1 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
3 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4 from all CPU target flags enable members.
6 2017-07-26 Richard Biener <rguenther@suse.de>
8 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9 (decision_tree::gen): Make 'type' const.
10 (write_predicate): Likewise.
12 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16 (rs6000_option_override_internal): Likewise.
17 (rs6000_expand_vector_set): Likewise.
18 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19 (TARGET_UPPER_REGS_SF): Likewise.
20 (TARGET_UPPER_REGS_DI): Likewise.
21 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
24 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
25 (Splitters for DI constants in Altivec registers): Likewise.
26 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
27 (vsx_set_v4sf_p9): Likewise.
28 (vsx_set_v4sf_p9_zero): Likewise.
29 (vsx_insert_extract_v4sf_p9): Likewise.
30 (vsx_insert_extract_v4sf_p9_2): Likewise.
32 2017-07-25 Carl Love <cel@us.ibm.com>
34 * doc/extend.texi: Update the built-in documentation file for the
35 existing built-in functions
36 vector signed char vec_cnttz (vector signed char);
37 vector unsigned char vec_cnttz (vector unsigned char);
38 vector signed short vec_cnttz (vector signed short);
39 vector unsigned short vec_cnttz (vector unsigned short);
40 vector signed int vec_cnttz (vector signed int);
41 vector unsigned int vec_cnttz (vector unsigned int);
42 vector signed long long vec_cnttz (vector signed long long);
43 vector unsigned long long vec_cnttz (vector unsigned long long);
45 2017-07-25 Andrew Pinski <apinski@cavium.com>
47 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
48 accesses where the use is for the first operand of a BIT_INSERT.
50 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
53 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
54 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
56 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
58 * config/i386/gstabs.h: Delete.
59 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
61 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
63 * config/i386/i386.c (ix86_decompose_address): Do not check for
64 register RTX when looking at index_reg or base_reg.
65 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
67 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
69 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
70 to update EH info here.
72 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
74 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
76 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
78 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
80 2017-07-25 Torsten Duwe <duwe@suse.de>
82 * common.opt: Introduce -fpatchable-function-entry
83 command line option, and its variables function_entry_patch_area_size
84 and function_entry_patch_area_start.
85 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
86 including a two-value parser.
87 * target.def (print_patchable_function_entry): New target hook.
88 * targhooks.h (default_print_patchable_function_entry): New function.
89 * targhooks.c (default_print_patchable_function_entry): Likewise.
90 * toplev.c (process_options): Switch off IPA-RA if
91 patchable function entries are being generated.
92 * varasm.c (assemble_start_function): Look at the
93 patchable-function-entry command line switch and current
94 function attributes and maybe generate NOP instructions by
95 calling the print_patchable_function_entry hook.
96 * doc/extend.texi: Document patchable_function_entry attribute.
97 * doc/invoke.texi: Document -fpatchable_function_entry
99 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
101 * doc/tm.texi: Re-generate.
103 2017-07-25 Jakub Jelinek <jakub@redhat.com>
106 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
107 TARGET_AVX512DQ rather than TARGET_AVX512BW.
109 2017-07-25 Tamar Christina <tamar.christina@arm.com>
111 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
113 2017-07-25 Richard Biener <rguenther@suse.de>
115 PR tree-optimization/81455
116 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
117 not walk in cycles when looking for guards.
119 2017-07-25 Richard Biener <rguenther@suse.de>
121 PR tree-optimization/81529
122 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
123 when optimizing backedge uses.
125 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
127 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
129 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
130 to dl_section_ref. On AIX, append an expression to subtract
131 the size of the section length to dl_section_ref.
133 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
135 * configure.ac: If any of the config.* scripts fail, exit 1.
136 * configure: Regenerate.
138 2017-07-25 Richard Biener <rguenther@suse.de>
141 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
142 of immediate uses, be more verbose on errors.
144 2017-07-25 Richard Biener <rguenther@suse.de>
146 PR tree-optimization/81510
147 * tree-vect-loop.c (vect_is_simple_reduction): When the
148 reduction stmt is not inside the loop bail out.
150 2017-07-25 Richard Biener <rguenther@suse.de>
152 PR tree-optimization/81303
153 * tree-vect-loop-manip.c (vect_loop_versioning): Build
154 profitability check against LOOP_VINFO_NITERSM1.
156 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
158 * domwalk.c (cmp_bb_postorder): Simplify.
159 (sort_bbs_postorder): New function. Use it...
160 (dom_walker::walk): ...here to optimize common cases.
162 2017-07-25 Martin Liska <mliska@suse.cz>
165 * ipa-visibility.c (function_and_variable_visibility): Make the
166 redirection just on target that supports aliasing.
167 Fix GNU coding style.
169 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
172 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
174 * config/arm/rtems.h: Likewise.
175 * config/bfin/rtems.h: Likewise.
176 * config/i386/rtemself.h: Likewise.
177 * config/lm32/rtems.h: Likewise.
178 * config/m32c/rtems.h: Likewise.
179 * config/m68k/rtemself.h: Likewise.
180 * config/microblaze/rtems.h: Likewise.
181 * config/mips/rtems.h: Likewise.
182 * config/moxie/rtems.h: Likewise.
183 * config/nios2/rtems.h: Likewise.
184 * config/powerpcspe/rtems.h: Likewise.
185 * config/rs6000/rtems.h: Likewise.
186 * config/rtems.h: Likewise.
187 * config/sh/rtems.h: Likewise.
188 * config/sh/rtemself.h: Likewise.
189 * config/sparc/rtemself.h: Likewise.
191 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
194 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
195 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
196 * tree-ssa-structalias.c (alias_get_name): Same.
198 2017-07-25 Bin Cheng <bin.cheng@arm.com>
201 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
202 instructions if no du chain is found.
204 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
206 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
208 2017-07-25 Richard Biener <rguenther@suse.de>
211 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
214 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
216 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
218 (ISA_2_7_MASKS_SERVER): Likewise.
219 (ISA_3_0_MASKS_IEEE): Likewise.
220 (OTHER_P8_VECTOR_MASKS): Likewise.
221 (OTHER_VSX_VECTOR_MASKS): Likewise.
222 (POWERPC_MASKS): Likewise.
223 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
224 duplicate list of options.
225 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
226 explicit -mupper-regs options.
227 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
228 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
229 alias for -mupper-regs-df.
230 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
231 (rs6000_init_hard_regno_mode_ok): Likewise.
232 (rs6000_option_override_internal): Likewise.
233 (rs6000_opt_masks): Likewise.
234 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
235 options in terms of whether -mvsx or -mpower8-vector was used.
236 (TARGET_UPPER_REGS_DI): Likewise.
237 (TARGET_UPPER_REGS_SF): Likewise.
238 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
239 -mupper-regs-* options.
241 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
243 * passes.c (emergency_dump_function): Print some empty lines and a
244 header before the RTL dump.
246 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
248 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
250 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
253 * config/aarch64/aarch64.c (aarch64_classify_symbol):
254 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
256 2017-07-24 Carl Love <cel@us.ibm.com>
258 * config/rs6000/rs6000-c.c: Add support for built-in functions
259 vector float vec_extract_fp32_from_shorth (vector unsigned short);
260 vector float vec_extract_fp32_from_shortl (vector unsigned short);
261 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
262 vec_extract_fp_from_shortl): Add defines for the two builtins.
263 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
264 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
266 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
267 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
268 * doc/extend.texi: Update the built-in documentation file for the
269 new built-in function.
271 2017-07-24 Jakub Jelinek <jakub@redhat.com>
274 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
276 * doc/generic.texi: Likewise.
277 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
278 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
280 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
282 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
283 (aarch64_mls_elt_merge<mode>): Likewise.
285 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
287 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
290 2017-07-23 Michael Collison <michael.collison@arm.com>
292 * config/arm/arm.c (arm_option_override): Deprecate
293 use of -mstructure-size-boundary.
294 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
295 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
297 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
300 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
301 Reduce cost estimate for direct moves.
303 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
306 * config/i386/i386.c (ix86_option_override_internal): Disable
307 BMI, BMI2 and TBM instructions for -m16.
309 2017-07-21 Carl Love <cel@us.ibm.com>
311 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
312 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
313 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
315 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
316 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
317 VMULOSW): New enum "unspec" values.
318 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
319 altivec_vmulosw): New patterns.
320 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
321 VMULOSW): Add definitions.
323 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
325 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
327 * config/aarch64/aarch64-options-extensions.def (rdma); New.
328 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
329 (AARCH64_FL_V8_1): Renumber.
330 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
331 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
332 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
333 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
334 rdma to feature modifiers list.
336 2017-07-21 Yury Gribov <tetra2005@gmail.com>
339 * ipa-visibility (function_and_variable_visibility): Convert
340 recursive PLT call to direct call if appropriate.
342 2017-07-21 Andrew Pinski <apinski@cavium.com>
344 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
345 operand 1 to see if the types precision matches.
346 * fold-const.c (operand_equal_p): Likewise.
348 2017-07-21 Richard Biener <rguenther@suse.de>
350 PR tree-optimization/81303
351 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
352 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
353 (vect_peeling_hash_get_lowest_cost): Adjust.
354 (vect_enhance_data_refs_alignment): Likewise. Use
355 vect_get_peeling_costs_all_drs to compute the penalty for no
356 peeling to match up costs.
358 2017-07-21 Richard Biener <rguenther@suse.de>
360 PR tree-optimization/81500
361 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
362 we didn't identify a reduction path.
364 2017-07-21 Tom de Vries <tom@codesourcery.com>
365 Cesar Philippidis <cesar@codesourcery.com>
367 PR gcov-profile/81442
368 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
371 2017-07-21 Tom de Vries <tom@codesourcery.com>
374 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
376 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
377 nvptx_override_options_after_change.
379 2017-07-21 Ulrich Drepper <drepper@redhat.com>
381 * dwarf2out.c (output_file_names): Avoid double testing for
384 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
386 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
388 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
390 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
392 (try_crossjump_to_edge): Do not punt on partitioned functions.
394 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
396 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
397 Put all BBs reachable only via paths crossing cold region to cold
399 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
401 2016-07-21 Richard Biener <rguenther@suse.de>
403 PR tree-optimization/81303
404 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
405 into account prologue and epilogue iterations when raising
406 min_profitable_iters to sth at least covering one vector iteration.
408 2017-07-21 Tamar Christina <tamar.christina@arm.com>
410 * config/arm/arm.c (arm_test_cpu_arch_dat):
413 2017-07-20 Nathan Sidwell <nathan@acm.org>
416 * tree.h (TYPE_METHODS): Delete.
417 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
418 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
419 (dbxout_type_methods): Scan TYPE_FIELDS.
420 (dbxout_type): Don't check TYPE_METHODS here.
421 * function.c (use_register_for_decl): Always ignore register for
422 class types when not optimizing.
423 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
424 * tree.c (free_lang_data_in_type): Stitch out member functions and
425 templates from TYPE_FIELDS.
426 (build_distinct_type_copy, verify_type_variant,
427 verify_type): Member fns are on TYPE_FIELDS.
428 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
429 * tree-pretty-print.c (dump_generic_node): Likewise.
431 2017-07-20 Jakub Jelinek <jakub@redhat.com>
434 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
435 V2TImode and V4TImode.
436 (ix86_expand_vector_extract): Likewise.
437 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
438 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
439 (ssescalarmode): Handle V4TImode and V2TImode.
440 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
441 (*vec_extractv2ti, *vec_extractv4ti): New insns.
442 (VEXTRACTI128_MODE): New mode iterator.
443 (splitter for *vec_extractv?ti first element): New.
444 (VEC_INIT_MODE): New mode iterator.
445 (vec_init<mode>): Consolidate 3 expanders into one using
446 VEC_INIT_MODE mode iterator.
448 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
450 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
451 non_spilled_static_chain_regno_p.
453 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
455 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
457 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
459 * bb-reorder.c (connect_traces): Allow copying of blocks within
462 2017-07-20 Richard Biener <rguenther@suse.de>
464 * gimple.h (gimple_phi_result): Add gphi * overload.
465 (gimple_phi_result_ptr): Likewise.
466 (gimple_phi_arg): Likewise. Adjust index assert to only
467 allow actual argument accesses rather than all slots available
469 (gimple_phi_arg_def): Add gphi * overload.
470 * tree-phinodes.c (make_phi_node): Initialize only actual
472 (resize_phi_node): Clear memory not covered by old node,
473 do not initialize excess argument slots.
474 (reserve_phi_args_for_new_edge): Initialize new argument slot
477 2017-07-20 Bin Cheng <bin.cheng@arm.com>
479 PR tree-optimization/81388
481 2016-07-21 Bin Cheng <bin.cheng@arm.com>
483 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
484 by removing computation of may_be_zero.
486 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
487 Tom de Vries <tom@codesourcery.com>
490 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
491 when gimple level profile disagrees with what RTL expander did.
493 2017-07-20 Richard Biener <rguenther@suse.de>
495 PR tree-optimization/61171
496 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
497 (vect_analyze_stmt): Add slp instance parameter.
498 (vectorizable_reduction): Likewise.
499 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
500 (vect_is_simple_reduction): Deal with chains not detected
501 as SLP reduction chain, specifically not properly associated
502 chains containing a mix of plus/minus.
503 (get_reduction_op): Remove.
504 (get_initial_defs_for_reduction): Simplify, pass in whether
505 this is a reduction chain, pass in the SLP node for the PHIs.
506 (vect_create_epilog_for_reduction): Get the SLP instance as
508 (vectorizable_reduction): Get the SLP instance as arg.
509 During analysis remember the SLP node with the PHIs in the
510 instance. Simplify getting at the vectorized reduction PHIs.
511 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
512 through SLP instance.
513 (vect_slp_analyze_operations): Likewise.
514 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
515 (vect_transform_stmt): Likewise.
517 2017-07-20 Tom de Vries <tom@codesourcery.com>
519 PR tree-optimization/81489
520 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
521 read of phi arg location to before loop that modifies phi.
523 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
525 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
528 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
531 * except.c (execute): Fix ordering issue.
533 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
535 PR rtl-optimization/81423
536 * combine.c (make_compound_operation_int): Don't try to optimize
537 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
539 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
541 PR rtl-optimization/81423
542 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
543 with a constant that is -1 in the truncated to mode.
545 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
547 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
548 (determine_unlikely_bbs): ... here.
549 * predict.h (propagate_unlikely_bbs_forward): Declare.
550 * cfgexpand.c (pass_expand::execute): Use it.
551 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
553 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
554 propagate_unlikely_bbs_forward.
556 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
559 * except.c (maybe_add_nop_after_section_switch): New function.
562 2017-07-19 Tom de Vries <tom@codesourcery.com>
564 * gimple.h (gimple_phi_set_arg): Make assert more strict.
566 2017-07-19 Tom de Vries <tom@codesourcery.com>
568 * gimple.h (gimple_phi_arg): Make assert more strict.
570 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
572 * config.gcc (powerpc*-*-*): Add mmintrin.h.
573 * config/rs6000/mmintrin.h: New file.
574 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
576 2017-07-19 Jakub Jelinek <jakub@redhat.com>
578 PR tree-optimization/81346
579 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
581 2017-07-19 Tom de Vries <tom@codesourcery.com>
583 * config/nvptx/nvptx.md (VECIM): Add V2DI.
585 2017-07-19 Tom de Vries <tom@codesourcery.com>
587 * config/nvptx/nvptx-modes.def: Add V2DImode.
588 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
589 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
590 (nvptx_output_mov_insn): Handle lack of mov.b128.
591 (nvptx_print_operand): Handle 'H' and 'L' codes.
592 (nvptx_vector_mode_supported): Allow V2DImode.
593 (nvptx_preferred_simd_mode): New function.
594 (nvptx_data_alignment): New function.
595 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
596 nvptx_preferred_simd_mode.
597 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
599 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
601 2017-07-19 Tom de Vries <tom@codesourcery.com>
603 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
604 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
605 (nvptx_vector_mode_supported): New function. Allow V2SImode.
606 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
607 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
608 (mov<VECIM>_insn): New define_insn.
609 (define_expand "mov<VECIM>): New define_expand.
611 2017-07-19 Tom de Vries <tom@codesourcery.com>
613 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
615 2017-07-19 Jakub Jelinek <jakub@redhat.com>
617 PR tree-optimization/81346
618 * fold-const.h (fold_div_compare, range_check_type): Declare.
619 * fold-const.c (range_check_type): New function.
620 (build_range_check): Use range_check_type.
621 (fold_div_compare): No longer static, rewritten into
622 a match.pd helper function.
623 (fold_comparison): Don't call fold_div_compare here.
624 * match.pd (X / C1 op C2): New optimization using fold_div_compare
627 2017-07-19 Nathan Sidwell <nathan@acm.org>
629 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
630 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
631 * tree.c (find_decls_types_r, verify_type): Use
632 TYPE_{MIN,MAX}_VALUE_RAW.
633 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
634 (hash_tree): Likewise.
635 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
637 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
640 2017-07-18 Tom de Vries <tom@codesourcery.com>
643 * omp-expand.c (expand_omp_for_static_chunk): Handle
644 equal-argument loop exit phi.
646 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
649 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
650 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
652 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
654 (ror,rol -> rorx splitters): Use const_int_operand as
657 2017-06-18 Richard Biener <rguenther@suse.de>
659 PR tree-optimization/81410
660 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
661 the gap in the ! slp_perm SLP case after each group.
663 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
666 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
669 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
672 * predict.c (set_even_probabilities): Cleanup; do not affect
673 probabilities that are already known.
674 (combine_predictions_for_bb): Call even when count is set.
676 2017-07-18 Nathan Sidwell <nathan@acm.org>
678 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
681 2017-07-18 Bin Cheng <bin.cheng@arm.com>
684 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
685 optimization for loop niter analysis.
687 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
690 * config/avr/avr.c (avr_optimize_casesi): Don't use
691 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
693 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
695 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
696 body_cost_vec from _vect_peel_extended_info.
697 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
698 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
701 2017-07-18 Bin Cheng <bin.cheng@arm.com>
703 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
705 2017-07-18 Richard Biener <rguenther@suse.de>
707 PR tree-optimization/80620
708 PR tree-optimization/81403
709 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
710 info when re-using a VN table entry.
712 2017-07-18 Richard Biener <rguenther@suse.de>
714 PR tree-optimization/81418
715 * tree-vect-loop.c (vectorizable_reduction): Properly compute
716 vectype_in. Verify that with lane-reducing reduction operations
717 we have a single def-use cycle.
719 2017-07-17 Carl Love <cel@us.ibm.com>
721 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
723 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
724 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
725 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
726 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
727 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
728 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
729 VMULOSW): New enum "unspec" values.
730 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
731 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
732 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
733 altivec_vmulosw): New patterns.
734 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
735 VMULOSW): Add definitions.
736 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
738 * config/alpha/alpha.c: Include predict.h.
740 2017-07-17 Yury Gribov <tetra2005@gmail.com>
742 * tree-vrp.c (compare_assert_loc): Fix comparison function
743 to return predictable results.
745 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
747 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
750 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
751 * doc/invoke.texi (mexpand-adddi): Update text.
753 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
755 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
756 that also clobbers the CC register. The old expand code is moved
758 (*arc_clzsi2): ... here.
759 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
760 the CC register. The old expand code is moved to ...
761 (arc_ctzsi2): ... here.
763 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
765 * config/arc/arc.opt (mindexed-loads): Use initial value
766 TARGET_INDEXED_LOADS_DEFAULT.
767 (mauto-modify-reg): Use initial value
768 TARGET_AUTO_MODIFY_REG_DEFAULT.
769 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
770 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
771 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
772 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
774 2017-07-17 Martin Liska <mliska@suse.cz>
777 * opts.c (finish_options): Do not allow -fgnu-tm
778 w/ -fsanitize={kernel-,}address. Say sorry.
780 2017-07-17 Bin Cheng <bin.cheng@arm.com>
783 * tree-loop-distribution.c (classify_partition): Only assert on
785 (merge_dep_scc_partitions): Delete prameter. Update function call.
786 (distribute_loop): Remove code handling loop with unknown niters.
787 (pass_loop_distribution::execute): Skip loop with unknown niters.
789 2017-07-17 Bin Cheng <bin.cheng@arm.com>
792 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
793 function sort_partitions_by_post_order.
795 2017-07-17 Bin Cheng <bin.cheng@arm.com>
797 PR tree-optimization/81374
798 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
799 the max index of basic blocks, rather than number of basic blocks.
801 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
803 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
805 (arc_legitimate_pic_operand_p): Likewise.
806 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
808 (arc_needs_pcl_p): Likewise.
809 (arc_legitimate_pc_offset_p): Likewise.
810 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
811 function is also used in constrains.md.
