h (EXPR_CILK_SPAWN): Define.
10113 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10115 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10116 to use new fixit_hint representation, using the "replace" logic.
10117 (get_line_span_for_fixit_hint): Likewise.
10118 (layout::print_any_fixits): Likewise.
10119 (selftest::test_one_liner_many_fixits): Rename to...
10120 (selftest::test_one_liner_many_fixits_1): ...this, and update
10121 comment and expected output to reflect that the multiple fix-it
10122 hints are now consolidated into one insertion.
10123 (selftest::test_one_liner_many_fixits_2): New test.
10124 (selftest::test_diagnostic_show_locus_one_liner): Update for
10126 (selftest::test_fixit_consolidation): Update for fix-it API
10128 * diagnostic.c (print_parseable_fixits): Likewise.
10129 * edit-context.c (edited_line::m_line_events): Convert from
10130 auto_vec <line_event *> to auto_vec <line_event>.
10131 (class line_event): Convert from abstract base class to a concrete
10132 class, taking over the role of replace_event.
10133 (class insert_event): Delete.
10134 (class replace_event): Rename to class line_event. Convert to
10136 (edit_context::add_fixits): Reimplement.
10137 (edit_context::apply_insert): Delete.
10138 (edit_context::apply_replace): Rename to...
10139 (edit_context::apply_fixit): ...this. Convert to half-open range.
10140 (edited_file::apply_insert): Delete.
10141 (edited_file::apply_replace): Rename to...
10142 (edited_file::apply_fixit): ...this.
10143 (edited_line::~edited_line): Drop deletion of events.
10144 (edited_line::apply_insert): Delete.
10145 (edited_line::apply_replace): Rename to...
10146 (edited_line::apply_fixit): ...this. Convert to half-open range.
10147 Update for change to type of m_line_events.
10148 * edit-context.h (edit_context::apply_insert): Delete.
10149 (edit_context::apply_replace): Rename to...
10150 (edit_context::apply_fixit): ...this.
10152 2017-05-01 Martin Sebor <msebor@redhat.com>
10154 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10157 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10160 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10161 __get_cpuid_max returns 0.
10162 (__get_cpuid_count): Ditto.
10164 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10166 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10167 replacement expression is another instance of one of its arguments.
10169 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10172 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10173 check for stack push/pop autoinc.
10174 * config/i386/i386.c (ix86_agi_dependent): Return false
10175 if the only reason why modified_in_p returned true is that
10176 addr is SP based and set_insn is a push or pop.
10178 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10180 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10183 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10186 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10187 (account_size_time): Use two predicates - exec_pred and
10189 (evaluate_conditions_for_known_args): Compute both clause and
10191 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10192 (inline_summary_t::duplicate): Update.
10193 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10195 (compute_inline_parameters): Likewise.
10196 (estimate_edge_size_and_time): Update caluclation of time.
10197 (estimate_node_size_and_time): Compute both time and nonspecialized
10199 (estimate_ipcp_clone_size_and_time): Update.
10200 (inline_merge_summary): Update.
10201 (do_estimate_edge_time): Update.
10202 (do_estimate_edge_size): Update.
10203 (do_estimate_edge_hints): Update.
10204 (inline_read_section, inline_write_summary): Stream both new predicates.
10205 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10207 (compute_inlined_call_time): Cleanup.
10208 (big_speedup_p): Update.
10209 (edge_badness): Update.
10210 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10211 (size_time_entry): Replace predicate by exec_predicate and
10212 nonconst_predicate.
10213 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10214 (estimate_edge_time): Return also nonspec_time.
10215 (reset_edge_growth_cache): Update.
10217 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10219 PR rtl-optimization/80491
10220 * ifcvt.c (noce_process_if_block): When looking for x setter
10221 with missing else_bb, don't check only the insn right before
10222 cond_earliest, but look for the last insn that x is modified in
10223 within the same bb.
10225 PR rtl-optimization/80491
10226 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10228 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10230 PR tree-optimization/80487
10231 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10233 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10235 PR tree-optimization/79697
10236 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10237 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10238 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10240 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10241 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10243 2017-04-28 Martin Sebor <msebor@redhat.com>
10245 PR tree-optimization/80523
10246 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10247 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10249 (maybe_warn, format_directive, parse_directive): Use new functions.
10250 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10252 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10254 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10256 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10258 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10259 target_header_dir): Set correctly.
10260 * configure: Regenerated.
10261 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10262 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10263 instead of SYSTEM_HEADER_DIR.
10265 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10267 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10268 (estimate_local_effects): Likewise.
10269 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10270 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10271 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10272 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10273 do_estimate_edge_time, estimate_edge_time): Likewise.
10274 * ipa-inline-analysis.c (estimate_node_size_and_time,
10275 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10276 (estimate_time_after_inlining): Remove.
10278 2017-04-28 Martin Liska <mliska@suse.cz>
10280 * doc/gcov.texi: Enhance documentation of gcov.
10282 2017-04-28 Martin Liska <mliska@suse.cz>
10284 * doc/gcov.texi: Sort options in alphabetic order.
10285 * doc/gcov-dump.texi: Likewise.
10286 * doc/gcov-tool.texi: Likewise.
10287 * gcov.c (print_usage): Likewise.
10288 * gcov-dump.c (print_usage): Likewise.
10289 * gcov-tool.c (print_merge_usage_message): Likewise.
10290 (print_rewrite_usage_message): Likewise.
10291 (print_overlap_usage_message): Likewise.
10293 2017-04-28 Martin Liska <mliska@suse.cz>
10295 PR gcov-profile/53915
10296 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10298 2017-04-28 Martin Liska <mliska@suse.cz>
10300 PR gcov-profile/79891
10301 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10302 is marked by compiler as living on a line.
10303 (get_cycles_count): Remove usage of the union.
10304 (output_intermediate_file): Likewise.
10305 (find_source): Fix GNU coding style.
10306 (accumulate_line_counts): Remove old non-all block mode.
10307 (output_lines): Remove usage of the union.
10308 * profile.c (output_location): Include all BBs, even if
10309 belonging to a same line (and file) as a previous BB.
10311 2017-04-28 Martin Liska <mliska@suse.cz>
10313 * gcov.c (process_args): Handle new argument 'w'.
10314 (read_graph_file): Assign ID to BBs.
10315 (output_branch_count): Display BB # if verbose flag is set.
10316 (output_lines): Likewise for arcs.
10317 (print_usage): Add '--verbose' option help.
10318 * doc/gcov.texi: Document --verbose (-w) option.
10320 2017-04-28 Martin Liska <mliska@suse.cz>
10322 * gcov.c (struct block_location_info): New struct.
10323 (process_file): Fill up the new structure.
10324 (read_graph_file): Replace usage of encoding by the newly added
10326 (add_line_counts): Likewise.
10327 (accumulate_line_counts): Remove usage of the union.
10328 (function_info::function_info): New function.
10329 (function_info::~function_info): Likewise.
10330 (process_file): Call delete instead of release_function.
10331 (release_function): Release the function.
10332 (release_structures): Call delete instead of release_function.
10333 (solve_flow_graph): Replace usage of num_blocks.
10334 (find_exception_blocks): Likewise.
10335 (output_lines): Fix GNU coding style.
10337 2017-04-28 Martin Liska <mliska@suse.cz>
10340 * coverage.c (coverage_remove_note_file): New function.
10341 * coverage.h: Declare the function.
10342 * toplev.c (finalize): Clean if an error has been seen.
10344 2017-04-28 Martin Liska <mliska@suse.cz>
10346 PR gcov-profile/80031
10347 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10348 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10349 * gcov.c (read_graph_file): Read just number of blocks.
10350 * profile.c (branch_prob): Do not stream 0 flags per a basic
10353 2017-04-28 Martin Liska <mliska@suse.cz>
10355 * gcov-dump.c (tag_*): Add new argument to declarations.
10356 (dump_gcov_file): Likewise.
10357 (tag_blocks): Add and use new argument depth.
10358 (tag_arcs): Likewise.
10359 (tag_lines): Likewise.
10360 (tag_counters): Likewise.
10361 (tag_summary): Likewise.
10362 (dump_working_sets): Use depth to do a proper indentation.
10364 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10367 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10368 * symtab.c (symtab_node::debug_symtab): Move definition here.
10370 2017-04-28 Richard Biener <rguenther@suse.de>
10372 * lto-streamer.h (LTO_major_version): Bump to 7.
10374 2017-04-28 Richard Biener <rguenther@suse.de>
10376 * tree-vrp.c (assert_info): New struct.
10377 (add_assert_info): New helper.
10378 (register_edge_assert_for_2): Refactor to add asserts to a vector
10380 (register_edge_assert_for_1): Likewise.
10381 (register_edge_assert_for): Likewise.
10382 (finish_register_edge_assert_for): New helper actually registering
10383 asserts where live on edge.
10384 (find_conditional_asserts): Adjust.
10385 (find_switch_asserts): Likewise.
10386 (evrp_dom_walker::try_find_new_range): Generalize.
10387 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10389 2017-04-27 Marek Polacek <polacek@redhat.com>
10392 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10393 arg10 and arg11 to itype.
10395 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10397 * doc/extend.texi (Object Size Checking): Improve grammar.
10399 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10402 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10403 that the logic for permitting reciprocal estimates matches that
10406 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10409 * tree.c (type_cache_hasher::equal): Only compare
10410 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10411 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10412 non-aggregate element types.
10413 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10414 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10416 2017-04-27 Richard Biener <rguenther@suse.de>
10418 PR middle-end/80533
10419 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10420 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10421 keeping a reference to a trailing array.
10423 2017-04-27 Richard Biener <rguenther@suse.de>
10425 PR middle-end/80539
10426 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10427 being in loop-closed SSA form conservatively.
10428 (chrec_fold_multiply_poly_poly): Likewise.
10430 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10432 PR middle-end/79665
10433 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10434 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10436 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10439 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10440 (aarch64_function_arg_alignment): Return unsigned int again, but still
10441 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10442 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10443 Don't emit -Wpsabi note.
10444 (aarch64_function_arg_boundary): Likewise.
10445 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10448 2017-04-26 Nathan Sidwell <nathan@acm.org>
10450 * tree.h (crc32_unsigned_n): Declare.
10451 (crc32_unsigned, crc32_unsigned): Make inline.
10452 * tree.c (crc32_unsigned_bits): Replace with ...
10453 (crc32_unsigned_n): ... this.
10454 (crc32_unsigned, crc32_byte): Remove.
10455 (crc32_string): Remove unnecessary braces.
10457 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10459 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10460 * ipa-inline-analysis.c (MAX_TIME): Remove.
10461 (account_size_time): Use sreal for time.
10462 (dump_inline_summary): Update.
10463 (estimate_function_body_sizes): Update.
10464 (estimate_edge_size_and_time): Update.
10465 (estimate_calls_size_and_time): Update.
10466 (estimate_node_size_and_time): Update.
10467 (inline_merge_summary): Update.
10468 (inline_update_overall_summary): Update.
10469 (estimate_time_after_inlining): Update.
10470 (inline_read_section): Update.
10471 (inline_write_summary): Update.
10472 * ipa-inline.c (compute_uninlined_call_time): Update.
10473 (compute_inlined_call_time): Update.
10474 (recursive_inlining): Update.
10475 (inline_small_functions): Update.
10476 (dump_overall_stats): Update.
10477 * ipa-inline.h: Include sreal.h.
10478 (size_time_entry): Turn time to sreal.
10479 (inline_summary): Turn self_time nad time to sreal.
10481 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10483 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10485 (sreal::stream_out, sreal::stream_in): New.
10486 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10488 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10490 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10493 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10496 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10498 (dimode_scalar_chain::compute_convert_gain): Ditto.
10499 (dimode_scalar_chain::make_vector_copies): Ditto.
10500 (dimode_scalar_chain::convert_reg): Ditto.
10501 (dimode_scalar_chain::convert_insn): Ditto.
10502 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10503 (VI248_AVX512BW_1): New mode iterator.
10504 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10505 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10508 2017-04-25 Martin Sebor <msebor@redhat.com>
10510 PR tree-optimization/80497
10511 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10512 constants are representable in HOST_WIDE_INT.
10513 (parse_directive): Ditto.
10515 2017-04-25 Martin Sebor <msebor@redhat.com>
10518 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10519 (new_zero_array): Adjust signature.
10520 (dom_info::dom_init): Used unsigned rather that size_t.
10521 (dom_info::dom_info): Same.
10523 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10524 Jakub Jelinek <jakub@redhat.com>
10527 * config/arm/arm.c: Include gimple.h.
10528 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10529 returns negative, increment ncrn only if it returned positive.
10530 (arm_needs_doubleword_align): Return int instead of bool,
10531 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10532 members, but if there is any such non-FIELD_DECL
10533 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10534 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10535 returns negative, increment nregs only if it returned positive.
10536 (arm_setup_incoming_varargs): Likewise.
10537 (arm_function_arg_boundary): Emit -Wpsabi note if
10538 arm_needs_doubleword_align returns negative, return
10539 DOUBLEWORD_ALIGNMENT only if it returned positive.
10541 2017-04-25 Marek Polacek <polacek@redhat.com>
10544 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10545 first argument to type.
10547 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10550 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10551 type checks to test for compatibility instead of equality.
10553 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10554 Jakub Jelinek <jakub@redhat.com>
10557 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10559 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10560 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10561 the alignment computation, but return their maximum in warn_alignment.
10562 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10563 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10565 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10566 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10569 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10571 * config/arc/simdext.md (dmpyh): Fix typo.
10573 2017-04-25 Richard Biener <rguenther@suse.de>
10575 PR tree-optimization/80492
10576 * alias.c (compare_base_decls): Handle registers with asm
10577 specification conservatively.
10578 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10579 compare_base_decls returning dont-know properly.
10581 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10583 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10584 (legitimate_offset_address_p): New function.
10585 (arc_legitimate_address_p): Use above function.
10587 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10589 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10591 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10593 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10594 ACCH registers whenever they are available.
10596 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10598 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10599 double regs fix when not used.
10601 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10603 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10605 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10606 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10608 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10610 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10611 when emitting short ADD instructions.
10613 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10615 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10617 (cmpsi_cc_c_insn): Likewise.
10618 (cbranchsi4_scratch): Compute proper instruction length using
10619 compact_hreg_operand.
10620 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10622 2017-04-25 Richard Biener <rguenther@suse.de>
10624 PR middle-end/80509
10625 * passes.c (pass_manager::pass_manager): Initialize
10626 m_name_to_pass_map.
10628 2017-04-25 Richard Biener <rguenther@suse.de>
10630 PR tree-optimization/79201
10631 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10633 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10636 * config/s390/vector.md: Split MEM->GPR vector moves for
10637 non-s_operand addresses.
10639 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10642 * config/s390/predicates.md (reload_const_wide_int_operand): New
10644 * config/s390/s390.md ("movti"): Remove d/P alternative.
10645 ("movti_bigconst"): New pattern definition.
10647 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10650 * s390-protos.h (s390_expand_cs_hqi): Removed.
10651 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10652 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10653 modes as well as CCZ1mode and CCZmode.
10654 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10655 signature of s390_emit_compare_and_swap.
10656 (s390_expand_cs_hqi): Likewise, make static.
10657 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10658 compare-and-swap, in some cases.
10659 (s390_expand_cs): Wrapper function.
10660 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10662 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10663 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10664 patterns for small and large integers. Forbid symref memory operands.
10665 Move expander to s390.c. Require cc register.
10666 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10667 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10668 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10669 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10670 symref memory operands. Remove CC mode and call s390_match_ccmode
10672 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10674 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10676 * config/s390/s390.md (define_peephole2): New peephole to help
10677 combining the load-and-test pattern with volatile memory.
10679 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10681 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10682 with CCZmode for TARGET_Z196.
10684 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10686 PR rtl-optimization/80501
10687 * combine.c (make_compound_operation_int): Set subreg_code to SET
10688 even for AND with mask of the sign bit of mode.
10690 PR rtl-optimization/80500
10691 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10692 sum's initial value.
10694 2017-04-25 Julian Brown <julian@codesourcery.com>
10695 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10697 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10699 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10701 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10703 2017-04-25 Julian Brown <julian@codesourcery.com>
10704 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10706 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10707 (thunderx2t99_sha): New Reservation.
10709 2017-04-25 Julian Brown <julian@codesourcery.com>
10710 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10712 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10713 type for 1-element load.
10715 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10717 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10719 2017-04-24 Martin Jambor <mjambor@suse.cz>
10721 PR tree-optimization/80293
10722 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10723 char arrays not totally scalarizable if it is false.
10724 (analyze_all_variable_accesses): Pass correct value in the new
10725 parameter. Add a statistics counter.
10727 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10729 PR middle-end/79931
10730 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10732 2017-04-24 Richard Biener <rguenther@suse.de>
10734 PR tree-optimization/80494
10735 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10736 out for complex types.
10738 2017-04-24 Richard Biener <rguenther@suse.de>
10740 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10741 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10742 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10743 (DFS): Adjust and never fail.
10744 (sccvn_dom_walker::fail): Remove.
10745 (sccvn_dom_walker::before_dom_children): Adjust.
10746 (run_scc_vn): Likewise and never fail.
10747 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10748 (pass_fre::execute): Likewise.
10750 2017-04-24 Richard Biener <rguenther@suse.de>
10752 PR tree-optimization/79725
10753 * tree-ssa-sink.c (statement_sink_location): Return whether
10754 failure reason was zero uses. Move that check later.
10755 (sink_code_in_bb): Deal with zero uses by removing the stmt
10758 2017-04-24 Richard Biener <rguenther@suse.de>
10761 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10762 pointer-based references.
10764 2017-04-24 Richard Biener <rguenther@suse.de>
10767 * pass_manager.h (pass_manager::operator new): Remove.
10768 (pass_manager::operator delete): Likewise.
10769 * passes.c (pass_manager::operator new): Remove.
10770 (pass_manager::operator delete): Likewise.
10771 (pass_manager::pass_manager): Zero individual pass members.
10773 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10776 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10777 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10778 Check "XEXP (src, 1)" operand here.
10779 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10780 Check "XEXP (src, 1)" operand here.
10781 (dimode_scalar_chain::make_vector_copies): Detect count register
10782 of a shift instruction. Zero extend count register from QImode
10783 to DImode to satisfy vector shift pattern count operand predicate.
10784 Substitute vector shift count operand with a DImode copy.
10785 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10788 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10790 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10791 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10792 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10793 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10794 (UNSPEC_NOREX_MEM): Remove definition.
10796 2017-04-21 Richard Biener <rguenther@suse.de>
10798 PR tree-optimization/79547
10799 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10800 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10801 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10802 without any constraints.
10804 2017-04-21 Richard Biener <rguenther@suse.de>
10806 PR tree-optimization/78847
10807 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10809 2017-04-21 Richard Biener <rguenther@suse.de>
10811 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10812 (build_distinct_type_copy): Likewise.
10813 (build_variant_type_copy): Likewise.
10814 * tree.c (build_qualified_type): Pass down mem-stat info.
10815 (build_distinct_type_copy): Likewise.
10816 (build_variant_type_copy): Likewise.
10818 2017-04-21 Richard Biener <rguenther@suse.de>
10820 PR tree-optimization/80237
10821 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10823 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10824 for a simplified result.
10826 2016-04-21 Richard Biener <rguenther@suse.de>
10828 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10829 sth as strict as a simple_iv but a chrec without symbols and an
10830 operand defined in the loop we are peeling (and not some subloop).
10831 (propagate_constants_for_unrolling): Propagate all constants.
10833 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10836 * config/i386/i386.c (print_reg): Remove assert for disalowed
10837 regno values, call output_operand_lossage instead.
10839 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10842 * config/i386/constraints.md (Yc): New register constraint.
10843 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10844 Use Yc constraint for alternative 2 of operand 0. Remove
10845 preferred_for_speed attribute.
10847 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10849 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10850 lastprivate clauses in SIMT case.
10852 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10854 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10856 2017-04-20 Richard Biener <rguenther@suse.de>
10858 PR tree-optimization/57796
10859 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10860 as N scalar stores.
10861 (vect_model_load_cost): Cost gathers as N scalar loads.
10863 2017-04-20 Richard Biener <rguenther@suse.de>
10865 * ggc-page.c (ggc_allocated_p): Rename to ...
10866 (safe_lookup_page_table_entry): ... this and return the lookup
10868 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10870 2017-04-20 Richard Biener <rguenther@suse.de>
10872 PR tree-optimization/80453
10873 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
10874 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
10875 from the conditions.
10876 (vn_phi_eq): Pass them down.
10877 (vn_phi_lookup): Record them.
10878 (vn_phi_insert): Likewise.
10880 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
10882 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
10883 uninitialized variable warning to avoid buffer overrun.
10885 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10888 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
10889 is suppressed for '{ 0 }' in C.
10891 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10893 * BASE-VER: Set to 8.0.0.
10895 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
10897 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
10898 priority .init_array and .fini_array section with SECTION_NOTYPE
10901 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10903 PR middle-end/80423
10904 * tree.h (build_array_type): Add typeless_storage default argument.
10905 * tree.c (type_cache_hasher::equal): Also compare
10906 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
10907 (build_array_type): Add typeless_storage argument, set
10908 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
10910 (build_nonshared_array_type): Adjust build_array_type_1 caller.
10911 (build_array_type): Likewise. Add typeless_storage argument.
10913 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10914 Jakub Jelinek <jakub@redhat.com>
10916 PR tree-optimization/80426
10917 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
10918 operation on symbolic operands, also compute the overflow for the
10919 invariant part when the operation degenerates into a negation.
10921 2017-04-19 Jakub Jelinek <jakub@redhat.com>
10924 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
10925 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
10928 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
10930 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
10933 * config/avr/avr.c (tree.h): Include it.
10934 (cgraph.h): Include it.
10935 (avr_encode_section_info): Don't warn for uninitialized progmem
10936 variable if it's just an alias.
10938 2017-04-19 Richard Biener <rguenther@suse.de>
10941 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
10942 when needed by AutoPGO.
10944 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
10947 * doc/lto.texi: Remove an extra 'that'.
10949 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
10951 PR rtl-optimization/80429
10952 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
10953 are only used in debug insns.
10955 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10956 Vladimir Makarov <vmakarov@redhat.com>
10958 * config/sparc/predicates.md (input_operand): Add comment. Return
10959 true for any memory operand when LRA is in progress.
10960 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
10962 2017-04-18 Jeff Law <law@redhat.com>
10965 * mips.md ({return,simple_return}_internal): Do not overwrite
10968 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10970 PR tree-optimization/80443
10971 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
10972 instead of adding 1, subtract -1 and similarly instead of subtracting
10975 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
10977 PR rtl-optimization/80357
10978 * haifa-sched.c (tmp_bitmap): New variable.
10979 (model_recompute): Handle duplicate use records.
10980 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
10981 (free_global_sched_pressure_data): Free it.
10983 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
10986 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10987 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10988 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10989 instead of SYSTEM_HEADER_DIR.
10991 2017-04-18 Jeff Law <law@redhat.com>
10993 PR middle-end/80422
10994 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
10995 predecessors after walking up the insn chain.
10997 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11000 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11001 sizetype type into debug info.
11003 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11006 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11007 unneeded test for TARGET_UPPER_REGS_SF.
11008 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11010 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11013 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11014 instead of gsi_after_labels.
11016 2017-04-18 Jeff Law <law@redhat.com>
11018 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11022 2017-04-13 Jeff Law <law@redhat.com>
11023 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11024 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11026 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11029 * config/avr/avr.c (intl.h): Include it.
11030 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11032 2017-04-18 Martin Liska <mliska@suse.cz>
11034 PR gcov-profile/78783
11035 * gcov-tool.c (gcov_output_files): Validate that destination
11036 file is either removed by the tool or by a user.
11038 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11039 Guy Benyei <guybe@mellanox.com>
11041 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11042 block, and do not negate it, the stored id is already negative.
11044 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11046 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11048 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11051 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11052 masks of options that should be turned off if the VSX vector
11053 options are turned off.
11054 (OTHER_P8_VECTOR_MASKS): Likewise.
11055 (OTHER_VSX_VECTOR_MASKS): Likewise.
11056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11057 rs6000_disable_incompatible_switches to validate no type switches
11059 (rs6000_incompatible_switch): New function to disallow turning on
11060 other vector options if -mno-vsx, -mno-power8-vector, or
11061 -mno-power9-vector are specified.
11063 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11065 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11067 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11069 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11070 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11071 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11072 (ARG_POINTER_CFA_OFFSET): Likewise.
11074 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11076 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11077 conditions to take advantage of various optimizations.
11079 2017-04-13 Jeff Law <law@redhat.com>
11081 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11082 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11083 (zero_extendsidi2_dext): Likewise.
11085 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11088 * fold-const.c (fold_ternary_loc): Revert
11089 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11092 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11094 PR rtl-optimization/80343
11095 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11096 new scratch pseudo.
11098 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11101 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11102 to ubsan_encode_value.
11104 2017-04-13 Jeff Law <law@redhat.com>
11106 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11107 appearing in DEBUG_INSNs.
11109 2017-04-13 Martin Liska <mliska@suse.cz>
11111 PR gcov-profile/80413
11112 * gcov-io.c (gcov_write_string): Copy to buffer just when
11113 allocated size is greater than zero.
11115 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11118 * dwarf2out.c (decls_for_scope): Ignore declarations of
11119 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11121 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11124 * ipa-visibility.c (non_local_p): Fix typos.
11125 (localize_node): When localizing symbol in same comdat group,
11126 dissolve the group only when we know external symbols are going
11128 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11130 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11132 PR tree-optimization/79390
11133 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11134 order does not result in usable sequence, retry with reversed operand
11140 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11141 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11142 op0 instead of fold_convert_loc (loc, type, arg0).
11144 2017-04-12 Jeff Law <law@redhat.com>
11146 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11147 has a delay slot in the generated code.
11149 * config/cris/cris.md (cris_preferred_reload_class): Return
11150 GENNONACR_REGS rather than GENERAL_REGS.
11152 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11155 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11156 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11157 signedness of the result type.
11159 2017-04-12 Richard Biener <rguenther@suse.de>
11160 Jeff Law <law@redhat.com>
11162 PR tree-optimization/80359
11163 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11164 trim stores to TARGET_MEM_REFs.
11166 2017-04-12 Richard Biener <rguenther@suse.de>
11168 PR tree-optimization/79390
11169 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11170 threading case even more.
11172 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11175 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11176 for quad_address_p for TImode, instead of just not indexed_address.
11178 2017-04-12 Richard Biener <rguenther@suse.de>
11179 Bernd Edlinger <bernd.edlinger@hotmail.de>
11181 PR middle-end/79671
11182 * alias.c (component_uses_parent_alias_set_from): Handle
11183 TYPE_TYPELESS_STORAGE.
11184 (get_alias_set): Likewise.
11185 * tree-core.h (tree_type_common): Add typeless_storage flag.
11186 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11187 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11188 for types containing members with TYPE_TYPELESS_STORAGE.
11189 (place_field): Likewise.
11190 (layout_type): Likewise for ARRAY_TYPE.
11191 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11192 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11193 TYPE_TYPELESS_STORAGE.
11194 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11196 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11199 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11200 first argument to type.
11202 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11206 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11207 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11208 (rs6000_expand_binop_builtin): Likewise.
11209 (rs6000_expand_ternop_builtin): Likewise; also add missing
11210 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11211 vshasigma built-ins.
11212 * doc/extend.texi: Document that vec_xxpermdi's third argument
11213 must be a constant.
11215 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11217 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11218 Use shift_const cost parameter when calculating gain of STV shifts.
11220 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11222 PR rtl-optimization/70478
11223 * lra-constraints.c (process_alt_operands): Check memory for
11224 disfavoring memory insn operand.
11226 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11228 PR middle-end/80100
11229 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11230 left shift in unsigned HOST_WIDE_INT type.
11232 PR rtl-optimization/80385
11233 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11234 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11237 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11238 if they have any depend clauses.
11240 2017-04-11 Martin Liska <mliska@suse.cz>
11243 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11244 * ipa-split.c (split_function): Create a local comdat symbol
11245 if caller is in a comdat group.
11247 2017-04-11 Martin Liska <mliska@suse.cz>
11250 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11253 2017-04-11 Martin Sebor <msebor@redhat.com>
11255 PR middle-end/80364
11256 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11257 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11259 (directive::set_width, directive::set_precision, format_character):
11261 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11264 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11267 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11268 conflict, set target->arch_name instead of target->cpu_name.
11270 2017-04-11 Richard Biener <rguenther@suse.de>
11272 PR tree-optimization/80374
11273 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11274 build_zero_cst, remove fold_convertible_p check again.
11276 2017-04-11 Martin Liska <mliska@suse.cz>
11279 * ubsan.c (instrument_object_size): Do not instrument register
11282 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11285 * config/i386/i386-builtin-types.def
11286 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11287 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11288 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11289 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11290 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11291 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11292 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11293 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11294 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11295 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11296 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11297 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11298 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11299 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11300 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11301 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11302 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11303 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11304 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11305 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11306 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11307 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11308 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11309 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11310 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11311 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11312 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11313 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11314 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11315 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11316 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11317 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11318 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11319 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11320 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11321 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11322 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11323 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11324 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11325 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11326 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11327 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11328 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11329 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11330 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11331 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11333 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11334 flag to second_arg_count, handle 4 argument function type _COUNT
11335 aliases, handle second_arg_count on second argument rather than last.
11337 2017-04-10 Jeff Law <law@redhat.com>
11339 PR tree-optimization/80374
11340 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11341 record anything if we can not convert integer_zero_node to the
11344 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11347 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11348 Enhance special handling given to the TARGET_P9_MINMAX option in
11349 relation to certain other options.
11351 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11353 PR tree-optimization/80153
11354 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11355 remove POINTER_PLUS_EXPR's base part directly, rather than through
11358 2017-04-10 Richard Biener <rguenther@suse.de>
11359 Bin Cheng <bin.cheng@arm.com>
11361 PR tree-optimization/80153
11362 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11363 the first element of pointer type aff_tree. Build result expr in
11365 (add_elt_to_tree): Convert to type unconditionally. Remove other
11366 fold_convert calls.
11367 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11368 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11370 2017-04-10 Richard Biener <rguenther@suse.de>
11372 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11375 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11377 PR rtl-optimization/70478
11378 * lra-constraints.c (curr_small_class_check): New.
11379 (update_and_check_small_class_inputs): New.
11380 (process_alt_operands): Update curr_small_class_check. Disfavor
11381 alternative insn memory operands. Check available regs for small
11384 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11387 * config/mips/mips.opt (-mvirt): Update description.
11388 * doc/invoke.texi (-mvirt): Likewise.
11390 2017-04-10 Richard Biener <rguenther@suse.de>
11392 PR middle-end/80362
11393 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11394 looking for NEGATE_EXPR in -A / -B to A / B folding.
11396 2017-04-10 Martin Liska <mliska@suse.cz>
11398 PR gcov-profile/80224
11399 * gcov.c (print_usage): Fix usage string.
11400 (get_gcov_intermediate_filename): Remove.
11401 (output_gcov_file): Use both for normal and intermediate format.
11402 (generate_results): Do not initialize special file for
11403 intermediate format.
11405 2017-04-10 Richard Biener <rguenther@suse.de>
11407 PR tree-optimization/80304
11408 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11411 2017-04-10 Nathan Sidwell <nathan@acm.org>
11414 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11415 (rs6000_init_builtins): Use it.
11417 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11419 * config/arm/arm.md (<mrc>): Add mode to SET source.
11420 (<mrrc>): Likewise.
11422 2017-04-10 Richard Biener <rguenther@suse.de>
11424 PR middle-end/80344
11425 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11427 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11430 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11431 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11432 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11433 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11434 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11435 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11436 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11437 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11438 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11439 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11440 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11441 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11442 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11443 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11444 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11445 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11446 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11447 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11448 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11449 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11450 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11451 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11452 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11454 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11456 PR rtl-optimization/70478
11457 * lra-constraints.c: Reverse the last patch.
11459 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11461 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11462 Add comment for WCHAR_T.
11464 2017-04-08 Martin Liska <mliska@suse.cz>
11467 2017-04-07 Martin Liska <mliska@suse.cz>
11470 * ipa-split.c (split_function): Add function part to a same comdat
11473 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11476 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11478 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11480 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11481 (vec_pairing): ...to this.
11482 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11483 (rs6000_sched_init): Adjust for name change.
11484 (struct rs6000_sched_context): Likewise.
11485 (rs6000_init_sched_context): Likewise.
11486 (rs6000_set_sched_context): Likewise.
11488 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11494 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11496 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11497 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11498 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11500 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11502 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11504 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11506 PR rtl-optimization/70703
11507 * ira-color.c (update_conflict_hard_regno_costs): Use
11508 int64_t instead of HOST_WIDE_INT.
11510 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11512 PR rtl-optimization/70478
11513 * lra-constraints.c (process_alt_operands): Disfavor alternative
11514 insn memory operands.
11516 2017-04-07 Jeff Law <law@redhat.com>
11518 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11519 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11521 2017-04-07 Martin Liska <mliska@suse.cz>
11524 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11525 Show error message instead of an ICE.
11527 2017-04-07 Martin Liska <mliska@suse.cz>
11530 * ipa-split.c (split_function): Add function part to a same comdat
11533 2017-04-07 Richard Biener <rguenther@suse.de>
11535 PR middle-end/80341
11536 * tree.c (get_unwidened): Also handle ! for_type case for
11538 * convert.c (do_narrow): Split out from ...
11539 (convert_to_integer_1): ... here. Do not pass final truncation
11540 type to get_unwidened for TRUNC_DIV_EXPR.
11542 2017-04-07 Richard Biener <rguenther@suse.de>
11544 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11546 (aff_combination_const): Adjust.
11547 (aff_combination_scale): Likewise.
11548 (aff_combination_add_elt): Likewise.
11549 (aff_combination_add_cst): Likewise.
11550 (aff_combination_convert): Likewise.
11551 (add_elt_to_tree): Likewise. Remove unused argument.
11552 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11554 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11556 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11558 * config/arm/arm.c (arm_default_short_enums): Use
11559 ARM_DEFAULT_SHORT_ENUMS.
11560 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11562 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11565 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11566 members with redundant out-of-class redeclaration.
11568 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11571 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11572 * config/i386/i386.md (*zero_extendsidi2):
11573 Add (?*x,*x) and (?*v,*v) alternatives.
11575 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11578 * config/i386/i386.c (ix86_expand_builtin)
11579 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11580 mode from insn data. Convert operands to insn operand mode.
11581 Copy operands that don't satisfy insn predicate to a register.
11583 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11585 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11588 2017-04-06 Richard Biener <rguenther@suse.de>
11590 PR tree-optimization/80334
11591 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11592 preserve alignment of accesses.
11594 2017-04-06 Richard Biener <rguenther@suse.de>
11596 PR tree-optimization/80262
11597 * tree-sra.c (build_ref_for_offset): Preserve address-space
11599 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11600 Drop useless address-space information on MEM_REF offsets.
11602 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11604 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11606 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11608 PR rtl-optimization/70703
11609 * ira-color.c (update_conflict_hard_regno_costs): Use
11610 HOST_WIDE_INT instead of long.
11612 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11615 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11616 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11618 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11619 for x86_64 target. Handle -m3dnowa option.
11621 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11623 PR rtl-optimization/70703
11624 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11625 (update_conflict_hard_regno_costs): Use long instead of unsigned
11626 arithmetic for cost calculation.
11628 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11629 Bernd Edlinger <bernd.edlinger@hotmail.de>
11632 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11635 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11638 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11639 ptr_mode with Pmode throughout.
11640 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11641 into probe_stack_range and use DImode.
11643 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11646 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11647 call_eh_return is true.
11649 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11651 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11652 Initialize last_match_fntype_index.
11654 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11657 * tree-nvr.c: Include internal-fn.h.
11658 (pass_return_slot::execute): Ignore internal calls without
11661 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11662 Richard Biener <rguenther@suse.de>
11665 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11666 captures used multiple times, except for the last use.
11667 * generic-match-head.c: Include gimplify.h.
11669 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11671 PR tree-optimization/79390
11672 * target.h (struct noce_if_info): Declare.
11673 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11674 * target.def (noce_conversion_profitable_p): New target hook.
11675 * ifcvt.h (struct noce_if_info): New type, moved from ...
11676 * ifcvt.c (struct noce_if_info): ... here.
11677 (noce_conversion_profitable_p): Renamed to ...
11678 (default_noce_conversion_profitable_p): ... this. No longer
11680 (noce_try_store_flag_constants, noce_try_addcc,
11681 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11682 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11683 instead of noce_conversion_profitable_p.
11684 * config/i386/i386.c: Include ifcvt.h.
11685 (ix86_option_override_internal): Don't override
11686 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11687 (ix86_noce_conversion_profitable_p): New function.
11688 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11689 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11690 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11691 * doc/tm.texi: Regenerated.
11693 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11695 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11698 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11701 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11702 instructions for small multiply cores.
11704 2017-04-04 Jeff Law <law@redhat.com>
11706 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11708 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11709 that are not set by the loop over the elements.
11711 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11714 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11715 int mode, convert_modes it to mode as unsigned, otherwise use
11716 lowpart_subreg to mode rather than SImode.
11717 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11718 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11719 Use DImode instead of SImode for the shift count operand.
11720 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11723 2017-04-04 Richard Biener <rguenther@suse.de>
11725 PR middle-end/80281
11726 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11727 arithmetic done for the negate or the plus. Simplify.
11728 (A - (-B) -> A + B): Likewise.
11729 * fold-const.c (split_tree): Make sure to not negate pointers.
11731 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11733 PR rtl-optimization/60818
11734 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11735 a compare of comparisons with the thing compared if this results
11736 in a different machine mode.
11738 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11740 * alias.c (base_alias_check): Fix typo in comment.
11741 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11742 * cgraphunit.c (symbol_table::compile): Likewise.
11743 * collect2.c (maybe_run_lto_and_relink): Likewise.
11744 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11745 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11746 * config/avr/avr.c (avr_map_op_t): Likewise.
11747 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11748 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11749 * config/epiphany/epiphany.md (movcc): Likewise.
11750 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11751 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11753 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11754 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11755 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11756 * config/sh/sh.c (sh_rtx_costs): Likewise.
11757 * fold-const.c (fold_truth_andor): Likewise.
11758 * genautomata.c (collapse_flag): Likewise.
11759 * gengtype.h (struct type::u::s): Likewise.
11760 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11761 * input.c (FORMAT_AMOUNT): Likewise.
11762 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11763 (known_aggs_to_agg_replacement_list): Likewise.
11764 * ipa-inline-analysis.c: Likewise.
11765 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11766 * ipa-polymorphic-call.c
11767 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11768 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11769 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11771 * modulo-sched.c (apply_reg_moves): Likewise.
11772 * omp-expand.c (build_omp_regions_1): Likewise.
11773 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11774 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11775 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11776 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11777 * value-prof.c: Likewise.
11778 * var-tracking.c (val_reset): Likewise.
11780 2017-04-03 Richard Biener <rguenther@suse.de>
11782 PR tree-optimization/80275
11783 * fold-const.c (split_address_to_core_and_offset): Handle
11786 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11788 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11789 descriptors is at least equal to that of functions.
11791 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11793 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11795 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11798 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11799 (mov<IMOD4:mode>): New expander.
11800 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11802 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11804 PR rtl-optimization/79405
11805 * fwprop.c (propagations_left): New variable.
11806 (forward_propagate_into): Decrement it.
11807 (fwprop_init): Initialize it.
11808 (fw_prop): If the variable has reached zero, stop propagating.
11809 (fwprop_addr): Ditto.
11811 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11814 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11815 a FUNCTION_DECL, pass it as decl instead of origin to
11818 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11820 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11823 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11826 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11827 TARGET_VSX_SMALL_INTEGER.
11829 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11831 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11832 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11834 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11836 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11837 extraction from odd-numbered MSA register.
11839 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11841 PR middle-end/80173
11842 * expmed.c (store_bit_field_1): Don't attempt to create
11843 a word subreg out of hard registers wider than word if they
11844 have HARD_REGNO_NREGS of 1 for their mode.
11846 PR middle-end/80163
11847 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11848 conversions to integer types wider than word and pointer.
11851 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11852 (rtx_equal_for_cselib_p): Pass 0 to it.
11853 * cselib.c (cselib_hasher::equal): Likewise.
11854 (rtx_equal_for_cselib_1): Add depth argument. If depth
11855 is 128, don't look up VALUE locs and punt. Increment
11856 depth in recursive calls when walking VALUE locs.
11858 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11860 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11861 (make_gcov_file_name): Use the canonical path name for generating
11863 (read_line): Fix handling of files with ascii null bytes.
11865 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11867 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11868 to initialise a vector register instead
11869 of using a const_int.
11871 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11873 PR translation/80189
11874 * gimplify.c (omp_default_clause): Use %qs instead of %s in
11875 diagnostic messages.
11877 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
11880 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
11881 (dfp_diex_<mode>): Update mode of operand 1.
11882 * doc/extend.texi (dxex, dxexq): Document change to return type.
11883 (diex, diexq): Document change to argument type.
11885 2017-03-30 Martin Jambor <mjambor@suse.cz>
11888 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
11889 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
11890 it reflects the signature changes performed at the callee side.
11891 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
11892 to cgraph_build_function_type_skip_args.
11893 (build_function_decl_skip_args): Adjust call to the above function.
11895 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11898 * config/i386/sse.md
11899 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
11900 register as dest whenever it is a MEM not rtx_equal_p to the
11901 corresponding dup operand, and when forcing into reg move the
11902 reg into the memory afterwards.
11903 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
11904 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
11905 for the force_reg mode.
11906 (avx512vl_vextractf128<mode>): Use register as dest either
11907 always when a MEM, or when it is a MEM not rtx_equal_p to the
11908 corresponding dup operand, or even not when it is a CONST_VECTOR
11909 depending on the mode and lo vs. hi.
11910 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
11912 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
11913 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
11914 Likewise. Require that operands[2] is even.
11915 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
11916 Remove extraneous parens. Require that operands[2] is a multiple
11918 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
11919 operands[0] is a MEM if <mask_applied>, the predicates/constraints
11920 disallow memory then.
11922 2017-03-30 Richard Biener <rguenther@suse.de>
11924 PR tree-optimization/77498
11925 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
11926 to non-constants over backedges.
11928 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
11930 PR rtl-optimization/80233
11931 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
11932 as last_combined_insn. Do not test for BARRIER_P separately.
11934 2017-03-29 Andreas Schwab <schwab@suse.de>
11937 * calls.c (prepare_call_address): Convert funexp to Pmode before
11938 copying to temp reg.
11940 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11942 PR tree-optimization/80158
11943 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
11944 Handle possible future case of more than one alternate
11946 (replace_rhs_if_not_dup): Likewise.
11947 (replace_one_candidate): Likewise.
11949 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
11951 PR rtl-optimization/80193
11952 * ira.c (ira): Do not check allocation for LRA.
11954 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
11956 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
11957 (nvptx_output_simt_exit): Declare.
11958 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
11959 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
11960 (init_softstack_frame): Move initialization of crtl->is_leaf to...
11961 (nvptx_declare_function_name): ...here. Emit declaration of local
11962 memory space buffer for omp_simt_enter insn.
11963 (nvptx_output_unisimt_switch): New.
11964 (nvptx_output_softstack_switch): New.
11965 (nvptx_output_simt_enter): New.
11966 (nvptx_output_simt_exit): New.
11967 * config/nvptx/nvptx.h (struct machine_function): New fields
11968 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
11969 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
11970 (UNSPECV_SIMT_EXIT): Ditto.
11971 (omp_simt_enter_insn): New insn.
11972 (omp_simt_enter): New expansion.
11973 (omp_simt_exit): New insn.
11974 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
11976 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
11977 (expand_GOMP_SIMT_ENTER_ALLOC): New.
11978 (expand_GOMP_SIMT_EXIT): New.
11979 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
11980 (GOMP_SIMT_ENTER_ALLOC): Ditto.
11981 (GOMP_SIMT_EXIT): Ditto.
11982 * target-insns.def (omp_simt_enter): New insn.
11983 (omp_simt_exit): Ditto.
11984 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
11986 (lower_rec_simd_input_clauses): Implement SIMT privatization.
11987 (lower_rec_input_clauses): Likewise.
11988 (lower_lastprivate_clauses): Handle SIMT privatization.
11990 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
11991 (ompdevlow_adjust_simt_enter): New.
11992 (find_simtpriv_var_op): New.
11993 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
11994 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
11996 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
11997 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
11998 (copy_decl_for_dup_finish): Ditto.
12000 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12002 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12005 * config/i386/i386.c (ix86_option_override_internal): Always
12006 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12008 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12010 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12012 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12014 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12015 mark new edge's irreducible flag accordign to it.
12016 (vect_do_peeling): Check loop preheader edge's irreducible flag
12017 and pass it to function slpeel_add_loop_guard.
12019 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12021 PR tree-optimization/80218
12022 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12023 Update block frequencies and counts.
12025 2017-03-28 Richard Biener <rguenther@suse.de>
12027 PR tree-optimization/78644
12028 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12029 of a simplification result we may not use it at all.
12031 2017-03-28 Richard Biener <rguenther@suse.de>
12034 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12035 without arguments, generate default definition of a SSA name.
12037 2017-03-28 Richard Biener <rguenther@suse.de>
12039 PR middle-end/80222
12040 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12041 TYPE_REF_CAN_ALIAS_ALL references.
12042 * fold-const.c (fold_indirect_ref_1): Likewise.
12044 2017-03-28 Martin Liska <mliska@suse.cz>
12047 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12048 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12050 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12051 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12053 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12054 (EXTRA_SPECS): Define.
12055 (SUBTARGET_EXTRA_SPECS): Likewise.
12056 (SUBTARGET_CPP_SPEC): Likewise.
12057 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12058 SUBTARGET_EXTRA_SPECS.
12059 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12061 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12063 * config/arc/simdext.md (vst64_insn): Update pattern.
12064 (vld32wh_insn): Likewise.
12065 (vld32wl_insn): Likewise.
12066 (vld64_insn): Likewise.
12067 (vld32_insn): Likewise.
12069 2017-03-28 Marek Polacek <polacek@redhat.com>
12072 * fold-const.c (fold_comparison): Use protected_set_expr_location
12073 instead of SET_EXPR_LOCATION.
12075 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12077 * tree.c (add_expr): Avoid name lookup warning.
12079 2017-03-27 Jeff Law <law@redhat.com>
12081 PR tree-optimization/80216
12082 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12083 function name. Limit recursion depth.
12084 (record_temporary_equivalences): Corresponding changes.
12086 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12088 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12091 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12094 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12096 * cfgcleanup.c (reg_note_cfa_p): New array.
12097 (insns_have_identical_cfa_notes): New function.
12098 (old_insns_match_p): Don't cross-jump in between /f
12099 and non-/f instructions. If both i1 and i2 are frame related,
12100 verify all CFA notes, their order and content.
12102 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12105 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12106 HImode and SImode with zero extend to DImode to one insn.
12107 (bswap<mode>2_extenddi): Likewise.
12108 (bswapsi2_extenddi): Likewise.
12109 (bswaphi2_extendsi): Likewise.
12110 (bswaphi2): Combine bswap HImode and SImode into one insn.
12111 Separate memory insns from swapping register.
12112 (bswapsi2): Likewise.
12113 (bswap<mode>2): Likewise.
12114 (bswaphi2_internal): Delete, no longer used.
12115 (bswapsi2_internal): Likewise.
12116 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12117 store, and gpr<-gpr swap insns.
12118 (bswap<mode>2_store): Likewise.
12119 (bswaphi2_reg): Register only splitter, combine with the splitter.
12120 (bswaphi2 splitter): Likewise.
12121 (bswapsi2_reg): Likewise.
12122 (bswapsi2 splitter): Likewise.
12123 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12124 the insns into load, store, and register/register insns.
12125 (bswapdi2_ldbrx): Likewise.
12126 (bswapdi2_load): Likewise.
12127 (bswapdi2_store): Likewise.
12128 (bswapdi2_reg): Likewise.
12130 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12132 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12133 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12135 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12138 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12140 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12141 special handling for target option conflicts between dform
12142 options (-mpower9-dform, -mpower9-dform-vector,
12143 -mpower9-dform-scalar) and -mno-direct-move.
12145 2017-03-27 Richard Biener <rguenther@suse.de>
12147 PR tree-optimization/80181
12148 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12150 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12152 * config/arc/predicates.md (move_double_src_operand): Replace the
12153 call to move_double_src_operand with a call to address_operand.
12155 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12157 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12158 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12159 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12161 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12163 * config/arc/predicates.md (long_immediate_loadstore_operand):
12164 Consider scaled addresses cases.
12166 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12168 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12169 restored when in interrupt.
12170 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12171 doesn't have delay slot.
12173 2017-03-27 Richard Biener <rguenther@suse.de>
12176 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12177 inlined thunk clones.
12179 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12182 * asan.c (instrument_derefs): Copy over last operand from
12183 original COMPONENT_REF to the new COMPONENT_REF with
12184 DECL_BIT_FIELD_REPRESENTATIVE.
12185 * ubsan.c (instrument_object_size): Likewise.
12187 2017-03-27 Richard Biener <rguenther@suse.de>
12189 PR tree-optimization/80170
12190 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12191 sure DR/SCEV didnt fold in constants we do not see when looking
12192 at the reference base alignment.
12194 2017-03-27 Richard Biener <rguenther@suse.de>
12196 PR middle-end/80171
12197 * gimple-fold.c (fold_ctor_reference): Properly guard against
12198 NULL return value from canonicalize_constructor_val.
12200 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12203 * config/i386/i386.c (ix86_expand_builtin)
12204 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12205 flags reg setting and flags reg using instructions.
12206 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12207 clobbering instructions to zero extend op2.
12209 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12211 * doc/install.texi (Configuration) <--with-aix-soname>:
12212 Update link to AIX ld.
12214 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12216 PR rtl-optimization/80160
12217 PR rtl-optimization/80159
12218 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12219 reg_alternate_class into account.
12221 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12224 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12225 to consider in curr_insn_transform.
12227 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12229 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12230 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12231 and emit_mode_inner.
12233 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12235 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12236 argument to the overloaded builtin variants. Use the new flag to
12237 deprecate certain builtin variants.
12238 * config/s390/s390-builtin-types.def: Add new builtin types.
12239 * config/s390/s390-builtins.h: Support new flags field for
12240 overloaded builtins.
12241 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12242 (s390_macro_to_expand): Enable vector float data type.
12243 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12244 builtins by incrementing the __VEC__ version number.
12245 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12247 (s390_resolve_overloaded_builtin): Emit error messages depending
12248 on the builtin flags.
12249 * config/s390/s390.c (s390_expand_builtin): Support additional
12250 flags argument. Change error message to match the messages
12251 emitted in s390-c.c.
12252 * config/s390/s390.md: New UNSPEC_* constants.
12253 (op_type): Add new instruction types.
12254 * config/s390/vecintrin.h: Add new builtins and test data class
12256 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12257 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12258 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12259 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12260 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12261 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12263 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12264 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12265 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12266 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12268 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12269 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12270 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12271 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12273 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12274 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12275 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12276 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12277 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12278 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12279 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12281 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12282 ("vec_scatter_element<V_HW_4:mode>_DI")
12283 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12284 ("vec_fpint<mode>", "vflls")
12285 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12286 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12287 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12288 ("*vec_cmphe<mode>_cc"): ... these.
12290 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12291 mode constant instead of magic value.
12293 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12295 * config/s390/s390.c (s390_expand_vec_compare): Support other
12296 vector floating point modes than just V2DF.
12297 (s390_expand_vcond): Likewise.
12298 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12299 (s390_cannot_change_mode_class): Prevent mode changes between TF
12300 and V1TF in vector registers.
12301 * config/s390/s390.md (DF, SF): New mode attributes.
12302 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12303 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12304 SFmode support for VRs.
12305 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12307 (VFT, VF_HW): New mode iterators.
12308 (vw, sdx): New mode attributes.
12309 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12310 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12311 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12312 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12313 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12314 also the new vector floating point modes. Renaming to ...
12316 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12317 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12318 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12319 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12320 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12321 ("vec_unordered<mode>"): ... these.
12323 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12324 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12325 ("*vec_extendv2df"): New insn definitions.
12327 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12329 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12330 ("mulditi3_2", "*muldi3_sign"): New patterns.
12331 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12332 rename the pattern definition.
12334 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12336 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12338 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12340 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12342 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12343 instruction if possible.
12344 * config/s390/vector.md (vec_halfnumelts): New mode
12346 ("*vec_vllezlf<mode>"): New pattern.
12348 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12350 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12351 ("popcountv4si2", "popcountv2di2"): Rename to ...
12352 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12353 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12355 ("popcount<mode>2_vxe"): New pattern.
12357 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12359 * common/config/s390/s390-common.c (processor_flags_table): Add
12361 * config.gcc: Add arch12.
12362 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12363 Default to arch12 for unknown CPU model numbers.
12364 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12365 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12366 PROCESSOR_max sanity check.
12367 * config/s390/s390-opts.h (enum processor_type): Add
12369 * config/s390/s390.c (processor_table): Add arch12.
12370 (s390_expand_builtin): Add check for B_VXE flag.
12371 (s390_issue_rate): Add PROCESSOR_ARCH12.
12372 (s390_get_sched_attrmask): Likewise.
12373 (s390_get_unit_mask): Likewise.
12374 (s390_sched_score): Enable z13 scheduling for arch12.
12375 (s390_sched_reorder): Likewise.
12376 (s390_sched_variable_issue): Likewise.
12377 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12379 (s390_tune_attr): Use z13 scheduling also for arch12.
12380 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12381 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12382 (TARGET_VXE_P): New macros.
12383 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12384 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12385 * config/s390/s390.opt: Add arch12 as processor_type.
12387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12389 * config/s390/s390.md
12390 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12391 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12392 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12394 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12395 Rename expanders to ...
12397 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12398 ("fixuns_truncdddi2_emu"): ... these.
12400 ("fixuns_trunc<mode>si2_emu"): New expander.
12402 ("*fixuns_truncdfdi2_z13"): Rename to ...
12403 ("*fixuns_truncdfdi2_vx"): ... this.
12405 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12407 * config/s390/2964.md: Remove the single element vector compare
12408 instructions which are no longer used.
12409 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12411 (s390_canonicalize_comparison): Remove handling of DFmode
12413 (s390_expand_vec_compare_scalar): Remove function.
12414 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12415 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12417 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12419 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12421 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12423 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12424 will anyway by matched by mov<mode>_64dfp.
12426 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12428 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12429 vlef/vstef. Add missing operand to vleif.
12431 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12433 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12434 pair for all vector types with 64 bit elements.
12435 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12436 * config/s390/vector.md (V_HW_64): ... here.
12437 (V_128_NOSINGLE): New mode iterator.
12438 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12439 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12440 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12441 ("*vec_load_pairv2di"): Change to ...
12442 ("*vec_load_pair<mode>"): ... this one.
12444 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12446 * config/s390/constraints.md: Add comments.
12447 (jKK): Reject element sizes > 8 bytes.
12448 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12450 * config/s390/s390.md: Add the s_operand checks formerly in
12451 s390_split_ok_p to various splitters where they are still
12453 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12454 for 128 bit vectors. Plus two splitters.
12456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12458 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12461 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12464 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12465 error if the boundary argument is not constant.
12467 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12469 PR rtl-optimization/80112
12470 * loop-doloop.c (doloop_condition_get): Don't check condition
12471 if cmp isn't SET with IF_THEN_ELSE src.
12473 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12475 PR tree-optimization/80158
12476 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12477 replacing a candidate statement, also replace it for the
12478 candidate's alternate interpretation.
12479 (replace_rhs_if_not_dup): Likewise.
12480 (replace_one_candidate): Likewise.
12482 2017-03-24 Richard Biener <rguenther@suse.de>
12484 PR tree-optimization/80167
12485 * graphite-isl-ast-to-gimple.c
12486 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12488 (translate_isl_ast_to_gimple::get_rename): Likewise.
12490 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12492 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12493 handling of certain combinations of target options, including the
12494 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12495 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12497 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12500 * config/arm/arm.md (*load_multiple): Add reload_completed to
12501 matching condition.
12503 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12504 Richard Biener <rguenth@suse.de>
12506 PR tree-optimization/79908
12507 PR tree-optimization/80136
12508 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12509 been cast away, gimplify_and_add suffices.
12511 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12513 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12515 2017-03-23 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/80032
12518 * gimplify.c (gimple_push_cleanup): Forced unconditional
12519 cleanups still have to go to the conditional_cleanups
12522 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12524 PR tree-optimization/80072
12525 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12527 (next_operand_entry_id): Change type to unsigned int.