812 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
813 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
814 PLUS. Only return true/false in known cases, otherwise assert.
815 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
816 is already called in arc_legitimate_constant_p.
817 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
819 (LEGITIMATE_PIC_OPERAND_P): Use
820 arc_raw_symbolic_reference_mentioned_p function.
821 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
826 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
827 Andrew Burgess <andrew.burgess@embecosm.com>
829 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
830 (arc_return_address_register): New function.
831 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
832 (arc_handle_fndecl_attribute): Add naked attribute.
833 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
834 (TARGET_WARN_FUNC_RETURN): Likewise.
835 (arc_allocate_stack_slots_for_args): New function.
836 (arc_warn_func_return): Likewise.
837 (machine_function): Change type fn_type.
838 (arc_compute_function_type): Consider new naked function type,
839 change function return type.
840 (arc_must_save_register): Adapt to handle new
841 arc_compute_function_type's return type.
842 (arc_expand_prologue): Likewise.
843 (arc_expand_epilogue): Likewise.
844 (arc_return_address_regs): Delete.
845 (arc_return_address_register): New function.
846 (arc_epilogue_uses): Use above function.
847 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
848 (arc_function_type): Change encoding, add naked type.
849 (ARC_INTERRUPT_P): Change to handle the new encoding.
850 (ARC_FAST_INTERRUPT_P): Likewise.
851 (ARC_NORMAL_P): Define.
852 (ARC_NAKED_P): Likewise.
853 (arc_compute_function_type): Delete prototype.
854 * config/arc/arc.md (in_ret_delay_slot): Use
855 arc_return_address_register function.
856 (simple_return): Likewise.
857 (p_return_i): Likewise.
859 2017-07-17 Jakub Jelinek <jakub@redhat.com>
861 PR tree-optimization/81428
862 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
863 can't be built for those types.
865 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
867 Remove stuff dead since r239246.
869 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
870 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
871 (avr_inform_devices): Remove dead stuff.
873 2017-07-17 Tamar Christina <tamar.christina@arm.com>
875 * config/arm/arm_neon.h: Fix softp typo.
877 2017-07-17 Jakub Jelinek <jakub@redhat.com>
879 PR tree-optimization/81365
880 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
881 aggregate moves onto bb predecessor edges, make sure there are no
882 loads that could alias the lhs in between the start of bb and the
885 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
888 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
889 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
890 [LSHIFTRT, outer_code = TRUNCATE]: Same.
892 2017-07-17 Jakub Jelinek <jakub@redhat.com>
894 PR tree-optimization/81396
895 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
896 (init_symbolic_number): Initialize it to 1.
897 (perform_symbolic_merge): Add n_ops from both operands into the new
899 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
900 without base_addr as useless if they need more than one operation.
901 (bswap_replace): Handle !bswap case for NULL base_addr.
903 2017-07-17 Tom de Vries <tom@codesourcery.com>
906 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
909 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
911 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
912 conditional builtin define __FIX_LEON3FT_B2BST.
914 2017-07-17 Daniel Cederman <cederman@gaisler.com>
916 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
917 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
920 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
922 PR rtl-optimization/81424
923 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
924 to remove potential trapping from operands if -fnon-call-exceptions.
926 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
928 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
929 profile_proability for scalling.
930 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
932 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
934 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
936 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
938 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
939 fixpoint arithmetics.
941 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
943 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
944 fixpoint arithmetics.
946 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
948 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
949 fixpoint arithmetics.
951 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
953 * profile-count.h (profile_probability::from_reg_br_prob_note,
954 profile_probability::to_reg_br_prob_note): New functions.
955 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
956 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
957 * predict.c (probability_reliable_p): Update.
958 (edge_probability_reliable_p): Update.
959 (br_prob_note_reliable_p): Update.
960 (invert_br_probabilities): Update.
961 (add_reg_br_prob_note): New function.
962 (combine_predictions_for_insn): Update.
963 * asan.c (asan_clear_shadow): Update.
964 * cfgbuild.c (compute_outgoing_frequencies): Update.
965 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
966 (update_br_prob_note): Update.
967 (rtl_verify_edges): Update.
968 (purge_dead_edges): Update.
969 (fixup_reorder_chain): Update.
970 * emit-rtl.c (try_split): Update.
971 * ifcvt.c (cond_exec_process_insns): Update.
972 (cond_exec_process_if_block): Update.
973 (dead_or_predicable): Update.
974 * internal-fn.c (expand_addsub_overflow): Update.
975 (expand_neg_overflow): Update.
976 (expand_mul_overflow): Update.
977 * loop-doloop.c (doloop_modify): Update.
978 * loop-unroll.c (compare_and_jump_seq): Update.
979 * optabs.c (emit_cmp_and_jump_insn_1): Update.
981 * reorg.c (mostly_true_jump): Update.
983 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
984 * config/alpha/alpha.c (emit_unlikely_jump): Update.
985 * config/arc/arc.c: (emit_unlikely_jump): Update.
986 * config/arm/arm.c: (emit_unlikely_jump): Update.
987 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
988 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
989 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
990 (ix86_print_operand): Update.
991 (ix86_split_fp_branch): Update.
992 (predict_jump): Update.
993 * config/ia64/ia64.c (ia64_print_operand): Update.
994 * config/mmix/mmix.c (mmix_print_operand): Update.
995 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
996 (rs6000_expand_split_stack_prologue): Update.
997 * config/rs6000/rs6000.c: Update.
998 * config/s390/s390.c (s390_expand_vec_strlen): Update.
999 (s390_expand_vec_movstr): Update.
1000 (s390_expand_cs_tdsi): Update.
1001 (s390_expand_split_stack_prologue): Update.
1002 * config/sh/sh.c (sh_print_operand): Update.
1003 (expand_cbranchsi4): Update.
1004 (expand_cbranchdi4): Update.
1005 * config/sparc/sparc.c (output_v9branch): Update.
1006 * config/spu/spu.c (get_branch_target): Update.
1007 (ea_load_store_inline): Update.
1008 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1009 * config/tilepro/tilepro.c: Update.
1011 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1013 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1014 (gimplify_save_expr): Likewise.
1016 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1018 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1020 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1022 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1024 * timevar.def (TV_IPA_FNSUMMARY): Define.
1026 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1028 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1029 to back store errata sensitive sequence from being generated.
1030 (sqrtdf2_fix): Likewise.
1032 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1034 * tree-ssa-threadupdate.c (compute_path_counts,
1035 update_joiner_offpath_counts): Use profile_probability.
1037 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1040 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1042 * config/arm/arm-c.c (arm_cpu_builtins): Define
1043 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1045 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1047 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1048 array entries to represent __ieee128 versions of the
1049 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1050 scalar_extract_sig, and scalar_insert_exp built-in functions.
1051 (altivec_resolve_overloaded_builtin): Add special case handling
1052 for the __builtin_scalar_insert_exp function, as represented by
1053 the P9V_BUILTIN_VEC_VSIEDP constant.
1054 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1055 exponent support for __ieee128 argument.
1056 (VSESQP): Add scalar extract signature support for __ieee128
1058 (VSTDCNQP): Add scalar test negative support for __ieee128
1060 (VSIEQP): Add scalar insert exponent support for __int128 argument
1061 with __ieee128 result.
1062 (VSIEQPF): Add scalar insert exponent support for __ieee128
1063 argument with __ieee128 result.
1064 (VSTDCQP): Add scalar test data class support for __ieee128
1066 (VSTDCNQP): Add overload support for scalar test negative with
1068 (VSTDCQP): Add overload support for scalar test data class
1070 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1072 (UNSPEC_VSX_SIEXPQP): New constant.
1073 (xsxexpqp): New insn for VSX scalar extract exponent quad
1075 (xsxsigqp): New insn for VSX scalar extract significand quad
1077 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1078 precision with floating point argument.
1079 (xststdcqp): New expand for VSX scalar test data class quad
1081 (xststdcnegqp): New expand for VSX scalar test negative quad
1083 (xststdcqp): New insn to match expansions for VSX scalar test data
1084 class quad precision and VSX scalar test negative quad precision.
1085 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1086 special case operand checking to enforce that second operand of
1087 VSX scalar test data class with quad precision argument is a 7-bit
1089 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1090 prototypes and descriptions of __ieee128 versions of
1091 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1092 scalar_test_data_class, and scalar_test_neg built-in functions.
1094 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1096 PR tree-optimization/81162
1097 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1098 replace a negate with an add.
1100 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1102 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1104 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1106 * config/arm/arm-c.c (arm_cpu_builtins): Define
1107 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1109 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1111 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1112 (armv8-r): Set ARM Cortex-R52 as default CPU.
1113 * config/arm/arm-tables.opt: Regenerate.
1114 * config/arm/arm-tune.md: Regenerate.
1115 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1117 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1118 extension for -mcpu=cortex-r52.
1120 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1122 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1123 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1124 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1125 (fp-armv8): Define it as FP_ARMv8 only.
1126 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1127 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1129 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1130 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1131 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1132 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1133 than TARGET_FPU_ARMV8.
1134 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1135 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1136 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1138 * config/arm/neon.md (neon_vrint): Likewise.
1139 (neon_vcvt): Likewise.
1140 (neon_<fmaxmin_op><mode>): Likewise.
1141 (<fmaxmin><mode>3): Likewise.
1142 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1143 * config/arm/predicates.md (arm_cond_move_operator): Check against
1144 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1146 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1148 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1151 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1153 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1154 loop in comment with memset.
1156 2017-07-14 Martin Liska <mliska@suse.cz>
1158 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1159 * dwarf2out.c (is_java): Remove the function.
1160 (output_pubname): Remove usage of the function.
1161 (lower_bound_default): Remove usage of DW_LANG_Java.
1162 (gen_compile_unit_die): Likewise.
1163 * gcc.c: Remove compiler defaults for .java and .zip files.
1164 * gimple-expr.c (remove_suffix): Change as there's no longer
1165 extension than 4-letter one.
1166 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1167 (gimplify_save_expr): Likewise.
1168 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1169 as it's possible even for other languages than Java.
1170 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1171 * lto-opts.c (lto_write_options): Remove reference to Java.
1172 * opts.c (strip_off_ending): Update file extension handling.
1173 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1174 * tree-eh.c (lower_resx): Likewise.
1175 * tree.c (free_lang_data_in_type): Remove dead code.
1176 (find_decls_types_r): Likewise.
1177 (build_common_builtin_nodes): Remove Java from a comment.
1178 (verify_type): Remove dead code.
1179 * varasm.c (assemble_external): Remove Java from a comment.
1181 2017-07-14 Martin Liska <mliska@suse.cz>
1183 * opts.c (finish_options): Add quotes.
1184 (common_handle_option): Likewise.
1186 2017-07-14 Martin Liska <mliska@suse.cz>
1188 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1189 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1191 * dwarf2out.c (lower_bound_default): Do not handle
1193 (gen_compile_unit_die): Likewise.
1194 * gcc.c: Remove default extension binding for GNU Pascal.
1195 * stmt.c: Remove Pascal language from a comment.
1196 * xcoffout.c: Likewise.
1198 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1201 * diagnostic-show-locus.c (fixit_cmp): New function.
1202 (layout::layout): Sort m_fixit_hints.
1203 (column_range::column_range): Assert that the values are valid.
1204 (struct char_span): New struct.
1205 (correction::overwrite): New method.
1206 (struct source_line): New struct.
1207 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1208 calls in terms of classes source_line and char_span, and
1209 correction::overwrite.
1210 (selftest::test_overlapped_fixit_printing_2): New function.
1211 (selftest::diagnostic_show_locus_c_tests): Call it.
1213 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1215 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1216 early if there is no lhs.
1218 2017-07-13 Martin Liska <mliska@suse.cz>
1220 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1221 (gen_reference_type_die): Likewise.
1222 * stor-layout.c: Remove Pascal-related comment.
1224 2017-07-13 Martin Liska <mliska@suse.cz>
1226 * opts.c (finish_options): Add quotes to error messages.
1227 (parse_sanitizer_options): Likewise.
1229 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1231 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1233 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1235 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1237 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1239 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1241 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1243 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1246 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1247 provides the hardware capability bits, define the macro
1248 __BUILTIN_CPU_SUPPORTS__.
1249 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1250 if GLIBC does not provide the hardware capability bits. Add a
1251 gcc_unreachable call if the built-in cpu function is neither
1252 __builtin_cpu_is nor __builtin_cpu_supports.
1253 (rs6000_get_function_versions_dispatcher): Change the warning
1254 that an old GLIBC is used which does not export the capability
1255 bits to be an error.
1256 * doc/extend.texi (target_clones attribute): Document the
1257 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1258 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1259 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1260 the macros defined by GCC if the newer GLIBC is available.
1262 2017-07-12 Jeff Law <law@redhat.com>
1264 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1265 remaining includes slightly.
1266 * config/riscv/riscv-builtins.c: Include profile-count.h.
1268 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1271 * config/avr/avr.c (avr_set_current_function): In diagnostic
1272 messages: Quote keywords and (parts of) identifiers.
1273 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1276 2017-07-12 Carl Love <cel@us.ibm.com>
1278 * config/rs6000/rs6000-c.c: Add support for built-in functions
1279 vector bool char vec_revb (vector bool char);
1280 vector bool short vec_revb (vector short char);
1281 vector bool int vec_revb (vector bool int);
1282 vector bool long long vec_revb (vector bool long long);
1283 * doc/extend.texi: Update the built-in documentation file for the
1284 new built-in functions.
1286 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1288 * config/s390/s390.md: Remove movcc splitter.
1290 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1292 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1293 load/store on condition.
1295 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1298 * config/avr/avr.c (avr_encode_section_info)
1299 [progmem && !TREE_READONLY]: Error if progmem object needs
1302 2017-07-11 Michael Collison <michael.collison@arm.com>
1304 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1307 2017-07-11 Carl Love <cel@us.ibm.com>
1309 * config/rs6000/rs6000-c.c: Add support for builtins
1310 vector unsigned int vec_parity_lsbb (vector signed int);
1311 vector unsigned int vec_parity_lsbb (vector unsigned int);
1312 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1313 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1314 vector unsigned long long vec_parity_lsbb (vector signed long long);
1315 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1316 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1317 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1318 * doc/extend.texi: Update the built-in documentation file for the
1319 new built-in functions.
1321 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1323 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1324 (layout::m_primary_loc): New field.
1325 (layout::layout): Initialize new field. Move location filtering
1326 logic from here to...
1327 (layout::maybe_add_location_range): ...this new method. Add
1328 support for filtering to just the lines already specified by other
1330 (layout::will_show_line_p): New method.
1331 (gcc_rich_location::add_location_if_nearby): New method.
1332 (selftest::test_add_location_if_nearby): New test function.
1333 (selftest::diagnostic_show_locus_c_tests): Call it.
1334 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1337 2017-07-11 Tom de Vries <tom@codesourcery.com>
1339 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1340 (bb_first_real_insn): New function.
1341 (nvptx_single): Add extra initialization of broadcasted condition
1344 2017-07-11 Nathan Sidwell <nathan@acm.org>
1346 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1348 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1350 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1351 Binutils doc as TEXI will mess them up.
1352 * doc/invoke.texi (AVR Options): Same here.
1354 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1356 * config/sparc/sparc.opt (mfix-ut700): New option.
1357 (mfix-gr712rc): Likewise.
1358 (sparc_fix_b2bst): New variable.
1359 * doc/invoke.texi (SPARC options): Document them.
1360 (ARM options): Fix warnings.
1361 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1362 instructions to prevent sequences that can trigger the store-store
1363 errata for certain LEON3FT processors.
1364 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1365 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1366 * config/sparc/sparc.md (fix_b2bst): New attribute.
1367 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1369 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1372 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1374 (*rsqrtsf2_sse): Ditto.
1376 (div<mode>3): Macroize insn from divdf3 and divsf3
1377 using MODEF mode iterator.
1379 2017-07-10 Martin Sebor <msebor@redhat.com>
1381 PR tree-optimization/80397
1382 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1383 instead of testing for equality to INTEGER_TYPE.
1385 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1387 * config.gcc: Remove uclibc from arc target spec.
1389 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1391 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1393 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1396 * lto-wrapper.c (remove_option): New function.
1397 (merge_and_complain): Merge PIC/PIE options more realistically.
1399 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1401 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1405 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1406 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1407 * config.in: Regenerate.
1408 * configure: Regenerate.
1409 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1410 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1411 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1412 (TARGET_GASISR_PROLOGUES): ...target mask.
1413 * common/config/avr/avr-common.c
1414 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1415 Set -mgas-isr-prologues.
1416 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1417 INSERT_PASS_BEFORE for it.
1418 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1419 * config/avr/avr.c (avr_option_override)
1420 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1421 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1422 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1423 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1424 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1426 (make_avr_pass_pre_proep): New function.
1427 (emit_push_sfr) <treg>: Add argument to function and use it
1429 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1430 and set machine->gasisr.yes.
1431 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1432 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1433 __gcc_isr.n_pushed to .L__stack_usage.
1434 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1435 (avr_asm_final_postscan_insn): ...this new static function.
1436 * config/avr/avr.h (machine_function)
1437 <is_no_gccisr, use_L__stack_usage>: New fields.
1438 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1439 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1440 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1441 (gasisr, *gasisr): New expander and insn.
1442 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1443 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1444 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1446 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1448 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1451 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1453 Move jump-tables out of .text again.
1456 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1457 (ASM_OUTPUT_ADDR_VEC): New function.
1458 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1459 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1460 INSN_ADDRESSes as asm comment.
1461 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1462 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1463 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1464 * config/avr/avr.md (*tablejump): Adjust comment.
1465 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1466 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1468 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1469 (avr_output_addr_vec): New proto.
1470 (avr_log_t) <insn_addresses>: New field.
1472 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1475 * config/i386/i386.c (ix86_function_arg_advance): Set
1476 outgoing_args_on_stack to true if there are outgoing arguments
1478 (ix86_function_arg): Likewise.
1479 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1480 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1481 * config/i386/i386.h (machine_function): Add
1482 outgoing_args_on_stack.
1484 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1486 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1488 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1490 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1492 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1493 (REAL_H): Remove $(MACHMODE_H).
1494 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1496 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1497 $(MACHMODE_H) and double-int.h.
1498 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1500 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1503 2017-07-07 Andrew Pinski <apinski@cavium.com>
1505 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1506 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1508 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1510 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1511 Add warning if GCC was not configured to link against a GLIBC that
1512 exports the hardware capability bits.
1513 (make_resolver_func): Make resolver function private and not a
1514 COMDAT function. Create the name with clone_function_name instead
1515 of make_unique_name.
1518 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1519 correct operand in doing the split.
1521 2017-07-07 Carl Love <cel@us.ibm.com>
1523 * config/rs6000/rs6000-c: Add support for built-in function
1524 vector unsigned short vec_pack_to_short_fp32 (vector float,
1526 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1527 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1528 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1529 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1530 (convert_4f32_8i16): Add define_expand.
1531 * doc/extend.texi: Update the built-in documentation file for the
1532 new built-in function.
1534 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1536 * config/sparc/m8.md: New file.
1537 * config/sparc/sparc.md: Include m8.md.
1539 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1541 * config/sparc/sparc.opt: New option -mvis4b.
1542 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1543 (sparc_option_override): Handle VIS4B.
1544 (enum sparc_builtins): Define
1545 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1546 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1547 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1548 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1549 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1550 (check_constant_argument): New function.
1551 (sparc_vis_init_builtins): Define builtins
1552 __builtin_vis_dictunpack{8,16,32},
1553 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1554 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1555 __builtin_vis_fpcmpde{8,16,32}shl and
1556 __builtin_vis_fpcmpur{8,16,32}shl.
1557 (sparc_expand_builtin): Check that the constant operands to
1558 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1559 constant and in range.
1560 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1562 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1563 (SPARC_IMM5_P): Likewise.
1564 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1565 (enabled): Handle vis4b.
1566 (UNSPEC_DICTUNPACK): New unspec.
1567 (UNSPEC_FPCMPSHL): Likewise.
1568 (UNSPEC_FPUCMPSHL): Likewise.
1569 (UNSPEC_FPCMPDESHL): Likewise.
1570 (UNSPEC_FPCMPURSHL): Likewise.
1571 (cpu_feature): New CPU feature `vis4b'.
1572 (dictunpack{8,16,32}): New insns.
1573 (FPCSMODE): New mode iterator.
1574 (fpcscond): New code iterator.
1575 (fpcsucond): Likewise.
1576 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1577 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1578 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1579 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1580 * config/sparc/constraints.md: Define constraints `q' for unsigned
1581 2-bit integer constants and `t' for unsigned 5-bit integer
1583 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1585 (imm5_operand_dictunpack16): Likewise.