12528 (sort_by_operand_rank): Make sure to return the right return value
12529 even if unsigned fields are bigger than INT_MAX.
12530 (struct oecount): Change cnt and id type to unsigned int.
12531 (oecount_hasher::equal): Formatting fix.
12532 (oecount_cmp): Make sure to return the right return value
12533 even if unsigned fields are bigger than INT_MAX.
12534 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12537 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12538 TREE_READONLY on result if writing it more than once.
12541 * doc/invoke.texi (-fsanitize=thread): Document that with
12542 -fnon-call-exceptions atomics are not able to throw
12546 * tsan.c: Include tree-eh.h.
12547 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12548 maybe_clean_or_replace_eh_stmt where needed.
12549 (instrument_memory_accesses): Add cfg_changed argument.
12550 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12551 if it returned true.
12552 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12554 PR rtl-optimization/63191
12555 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12556 wrapper function, moved the whole old content into ...
12557 (ix86_delegitimize_address_1): ... this. New inline function.
12558 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12559 true as last argument instead of ix86_delegitimize_address.
12561 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12563 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12564 cortexa57_branch_cost.
12566 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12568 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12570 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12573 * doc/md.texi (Constraints): Document wA constraint.
12574 * config/rs6000/constraints.md (wA): New.
12575 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12576 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12577 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12578 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12580 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12583 * gimplify.c (is_oacc_declared): New function.
12584 (oacc_default_clause): Use it to set default flags for acc declared
12585 variables inside parallel regions.
12586 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12587 declared variables.
12588 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12589 declare attribute to any decl as necessary.
12591 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12594 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12595 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12596 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12597 (arm_arch_lpae): This.
12598 * config/arm/arm.c (arm_arch7ve): Rename into ...
12599 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12600 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12603 2017-03-22 Martin Liska <mliska@suse.cz>
12606 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12607 error message instead of an ICE.
12609 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12611 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12613 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12615 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12617 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12620 2017-03-21 Martin Sebor <msebor@redhat.com>
12622 * doc/extend.texi: Use "cannot" instead of "can't."
12623 * doc/hostconfig.texi: Same.
12624 * doc/install.texi: Same.
12625 * doc/invoke.texi: Same.
12626 * doc/loop.texi: Same.
12627 * doc/md.texi: Same.
12628 * doc/objc.texi: Same.
12629 * doc/rtl.texi: Same.
12630 * doc/tm.texi: Same.
12631 * doc/tm.texi.in: Same.
12632 * doc/trouble.texi: Same.
12634 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12637 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12638 (collect_checksum_attributes): Set it.
12639 (die_checksum_ordered): Use it.
12641 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12643 PR tree-optimization/79908
12644 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12645 change: For a VA_ARG whose LHS has been cast away, use
12646 force_gimple_operand to construct the side effects.
12648 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12650 PR translation/80001
12651 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12652 more amenable to translation.
12653 (oacc_loop_auto_partitions): Likewise.
12655 2017-03-21 Marek Polacek <polacek@redhat.com>
12656 Martin Sebor <msebor@redhat.com>
12658 PR tree-optimization/80109
12659 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12660 on INTEGRAL_TYPE_P.
12662 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12663 Segher Boessenkool <segher@kernel.crashing.org>
12666 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12667 check reg_used_between_p between insn and one of succ or succ2
12668 depending on if succ is artificial insn not inserted into insn
12671 2017-03-21 Martin Liska <mliska@suse.cz>
12673 PR gcov-profile/80081
12674 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12675 * doc/gcc.texi: Include gcov-dump stuff.
12676 * doc/gcov-dump.texi: New file.
12678 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12680 PR rtl-optimization/79150
12681 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12682 conditional jump, if the jump is the last insn of the loop.
12684 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12685 Richard Biener <rguenth@suse.de>
12687 PR tree-optimization/79908
12688 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12689 been cast away, use force_gimple_operand to construct the side
12692 2017-03-21 Martin Liska <mliska@suse.cz>
12694 PR libfortran/79956
12695 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12698 2017-03-21 Brad Spengler <spender@grsecurity.net>
12701 * plugin.c (htab_hash_plugin): New function.
12702 (add_new_plugin): Use it and adjust.
12703 (parse_plugin_arg_opt): Adjust.
12704 (init_one_plugin): Likewise.
12706 2017-03-21 Richard Biener <rguenther@suse.de>
12708 PR tree-optimization/80032
12709 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12710 if set force the cleanup to happen unconditionally.
12711 (gimplify_target_expr): Push inserted clobbers with force_uncond
12712 to avoid them being removed by control-dependent DCE.
12714 2017-03-21 Richard Biener <rguenther@suse.de>
12716 PR tree-optimization/80122
12717 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12718 va_arg_pack_len when the inlined call stmt requires pack
12720 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12722 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12725 * tsan.c (instrument_builtin_call): If the memory model argument
12726 is not a constant, assume it is valid.
12729 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12732 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12734 PR rtl-optimization/79910
12735 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12736 if its dest is used by an insn before I2 (other than the combined
12737 insns themselves, which are properly handled already).
12739 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12742 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12744 * combine.c (record_used_regs): New static function.
12745 (try_combine): Handle situations where there is an additional
12746 instruction between I2 and I3 which needs to have a LOG_LINK
12750 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12752 * combine.c (try_combine): Delete redundant i1 test. Call
12753 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12755 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12758 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12759 alternatives 13/14.
12761 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12763 PR tree-optimization/80054
12764 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12765 the optimization if a PHI or any of its arguments is not dominated
12766 by the candidate's basis. Use gphi* rather than gimple* as
12768 (replace_profitable_candidates): Clean up a gimple* variable that
12769 should be a gphi* variable.
12771 2017-03-20 Martin Sebor <msebor@redhat.com>
12774 * doc/extend.texi (attribute constructor): Document present limitation.
12776 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12779 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12780 __POWER9_VECTOR__ #ifdef control, change template definition to
12781 use Power9-specific built-in function.
12782 (vec_any_eq): Likewise.
12783 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12784 to control outcomes from this test.
12785 (vector_ae_<mode>p): For VEC_F modes, likewise.
12787 2017-03-20 Ian Lance Taylor <iant@google.com>
12789 * config/i386/i386.c (ix86_function_regparm): Save an extra
12790 register for -fsplit-stack with DECL_STATIC_CHAIN.
12792 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12795 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12796 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12798 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12800 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12802 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12805 2017-03-20 Marek Polacek <polacek@redhat.com>
12808 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12810 2017-03-20 Richard Biener <rguenther@suse.de>
12812 PR tree-optimization/80113
12813 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12814 allocate extra SSA name for PHI def.
12815 (add_close_phis_to_outer_loops): Likewise.
12816 (add_close_phis_to_merge_points): Likewise.
12817 (copy_loop_close_phi_args): Likewise.
12818 (copy_cond_phi_nodes): Likewise.
12820 2017-03-20 Martin Liska <mliska@suse.cz>
12822 PR middle-end/79753
12823 * tree-chkp.c (chkp_build_returned_bound): Do not build
12824 returned bounds for a LHS that's not a BOUNDED_P type.
12826 2017-03-20 Martin Liska <mliska@suse.cz>
12830 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12831 COMPLEX_CST and VECTOR_CST.
12833 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12836 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12837 target operand. A new splitter adds the clobber statement in case
12838 the target operand is dead anyway.
12840 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12842 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12843 to age-old versions of binutils and glibc.
12845 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12847 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12849 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12851 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12853 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12855 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12856 requirement for binutils 2.13.
12858 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12860 * combine.c (try_combine): Delete redundant i1 test. Call
12861 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12863 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12865 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12866 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12868 <riscv64-*-elf>: Re-arrange section
12869 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12870 <riscv32-*-linux>: Likewise.
12871 <riscv64-*-elf>: Likewise
12872 <riscv64-*-linux>: Likewise.
12874 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
12877 * aarch64.opt(verbose-cost-dump): Fix typo.
12879 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
12882 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
12883 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
12885 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12887 * reload.c (find_reloads): When reloading a nonoffsettable address,
12888 use RELOAD_OTHER for it and its address reloads.
12890 PR rtl-optimization/79910
12891 * combine.c (record_used_regs): New static function.
12892 (try_combine): Handle situations where there is an additional
12893 instruction between I2 and I3 which needs to have a LOG_LINK
12896 2017-03-17 Jeff Law <law@redhat.com>
12898 PR tree-optimization/71437
12899 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
12900 conditional in the hash table first.
12901 (vrp_dom_walker::before_dom_children): Extract condition from
12902 ASSERT_EXPR. Record condition, its inverion and any implied
12903 conditions as well.
12905 2017-03-17 Marek Polacek <polacek@redhat.com>
12906 Markus Trippelsdorf <markus@trippelsdorf.de>
12908 PR tree-optimization/80079
12909 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
12912 2017-03-17 Richard Biener <rguenther@suse.de>
12914 PR middle-end/80075
12915 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
12916 Properly verify the LHS before the RHS possibly claims to be
12918 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
12921 2017-03-17 Martin Jambor <mjambor@suse.cz>
12923 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
12924 (List of -O2 options): Likewise.
12925 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
12928 2017-03-17 Tom de Vries <tom@codesourcery.com>
12930 * gcov-dump.c (print_usage): Print bug_report_url.
12932 2017-03-17 Richard Biener <rguenther@suse.de>
12934 PR middle-end/80050
12935 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
12936 (parser::peek): Likewise.
12938 2017-03-17 Richard Biener <rguenther@suse.de>
12940 PR tree-optimization/80048
12941 * sese.c (free_sese_info): Properly release rename_map and
12942 copied_bb_map elements.
12944 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
12946 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
12947 Add linked-list forward and backlinks. Insert on
12948 construction, remove on destruction.
12949 (class pass_store_merging): Add m_stores_head field.
12950 (pass_store_merging::terminate_and_process_all_chains):
12951 Iterate over m_stores_head list.
12952 (pass_store_merging::terminate_all_aliasing_chains):
12954 (pass_store_merging::execute): Check for debug stmts first.
12955 Push new chains onto the m_stores_head stack.
12957 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
12960 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
12961 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
12962 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
12964 2017-03-16 Jeff Law <law@redhat.com>
12966 PR tree-optimization/71437
12967 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
12968 member function. Implementation moved into after_dom_children
12969 member function and into the threader's thread_outgoing_edges
12971 (dom_opt_dom_walker::after_dom_children): Simplify by moving
12972 some code into new thread_outgoing_edges.
12973 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
12974 definition. Simplify marker handling (do it here). Assume we always
12975 have the available expression and the const/copies tables.
12976 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
12978 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
12979 * tree-vrp.c (equiv_stack): No longer file scoped.
12980 (vrp_dom_walker): New class.
12981 (vrp_dom_walker::before_dom_children): New member function.
12982 (vrp_dom_walker::after_dom_children): Likewise.
12983 (identify_jump_threads): Setup domwalker. Use it rather than
12984 walking edges in a random order by hand. Simplify setup/finalization.
12985 (finalize_jump_threads): Remove.
12986 (vrp_finalize): Do not call identify_jump_threads here.
12987 (execute_vrp): Do it here instead and call thread_through_all_blocks
12990 PR tree-optimization/71437
12991 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
12993 (simplify_stmt_for_jump_threading): Add basic_block argument. All
12995 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
12996 (dom_opt_dom_walker::thread_across_edge): Remove
12997 handle_dominating_asserts argument. All callers changed.
12998 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
12999 changes. Remove calls to lhs_of_dominating_assert. Other
13000 uses of handle_dominating_asserts turn into unconditional code
13001 (simplify_control_stmt_condition_1): Likewise.
13002 (simplify_control_stmt_condition): Likewise.
13003 (thread_through_normal_block, thread_across_edge): Likewise.
13004 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13005 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13006 object if it is not an SSA_NAME.
13007 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13008 before calling into the VRP specific simplifiers.
13009 (identify_jump_threads): Remove handle_dominating_asserts
13012 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13015 * tree-diagnostic.c (default_tree_printer): No longer static.
13016 * tree-diagnostic.h (default_tree_printer): New prototype.
13018 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13020 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13021 Change ins into fmov.
13023 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13025 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13026 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13027 Use h_con constraint for operand 1.
13028 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13029 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13031 2017-03-15 Jeff Law <law@redhat.com>
13033 PR tree-optimization/71437
13034 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13035 (record_temporary_equivalences): Use it.
13037 PR tree-optimization/71437
13038 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13039 tree-ssa-scopedtables.
13040 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13041 (record_conditions, record_cond, vuse_eq): Likewise.
13042 (record_edge_info): Adjust to API tweak of record_conditions.
13043 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13044 (record_temporary_equivalences, optimize_stmt): Likewise.
13045 (eliminate_redundant_computations): Likewise.
13046 (record_equivalences_from_stmt): Likewise.
13047 * tree-ssa-scopedtables.c: Include options.h and params.h.
13048 (vuse_eq): New function, moved from tree-ssa-dom.c
13049 (build_and_record_new_cond): Likewise.
13050 (record_conditions): Likewise. Accept vector of conditions rather
13051 than edge_equivalence structure for first argument.
13052 for the first argument.
13053 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13054 from tree-ssa-dom.c.
13055 (avail_exprs_stack::record_cond): Likewise.
13056 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13057 from tree-ssa-dom.c.
13058 (avail_exprs_stack): Add new member functions lookup_avail_expr
13060 (record_conditions): Declare.
13062 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13065 * lra-constraints.c (process_alt_operands): Increase reject for
13066 reloading an input/output operand.
13068 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13071 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13072 insns to convert from signed/unsigned char/short to IEEE 128-bit
13074 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13076 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13079 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13080 subreg of inner mode for values already in registers.
13082 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13084 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13085 iteration reg is used after the loop.
13087 2017-03-14 Martin Sebor <msebor@redhat.com>
13089 PR tree-optimization/79800
13090 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13091 precision in negative-positive range.
13092 (format_floating): Call non-const overload with adjusted precision.
13094 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13097 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13100 2017-03-14 Martin Sebor <msebor@redhat.com>
13102 PR middle-end/80020
13103 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13104 * builtins.def (aligned_alloc): Use it.
13107 * Makefile.in (GTFILES): Add calls.c.
13108 * calls.c: Include "gt-calls.h".
13110 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13112 PR rtl-optimization/79728
13113 * regs.h (struct target_regs): New field
13114 x_contains_allocatable_regs_of_mode.
13115 (contains_allocatable_regs_of_mode): New macro.
13116 * reginfo.c (init_reg_sets_1): Initialize it, and change
13117 contains_reg_of_mode so it includes global regs as well.
13118 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13119 rather than contains_regs_of_mode.
13121 2017-03-14 Martin Liska <mliska@suse.cz>
13123 * doc/invoke.texi: Document options that can't be combined with
13124 -fcheck-pointer-bounds.
13126 2017-03-14 Martin Liska <mliska@suse.cz>
13128 PR middle-end/79831
13129 * doc/invoke.texi (-Wchkp): Document the option.
13131 2017-03-14 Martin Liska <mliska@suse.cz>
13133 * Makefile.in: Install gcov-dump.
13135 2017-03-14 Martin Liska <mliska@suse.cz>
13137 * multiple_target.c (expand_target_clones): Bail out for
13138 an invalid attribute.
13140 2017-03-14 Richard Biener <rguenther@suse.de>
13142 * alias.c (struct alias_set_entry): Pack properly.
13143 * cfgloop.h (struct loop): Likewise.
13144 * cse.c (struct set): Likewise.
13145 * ipa-utils.c (struct searchc_env): Likewise.
13146 * loop-invariant.c (struct invariant): Likewise.
13147 * lra-remat.c (struct cand): Likewise.
13148 * recog.c (struct change_t): Likewise.
13149 * rtl.h (struct address_info): Likewise.
13150 * symbol-summary.h (function_summary): Likewise.
13151 * tree-loop-distribution.c (struct partition): Likewise.
13152 * tree-object-size.c (struct object_size_info): Likewise.
13153 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13154 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13155 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13156 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13157 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13158 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13159 (struct _stmt_vec_info): Likewise.
13161 2017-03-14 Martin Liska <mliska@suse.cz>
13164 * multiple_target.c (create_dispatcher_calls): Check that
13165 a target can create a function dispatcher.
13167 2017-03-14 Martin Liska <mliska@suse.cz>
13170 * multiple_target.c (expand_target_clones): Drop local.local
13171 flag for default implementation.
13173 2017-03-14 Richard Biener <rguenther@suse.de>
13175 PR tree-optimization/80030
13176 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13178 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13180 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13181 gcc_fallthrough() instead of __attribute__((fallthrough));
13183 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13185 * doc/gcc.texi: Remove "up" link to (DIR).
13186 * doc/gccint.texi: Ditto.
13188 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13190 * doc/install.texi (Specific) <avr>: Remove reference to
13193 2017-03-13 Jeff Law <law@redhat.com>
13195 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13196 attribute rather than comments.
13198 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13199 match_scratch operand is highest.
13201 2017-03-13 Martin Liska <mliska@suse.cz>
13203 PR middle-end/78339
13204 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13205 is a CHKP clone, use original declaration.
13207 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13209 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13210 (arc_conditional_register_usage): Use a different allocation order
13211 when optimizing for size.
13212 * common/config/arc/arc-common.c (arc_option_optimization_table):
13213 Section anchors default on when optimizing for size.
13215 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13217 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13219 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13221 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13222 * config/arc/arc.md (cpu_facility): Add cd variant.
13223 (*movqi_insn): Add code density variant.
13224 (*movhi_insn): Likewise.
13225 (*movqi_insn): Likewise.
13226 (*addsi3_mixed): Likewise.
13227 (subsi3_insn): Likewise.
13229 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13231 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13233 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13235 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13236 expressions with MINUS and UNARY ops.
13238 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13241 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13243 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13244 between vec_select and vector argument.
13245 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13246 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13247 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13248 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13249 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13250 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13252 2017-03-13 Richard Biener <rguenther@suse.de>
13255 * params.def (vect-max-peeling-for-alignment): Fix typo.
13257 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13259 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13260 issue that only occurred with binutils below 2.18.
13262 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13264 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13265 refer to binutils 2.11/2.12 minimum.
13267 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13269 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13270 ftp.kernel.org and simplify binutils requirement.
13272 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13274 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13276 (Optimize Options): Ditto. Also remove redundancy.
13278 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13280 PR translation/79848
13281 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13283 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13284 to G_ to avoid double translation.
13286 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13288 PR translation/79923
13289 * auto-profile.c (get_combined_location): Convert leading
13290 character of diagnostics to lower case and remove trailing period.
13291 (read_profile): Likewise for various diagnostics.
13292 * config/arm/arm.c (arm_option_override): Remove trailing period
13293 from various diagnostics.
13294 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13295 (msp430_expand_delay_cycles): Likewise.
13297 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13300 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13301 full command-line argument, rather than just "str".
13302 (aarch64_validate_march): Likewise.
13303 (aarch64_validate_mtune): Likewise.
13305 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13307 PR rtl-optimization/78911
13308 * lra-assigns.c (must_not_spill_p): New function.
13309 (spill_for): Use it.
13311 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13313 PR tree-optimization/79981
13314 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13315 ATOMIC_COMPARE_EXCHANGE ifn result.
13316 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13317 IFN_ATOMIC_COMPARE_EXCHANGE.
13319 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13322 * opts.c (parse_sanitizer_options): Add missing question mark to
13323 "did you mean" message.
13325 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13327 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13329 (VMULEUH_UNS): Likewise.
13330 (VMULOUB_UNS): Likewise.
13331 (VMULOUH_UNS): Likewise.
13332 * config/rs6000/rs6000.c (builtin_function_type): Remove
13333 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13335 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13338 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13339 x with result of extra_parsing_for_operand_code_0.
13340 (function_reader::extra_parsing_for_operand_code_0): Convert
13341 return type from void to rtx, returning x. When reading
13342 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13343 larger size containing struct block_symbol.
13345 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13348 -mfloat128-hardware without -m64.
13350 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13353 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13354 entries to the case statement that marks unsigned arguments to
13355 overloaded functions.
13357 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13359 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13360 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13362 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13365 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13366 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13368 2017-03-10 Martin Liska <mliska@suse.cz>
13372 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13373 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13376 2017-03-10 Olivier Hainque <hainque@adacore.com>
13378 * tree-switch-conversion (array_value_type): Start by resetting
13379 candidate type to it's main variant.
13381 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13383 PR rtl-optimization/79909
13384 * combine.c (try_combine): Use simplify_replace_rtx on individual
13385 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13386 of the whole CALL_INSN_FUNCTION_USAGE.
13388 PR tree-optimization/79972
13389 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13390 get_range_info on SSA_NAMEs. Formatting fixes.
13392 2017-03-10 Richard Biener <rguenther@suse.de>
13393 Jakub Jelinek <jakub@redhat.com>
13395 PR tree-optimization/77975
13396 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13397 edge to be constant.
13398 (get_val_for): For constant x return it. Formatting fix.
13399 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13400 would use the same bases as the current one.
13402 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13404 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13405 instead of vec_select for V1TImode.
13406 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13408 (VSX_LE_128): Add V1TI to this mode iterator.
13409 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13410 (*vsx_le_perm_store_<mode>): Likewise.
13411 (pre-reload splitter for VSX stores): Likewise.
13412 (post-reload splitter for VSX stores): Likewise.
13413 (*vsx_xxpermdi2_le_<mode>): Likewise.
13414 (*vsx_lxvd2x2_le_<mode>): Likewise.
13415 (*vsx_stxvd2x2_le_<mode>): Likewise.
13417 2017-03-09 Michael Eager <eager@eagercon.com>
13419 Correct failures with --enable-checking=yes,rtl.
13421 * config/microblaze/microblaze.c (microblaze_expand_shift):
13422 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13423 test for const0_rtx.
13424 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13425 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13427 2017-03-09 Richard Biener <rguenther@suse.de>
13429 PR tree-optimization/79977
13430 * graphite-scop-detection.c (scop_detection::merge_sese):
13431 Handle the case of extra exits to blocks dominating the entry.
13433 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13435 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13438 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13440 PR rtl-optimization/79949
13441 * lra-constraints.c (process_alt_operands): Check memory when
13442 trying to predict a cycle. Print about the overall increase.
13444 2017-03-09 Richard Biener <rguenther@suse.de>
13446 PR middle-end/79971
13447 * gimple-expr.c (useless_type_conversion_p): Preserve
13448 TYPE_SATURATING for fixed-point types.
13450 2017-03-09 Richard Biener <rguenther@suse.de>
13453 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13454 alignment of BLKmode params.
13456 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13459 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13460 (VALL_NO_V2Q): Likewise.
13462 * config/aarch64/aarch64-simd.md
13463 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13465 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13466 VALL_NO_V2Q mode iterator.
13467 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13469 2017-03-09 Martin Liska <mliska@suse.cz>
13471 PR tree-optimization/79631
13472 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13473 tree_int_cst_sign_bit just for INTEGER constants.
13475 2017-03-09 Martin Liska <mliska@suse.cz>
13479 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13482 2017-03-09 Marek Polacek <polacek@redhat.com>
13485 * tree.c (inchash::add_expr): Handle TREE_VEC.
13487 2017-03-09 Martin Liska <mliska@suse.cz>
13490 (chkp_narrow_size_and_offset): New function.
13491 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13492 (void chkp_parse_bit_field_ref): New function.
13493 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13494 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13496 2017-03-09 Martin Liska <mliska@suse.cz>
13499 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13500 (chkp_find_bounds_1): Remove gcc_unreachable.
13502 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13505 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13506 BUILT_IN_SYNC*, determine the access type from the size suffix and
13507 always build a MEM_REF with that type. Handle forgotten
13508 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13511 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13512 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13513 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13514 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13515 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13516 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13517 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13518 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13519 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13520 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13521 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13522 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13523 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13524 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13525 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13526 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13527 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13528 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13529 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13530 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13531 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13532 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13533 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13534 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13535 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13536 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13537 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13538 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13539 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13540 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13541 definitions outside of __OPTIMIZE__ guarded section.
13544 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13545 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13546 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13547 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13550 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13552 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13553 ("vfenez<mode>"): Add missing constraints.
13555 2017-03-08 Martin Sebor <msebor@redhat.com>
13558 * config/nds32/nds32.c (nds32_option_override):
13559 Fix misspelled diagnostic.
13561 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13564 * gimplify.c (gimplify_omp_for): Replace index var in outer
13565 taskloop statement with an artificial variable and add
13566 OMP_CLAUSE_PRIVATE clause for it.
13568 2017-03-08 Richard Biener <rguenther@suse.de>
13570 PR tree-optimization/79955
13571 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13572 for accesses that are completely outside of the variable.
13574 2017-03-08 Andrew Haley <aph@redhat.com>
13576 PR tree-optimization/79943
13577 * tree-ssa-loop-split.c (compute_new_first_bound): When
13578 calculating the new upper bound, (END-BEG) should be added, not
13581 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13583 * config/avr/avr.md (setmemhi): Make sure match_dup
13584 operand number comes before match_scratch.
13586 2017-03-08 Richard Biener <rguenther@suse.de>
13588 PR tree-optimization/79920
13589 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13590 with ncopies == 1 to ...
13591 (vect_transform_slp_perm_load): ... here. Properly compute
13592 all element loads by iterating VF times over the group. Do
13593 not handle ncopies (computed in a broken way) in
13594 vect_create_mask_and_perm.
13596 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13599 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13600 is a uniform vector, use uniform_vector_p return value instead of
13601 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13603 2017-03-07 Marek Polacek <polacek@redhat.com>
13605 PR middle-end/79809
13606 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13607 (alloca_call_type): Likewise.
13609 2017-03-07 Martin Liska <mliska@suse.cz>
13611 * gcov.c (process_args): Put comment to correct location.
13613 2017-03-07 Martin Liska <mliska@suse.cz>
13615 PR middle-end/68270
13616 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13617 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13618 (chkp_narrow_bounds_for_field): Likewise.
13619 (chkp_parse_array_and_component_ref): Pass one more argument to
13622 2017-03-07 Richard Biener <rguenther@suse.de>
13624 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13627 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13629 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13630 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13632 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13635 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13636 to end of description.
13637 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13639 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13641 PR rtl-optimization/79901
13642 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13644 (*avx512f_<code><mode>3<mask_name>): ... this.
13645 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13646 iterator instead of VI8_AVX2_AVX512BW.
13648 PR rtl-optimization/79901
13649 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13650 min/max expander, expand it using expand_vec_cond_expr.
13653 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13656 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13659 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13660 to void * for PCH reasons.
13661 * dwarf2out.c (output_loc_operands, output_die): Cast
13662 v.val_vec.array to unsigned char *.
13664 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13667 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13670 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13672 PR rtl-optimization/79571
13673 * lra-constraints.c (process_alt_operands): Calculate static
13674 reject and subtract it from overall when only addresses will be
13677 2017-03-06 Julia Koval <julia.koval@intel.com>
13680 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13681 incoming stack boundary to 128 for 64-bit targets.
13683 2017-03-06 Richard Biener <rguenther@suse.de>
13685 PR tree-optimization/79894
13686 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13687 to NULL after folding it.
13689 2017-03-06 Richard Biener <rguenther@suse.de>
13691 PR tree-optimization/79824
13692 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13693 check disabling peeling for gaps.
13695 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13697 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13698 attributes): Document gettimeofday.
13700 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13702 * config/s390/s390.c (s390_option_override_internal): Set
13703 PARAM_MIN_VECT_LOOP_BOUND
13705 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13707 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13708 * config/s390/s390.md: Likewise.
13710 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13713 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13714 (<avx2_avx512>_perm<mode>): Rename to ...
13715 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13717 (<avx512>_perm<mode>_mask): Rename to ...
13718 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13720 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13721 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13722 instead of VI8F_256_512.
13723 (avx512f_perm<mode>): New define_expand.
13724 (avx512f_perm<mode>_mask): Likewise.
13725 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13726 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13728 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13730 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13731 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13733 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13735 2017-03-06 Martin Liska <mliska@suse.cz>
13738 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13739 when having a SSA NAME w/o VAR_DECL assigned to it.
13741 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13743 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13744 msa_dpsub_<su>_d): Fix MODE for vec_select.
13746 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13748 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13750 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13752 2017-03-06 Richard Biener <rguenther@suse.de>
13754 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13755 * plugin.c (register_plugin_info): Likewise.
13756 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13758 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13760 * config/i386/sse.md (sse_storehps, sse_storelps,
13761 avx_<castmode><avxsizesuffix>_<castmode>,
13762 avx512f_<castmode><avxsizesuffix>_<castmode>,
13763 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13764 in condition that at least one operand is not a MEM.
13766 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13768 PR middle-end/79805
13769 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13770 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13772 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13773 gimple_call_nothrow_p flag based on whether original builtin can throw.
13774 If it can, emit following stmts on the fallthrough edge.
13775 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13776 don't create new bb if inserting just debug stmts on the edge, try to
13777 insert them on the fallthru bb or just reset debug stmts.
13779 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13782 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13783 restore recog_data (including the operand rtxes inside it) around
13784 the call to get_insn_template.
13786 2017-03-03 Martin Sebor <msebor@redhat.com>
13788 PR tree-optimization/79699
13789 * context.c (context::~context): Free MPFR caches to avoid
13790 a memory leak on program exit.
13792 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13794 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13795 Use wide_int::ulow () instead of .elt (0).
13797 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13799 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13800 (*pushxf): Limit oF constraint to 32bit targets and add oC
13801 constraint for 64bit targets.
13802 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13803 (*pushdf): Change rmF constraint to rmC.
13805 2017-03-03 Martin Liska <mliska@suse.cz>
13807 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13808 Remove unused variable.
13810 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13813 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13814 is a memory operand, increase num_memory.
13815 (ix86_expand_args_builtin): Likewise.
13817 2017-03-03 Jan Hubicka <jh@suse.cz>
13820 * ipa-devirt.c (maybe_record_node): Properly handle
13821 __cxa_pure_virtual visibility.
13823 2017-03-03 Martin Liska <mliska@suse.cz>
13825 PR tree-optimization/79803
13826 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13828 (pass_loop_prefetch::execute): Disabled optimization if an
13829 assumption about L1 cache size is not met.
13831 2017-03-03 Martin Liska <mliska@suse.cz>
13833 PR rtl-optimization/79574
13834 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13835 (hash_scan_set): Likewise.
13836 (dump_hash_table): Likewise.
13837 (hoist_code): Likewise.
13839 2017-03-03 Richard Biener <rguenther@suse.de>
13841 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13842 in place of uhigh ().
13843 (fixed_convert_from_real): Likewise.
13845 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13848 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13850 2017-03-03 Richard Biener <rguenther@suse.de>
13852 PR middle-end/79818
13853 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13854 TYPE_OVERFLOW_UNDEFINED check.
13856 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13858 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13860 (vector_ae_<mode>_p): Likewise.
13861 (vector_nez_<mode>_p): Likewise.
13862 (vector_ne_v2di_p): Likewise.
13863 (vector_ae_v2di_p): Likewise.
13864 (vector_ne_<mode>_p): Likewise.
13865 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13867 (vsx_tsqrt<mode>2_fe): Likewise.
13869 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13872 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
13874 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13876 PR rtl-optimization/79780
13877 * cprop.c (one_cprop_pass): When second and further conditional trap
13878 in a single basic block is turned into an unconditional trap, turn it
13879 into a deleted note to avoid RTL verification failures.
13881 2017-03-02 Richard Biener <rguenther@suse.de>
13883 * fold-const.c (const_binop): Use ulow () instead of elt (0).
13885 2017-03-02 Richard Biener <rguenther@suse.de>
13887 PR tree-optimization/79345
13889 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
13890 param and abort the walk, returning -1 if it is hit.
13891 (walk_aliased_vdefs): Take a limit param and pass it on.
13892 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
13893 defaulting to 0 and return a signed int.
13894 * tree-ssa-uninit.c (struct check_defs_data): New struct.
13895 (check_defs): New helper.
13896 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
13897 about uninitialized memory.
13898 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
13899 bogus uninitialized warning.
13900 (fixed_convert_from_real): Likewise.
13902 2017-03-02 Bin Cheng <bin.cheng@arm.com>
13904 PR tree-optimization/66768
13905 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
13906 iv_use if base object can't be determined.
13908 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13910 PR tree-optimization/79345
13911 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
13912 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
13913 (get_pattern_stats): Initialize it.
13914 * genemit.c (gen_expand): Verify match_scratch numbers come after
13915 match_operand/match_dup numbers.
13916 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
13917 match_scratch numbers.
13918 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
13920 * config/s390/s390.md (trunctdsd2): Likewise.
13922 2017-03-02 Richard Biener <rguenther@suse.de>
13924 * wide-int.h (wide_int_storage::operator=): Implement in terms
13927 2017-03-02 Richard Biener <rguenther@suse.de>
13929 PR tree-optimization/79777
13930 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
13931 the to insert expression to sth existing.
13933 2017-03-01 Martin Sebor <msebor@redhat.com>
13935 PR middle-end/79692
13936 * gimple-ssa-sprintf.c
13937 (directive::known_width_and_precision): New function.
13938 (format_integer): Use it.
13939 (get_mpfr_format_length): Consider the full range of precision
13940 when computing %g output with the # flag. Set the likely byte
13941 count to 3 rather than 1 when precision is indeterminate.
13942 (format_floating): Correct the lower bound of precision.
13944 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13946 * doc/invoke.texi: Document default code model for 64-bit Linux.
13948 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13951 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
13952 udiv rather than div since input pattern is unsigned.
13954 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
13956 * config/i386/i386.c (print_reg): Warn for values of
13957 unsupported size in integer register.
13959 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
13962 * config/rs6000/predicates.md (current_file_function_operand): Do
13963 not allow self calls to be local if the function is replaceable.
13965 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
13968 * config/rs6000/altivec.h (vec_ctz and others): Change the
13969 preprocessor macro that controls conditional compilation from
13970 _ARCH_PWR9 to __POWER9_VECTOR__.
13971 (vec_all_ne): Change parameterization of __altivec_scalar_pred
13972 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
13973 control (instead of _ARCH_PWR9 control) so that template
13974 definition uses power9-specific function.
13975 (vec_any_eq): Likewise.
13976 (vec_all_ne): Change macro definition to use a power9-specific
13977 expansion under #ifdef __POWER9_VECTOR__ control (instead of
13978 _ARCH_PWR9 control).
13979 (vec_any_eq) Likewise.
13980 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
13981 expansion for CMPNEF to remove support for xvcmpnesp instruction.
13982 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
13983 support for xvcmpnedp instruction.
13984 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
13985 macro expansion so that Power9 implementation of vec_all_ne does
13986 not use the AltiVec predicate framework.
13987 (VCMPNEH_P): Likewise.
13988 (VCMPNEW_P): Likewise.
13989 (VCMPNED_P): Likewise.
13990 (VCMPNEFP_P): Likewise.
13991 (VCMPNEDP_P): Likewise.
13992 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
13993 implementation of vec_any_eq to not use AltiVec predicate
13995 (VCMPAEH_P): Likewise.
13996 (VCMPAEW_P): Likewise.
13997 (VCMPAED_P): Likewise.
13998 (VCMPAEFP_P): Likewise.
13999 (VCMPAEDP_P): Likewise.
14000 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14001 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14002 not use the AltiVec predicate framework.
14003 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14004 of vec_any_eq to not use AltiVec predicate framework.
14005 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14006 support for predefined __POWER9_VECTOR__ macro to indicate that
14007 Power9 instruction selection is enabled.
14008 (altivec_overloaded_builtins): Remove extraneous
14009 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14010 function argument types RS6000_BTI_bool_V16QI and
14011 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14012 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14013 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14014 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14015 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14016 Power9 for implementations of vec_cmpne. Change the signature for
14017 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14018 (representing vec_all_ne) to remove the previously described first
14019 argument of type RS6000_BTI_INTSI, as this was an artifact of
14020 reliance on the AltiVec predicate framework, which is no longer
14021 used in the implementation of these functions. Add
14022 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14023 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14024 since, unlike the AltiVec predicate framework implementation, we
14025 do not share function descriptors between vec_alle and vec_anyeq.
14026 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14027 set of modes that receive special treatment even when
14028 TARGET_P9_VECTOR is true. The special treatment emits code that
14029 does not depend on Power9 instructions.
14030 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14031 define_expand to not rely on AltiVec predicate framework.
14032 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14034 (vector_ne_v2di_p): Change this define_expand to not rely on
14035 AltiVec predicate framework.
14036 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14038 (vector_ne_<mode>_p): Change this define_expand to not rely on
14039 AltiVec predicate framework.
14040 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14042 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14043 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14044 define_insn pattern.
14045 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14046 define_insn pattern because the xvcmpne<VSs>. instruction is not
14048 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14049 instruction is not supported.
14051 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14053 * config/nvptx/nvptx.c: Include intl.h.
14055 2017-03-01 Martin Jambor <mjambor@suse.cz>
14058 * ipa-prop.h (ipa_bits): Removed field known.
14059 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14060 to pointers. Adjusted their comments to warn about their sharing.
14061 (ipcp_transformation_summary): Change bits to a vector of pointers.
14062 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14063 (ipa_get_ipa_bits_for_value): Declare.
14064 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14065 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14066 (ipa_bits_hash_table): Likewise.
14067 (ipa_vr_ggc_hash_traits): Likewise.
14068 (ipa_vr_hash_table): Likewise.
14069 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14070 being pointers and vr_known being removed.
14071 (ipa_set_jf_unknown): Likewise.
14072 (ipa_get_ipa_bits_for_value): New function.
14073 (ipa_set_jfunc_bits): Likewise.
14074 (ipa_get_value_range): New overloaded functions.
14075 (ipa_set_jfunc_vr): Likewise.
14076 (ipa_compute_jump_functions_for_edge): Use the above functions to
14077 construct bits and vr parts of jump functions.
14078 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14079 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14081 (ipcp_grow_transformations_if_necessary): Also allocate
14082 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14084 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14085 them. Fix too long lines.
14086 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14087 vr_known being removed.
14088 (ipa_read_jump_function): Use new setter functions to construct bits
14089 and vr parts of jump functions or set them to NULL.
14090 (write_ipcp_transformation_info): Adjust for bits being pointers.
14091 (read_ipcp_transformation_info): Likewise.
14092 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14094 Include gt-ipa-prop.h.
14095 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14097 (ipcp_store_bits_results): Likewise.
14098 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14099 Do not write to existing jump functions but use a temporary instead.
14101 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14104 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14105 attempt to use its first operand as BIT_FIELD_REF base.
14107 2017-03-01 Richard Biener <rguenther@suse.de>
14109 PR middle-end/79721
14110 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14111 interpolating formula in wrapping arithmetic.
14112 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14114 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14116 PR tree-optimization/79734
14117 * tree-vect-generic.c (expand_vector_condition): Optimize
14118 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14119 Handle VEC_COND_EXPR where comparison has different inner width from
14120 type's inner width.
14122 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14124 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14125 markup, and similar issues. Remove @opindex entries for things
14126 that aren't options. Add missing -mmpy-option entries.
14128 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14130 PR tree-optimization/79737
14131 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14132 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14133 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14134 instead of byte_size. Formatting fix.
14135 (shift_bytes_in_array_right): Formatting fix.
14137 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14140 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14141 condition on optimize for the leaf function test.
14143 2017-02-28 Martin Liska <mliska@suse.cz>
14146 * read-rtl-function.c (function_reader::handle_unknown_directive):
14147 Bail out when one uses -flto.
14149 2017-02-28 Martin Liska <mliska@suse.cz>
14151 * common.opt: Replace space with tabular for options of <number>
14153 * config/i386/i386.opt: Show <number> value for
14154 -mlarge-data-threshold.
14155 * opts.c (print_filtered_help): Do not display number in hexadecimal
14158 2017-02-28 Martin Liska <mliska@suse.cz>
14160 * common.opt: Fix --help=option -Q for options which are of
14163 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14165 * config/i386/i386.c (print_reg): Error out for values
14166 of 8-bit size in invalid integer register.
14168 2017-02-28 Martin Sebor <msebor@redhat.com>
14170 PR tree-optimization/79691
14171 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14173 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14176 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14177 gcc_unreachable with output_operand_lossage.
14179 2017-02-28 Richard Biener <rguenther@suse.de>
14181 PR tree-optimization/79740
14182 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14184 (visit_nary_op): Insert the nary into the hashtable if we
14185 pattern-matched sth.
14186 * tree-ssa-pre.c (eliminate_insert): Robustify.
14188 2017-02-28 Richard Biener <rguenther@suse.de>
14190 PR middle-end/79731
14191 * fold-const.c (decode_field_reference): Reject out-of-bound
14194 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14196 * config/i386/i386.c: Include intl.h.
14197 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14198 instead of just cond ? "..." : "...".
14199 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14200 * coverage.c (read_counts_file): Likewise.
14201 * omp-offload.c: Include intl.h.
14202 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14203 of just cond ? "..." : "...".
14204 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14205 of just cond ? "..." : "...".
14207 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14210 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14212 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14213 'tune for' CPU name.
14214 * config/arm/arm-cpu-data.h: Regenerated.
14216 2017-02-28 Richard Biener <rguenther@suse.de>
14218 PR tree-optimization/79732
14219 * tree-inline.c (expand_call_inline): Do not shadow var.
14221 2017-02-28 Richard Biener <rguenther@suse.de>
14223 PR tree-optimization/79723
14224 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14225 address-space properly.
14227 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14229 * doc/optinfo.texi (Optimization groups): Fix option used for
14231 * doc/invoke.texi (-fopt-info): Document "omp".
14232 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14233 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14234 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14235 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14236 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14238 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14240 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14241 OPTGROUP_OMP with "omp".
14243 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14246 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14247 for arithmetic shift of unsigned V2DI.
14249 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14251 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14252 arc/linux.h headers.
14253 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14254 (LINK_SPEC): Likewise.
14255 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14256 (EXTRA_SPECS): Likewise.
14257 (STARTFILE_SPEC): Likewise.
14258 (ENDFILE_SPEC): Likewise.
14259 (LIB_SPEC): Likewise.
14260 (TARGET_SDATA_DEFAULT): Likewise.
14261 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14262 (MULTILIB_DEFAULTS): Likewise.
14263 (DWARF2_UNWIND_INFO): Likewise.
14264 * config/arc/big.h: New file.
14265 * config/arc/elf.h: Likewise.
14266 * config/arc/linux.h: Likewise.
14267 * config/arc/t-uClibc: Remove.
14269 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14271 PR tree-optimization/77536
14272 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14273 (tree_transform_and_unroll_loop): Use above function to compute the
14274 estimated niter of unrolled loop and use it when scaling profile.
14275 Also use count info rather than frequency if it's non-zero.
14276 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14277 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14278 (vect_transform_loop): Call above function.
14280 2017-02-27 Richard Biener <rguenther@suse.de>
14282 PR tree-optimization/45397
14283 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14284 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14285 (visit_nary_op): Add pattern matching for CSEing sign-changed
14286 or truncated operations with wider ones.
14288 2017-02-27 Richard Biener <rguenther@suse.de>
14290 PR tree-optimization/79690
14291 * tree-vect-stmts.c (vectorizable_store): Use vector type
14292 built from the DR with address-space.
14294 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14296 * doc/invoke.texi (Optimize Options): Refine the description
14297 of asan-use-after-return.
14299 2017-02-25 Alan Modra <amodra@gmail.com>
14301 PR rtl-optimization/79584
14302 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14303 base, not ad->base_term, the reg within base. Remove assertion
14304 that ad->base == ad->base_term. Replace gen_int_mode using
14305 bogus mode with const0_rtx.
14307 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14309 PR middle-end/79396
14310 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14311 FMA_EXPR like tcc_binary or tcc_unary.
14313 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14316 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14318 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14319 (output_loc_operands): Handle DW_OP_call_ref and
14320 DW_OP_GNU_variable_value.
14321 (struct variable_value_struct): New type.
14322 (struct variable_value_hasher): Likewise.
14323 (variable_value_hash): New variable.
14324 (string_types): Remove.
14325 (copy_loc_descr): New function.
14326 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14327 (prepend_loc_descr_to_each): New function.
14328 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14329 instead of add_loc_descr_to_each if the first argument is single
14330 location list and the second has multiple.
14331 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14332 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14333 when looking for variable value which doesn't have other location info.
14334 (loc_list_from_tree): Formatting fix.
14335 (gen_array_type_die): Simplify DW_AT_string_length handling.
14336 (adjust_string_types): Remove.
14337 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14338 string_types. Call resolve_variable_values.
14339 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14340 (resolve_addr_in_expr): Likewise. Add A argument.
14341 (copy_deref_exprloc): Remove deref argument. Adjust for the
14342 original expression being DW_OP_GNU_variable_value with optionally
14343 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14344 optionally after it.
14345 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14346 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14347 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14348 (variable_value_hasher::hash, variable_value_hasher::equal): New
14350 (resolve_variable_value_in_expr, resolve_variable_value,
14351 resolve_variable_values, note_variable_value_in_expr,
14352 note_variable_value): New functions.
14353 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14356 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14359 * opts.h (handle_generated_option): Add GENERATED_P argument.
14360 * opts-common.c (handle_option): Adjust function comment.
14361 (handle_generated_option): Add GENERATED_P argument, pass it to
14363 (control_warning_option): Pass false to handle_generated_option
14365 * opts.c (maybe_default_option): Pass true to handle_generated_option
14367 * optc-gen.awk: Likewise.
14369 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14371 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14372 a REG, look at the REG it is a SUBREG of.
14373 (splitter for cmpeqsi_t): Ditto.
14375 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14377 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14378 the special USEs with the pattern of the insn, not the insn itself.
14380 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14383 * doc/invoke.texi: Document -mload-store-pairs.
14385 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14386 Sandra Loosemore <sandra@codesourcery.com>
14388 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14389 argument isn't a CONST_INT.
14390 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14391 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14392 (nios2_validate_compare): Bypass alternate compare logic if *op2
14393 is not a CONST_INT.
14394 (ldstwm_operation_p): Return false if first_base is not a REG or
14395 if first_offset is not a CONST_INT.
14397 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14399 * config/cris/cris.md: Use correct operand in a define_peephole2.
14401 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14403 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14405 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14407 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14408 this_insn if it is an INSN or JUMP_INSN.
14409 (force_offsettable): Look at base, not at addr.
14410 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14411 on things that aren't necessarily CONST_INTs.
14413 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14415 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14416 -mfpmath=sse is the default also for x86-32 targets with SSE2
14417 instruction set when @option{-ffast-math} is enabled
14419 2017-02-24 Jeff Law <law@redhat.com>
14421 PR rtl-optimizatoin/79286
14422 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14425 2017-02-24 Richard Biener <rguenther@suse.de>
14427 PR tree-optimization/79389
14428 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14431 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14433 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14434 function comment to reflect reality.
14435 (loop_exits_before_overflow): Fix typo in function description.
14437 2017-02-24 Richard Biener <rguenther@suse.de>
14439 PR tree-optimization/79389
14440 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14441 properly that a threading opportunity exists. Detect conditional
14442 copy/constant propagation opportunities.
14444 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14446 * config/visium/visium.md (type): Add trap.
14447 (b): New mode attribute.
14448 (*btst): Rename into...
14449 (*btst<mode>): ...this and adjust.
14450 (*cbranchsi4_btst_insn): Rename into...
14451 (*cbranch<mode>4_btst_insn): ...this and adjust.
14452 (trap): New define_insn.
14454 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14456 PR tree-optimization/79389
14457 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14458 (noce_reversed_cond_code): New function.
14459 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14460 reversed_comparison_code. Formatting fix.
14461 (noce_try_store_flag): Test rev_cond != NULL in addition to
14462 reversed_comparison_code.
14463 (noce_try_store_flag_constants): Likewise.
14464 (noce_try_store_flag_mask): Likewise.
14465 (noce_try_addcc): Use rev_cond if non-NULL instead of
14466 reversed_comparison_code.
14467 (noce_try_cmove_arith): Likewise. Formatting fixes.
14468 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14469 (noce_find_if_block): Initialize rev_cond.
14470 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14471 instead of false as last argument never attempt to reverse it
14474 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14476 PR tree-optimization/79663
14477 * tree-predcom.c (combine_chains): Process refs in reverse order
14478 only for ZERO length chains, and add explaining comment.
14480 2017-02-23 Jeff Law <law@redhat.com>
14482 PR tree-optimization/79578
14483 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14484 in call to operand_equal_p.
14486 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14489 * config/i386/cpuid.h: Fix another undefined behavior.
14491 2017-02-23 Richard Biener <rguenther@suse.de>
14493 PR tree-optimization/79683
14494 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14495 vector types for data-refs.
14497 2017-02-23 Martin Liska <mliska@suse.cz>
14499 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14501 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14503 PR middle-end/79665
14504 * internal-fn.c (get_range_pos_neg): Moved to ...
14505 * tree.c (get_range_pos_neg): ... here. No longer static.
14506 * tree.h (get_range_pos_neg): New prototype.
14507 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14508 are known to be in between 0 and signed maximum inclusive, try to
14509 expand both unsigned and signed divmod and use the cheaper one from
14512 2017-02-22 Jeff Law <law@redhat.com>
14514 PR tree-optimization/79578
14515 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14516 to compare base operands.
14518 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14521 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14522 gpc_reg_operand instead of fpr_reg_operand.
14524 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14526 * config/mips/mips.c (mips_return_in_memory): Force FP
14527 vector types to be returned in memory for o32 ABI.
14529 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14531 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14532 instead of DW_TAG_member for static data member declarations and don't
14533 set no_linkage_name for static inline data members.
14534 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14537 2017-02-22 Martin Liska <mliska@suse.cz>
14539 * doc/invoke.texi: Replace inequality signs with square brackets
14542 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14544 PR tree-optimization/68644
14545 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14547 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14550 * lra-constraints.c (simplify_operand_subreg): Handle
14551 WORD_REGISTER_OPERATIONS targets.
14553 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14556 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14557 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14558 elimination by swapping fld*.
14560 2017-02-22 Richard Biener <rguenther@suse.de>
14562 PR tree-optimization/79673
14563 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14564 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14565 irrelevant address-space qualifiers and avoiding a
14566 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14568 2017-02-22 Richard Biener <rguenther@suse.de>
14570 PR tree-optimization/79666
14571 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14572 to not symbolically negate if that may introduce undefined
14575 2017-02-22 Martin Liska <mliska@suse.cz>
14578 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14579 * data-streamer-out.c (streamer_write_gcov_count_stream):
14581 * value-prof.c (stream_out_histogram_value): Make assert more
14582 precise based on type of counter.
14584 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14587 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14588 Use nonimmediate_operand instead of memory_operand for operand 1.
14589 (float-extend standard_x87sse_constant_load splitter): Ditto.
14591 2017-02-21 Jeff Law <law@redhat.com>
14593 PR tree-optimization/79621
14594 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14595 blocks with edges to themselves.
14597 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14600 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14601 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14602 Use gimple_call_builtin_p.
14605 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14606 on temporarily removed DEBUG_INSNs.
14608 PR tree-optimization/79649
14609 * tree-loop-distribution.c (classify_partition): Give up on
14610 non-generic address space loads/stores.
14612 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14614 * doc/loop.texi (Loop manipulation): Remove nonexistent
14615 tree_ssa_loop_version from the documentation.
14616 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14618 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14621 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14622 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14623 * config/rs6000/rs6000.c: Include except.h.
14624 (rs6000_expand_split_stack_prologue): Call
14625 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14627 2017-02-21 Martin Jambor <mjambor@suse.cz>
14630 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14631 have been analyzed.
14633 2017-02-21 Martin Jambor <mjambor@suse.cz>
14635 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14636 for backward compatibility only.
14637 * doc/invoke.texi (Option Summary): Remove all references to
14638 -fipa-cp-alignment.
14640 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14644 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14646 * lra-constraints.c (curr_insn_transform): Handle
14647 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14649 2017-02-21 Martin Liska <mliska@suse.cz>
14651 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14653 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14655 PR translation/79638
14656 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14658 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14661 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14662 (arm_function_ok_for_sibcall): Return false for an indirect call by
14663 descriptor if all the argument registers are used.
14664 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14665 alignment of the function.
14667 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14669 PR tree-optimization/61441
14670 * simplify-rtx.c (simplify_const_unary_operation): For
14671 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14672 the sNaN unmodified.
14674 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14676 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14677 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14678 instead of SYSTEM_HEADER_DIR.
14680 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14681 Martin Liška <mliska@suse.cz>
14683 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14684 Fix typos and grammar, use active voice, and clarify.
14686 2017-02-20 Marek Polacek <polacek@redhat.com>
14688 PR middle-end/79537
14689 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14692 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14694 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14697 * config/i386/i386.c (ix86_expand_builtin): Handle
14698 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14699 ix86_builtins_isa[fcode].isa as a requirement of those
14700 flags and any other flag in the bitmask.
14701 (ix86_init_mmx_sse_builtins): Use 0 instead of
14702 ~OPTION_MASK_ISA_64BIT as mask.
14703 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14704 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14705 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14706 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14708 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14711 * lra-constraints.c (split_reg): Check requested split mode
14712 is supported by the register.
14714 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14716 * lra-constraints.c (simplify_operand_subreg): Remove early
14719 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14722 * lra-constraints.c (curr_insn_transform): Tighten condition
14723 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14725 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14728 * lra-constraints.c (curr_insn_transform): Handle
14729 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14731 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14734 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14736 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14738 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14741 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14744 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14746 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14749 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14751 * final.c (last_columnnum, override_columnnum): New variables.
14752 (final_start_function): Set last_columnnum, pass it to begin_prologue
14753 hook and pass 0 to dwarf2out_begin_prologue.
14754 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14755 to source_line debug hook.
14756 (notice_source_line): Compute last_columnnum and for debug_column_info
14757 return true on column changes.
14758 * debug.h (struct gcc_debug_hooks): Add column argument to
14759 source_line and begin_prologue hooks.
14760 (debug_nothing_int_charstar_int_bool): Remove prototype.
14761 (debug_nothing_int_int_charstar,
14762 debug_nothing_int_int_charstar_int_bool): New prototypes.
14763 (dwarf2out_begin_prologue): Add column argument.
14764 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14765 begin_prologue hooks.
14766 (debug_nothing_int_charstar_int_bool): Remove.
14767 (debug_nothing_int_int_charstar,
14768 debug_nothing_int_int_charstar_int_bool): New functions.
14769 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14770 through to dwarf2out_source_line.
14771 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14772 (dwarf2out_source_line): Add column argument, emit it if requested.
14773 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14775 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14776 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14777 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14778 through to dwarf2out_begin_prologue.
14779 (vmsdbgout_source_line): Add column argument, pass it through to
14780 dwarf2out_source_line.
14781 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14782 dbxout_source_line caller.
14783 (dbxout_source_line): Add column argument.
14785 * common.opt (gno-column-info, gcolumn-info): New options.
14786 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14787 (check_die): Also test for multiple DW_AT_decl_column attributes.
14788 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14789 DW_AT_decl_column if requested.
14790 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14792 (gen_variable_die): Likewise.
14793 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14794 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14797 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14798 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14799 (ix86_handle_option): Handle OPT_m3dnowa.
14800 * doc/invoke.texi (-m3dnowa): Document.
14801 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14802 -m3dnowa instead of -m3dnow -march=athlon.
14805 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14806 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14808 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14811 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14812 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14813 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14814 generator for vsx_xxpermdi_<mode>_be.
14815 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14816 force big-endian semantics.
14817 (vsx_xxpermdi_<mode>_be): New define_expand with same
14818 implementation as previous version of vsx_xxpermdi_<mode>.
14820 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14822 PR tree-optimization/79327
14823 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14824 variable, its initialization and use.
14826 2017-02-17 Julia Koval <julia.koval@intel.com>
14828 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14829 (OPTION_MASK_ISA_PKU_UNSET): New.
14830 (ix86_handle_option): Handle -mrdpid.
14831 * config/i386/cpuid.h (bit_RDPID): New.
14832 * config/i386/driver-i386.c (host_detect_local_cpu):
14833 Detect RDPID feature.
14834 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14835 * config/i386/i386-c.c (ix86_target_macros_internal):
14837 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14838 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14839 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14840 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14841 * config/i386/i386.md (define_insn "rdpid"): New.
14842 * config/i386/i386.opt Add -mrdpid.
14843 * config/i386/immintrin.h (_rdpid_u32): New.
14845 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14847 PR rtl-optimization/79541
14848 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14849 instead of transforming it into USE.
14851 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14853 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14854 If HONOR_SNANS (SFmode) force the input to a register.
14855 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14856 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14857 an frsp or similar insn.