1586 (imm5_operand_dictunpack32): Likewise.
1587 (imm2_operand): Likewise.
1588 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1589 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1590 ditunpack* and fpcmp*shl builtins.
1592 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1594 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1595 * config.in: Add HAVE_AS_SPARC6 define.
1596 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1598 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1600 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1601 (CPP_CPU_SPEC): Handle m8.
1602 (ASM_CPU_SPEC): Likewise.
1603 * config/sparc/sparc-opts.h (enum processor_type): Add
1605 * config/sparc/sparc.c (m8_costs): New struct.
1606 (sparc_option_override): Handle TARGET_CPU_m8.
1607 (sparc32_initialize_trampoline): Likewise.
1608 (sparc64_initialize_trampoline): Likewise.
1609 (sparc_issue_rate): Likewise.
1610 (sparc_register_move_cost): Likewise.
1611 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1612 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1613 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1614 (CPP_CPU_SPEC): Handle M8.
1615 (ASM_CPU_SPEC): Likewise.
1616 (AS_M8_FLAG): Define.
1617 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1618 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1619 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1621 * configure: Regenerate.
1622 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1625 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1627 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1629 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1630 ("*movdi_insn_sp32"): Do not set v3pipe.
1631 ("*movsi_insn"): Likewise.
1632 ("*movdi_insn_sp64"): Likewise.
1633 ("*movsf_insn"): Likewise.
1634 ("*movdf_insn_sp32"): Likewise.
1635 ("*movdf_insn_sp64"): Likewise.
1636 ("*zero_extendsidi2_insn_sp64"): Likewise.
1637 ("*sign_extendsidi2_insn"): Likewise.
1638 ("*mov<VM32:mode>_insn"): Likewise.
1639 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1640 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1641 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1642 ("<vlop:code><VL:mode>3"): Likewise.
1643 ("*not_<vlop:code><VL:mode>3"): Likewise.
1644 ("*nand<VL:mode>_vis"): Likewise.
1645 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1646 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1647 ("one_cmpl<VL:mode>2"): Likewise.
1648 ("faligndata<VM64:mode>_vis"): Likewise.
1649 ("alignaddrsi_vis"): Likewise.
1650 ("alignaddrdi_vis"): Likweise.
1651 ("alignaddrlsi_vis"): Likewise.
1652 ("alignaddrldi_vis"): Likewise.
1653 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1654 ("bmaskdi_vis"): Likewise.
1655 ("bmasksi_vis"): Likewise.
1656 ("bshuffle<VM64:mode>_vis"): Likewise.
1657 ("cmask8<P:mode>_vis"): Likewise.
1658 ("cmask16<P:mode>_vis"): Likewise.
1659 ("cmask32<P:mode>_vis"): Likewise.
1660 ("pdistn<P:mode>_vis"): Likewise.
1661 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1663 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1665 * config/sparc/sparc.md ("subtype"): New insn attribute.
1666 ("*wrgsr_sp64"): Set insn subtype.
1667 ("*rdgsr_sp64"): Likewise.
1668 ("alignaddrsi_vis"): Likewise.
1669 ("alignaddrdi_vis"): Likewise.
1670 ("alignaddrlsi_vis"): Likewise.
1671 ("alignaddrldi_vis"): Likewise.
1672 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1673 ("fexpand_vis"): Likewise.
1674 ("fpmerge_vis"): Likewise.
1675 ("faligndata<VM64:mode>_vis"): Likewise.
1676 ("bshuffle<VM64:mode>_vis"): Likewise.
1677 ("cmask8<P:mode>_vis"): Likewise.
1678 ("cmask16<P:mode>_vis"): Likewise.
1679 ("cmask32<P:mode>_vis"): Likewise.
1680 ("fchksm16_vis"): Likewise.
1681 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1682 ("fmean16_vis"): Likewise.
1683 ("fp<plusminus_insn>64_vis"): Likewise.
1684 ("<plusminus_insn>v8qi3"): Likewise.
1685 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1686 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1687 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1688 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1689 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1690 ("*movqi_insn"): Likewise.
1691 ("*movhi_insn"): Likewise.
1692 ("*movsi_insn"): Likewise.
1693 ("movsi_pic_gotdata_op"): Likewise.
1694 ("*movdi_insn_sp32"): Likewise.
1695 ("*movdi_insn_sp64"): Likewise.
1696 ("movdi_pic_gotdata_op"): Likewise.
1697 ("*movsf_insn"): Likewise.
1698 ("*movdf_insn_sp32"): Likewise.
1699 ("*movdf_insn_sp64"): Likewise.
1700 ("*zero_extendhisi2_insn"): Likewise.
1701 ("*zero_extendqihi2_insn"): Likewise.
1702 ("*zero_extendqisi2_insn"): Likewise.
1703 ("*zero_extendqidi2_insn"): Likewise.
1704 ("*zero_extendhidi2_insn"): Likewise.
1705 ("*zero_extendsidi2_insn_sp64"): Likewise.
1706 ("ldfsr"): Likewise.
1707 ("prefetch_64"): Likewise.
1708 ("prefetch_32"): Likewise.
1709 ("tie_ld32"): Likewise.
1710 ("tie_ld64"): Likewise.
1711 ("*tldo_ldub_sp32"): Likewise.
1712 ("*tldo_ldub1_sp32"): Likewise.
1713 ("*tldo_ldub2_sp32"): Likewise.
1714 ("*tldo_ldub_sp64"): Likewise.
1715 ("*tldo_ldub1_sp64"): Likewise.
1716 ("*tldo_ldub2_sp64"): Likewise.
1717 ("*tldo_ldub3_sp64"): Likewise.
1718 ("*tldo_lduh_sp32"): Likewise.
1719 ("*tldo_lduh1_sp32"): Likewise.
1720 ("*tldo_lduh_sp64"): Likewise.
1721 ("*tldo_lduh1_sp64"): Likewise.
1722 ("*tldo_lduh2_sp64"): Likewise.
1723 ("*tldo_lduw_sp32"): Likewise.
1724 ("*tldo_lduw_sp64"): Likewise.
1725 ("*tldo_lduw1_sp64"): Likewise.
1726 ("*tldo_ldx_sp64"): Likewise.
1727 ("*mov<VM32:mode>_insn"): Likewise.
1728 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1729 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1731 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1733 * config/sparc/sparc.md ("type"): New insn type viscmp.
1734 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1736 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1737 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1738 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1739 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1741 ("n7_vis_logical_11cycle"): Likewise.
1742 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1743 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1744 * config/sparc/niagara.md ("niag_vis"): Likewise.
1745 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1746 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1748 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1750 * config/sparc/sparc.md: New instruction type `bmask'.
1751 (bmaskdi_vis): Use the `bmask' type.
1752 (bmasksi_vis): Likewise.
1753 * config/sparc/ultra3.md (us3_array): Likewise.
1754 * config/sparc/niagara7.md (n7_array): Likewise.
1755 * config/sparc/niagara4.md (n4_array): Likewise.
1756 * config/sparc/niagara2.md (niag2_vis): Likewise.
1757 (niag3_vis): Likewise.
1758 * config/sparc/niagara.md (niag_vis): Likewise.
1760 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1762 * ipa-comdats.c: Remove optimize check from gate.
1763 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1764 for functions not optimized.
1765 (ipa_fn_summary_read): Skip optimize check.
1766 (ipa_fn_summary_write): Likewise.
1767 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1769 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1771 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1772 (check_callers): Give up on caller which is not optimized.
1773 (inline_small_functions): Likewise.
1774 (ipa_inline): Do not give up when not optimizing.
1775 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1776 away unoptimizes cdtors.
1777 (whole_program_function_and_variable_visibility): Do
1778 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1779 * ipa.c (process_references): Do not check optimize.
1780 (symbol_table::remove_unreachable_nodes): Update optimize check.
1781 (set_writeonly_bit): Update optimize check.
1782 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1783 (pass_ipa_single_use::gate): Remove.
1785 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1787 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1788 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1789 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1790 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1791 permute_load, permute_store, adjust_extract, adjust_splat,
1792 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1793 replace_swap_with_copy, dump_swap_insn_table,
1794 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1795 recombine_lvx_pattern, recombine_stvx_pattern,
1796 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1797 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1798 to file rs6000-p8swap.c.
1799 * config/rs6000/rs6000-p8swap.c: New file.
1800 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1801 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1802 and rs6000*-*-* targets.
1804 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1806 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1808 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1810 * lto-wrapper.c (merge_and_complain): Do not merge
1811 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
1812 fsigned_zeros, ftrapping_math, fwrapv.
1813 (append_compiler_options): Do not track these options.
1814 (append_linker_options): Likewie
1816 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1818 * cgraphunit.c (cgraph_node::finalize_function): When
1819 !flag_toplevel_reorde set no_reorder flag.
1820 (varpool_node::finalize_decl): Likewise.
1821 (symbol_table::compile): Drop no toplevel reorder path.
1823 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1825 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
1826 edges; zero probability is not better than uninitialized.
1828 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1830 * asan.h (asan_sanitize_allocas_p): Declare.
1831 * asan.c (asan_sanitize_allocas_p): New function.
1832 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
1833 (handle_builtin_alloca): Likewise.
1834 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
1835 if !asan_sanitize_allocas_p.
1836 * params.def (asan-instrument-allocas): Add new option.
1837 * params.h (ASAN_PROTECT_ALLOCAS): Define.
1838 * opts.c (common_handle_option): Disable allocas sanitization for
1841 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1843 * asan.c: Include gimple-fold.h.
1844 (get_last_alloca_addr): New function.
1845 (handle_builtin_stackrestore): Likewise.
1846 (handle_builtin_alloca): Likewise.
1847 (asan_emit_allocas_unpoison): Likewise.
1848 (get_mem_refs_of_builtin_call): Add new parameter, remove const
1849 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
1850 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
1851 (instrument_builtin_call): Pass gimple iterator to
1852 get_mem_refs_of_builtin_call.
1853 (last_alloca_addr): New global.
1854 * asan.h (asan_emit_allocas_unpoison): Declare.
1855 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
1856 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
1857 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
1858 if function calls alloca.
1859 * gimple-fold.c (replace_call_with_value): Remove static keyword.
1860 * gimple-fold.h (replace_call_with_value): Declare.
1861 * internal-fn.c: Include asan.h.
1862 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
1863 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
1865 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1867 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
1868 (C_SELFTEST_FLAGS): New.
1869 (CPP_SELFTEST_FLAGS): New.
1870 (SELFTEST_DEPS): New, from deps of s-selftest.
1871 (C_SELFTEST_DEPS): New, from deps of s-selftest.
1872 (CPP_SELFTEST_DEPS): New.
1873 (selftest): Add dependency on s-selftest-c++.
1874 (s-selftest): Rename to...
1875 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
1876 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
1877 than SELFTEST_FLAGS.
1878 (selftest-gdb): Rename to...
1879 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
1881 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
1882 (selftest-valgrind): Rename to...
1883 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
1885 (selftest-valgrind): Reintroduce as an alias for
1886 selftest-c-valgrind.
1887 (s-selftest-c++): New.
1888 (selftest-c++-gdb): New.
1889 (selftest-c++-valgrind): New.
1891 2017-07-06 Olivier Hainque <hainque@adacore.com>
1893 * gcc.c (process_command): When deciding if undefined variables
1894 should be ignored when processing specs, accept "gcc -v" as well.
1896 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1898 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
1899 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
1901 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1903 * config/arm/arm-cpus.in (armv8-r): Add new entry.
1904 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
1905 * config/arm/arm-tables.opt: Regenerate.
1906 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
1908 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
1910 2017-07-06 Carl Love <cel@us.ibm.com>
1912 * ChangeLog: Clean up from mid air collision
1914 2017-07-06 Carl Love <cel@us.ibm.com>
1916 * config/rs6000/rs6000-c.c: Add support for built-in functions
1917 vector signed int vec_subc (vector signed int, vector signed int);
1918 vector signed __int128 vec_subc (vector signed __int128,
1919 vector signed __int128);
1920 vector unsigned __int128 vec_subc (vector unsigned __int128,
1921 vector unsigned __int128);
1922 vector signed int vec_sube (vector signed int, vector signed int,
1924 vector unsigned int vec_sube (vector unsigned int,
1925 vector unsigned int,
1926 vector unsigned int);
1927 vector signed __int128 vec_sube (vector signed __int128,
1928 vector signed __int128,
1929 vector signed__int128);
1930 vector unsigned __int128 vec_sube (vector unsigned __int128,
1931 vector unsigned __int128,
1932 vector unsigned __int128);
1933 vector signed int vec_subec (vector signed int, vector signed int,
1935 vector unsigned int vec_subec (vector unsigned int,
1936 vector unsigned int,
1937 vector unsigned int);
1938 vector signed __int128 vec_subec (vector signed __int128,
1939 vector signed __int128,
1940 vector signed__int128);
1941 vector unsigned __int128 vec_subec (vector unsigned __int128,
1942 vector unsigned __int128,
1943 vector unsigned __int128);
1944 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
1945 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
1946 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
1947 BU_ALTIVEC_OVERLOAD_X definitions.
1948 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
1949 * doc/extend.texi: Update the built-in documentation file for the new
1952 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1955 * diagnostic-show-locus.c (layout::layout): Use start and finish
1956 spelling location for the start and finish of each range.
1957 * genmatch.c (linemap_client_expand_location_to_spelling_point):
1958 Add unused aspect param.
1959 * input.c (expand_location_1): Add "aspect" param, and use it
1960 to access the correct part of the location.
1961 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
1963 (expand_location_to_spelling_point): Likewise.
1964 (linemap_client_expand_location_to_spelling_point): Add "aspect"
1965 param, and pass it to expand_location_1.
1967 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
1969 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
1970 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
1971 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
1972 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
1973 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
1974 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
1975 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
1976 _mm_maskz_getmant_ss): New intrinsics.
1977 (__builtin_ia32_getexpss128_mask): Changed to ...
1978 __builtin_ia32_getexpss128_round ... this.
1979 (__builtin_ia32_getexpsd128_mask): Changed to ...
1980 __builtin_ia32_getexpsd128_round ... this.
1981 * config/i386/i386-builtin-types.def
1982 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
1983 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
1984 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
1985 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
1986 __builtin_ia32_getmantss_mask_round): New builtins.
1987 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
1988 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
1989 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
1990 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
1991 * config/i386/sse.md
1992 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
1993 avx512f_sgetexp<mode><mask_scalar_name>
1994 <round_saeonly_scalar_name> ... this.
1995 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
1996 %0, %1, %2<round_saeonly_op3>}): Changed to ...
1997 vgetexp<ssescalarmodesuffix>
1998 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1999 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2000 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2001 avx512f_vgetmant<mode><mask_scalar_name>
2002 <round_saeonly_scalar_name> ... this.
2003 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2004 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2005 vgetmant<ssescalarmodesuffix>
2006 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2007 %0<mask_scalar_operand4>, %1, %2
2008 <round_saeonly_scalar_mask_op4>, %3} ... this.
2009 * config/i386/subst.md (mask_scalar_operand4,
2010 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2011 round_saeonly_scalar_nimm_predicate): New subst attributes.
2013 2017-07-06 Julia Koval <julia.koval@intel.com>
2015 * config/i386/i386.c (ix86_erase_embedded_rounding):
2016 Remove code for old rounding pattern.
2018 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2020 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2022 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2024 * doc/sourcebuild.texi (Test Directives, Variants of
2025 dg-require-support): Add documentation for dg-require-stack-check.
2027 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2029 * config/i386/subst.md (mask_scalar, round_scalar,
2030 round_saeonly_scalar): New meta-templates.
2031 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2032 round_scalar_mask_operand3, round_scalar_mask_op3,
2033 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2034 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2035 round_saeonly_scalar_constraint,
2036 round_saeonly_scalar_prefix): New subst attribute.
2037 * config/i386/sse.md
2038 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2039 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2040 <round_scalar_name> ... this.
2041 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2042 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2043 <round_scalar_name> ... this.
2044 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2045 <sse>_vm<code><mode>3<mask_scalar_name>
2046 <round_saeonly_scalar_name> ... this.
2047 (v<plusminus_mnemonic><ssescalarmodesuffix>
2048 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2049 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2050 v<plusminus_mnemonic><ssescalarmodesuffix>
2051 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2052 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2053 (v<multdiv_mnemonic><ssescalarmodesuffix>
2054 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2055 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2056 v<multdiv_mnemonic><ssescalarmodesuffix>
2057 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2058 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2059 (v<maxmin_float><ssescalarmodesuffix>
2060 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2061 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2062 v<maxmin_float><ssescalarmodesuffix>
2063 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2064 %0<mask_scalar_operand3>, %1, %<iptr>2
2065 <round_saeonly_scalar_mask_op3>} ... this.
2067 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2069 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2070 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2072 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2073 Alan Hayward <alan.hayward@arm.com>
2074 David Sherwood <david.sherwood@arm.com>
2076 * combine.c (simplify_if_then_else): Remove "enum" before
2078 * compare-elim.c (can_eliminate_compare): Likewise.
2079 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2081 (aarch64_lookup_simd_builtin_type): Likewise.
2082 (aarch64_simd_builtin_type): Likewise.
2083 (aarch64_init_simd_builtin_types): Likewise.
2084 (aarch64_simd_expand_args): Likewise.
2085 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2087 (aarch64_reverse_mask): Likewise.
2088 (aarch64_simd_emit_reg_reg_move): Likewise.
2089 (aarch64_gen_adjusted_ldpstp): Likewise.
2090 (aarch64_ccmp_mode_to_code): Likewise.
2091 (aarch64_operands_ok_for_ldpstp): Likewise.
2092 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2093 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2095 (aarch64_min_divisions_for_recip_mul): Likewise.
2096 (aarch64_reassociation_width): Likewise.
2097 (aarch64_get_condition_code_1): Likewise.
2098 (aarch64_simd_emit_reg_reg_move): Likewise.
2099 (aarch64_simd_attr_length_rglist): Likewise.
2100 (aarch64_reverse_mask): Likewise.
2101 (aarch64_operands_ok_for_ldpstp): Likewise.
2102 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2103 (aarch64_gen_adjusted_ldpstp): Likewise.
2104 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2106 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2107 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2108 (arm_lookup_simd_builtin_type): Likewise.
2109 (arm_simd_builtin_type): Likewise.
2110 (arm_init_simd_builtin_types): Likewise.
2111 (arm_expand_builtin_args): Likewise.
2112 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2113 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2114 (ft32_setup_incoming_varargs): Likewise.
2115 (ft32_function_arg): Likewise.
2116 (ft32_function_arg_advance): Likewise.
2117 (ft32_pass_by_reference): Likewise.
2118 (ft32_arg_partial_bytes): Likewise.
2119 (ft32_valid_pointer_mode): Likewise.
2120 (ft32_addr_space_pointer_mode): Likewise.
2121 (ft32_addr_space_legitimate_address_p): Likewise.
2122 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2124 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2125 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2126 (iamcu_alignment): Likewise.
2127 (canonicalize_vector_int_perm): Likewise.
2128 (ix86_noce_conversion_profitable_p): Likewise.
2129 (ix86_mpx_bound_mode): Likewise.
2130 (ix86_operands_ok_for_move_multiple): Likewise.
2131 * config/microblaze/microblaze-protos.h
2132 (microblaze_expand_conditional_branch_reg): Likewise.
2133 * config/microblaze/microblaze.c
2134 (microblaze_expand_conditional_branch_reg): Likewise.
2135 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2137 (rs6000_reassociation_width): Likewise.
2138 (rs6000_invalid_binary_op): Likewise.
2139 (fusion_p9_p): Likewise.
2140 (emit_fusion_p9_load): Likewise.
2141 (emit_fusion_p9_store): Likewise.
2142 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2144 (riscv_hard_regno_mode_ok_p): Likewise.
2145 (riscv_address_insns): Likewise.
2146 (riscv_split_symbol): Likewise.
2147 (riscv_legitimize_move): Likewise.
2148 (riscv_function_value): Likewise.
2149 (riscv_hard_regno_nregs): Likewise.
2150 (riscv_expand_builtin): Likewise.
2151 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2152 (riscv_build_integer): Likewise.
2153 (riscv_split_integer): Likewise.
2154 (riscv_legitimate_constant_p): Likewise.
2155 (riscv_cannot_force_const_mem): Likewise.
2156 (riscv_regno_mode_ok_for_base_p): Likewise.
2157 (riscv_valid_base_register_p): Likewise.
2158 (riscv_valid_offset_p): Likewise.
2159 (riscv_valid_lo_sum_p): Likewise.
2160 (riscv_classify_address): Likewise.
2161 (riscv_legitimate_address_p): Likewise.
2162 (riscv_address_insns): Likewise.