14859 2017-02-17 Martin Liska <mliska@suse.cz>
14861 PR rtl-optimization/79577
14862 * params.def (selsched-max-sched-times): Increase minimum to 1.
14864 2017-02-17 Martin Liska <mliska@suse.cz>
14866 PR rtl-optimization/79574
14867 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14869 2017-02-17 Martin Liska <mliska@suse.cz>
14871 PR tree-optimization/79529
14872 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
14873 ssa_defined_default_def_p to handle cases which are implicitly
14875 * tree-ssa.c (ssa_defined_default_def_p): New function.
14876 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
14877 which are implicitly defined.
14878 * tree-ssa.h (ssa_defined_default_def_p): Declare.
14880 2017-02-17 Richard Biener <rguenther@suse.de>
14882 PR middle-end/79576
14883 * params.def (max-ssa-name-query-depth): Limit to 10.
14885 2017-02-17 Richard Biener <rguenther@suse.de>
14887 PR tree-optimization/79552
14888 * tree-ssa-structalias.c (visit_loadstore): Properly verify
14891 2017-02-17 Richard Biener <rguenther@suse.de>
14894 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
14896 2017-02-17 Marek Polacek <polacek@redhat.com>
14898 PR middle-end/79536
14899 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
14900 (fold_negate_expr): New wrapper.
14902 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
14904 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
14905 Correct terminology and de-emphasize pre-standard behavior.
14907 2017-02-16 Alan Modra <amodra@gmail.com>
14909 PR rtl-optimization/79286
14910 * ira.c (def_dominates_uses): New function.
14911 (update_equiv_regs): Don't create an equivalence for insns that
14912 may trap where the register def does not dominate the use.
14914 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
14916 PR rtl-optimization/78127
14917 * lra.c (lra): Call lra_eliminate before finish the loop after
14920 2017-02-16 Richard Biener <rguenther@suse.de>
14922 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
14924 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
14925 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
14926 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
14927 (isl_val_int_from_wi): New function.
14928 (extract_affine_gmp): Rename to ...
14929 (extract_affine_wi): ... this, take a widest_int.
14930 (extract_affine_int): Just wrap extract_affine_wi.
14931 (add_param_constraints): Use isl_val_int_from_wi.
14932 (add_loop_constraints): Likewise, and extract_affine_wi.
14934 2017-02-15 Jeff Law <law@redhat.com>
14936 PR middle-end/79521
14937 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
14938 ira_init_register_move_cost_if_necessary.
14940 2017-02-15 Martin Sebor <msebor@redhat.com>
14942 PR middle-end/32003
14943 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
14944 removed in a prior commit.
14946 2017-02-15 Bin Cheng <bin.cheng@arm.com>
14948 PR tree-optimization/79347
14949 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
14950 counters during peeling.
14952 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
14954 * Makefile.in (site.exp): Remove "set ISLVER".
14956 2017-02-15 Jakub Jelinek <jakub@redhat.com>
14959 * real.c (real_from_integer): Call real_convert even for decimal.
14961 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
14964 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
14966 2017-02-14 Andrew Pinski <apinski@cavium.com>
14968 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
14969 cores and change the partno/implementer to be correct.
14970 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
14971 the 'B" as the implementer.
14972 * config/aarch64/aarch64-tune.md: Regenerate.
14974 2017-02-14 Carl Love <cel@us.ibm.com>
14976 * config/rs6000/rs6000.c: Add case statement entry to make the
14977 xvcvuxdsp built-in argument unsigned.
14978 * config/rs6000/vsx.md: Fix the source and return operand types so they
14979 match the instruction definitions from the ISA document. Fix typo
14980 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
14983 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
14986 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
14987 member early_clobber_alts.
14988 * lra-lives.c (reg_early_clobber_p): New.
14989 (process_bb_lives): Use it.
14990 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
14991 (debug_operand_data): Initialize early_clobber_alts.
14992 (setup_operand_alternative): Set up early_clobber_alts.
14993 (collect_non_operand_hard_regs): Ditto. Pass early clobber
14994 alternatives to new_insn_reg.
14995 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
14997 (lra_update_insn_regno_info): Pass the new arg.
14999 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15001 PR middle-end/79505
15002 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15003 (new_oacc_loop_raw): Don't clear already cleared fields.
15006 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15007 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15008 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15010 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15013 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15015 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15018 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15019 the extra instruction to the right place to store 128-bit constant
15022 2017-02-14 Martin Sebor <msebor@redhat.com>
15024 PR middle-end/79448
15025 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15026 warning for strings of unknown length.
15028 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15030 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15032 2017-02-14 Jeff Law <law@redhat.com>
15035 * ira-costs.c (scan_one_insn): Initialize register move costs
15036 for pseudos seen in USE/CLOBBER insns.
15038 PR tree-optimization/79095
15039 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15040 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15041 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15042 if the operands are known to be not equal, then the resulting range
15044 (intersect_ranges): If the new range is ~[0,0] and the old range is
15045 wide, then prefer ~[0,0].
15046 * tree-vrp.c (overflow_comparison_p_1): New function.
15047 (overflow_comparison_p): New function.
15048 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15049 if NAME is used in an overflow test.
15050 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15051 overflow check that can be expressed as an equality test, then adjust
15052 ops to be that equality test.
15054 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15056 * config/s390/s390-builtin-types.def: Remove flags argument.
15057 * config/s390/s390.c (s390_init_builtins): Likewise.
15059 2017-02-14 Martin Liska <mliska@suse.cz>
15061 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15062 vector. Fix trailing white spaces.
15064 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15066 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15069 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15071 PR rtl-optimization/68664
15072 * config/arm/arm.c (arm_sched_can_speculate_insn):
15073 New function. Declare prototype.
15074 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15076 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15078 PR rtl-optimization/68664
15079 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15081 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15083 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15085 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15086 max skip bytes for function, loop and jump.
15088 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15090 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15091 ABS_EXPR for gimple dump.
15093 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15096 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15098 PR tree-optimization/79408
15099 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15100 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15101 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15102 also if rhs1 is INTEGER_CST.
15104 2017-02-14 Richard Biener <rguenther@suse.de>
15106 PR middle-end/79432
15107 * tree-into-ssa.c (insert_phi_nodes): When the function can
15108 have abnormal edges rewrite SSA names with broken use-def
15109 dominance out of SSA and register them for PHI insertion.
15111 2017-02-13 Martin Sebor <msebor@redhat.com>
15113 PR middle-end/79496
15114 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15115 clearing info.nowrite flag when snprintf size argument is a range.
15117 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15119 * cprop.c (cprop_jump): Add missing space in string literal.
15120 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15121 (get_constraint_for_component_ref): Likewise.
15122 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15123 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15124 * lra-constraints.c (process_alt_operands): Likewise.
15125 * ipa-inline.c (inline_small_functions): Likewise.
15126 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15127 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15128 * trans-mem.c (diagnose_tm_1_op): Likewise.
15129 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15130 (grid_parallel_clauses_gridifiable): Likewise.
15132 * config/nvptx/mkoffload.c (process): Add space in between
15135 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15136 "MOD4_SSE_REGS" and "ALL_REGS".
15138 * spellcheck.c (test_data): Add , in between "foo" and "food".
15140 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15143 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15144 boundary crossing check and subsequent code generation agree.
15146 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15148 * config/aarch64/aarch64.c (has_memory_op): Delete.
15149 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15152 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15154 PR rtl-optimization/79388
15155 PR rtl-optimization/79450
15156 * combine.c (distribute_notes): When removing TEM_INSN for which
15157 corresponding dest has last value recorded, invalidate that last
15160 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15162 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15163 of explicit '@'. Add missing assembly comment marker on branch costs
15166 2017-02-13 Nathan Sidwell <nathan@acm.org>
15168 * gengtype-lex.l (<in_struct>): Add '/'.
15170 2017-02-13 Martin Liska <mliska@suse.cz>
15173 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15175 2017-02-13 Richard Biener <rguenther@suse.de>
15177 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15179 * configure: Re-generate.
15180 * config.in: Likewise.
15181 * graphite-dependences.c: Simplify as if
15182 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15183 * graphite-isl-ast-to-gimple.c: Likewise.
15184 * graphite-optimize-isl.c: Likewise.
15185 * graphite-poly.c: Likewise.
15186 * graphite-sese-to-poly.c: Likewise.
15187 * graphite.h: Likewise.
15188 * toplev.c: Include isl/version.h and use isl_version () for
15189 printing the ISL version.
15190 * doc/install.texi: Update ISL requirement.
15192 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15194 * doc/standards.texi (Standards): Update reference to
15197 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15199 * doc/extend.texi (Named Address Spaces): sourceware.org now
15201 * doc/install.texi (Binaries): Ditto.
15204 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15206 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15207 standard terminology "stringize"/"stringizing" throughout.
15208 * doc/cppinternals.texi: Likewise.
15210 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15212 * doc/extend.texi: Fix some spelling mistakes and typos.
15213 * doc/invoke.texi: Likewise.
15215 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15218 * params.def (inline-min-speedup) Change from 10 to 8.
15220 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15222 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15225 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15228 * ipa-inline-analysis.c (get_minimal_bb): New function.
15229 (record_modified): Use it.
15230 (remap_edge_change_prob): Handle also ancestor functions.
15232 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15234 * doc/contrib.texi (Contributors): Remove broken link into
15235 the Mauve CVS repository.
15237 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15239 PR middle-end/79454
15240 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15241 result computation whenever lhs doesn't have vector mode, not
15242 just when it has BLKmode.
15244 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15246 * doc/makefile.texi (profiledbootstrap): Refer to the
15247 installation instructions only in textual form.
15249 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15252 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15254 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15256 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15257 (Specific): Update mingw-w64 reference.
15259 (Specific): Remove broken link to Renesas RX processor.
15261 2017-02-10 Richard Biener <rguenther@suse.de>
15263 * toplev.c (process_options): Do not mention obsolete graphite
15264 options when printing sorry message about missing graphite support.
15265 Mention -floop-nest-optimize.
15267 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15269 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15270 (vtst_p16): Likewise.
15271 (vtstq_p8): Likewise.
15272 (vtstq_p16): Likewise.
15274 (vtstq_p64): Likewise.
15275 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15276 (vset_lane_p64): New.
15277 (vsetq_lane_p64): New.
15279 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15281 PR tree-optimization/79411
15282 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15283 stmt operands are SSA_NAMEs used in abnormal phis.
15284 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15287 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15290 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15293 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15295 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15297 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15299 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15303 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15304 not arbitrary TREE_CONSTANT.
15307 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15308 "omp declare target link" attribute unless is_global_var.
15309 * omp-offload.c (find_link_var_op): Likewise.
15311 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15312 Chung-Lin Tang <cltang@codesourcery.com>
15314 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15316 (gimplify_adjust_omp_clauses): Don't delete TILE.
15317 (gimplify_omp_for): Deal with TILE.
15318 * internal-fn.c (expand_GOACC_TILE): New function.
15319 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15321 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15322 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15324 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15325 avoid DIV for outermost collapse var.
15326 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15327 Remove out of date comments, fix whitespace.
15328 * omp-general.c (omp_extract_for_data): Deal with tiling.
15329 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15330 adjust OLF_DIM_BASE value.
15331 (struct omp_for_data): Add tiling field.
15332 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15333 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15334 for auto loops. Remove default auto determining, moved to
15335 oacc_loop_fixed_partitions.
15336 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15337 stmts, add e_mask field.
15338 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15339 (oacc_thread_numbers): Use oacc_dim_call.
15340 (oacc_xform_tile): New.
15341 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15342 (finish_oacc_loop): Adjust for ifns vector.
15343 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15344 add case for GOACC_TILE fns.
15345 (oacc_loop_xform_loop): Delete.
15346 (oacc_loop_process): Iterate over call list directly, and add
15347 handling for GOACC_TILE fns.
15348 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15350 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15351 vector partitioning to outer loops. Assign 2 partitions to loops
15352 when available. Add TILE handling.
15353 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15354 (execite_oacc_device_lower): Process GOACC_TILE fns,
15355 ignore unknown specs.
15356 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15357 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15358 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15360 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15362 * configure.ac (ACX_BUGURL): Update.
15363 * configure: Regenerate.
15365 2017-02-09 Richard Biener <rguenther@suse.de>
15367 PR tree-optimization/69823
15368 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15369 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15371 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15373 * config/arc/arc-c.def: Add __NPS400__ definition.
15374 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15375 (TARGET_NPS400): Define.
15377 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15379 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15381 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15382 pointer, arch_info.
15383 (arc_cpu_types): Fill the arch_info field with a pointer into the
15384 arc_arch_types table.
15385 (arc_selected_cpu): Declare.
15386 * config/arc/arc.c (arc_selected_cpu): Make global.
15387 (arc_selected_arch): Delete.
15388 (arc_base_cpu): Delete.
15389 (arc_override_options): Remove references to deleted variables,
15390 update access to arch information.
15391 (ARC_OPT): Update access to arch information.
15392 (ARC_OPTX): Likewise.
15393 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15394 (TARGET_ARC600): Update access to arch information.
15395 (TARGET_ARC601): Likewise.
15396 (TARGET_ARC700): Likewise.
15397 (TARGET_EM): Likewise.
15398 (TARGET_HS): Likewise.
15399 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15402 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15405 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15406 condition/operands for integer GE/LE/GEU/LEU operations.
15408 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15410 PR translation/79397
15411 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15414 2017-02-08 Martin Jambor <mjambor@suse.cz>
15417 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15418 whether allocation happened.
15419 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15420 nothing was allocated.
15422 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15424 PR tree-optimization/79408
15425 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15426 constant, but SSA_NAME with a known integer range, use the minimum
15427 of that range instead of op1 to determine if modulo can be replaced
15428 with its first operand.
15430 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15432 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15434 2017-02-08 Richard Biener <rguenther@suse.de>
15436 PR tree-optimization/71824
15437 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15438 Check all loops contained in the merged region.
15440 2017-02-07 Andrew Pinski <apinski@cavium.com>
15442 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15444 2017-02-07 Andrew Pinski <apinski@cavium.com>
15446 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15447 (thunderxt88): Likewise.
15448 (thunderxt81): Disable LSE and change v8.1 to v8.
15449 (thunderxt83): Likewise.
15451 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15452 Richard Biener <rguenther@suse.de>
15454 PR middle-end/79399
15455 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15456 type from int to size_t.
15457 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15459 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15461 PR rtl-optimization/79386
15462 * cprop.c (bypass_conditional_jumps): Initialize
15463 bypass_last_basic_block already before splitting bbs after
15464 unconditional traps...
15465 (bypass_conditional_jumps): ... rather than here.
15468 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15469 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15470 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15471 fix -masm=intel patterns.
15473 2017-02-07 Richard Biener <rguenther@suse.de>
15475 PR tree-optimization/79256
15476 PR middle-end/79278
15477 * builtins.c (get_object_alignment_2): Use min_align_of_type
15478 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15479 and ADJUST_FIELD_ALIGN.
15481 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15483 * doc/tm.texi: Regenerate.
15484 * stor-layout.c (layout_decl): Adjust.
15485 (update_alignment_for_field): Likewise.
15486 (place_field): Likewise.
15487 (min_align_of_type): Likewise.
15488 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15489 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15490 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15491 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15492 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15493 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15494 * config/i386/i386.c (x86_field_alignment): Likewise.
15495 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15496 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15497 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15498 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15499 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15500 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15504 2017-01-30 Richard Biener <rguenther@suse.de>
15506 PR tree-optimization/79256
15507 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15508 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15511 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15513 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15514 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15515 builtins to SImode and emit a zero-extend, if necessary.
15517 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15519 * docs/invoke.texi (RISC-V Options): Alphabetize.
15521 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15523 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15526 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15528 * config/riscv/riscv.c: New file.
15529 * common/config/riscv/riscv-common.c: Likewise.
15530 * config.gcc: Likewise.
15531 * config/riscv/constraints.md: Likewise.
15532 * config/riscv/elf.h: Likewise.
15533 * config/riscv/generic.md: Likewise.
15534 * config/riscv/linux.h: Likewise.
15535 * config/riscv/multilib-generator: Likewise.
15536 * config/riscv/peephole.md: Likewise.
15537 * config/riscv/pic.md: Likewise.
15538 * config/riscv/predicates.md: Likewise.
15539 * config/riscv/riscv-builtins.c: Likewise.
15540 * config/riscv/riscv-c.c: Likewise.
15541 * config/riscv/riscv-ftypes.def: Likewise.
15542 * config/riscv/riscv-modes.def: Likewise.
15543 * config/riscv/riscv-opts.h: Likewise.
15544 * config/riscv/riscv-protos.h: Likewise.
15545 * config/riscv/riscv.h: Likewise.
15546 * config/riscv/riscv.md: Likewise.
15547 * config/riscv/riscv.opt: Likewise.
15548 * config/riscv/sync.md: Likewise.
15549 * config/riscv/t-elf-multilib: Likewise.
15550 * config/riscv/t-linux: Likewise.
15551 * config/riscv/t-linux-multilib: Likewise.
15552 * config/riscv/t-riscv: Likewise.
15553 * configure.ac: Likewise.
15554 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15555 Waterman as RISC-V maintainers.
15556 * doc/install.texi: Add RISC-V entries.
15557 * doc/invoke.texi: Add RISC-V options section.
15558 * doc/md.texi: Add RISC-V constraints section.
15559 * configure: Regenerated.
15561 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15564 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15565 false values to be constant vectors with all 0 or all 1 bits set.
15566 (vcondu<mode><mode>): Likewise.
15567 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15569 (fpmask_comparison_operator): Update comment.
15570 (vecint_comparison_operator): New predicate.
15571 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15572 vector conditionals when the true and false values are constant
15573 vectors with all 0 bits or all 1 bits set.
15575 2017-02-06 Martin Sebor <msebor@redhat.com>
15577 PR tree-optimization/79376
15578 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15580 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15582 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15583 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15584 to simplify split condition.
15586 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15588 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15589 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15592 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15594 PR rtl-optimization/68664
15595 * target.def (can_speculate_insn): New hook.
15596 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15597 * doc/tm.texi: Regenerate.
15598 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15599 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15600 (rs6000_sched_can_speculate_insn): New function.
15602 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15604 PR tree-optimization/79284
15605 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15606 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15607 vectorizable_mask_load_store, vectorizable_operation,
15608 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15609 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15610 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15611 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15612 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15613 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15614 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15615 is_gimple_assign (stmt). Replace another such test with
15616 is_gimple_assign (stmt).
15618 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15621 * config/avr/avr.c (rtl-iter.h): Include it.
15622 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15623 (avr_legitimate_combined_insn): ...and implementation.
15625 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15627 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15628 * config/s390/s390.c (s390_const_operand_ok)
15629 (s390_canonicalize_comparison, s390_extract_part)
15630 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15631 (s390_contiguous_bitmask_p, s390_rtx_costs)
15632 (legitimize_pic_address): Likewise.
15633 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15634 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15635 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15636 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15637 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15639 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15641 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15642 REGNO($0) == REGNO($1).
15644 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15646 * config/s390/linux.h(SIZE_TYPE): Add comment.
15648 2017-02-06 Julian Brown <julian@codesourcery.com>
15649 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15650 Virendra Pathak <virendra.pathak@broadcom.com>
15652 * config/aarch64/aarch64-cores.def: Change the scheduler
15654 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15655 * config/aarch64/thunderx2t99.md: New file.
15657 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15659 * doc/standards.texi (Go Language): Update link to language
15662 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15664 * tree-eh.c (lower_resx): Sanitize profile.
15665 (cleanup_empty_eh_move_lp): Likewise.
15667 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15670 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15672 * cfgloopmanip.h (loop_version): Update prototype.
15673 * modulo-sched.c (sms_schedule): Update call of loop_version.
15674 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15675 * tree-parloops.c (gen_parallel_loop): Likewise.
15676 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15677 * tree-ssa-loop-split.c (split_loop): Likewise.
15678 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15679 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15681 2017-02-05 Martin Liska <mliska@suse.cz>
15684 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15686 (print_operand_address): Initialize a struct to zero.
15688 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15690 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15691 garbage collector only in textual form.
15693 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15695 * doc/extend.texi (x86 specific memory model extensions for
15696 transactional memory): Simplify a phrase.
15698 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15701 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15702 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15703 (atomic_storedi_1): Likewise.
15705 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15707 PR tree-optimization/79338
15708 * tree-parloops.c (gather_scalar_reductions): Don't call
15709 vect_analyze_loop_form for loop->inner before destroying loop's
15712 2017-02-03 Martin Sebor <msebor@redhat.com>
15714 PR tree-optimization/79327
15715 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15716 when precision has resulted in leading zeros.
15717 (format_integer): Adjust the likely counter to assume an unknown
15718 argument that may be zero is non-zero.
15720 2017-02-03 Jason Merrill <jason@redhat.com>
15723 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15724 avoid copying non-taken branch.
15726 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15728 PR tree-optimization/79340
15729 * tree-vect-loop.c (vectorizable_reduction): Release
15730 vec_defs elements after safe_splicing them into other vectors.
15733 PR tree-optimization/79327
15734 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15735 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15737 (format_integer): Use wide_int_to_tree instead of build_int_cst
15738 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15739 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15740 of shortest and longest sequence.
15742 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15744 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15745 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15747 2017-02-03 Walter Lee <walt@tilera.com>
15750 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15751 after initial stackframe link reg save.
15752 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15754 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15757 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15758 wu for stxssp alternative.
15760 2017-02-03 Martin Sebor <msebor@redhat.com>
15762 PR tree-optimization/79352
15763 * gimple-fold.c (get_range_strlen): Add argument.
15764 (get_range_strlen): Change return type to bool.
15765 (get_maxval_strlen): Pass in a dummy argument.
15766 * gimple-fold.h (get_range_strlen): Change return type to bool.
15767 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15768 * tree.h (array_at_struct_end_p): Add argument.
15769 * tree.c (array_at_struct_end_p): Handle it.
15771 2017-02-03 Martin Liska <mliska@suse.cz>
15774 * multiple_target.c (create_dispatcher_calls): Redirect edge
15775 from a caller of a dispatcher.
15776 (expand_target_clones): Make the clones local.
15777 (ipa_target_clone): Do both target clones and resolvers.
15778 (ipa_dispatcher_calls): Remove the pass.
15779 (pass_dispatcher_calls::gate): Likewise.
15780 (make_pass_dispatcher_calls): Likewise.
15781 * passes.def (pass_target_clone): Put as very first IPA early
15784 2017-02-03 Martin Liska <mliska@suse.cz>
15786 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15787 in case of a function with ifunc attribute.
15789 2017-02-03 Martin Liska <mliska@suse.cz>
15791 * cgraph.c (cgraph_node::dump): Dump function version info.
15792 * symtab.c (symtab_node::dump_base): Add missing new line.
15794 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15796 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15797 (ifcombine_ifandif): Use it.
15799 2017-02-03 Martin Liska <mliska@suse.cz>
15801 * doc/invoke.texi: Document default value for
15802 use-after-scope-direct-emission-threshold.
15804 2017-02-03 Martin Liska <mliska@suse.cz>
15806 PR tree-optimization/79339
15807 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15808 (format_floating): Likewise.
15810 2017-02-03 Martin Liska <mliska@suse.cz>
15813 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15815 (ipa_node_params_t::remove): Likewise.
15816 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15817 initialization from removed ipa_node_params_t::insert.
15818 (ipa_node_params::~ipa_node_params): Move from removed
15819 ipa_node_params_t::release.
15820 * symbol-summary.h (symbol_summary::m_released): New member.
15821 Do not release a summary twice. Do not allow to call finalizer
15822 for types of a summary that live in GGC memory.
15824 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15826 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15829 2017-02-02 Martin Sebor <msebor@redhat.com>
15831 PR middle-end/79275
15832 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15833 (format_string): Tighten up the range of output for non-constant
15834 strings and correct the expected range for wide non-constant strings.
15836 2017-02-02 Martin Sebor <msebor@redhat.com>
15838 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15840 PR middle-end/32003
15841 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15843 (-fdump-tree-@var): Add to index and document how to come up
15844 with pass-specific option and dump file names.
15845 (-fdump-passes): Clarify where to look for output.
15847 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15849 PR middle-end/77445
15850 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15851 statistics of the analyzed path; allow threading for speed when
15852 any of BBs along the path are optimized for speed.
15854 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15856 PR middle-end/78468
15857 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15858 settings of the virtual registers.
15861 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15863 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15864 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15867 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15869 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15870 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15872 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15874 * config/s390/s390.md: Add missing comments with the expanded
15876 * config/s390/vector.md: Likewise.
15877 * config/s390/vx-builtins.md: Likewise.
15879 2017-02-02 Jakub Jelinek <jakub@redhat.com>
15882 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
15883 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
15884 conditions on a single line.
15886 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15888 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15889 __S390_VX__ to __VX__.
15891 2017-02-01 Andrew Pinski <apinski@cavium.com>
15893 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
15894 stmt_info to record_stmt_cost.
15895 (vect_get_known_peeling_cost): Pass stmt_info if known to
15897 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
15898 cpu_vector_cost field into
15899 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
15900 field into vec_int_stmt_cost and vec_fp_stmt_cost.
15901 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
15902 splitting of scalar_stmt_cost and vec_stmt_cost.
15903 (thunderx_vector_cost): Likewise.
15904 (cortexa57_vector_cost): LIkewise.
15905 (exynosm1_vector_cost): Likewise.
15906 (xgene1_vector_cost): Likewise.
15907 (thunderx2t99_vector_cost): Improve after the splitting of the two
15909 (aarch64_builtin_vectorization_cost): Update for the splitting of
15910 scalar_stmt_cost and vec_stmt_cost.
15912 2017-02-01 Torvald Riegel <triegel@redhat.com>
15913 Richard Henderson <rth@redhat.com>
15915 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
15916 conditional on existance of a fast atomic load.
15917 * optabs-query.c (can_atomic_load_p): New function.
15918 * optabs-query.h (can_atomic_load_p): Declare it.
15919 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
15920 no fast atomic load is available for the particular size of access.
15921 (expand_atomic_compare_and_swap): Likewise.
15922 (expand_atomic_load): Likewise.
15923 (expand_atomic_store): Likewise.
15924 (expand_atomic_fetch_op): Likewise.
15925 * testsuite/lib/target-supports.exp
15926 (check_effective_target_sync_int_128): Remove x86 because it provides
15927 no fast atomic load.
15928 (check_effective_target_sync_int_128_runtime): Likewise.
15930 2017-02-01 Richard Biener <rguenther@suse.de>
15932 * graphite.c: Include tree-vectorizer.h for find_loop_location.
15933 (graphite_transform_loops): Provide opt-info for optimized nests.
15934 * tree-parloop.c (parallelize_loops): Provide opt-info for
15935 parallelized loops.
15937 2017-02-01 Richard Biener <rguenther@suse.de>
15939 PR middle-end/79315
15940 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
15941 was not set before.
15943 2017-02-01 Richard Biener <rguenther@suse.de>
15945 PR tree-optimization/71824
15946 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15947 Verify the loops are valid in the merged SESE region.
15948 (scop_detection::can_represent_loop_1): Check analyzing the
15949 evolution of the number of iterations in the region succeeds.
15951 2017-01-31 Ian Lance Taylor <iant@golang.org>
15953 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
15954 REG_ARGS_SIZE note to 32-bit push insns and call insn.
15956 2017-01-31 David Malcolm <dmalcolm@redhat.com>
15958 PR preprocessor/79210
15959 * input.c (get_substring_ranges_for_loc): Replace line_width
15960 assertion with error-handling.
15962 2017-01-31 Richard Biener <rguenther@suse.de>
15964 PR tree-optimization/77318
15965 * graphite-sese-to-poly.c (extract_affine): Fix assert.
15966 (create_pw_aff_from_tree): Take loop parameter.
15967 (add_condition_to_pbb): Pass loop of the condition to
15968 create_pw_aff_from_tree.
15970 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15972 * config/s390/s390.c (s390_asan_shadow_offset): New function.
15973 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
15975 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
15979 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
15981 (convert_int_to_float128): Likewise.
15982 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
15983 (convert_int_to_float128): Likewise.
15984 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
15985 (UNSPEC_IEEE128_CONVERT): Likewise.
15986 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
15987 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
15988 Use local variables for IBM extended format.
15989 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
15990 (fix_trunc<mode>si2_fprs): Likewise.
15991 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
15992 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
15993 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
15994 to know that we can now have integers of all sizes in vector
15996 (fix<uns>_<mode>di2_hw): Likewise.