2163 (riscv_load_store_insns): Likewise.
2164 (riscv_force_binary): Likewise.
2165 (riscv_split_symbol): Likewise.
2166 (riscv_force_address): Likewise.
2167 (riscv_legitimize_address): Likewise.
2168 (riscv_move_integer): Likewise.
2169 (riscv_legitimize_const_move): Likewise.
2170 (riscv_legitimize_move): Likewise.
2171 (riscv_address_cost): Likewise.
2172 (riscv_subword): Likewise.
2173 (riscv_output_move): Likewise.
2174 (riscv_canonicalize_int_order_test): Likewise.
2175 (riscv_emit_int_order_test): Likewise.
2176 (riscv_function_arg_boundary): Likewise.
2177 (riscv_pass_mode_in_fpr_p): Likewise.
2178 (riscv_pass_fpr_single): Likewise.
2179 (riscv_pass_fpr_pair): Likewise.
2180 (riscv_get_arg_info): Likewise.
2181 (riscv_function_arg): Likewise.
2182 (riscv_function_arg_advance): Likewise.
2183 (riscv_arg_partial_bytes): Likewise.
2184 (riscv_function_value): Likewise.
2185 (riscv_pass_by_reference): Likewise.
2186 (riscv_setup_incoming_varargs): Likewise.
2187 (riscv_print_operand): Likewise.
2188 (riscv_elf_select_rtx_section): Likewise.
2189 (riscv_save_restore_reg): Likewise.
2190 (riscv_for_each_saved_reg): Likewise.
2191 (riscv_register_move_cost): Likewise.
2192 (riscv_hard_regno_mode_ok_p): Likewise.
2193 (riscv_hard_regno_nregs): Likewise.
2194 (riscv_class_max_nregs): Likewise.
2195 (riscv_memory_move_cost): Likewise.
2196 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2197 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2198 (rl78_addr_space_address_mode): Likewise.
2199 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2201 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2202 (rs6000_reassociation_width): Likewise.
2203 (rs6000_invalid_binary_op): Likewise.
2204 (fusion_p9_p): Likewise.
2205 (emit_fusion_p9_load): Likewise.
2206 (emit_fusion_p9_store): Likewise.
2207 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2208 (ok_for_simple_move_operands): Likewise.
2209 (ok_for_simple_move_strict_operands): Likewise.
2210 (ok_for_simple_arith_logic_operands): Likewise.
2211 (visium_legitimize_reload_address): Likewise.
2212 (visium_select_cc_mode): Likewise.
2213 (output_cbranch): Likewise.
2214 (visium_split_double_move): Likewise.
2215 (visium_expand_copysign): Likewise.
2216 (visium_expand_int_cstore): Likewise.
2217 (visium_expand_fp_cstore): Likewise.
2218 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2219 (visium_function_arg): Likewise.
2220 (visium_function_arg_advance): Likewise.
2221 (visium_libcall_value): Likewise.
2222 (visium_setup_incoming_varargs): Likewise.
2223 (visium_legitimate_constant_p): Likewise.
2224 (visium_legitimate_address_p): Likewise.
2225 (visium_legitimize_address): Likewise.
2226 (visium_secondary_reload): Likewise.
2227 (visium_register_move_cost): Likewise.
2228 (visium_memory_move_cost): Likewise.
2229 (prepare_move_operands): Likewise.
2230 (ok_for_simple_move_operands): Likewise.
2231 (ok_for_simple_move_strict_operands): Likewise.
2232 (ok_for_simple_arith_logic_operands): Likewise.
2233 (visium_function_value_1): Likewise.
2234 (rtx_ok_for_offset_p): Likewise.
2235 (visium_legitimize_reload_address): Likewise.
2236 (visium_split_double_move): Likewise.
2237 (visium_expand_copysign): Likewise.
2238 (visium_expand_int_cstore): Likewise.
2239 (visium_expand_fp_cstore): Likewise.
2240 (visium_split_cstore): Likewise.
2241 (visium_select_cc_mode): Likewise.
2242 (visium_split_cbranch): Likewise.
2243 (output_cbranch): Likewise.
2244 (visium_print_operand_address): Likewise.
2245 * expmed.c (flip_storage_order): Likewise.
2246 * expmed.h (emit_cstore): Likewise.
2247 (flip_storage_order): Likewise.
2248 * genrecog.c (validate_pattern): Likewise.
2249 * hsa-gen.c (gen_hsa_addr): Likewise.
2250 * internal-fn.c (expand_arith_overflow): Likewise.
2251 * ira-color.c (allocno_copy_cost_saving): Likewise.
2252 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2253 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2254 (process_invariant_for_inheritance): Likewise.
2255 * lra-eliminations.c (move_plus_up): Likewise.
2256 * omp-low.c (lower_oacc_reductions): Likewise.
2257 * simplify-rtx.c (simplify_subreg): Likewise.
2258 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2259 (TARGET_CHKP_BOUND_MODE): Likewise..
2260 * targhooks.c (default_chkp_bound_mode): Likewise.
2261 (default_setup_incoming_vararg_bounds): Likewise.
2262 * targhooks.h (default_chkp_bound_mode): Likewise.
2263 (default_setup_incoming_vararg_bounds): Likewise.
2264 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2265 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2266 (have_whole_vector_shift): Likewise.
2267 * tree-vect-stmts.c (vectorizable_load): Likewise.
2268 * doc/tm.texi: Regenerate.
2270 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2272 Graceful degrade if Binutils PR21472 is not available.
2275 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2276 .rodata in flash test fails.
2277 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2278 * confgure: Regenerate.
2279 * config.in: Regenerate.
2280 * config/avr/avr.c (avr_asm_named_section)
2281 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2282 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2283 (avr_asm_init_sections): Same.
2285 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2287 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2288 (fma<VH:mode>4_intrinsic): Likewise.
2289 (*fmsub<VCVTF:mode>4): Likewise.
2290 (*fmsub<VH:mode>4_intrinsic): Likewise.
2292 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2295 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2296 Don't depend on "optimize > 0".
2297 (out_movhi_r_mr, out_movqi_mr_r): Same.
2298 (out_movhi_mr_r, out_movqi_r_mr): Same.
2299 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2300 io_address_operand on "optimize > 0".
2302 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2304 * tree-loop-distribution.c: Add general explanantion on the pass.
2305 (generate_loops_for_partition): Mark distributed loop.
2306 (pg_add_dependence_edges): New parameter. Handle alias data
2307 dependence specially and record it in the parameter if asked.
2308 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2309 (init_partition_graph_vertices, add_partition_graph_edge): New.
2310 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2311 (free_partition_graph_vdata, build_partition_graph): New.
2312 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2313 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2314 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2315 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2316 (version_for_distribution_p, finalize_partitions): New.
2317 (distribute_loop): Handle alias data dependence specially. Factor
2318 out loop fusion code as functions and call these functions.
2320 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2322 * tree-loop-distribution.c (classify_partition): New parameter and
2323 better handle reduction statement.
2324 (rdg_build_partitions): Revise comment.
2325 (distribute_loop): Compute statements in all partitions and pass it
2326 to classify_partition.
2328 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2330 * tree-loop-distribution.c (enum partition_type): New.
2331 (struct partition): New field type.
2332 (partition_merge_into): Add parameter. Update partition type.
2333 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2334 (build_rdg_partition_for_vertex): Compute partition type.
2335 (rdg_build_partitions): Dump partition type.
2336 (distribute_loop): Update calls to partition_merge_into.
2338 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2340 * tree-loop-distribution.c (struct ddr_hasher): New.
2341 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2343 (classify_partition): Call get_data_dependence.
2344 (pg_add_dependence_edges): Ditto.
2345 (distribute_loop): Release data dependence hash table.
2347 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2349 * tree-loop-distribution.c (ref_base_address): Delete.
2350 (similar_memory_accesses): Rename ...
2351 (share_memory_accesses): ... to this. Check if partitions access
2352 the same memory reference.
2353 (distribute_loop): Call share_memory_accesses.
2355 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2357 * tree-loop-distribution.c (struct partition): New field recording
2359 (partition_alloc, partition_free): Init and release data refs.
2360 (partition_merge_into): Merge data refs.
2361 (build_rdg_partition_for_vertex): Collect data refs for partition.
2362 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2364 (distribute_loop): Remve data refs from vertice data of partition
2367 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2369 * tree-loop-distribution.c (params.h): Include header file.
2370 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2371 (datarefs_vec): New global var.
2372 (create_rdg_vertices): Use datarefs_vec directly.
2373 (free_rdg): Don't free data references.
2374 (build_rdg): Update use. Don't free data references.
2375 (distribute_loop): Compute global variable for data references.
2376 Bail out if there are too many data references.
2378 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2380 * tree-loop-distribution.c (loop_nest): New global var.
2381 (build_rdg): Use loop directly, rather than loop nest.
2382 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2384 (distribute_loop): Compute global variable loop nest. Update use.
2386 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2388 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2389 (partition_merge_into): New parameter. Dump reason for fusion.
2390 (distribute_loop): Update use of partition_merge_into.
2392 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2394 * tree-loop-distribution.c (bb_top_order_index): New.
2395 (bb_top_order_index_size, bb_top_order_cmp): New.
2396 (stmts_from_loop): Use topological order.
2397 (pass_loop_distribution::execute): Compute and release topological
2398 order for basic blocks.
2400 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2402 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2405 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2407 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2408 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2409 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2410 * internal-fn.def (LOOP_DIST_ALIAS): New.
2411 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2412 (fold_loop_internal_call): ... this.
2413 (vect_loop_dist_alias_call): New function.
2414 (set_uid_loop_bbs): Call fold_loop_internal_call.
2415 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2418 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2421 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2422 Require dead FLAGS_REG at the beginning of a peephole.
2424 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2427 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2428 arguments in the call to __builtin_ia32_sbb_u32.
2429 (_subborrow_u64): Swap _X and _Y arguments in the call to
2430 __builtin_ia32_sbb_u64.
2432 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2435 * tree-vrp.c (compare_assert_loc): Turn into a function template
2436 with stable template parameter. Only test if a->e is NULL,
2437 !a->e == !b->e has been verified already. Use e == NULL or
2438 e != NULL instead of e or ! e tests. If stable is true, don't use
2439 iterative_hash_expr, on the other side allow a or b or both NULL
2440 and sort the NULLs last.
2441 (process_assert_insertions): Sort using compare_assert_loc<false>
2442 instead of compare_assert_loc, later sort using
2443 compare_assert_loc<true> before calling process_assert_insertions_for
2444 in a loop. Use break instead of continue once seen NULL pointer.
2446 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2448 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2449 Cortex-R7 and Cortex-R8 processors.
2451 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2453 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2454 uninitialized while src is not.
2456 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2458 * common/config/arm/arm-common.c: Adjust include path for
2460 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2461 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2462 (arm-cpu-data.h): Likewise.
2463 (arm-cpu-cdata.h): Likewise.
2464 * config/arm/arm-cpu.h: Delete.
2465 * config/arm/arm-cpu-cdata.h: Delete.
2466 * config/arm/arm-cpu-data.h: Delete.
2468 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2470 * config/arm/arm-cpus.in (cortex-a55): New.
2471 (cortex-a75): Likewise.
2472 (cortex-a75.cortex-a55): Likewise.
2473 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2475 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2476 * config/arm/arm-cpu-cdata.h: Regenerate.
2477 * config/arm/arm-cpu-data.h: Regenerate.
2478 * config/arm/arm-cpu.h: Regenerate.
2479 * config/arm/arm-tables.opt: Regenerate.
2480 * config/arm/arm-tune.md: Regenerate.
2482 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2484 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2486 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2488 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2491 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2493 PR tree-optimization/81292
2494 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2495 full_string_p, also call adjust_related_strinfos if the adjustment
2496 is simple, otherwise invalidate related strinfos.
2498 2017-07-04 Martin Liska <mliska@suse.cz>
2501 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2502 newly created variable as DECL_IGNORED_P.
2504 2017-07-04 Martin Liska <mliska@suse.cz>
2507 * ipa-inline.c (inline_small_functions):
2508 Use xstrdup_for_dump.
2510 2017-07-04 Tom de Vries <tom@codesourcery.com>
2512 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2514 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2517 * config/darwin.c (darwin_function_switched_text_sections):
2520 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2522 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2524 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2526 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2528 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2530 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2531 min_profitable_iters, and th as inclusive lower bounds.
2532 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2533 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2534 for min_profitable_iters and min_profitable_estimate.
2535 (vect_transform_loop): Treat th as an inclusive lower bound.
2536 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2538 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2541 * config/darwin.c (darwin_function_switched_text_sections):
2542 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2543 in two pieces, and suppress the use of buf.
2545 2017-07-03 Nathan Sidwell <nathan@acm.org>
2547 * hash-table.h (hash_table_mod1): Fix indentation.
2549 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2552 * predict.c (force_edge_cold): Be more careful about propagation
2554 * profile-count.h (profile_probability::guessed,
2555 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2557 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2559 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2561 * doc/invoke.texi (rcpc architecture extension): Document it.
2563 2017-07-03 Richard Biener <rguenther@suse.de>
2565 PR tree-optimization/60510
2566 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2567 the scalar reduction PHI and use it.
2568 (vectorizable_reduction): Properly guard the single_defuse_cycle
2569 path for non-SLP reduction chains where we cannot use it.
2570 Rework reduc_def/index and vector type deduction. Rework
2571 vector operand gathering during reduction op code-gen.
2572 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2573 chains dissolve the chain and leave it to non-SLP reduction
2576 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2578 * tree-data-ref.h (dr_alignment): Declare.
2579 * tree-data-ref.c (dr_alignment): New function.
2580 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2581 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2583 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2585 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2587 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2588 and base_misalignment fields.
2589 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2590 * tree-data-ref.c: Include builtins.h.
2591 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2592 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2593 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2594 * tree-vect-data-refs.c: Include tree-cfg.h.
2595 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2596 fields instead of calculating an alignment here.
2597 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2598 innermost_loop_behavior fields.
2600 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2602 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2604 (DR_STEP_ALIGNMENT): New macro.
2605 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2606 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2607 (create_data_ref): Print it.
2608 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2609 to tell whether the step preserves vector (mis)alignment.
2610 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2611 Move the check for an integer step and generalise to all INTEGER_CST.
2612 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2613 Print the outer step alignment.
2615 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2617 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2618 with offset_alignment.
2619 (DR_ALIGNED_TO): Delete.
2620 (DR_OFFSET_ALIGNMENT): New macro.
2621 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2622 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2623 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2624 (create_data_ref): Likewise.
2625 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2626 (vect_analyze_data_refs): Likewise.
2627 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2628 creating dummy innermost behavior.
2630 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2632 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2633 with a "innermost_loop_behavior *" and refeence tree.
2634 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2635 (create_data_ref): Update call accordingly.
2636 * tree-predcom.c (find_looparound_phi): Likewise.
2638 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2640 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2641 fields with dr_wrt_vec_loop.
2642 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2643 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2644 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2645 (vect_dr_behavior): New function.
2646 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2647 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2648 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2649 track whether the step preserves the misalignment.
2650 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2651 Use vect_dr_behavior.
2652 (vect_setup_realignment): Update call accordingly.
2653 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2654 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2655 call to vect_create_addr_base_for_vector_ref.
2656 (vect_create_cond_for_align_checks): Likewise.
2657 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2658 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2659 (vect_recog_mask_conversion_pattern): Likewise.
2660 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2661 (new_stmt_vec_info): Remove redundant zeroing.
2663 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2665 * common/config/arm/arm-common.c (arm_be8_option): New function.
2666 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2667 (ISA_ARMv6): Add isa_bit_be8.
2668 * config/arm/arm.h (arm_be8_option): Add prototype.
2669 (BE8_SPEC_FUNCTION): New define.
2670 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2671 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2672 (mlittle-endian): Similarly.
2673 (mbe8, mbe32): New options.
2674 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2675 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2677 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2679 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2681 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2683 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2684 (cleanup_tree_cfg_bb): Use it.
2685 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2687 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2689 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2692 * loop-doloop.c (add_test): Update profile.
2694 2017-07-03 Martin Liska <mliska@suse.cz>
2697 * sanopt.c (rewrite_usage_of_param): New function.
2698 (sanitize_rewrite_addressable_params): Likewise.
2699 (pass_sanopt::execute): Call rewrite_usage_of_param.
2701 2017-07-03 Richard Biener <rguenther@suse.de>
2703 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2704 back to using VIEW_CONVERT_EXPR.
2706 2017-07-03 Martin Liska <mliska@suse.cz>
2709 * doc/extend.texi: Document when a resolver function is
2710 generated for target_clones.
2712 2017-07-03 Martin Liska <mliska@suse.cz>
2714 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2715 and shadow memory of auto variables which are subject of
2716 use-after-scope sanitization.
2717 (asan_expand_mark_ifn): Add do set only when is_poison.
2719 2016-07-03 Richard Biener <rguenther@suse.de>
2721 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2723 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2724 (vectorizable_reduction): Perform reduction PHI creation when
2725 visiting a reduction PHI and adjust and simplify code generation
2726 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2727 (vect_transform_loop): Visit reduction PHIs.
2728 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2729 defs into the SLP tree.
2730 (vect_build_slp_tree): Reduction defs terminate the recursion.
2731 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2733 (vect_get_vec_defs_for_stmt_copy): Export.
2734 (vect_get_vec_defs): Likewise.
2735 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2737 (vect_get_vec_defs_for_stmt_copy): Declare.
2738 (vect_get_vec_defs): Likewise.
2740 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2742 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2743 parameter with a "loop" parameter and use it instead of the
2744 loop containing DR_STMT. Don't check simple_iv when doing
2745 BB analysis. Describe the two analysis modes in the comment.
2747 2017-07-03 Tom de Vries <tom@codesourcery.com>
2749 PR tree-optimization/69468
2750 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2751 (find_same_succ_bb): Handle ignore_edge_flags.
2753 2017-07-03 Tom de Vries <tom@codesourcery.com>
2755 PR tree-optimization/81192
2756 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2758 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2760 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2762 2017-07-03 Tom de Vries <tom@codesourcery.com>
2764 PR tree-optimization/81192
2765 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2766 BB_SAME_SUCC (bb) == NULL.
2768 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2770 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2773 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2775 * dumpfile.c: Include profile-count.h
2776 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2778 (insert_cond_bb): Update profile.
2779 * tree-cfg.h (insert_cond_bb): Update prototype.
2780 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2781 * tree-dump.c: Do not include tree-cfg.
2783 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2785 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2787 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2789 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2792 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2794 * tree-complex.c (expand_complex_div_wide): update profile.
2796 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2797 Alan Hayward <alan.hayward@arm.com>
2798 David Sherwood <david.sherwood@arm.com>
2800 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2801 (CORETYPES_H): New define.
2802 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2803 (insn-modes-inline.h, s-modes-inline-h): New rules.
2804 (generated_files): Add insn-modes-inline.h.
2805 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2806 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2807 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2808 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2809 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2810 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2811 (build/gencodes.o, build/genconditions.o): Likewise.
2812 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
2813 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
2814 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
2815 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
2816 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
2817 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
2818 * coretypes.h: Include everything up to real.h for generators.
2819 Include insn-modes.h first. Include wide-int-print.h after
2820 wide-int.h. Include insn-modes-inline.h and then machmode.h.
2821 * machmode.h: Don't include insn-modes.h here.
2822 * function-tests.c: Remove includes of signop.h, machmode.h,
2823 double-int.h and wide-int.h.
2825 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
2827 * optc-save-gen.awk: Likewise.
2828 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
2829 * godump.c: Remove include of wide-int-print.h.
2830 * pretty-print.h: Likewise.
2831 * wide-int-print.cc: Likewise.
2832 * wide-int.cc: Likewise.
2833 * hash-map-tests.c: Remove include of signop.h.
2834 * hash-set-tests.c: Likewise.
2835 * rtl-tests.c: Likewise.
2836 * mkconfig.sh: Remove include of machmode.h.
2837 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
2839 (emit_insn_modes_inline_h): ...this new function. Emit the code
2840 into an insn-modes-inline.h header file, adding appropriate
2841 include guards and end comments.
2842 (emit_insn_modes_c_header): Remove include of machmode.h.
2843 (emit_min_insn_modes_c_header): Include coretypes.h rather than
2845 (main): Handle -i flag and call emit_insn_modes_inline_h when
2848 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2850 * tree-ssa-strlen.c (strinfo): Rename the length field to
2851 nonzero_chars. Add a full_string_p field.
2852 (compare_nonzero_chars, zero_length_string_p): New functions.
2853 (get_addr_stridx): Add an offset_out parameter.
2854 Use compare_nonzero_chars.
2855 (get_stridx): Update accordingly. Use compare_nonzero_chars.