15997 (float<uns>_<mode>si2_hw): Likewise.
15998 (fix_<mode>si2_hw): Likewise.
15999 (fixuns_<mode>si2_hw): Likewise.
16000 (float<uns>_<mode>di2_hw): Likewise.
16001 (float_<mode>di2_hw): Likewise.
16002 (float_<mode>si2_hw): Likewise.
16003 (floatuns_<mode>di2_hw): Likewise.
16004 (floatuns_<mode>si2_hw): Likewise.
16005 (xscvqp<su>wz_<mode>): Delete, no longer used.
16006 (xscvqp<su>dz_<mode>): Likewise.
16007 (xscv<su>dqp_<mode>): Likewise.
16008 (ieee128_mfvsrd_64bit): Likewise.
16009 (ieee128_mfvsrd_32bit): Likewise.
16010 (ieee128_mfvsrwz): Likewise.
16011 (ieee128_mtvsrw): Likewise.
16012 (ieee128_mtvsrd_64bit): Likewise.
16013 (ieee128_mtvsrd_32bit): Likewise.
16015 2017-01-31 Martin Liska <mliska@suse.cz>
16018 * ipa-prop.c (ipa_free_all_node_params): Call release method
16019 instead of ~sumbol_summary to not to trigger double times
16022 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16024 PR tree-optimization/71691
16025 * bitmap.h (class auto_bitmap): New.
16026 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16027 is_maybe_undefined instead of ssa_undefined_value_p.
16029 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16031 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16032 __S390_ARCH_LEVEL__ to __ARCH__.
16034 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16036 PR tree-optimization/79267
16037 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16038 if should_remove_lhs_p is true.
16040 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16043 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16044 (add_alignment_attribute): New.
16045 (base_type_die): Add alignment attribute.
16046 (subrange_type_die): Likewise.
16047 (modified_type_die): Likewise.
16048 (gen_array_type_die): Likewise.
16049 (gen_descr_array_type_die: Likewise.
16050 (gen_enumeration_type_die): Likewise.
16051 (gen_subprogram_die): Likewise.
16052 (gen_variable_die): Likewise.
16053 (gen_field_die): Likewise.
16054 (gen_ptr_to_mbr_type_die): Likewise.
16055 (gen_struct_or_union_type_die): Likewise.
16056 (gen_subroutine_type_die): Likewise.
16057 (gen_typedef_die): Likewise.
16058 (base_type_cmp): Compare alignment attribute.
16060 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16063 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16064 (setb_unsigned) New pattern for setb with CCUNS.
16065 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16066 subfc./subfe sequence to avoid overflow problems. Generate a
16067 shorter sequence with cmpld/setb for power9.
16068 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16069 for generating subfc. instruction.
16070 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16071 now uses this instruction.
16073 2017-01-30 Ian Lance Taylor <iant@google.com>
16076 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16077 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16079 2017-01-30 Martin Sebor <msebor@redhat.com>
16081 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16082 Move constant to the right of a relational operator.
16083 (get_mpfr_format_length, format_character, format_string): Ditto.
16084 (should_warn_p, maybe_warn): Same.
16086 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16088 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16091 * asan.c (get_translation_unit_decl): Remove function.
16092 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16094 2017-01-30 Martin Liska <mliska@suse.cz>
16096 PR gcov-profile/79259
16097 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16098 -fprofile-generate.
16100 2017-01-30 Martin Liska <mliska@suse.cz>
16103 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16104 Initialize variables with NULL value.
16106 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16109 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16111 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16113 2017-01-30 Richard Biener <rguenther@suse.de>
16115 PR tree-optimization/79276
16116 * tree-vrp.c (process_assert_insertions): Properly adjust common
16117 when removing a duplicate.
16119 * gcc.dg/torture/pr79276.c: New testcase.
16121 2017-01-30 Richard Biener <rguenther@suse.de>
16123 PR tree-optimization/79256
16124 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16125 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16127 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16129 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16132 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16133 ("*r<noxa>sbg_<mode>_sll_bitmask")
16134 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16135 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16136 Use contiguous_bitmask_nowrap_operand.
16138 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16141 * config/rs6000/altivec.h (vec_xl): Revise #define.
16142 (vec_xst): Likewise.
16144 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16146 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16148 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16150 PR rtl-optimization/79194
16151 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16152 traps before call to bypass_conditional_jumps.
16154 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16156 PR tree-optimization/71374
16157 * lra-constraints.c (check_conflict_input_operands): New.
16158 (match_reload): Use it.
16160 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16163 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16164 account to calculate conflict_set.
16166 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16168 PR rtl-optimization/78559
16169 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16170 other_insn in combine.
16172 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16174 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16175 uint16_type_node for BT_UINT16.
16177 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16179 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16180 "RTL Tests" to menu.
16181 (GIMPLE Tests): New node.
16182 (RTL Tests): New node.
16184 2017-01-27 Richard Biener <rguenther@suse.de>
16186 PR tree-optimization/79245
16187 * tree-loop-distribution.c (distribute_loop): Apply cost
16188 modeling also to detected patterns.
16190 2017-01-27 Richard Biener <rguenther@suse.de>
16192 PR tree-optimization/71433
16193 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16194 (compare_assert_loc): New function.
16195 (process_assert_insertions): Sort and optimize assert locations
16196 to remove duplicates and push down identical assertions on
16197 edges to their destination block.
16199 2017-01-27 Richard Biener <rguenther@suse.de>
16201 PR tree-optimization/79244
16202 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16203 out SSA names even if abnormal.
16205 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16207 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16208 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16209 instead of MPFR_RNDN.
16211 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16214 * arm.c (arm_option_override): Don't call build_target_option_node
16215 until after doing all option overrides.
16216 (arm_valid_target_attribute_tree): Likewise.
16218 2017-01-27 Martin Liska <mliska@suse.cz>
16220 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16221 for {cd}tors and C++ {cd}tors.
16223 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16225 * config/s390/s390.md ("*setmem_long_and")
16226 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16228 2017-01-26 Martin Sebor <msebor@redhat.com>
16230 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16233 2017-01-26 Martin Sebor <msebor@redhat.com>
16235 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16236 HAVE_DFmode before using XFmode or DFmode.
16237 (parse_directive): Avoid using the z length modifier to avoid
16238 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16240 PR middle-end/78703
16241 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16242 to accept adjustment as an array.
16243 (get_int_range): New function.
16244 (struct directive): Make width and prec arrays.
16245 (directive::set_width, directive::set_precision): Call get_int_range.
16246 (format_integer, format_floating): Handle width and precision ranges.
16247 (format_string, parse_directive): Same.
16249 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16252 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16253 comdat_type_p set, just clone them, but keep the children in the
16257 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16258 which have direct callers with -fvar-tracking-assignments enabled
16260 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16261 inside of type units.
16263 2017-01-26 Martin Sebor <msebor@redhat.com>
16265 PR middle-end/78703
16266 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16268 (struct format_result): Replace number_chars, number_chars_min,
16269 and number_chars_max with a single member of struct result_range.
16271 (format_result::operator+=): Adjust.
16272 (struct fmtresult): Remove bounded. Handle likely and unlikely
16274 (fmtresult::adjust_for_width_or_precision): New function.
16275 (fmtresult:type_max_digits): New function.
16276 (bytes_remaining): Handle likely and unlikely counters.
16277 (min_bytes_remaining): Remove.
16278 (format_percent): Simplify.
16279 (format_integer, format_floating): Set likely and unlikely counters.
16280 (get_string_length, format_character, format_string): Same.
16281 (format_plain, should_warn_p): New function.
16282 (maybe_warn): Call should_warn_p. Update diagnostic messages
16283 and handle those for all directives, including plain strings.
16284 (format_directive): Handle likely and unlikely counters.
16285 Remove unnecessary quoting from diagnostics. Add an informational
16287 (add_bytes): Remove.
16288 (pass_sprintf_length::compute_format_length): Simplify.
16289 (try_substitute_return_value): Handle likely and unlikely counters.
16291 2017-01-26 Carl Love <cel@us.ibm.com>
16293 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16294 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16296 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16299 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16300 endianess for subregs into account.
16301 * lra-constraints.c (lra_constraints): Do risky transformations
16302 always on the first iteration.
16303 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16304 last_call_used_reg_set.
16305 (process_bb_lives): Define and use last_call_used_reg_set.
16306 * lra.c (lra): Always continue after lra_constraints on the first
16309 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16311 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16313 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16315 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16317 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16318 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16319 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16320 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16321 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16322 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16323 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16324 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16325 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16327 2017-01-26 Marek Polacek <polacek@redhat.com>
16330 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16331 for the third operand.
16333 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16335 PR middle-end/79236
16336 * omp-low.c (struct omp_context): Add simt_stmt field.
16337 (scan_omp_for): Return omp_context *.
16338 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16339 context to the _simt_ SIMD stmt.
16340 (lower_omp_for): For combined SIMD with sibling _simt_
16341 SIMD, make sure to use the same decls in _looptemp_
16342 clauses as in the sibling.
16344 2017-01-26 David Sherwood <david.sherwood@arm.com>
16346 PR middle-end/79212
16347 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16350 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16353 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16354 emit fld b; fld a; if possible.
16356 * brig-builtins.def: Update copyright years.
16357 * config/arm/arm_acle_builtins.def: Update copyright years.
16359 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16362 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16363 constraint instead of o for the stxsd instruction.
16365 2017-01-25 Carl Love <cel@us.ibm.com>
16367 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16368 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16370 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16372 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16374 2017-01-25 Richard Biener <rguenther@suse.de>
16376 PR tree-optimization/69264
16377 * target.def (vector_alignment_reachable): Improve documentation.
16378 * doc/tm.texi: Regenerate.
16379 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16381 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16382 earlier changes with respect to TYPE_USER_ALIGN.
16383 (vector_alignment_reachable_p): Likewise. Improve dumping.
16385 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16388 * config/arm/arm.md (xordi3): Force constant operand into a register
16391 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16393 * doc/invoke.texi (-fstore-merging): Correct default optimization
16394 levels at which it is enabled.
16395 (-O): Move -fstore-merging from list to...
16398 2017-01-25 Richard Biener <rguenther@suse.de>
16401 * omp-expand.c: Include debug.h.
16402 (expand_omp_taskreg): Make sure to generate early debug before
16403 outlining anything from a function.
16404 (expand_omp_target): Likewise.
16405 (grid_expand_target_grid_body): Likewise.
16407 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16410 * asan.c (get_translation_unit_decl): New function.
16411 (asan_add_global): Extract modules file name from globals
16412 TRANSLATION_UNIT_DECL name.
16414 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16417 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16418 for long calls with APCS frame and VFP.
16420 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16422 * cfg.c (original_copy_tables_initialized_p): New function.
16423 * cfg.h (original_copy_tables_initialized_p): New decl.
16424 * cfgrtl.c (relink_block_chain): Guard the call to
16425 free_original_copy_tables with a call to
16426 original_copy_tables_initialized_p.
16427 * cgraph.h (symtab_node::native_rtl_p): New decl.
16428 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16429 (symtab_node::needed_p): Don't assert for early assembly output
16430 for __RTL functions.
16431 (cgraph_node::finalize_function): Set "force_output" for __RTL
16433 (cgraph_node::analyze): Bail out early for __RTL functions.
16434 (analyze_functions): Update assertion to support __RTL functions.
16435 (cgraph_node::expand): Bail out early for __RTL functions.
16436 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16438 * function.h (struct function): Update comment for field
16440 * gimple-expr.c: Include "tree-pass.h".
16441 (gimple_has_body_p): Return false for __RTL functions.
16442 * Makefile.in (OBJS): Add run-rtl-passes.o.
16443 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16445 (gcc::pass_manager::get_clean_slate): New accessor.
16446 * passes.c: Include "insn-addr.h".
16447 (should_skip_pass_p): Add logging. Update logic for running
16448 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16449 property-provider override so it is only done for gimple passes.
16451 (skip_pass): New function.
16452 (execute_one_pass): Call skip_pass when skipping passes.
16453 * read-md.c (md_reader::read_char): Support filtering
16454 the input to a subset of line numbers.
16455 (md_reader::md_reader): Initialize fields
16456 m_first_line and m_last_line.
16457 (md_reader::read_file_fragment): New function.
16458 * read-md.h (md_reader::read_file_fragment): New decl.
16459 (md_reader::m_first_line): New field.
16460 (md_reader::m_last_line): New field.
16461 * read-rtl-function.c (function_reader::create_function): Only
16462 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16463 curr_properties. Set DECL_INITIAL to a dummy block.
16464 (read_rtl_function_body_from_file_range): New function.
16465 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16467 * run-rtl-passes.c: New file.
16468 * run-rtl-passes.h: New file.
16470 2017-01-24 Jeff Law <law@redhat.com>
16472 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16475 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16477 PR tree-optimization/79159
16478 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16479 (record_nonwrapping_iv): Improve boundary using above function if no
16480 value range information.
16482 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16483 Martin Jambor <mjambor@suse.cz>
16485 * brig-builtins.def: New file.
16486 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16487 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16488 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16489 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16490 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16491 * builtin-types.def (BT_INT8): New.
16492 (BT_INT16): Likewise.
16493 (BT_UINT8): Likewise.
16494 (BT_UINT16): Likewise.
16495 (BT_FN_ULONG): Likewise.
16496 (BT_FN_UINT_INT): Likewise.
16497 (BT_FN_UINT_ULONG): Likewise.
16498 (BT_FN_UINT_LONG): Likewise.
16499 (BT_FN_UINT_PTR): Likewise.
16500 (BT_FN_ULONG_PTR): Likewise.
16501 (BT_FN_INT8_FLOAT): Likewise.
16502 (BT_FN_INT16_FLOAT): Likewise.
16503 (BT_FN_UINT32_FLOAT): Likewise.
16504 (BT_FN_UINT16_FLOAT): Likewise.
16505 (BT_FN_UINT8_FLOAT): Likewise.
16506 (BT_FN_UINT64_FLOAT): Likewise.
16507 (BT_FN_UINT16_UINT32): Likewise.
16508 (BT_FN_UINT32_UINT16): Likewise.
16509 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16510 (BT_FN_INT_PTR_INT): Likewise.
16511 (BT_FN_UINT_PTR_UINT): Likewise.
16512 (BT_FN_LONG_PTR_LONG): Likewise.
16513 (BT_FN_ULONG_PTR_ULONG): Likewise.
16514 (BT_FN_VOID_UINT64_UINT64): Likewise.
16515 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16516 (BT_FN_INT8_INT8_INT8): Likewise.
16517 (BT_FN_INT16_INT16_INT16): Likewise.
16518 (BT_FN_INT_INT_INT): Likewise.
16519 (BT_FN_UINT_FLOAT_UINT): Likewise.
16520 (BT_FN_FLOAT_UINT_UINT): Likewise.
16521 (BT_FN_ULONG_UINT_UINT): Likewise.
16522 (BT_FN_ULONG_UINT_PTR): Likewise.
16523 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16524 (BT_FN_UINT_UINT_UINT): Likewise.
16525 (BT_FN_VOID_UINT_PTR): Likewise.
16526 (BT_FN_UINT_UINT_PTR: Likewise.
16527 (BT_FN_UINT32_UINT64_PTR): Likewise.
16528 (BT_FN_INT_INT_UINT_UINT): Likewise.
16529 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16530 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16531 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16532 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16533 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16534 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16535 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16536 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16537 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16538 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16539 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16540 * doc/frontends.texi: List BRIG FE.
16541 * doc/install.texi (Testing): Add BRIG tesring requirements.
16542 * doc/invoke.texi (Overall Options): Mention BRIG.
16543 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16545 2017-01-24 Richard Biener <rguenther@suse.de>
16547 PR translation/79208
16548 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16550 2017-01-24 Martin Jambor <mjambor@suse.cz>
16553 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16554 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16555 and known_contexts.
16557 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16559 PR middle-end/79123
16560 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16561 casts from signed to unsigned really don't have a range.
16563 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16565 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16566 GMP_RNDx for compatiblity.
16568 2017-01-24 Martin Liska <mliska@suse.cz>
16571 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16572 that would prevent us to call alloca with -1 as argument.
16574 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16576 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16577 Avoid -Wformat-security warning.
16579 2017-01-23 Andrew Pinski <apinski@cavium.com>
16581 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16584 2017-01-23 Martin Sebor <msebor@redhat.com>
16586 PR middle-end/78703
16587 * gimple-ssa-sprintf.c (warn_level): New global.
16588 (format_integer): Use it here and throughout the rest of the file.
16589 Use the same switch to compute sign as base.
16590 (maybe_warn): New function.
16591 (format_directive): Factor out warnings into maybe_warn.
16592 Add debugging output. Use warn_level.
16593 (add_bytes): Use warn_level.
16594 (pass_sprintf_length::compute_format_length): Add debugging output.
16595 (try_substitute_return_value): Same.
16596 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16598 PR middle-end/78703
16599 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16600 (struct fmtresult, format_integer, format_floating): Adjust.
16601 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16602 (get_string_length, format_string,format_directive): Same.
16603 (pass_sprintf_length::compute_format_length): Same.
16604 (try_substitute_return_value): Simplify slightly.
16606 PR middle-end/78703
16607 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16608 (fmtresult::operator+=): Outlined.
16609 (struct fmtresult): Add ctors.
16610 (struct conversion_spec): Rename...
16611 (struct directive): ...to this. Add and remove data members.
16612 (directive::set_width, directive::set_precision): New functions.
16613 (format_percent): Use fmtresult ctor.
16614 (get_width_and_precision): Remove.
16615 (format_integer): Make naming changes. Avoid computing width and
16617 (format_floating): Same. Adjust indentation.
16618 (format_character, format_none): New functions.
16619 (format_string): Moved character handling to format_character.
16620 (format_directive): Remove arguments, change return type.
16621 (parse_directive): New function.
16622 (pass_sprintf_length::compute_format_length): Move directive
16623 parsing to parse_directive.
16625 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16627 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16628 (assign_assembler_name_if_needed): ... this.
16629 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16630 (assign_assembler_name_if_needed): ... this.
16631 (free_lang_data_in_cgraph): Adjust callers.
16632 * cgraphunit.c (cgraph_node::analyze): Likewise.
16633 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16636 2017-01-23 Richard Biener <rguenther@suse.de>
16638 PR tree-optimization/79088
16639 PR tree-optimization/79188
16640 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16641 resetting loop bounds after last path deletion. Reset loop
16642 bounds of the target loop, make code match the comments.
16643 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16644 Make sure loops need no fixups.
16646 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16648 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16649 exponent support with double type for first argument.
16650 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16651 type returned by __builtin_vec_extract_sig,
16652 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16653 functions from "vector int" to "vector unsigned int" or from
16654 "vector long long int" to "vector unsigned long long int".
16655 Changed type returned by __builtin_vec_extract_exp,
16656 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16657 functions from "vector int" to "vector unsigned int" or from
16658 "vector long long int" to "vector unsigned long long int".
16659 Changed return type of __builtin_vec_test_data_class,
16660 __builtin_vec_test_data_class_sp, and
16661 __builtin_vec_test_data_class_dp from "vector int" to
16662 "vector bool int" or from "vector long long int" to "vector bool
16663 long long int" and changed second argument type from "unsigned
16664 int" to "int". Added new overloaded function forms "vector float
16665 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16666 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16667 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16668 double, vector unsigned long long int)" and "vector double
16669 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16670 long int)". Changed return type of
16671 __builtin_scalar_test_data_class and
16672 __builtin_scalar_test_data_class_sp and
16673 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16674 int" and changed second argument from "unsigned int" to "int".
16675 Changed type returned by __builtin_scalar_test_neg,
16676 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16677 from "int" to "bool int". Added new overloaded function form
16678 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16679 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16680 exponent double-precision with floating point first argument.
16681 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16682 documentation of scalar_test_data_class, scalar_test_neg,
16683 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16684 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16685 vec_test_data_class built-in functions to reflect refinements in
16686 their type signatures.
16688 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16690 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16692 (aarch64_elf_asm_destructor): Likewise.
16694 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16696 PR rtl-optimization/78634
16697 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16698 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16699 * ifcvt.c (noce_try_cmove): Add missing cost check.
16701 PR rtl-optimization/71724
16702 * combine.c (if_then_else_cond): Look for situations where it is
16703 beneficial to undo the work of one of the recursive calls.
16705 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16707 PR tree-optimization/70754
16708 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16709 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16710 combined stmt before it if not NULL.
16711 (combine_chains): Process refs reversely and compute dominance point
16714 2017-01-23 Martin Liska <mliska@suse.cz>
16716 PR tree-optimization/79196
16717 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16718 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16720 (strlen_optimize_stmt): Call the renamed function.
16722 2017-01-23 Michael Matz <matz@suse.de>
16724 PR tree-optimization/78384
16725 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16727 2017-01-23 Richard Biener <rguenther@suse.de>
16729 PR tree-optimization/79186
16730 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16731 both incoming edges before moving an assert.
16733 2017-01-23 Martin Jambor <mjambor@suse.cz>
16735 * ipa-prop.c (load_from_param_1): Removed.
16736 (load_from_unmodified_param): Bits from load_from_param_1 put back
16738 (load_from_param): Removed.
16739 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16740 with stmt. Reverted back to use of load_from_unmodified_param.
16742 2017-01-23 Martin Jambor <mjambor@suse.cz>
16745 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16746 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16747 field a pointer to garbage collected vector, mark lattices and
16748 ipcp_orig_node with GTY((skip)).
16749 (ipa_get_param_count): Adjust to descriptors being a pointer.
16750 (ipa_get_param): Likewise.
16751 (ipa_get_type): Likewise.
16752 (ipa_get_param_move_cost): Likewise.
16753 (ipa_set_param_used): Likewise.
16754 (ipa_get_controlled_uses): Likewise.
16755 (ipa_set_controlled_uses): Likewise.
16756 (ipa_is_param_used): Likewise.
16757 (ipa_node_params_t): Move into garbage collector. New methods insert
16759 (ipa_node_params_sum): Annotate wth GTY(()).
16760 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16762 (ipa_load_from_parm_agg): Adjust declaration.
16763 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16764 * ipa-profile.c (ipa_profile): Likewise.
16765 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16766 (ipa_populate_param_decls): Make descriptors parameter garbage
16768 (ipa_dump_param): Adjust to descriptors being a pointer.
16769 (ipa_alloc_node_params): Likewise.
16770 (ipa_initialize_node_params): Likewise.
16771 (load_from_param_1): Make descriptors parameter garbage collected.
16772 (load_from_unmodified_param): Likewise.
16773 (load_from_param): Likewise.
16774 (ipa_load_from_parm_agg): Likewise.
16775 (ipa_node_params::~ipa_node_params): Removed.
16776 (ipa_free_all_node_params): Remove call to delete operator.
16777 (ipa_node_params_t::insert): New.
16778 (ipa_node_params_t::remove): Likewise.
16779 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16780 copy known_csts and known_contexts vectors.
16781 (ipa_read_node_info): Adjust to descriptors being a pointer.
16782 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16784 (ipcp_transform_function): Make descriptors variable garbage
16787 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16789 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16790 * config/i386/avx512dqintrin.h: Ditto.
16791 * config/i386/avx512fintrin.h: Ditto.
16792 * config/i386/i386.c: Handle new builtins.
16793 * config/i386/i386-builtin.def: Add new builtins.
16794 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16795 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16797 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16798 Martin Liska <mliska@suse.cz>
16800 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16801 * asan.c (asan_expand_poison_ifn): Support stores and use
16802 appropriate ASAN report function.
16803 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16804 * internal-fn.def (ASAN_POISON_USE): Declare.
16805 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16806 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16807 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16808 ASAN_POISON calls w/o LHS.
16809 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16810 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16811 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16812 * gimplify.c (asan_poison_variables): Add attribute
16813 use_after_scope_memory to variables that really needs to live
16815 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16816 having the attribute.
16818 2017-01-23 Martin Liska <mliska@suse.cz>
16820 * asan.c (create_asan_shadow_var): New function.
16821 (asan_expand_poison_ifn): Likewise.
16822 * asan.h (asan_expand_poison_ifn): New declaration.
16823 * internal-fn.c (expand_ASAN_POISON): Likewise.
16824 * internal-fn.def (ASAN_POISON): New builtin.
16825 * sanopt.c (pass_sanopt::execute): Expand
16826 asan_expand_poison_ifn.
16827 * tree-inline.c (copy_decl_for_dup_finish): Make function
16829 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16830 * tree-ssa.c (is_asan_mark_p): New function.
16831 (execute_update_addresses_taken): Rewrite local variables
16832 (identified just by use-after-scope as addressable) into SSA.
16834 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16836 * doc/install.texi (Specific): opensource.apple.com uses https
16837 now. Remove trailing slash.
16839 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16841 * README.Portability: Remove note on an Irix compatibility issue.
16843 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16845 * gcov.c (INCLUDE_ALGORITHM): Define.
16846 (INCLUDE_VECTOR): Define.
16847 No longer include <vector> and <algorithm> directly.
16849 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16851 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16853 * doc/invoke.texi (Code Gen Options): Ditto.
16855 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16858 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16860 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16862 rtl-optimization/79125
16863 * cprop.c (local_cprop_pass): Handle cases where we make an
16864 unconditional trap.
16866 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16870 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16871 read from, for big endian.
16873 2017-01-20 Jiong Wang <jiong.wang@arm.com>
16875 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
16876 register pauth builtins for LP64 only.
16878 2017-01-20 Marek Polacek <polacek@redhat.com>
16881 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
16884 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16886 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
16888 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
16889 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
16890 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
16892 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16895 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
16896 in big-endian mode when they are not a single duplicated value.
16898 2017-01-20 Richard Biener <rguenther@suse.de>
16900 * BASE-VER: Bump to 7.0.1.
16902 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16904 * omp-low.c (omplow_simd_context): New struct. Use it...
16905 (lower_rec_simd_input_clauses): ...here and...
16906 (lower_rec_input_clauses): ...here to hold common data. Adjust all
16907 references to idx, lane, max_vf, is_simt.
16909 2017-01-20 Graham Markall <graham.markall@embecosm.com>
16911 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
16914 2017-01-20 Martin Jambor <mjambor@suse.cz>
16916 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
16917 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
16919 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
16920 (GTFILES): Rename hsa.c to hsa-common.c.
16921 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
16922 * hsa-dump.c: Likewise.
16923 * hsa-gen.c: Likewise.
16924 * hsa-regalloc.c: Likewise.
16925 * ipa-hsa.c: Likewise.
16926 * omp-expand.c: Likewise.
16927 * omp-low.c: Likewise.
16928 * toplev.c: Likewise.
16930 2017-01-20 Marek Polacek <polacek@redhat.com>
16933 * doc/invoke.texi: Document -Wduplicated-branches.
16934 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
16935 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
16936 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
16937 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
16938 return 0 only when not OEP_LEXICOGRAPHIC.
16939 (fold_build_cleanup_point_expr): Use the expression
16940 location when building CLEANUP_POINT_EXPR.
16941 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
16942 * tree.c (add_expr): Handle error_mark_node.
16944 2017-01-20 Martin Liska <mliska@suse.cz>
16947 * tree-profile.c (init_ic_make_global_vars): Do not call
16949 (gimple_init_gcov_profiler): Likewise.
16951 2017-01-20 Martin Liska <mliska@suse.cz>
16954 * cgraph.h (maybe_create_reference): Remove argument and
16956 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
16958 * ipa-cp.c (create_specialized_node): Likewise.
16959 * symtab.c (symtab_node::maybe_create_reference): Handle
16960 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
16962 2017-01-20 Martin Liska <mliska@suse.cz>
16964 * read-rtl-function.c (function_reader::create_function): Use
16965 build_decl instread of build_decl_stat.
16967 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
16969 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
16970 * config/i386/avx512dqintrin.h: Ditto.
16971 * config/i386/avx512fintrin.h: Ditto.
16972 * config/i386/i386-builtin-types.def: Add new types.
16973 * config/i386/i386.c: Handle new types.
16974 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
16975 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
16976 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
16977 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
16978 (__builtin_ia32_kshiftridi): New.
16979 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
16981 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
16985 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
16986 define to rs6000_init_stack_protect_guard.
16987 (rs6000_init_stack_protect_guard): New function.
16989 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16990 Yunqiang Su <yunqiang.su@imgtec.com>
16992 * config.gcc (supported_defaults): Add madd4.
16993 (with_madd4): Add validation.
16994 (all_defaults): Add madd4.
16995 * config/mips/mips.opt (mmadd4): New option.