2856 (new_strinfo): Update after above changes to strinfo.
2857 (set_endptr_and_length): Set full_string_p.
2858 (get_string_length): Update after above changes to strinfo.
2859 (unshare_strinfo): Update call to new_strinfo.
2860 (maybe_invalidate): Likewise.
2861 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
2862 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
2863 as a uhwi instead of an shwi. Update after above changes to
2864 strinfo and new_strinfo.
2865 (zero_length_string): Assert that chainsi contains full strings.
2866 Use zero_length_string_p. Update call to new_strinfo.
2867 (adjust_related_strinfos): Update after above changes to strinfo.
2868 Copy full_string_p from origsi.
2869 (adjust_last_stmt): Use zero_length_string_p.
2870 (handle_builtin_strlen): Update after above changes to strinfo and
2871 new_strinfo. Install the lhs as the string length if the previous
2872 entry didn't describe a full string.
2873 (handle_builtin_strchr): Update after above changes to strinfo
2875 (handle_builtin_strcpy): Likewise.
2876 (handle_builtin_strcat): Likewise.
2877 (handle_builtin_malloc): Likewise.
2878 (handle_pointer_plus): Likewise.
2879 (handle_builtin_memcpy): Likewise. Track nonzero characters
2880 that aren't necessarily followed by a nul terminator.
2881 (handle_char_store): Likewise.
2883 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2885 PR tree-optimization/80769
2886 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
2887 for malloc and calloc. Document the new invariant that all related
2888 strinfos have delayed lengths or none do.
2889 (verify_related_strinfos): Move earlier in file.
2890 (set_endptr_and_length): New function, split out from...
2891 (get_string_length): ...here. Also set the lengths of related
2893 (zero_length_string): Assert that chainsi has known (rather than
2895 (adjust_related_strinfos): Likewise.
2897 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2899 PR tree-optimization/81136
2900 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
2901 assert that two references with the same misalignment have the same
2902 compile-time misalignment if those compile-time misalignments
2905 2017-07-01 Andi Kleen <ak@linux.intel.com>
2907 * print-tree.c (print_node): Print all attributes.
2909 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2911 * cfg.c (scale_bbs_frequencies): New function.
2912 * cfg.h (scale_bbs_frequencies): Declare it.
2913 * cfgloopanal.c (single_likely_exit): Cleanup.
2914 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
2916 (scale_loop_profile): Likewise.
2917 (loop_version): Likewise.
2918 (create_empty_loop_on_edge): Update.
2919 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
2920 scale_loop_frequencies, scale_loop_profile, loopify,
2921 loop_version): Update prototypes.
2922 * modulo-sched.c (sms_schedule): Update.
2923 * predict.c (unlikely_executed_edge_p): Also check probability.
2924 (probably_never_executed_edge_p): Fix typo.
2925 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2926 * tree-parloops.c (gen_parallel_loop): Update.
2927 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
2928 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2929 * tree-ssa-loop-split.c (split_loop): Update.
2930 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2931 * tree-vect-loop-manip.c (vect_do_peeling): Update.
2932 (vect_loop_versioning): Update.
2933 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2935 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2937 * trans-mem.c (split_bb_make_tm_edge): Update profile.
2939 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2941 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
2942 to keep profile consistent.
2944 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2946 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
2947 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
2948 * profile-count.h (max_safe_multiplier): Make unsigned.
2949 (profile_count::guessed_zero): New.
2951 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2953 * bb-reorder.c (fix_up_crossing_landing_pad,
2954 fix_crossing_conditional_branches): Use make_single_succ_edge
2955 to keep profile consistent.
2957 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2959 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
2962 2017-07-01 Jakub Jelinek <jakub@redhat.com>
2965 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
2966 the right scopes, make sure cond_jump isn't preserved between multiple
2967 iterations. Search for fallthru edge whenever there are 3+ edges and
2968 use find_fallthru_edge for it.
2970 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2972 Patch by Alexander Monakov <amonakov@ispras.ru>
2973 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
2974 probabilities consistently.
2976 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2978 * pa.c (pa_expand_compare_and_swap_loop): Update call of
2979 emit_cmp_and_jump_insns.
2981 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2984 * tree-inline.c (expand_call_inline): Combine profile statuses.
2986 2017-06-30 Andrew Pinski <apinski@cavium.com>
2988 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
2989 fold_stmt returned true.
2991 2017-06-30 Nathan Sidwell <nathan@acm.org>
2993 * ggc.h (empty_string): Delete.
2994 * cfgexpand.c (expand_asm_stmt): Use plain "".
2995 * optabs.c (expand_asm_memory_barrier): Likewise.
2996 * stringpool.c (empty_string): Delete.
2997 (digit_vector, digit_string): Delete.
2998 (ggc_alloc_string): Use plain "", don't optimize single digit
2999 strings. Use ggc_alloc_atomic.
3001 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3003 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3004 comparison set and one other set, use the cost of the non-comparison
3007 2017-06-30 Nathan Sidwell <nathan@acm.org>
3009 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3012 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3014 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3015 loops. Remove now unneeded calls to gimple_switch_set_label() that
3016 just set removed labels to NULL_TREE.
3018 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3020 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3021 (set_range_info): ...here. Only call set_range_info_raw if domain
3023 (set_nonzero_bits): Call set_range_info_raw.
3024 * tree-ssanames.h (set_range_info_raw): New.
3026 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3029 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3030 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3031 of nonimmediate_operand and <store_mask_constraint> instead of m
3032 for the input operand. For V8FI iterator, always split if input
3033 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3034 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3035 <store_mask_predicate> instead of register_operand and
3036 <store_mask_constraint> instead of v for the input operand. Make
3037 sure both operands aren't MEMs for if not <mask_applied>.
3039 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3041 * lto-wrapper.c (copy_file) Close both file descriptors before
3044 2017-06-30 Martin Liska <mliska@suse.cz>
3047 * multiple_target.c (create_dispatcher_calls): Make ifunc
3048 also for function that don't have calls or are not referenced.
3050 2017-06-30 Richard Biener <rguenther@suse.de>
3052 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3053 analyze the first scalar stmt. Move vector type computation
3054 for the BB case here from ...
3055 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3056 live operation processing in the SLP case properly.
3058 2017-06-30 Richard Biener <rguenther@suse.de>
3060 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3062 2017-06-30 Martin Liska <mliska@suse.cz>
3065 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3066 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3068 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3070 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3073 2017-06-30 Richard Biener <rguenther@suse.de>
3075 PR tree-optimization/81249
3076 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3077 condition reduction result to original scalar type.
3079 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3081 * profile-count.h (enum profile_quality): Fix typos and whitespace
3084 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3086 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3087 type for branch probabilities.
3089 2017-06-29 Julian Brown <julian@codesourcery.com>
3090 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3092 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3093 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3094 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3095 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3097 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3099 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3100 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3101 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3102 CC usage from generic code to here.
3103 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3104 CC usage into the target macros.
3106 2017-06-29 Maya Rashish <coypu@sdf.org>
3108 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3111 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3113 * arm/arm-builtins.c: Include profile-count.h
3114 * except.c (sjlj_emit_function_enter): Use
3115 profile_probability::unlikely.
3117 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3119 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3120 and tocrel_offset be pointer args rather than implicitly using
3122 (legitimate_constant_pool_address_p, rs6000_emit_move,
3123 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3124 tocrel_offset and use in toc_relative_expr_p call.
3125 (print_operand, print_operand_address): Use static tocrel_base_oac
3126 and tocrel_offset_oac.
3127 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3130 2017-06-29 Maya Rashish <coypu@sdf.org>
3132 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3134 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3136 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3137 objects, take into account only the alignment of 'op0' and 'mode1' if
3140 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3142 * ccmp.c (ccmp_tree_comparison_p): New function.
3143 (ccmp_candidate_p): Update to use above function.
3144 (get_compare_parts): New function.
3145 (expand_ccmp_next): Update to use new functions.
3146 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3148 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3149 take mode as argument.
3150 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3151 * expr.c (expand_expr_real_1): Pass mode as argument.
3153 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3155 * combine.c (combine_instructions): Print insns to dump_file, together
3158 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3160 * asan.c (asan_emit_stack_protection): Update.
3161 (create_cond_insert_point): Update.
3162 * auto-profile.c (afdo_propagate_circuit): Update.
3163 * basic-block.h (struct edge_def): Turn probability to
3164 profile_probability.
3165 (EDGE_FREQUENCY): Update.
3166 * bb-reorder.c (find_traces_1_round): Update.
3167 (better_edge_p): Update.
3168 (sanitize_hot_paths): Update.
3169 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3170 (make_single_succ_edge): Update.
3171 (check_bb_profile): Update.
3172 (dump_edge_info): Update.
3173 (update_bb_profile_for_threading): Update.
3174 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3175 probabilitycount to 0.
3176 * cfgbuild.c (compute_outgoing_frequencies): Update.
3177 * cfgcleanup.c (try_forward_edges): Update.
3178 (outgoing_edges_match): Update.
3179 (try_crossjump_to_edge): Update.
3180 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3181 (expand_gimple_tailcall): Update.
3182 (construct_init_block): Use make_single_succ_edge.
3183 (construct_exit_block): Use make_single_succ_edge.
3184 * cfghooks.c (verify_flow_info): Update.
3185 (redirect_edge_succ_nodup): Update.
3186 (split_edge): Update.
3187 (account_profile_record): Update.
3188 * cfgloopanal.c (single_likely_exit): Update.
3189 * cfgloopmanip.c (scale_loop_profile): Update.
3190 (set_zero_probability): Remove.
3191 (duplicate_loop_to_header_edge): Update.
3192 * cfgloopmanip.h (loop_version): Update prototype.
3193 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3194 (force_nonfallthru_and_redirect): Update.
3195 (update_br_prob_note): Update.
3196 (rtl_verify_edges): Update.
3197 (purge_dead_edges): Update.
3198 (rtl_lv_add_condition_to_bb): Update.
3199 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3200 * cgraphunit.c (init_lowered_empty_function): Update.
3201 (cgraph_node::expand_thunk): Update.
3202 * cilk-common.c: Include profile-count.h
3203 * dojump.c (inv): Remove.
3204 (jumpifnot): Update.
3205 (jumpifnot_1): Update.
3206 (do_jump_1): Update.
3208 (do_jump_by_parts_greater_rtx): Update.
3209 (do_compare_rtx_and_jump): Update.
3210 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3211 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3212 * dwarf2cfi.c: Include profile-count.h
3213 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3214 (sjlj_emit_dispatch_table): Likewise.
3215 * explow.c: Include profile-count.h
3216 * expmed.c (emit_store_flag_force): Update.
3217 (do_cmp_and_jump): Update.
3218 * expr.c (compare_by_pieces_d::generate): Update.
3219 (compare_by_pieces_d::finish_mode): Update.
3220 (emit_block_move_via_loop): Update.
3221 (store_expr_with_bounds): Update.
3222 (store_constructor): Update.
3223 (expand_expr_real_2): Update.
3224 (expand_expr_real_1): Update.
3225 * expr.h (try_casesi, try_tablejump): Update prototypes.
3226 * gimple-pretty-print.c (dump_probability): Update.
3227 (dump_profile): New.
3228 (dump_gimple_label): Update.
3229 (dump_gimple_bb_header): Update.
3230 * graph.c (draw_cfg_node_succ_edges): Update.
3231 * hsa-gen.c (convert_switch_statements): Update.
3232 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3233 (find_if_case_1): Update.
3234 (find_if_case_2): Update.
3235 * internal-fn.c (expand_arith_overflow_result_store): Update.
3236 (expand_addsub_overflow): Update.
3237 (expand_neg_overflow): Update.
3238 (expand_mul_overflow): Update.
3239 (expand_vector_ubsan_overflow): Update.
3240 * ipa-cp.c (good_cloning_opportunity_p): Update.
3241 * ipa-split.c (split_function): Use make_single_succ_edge.
3242 * ipa-utils.c (ipa_merge_profiles): Update.
3243 * loop-doloop.c (add_test): Update.
3244 (doloop_modify): Update.
3245 * loop-unroll.c (compare_and_jump_seq): Update.
3246 (unroll_loop_runtime_iterations): Update.
3247 * lra-constraints.c (lra_inheritance): Update.
3248 * lto-streamer-in.c (input_cfg): Update.
3249 * lto-streamer-out.c (output_cfg): Update.
3250 * mcf.c (adjust_cfg_counts): Update.
3251 * modulo-sched.c (sms_schedule): Update.
3252 * omp-expand.c (expand_omp_for_init_counts): Update.
3253 (extract_omp_for_update_vars): Update.
3254 (expand_omp_ordered_sink): Update.
3255 (expand_omp_for_ordered_loops): Update.
3256 (expand_omp_for_generic): Update.
3257 (expand_omp_for_static_nochunk): Update.
3258 (expand_omp_for_static_chunk): Update.
3259 (expand_cilk_for): Update.
3260 (expand_omp_simd): Update.
3261 (expand_omp_taskloop_for_outer): Update.
3262 (expand_omp_taskloop_for_inner): Update.
3263 * omp-simd-clone.c (simd_clone_adjust): Update.
3264 * optabs.c (expand_doubleword_shift): Update.
3265 (expand_abs): Update.
3266 (emit_cmp_and_jump_insn_1): Update.
3267 (expand_compare_and_swap_loop): Update.
3268 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3269 * predict.c (predictable_edge_p): Update.
3270 (edge_probability_reliable_p): Update.
3271 (set_even_probabilities): Update.
3272 (combine_predictions_for_insn): Update.
3273 (combine_predictions_for_bb): Update.
3274 (propagate_freq): Update.
3275 (estimate_bb_frequencies): Update.
3276 (force_edge_cold): Update.
3277 * profile-count.c (profile_count::dump): Add missing space into dump.
3278 (profile_count::debug): Add newline.
3279 (profile_count::differs_from_p): Explicitly convert to unsigned.
3280 (profile_count::stream_in): Update.
3281 (profile_probability::dump): New member function.
3282 (profile_probability::debug): New member function.
3283 (profile_probability::differs_from_p): New member function.
3284 (profile_probability::differs_lot_from_p): New member function.
3285 (profile_probability::stream_in): New member function.
3286 (profile_probability::stream_out): New member function.
3287 * profile-count.h (profile_count_quality): Rename to ...
3288 (profile_quality): ... this one.
3289 (profile_probability): New.
3290 (profile_count): Update.
3291 * profile.c (compute_branch_probabilities): Update.
3292 * recog.c (peep2_attempt): Update.
3293 * sched-ebb.c (schedule_ebbs): Update.
3294 * sched-rgn.c (find_single_block_region): Update.
3295 (compute_dom_prob_ps): Update.
3296 (schedule_region): Update.
3297 * sel-sched-ir.c (compute_succs_info): Update.
3298 * stmt.c (struct case_node): Update.
3299 (do_jump_if_equal): Update.
3300 (get_outgoing_edge_probs): Update.
3301 (conditional_probability): Update.
3302 (emit_case_dispatch_table): Update.
3303 (expand_case): Update.
3304 (expand_sjlj_dispatch_table): Update.
3305 (emit_case_nodes): Update.
3306 * targhooks.c: Update.
3307 * tracer.c (better_p): Update.
3308 (find_best_successor): Update.
3309 * trans-mem.c (expand_transaction): Update.
3310 * tree-call-cdce.c: Update.
3311 * tree-cfg.c (gimple_split_edge): Upate.
3312 (move_sese_region_to_fn): Upate.
3313 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3314 * tree-eh.c (lower_resx): Upate.
3315 (cleanup_empty_eh_move_lp): Upate.
3316 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3317 * tree-inline.c (copy_edges_for_bb): Update.
3318 (copy_cfg_body): Update.
3319 * tree-parloops.c (gen_parallel_loop): Update.
3320 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3321 (gimple_gen_time_profiler): Update.
3322 * tree-ssa-dce.c (remove_dead_stmt): Update.
3323 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3324 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3325 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3326 (unloop_loops): Update.
3327 (try_peel_loop): Update.
3328 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3329 * tree-ssa-loop-split.c (connect_loops): Update.
3330 (split_loop): Update.
3331 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3332 (hoist_guard): Update.
3333 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3334 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3335 (value_replacement): Update.
3336 * tree-ssa-reassoc.c (branch_fixup): Update.
3337 * tree-ssa-tail-merge.c (replace_block_by): Update.
3338 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3339 (create_edge_and_update_destination_phis): Update.
3340 (compute_path_counts): Update.
3341 (recompute_probabilities): Update.
3342 (update_joiner_offpath_counts): Update.
3343 (freqs_to_counts_path): Update.
3344 (duplicate_thread_path): Update.
3345 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3346 (struct switch_conv_info): Update.
3347 (gen_inbound_check): Update.
3348 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3349 (vect_do_peeling): Update.
3350 (vect_loop_versioning): Update.
3351 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3352 (optimize_mask_stores): Update.
3353 * ubsan.c (ubsan_expand_null_ifn): Update.
3354 * value-prof.c (gimple_divmod_fixed_value): Update.
3355 (gimple_divmod_fixed_value_transform): Update.
3356 (gimple_mod_pow2): Update.
3357 (gimple_mod_pow2_value_transform): Update.
3358 (gimple_mod_subtract): Update.
3359 (gimple_mod_subtract_transform): Update.
3360 (gimple_ic): Update.
3361 (gimple_stringop_fixed_value): Update.
3362 (gimple_stringops_transform): Update.
3363 * value-prof.h: Update.
3365 2017-06-29 Carl Love <cel@us.ibm.com>
3367 * config/rs6000/rs6000-c.c: Add support for built-in functions
3368 vector signed int vec_signed (vector float);
3369 vector signed long long vec_signed (vector double);
3370 vector signed int vec_signed2 (vector double, vector double);
3371 vector signed int vec_signede (vector double);
3372 vector signed int vec_signedo (vector double);
3373 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3374 instruction generator.
3375 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3376 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3377 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3379 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3380 vunsignede_v2df): Add define_expands.
3381 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3382 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3383 VEC_UNSIGNEDO): Add definitions.
3384 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3385 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3386 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3387 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3388 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3389 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3390 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3391 vec_unsignede, vec_unsignedo): Add builtin defines.
3392 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3394 * doc/extend.texi: Update the built-in documentation file for the
3395 new built-in functions.
3397 2017-06-29 Richard Biener <rguenther@suse.de>
3399 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3400 reduction chains to LOOP_VINFO_REDUCTIONS.
3401 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3402 SLP reductions after processing reduction chains.
3404 2017-06-29 Nathan Sidwell <nathan@acm.org>
3406 * builtins.c (fold_builtin_FUNCTION): Use
3407 lang_hooks.decl_printable_name.
3409 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3412 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3413 with only one label.
3414 * stmt.c (expand_case): Assert NCASES is greater than one.
3416 2017-06-29 Richard Biener <rguenther@suse.de>
3418 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3420 (group_case_labels): Likewise.
3421 (find_taken_edge): Push sanity checking on val to workers...
3422 (find_taken_edge_cond_expr): ... here
3423 (find_taken_edge_switch_expr): ... and here, handle cases
3424 with just a default label.
3425 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3426 (group_case_labels): Likewise.
3427 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3428 group_case_labels does anything cleanup the CFG again.
3430 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3432 PR tree-optimization/81196
3433 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3434 exit condition comparing two IVs.
3436 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3438 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3439 profile to the dummy entry at the end of the list of architectures.
3440 * config/arm/arm-cpu-cdata.h: Regenerated.
3442 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3443 Michael Collison <michael.collison@arm.com>
3446 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3448 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3449 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3450 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3451 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3452 of shift when the shift amount is masked with constant equal to
3453 the size of the mode.
3454 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3457 2017-06-29 Martin Liska <mliska@suse.cz>
3459 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3461 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3464 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3467 (edge_growth_cache_entry::edge_growth_cache_entry): New
3469 (reset_edge_growth_cache): Update to use constructor.
3471 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3473 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3474 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3475 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3477 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3479 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3480 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3482 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3484 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3485 (*-linux-uclibc*): Add t-uclibc tmake_file.
3486 * config/t-musl: New.
3487 * config/t-uclibc: New.
3489 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3491 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3493 (gen_comm_data): Emit architectural setting of arch_prof.
3494 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3496 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3497 (armv8-m.base, armv8-m.main): Likewise.
3498 * arm-protos.h (arm_build_target): Add profile field.
3499 (arch_option): Likewise.
3500 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3502 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3503 arm_active_target.profile.
3505 2017-06-28 Richard Biener <rguenther@suse.de>
3508 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3509 TYPE_OVERFLOW_WRAPS.
3510 * match.pd (negate_expr_p): Likewise.
3511 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3512 fold_build2, not fold_binary.