16996 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16998 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17000 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17001 (ISA_HAS_FUSED_MADD4): Likewise.
17002 * doc/invoke.texi (-mmadd4): Document the new option.
17003 * doc/install.texi (--with-madd4): Document the new option.
17005 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17007 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17008 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17009 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17010 (aarch64_init_pauth_hint_builtins): New.
17011 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17012 (aarch64_expand_builtin): Expand new builtins.
17014 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17016 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17017 * combine-stack-adj.c (no_unhandled_cfa): Handle
17018 REG_CFA_TOGGLE_RA_MANGLE.
17019 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17020 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17021 info for return address signing.
17022 (aarch64_expand_epilogue): Likewise.
17024 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17026 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17027 * config/aarch64/aarch64-protos.h
17028 (aarch64_return_address_signing_enabled): New declaration.
17029 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17031 (aarch64_expand_prologue): Sign return address before it's pushed onto
17033 (aarch64_expand_epilogue): Authenticate return address fetched from
17035 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17036 (aarch64_attributes): New function attributes for "sign-return-address".
17037 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17038 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17039 ("*do_return"): Generate combined instructions according to key index.
17040 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17041 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17043 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17044 * config/aarch64/aarch64.opt (msign-return-address=): New.
17045 * doc/extend.texi (AArch64 Function Attributes): Documents
17046 "sign-return-address=".
17047 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17049 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17051 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17052 overall option summary.
17054 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17056 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17057 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17058 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17059 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17061 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17063 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17064 -mpower9-minmax by default for -mcpu=power9.
17065 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17066 128-bit floating point.
17068 2017-01-20 Alan Modra <amodra@gmail.com>
17070 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17071 optimizing for size.
17073 2017-01-20 Alan Modra <amodra@gmail.com>
17076 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17077 for strcmp and strncmp from corresponding builtin decl.
17079 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17081 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17082 instead of i386/rtems-64.h.
17083 * config/i386/rtems-64.h: Remove.
17085 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17089 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17091 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17093 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17095 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17096 Change int to HOST_WIDE_INT.
17097 * config/aarch64/aarch64-protos.h
17098 (aarch64_simd_gen_const_vector_dup): Likewise.
17099 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17101 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17103 * langhooks-def.h (lhd_type_for_size): New decl.
17104 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17105 * langhooks.c (lhd_type_for_size): New function, taken from
17108 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17110 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17111 define_bypass for CR latency.
17112 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17113 (power9-alu2): Add define_bypass for CR latency.
17115 (power9-mul): Update insn latency.
17116 (power9-mul-compare): Update insn latency, bypass latency and remove
17119 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17121 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17123 * config/aarch64/aarch64.md
17124 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17125 aarch64_nopcrelative_literal_loads.
17126 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17128 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17130 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17131 TARGET_LOONGSON_3A.
17132 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17134 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17137 * config.gcc (supported_defaults): Add lxc1-sxc1.
17138 (with_lxc1_sxc1): Add validation.
17139 (all_defaults): Add lxc1-sxc1.
17140 * config/mips/mips.opt (mlxc1-sxc1): New option.
17141 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17143 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17144 __mips_no_lxc1_sxc1.
17145 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17146 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17147 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17149 2017-01-19 Richard Biener <rguenther@suse.de>
17151 PR tree-optimization/72488
17152 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17153 sure to restore SSA info.
17154 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17156 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17158 PR rtl-optimization/79121
17159 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17160 of the inner type when shifting an extended value.
17162 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17165 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17166 interposable aliases.
17168 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17171 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17172 Use the evmergelohi instruction.
17173 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17174 (mov_sitf_e500_subreg8_2_be): Likewise.
17175 (mov_sitf_e500_subreg12_2_le): Likewise.
17176 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17177 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17178 (mov_sitf_e500_subreg8_2_le): Likewise.
17179 (mov_sitf_e500_subreg12_2_be): Likewise.
17181 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17183 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17184 attribute from vecsimple to vecperm.
17185 (altivec_vbpermq2): Likewise.
17187 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17190 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17192 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17193 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17194 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17195 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17196 case where N arg is SIZE_MAX.
17197 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17198 (cmpstrsi): Add pattern.
17200 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17202 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17203 __builtin_vec_revb builtins.
17204 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17205 built-in functions to support generation of the ISA 3.0 XXBR<x>
17206 vector byte reverse instructions.
17207 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17208 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17209 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17210 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17211 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17212 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17213 (P9V_BUILTIN_VEC_REVB): Likewise.
17214 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17215 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17216 (p9_xxbrq_v16qi): Likewise.
17217 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17218 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17219 (p9_xxbrh_v8hi): Likewise.
17220 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17221 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17222 vec_revb built-in functions.
17224 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17226 PR rtl-optimization/78952
17227 * config/i386/i386.md (any_extract): New code iterator.
17228 (*insvqi_2): Use any_extract for source operand.
17229 (*insvqi_3): Use any_shiftrt for source operand.
17231 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17233 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17235 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17237 2017-01-18 Matthias Klose <doko@ubuntu.com>
17239 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17241 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17243 * config/rs6000/altivec.h (vec_bperm): Change #define.
17244 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17245 (altivec_vbpermq2): New define_insn.
17246 (altivec_vbpermd): Likewise.
17247 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17248 function interface.
17249 (VBPERMD): Likewise.
17250 (VBPERM): New polymorphic function interface.
17251 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17252 Add entries for P9V_BUILTIN_VEC_VBPERM.
17253 * doc/extend.texi: Add interfaces for vec_bperm.
17255 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17257 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17258 first letter of error messages.
17259 (s390_resolve_overloaded_builtin): Likewise.
17260 * config/s390/s390.c (s390_expand_builtin): Likewise.
17261 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17262 (s390_valid_target_attribute_inner_p): Likewise.
17263 * config/s390/s390.md ("tabort"): Likewise.
17265 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17267 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17268 (ISA_AVOID_DIV_HILO): New macro.
17269 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17270 (ISA_HAS_DDIV): Likewise.
17272 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17274 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17276 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17278 * doc/invoke.texi (fabi-version): Spelling fix.
17280 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17283 * doc/invoke.texi (fabi-version): Mention mangling fix for
17286 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17289 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17291 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17294 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17297 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17299 (stack_protect_test): Ditto.
17300 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17301 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17303 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17304 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17305 -mstack-protector-guard-offset=.
17306 (RS/6000 and PowerPC Options): Ditto.
17308 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17310 * config/i386/i386.h (MASK_CLASS_P): New define.
17311 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17312 there are no registers from different register sets also when
17313 mask registers are used. Update function comment.
17314 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17315 to (*k/*r) and (*k/*km) alternatives.
17317 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17319 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17320 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17321 (EH_RETURN_HANDLER_RTX): New define.
17322 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17323 Force frame pointer in EH return functions.
17324 (aarch64_expand_epilogue): Add barrier for eh_return.
17325 (aarch64_final_eh_return_addr): Remove.
17326 (aarch64_eh_return_handler_rtx): New function.
17327 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17329 (aarch64_eh_return_handler_rtx): New prototype.
17331 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17333 * config/rs6000/altivec.h (vec_rlmi): New #define.
17334 (vec_vrlnm): Likewise.
17335 (vec_rlnm): Likewise.
17336 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17337 (UNSPEC_VRLNM): Likewise.
17338 (VIlong): New mode iterator.
17339 (altivec_vrl<VI_char>mi): New define_insn.
17340 (altivec_vrl<VI_char>nm): Likewise.
17341 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17343 (VRLDNM): Likewise.
17344 (RLNM): New polymorphic function entry.
17345 (VRLWMI): New monomorphic function entry.
17346 (VRLDMI): Likewise.
17347 (RLMI): New polymorphic function entry.
17348 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17349 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17350 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17353 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17356 * dwarf2out.c (field_byte_offset): Restore the
17357 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17358 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17359 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17362 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17365 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17367 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17370 * dwarf2out.c (add_data_member_location_attribute): For constant
17371 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17372 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17373 DW_AT_byte_size attributes.
17375 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17377 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17378 after forcing to constant memory when the code model is medium.
17380 2017-01-17 Julia Koval <julia.koval@intel.com>
17383 * config/i386/avx512fintrin.h
17384 (_mm512_i32gather_ps): Change __addr type to void const*.
17385 (_mm512_mask_i32gather_ps): Ditto.
17386 (_mm512_i32gather_pd): Ditto.
17387 (_mm512_mask_i32gather_pd): Ditto.
17388 (_mm512_i64gather_ps): Ditto.
17389 (_mm512_mask_i64gather_ps): Ditto.
17390 (_mm512_i64gather_pd): Ditto.
17391 (_mm512_mask_i64gather_pd): Ditto.
17392 (_mm512_i32gather_epi32): Ditto.
17393 (_mm512_mask_i32gather_epi32): Ditto.
17394 (_mm512_i32gather_epi64): Ditto.
17395 (_mm512_mask_i32gather_epi64): Ditto.
17396 (_mm512_i64gather_epi32): Ditto.
17397 (_mm512_mask_i64gather_epi32): Ditto.
17398 (_mm512_i64gather_epi64): Ditto.
17399 (_mm512_mask_i64gather_epi64): Ditto.
17400 (_mm512_i32scatter_ps): Change __addr type to void*.
17401 (_mm512_mask_i32scatter_ps): Ditto.
17402 (_mm512_i32scatter_pd): Ditto.
17403 (_mm512_mask_i32scatter_pd): Ditto.
17404 (_mm512_i64scatter_ps): Ditto.
17405 (_mm512_mask_i64scatter_ps): Ditto.
17406 (_mm512_i64scatter_pd): Ditto.
17407 (_mm512_mask_i64scatter_pd): Ditto.
17408 (_mm512_i32scatter_epi32): Ditto.
17409 (_mm512_mask_i32scatter_epi32): Ditto.
17410 (_mm512_i32scatter_epi64): Ditto.
17411 (_mm512_mask_i32scatter_epi64): Ditto.
17412 (_mm512_i64scatter_epi32): Ditto.
17413 (_mm512_mask_i64scatter_epi32): Ditto.
17414 (_mm512_i64scatter_epi64): Ditto.
17415 (_mm512_mask_i64scatter_epi64): Ditto.
17416 * config/i386/avx512pfintrin.h
17417 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17418 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17419 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17420 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17421 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17422 (_mm512_prefetch_i32scatter_ps): Ditto.
17423 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17424 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17425 (_mm512_prefetch_i64scatter_pd): Ditto.
17426 (_mm512_prefetch_i64scatter_ps): Ditto.
17427 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17428 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17429 * config/i386/avx512vlintrin.h
17430 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17431 (_mm_mmask_i32gather_ps): Ditto.
17432 (_mm256_mmask_i32gather_pd): Ditto.
17433 (_mm_mmask_i32gather_pd): Ditto.
17434 (_mm256_mmask_i64gather_ps): Ditto.
17435 (_mm_mmask_i64gather_ps): Ditto.
17436 (_mm256_mmask_i64gather_pd): Ditto.
17437 (_mm_mmask_i64gather_pd): Ditto.
17438 (_mm256_mmask_i32gather_epi32): Ditto.
17439 (_mm_mmask_i32gather_epi32): Ditto.
17440 (_mm256_mmask_i32gather_epi64): Ditto.
17441 (_mm_mmask_i32gather_epi64): Ditto.
17442 (_mm256_mmask_i64gather_epi32): Ditto.
17443 (_mm_mmask_i64gather_epi32): Ditto.
17444 (_mm256_mmask_i64gather_epi64): Ditto.
17445 (_mm_mmask_i64gather_epi64): Ditto.
17446 (_mm256_i32scatter_ps): Change __addr type to void*.
17447 (_mm256_mask_i32scatter_ps): Ditto.
17448 (_mm_i32scatter_ps): Ditto.
17449 (_mm_mask_i32scatter_ps): Ditto.
17450 (_mm256_i32scatter_pd): Ditto.
17451 (_mm256_mask_i32scatter_pd): Ditto.
17452 (_mm_i32scatter_pd): Ditto.
17453 (_mm_mask_i32scatter_pd): Ditto.
17454 (_mm256_i64scatter_ps): Ditto.
17455 (_mm256_mask_i64scatter_ps): Ditto.
17456 (_mm_i64scatter_ps): Ditto.
17457 (_mm_mask_i64scatter_ps): Ditto.
17458 (_mm256_i64scatter_pd): Ditto.
17459 (_mm256_mask_i64scatter_pd): Ditto.
17460 (_mm_i64scatter_pd): Ditto.
17461 (_mm_mask_i64scatter_pd): Ditto.
17462 (_mm256_i32scatter_epi32): Ditto.
17463 (_mm256_mask_i32scatter_epi32): Ditto.
17464 (_mm_i32scatter_epi32): Ditto.
17465 (_mm_mask_i32scatter_epi32): Ditto.
17466 (_mm256_i32scatter_epi64): Ditto.
17467 (_mm256_mask_i32scatter_epi64): Ditto.
17468 (_mm_i32scatter_epi64): Ditto.
17469 (_mm_mask_i32scatter_epi64): Ditto.
17470 (_mm256_i64scatter_epi32): Ditto.
17471 (_mm256_mask_i64scatter_epi32): Ditto.
17472 (_mm_i64scatter_epi32): Ditto.
17473 (_mm_mask_i64scatter_epi32): Ditto.
17474 (_mm256_i64scatter_epi64): Ditto.
17475 (_mm256_mask_i64scatter_epi64): Ditto.
17476 (_mm_i64scatter_epi64): Ditto.
17477 (_mm_mask_i64scatter_epi64): Ditto.
17478 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17479 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17480 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17481 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17482 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17483 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17484 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17485 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17486 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17487 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17488 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17489 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17490 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17491 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17492 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17493 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17494 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17495 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17496 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17497 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17498 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17499 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17500 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17501 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17502 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17503 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17504 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17505 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17506 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17507 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17508 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17509 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17510 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17511 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17512 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17513 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17514 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17515 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17516 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17517 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17518 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17519 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17520 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17521 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17522 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17523 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17524 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17525 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17526 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17527 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17528 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17529 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17530 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17531 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17532 definitions accordingly.
17534 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17535 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17538 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17541 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17544 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17545 hard regs for inner regno.
17547 2017-01-17 Martin Liska <mliska@suse.cz>
17550 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17551 assumption and add comment.
17553 2017-01-17 Nathan Sidwell <nathan@acm.org>
17555 * ipa-visibility.c (localize_node): New function, broken out of ...
17556 (function_and_variable_visibility): ... here. Call it.
17558 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17560 PR middle-end/77445
17561 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17562 correctly set frequency of oudgoing edge.
17563 (duplicate_thread_path): Fix profile updating.
17565 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17568 * configure.ac: Add GCC_BASE_VER.
17569 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17570 version from BASE-VER file.
17571 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17572 (gcc.o): Depend on $(BASEVER).
17573 * common.opt (dumpfullversion): New option.
17574 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17575 * doc/invoke.texi: Document -dumpfullversion.
17576 * doc/install.texi: Document --with-gcc-major-version-only.
17577 * configure: Regenerated.
17579 2017-01-17 Richard Biener <rguenther@suse.de>
17581 PR tree-optimization/71433
17582 * tree-vrp.c (register_new_assert_for): Merge same asserts
17583 on all incoming edges.
17584 (process_assert_insertions_for): Handle insertions at the
17587 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17589 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17590 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17592 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17595 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17598 2017-01-17 Alan Modra <amodra@gmail.com>
17601 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17602 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17603 symbolic stack limit when pic.
17605 2017-01-16 Martin Sebor <msebor@redhat.com>
17607 PR tree-optimization/78608
17608 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17610 2017-01-16 Jeff Law <law@redhat.com>
17613 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17614 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17615 for several include directories that may be relative to sysroot.
17616 * config/i386/x-mingw32 (gplus_includedir): Define.
17617 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17618 (native_system_includedir): Likewise.
17619 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17620 override if TARGET_SYSTEM_ROOT is defined.
17621 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17623 PR tree-optimization/79090
17624 PR tree-optimization/33562
17625 PR tree-optimization/61912
17626 PR tree-optimization/77485
17627 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17628 and computed trims into the dump file.
17630 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17632 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17634 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17637 * gimplify.c (gimplify_init_constructor): If want_value and
17638 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17642 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17643 sequence. Formatting fixes.
17644 (doloop_optimize): Formatting fixes.
17647 * gcc.c (debug_level_greater_than_spec_func): New function.
17648 (static_spec_functions): Add debug-level-gt spec function.
17649 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17651 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17652 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17653 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17654 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17655 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17656 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17658 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17660 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17661 QImode fixups to general and mask registers only.
17663 2017-01-16 Carl Love <cel@us.ibm.com>
17665 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17666 for built-in functions
17667 vector signed char vec_nabs (vector signed char)
17668 vector signed short vec_nabs (vector signed short)
17669 vector signed int vec_nabs (vector signed int)
17670 vector signed long long vec_nabs (vector signed long long)
17671 vector float vec_nabs (vector float)
17672 vector double vec_nabs (vector double)
17673 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17675 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17676 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17677 * doc/extend.texi: Update the documentation file for the new built-in
17680 2017-01-16 Martin Sebor <msebor@redhat.com>
17682 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17685 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17687 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17688 UNSPEC_VSX__XXSPLTD to require special splat handling.
17690 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17693 * system.h: Poison strndup.
17695 2017-01-16 Alan Modra <amodra@gmail.com>
17698 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17701 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17703 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17705 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17707 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17708 call recog here. Assert that INSN_CODE (insn) is non-negative.
17710 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17713 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17715 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17716 in the currently scheduled RTL fragment.
17718 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17720 PR rtl-optimization/78751
17721 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17724 2017-01-14 Jeff Law <law@redhat.com>
17726 PR tree-optimization/79090
17727 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17728 variable length stores.
17729 (compute_trims): Delete dead assignment to *trim_tail.
17730 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17733 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17735 PR rtl-optimization/78626
17736 PR rtl-optimization/78727
17737 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17738 of a block, and split such blocks after everything else is finished.
17740 2017-01-14 Alan Modra <amodra@gmail.com>
17743 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17744 target legitimate_combined_insn.
17745 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17746 (rs6000_legitimate_combined_insn): New function.
17747 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17749 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17750 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17751 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17753 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17755 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17757 2017-01-13 Jeff Law <law@redhat.com>
17759 PR tree-optimization/33562
17760 PR tree-optimization/61912
17761 PR tree-optimization/77485
17762 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17764 (delete_dead_assignment): Likewise.
17765 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17766 statement to delete_dead_call and delete_dead_assignment.
17768 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17771 * substring-locations.c (format_warning_va): Strengthen case 1 so
17772 that both endpoints of the substring must be within the format
17773 range for just the substring to be printed.
17775 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17777 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17778 * config/i386/i386.c (ix86_target_string): Add missing options
17779 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17780 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17781 flags_other and ix86_target_other to flags2_other. Display unknown
17783 (ix86_valid_target_attribute_inner_p): Add missing options and
17784 reorder options by implied ISAs, as in ix86_target_string.
17786 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17788 * hash-table.h (hash_table::too_empty_p): New function.
17789 (hash_table::expand): Use it.
17790 (hash_table::traverse): Likewise.
17791 (hash_table::empty_slot): Use sizeof (value_type) instead of
17792 sizeof (PTR) to convert bytes to elements. Shrink the table
17793 if the current size is excessive for the current number of
17796 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17798 * ira-costs.c (record_reg_classes): Break from the inner loop
17799 early once alt_fail is known to be true. Update outer loop
17800 handling accordingly.
17802 2017-01-13 Jeff Law <law@redhat.com>
17804 * tree-ssa-dse.c (decrement_count): New function.
17805 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17806 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17807 when we know the partially dead statement is a mem* function.
17809 PR tree-optimization/61912
17810 PR tree-optimization/77485
17811 * tree-ssa-dse.c: Include expr.h.
17812 (maybe_trim_constructor_store): New function.
17813 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17815 PR tree-optimization/33562
17816 PR tree-optimization/61912
17817 PR tree-optimization/77485
17818 * doc/invoke.texi: Document new dse-max-object-size param.
17819 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17820 * tree-ssa-dse.c: Include params.h.
17821 (dse_store_status): New enum.
17822 (initialize_ao_ref_for_dse): New, partially extracted from
17824 (valid_ao_ref_for_dse, normalize_ref): New.
17825 (setup_live_bytes_from_ref, compute_trims): Likewise.
17826 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17827 (maybe_trim_partially_dead_store): Likewise.
17828 (maybe_trim_complex_store): Likewise.
17829 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17830 Track what bytes live from the original store. Return tri-state
17831 for dead, partially dead or live.
17832 (dse_dom_walker): Add constructor, destructor and new private members.
17833 (delete_dead_call, delete_dead_assignment): New extracted from
17835 (dse_optimize_stmt): Make a member of dse_dom_walker.
17836 Use initialize_ao_ref_for_dse.
17838 PR tree-optimization/33562
17839 PR tree-optimization/61912
17840 PR tree-optimization/77485
17841 * sbitmap.h (bitmap_count_bits): Prototype.
17842 (bitmap_clear_range, bitmap_set_range): Likewise.
17843 * sbitmap.c (bitmap_clear_range): New function.
17844 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17846 2017-01-13 Martin Liska <mliska@suse.cz>
17849 * function.c (set_cfun): Add new argument force.
17850 * function.h (set_cfun): Likewise.
17851 * ipa-inline-transform.c (inline_call): Use the function when
17852 strict alising from is dropped for function we inline to.
17854 2017-01-13 Richard Biener <rguenther@suse.de>
17856 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17857 for dumping GIMPLE INTEGER_CSTs.
17859 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17861 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17862 to 201112L since C++17.
17864 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17867 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17868 if -fsanitize=kernel-address is present.
17870 2017-01-13 Richard Biener <rguenther@suse.de>
17872 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
17873 as _Literal ( type ) number in case usual suffixes do not
17874 preserve all information.
17876 2017-01-13 Richard Biener <rguenther@suse.de>
17878 PR tree-optimization/77283
17879 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
17880 and ssa-iterators.h.
17881 (is_feasible_trace): Implement a cost model based on joiner
17884 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17887 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
17888 char or short to __float128/_Float128 directly.
17890 2017-01-12 Martin Sebor <msebor@redhat.com>
17892 to -Wformat-overflow.
17893 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
17894 (min_bytes_remaining): Same.
17895 (get_string_length): Same.
17896 (format_string): Same.
17897 (format_directive): Same.
17899 (pass_sprintf_length::handle_gimple_call): Same.
17901 2017-01-12 Jakub Jelinek <jakub@redhat.com>
17903 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
17904 info.nowrite calls with no lhs that can't throw. Return bool
17905 whether gsi_remove has been called or not.
17906 (pass_sprintf_length::handle_gimple_call): Return bool whether
17907 try_substitute_return_value called gsi_remove. Formatting fix.
17908 (pass_sprintf_length::execute): Don't use gsi_remove if
17909 handle_gimple_call returned true.
17912 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
17913 be removed due to side-effects, don't remove following barrier nor
17914 turn the successor edge into fallthru edge.
17916 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17919 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
17920 element-reversing loads and stores as not swappable.
17922 2017-01-12 Nathan Sidwell <nathan@acm.org>
17923 Nicolai Stange <nicstange@gmail.com>
17925 * combine.c (try_combine): Don't ignore result of overlap checking
17926 loop. Combine overlap & asm check into single loop.
17928 2017-01-12 Richard Biener <rguenther@suse.de>
17930 * tree-pretty-print.c (dump_generic_node): Provide -gimple
17931 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
17933 2017-01-12 Richard Biener <rguenther@suse.de>
17935 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
17936 and TS_TARGET_OPTION directly derive from TS_BASE.
17937 * tree-core.h (tree_optimization_option): Derive from tree_base.
17938 (tree_target_option): Likewise.
17940 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
17942 * config/i386/i386.c (memory_address_length): Increase len
17943 only when rip_relative_addr_p returns false.
17945 2017-01-11 Julia Koval <julia.koval@intel.com>
17947 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
17948 (OPTION_MASK_ISA_SGX_SET): New.
17949 (ix86_handle_option): Handle OPT_msgx.
17950 * config.gcc: Added sgxintrin.h.
17951 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
17952 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
17953 * config/i386/i386.c (ix86_target_string): Add -msgx.
17955 (ix86_option_override_internal): Handle new options.
17956 (ix86_valid_target_attribute_inner_p): Add sgx.
17957 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
17958 * config/i386/i386.opt: Add msgx.
17959 * config/i386/sgxintrin.h: New file.
17960 * config/i386/x86intrin.h: Add sgxintrin.h.
17962 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17965 * fold-const.c (maybe_nonzero_address): Return 1 for function
17967 (tree_single_nonzero_warnv_p): Don't handle function local objects
17971 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
17974 2017-01-11 David Malcolm <dmalcolm@redhat.com>
17977 * opts.c: Include "spellcheck.h"
17978 (struct string_fragment): New struct.
17979 (struct edit_distance_traits<const string_fragment &>): New
17981 (get_closest_sanitizer_option): New function.
17982 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
17984 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17986 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
17988 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
17989 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
17990 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
17991 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
17992 for initial die_offset if dwarf_split_debug_info.
17993 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
17994 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
17996 (output_skeleton_debug_sections): Formatting fix. Use
17997 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
17998 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18000 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18002 * config/arm/cortex-a53.md: Add bypasses for
18003 cortex_a53_r2f_cvt.
18004 (cortex_a53_r2f): Only use for transfers.
18005 (cortex_a53_f2r): Likewise.
18006 (cortex_a53_r2f_cvt): Add reservation for conversions.
18007 (cortex_a53_f2r_cvt): Likewise.
18009 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18011 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18012 to all inlined functions, change static to extern.
18014 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18017 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18019 (arm_assemble_integer): Likewise.
18021 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18023 * config.gcc: Use new awk script to check CPU, FPU and architecture
18024 parameters for --with-... options.
18025 * config/arm/parsecpu.awk: New file
18026 * config/arm/arm-cpus.in: New file.
18027 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18029 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18031 * config/arm/t-arm: Update dependency rules.
18032 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18033 of processing .def files.
18034 * config/arm/genopt.sh: Deleted.
18035 * config/arm/gentune.sh: Deleted.
18036 * config/arm/arm-cores.def: Deleted.
18037 * config/arm/arm-arches.def: Deleted.
18038 * config/arm/arm-fpus.def: Deleted.
18039 * config/arm/arm-tune.md: Regenerated.
18040 * config/arm/arm-tables.opt: Regenerated.
18041 * config/arm/arm-cpu.h: New generated file.
18042 * config/arm/arm-cpu-data.h: New generated file.
18043 * config/arm/arm-cpu-cdata.h: New generated file.
18045 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18048 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18050 (input_varpool_node): Unpack dynamically_initialized bit.
18052 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18054 PR rtl-optimization/79032
18055 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18056 the alignment of the adjusted memory reference against that of MODE,
18057 instead of the alignment of the original memory reference.
18059 2017-01-11 Martin Jambor <mjambor@suse.cz>
18061 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18063 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18064 decorated functions.
18066 2017-01-11 Richard Biener <rguenther@suse.de>
18068 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18069 set range/nonnull info for PHI results. Do not set it on
18070 stmts marked for removal.
18072 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18074 * expr.c (store_field): In the bitfield case, fetch the return value
18075 from the registers before applying a single big-endian adjustment.
18076 Always do a final load for a BLKmode value not larger than a word.
18078 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18081 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18082 that we correctly handle column numbers greater than
18083 LINE_MAP_MAX_COLUMN_NUMBER.
18085 2017-01-10 Martin Sebor <msebor@redhat.com>
18087 PR middle-end/78245
18088 * gimple-ssa-sprintf.c (get_destination_size): Call
18089 {init,fini}object_sizes.
18090 * tree-object-size.c (addr_object_size): Adjust.
18091 (pass_through_call): Adjust.
18092 (pass_object_sizes::execute): Adjust.
18093 * tree-object-size.h (fini_object_sizes): Declare.
18095 2017-01-10 Martin Sebor <msebor@redhat.com>
18097 PR tree-optimization/78775
18098 * builtins.c (get_size_range): Move...
18099 * calls.c: ...to here.
18100 (alloc_max_size): Accept zero argument.
18101 (operand_signed_p): Remove.
18102 (maybe_warn_alloc_args_overflow): Call get_size_range.
18103 * calls.h (get_size_range): Declare.
18105 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18107 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18108 from TI's devices.csv file as of September 2016.