3514 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3516 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3517 Convert memory address to Pmode.
3518 (aarch64_print_operand): Assert MEM operands are always Pmode.
3520 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3523 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3524 Remove redundant if.
3525 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3526 * config/arm/aarch-common-protos.h
3527 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3528 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3530 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3533 * multiple_target.c (create_dispatcher_calls): Set the default
3534 clone to be static, not public.
3536 2017-06-28 Richard Biener <rguenther@suse.de>
3538 * tree-vect-loop.c (vectorizable_reduction): Move special
3539 cond reduction IV var creation ...
3540 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3541 parameter. Use STMT_VINFO_VECTYPE.
3542 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3545 2017-06-28 Martin Liska <mliska@suse.cz>
3548 * ipa-visibility.c (non_local_p): Handle visibility.
3550 2017-06-28 Martin Liska <mliska@suse.cz>
3553 * common.opt: Add IntegerRange to various options.
3554 * opt-functions.awk (integer_range_info): New function.
3555 * optc-gen.awk: Add integer_range_info to cl_options struct.
3556 * opts-common.c (decode_cmdline_option): Handle
3557 CL_ERR_INT_RANGE_ARG.
3558 (cmdline_handle_error): Likewise.
3559 * opts.c (print_filtered_help): Show valid interval in
3560 when --help is provided.
3561 * opts.h (struct cl_option): Add range_min and range_max fields.
3562 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3564 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3566 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3567 (x * C EQ/NE y * C): New transformation.
3569 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3571 * genmultilib (combination_space): Accept '+' in option names.
3573 2017-06-28 Martin Liska <mliska@suse.cz>
3576 * asan.c (instrument_derefs): Bail out inner references
3577 that are hard register variables.
3579 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3582 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3583 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3585 2017-06-28 Richard Biener <rguenther@suse.de>
3587 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3588 (vect_get_slp_defs): Adjust.
3589 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3591 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3593 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3594 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3595 (vectorizable_reduction): Adjust.
3596 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3598 (vect_get_slp_defs): Likewise.
3599 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3600 (vectorizable_bswap): Adjust.
3601 (vectorizable_call): Likewise.
3602 (vectorizable_conversion): Likewise.
3603 (vectorizable_assignment): Likewise.
3604 (vectorizable_shift): Likewise.
3605 (vectorizable_operation): Likewise.
3606 (vectorizable_store): Likewise.
3607 (vectorizable_condition): Likewise.
3608 (vectorizable_comparison): Likewise.
3610 2017-06-28 Michael Collison <michael.collison@arm.com>
3613 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3615 (arm_gen_movmemqi): Removed unused variable 'i'.
3616 Convert 'for' loop into 'while' loop.
3617 (arm_expand_prologue): Remove last unnecessary set of insn.
3618 (thumb_pop): Remove unused variable 'pushed_words'.
3619 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3621 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3623 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3624 * config/s390/s390-protos.h: Add prototype of
3625 s390_rel_address_ok_p.
3626 * config/s390/s390.c (s390_got_symbol): New function.
3627 (s390_rel_address_ok_p): New function.
3628 (legitimize_pic_address): Use s390_rel_address_ok_p.
3629 (s390_load_got): Use s390_got_symbol.
3630 (s390_option_override): Issue error if
3631 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3632 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3634 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3636 2017-06-27 Andrew Pinski <apinski@cavium.com>
3638 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3639 (X * copysign (1.0, X)): New pattern.
3640 (X * copysign (1.0, -X)): New pattern.
3641 (copysign (-1.0, CST)): New pattern.
3643 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3645 * genmultilib (combination_space): Remove variable.
3646 Validate reuse rules against regular expression for any sequence
3647 of multilib options in any order.
3649 2017-06-27 Michael Collison <michael.collison@arm.com>
3651 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3652 call aarch64_split_simd_combine.
3653 * (aarch64_combine_internal<mode>): Delete pattern.
3654 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3655 Allow register and subreg operands.
3657 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3659 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3660 specific need, just fallback on defaults.
3661 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3663 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3664 Olivier Hainque <hainque@adacore.com>
3666 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3668 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3669 targets. Pick a default if no particular attempt applied.
3670 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3673 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3675 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3676 (x86_64-wrs-vxworks7): Likewise.
3678 2017-06-27 Marek Polacek <polacek@redhat.com>
3681 * ubsan.c (instrument_null): Check get_base_address's result for null.
3683 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3685 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3687 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3689 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3690 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3691 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3692 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3694 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3695 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3696 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3697 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3698 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3699 BUILT_IN_FEUPDATEENV): New builtins.
3700 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3701 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3702 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3703 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3705 (builtin_structptr_types): Adjust size.
3706 * tree.c (builtin_structptr_types): Add four entries.
3708 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3709 Olivier Hainque <hainque@adacore.com>
3711 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3712 (TLS_SYM): New local macro, forcing reference to __tls__ on
3713 link command lines for VxWorks 7 RTPs, triggering initialization
3715 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3716 OS features TLS support, true for RTPs on VxWorks 7.
3717 * config/vxworks.c (vxworks_override_options): Setup emutls
3720 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3722 * predict.c (test_prediction_value_range): Use -1U instead of -1
3723 to avoid narrowing conversion warning.
3724 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3725 to avoid narrowing conversion warning.
3726 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3728 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3730 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3732 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3733 64bit configurations.
3734 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3735 (SIZE_TYPE): Likewise.
3736 * config/vxworks.c (vxworks_emutls_var_fields): Use
3737 long_unsigned_type_node instead of unsigned_type_node as the offset
3738 field type, which is "pointer" mode in emutls.c.
3740 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3743 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3746 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3747 gimple_vuse copying separately from gimple_vdef copying.
3749 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3751 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3754 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3755 Olivier Hainque <hainque@adacore.com>
3757 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3758 for all vxworks7 targets.
3759 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3760 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3761 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3762 variations for VX6/VX7 and 32/64bits later on in ...
3763 (VXWORKS_LIB_SPEC): Leverage new macros.
3764 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3765 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3767 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3769 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3770 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3772 2017-06-26 Carl Love <cel@us.ibm.com>
3774 * config/rs6000/rs6000-c.c: Add support for built-in functions
3775 vector bool char vec_reve (vector bool char);
3776 vector signed char vec_reve (vector signed char);
3777 vector unsigned char vec_reve (vector unsigned char);
3778 vector bool int vec_reve (vector bool int);
3779 vector signed int vec_reve (vector signed int);
3780 vector unsigned int vec_reve (vector unsigned int);
3781 vector bool long long vec_reve (vector bool long long);
3782 vector signed long long vec_reve (vector signed long long);
3783 vector unsigned long long vec_reve (vector unsigned long long);
3784 vector bool short vec_reve (vector bool short);
3785 vector signed short vec_reve (vector signed short);
3786 vector double vec_reve (vector double);
3787 vector float vec_reve (vector float);
3788 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3789 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3790 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3791 (altivec_vreve): New pattern.
3792 * config/rs6000/altivec.h (vec_reve): New define.
3793 * doc/extend.texi (vec_rev): Update the built-in documentation file
3794 for the new built-in functions.
3796 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3798 PR tree-optimization/71815
3799 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3801 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3803 (slsr_process_phi): Likewise.
3804 (replace_uncond_cands_and_profitable_phis): Don't replace a
3805 multiply candidate with a stride of 1 (copy or cast).
3806 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3808 (lowest_cost_path): Likewise.
3809 (total_savings): Likewise.
3811 2017-06-26 Richard Biener <rguenther@suse.de>
3814 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
3815 Use def_builtin_pure for all gather builtins.
3817 2017-06-26 Richard Biener <rguenther@suse.de>
3819 PR tree-optimization/81203
3820 * tree-tailcall.c (find_tail_calls): Do not move stmts into
3823 2017-06-26 Marek Polacek <polacek@redhat.com>
3826 * doc/invoke.texi: Document -Wmultistatement-macros.
3828 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
3830 * doc/sourcebuild.texi (ARM-specific attributes): Document new
3831 arm_neon_ok_no_float_abi effective target.
3833 2017-06-26 Richard Biener <rguenther@suse.de>
3835 PR tree-optimization/80928
3836 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
3837 (copy_bbs): Set BB_DUPLICATED flag early.
3838 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
3840 (execute_on_shrinking_pred): Likewise.
3841 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
3842 BB_DUPLICATED blocks.
3843 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
3844 iterate over all PHIs considering removal of *gsi.
3846 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
3848 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
3851 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3853 * config/rs6000/rs6000-string.c: (expand_block_clear,
3854 do_load_for_compare, select_block_compare_mode,
3855 compute_current_alignment, expand_block_compare,
3856 expand_strncmp_align_check, expand_strn_compare,
3857 expand_block_move, rs6000_output_load_multiple)
3858 Move functions related to string/block move/compare
3860 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
3861 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
3862 for this function which is now used in two files.
3863 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
3864 * config.gcc: Add rs6000-string.o to extra_objs for
3865 targets powerpc*-*-* and rs6000*-*-*.
3867 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3870 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
3871 32-bit, since indexed is not valid for DImode.
3872 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
3873 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
3874 (define_peephole2 for Altivec d-form load): Add 32-bit support.
3875 (define_peephole2 for Altivec d-form store): Likewise.
3878 * multiple_target.c (create_dispatcher_calls): Only create the
3879 dispatcher call if the function is the default clone of a
3882 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
3885 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
3886 a call, force the call to not be a tail call.
3888 2017-06-23 Jeff Law <law@redhat.com>
3890 * doc/contrib.texi: Add entry for Steven Pemberton's work on
3893 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
3895 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
3896 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
3897 handling for early expansion of vector shifts (sl,sr,sra,rl).
3898 (builtin_function_type): Add vector shift right instructions
3899 to the unsigned argument list.
3901 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3903 rtl-optimizatoin/79286
3904 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
3905 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
3906 trap. PIC register plus a const unspec without offset can never trap.
3908 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
3910 * tree.h (builtin_structptr_type): New type.
3911 (builtin_structptr_types): Declare new array.
3912 * tree.c (builtin_structptr_types): New array.
3913 (free_lang_data, build_common_tree_nodes): Use it.
3915 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
3918 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
3921 2017-06-22 Matt Turner <mattst88@gmail.com>
3923 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
3924 Lake models to skylake case. Assume skylake for unknown
3925 models with clflushopt.
3927 2017-06-22 Jeff Law <law@redhat.com>
3929 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
3930 frame sizes that do not satisfy aarch64_uimm12_shift.
3932 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
3934 * profile-count.h (apply_probability,
3935 apply_scale, probability_in): Fix checks for zero.
3937 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3939 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
3940 * doc/cppdiropts.texi (-I @var{dir}): Document it.
3942 2016-06-22 Richard Biener <rguenther@suse.de>
3944 * tree-vect-loop.c (vect_model_reduction_cost): Handle
3945 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
3946 REDUC_MAX_EXPR support.
3947 (vectorizable_reduction): Likewise.
3948 (vect_create_epilog_for_reduction): Likewise.
3950 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
3952 * match.pd (A / (1 << B) -> A >> B): New.
3953 * generic-match-head.c: Include optabs-tree.h.
3954 * gimple-match-head.c: Likewise.
3955 * optabs-tree.h (target_supports_op_p): New.
3956 * optabs-tree.c (target_supports_op_p): New.
3958 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3960 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
3961 $gcc_cv_ld --help output.
3962 (gcc_cv_ld_demangle): Likewise.
3963 (gcc_cv_ld_eh_frame_hdr): Likewise.
3964 (gcc_cv_ld_pie): Likewise.
3965 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
3966 (gcc_cv_ld_buildid): Likewise.
3967 (gcc_cv_ld_sysroot): Likewise.
3968 (ld_bndplt_support): Likewise.
3969 (ld_pushpopstate_support): Likewise.
3970 * configure: Regenerate.
3971 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
3973 2017-06-21 Jakub Jelinek <jakub@redhat.com>
3976 * config/i386/sse.md (round<mode>2): Renumber match_dup and
3977 operands indexes to avoid gap between operands and match_dups.
3979 2017-06-21 Andrew Pinski <apinski@cavium.com>
3981 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
3982 Increment Arith_shift and Arith_shift_reg by 1.
3983 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
3985 * config/aarch64/aarch64.c (thunderx_tunings): Enable
3986 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
3987 (aarch64_strip_extend): Add new argument and test for it.
3988 (aarch64_cheap_mult_shift_p): New function.
3989 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
3990 add a cost if it is true.
3991 Update calls to aarch64_strip_extend.
3992 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
3994 2017-06-21 Andrew Pinski <apinski@cavium.com>
3996 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
3998 (thunderxt88): Likewise.
3999 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4000 (thunderx_prefetch_tune): New variable.
4001 (thunderx2t99_prefetch_tune): Update for the correct values.
4002 (thunderxt88_tunings): New variable.
4003 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4004 generic_prefetch_tune.
4005 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4007 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4009 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4010 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4011 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4012 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4013 (aarch64_atomic_cas<mode>, GPI): Likewise.
4015 2017-06-21 Martin Liska <mliska@suse.cz>
4017 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4018 statements on cold and hot labels.
4019 * predict.c (tree_estimate_probability_bb): Remove the
4020 prediction from this place.
4022 2017-06-21 Martin Liska <mliska@suse.cz>
4024 PR tree-optimization/79489
4025 * gimplify.c (maybe_add_early_return_predict_stmt): New
4027 (gimplify_return_expr): Call the function.
4028 * predict.c (tree_estimate_probability_bb): Remove handling
4030 * predict.def: Update comment about early return predictor.
4031 * gimple-predict.h (is_gimple_predict): New function.
4032 * predict.def: Change default value of early return to 66.
4033 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4035 * passes.def: Put pass_strip_predict_hints to the beginning of
4038 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4040 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4041 FUNCTION_DECL declarations.
4042 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4044 (dwaf2out_decl): Likewise.
4045 * godump.c (go_early_global_decl): Skip call to the real debug hook
4046 for FUNCTION_DECL declarations.
4047 * passes.c (rest_of_decl_compilation): Skip call to the
4048 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4049 -fdump-go-spec is passed.
4051 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4053 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4054 Reorder for compactness.
4055 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4056 (def_builtin_pure, def_builtin_pure2): New functions.
4057 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4059 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4061 * match.pd (nop_convert): New predicate.
4062 ((A +- CST1) +- CST2): Allow some NOP conversions.
4064 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4067 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4068 with ctors/dtors if GOVD_SHARED is set.
4070 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4072 * config/aarch64/aarch64.md (movti_aarch64):
4073 Emit mov rather than orr.
4074 (movtf_aarch64): Likewise.
4075 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4076 Emit mov rather than orr.
4078 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4080 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4081 Swap alternatives, make integer dup more expensive.
4083 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4085 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4086 Return true for non-tls symbols.
4088 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4090 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4091 (cortex-a75): Likewise.
4092 (cortex-a75.cortex-a55): Likewise.
4093 * config/aarch64/aarch64-tune.md: Regenerate.
4094 * doc/invoke.texi (-mtune): Document new values for -mtune.
4096 2017-06-21 Tom de Vries <tom@codesourcery.com>
4098 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4100 (Effective-Target Keywords, Other attributes): Suggest using
4101 dg-add-options stack_size feature to get stack limit in stack_size
4102 effective target documentation.
4104 2017-06-21 Julian Brown <julian@codesourcery.com>
4105 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4107 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4108 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4109 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4111 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4112 attribute type list for neon_multiply.
4113 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4114 attribute type list for neon_multiply.
4115 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4116 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4117 attribute type list for neon_multiply.
4118 * config/arm/types.md (crypto_pmull): Add.
4119 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4120 attribute type list.
4122 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4124 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4127 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4129 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4130 to make sure not to dereference a NULL cost_classes_ptr pointer.
4132 2017-06-20 Carl Love <cel@us.ibm.com>
4134 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4135 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4136 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4137 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4138 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4139 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4140 VMULOSW): New enum "unspec" values.
4141 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4142 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4143 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4144 altivec_vmulosw): New patterns.
4145 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4146 VMULOSW): Add definitions.
4148 2017-06-20 Julia Koval <julia.koval@intel.com>
4150 * config/i386/i386.c: Fix rounding expand for new pattern.
4151 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4153 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4155 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4156 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4158 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4160 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4163 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4165 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4166 architecture, then by implementer ID.
4167 * config/aarch64/aarch64-tune.md: Regenerate.
4169 2017-06-20 Richard Biener <rguenther@suse.de>
4172 * fold-const.c (split_tree): Fold to type before negating.
4174 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4176 * diagnostic-show-locus.c
4177 (selftest::test_fixit_deletion_affecting_newline): New function.
4178 (selftest::diagnostic_show_locus_c_tests): Call it.
4180 2017-06-20 Andreas Schwab <schwab@suse.de>
4183 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4186 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4188 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4189 __ARM_FEATURE_COPROC according to support.
4191 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4193 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4194 Rewritten to avoid overflow for > 32-bit pointers.
4197 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4198 by removing enum keyword.
4199 (ubsan_type_descriptor): Likewise. Formatting fix.
4202 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4203 splitter): Require TARGET_SSE2 in the condition.
4205 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4208 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4209 for doing vector set of SFmode on ISA 3.0.
4210 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4211 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4213 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4214 SFmode value into a V4SF variable that was extracted from another
4215 V4SF variable without converting the element to double precision
4216 and back to single precision vector format.
4217 (vsx_insert_extract_v4sf_p9_2): Likewise.
4219 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4221 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4222 in UWHI to avoid undefined overflow.
4225 * ubsan.h (enum ubsan_encode_value_phase): New.
4226 (ubsan_encode_value): Change second argument to
4227 enum ubsan_encode_value_phase with default value of
4228 UBSAN_ENCODE_VALUE_GENERIC.
4229 * ubsan.c (ubsan_encode_value): Change second argument to
4230 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4231 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4232 create_tmp_var_raw instead of create_tmp_var and use a
4234 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4235 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4236 ubsan_encode_value callers.
4239 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4240 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4241 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4243 2017-06-19 Richard Biener <rguenther@suse.de>
4246 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4247 estimates if we changed anything.
4249 2017-06-19 Richard Biener <rguenther@suse.de>
4251 PR tree-optimization/80887
4252 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4253 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4254 simplified lookups, then reset mprts_hook.
4255 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4257 (try_to_simplify): Likewise.
4259 2017-06-19 Martin Liska <mliska@suse.cz>
4262 * gimplify.c (gimplify_switch_expr):
4263 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4265 2017-06-19 Martin Liska <mliska@suse.cz>
4267 * doc/install.texi: Document that PGO runs in 4 stages.
4269 2017-06-19 Martin Liska <mliska@suse.cz>
4272 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4273 to dispatcher function name.
4274 * multiple_target.c (replace_function_decl): New function.
4275 (create_dispatcher_calls): Redirect both edges and references.
4277 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4279 * profile-count.c (profile_count::dump): Dump quality.
4280 (profile_count::differs_from_p): Update for unsigned val.
4281 * profile-count.h (profile_count_quality): New enum.
4282 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4284 2017-06-19 Richard Biener <rguenther@suse.de>
4286 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4287 struct function as arg.
4288 (estimate_numbers_of_iterations): Export overload with loop arg.
4289 (free_numbers_of_iterations_estimates_loop): Use an overload of
4290 free_numbers_of_iterations_estimates instead.
4291 * tree-cfg.c (remove_bb): Adjust.
4292 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4293 * tree-parloops.c (gen_parallel_loop): Likewise.
4294 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4296 (tree_unroll_loops_completely): Likewise.
4297 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4298 Use an overload instead and export.
4299 (estimated_loop_iterations): Adjust.
4300 (max_loop_iterations): Likewise.
4301 (likely_max_loop_iterations): Likewise.
4302 (estimate_numbers_of_iterations): Take struct function as arg
4304 (loop_exits_before_overflow): Adjust.
4305 (free_numbers_of_iterations_estimates_loop): Use an overload.
4306 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4307 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4309 2017-06-19 Richard Biener <rguenther@suse.de>
4312 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4313 RANGE_EXPR conservatively.
4315 2017-06-16 Carl Love <cel@us.ibm.com>
4317 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4318 definitions for vec_float, vec_float2, vec_floato,
4319 vec_floate built-ins.
4320 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4321 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4323 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4324 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4325 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4326 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4327 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4328 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4329 vec_floato): Add builtin defines.
4330 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4331 Update the built-in documentation file for the new built-in
4334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4336 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4337 (mthumb): Mark as the negative of -marm.
4339 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4341 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4343 (ARM Options, -mtune): Document that this accepts the same
4344 extension options as -mcpu.
4345 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4347 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4349 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4350 permitted extensions.
4352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4354 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4355 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4356 (armv8-m.main): Add option +nodsp.
4357 * config/arm/arm-cpu-cdata.h: Regenerated.
4359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4361 * config/arm/t-fuchsia: New file.
4362 * config.gcc (arm*-*-fuchsia*): Use it.
4364 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4366 * config/arm/t-symbian: Rewrite for new option infrastructure.
4368 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4370 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4371 (MULTILIB_REQUIRED): Likewise.
4373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4375 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4376 (MULTILIB_RESUE): Likewise.
4377 (MULTILIB_MATCHES): Likewise.
4378 (MULTLIB_REQUIRED): Likewise.
4380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4382 * config/arm/t-rtems: Rewrite for new option framework.
4384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4386 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4387 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4388 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4389 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4390 * config/arm/t-multilib: ... here.
4391 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4392 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4393 armv7-a and armv8*-a when A-profile libraries have not been built.
4394 * config/arm/t-rmprofile: Rewrite.
4396 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4398 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4399 with a backslash. Remove the backslash after substituting unescaped
4401 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4405 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4406 the driver to pass through the default setting of -mfloat-abi.
4407 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4409 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4410 (all_feat_combs): New rule.
4411 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4413 * config/arm/t-aprofile: Rewrite.
4415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4417 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4418 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4419 fpu is not specified by the user/command-line.
4420 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4421 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4422 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4423 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4424 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4425 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4427 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4429 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4430 * config/arm/t-arm-elf: Rewritten.
4432 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4434 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4435 have some floating-point instructions.
4436 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4437 (TARGET_MAYBE_HARD_FLOAT): New macro.
4438 * config/arm/arm-builtins.c (arm_init_builtins): Use
4439 TARGET_MAYBE_HARD_FLOAT.
4440 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4444 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4445 (configargs.h): Include it.
4446 (arm_print_hint_for_fpu_option): New function.
4447 (arm_parse_fpu_option): New function.
4448 (candidate_extension): New class.
4449 (arm_canon_for_multilib): New function.
4450 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4451 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4452 (ARCH_CANONICAL_SPECS): New macro.
4453 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4457 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4458 are set after handling multilib fragments. Set target_cpu_default2
4461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4463 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4465 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4466 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4468 (check_arch): Likewise.
4469 * config/arm/arm.c (arm_configure_build_target): Handle
4470 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4471 options in the default.
4473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4475 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4476 when an option is an alias of another.
4477 * config/arm/parsecpu.awk (optalias): New parser token.
4478 (gen_comm_data): Mark non-alias options as such. Emit entries
4479 for extension aliases.
4480 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4481 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4482 (armv6kz, armv6zk, armv6t2): Likewise.
4483 (armv7): Make vfpv3-d16 an alias.
4484 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4486 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4487 Sort in canonical order.
4488 (armv8-a): Sort in canonical order.
4489 (armv8.1-a, armv8.2-a): Likewise.
4490 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4492 (cortex-a9): Sort in canonical order.
4493 * config/arm/arm.c (selftests.h): Include it.
4494 (arm_test_cpu_arch_data): New function.
4495 (arm_run_self_tests): New function.
4496 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4497 (targetm): Move declaration to the end of the file.
4498 * arm-cpu-cdata.h: Regenerated.
4500 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4502 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4503 call to target_mode_check describing the type of option passed.
4504 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4505 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4506 arm_parse_cpu_option_name to match parameters against list of
4508 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4509 arm_arch_core_flags data structure.
4510 * config/arm/arm-cpu_cdata.h: Regenerated.
4512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4514 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4516 (arm_print_hint_for_cpu_option): Likewise.
4517 (arm_print_hint_for_arch_option): Likewise.
4518 (arm_parse_cpu_option_name): Likewise.
4519 (arm_parse_arch_option_name): Likewise.
4520 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4521 of entries in the all_fpus list.
4522 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4523 (arm_parse_cpu_option_name): Declare.
4524 (arm_parse_arch_option_name): Declare.
4525 (arm_parse_option_features): Declare.
4526 (arm_intialize_isa): Declare.
4527 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4529 (gen_comm_data): ... here. Make definitions non-static.
4530 * config/arm/arm-cpu-data.h: Regenerated.
4531 * config/arm/arm-cpu-cdata.h: Regenerated.
4533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4535 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4536 (cpu_arch_extension): New structure.
4537 (cpu_arch_option, arch_option, cpu_option): New structures.
4538 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4540 (gen_data): Generate new format data tables.
4541 * config/arm/arm.c (cpu_tune): New structure.
4542 (cpu_option, processors): Delete.
4543 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4544 (arm_print_hint_for_cpu_option): ... this and ...
4545 (arm_print_hint_for_arch_option): ... this.
4546 (arm_parse_arch_cpu_name): Delete. Replace with ...
4547 (arm_parse_cpu_option_name): ... this and ...
4548 (arm_parse_arch_option_name): ... this.
4549 (arm_unrecognized_feature): Change type of target parameter to
4551 (arm_parse_arch_cpu_features): Delete. Replace with ...
4552 (arm_parse_option_features): ... this.
4553 (arm_configure_build_target): Rework to use new configuration data
4555 (arm_print_tune_info): Rework for new configuration data tables.
4556 * config/arm/arm-cpu-data.h: Regenerated.
4557 * config/arm/arm-cpu.h: Regenerated.
4559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4561 * Makefile.in (OBJS): Move sbitmap.o from here ...
4562 (OBJS-libcommon): ... to here.
4564 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4566 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4567 (ISA_ALL_CRYPTO): New macro.
4568 (ISA_ALL_SIMD): New macro
4569 (ISA_ALL_FP): New macro.
4570 * config/arm/arm.c (fpu_bitlist): Update initializer.
4571 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4573 (arm9e): Add fpu. Add option for nofp
4574 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4575 (arm926ej-s, arm1026ej-s): Likewise.
4576 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4577 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4578 neon-fp16, neon-vfpv4, nofp and nosimd.
4579 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4580 (cortex-a8): Add fpu. Add option for nofp.
4581 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4582 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4583 (cortex-r4f): Add fpu.
4584 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4585 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4587 (cortex-r8): Likewise.
4588 (cortex-m4): Add fpu. Add option for nofp.
4589 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4590 (cortex-a17.cortex-a7): Likewise.
4591 (cortex-a32): Add fpu. Add options for crypto and nofp.
4592 (cortex-a35, cortex-a53): Likewise.
4593 (cortex-a57): Add fpu. Add option for crypto.
4594 (cortex-a72, cortex-a73): Likewise.
4595 (exynos-m1): Likewise.
4596 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4597 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4598 (cortex-m33): Add fpu. Add option for nofp.
4599 * config/arm/arm-cpu-cdata.h: Regenerated
4600 * config/arm/arm-cpu-data.h: Regenerated.
4602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4604 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4605 (armv5te, armv5tej): Likewise.
4606 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4607 (armv7): Add options fp and vfpv3-d16.
4608 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4609 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4611 (armv7ve): Likewise.
4612 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4613 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4614 (armv8-a): Add nocrypto option.
4615 (armv8.1-a, armv8.2-a): Likewise.
4616 (armv8-m.main): add options fp, fp.dp and nofp.
4618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4620 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4622 (armv8-a+crc): Delete.
4623 (armv8.1-a): Add options simd, crypto and nofp.
4624 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4625 (armv8.2-a+fp16): Delete.
4626 (armv8-m.main): Add option dsp.
4627 (armv8-m.main+dsp): Delete.
4628 (cortex-a8): Add fpu. Add nofp option.
4629 (cortex-a9): Add fpu. Add nofp and nosimd options.
4630 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4631 link to main cpu and architecture data structures.
4632 (gen_comm_data): Only put isa attributes from the main architecture
4634 (option): New statement for architecture and CPU entries.
4635 * arm.c (struct cpu_option): New structure.
4636 (struct processors): Add entry for options.
4637 (arm_unrecognized_feature): New function.
4638 (arm_parse_arch_cpu_name): Ignore any characters after the first
4640 (arm_parse_arch_cpu_feature): New function.
4641 (arm_configure_build_target): Separate out any CPU and architecture
4642 features and parse separately. Don't error out if -mfpu=auto is
4643 used with only an architecture string.
4644 (arm_print_asm_arch_directives): New function.
4645 (arm_file_start): Call it.
4646 * config/arm/arm-cpu-cdata.h: Regenerated.
4647 * config/arm/arm-cpu-data.h: Likewise.
4648 * config/arm/arm-tables.opt: Likewise.
4650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4652 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4653 assembler when it is not -mfpu=auto.
4655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4657 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4658 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4659 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4660 (ASM_CPU_SPEC): Rewrite.
4661 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4662 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4663 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4664 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4665 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4666 copied string is NUL-terminated. Also strip any characters prefixed
4668 (arm_rewrite_selected_arch): New function.
4669 (arm_rewrite_march): New function.
4671 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4673 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4674 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4675 (march, mcpu, mtune): Convert to string-based options.
4676 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4677 (arm_parse_arch_cpu_name): New function.
4678 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4679 identify selected architecture or CPU.
4680 (arm_option_save): New function.
4681 (TARGET_OPTION_SAVE): Redefine.
4682 (arm_option_restore): Restore string options.
4683 (arm_option_print): Print string options.
4685 2017-06-16 Martin Sebor <msebor@redhat.com>
4687 PR tree-optimization/80933
4688 PR tree-optimization/80934
4689 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4690 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4691 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4692 (gimple_fold_builtin): Call them.
4694 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4696 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4697 as unlikely; update profile.
4699 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4701 * predict.c (force_edge_cold): Handle declaring edges impossible
4704 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4706 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4708 (try_unroll_loop_completely): Fix reporting.
4710 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4712 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4714 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4717 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4718 if given a non-constant argument for an intrinsic which requires a
4721 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4723 * profile.c (compare_freqs): New function.
4724 (branch_prob): Sort edge list.
4725 (find_spanning_tree): Assume that the list is priority sorted.
4727 2017-06-16 Richard Biener <rguenther@suse.de>
4729 PR tree-optimization/81090
4730 * passes.def (pass_record_bounds): Remove.
4731 * tree-pass.h (make_pass_record_bounds): Likewise.
4732 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4733 make_pass_record_bounds): Likewise.
4734 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4735 not free niter estimates at the beginning but at the end.
4736 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4738 2017-06-16 Richard Biener <rguenther@suse.de>
4740 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4741 initializer to workaround ICE in host GCC 4.8.
4743 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4745 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4747 (clone_inlined_nodes): Update.
4749 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4751 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4752 prefetch settings, and enable prefetching by default at -O3.
4754 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4756 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4757 Set flag_prefetch_loop_arrays according to tuning data.
4759 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4761 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4763 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4764 [Unrelated to main purpose of the patch] Place the pointer field last
4765 to enable type checking errors when tune structure are wrongly merged.
4766 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4767 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4768 (thunderx2t99_prefetch_tune): New tune constants.
4769 (tune_params *_tunings): Update all tunings (no functional change).
4770 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4771 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4772 from tunings structures.
4774 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4777 * ubsan.c (instrument_null): Add T argument, use it instead
4778 of computing it based on IS_LHS.
4779 (instrument_object_size): Likewise.
4780 (pass_ubsan::execute): Adjust instrument_null and
4781 instrument_object_size callers to pass gimple_get_lhs or
4782 gimple_assign_rhs1 result to it. Use instrument_null instead of
4783 calling get_base_address and instrument_mem_ref. Handle
4784 aggregate call arguments for object-size sanitization.
4786 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4788 PR tree-optimization/81089
4789 * tree-vrp.c (is_masked_range_test): Validate operands of
4792 2017-06-15 Martin Sebor <msebor@redhat.com>
4795 * dumpfile.c (dump_register): Avoid calling memset to initialize
4796 a class with a default ctor.
4797 * gcc.c (struct compiler): Remove const qualification.
4798 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4799 * hash-table.h: Ditto.
4800 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4802 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4803 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4805 * params.h (struct param_info): Make struct members non-const.
4806 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4807 with default initialization.
4808 * vec.h (vec_copy_construct, vec_default_construct): New helper
4810 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4811 with vec_copy_construct.
4812 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
4813 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
4814 * doc/invoke.texi (-Wclass-memaccess): Document.
4816 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4818 * emit-rtl.h (is_leaf): Update comment about local
4821 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4824 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
4825 for a variable to have a section before checking if the section has a
4827 Set section to.persistent if persistent attribute is set.
4828 Warn if .persistent attribute is used on an automatic variable.
4830 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
4832 PR rtl-optimization/80474
4833 * reorg.c (update_block): Do not ignore instructions in a delay slot.
4835 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
4837 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
4840 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
4842 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
4843 (casesi): Emit bounds checking as RTL.
4844 (casesi_internal_mips16_<mode>): Remove bounds checking.
4846 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
4848 * config/xtensa/xtensa.c (xtensa_option_override): Append
4849 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
4850 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
4851 xtensa_doloop_hooks): Define unconditionally.
4852 (xtensa_reorg_loops): Only call reorg_loops in the presence of
4854 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
4855 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
4856 for it in xtensa_option_override.
4857 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
4858 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
4860 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
4862 * doc/cppopts.texi: Document '-' special value to -MF.
4864 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
4866 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
4867 (cortex_a53_fconst): Likewise.
4868 (cortex_a53_fpmul): Likewise.
4869 (cortex_a53_f_load_64): Likewise.
4870 (cortex_a53_f_load_many): Likewise.
4871 (cortex_a53_advsimd_alu): Likewise.
4872 (cortex_a53_advsimd_alu_q): Likewise.
4873 (cortex_a53_advsimd_mul): Likewise.
4874 (cortex_a53_advsimd_mul_q): Likewise.
4875 (fpmac bypass): Add new bypass for fpmac-fpmac case.
4876 Add missing fmul, r2f_cvt and fconst cases.
4878 2017-06-14 Richard Biener <rguenther@suse.de>
4881 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
4883 (fold_binary_loc): When associating do not treat pre-existing
4884 TREE_OVERFLOW on literal constants as a reason to allow
4885 TREE_OVERFLOW on associated literal constants.
4887 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4889 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
4890 (MASK_FEATURES): New macro.
4891 * config/sparc/sparc.c (sparc_option_override): Remove the special
4892 handling of -mfpu and generalize it to all MASK_FEATURES switches.
4894 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4896 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
4897 a division of 0 if non-call exceptions are enabled.
4899 2017-06-14 Andrew Pinski <apinski@cavium.com>
4900 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4903 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
4904 Improve vector initialization code gen for only variable case.
4906 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4908 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
4910 2017-06-14 Richard Biener <rguenther@suse.de>
4912 PR tree-optimization/81083
4913 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
4916 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4918 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
4919 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
4920 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
4921 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
4922 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
4923 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
4925 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4927 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
4928 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
4930 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4932 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
4934 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4936 * config/rs6000/t-rtems: Don't handle SPE.
4938 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4940 * config/rs6000/t-linux: Don't handle SPE.
4942 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4944 * config/rs6000/eabispe.h: Delete file.
4946 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4948 * config/rs6000/t-spe: Delete file.
4950 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4952 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
4953 (rs6000_legitimate_offset_address_p): Return false for anything in
4954 V2SImode or V2SFmode.
4956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4958 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
4959 except V2SF and V2SI. Rearrange the vector modes, and add comments.
4960 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
4962 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
4963 (rs6000_legitimate_offset_address_p): Ditto.
4964 (rs6000_emit_move): Ditto.
4965 (rs6000_init_builtins): Remove V4HI_type_node.
4967 2017-06-13 Martin Liska <mliska@suse.cz>
4970 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
4971 (gate_asan): Likewise.
4972 * asan.h (asan_no_sanitize_address_p): Remove the function.
4973 (sanitize_flags_p): New function.
4974 * builtins.def: Fix coding style.
4975 * common.opt: Use renamed enum value.
4976 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
4977 * doc/extend.texi: Document no_sanitize attribute.
4978 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
4979 to SANITIZE_UNDEFINED_NONDEFAULT.
4980 * gcc.c (sanitize_spec_function): Use the renamed enum value.
4981 * gimple-fold.c (optimize_atomic_compare_exchange_p):
4982 Use sanitize_flags_p.
4983 * gimplify.c (gimplify_function_tree): Likewise.
4984 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
4985 * opts.c (parse_no_sanitize_attribute): New function.
4986 (common_handle_option): Use renamed enum value.
4987 * opts.h (parse_no_sanitize_attribute): Declare.
4988 * tree.c (sanitize_flags_p): New function.
4989 * tree.h: Declared here.
4990 * tsan.c: Use sanitize_flags_p.
4991 * ubsan.c (ubsan_expand_null_ifn): Likewise.
4992 (instrument_mem_ref): Likewise.
4993 (instrument_bool_enum_load): Likewise.
4994 (do_ubsan_in_current_function): Remove the function.
4995 (pass_ubsan::execute): Use sanitize_flags_p.
4996 * ubsan.h: Remove do_ubsan_in_current_function
4997 * tree-cfg.c (print_no_sanitize_attr_value): New function.
4998 (dump_function_to_file): Use it here.
5000 2017-06-13 Martin Jambor <mjambor@suse.cz>
5002 PR tree-optimization/80803
5003 PR tree-optimization/81063
5004 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5005 (propagate_subaccesses_across_link): Enqueue subtree whenever
5006 necessary instead of relying on the caller.
5008 2017-06-13 Martin Jambor <mjambor@suse.cz>
5010 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5011 that have a first_link.
5012 (sort_and_splice_var_accesses): Do not check first_link before
5014 (subtree_mark_written_and_enqueue): Likewise.
5015 (propagate_all_subaccesses): Likewise and do not stop at first
5016 parent with a first_link.
5018 2017-06-13 Martin Jambor <mjambor@suse.cz>
5020 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5023 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5025 * match.pd: New pattern.
5027 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5029 * tree-vrp.c (is_masked_range_test): New function.
5030 (register_edge_assert_for): Determine ranges for
5033 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5035 PR tree-optimization/67328
5036 * fold-const.c (maskable_range_p): New function.
5037 (build_range_check): Generate bittests if possible.
5039 2017-06-13 Martin Liska <mliska@suse.cz>
5041 * gimple-pretty-print.c (dump_probability): Add new argument.
5042 (dump_edge_probability): Dump both probability and count.
5043 (dump_gimple_label): Likewise.
5044 (dump_gimple_bb_header): Likewise.
5046 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5049 * config/avr/avr-devices.c: Fix indentation.
5050 * config/avr/gen-avr-mmcu-specs.c: Dito.
5052 2017-06-13 Richard Biener <rguenther@suse.de>
5054 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5055 instead get vector type from stmt_info.
5056 (vectorizable_reduction): Adjust. Remove dead code.
5058 2017-06-13 Richard Biener <rguenther@suse.de>
5061 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5062 case of C * (x * C2 + C3).
5063 (fold_addr_of_array_ref_difference): Properly fold index difference.
5065 2017-06-12 David S. Miller <davem@davemloft.net>
5068 * config/sparc/sparc.md (return expander): Emit frame blockage if
5069 function uses alloca.
5071 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5073 * combine.c (make_field_assignment): Check len rather than the mode
5074 precision when calling force_to_mode.
5076 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5078 Support multilibs and devices that see flash in RAM address range.
5081 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5082 (avr_mcu_t) <flash_pm_offset>: New field.
5083 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5084 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5085 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5086 (AVR_TINY_PM_OFFSET): Remove macro.
5087 * config/avr/avr.opt (-mshort-calls): New option.
5088 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5089 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5090 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5091 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5092 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5093 instead of avr_arch->have_jmp_call.
5094 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5095 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5096 avr_arch->flash_pm_offset to define.
5097 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5098 new field flash_pm_offset. Add entry for avrxmega3.
5099 (avr_texinfo): Add entry for avrxmega3.
5100 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5101 attiny212, attiny214,
5102 attiny412, attiny414, attiny416, attiny417,
5103 attiny814, attiny816, attiny817,
5104 attiny1614, attiny1616, attiny1617,
5105 attiny3214, attiny3216, attiny3217.
5106 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5107 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5108 (avr_print_operand_address) [AVR_TINY]: Same.
5109 (avr_asm_init_sections) <readonly_data_section>: Only patch
5110 callback if avr_arch->flash_pm_offset = 0.
5111 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5112 for rodata if avr_arch->flash_pm_offset != 0.
5113 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5114 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5115 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5116 (m_options): Append opt_rcall.
5117 (m_dirnames): Append dir_rcall.
5118 * config/avr/t-multilib: Regenerate.
5120 * configure.ac [target=avr]: Check whether avrxmega3 default
5121 linker description file works as needed.
5122 * configure: Regenerate.
5123 * doc/avr-mmcu.texi: Regenerate.
5124 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5125 <__AVR_ARCH__>: Document avrxmega3 and 103.