18109 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18111 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18113 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18114 * doc/invoke.texi: Likewise.
18115 * doc/md.texi: Likewise.
18116 * doc/objc.texi: Likewise.
18118 2017-01-10 Joshua Conner <joshconner@google.com>
18120 * config/arm/fuchsia-elf.h: New file.
18121 * config/fuchsia.h: New file.
18122 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18123 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18125 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18127 2016-01-10 Richard Biener <rguenther@suse.de>
18129 PR tree-optimization/79034
18130 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18131 Propagate out degenerate PHIs in the joiner.
18133 2017-01-10 Martin Liska <mliska@suse.cz>
18135 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18136 (sort_congruence_classes_by_decl_uid): Likewise.
18137 (sort_congruence_class_groups_by_decl_uid): Likewise.
18138 (sem_item_optimizer::merge_classes): Sort class, groups in these
18139 classes and members in the groups by DECL_UID of declarations.
18140 This would make merge operations stable.
18142 2017-01-10 Martin Liska <mliska@suse.cz>
18144 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18145 usage of m_classes_vec.
18146 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18147 (sem_item_optimizer::get_group_by_hash): Likewise.
18148 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18149 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18150 (sem_item_optimizer::verify_classes): Likewise.
18151 (sem_item_optimizer::process_cong_reduction): Likewise.
18152 (sem_item_optimizer::dump_cong_classes): Likewise.
18153 (sem_item_optimizer::merge_classes): Likewise.
18154 * ipa-icf.h (congruence_class_hash): Rename from
18155 congruence_class_group_hash. Remove declaration of m_classes_vec.
18157 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18159 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18160 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18161 * config.gcc: Add avx512vpopcntdqintrin.h.
18162 * config/i386/avx512vpopcntdqintrin.h: New.
18163 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18164 * config/i386/i386-builtin-types.def: Add new types.
18165 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18166 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18167 __builtin_ia32_vpopcountq_v8di_mask): New.
18168 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18169 __AVX512VPOPCNTDQ__.
18170 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18171 (PTA_AVX512VPOPCNTDQ): Define.
18172 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18173 TARGET_AVX512VPOPCNTDQ_P): Define.
18174 * config/i386/i386.opt: Add mavx512vpopcntdq.
18175 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18176 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18178 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18180 PR middle-end/77484
18181 * predict.def (PRED_CALL): Set to 67.
18183 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18185 * expr.c (store_field): In the bitfield case, if the value comes from
18186 a function call and is of an aggregate type returned in registers, do
18187 not modify the field mode; extract the value in all cases if the mode
18188 is BLKmode and the size is not larger than a word.
18190 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18193 * config/i386/cpuid.h: Fix undefined behavior.
18195 2017-01-04 Jeff Law <law@redhat.com>
18197 PR tree-optimization/79007
18198 PR tree-optimization/67955
18199 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18200 conservative for pt.null when flag_non_call_exceptions is on.
18202 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18204 PR translation/79019
18205 PR translation/79020
18206 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18207 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18208 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18210 * config/avr/avr.opt (maccumulate-args): Likewise.
18211 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18212 * common.opt (freport-bug): Likewise.
18213 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18214 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18215 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18216 translatable string.
18217 * config/i386/i386.c (function_value_32): Likewise.
18218 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18219 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18221 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18222 * common/config/msp430/msp430-common.c (msp430_handle_option):
18224 * symtab.c (symtab_node::verify_base): Likewise.
18225 * opts.c (set_debug_level): Likewise.
18226 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18227 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18228 missing whitespace to translatable strings.
18229 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18230 * config/sh/superh.h: Likewise.
18231 * config/i386/xopintrin.h: Likewise.
18232 * config/i386/znver1.md: Likewise.
18233 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18234 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18235 * double-int.h (struct double_int): Likewise.
18236 * double-int.c (div_and_round_double): Likewise.
18237 * wide-int.cc: Likewise.
18238 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18239 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18240 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18241 (crossjumps_occurred): ... this.
18242 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18245 PR tree-optimization/78899
18246 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18247 returning bool return struct loop *, NULL for failure and the new
18249 (versionable_outer_loop_p): Don't version outer loop if it has
18250 dont_vectorized bit set.
18251 (tree_if_conversion): When versioning outer loop, ensure
18252 tree_if_conversion is performed also on the inner loop of the
18253 non-vectorizable outer loop copy.
18254 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18255 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18256 prevent vectorization of it.
18257 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18258 the outer loop vectorization of the non-scalar version is attempted
18259 before vectorization of the inner loop in scalar version. If
18260 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18261 vectorization of its inner loop.
18262 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18263 has 2 inner loops, rename also on edges from bb whose single pred
18264 is outer_loop->header. Fix typo in function comment.
18266 2017-01-09 Martin Sebor <msebor@redhat.com>
18269 * asan.c (asan_emit_stack_protection): Increase local buffer size
18270 to avoid snprintf truncation warning.
18272 2017-01-09 Andrew Pinski <apinski@cavium.com>
18274 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18275 to reference thunderx2t99 for the tuning structure
18276 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18278 (thunderx2t99_extra_costs): This.
18279 * config/aarch64/aarch64-tune.md: Regenerate.
18280 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18281 (vulcan_addrcost_table): This.
18282 (vulcan_regmove_cost): Rename to ...
18283 (thunderx2t99_regmove_cost): This.
18284 (vulcan_vector_cost): Rename to ...
18285 (thunderx2t99_vector_cost): this.
18286 (vulcan_branch_cost): Rename to ...
18287 (thunderx2t99_branch_cost): This.
18288 (vulcan_tunings): Rename to ...
18289 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18290 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18292 2017-01-09 Martin Jambor <mjambor@suse.cz>
18296 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18297 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18298 (propagate_vr_accross_jump_function): Use the above function for all
18299 value range computations for pass-through jump functions and type
18300 converasion from explicit value range values.
18301 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18302 parameters from TYPE_ARG_TYPES.
18303 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18304 (ipa_write_node_info): Stream type of the actual argument.
18305 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18307 2017-01-09 Martin Liska <mliska@suse.cz>
18310 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18311 (lookup_compiler): Do not show error message with have_E.
18313 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18315 PR tree-optimization/78938
18316 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18317 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18318 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18319 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18322 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18324 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18327 2017-01-09 Richard Biener <rguenther@suse.de>
18329 PR tree-optimization/78997
18330 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18331 name condition properly.
18333 2017-01-09 Richard Biener <rguenther@suse.de>
18336 * dwarf2out.c (is_cxx): New overload with context.
18337 (is_naming_typedef_decl): Use it.
18339 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18341 * invoke.texi (Option Summary): Correct spacing in option lists
18342 and add line breaks to fix over-long lines.
18344 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18346 PR middle-end/17660
18348 * extend.texi (Common Variable Attributes): Add xref to GCC
18349 Internals manual to explain mode attribute keywords.
18351 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18354 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18355 and Preprocessor Options.
18356 (Options for Linking): Document -pthread here....
18357 (RS/6000 and PowerPC Options): ...not here.
18358 (Solaris 2 Options): ...or here.
18359 * doc/cppopts.texi: Document -pthread.
18361 2017-01-08 Martin Sebor <msebor@redhat.com>
18363 PR middle-end/77708
18364 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18365 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18366 New member functions.
18367 (format_directive): Used them.
18369 (pass_sprintf_length::handle_gimple_call): Same.
18370 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18371 to avoid truncation for any argument.
18372 (extract_affine_mul): Same.
18373 * tree.c (get_file_function_name): Same.
18375 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18377 PR middle-end/77484
18378 * predict.def (PRED_INDIR_CALL): Set to 86.
18380 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18382 PR preprocessor/54124
18383 * doc/cppopts.texi: Reformat -d subtable to list the full name
18384 of the options. Add cross-reference to the docs for the general
18385 compiler -d options.
18386 * doc/invoke.texi (Developer Options): Add cross-reference to the
18387 preprocessor-specific -d option documentation.
18389 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18391 PR preprocessor/13498
18392 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18393 redudant material, and reflect new command-line options.
18394 (System Headers): Likewise.
18396 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18398 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18399 -isystem, and -idirafter. Copy-edit.
18400 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18401 default for -ftrack-macro-expansion. Delete obsolete and
18402 badly-formatted implementation details about -fdebug-cpp output.
18403 * doc/cppwarnopts.texi: Copy-edit.
18405 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18408 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18409 that the transition from a max line width >= 1<<10 to narrower
18410 lines works correctly.
18412 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18414 * doc/options.texi (PerFunction): New.
18415 * opt-functions.awk (switch_flags): Map both Optimization and
18416 PerFunction to CL_OPTIMIZATION.
18417 * opth-gen.awk: Test for PerFunction flag along with
18419 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18420 it only when the latter is present. Skip those that don't in
18421 the hash function generator.
18422 * common.opt (fvar-tracking): Mark as PerFunction instead of
18424 (fvar-tracking-assignments): Likewise.
18425 (fvar-tracking-assignments-toggle): Likewise.
18426 (fvar-tracking-uninit): Likewise.
18428 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18430 PR translation/79018
18431 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18434 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18437 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18438 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18439 TARGET_LONG_JUMP_TABLE_OFFSETS.
18440 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18441 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18442 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18443 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18444 * config/m68k/m68k.md (tablejump expander): Likewise.
18445 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18446 TARGET_LONG_JUMP_TABLE_OFFSETS.
18447 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18448 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18450 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18451 David Holsgrove <david.holsgrove@xilinx.com>
18453 * common/config/microblaze/microblaze-common.c
18454 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18455 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18457 * config/microblaze/microblaze.c (microblaze_must_save_register)
18458 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18460 (microblaze_eh_return): New function.
18461 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18462 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18463 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18464 * config/microblaze/microblaze.md (eh_return): New pattern.
18466 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18468 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18469 GCC_DIAGNOSTIC_STRINGIFY): Define.
18471 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18473 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18475 * config/arm/arm.md (<mcrr>): New.
18477 * config/arm/arm.c (arm_arch5te): New.
18478 (arm_option_override): Set arm_arch5te.
18479 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18481 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18482 (arm_mcrr_qualifiers): ... this. New.
18483 (MRRC_QUALIFIERS): Define to...
18484 (arm_mrrc_qualifiers): ... this. New.
18485 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18487 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18488 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18489 (MRRCI, mrrc, MRRC): New.
18490 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18491 VUNSPEC_MRRC2): New.
18493 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18495 * config/arm/arm.md (<mcr>): New.
18497 * config/arm/arm.c (arm_coproc_builtin_available): Add
18498 support for mcr, mrc, mcr2 and mrc2.
18499 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18500 (arm_mcr_qualifiers): ... this. New.
18501 (MRC_QUALIFIERS): Define to ...
18502 (arm_mrc_qualifiers): ... this. New.
18503 (MCR_QUALIFIERS): Define to ...
18504 (arm_mcr_qualifiers): ... this. New.
18505 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18507 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18508 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18509 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18510 VUNSPEC_MRC2): New.
18512 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18514 * config/arm/arm.md (*ldc): New.
18518 * config/arm/arm.c (arm_coproc_builtin_available): Add
18519 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18520 (arm_coproc_ldc_stc_legitimate_address): New.
18521 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18522 'qualifier_const_pointer'.
18523 (LDC_QUALIFIERS): Define to...
18524 (arm_ldc_qualifiers): ... this. New.
18525 (STC_QUALIFIERS): Define to...
18526 (arm_stc_qualifiers): ... this. New.
18527 * config/arm/arm-protos.h
18528 (arm_coproc_ldc_stc_legitimate_address): New.
18529 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18530 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18531 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18532 stc2, stcl, stc2l): New.
18533 * config/arm/constraints.md (Uz): New.
18534 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18535 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18536 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18537 VUNSPEC_STC2L): New.
18539 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18541 * config/arm/arm.md (<cdp>): New.
18542 * config/arm/arm.c (neon_const_bounds): Rename this ...
18543 (arm_const_bounds): ... this.
18544 (arm_coproc_builtin_available): New.
18545 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18546 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18547 (CDP_QUALIFIERS): Define to...
18548 (arm_cdp_qualifiers): ... this. New.
18550 (arm_expand_builtin_args): Add case for 6 arguments.
18551 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18552 (arm_const_bounds): ... this.
18553 (arm_coproc_builtin_available): New.
18554 * config/arm/arm_acle.h (__arm_cdp): New.
18556 * config/arm/arm_acle_builtins.def (cdp): New.
18558 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18559 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18560 'arm_const_bounds'.
18561 * config/arm/types.md (coproc): New.
18562 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18563 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18564 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18565 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18567 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18569 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18570 (UBINOP_QUALIFIERS): New.
18572 (acle_builtin_data): New. Change comment.
18573 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18574 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18575 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18576 arm_acle_builtins.def.
18577 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18578 (arm_init_acle_builtins): New.
18579 (CRC32_BUILTIN): Remove.
18580 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18581 crc32cb, crc32ch and crc32cw.
18582 (arm_init_crc32_builtins): Remove.
18583 (arm_init_builtins): Use arm_init_acle_builtins rather
18584 than arm_init_crc32_builtins.
18585 (arm_expand_acle_builtin): New.
18586 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18587 * config/arm/arm_acle_builtins.def: New.
18589 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18591 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18592 (arm_builtin_datum): ... this.
18593 (arm_init_neon_builtin): Rename to ...
18594 (arm_init_builtin): ... this. Add a new parameters PREFIX
18595 and USE_SIG_IN_NAME.
18596 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18597 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18598 'arm_builtin_datum'.
18599 (arm_init_vfp_builtins): Likewise.
18600 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18601 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18602 (arm_expand_neon_args): Rename to ...
18603 (arm_expand_builtin_args): ... this. Rename builtin_arg
18604 enum values and differentiate between ARG_BUILTIN_MEMORY
18605 and ARG_BUILTIN_NEON_MEMORY.
18606 (arm_expand_neon_builtin_1): Rename to ...
18607 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18608 values, arm_expand_builtin_args and add bool parameter NEON.
18609 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18610 (arm_expand_vfp_builtin): Likewise.
18611 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18613 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18615 PR middle-end/77484
18616 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18617 * predict.c (tree_estimate_probability_bb): Reverse direction of
18618 polymorphic call predictor.
18620 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18622 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18623 (determine_pass_name_match): ...this new function and...
18624 (should_skip_pass_p): ...this new function.
18626 2017-01-06 Nathan Sidwell <nathan@acm.org>
18628 * ipa-visibility.c (function_and_variable_visibility): Reformat
18629 comments and long lines. Remove extrneous if.
18630 * symtab.c (symtab_node::make_decl_local): Fix code format.
18631 (symtab_node::set_section_for_node): Fix comment typo.
18633 2017-01-06 Martin Liska <mliska@suse.cz>
18636 * lra-constraints.c: Rename invariant to lra_invariant.
18637 * predict.c (set_even_probabilities): Initialize e to NULL.
18639 2017-01-05 Martin Sebor <msebor@redhat.com>
18641 PR tree-optimization/78910
18642 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18643 (format_integer): Correct off-by-one error in the handling
18644 of precision with negative numbers in signed conversions..
18646 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18648 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18650 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18652 PR tree-optimization/71016
18653 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18654 factor_out_conditional_conversion. Formatting fix.
18655 (factor_out_conditional_conversion): Add cond_stmt argument.
18656 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18657 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18660 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18662 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18663 read-rtl-function.o, and selftest-rtl.o.
18664 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18665 (selftest::aarch64_test_loading_full_dump): New function.
18666 (selftest::aarch64_run_selftests): New function.
18667 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18668 selftest::aarch64_run_selftests.
18669 * config/i386/i386.c
18670 (selftest::ix86_test_loading_dump_fragment_1): New function.
18671 (selftest::ix86_test_loading_call_insn): New function.
18672 (selftest::ix86_test_loading_full_dump): New function.
18673 (selftest::ix86_test_loading_unspec): New function.
18674 (selftest::ix86_run_selftests): Call the new functions.
18675 * emit-rtl.c (maybe_set_max_label_num): New function.
18676 * emit-rtl.h (maybe_set_max_label_num): New decl.
18677 * function.c (instantiate_decls): Guard call to
18678 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18679 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18681 * gensupport.c (gen_reader::gen_reader): Pass "false"
18682 for new "compact" param of rtx_reader.
18683 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18684 rather than an empty string for NULL strings.
18685 * read-md.c: Potentially include config.h rather than bconfig.h.
18686 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18687 (have_error): New global, copied from errors.c.
18688 (md_reader::read_name): Rename to...
18689 (md_reader::read_name_1): ...this, adding "out_loc" param,
18690 and converting "missing name or number" to returning false, rather
18692 (md_reader::read_name): Reimplement in terms of read_name_1.
18693 (md_reader::read_name_or_nil): New function.
18694 (md_reader::read_string): Handle "(nil)" by returning NULL.
18695 (md_reader::md_reader): Add new param "compact".
18696 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18697 (md_reader::read_file): New method.
18698 * read-md.h (md_reader::md_reader): Add new param "compact".
18699 (md_reader::read_file): New method.
18700 (md_reader::is_compact): New accessor.
18701 (md_reader::read_name): Convert return type from void to file_location.
18702 (md_reader::read_name_or_nil): New decl.
18703 (md_reader::read_name_1): New decl.
18704 (md_reader::m_compact): New field.
18705 (noop_reader::noop_reader): Pass "false" for new "compact" param
18707 (rtx_reader::rtx_reader): Add new "compact" param.
18708 (rtx_reader::read_rtx_operand): Make virtual and convert return
18709 type from void to rtx.
18710 (rtx_reader::read_until): New decl.
18711 (rtx_reader::handle_any_trailing_information): New virtual function.
18712 (rtx_reader::postprocess): New virtual function.
18713 (rtx_reader::finalize_string): New virtual function.
18714 (rtx_reader::m_in_call_function_usage): New field.
18715 (rtx_reader::m_reuse_rtx_by_id): New field.
18716 * read-rtl-function.c: New file.
18717 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18718 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18719 (selftest::verify_three_block_rtl_cfg): New decl.
18720 * read-rtl-function.h: New file.
18721 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18722 For host, include function.h, memmodel.h, and emit-rtl.h.
18723 (one_time_initialization): New function.
18724 (struct compact_insn_name): New struct.
18725 (compact_insn_names): New array.
18726 (find_code): Handle insn codes in compact dumps.
18727 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18728 (bind_subst_iter_and_attr): Likewise.
18729 (add_condition_to_string): Likewise.
18730 (add_condition_to_rtx): Likewise.
18731 (apply_attribute_uses): Likewise.
18732 (add_current_iterators): Likewise.
18733 (apply_iterators): Likewise.
18734 (initialize_iterators): Guard usage of apply_subst_iterator with
18735 #ifdef GENERATOR_FILE.
18736 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18737 (md_reader::read_mapping): Likewise.
18738 (add_define_attr_for_define_subst): Likewise.
18739 (add_define_subst_attr): Likewise.
18740 (read_subst_mapping): Likewise.
18741 (check_code_iterator): Likewise.
18742 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18744 (one_time_initialization): New function.
18745 (rtx_reader::read_until): New method.
18746 (read_flags): New function.
18747 (parse_reg_note_name): New function.
18748 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18749 Handle reuse_rtx ids.
18750 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18751 Add parsing support for RTL dumps, mirroring the special-cases in
18752 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18753 values, and calling handle_any_trailing_information.
18754 (rtx_reader::read_rtx_operand): Convert return type from void
18755 to rtx, returning return_rtx. Handle case 'e'. Call
18756 finalize_string on XSTR and XTMPL fields.
18757 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18758 "(nil)" values were omitted. Call the postprocess vfunc on the
18760 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18761 class ctor. Initialize m_in_call_function_usage. Call
18762 one_time_initialization.
18763 * rtl-tests.c (selftest::test_uncond_jump): Call
18764 set_new_first_and_last_insn.
18765 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18766 * selftest-rtl.c: New file.
18767 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18768 (selftest::get_insn_by_uid): New decl.
18769 * selftest-run-tests.c (selftest::run_tests): Call
18770 read_rtl_function_c_tests.
18771 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18772 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18775 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18777 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18778 operands in a special way. Assert that pos+len <= mode precision.
18780 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18782 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18783 3 argument Alias with unlimited for the negative form.
18784 (fno-vect-cost-model): Removed.
18786 2017-01-05 Martin Liska <mliska@suse.cz>
18788 * hsa-gen.c (gen_hsa_divmod): New function.
18789 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18791 2017-01-05 Martin Liska <mliska@suse.cz>
18794 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18797 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18799 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18800 small constant length operands.
18802 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18804 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18805 between loop iterations.
18807 2017-01-05 Martin Liska <mliska@suse.cz>
18810 * gimplify.c (gimplify_decl_expr): Compare to
18811 asan_poisoned_variables instread of checking flags.
18812 (gimplify_target_expr): Likewise.
18813 (gimplify_expr): Likewise.
18814 (gimplify_function_tree): Conditionally initialize
18815 asan_poisoned_variables.
18817 2017-01-04 Jeff Law <law@redhat.com>
18819 PR tree-optimizatin/78812
18820 * rtl.h (contains_mem_rtx_p): Prototype.
18821 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18822 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18823 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18824 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18825 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18827 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18829 * input.c (assert_char_at_range): Default-initialize actual_range.
18831 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18833 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18834 to match the caller.
18836 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18838 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18839 insns after final jump in test to emit dummy move.
18841 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18843 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18844 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18846 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18848 * multiple_target.c (create_dispatcher_calls): Init e_next.
18849 * tree-ssa-loop-split.c (split_loop): Init border.
18850 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18853 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18858 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18859 (altivec_register_operand): Do not return true if the operand
18860 contains a SUBREG mixing SImode and SFmode.
18861 (vsx_register_operand): Likewise.
18862 (vsx_reg_sfsubreg_ok): New predicate.
18863 (vfloat_operand): Do not return true if the operand contains a
18864 SUBREG mixing SImode and SFmode.
18865 (vint_operand): Likewise.
18866 (vlogical_operand): Likewise.
18867 (gpc_reg_operand): Likewise.
18868 (int_reg_operand): Likewise.
18869 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18870 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18871 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
18873 (rs6000_emit_move_si_sf_subreg): New helper function.
18874 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
18875 fixup SUBREGs involving SImode and SFmode.
18876 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
18877 numbers for the new peephole2 optimization.
18878 (peephole2 for SFmode unions): New peephole2 to optimize cases in
18879 the GLIBC math library that do AND/IOR/XOR operations on single
18880 precision floating point.
18881 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
18882 target macros to say whether we need to avoid SUBREGs mixing
18884 (TARGET_ALLOW_SF_SUBREG): Likewise.
18885 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
18886 (UNSPEC_SI_FROM_SF): Likewise.
18887 (iorxor): Change spacing.
18888 (and_ior_xor): New iterator for AND, IOR, and XOR.
18889 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
18890 (movdi_from_sf_zero_ext): Likewise.
18891 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
18892 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
18893 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
18894 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
18895 (fms<mode>4): Likewise.
18896 (fnma<mode>4): Likewise.
18897 (fnms<mode>4): Likewise.
18898 (nfma<mode>4): Likewise.
18899 (nfms<mode>4): Likewise.
18901 2017-01-04 Marek Polacek <polacek@redhat.com>
18904 * doc/invoke.texi: Document -Wpointer-compare.
18906 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18908 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
18911 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
18912 descriptions for -gdwarf-5 and emit them as uleb128 instead of
18915 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
18918 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
18919 documentation of the powerpc_popcntb_ok attribute.
18920 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18921 code to issue warning messages if a requested CPU configuration is
18922 not supported by the binary (assembler and loader) toolchain.
18923 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
18924 made to define a built-in function that has been disabled.
18925 (paired_init_builtins): Add assertion to prevent ICE if attempt is
18926 made to define a built-in function that has been disabled.
18927 (altivec_init_builtins): Add comment explaining why definition
18928 of the DST built-in functions is not preceded by an assertion
18929 check. Add assertions to prevent ICE if attempts are made to
18930 define an altivec predicate or an abs* built-in function that has
18932 (htm_init_builtins): Add comment explaining why definition of the
18933 htm built-in functions is not preceded by an assertion check.
18935 2017-01-04 Jeff Law <law@redhat.com>
18937 PR tree-optimizatin/67955
18938 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
18939 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
18940 the points-to solution does not include pt_null. Use DECL_PT_UID
18943 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
18945 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
18946 Use gen_int_mode instead of gen_lopwart for const_int operands.
18948 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18950 PR tree-optimization/71563
18951 * match.pd: Simplify X << Y into X if Y is known to be 0 or
18952 out of range value - has low bits known to be zero.
18954 2017-01-04 Alan Modra <amodra@gmail.com>
18956 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
18957 * configure: Regenerate.
18958 * config.in: Regenerate.
18960 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18963 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
18964 a substring of the message, but strcmp with the whole message. Ifdef
18965 ENABLE_NLS, translate the message first using dgettext.
18967 2017-01-03 Jeff Law <law@redhat.com>
18969 PR tree-optimizatin/78856
18970 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
18971 (mark_threaded_blocks): Remove code to truncate thread paths that
18972 cross multiple loop headers. Instead invalidate the cached loop
18973 iteration information and handle case of a thread path walking
18974 into an irreducible region.
18976 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
18979 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
18980 assertions. Add support for doing the signbit if the IEEE 128-bit
18981 floating point value is in a GPR.
18982 * config/rs6000/rs6000.md (Fsignbit): Delete.
18983 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
18984 Update the length attribute if the value is in a GPR.
18985 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
18986 the sign or zero extension instruction, since the value is always 0/1.
18987 (signbit<mode>2_dm2): Delete using <Fsignbit>.
18990 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
18991 extracting SImode to a GPR register so that we can generate a
18992 store, limit the vector to be in a traditional Altivec register
18993 for the vextuwrx instruction.
18995 2017-01-03 Ian Lance Taylor <iant@google.com>
18997 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
18999 2017-01-03 Martin Sebor <msebor@redhat.com>
19001 PR tree-optimization/78696
19002 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19003 precision. Use MPFR for %f for greater fidelity. Correct handling
19005 (pass_sprintf_length::compute_format_length): Set width and precision
19006 specified by asrerisk to void_node for vararg functions.
19007 (try_substitute_return_value): Adjust dump output.
19009 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19011 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19013 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19015 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19016 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19017 -mlra/-mno-lra was passed to the compiler.
19019 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19021 PR rtl-optimization/65618
19022 * emit-rtl.c (try_split): Move initialization of "before" and
19023 "after" to just before the call to emit_insn_after_setloc.
19025 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19027 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19029 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19031 * dwarf2out.c (gen_enumeration_type_die): When
19032 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19034 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19036 PR tree-optimization/78965
19037 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19038 Change first argument from const call_info & to call_info &. For %n
19039 set info.nowrite to false.
19041 PR middle-end/78901
19042 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19043 possibly throwing calls.
19045 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19046 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19047 and fns handling, rather than in a separate case SSA_NAME.
19049 2017-01-02 Jeff Law <law@redhat.com>
19051 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19052 fixes for first_period and second_period variables.
19054 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19057 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19058 (*insvqi_1): New insn pattern.
19059 (*insvqi_1_mem_rex64): Ditto.
19060 (*insvqi_2): Ditto.
19061 (*insvqi_3): Rename from *insvqi.
19063 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19065 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19067 * doc/cfg.texi (Edges): Remove reference to Java.
19068 (Maintaining the CFG): Ditto.
19070 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19072 PR middle-end/77674
19073 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19074 transparent aliases.
19076 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19078 PR middle-end/77484
19079 * predict.def (PRED_CALL): Update hitrate.
19080 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19081 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19082 into direct/indirect/polymorphic variants.
19084 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19086 Update copyright years.
19088 * gcc.c (process_command): Update copyright notice dates.
19089 * gcov-dump.c (print_version): Ditto.
19090 * gcov.c (print_version): Ditto.
19091 * gcov-tool.c (print_version): Ditto.
19092 * gengtype.c (create_file): Ditto.
19093 * doc/cpp.texi: Bump @copying's copyright year.
19094 * doc/cppinternals.texi: Ditto.
19095 * doc/gcc.texi: Ditto.
19096 * doc/gccint.texi: Ditto.
19097 * doc/gcov.texi: Ditto.
19098 * doc/install.texi: Ditto.
19099 * doc/invoke.texi: Ditto.
19101 Copyright (C) 2017 Free Software Foundation, Inc.
19103 Copying and distribution of this file, with or without modification,
19104 are permitted in any medium without royalty provided the copyright
19105 notice and this notice are preserved.