5126 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5127 <__AVR_SHORT_CALLS__>: Document it.
5128 <__AVR_PM_BASE_ADDRESS__>: Document it.
5129 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5130 (AVR Variable Attributes) <progmem>: Document this is
5131 not needed for avrxmega3.
5132 (AVR Named Address Spaces) <__flash>: Dito.
5134 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5136 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5138 2017-06-12 Doug Rupp <rupp@adacore.com>
5140 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5141 Append vxworks-stdint.h to the tm_file list.
5142 * config/vxworks-stdint.h: New file.
5144 2017-06-12 Martin Liska <mliska@suse.cz>
5146 PR tree-optimization/81041
5147 * tree-profile.c (gimple_gen_ic_func_profiler):
5148 Create an extra BB in profile-generate
5149 (gimple_gen_time_profiler): Likewise.
5151 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5153 PR tree-optimization/81003
5154 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5155 (update_range_test): Use it instead of force_gimple_operand_gsi.
5157 2017-06-12 Richard Biener <rguenther@suse.de>
5159 PR tree-optimization/81053
5160 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5161 with backedge value not defined in loop. Simplify def stmt
5164 2017-06-11 Tom de Vries <tom@codesourcery.com>
5167 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5169 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5170 nvptx_cannot_force_const_mem.
5172 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5174 * opts.c (finish_options): Move test for flag_split_stack after
5175 it has been initialized.
5177 2017-06-11 Jason Merrill <jason@redhat.com>
5179 * tree.h (id_equal): New.
5180 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5181 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5182 instead of strcmp of IDENTIFIER_POINTER.
5184 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5186 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5187 (mark_all_inlined_calls_cdtor): Fix formating.
5188 (inline_transform): Rescale profile before inlining.
5190 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5192 * cgraph.h (cgraph_edge::clone): Update prototype.
5193 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5194 (cgraph_node::create_clone): Update.
5195 (cgraph_node::create_version_clone): Update.
5196 * tree-inline.c (copy_bb): Update.
5197 (expand_call_inline): Update.
5199 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5201 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5202 factored out from ...
5203 (rs6000_emit_prologue): ... here.
5205 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5207 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5208 factored out from ...
5209 (rs6000_emit_prologue): ... here.
5211 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5213 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5215 (handle_missing_profiles): Fix computation of tp_first_run.
5216 (counts_to_freqs): Do not touch freqs when count is 0.
5218 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5220 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5223 2017-06-10 Tom de Vries <tom@codesourcery.com>
5225 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5226 attributes): Document signal effective target.
5228 2017-06-10 Tom de Vries <tom@codesourcery.com>
5230 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5231 Document effective target stack_size.
5233 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5235 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5236 to the edit_context if they can be auto-applied.
5238 2017-06-9 Ian Lance Taylor <iant@golang.org>
5240 * opts.c (finish_options): If -fsplit-stack, disable implicit
5241 -forder-blocks-and-partition.
5242 * doc/invoke.texi (Optimize Options): Document that when using
5243 -fsplit-stack -forder-blocks-and-partition is not implicitly
5246 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5248 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5249 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5250 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5251 * builtins.def (abort, trap, unreachable): Declare cold.
5252 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5253 * tree-core.h (ECF_COLD): New.
5254 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5255 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5257 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5259 * predict.c (unlikely_executed_stmt_p): Cleanup.
5261 2017-06-09 Richard Biener <rguenther@suse.de>
5263 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5264 model if the ref is always written to.
5266 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5268 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5270 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5272 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5275 2017-06-09 Tom de Vries <tom@codesourcery.com>
5278 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5279 "target cannot support label values" when encountering LABEL_REF.
5281 2017-06-09 Martin Liska <mliska@suse.cz>
5283 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5284 (gimple_gen_ic_func_profiler): Emit direct comparison
5285 of __gcov_indirect_call_callee with NULL.
5286 (gimple_gen_time_profiler): Change probability from
5287 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5289 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5291 * profile.c (edge_gcov_counts): Turn to pointer.
5292 (compute_branch_probabilities, compute_branch_probabilities): Update.
5293 (branch_prob): Do not clear edge_gcov_count.
5294 * profile.h (edge_gcov_counts): Turn to pointer.
5295 (edge_gcov_count): Update.
5297 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5299 * gimple.h (gimple_check_failed): Mark cold.
5301 2017-06-09 Richard Biener <rguenther@suse.de>
5303 PR tree-optimization/66623
5304 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5305 refactor check_reduction into two parts, properly computing
5306 whether we have to check reduction validity for outer loop
5309 2017-06-09 Richard Biener <rguenther@suse.de>
5311 PR tree-optimization/79483
5312 * graphite-scop-detection.c (order): New global.
5313 (get_order): Compute bb to order mapping that satisfies code
5314 generation constraints.
5315 (cmp_pbbs): New helper.
5316 (build_scops): Start domwalk at entry block, sort generated
5319 2017-06-09 Richard Biener <rguenther@suse.de>
5322 * ipa-polymorphic-call.c
5323 (ipa_polymorphic_call_context::restrict_to_inner_class):
5324 Skip FIELD_DECLs with error_mark_node type.
5325 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5328 2017-06-09 Martin Liska <mliska@suse.cz>
5330 * predict.c (struct branch_predictor): New struct.
5331 (test_prediction_value_range): New test.
5332 (predict_c_tests): New function.
5333 * selftest-run-tests.c (selftest::run_tests): Run the function.
5334 * selftest.h: Declare new tests.
5336 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5339 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5340 gen_add3_insn did not fail.
5341 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5342 r0, construct that number in a temporary reg and add that reg to r0.
5343 If asked to put the result in r0 as well, fail.
5345 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5347 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5348 for early expansion of vec_eqv.
5350 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5353 * ubsan.c (instrument_null): Avoid pointless code temporary.
5354 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5356 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5360 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5363 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5366 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5368 * predict.c (unlikely_executed_edge_p): Move ahead.
5369 (probably_never_executed_edge_p): Use it.
5371 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5374 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5375 gimple_call_builtin_p call.
5377 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5379 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5380 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5381 rtl_check_failed_type2, rtl_check_failed_code1,
5382 rtl_check_failed_code2, rtl_check_failed_code_mode,
5383 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5384 rtvec_check_failed_bounds, rtl_check_failed_flag,
5385 _fatal_insn_not_found, _fatal_insn): Likewise.
5386 * tree.h (tree_contains_struct_check_failed,
5387 tree_check_failed, tree_not_check_failed,
5388 tree_class_check_failed, tree_range_check_failed,
5389 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5390 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5391 tree_operand_check_failed, omp_clause_check_failed,
5392 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5395 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5397 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5398 flag_branch_probabilities.
5399 * ipa-inline.c (edge_badness): Likewise.
5400 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5401 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5402 * predict.c (maybe_hot_frequency_p): Likewise.
5403 (probably_never_executed): Likewise.
5404 * sched-ebb.c (schedule_ebbs): Likewise.
5405 * sched-rgn.c (find_single_block_region): Likewise.
5406 * tracer.c (tail_duplicate): Likewise.
5408 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5410 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5411 longer requires x_flag_profile_use.
5413 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5415 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5416 instead of flag_reorder_blocks_and_partition.
5417 * dbxout.c (dbxout_function_end): Likewise.
5418 * dwarf2out.c (gen_subprogram_die): Likewise.
5419 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5420 * hw-doloop.c (reorg_loops): Likewise.
5421 * varasm.c (assemble_start_function,
5422 assemble_end_function): Likewise.
5423 (decide_function_section): Do not check for
5424 flag_reorder_blocks_and_partition.
5426 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5428 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5430 (chkp_get_hard_register_fake_addr_expr): Ditto.
5431 (chkp_build_addr_expr): Add check for hard reg case.
5432 (chkp_parse_array_and_component_ref): Ditto.
5433 (chkp_find_bounds_1): Ditto.
5434 (chkp_process_stmt): Don't generate bounds store for
5437 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5439 * predict.c (maybe_hot_bb_p): Do not check profile status.
5440 (maybe_hot_edge_p): Likewise.
5441 (probably_never_executed): Check for zero counts even if profile
5443 (unlikely_executed_edge_p): New function.
5444 (unlikely_executed_stmt_p): New function.
5445 (unlikely_executed_bb_p): New function.
5446 (set_even_probabilities): Use unlikely predicates.
5447 (combine_predictions_for_bb): Likewise.
5448 (predict_paths_for_bb): Likewise.
5449 (predict_paths_leading_to_edge): Likewise.
5450 (determine_unlikely_bbs): New function.
5451 (estimate_bb_frequencies): Use it.
5452 (compute_function_frequency): Use zero counts even if profile is
5454 * profile-count.h: Fix typo.
5456 2017-08-08 Julia Koval <julia.koval@intel.com>
5458 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5459 _mm512_mask_cvtsepi16_storeu_epi8,
5460 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5461 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5462 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5463 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5464 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5465 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5466 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5467 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5468 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5469 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5470 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5471 __builtin_ia32_pmovuswb256mem_mask,
5472 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5473 __builtin_ia32_pmovwb512mem_mask): New builtins.
5475 2017-08-08 Julia Koval <julia.koval@intel.com>
5477 PR target/73350,80862
5478 * config/i386/subst.md (round): Fix round pattern.
5479 * config/i386/i386.c (ix86_erase_embedded_rounding):
5480 Fix erasing rounding for the fixed pattern.
5482 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5484 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5486 2017-06-08 Martin Liska <mliska@suse.cz>
5488 PR gcov-profile/80911
5489 * gcov.c (block_info::block_info): New constructor.
5491 2017-06-07 Carl Love <cel@us.ibm.com>
5493 * config/rs6000/rs6000-c: The return type of the following
5494 built-in functions was implemented as int not long long. Fix sign
5495 of return value for the unsigned version of vec_mulo and vec_mule.
5496 vector unsigned long long vec_bperm (vector unsigned long long,
5497 vector unsigned char)
5498 vector signed long long vec_mule (vector signed int,
5500 vector unsigned long long vec_mule (vector unsigned int,
5501 vector unsigned int)
5502 vector signed long long vec_mulo (vector signed int,
5504 vector unsigned long long vec_mulo (vector unsigned int,
5505 vector unsigned int)
5506 * doc/extend.texi: Fix the documentation for the built-in
5509 2017-06-07 Carl Love <cel@us.ibm.com>
5512 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5515 2017-06-07 Carl Love <cel@us.ibm.com>
5517 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5518 support, Generate doublehv for signed int/float for BE case only.
5520 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5522 * doc/invoke.texi (mcx16): Rewrite.
5524 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5526 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5527 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5528 *mov<mode>_softfloat, and an anonymous splitter): Use
5529 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5531 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5533 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5535 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5536 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5537 (rs6000_debug_reg_global): Adjust.
5538 (rs6000_init_hard_regno_mode_ok): Adjust.
5539 (rs6000_dbx_register_number): Adjust.
5540 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5541 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5542 Remove SPE_ACC and SPEFSCR.
5543 (REG_ALLOC_ORDER): Ditto.
5544 (FRAME_POINTER_REGNUM): Change to 111.
5545 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5546 (REG_CLASS_NAMES): Ditto.
5547 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5548 (REGISTER_NAMES): Ditto.
5549 (ADDITIONAL_REG_NAMES): Ditto.
5550 (rs6000_reg_names): Ditto.
5551 * config/rs6000/rs6000.md: Renumber some register number
5554 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5556 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5558 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5559 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5561 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5562 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5563 Delete the SPE high registers.
5564 (REG_ALLOC_ORDER): Ditto.
5565 (enum reg_class): Remove SPE_HIGH_REGS.
5566 (REG_CLASS_NAMES): Ditto.
5567 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5568 (REGISTER_NAMES): Ditto.
5569 (rs6000_reg_names): Ditto.
5570 * doc/tm.texi.in: Remove SPE as example.
5571 * doc/tm.texi: Regenerate.
5573 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5575 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5576 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5577 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5578 * config/rs6000/rs6000.md (type): Remove "brinc".
5580 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5582 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5583 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5584 * config/rs6000/linuxspe.h: Delete file.
5585 * config/rs6000/rs6000.md: Don't include spe.md.
5586 * config/rs6000/spe.h: Delete file.
5587 * config/rs6000/spe.md: Delete file.
5588 * config/rs6000/t-rs6000: Remove spe.md.
5590 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5592 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5593 (reg_or_none500mem_operand): Delete.
5594 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5595 instead of reg_or_none500mem_operand.
5597 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5599 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5600 handling of SPE flags.
5601 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5603 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5605 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5607 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5608 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5609 paired_divv2sf3): Similar.
5610 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5611 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5612 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5614 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5615 Rename the paired_* instruction patterns.
5616 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5618 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5619 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5620 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5621 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5622 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5624 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5625 (spe_func_has_64bit_regs_p): Delete.
5626 (spe_expand_predicate_builtin): Delete.
5627 (spe_expand_evsel_builtin): Delete.
5628 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5629 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5630 (invalid_e500_subreg): Delete.
5631 (rs6000_legitimize_address): Always force_reg op2 as well, for
5632 paired single memory accesses.
5633 (rs6000_member_type_forces_blk): Delete.
5634 (rs6000_spe_function_arg): Delete.
5635 (rs6000_expand_unop_builtin): Delete SPE handling.
5636 (rs6000_expand_binop_builtin): Ditto.
5637 (spe_expand_stv_builtin): Delete.
5638 (bdesc_2arg_spe): Delete.
5639 (spe_expand_builtin): Delete.
5640 (spe_expand_predicate_builtin): Delete.
5641 (spe_expand_evsel_builtin): Delete.
5642 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5643 (spe_init_builtins): Delete.
5644 (spe_func_has_64bit_regs_p): Delete.
5645 (savres_routine_name): Delete "info" parameter. Adjust callers.
5646 (rs6000_emit_stack_reset): Ditto.
5647 (rs6000_dwarf_register_span): Delete.
5648 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5649 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5650 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5652 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5654 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5655 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5656 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5657 mulv2sf3, divv2sf3): Delete expanders.
5659 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5661 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5663 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5665 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5666 * config/rs6000/rs6000.c: Ditto.
5668 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5670 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5671 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5672 comparison_operator.
5674 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5676 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5677 * config/rs6000/rs6000.opt: Ditto.
5678 * config/rs6000/t-rtems: Ditto.
5680 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5682 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5683 TARGET_E500_SINGLE by 0, simplify.
5684 * config/rs6000/rs6000.c: Ditto.
5685 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5686 (spe_build_register_parallel): Delete.
5687 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5688 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5689 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5690 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5691 (E500_CONVERT): Delete.
5692 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5696 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5697 * config/rs6000/dfp.md: Ditto.
5698 (negdd2, *negdd2_fpr): Merge.
5699 (absdd2, *absdd2_fpr): Merge.
5700 (negtd2, *negtd2_fpr): Merge.
5701 (abstd2, *abstd2_fpr): Merge.
5702 * config/rs6000/e500.h: Delete file.
5703 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5704 TARGET_FPRS by 1 and simplify.
5705 * config/rs6000/rs6000-c.c: Ditto.
5706 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5708 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5710 * config/rs6000/rs6000.md: Ditto.
5711 (floatdidf2, *floatdidf2_fpr): Merge.
5712 (move_from_CR_gt_bit): Delete.
5713 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5714 (E500_CR_IOR_COMPARE): Delete.
5715 (All patterns that require !TARGET_FPRS): Delete.
5716 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5718 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5720 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5722 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5724 * graphds.c (add_edge): Intitialize edge's attached data.
5725 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5726 pointer parameter. Call pointed function on each edge during
5727 graph traversing. Skip traversing the edge when the function
5729 (graphds_dfs, graphds_scc): Ditto.
5730 (for_each_edge): New parameter. Pass the new parameter to callback
5732 * graphds.h (skip_edge_callback): New function pointer type.
5733 (graphds_dfs, graphds_scc): New function pointer parameter.
5734 (graphds_edge_callback, for_each_edge): New parameter.
5736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5738 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5739 out code checking if runtime alias check is possible to below ...
5740 Call the new function.
5741 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5742 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5744 2017-06-07 Marek Polacek <polacek@redhat.com>
5747 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5748 TYPE_OVERFLOW_WRAPS check.
5750 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5752 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5753 if versioning is required.
5754 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5755 peeling with the check for versioning.
5757 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5759 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5760 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5761 Set true to new parameter if new ssa variable is defined.
5762 (vect_gen_vector_loop_niters): Refactor. Set range information
5763 for the new vector loop bound variable.
5764 (vect_do_peeling): Ditto.
5766 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5768 * tree-affine.c (ssa.h): Include header file.
5769 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5770 has wrapping overflow behavior.
5772 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5774 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5776 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5778 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5779 (tree_to_aff_combination): ... here.
5781 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5783 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5784 reg_pressure model function.
5785 (ivopts_global_cost_for_size): Delete.
5786 (determine_set_costs, iv_ca_recount_cost): Call new model function
5787 ivopts_estimate_reg_pressure.
5789 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5791 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5792 expensive than udiv. Remove floating point cases from mod.
5794 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5796 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5799 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5801 * config/aarch64/aarch64.md
5802 (copysignsf3): Fix mask generation.
5804 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5806 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5808 (class dump_manager): Add register_dumps method.
5809 * dumpfile.c: Include langhooks.h.
5810 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5811 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
5812 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
5813 (dump_manager::dump_register): Start with 512 entries instead of 32.
5814 (dump_manager::register_dumps): New method.
5815 * toplev.c (general_init): Instead of invoking register_dumps
5816 langhook, invoke register_dumps method on the dump manager.
5817 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
5820 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
5822 * doc/md.texi: Clarify the restrictions on a define_insn condition.
5823 Say that # requires an associated define_split to exist, and that
5824 the define_split must be suitable for use after register allocation.
5826 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5828 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
5829 (compute_outgoing_frequencies): Also initialize zero counts.
5830 (find_many_sub_basic_blocks): Do not produce uninitialized profile
5831 around loops; preserve more of profile when nothing changes.
5833 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
5835 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
5837 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
5838 * config/arm/arm-cpu-cdata.h: Regenerate.
5839 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
5840 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
5841 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
5842 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
5843 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
5845 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
5847 * config/arm/t-rmprofile: Likewise.
5848 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
5850 2017-06-06 David S. Miller <davem@davemloft.net>
5853 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
5854 blockage if function uses alloca.
5856 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5858 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
5859 New "uid" fields to hold pretty-print IDs of group and ref.
5860 Memory references are now identified as <group_id>:<ref_id>
5861 instead of using [random] addresses.
5862 (dump_mem_details): Simplify, no functional change.
5863 (dump_mem_ref): Simplify and make output more concise.
5864 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
5865 (find_or_create_group): Initialize group uid.
5866 (record_ref): Initialize ref uid. Improve debug output.
5867 (prune_group_by_reuse, should_issue_prefetch_p,)
5868 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
5869 (mark_nontemporal_store, determine_loop_nest_reuse):
5870 Improve debug output.
5872 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5874 * dbgcnt.def (prefetch): New debug counter.
5875 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
5876 (schedule_prefetches): Stop issueing prefetches if debug counter
5879 2017-06-06 Tom de Vries <tom@codesourcery.com>
5881 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
5882 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
5884 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5886 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
5887 Use aarch64_reg_or_zero predicate for operand 4.
5888 (aarch64_compare_and_swap<mode> define_insn_and_split):
5889 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
5890 (aarch64_store_exclusive<mode>): Likewise for operand 2.
5892 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5894 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
5895 (arm_compute_save_core_reg_mask): This.
5896 (thumb1_compute_save_reg_mask): Rename into ...
5897 (thumb1_compute_save_core_reg_mask): This.
5898 (arm_compute_save_reg0_reg12_mask): Adapt comment.
5899 (arm_compute_frame_layout): Likewise.
5901 2017-06-06 Richard Biener <rguenther@suse.de>
5903 PR tree-optimization/80974
5904 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
5905 keep or clear leaders SSA info.
5907 2017-06-06 Tom de Vries <tom@codesourcery.com>
5909 * config/nvptx/nvptx.c (split_mode_p): New function.
5910 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
5912 2017-06-06 Tom de Vries <tom@codesourcery.com>
5914 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
5916 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5919 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
5922 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5924 * shrink-wrap.c (handle_simple_exit): Update profile.
5925 (try_shrink_wrapping): Upate profile.
5927 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5929 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
5930 (tree_guess_outgoing_edge_probabilities): New.
5931 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
5932 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
5934 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5936 * ipa-split.c (split_function): Initialize return bb profile.
5938 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5940 * profile.c (compute_branch_probabilities): Also initialize
5943 2017-06-06 Richard Biener <rguenther@suse.de>
5945 PR tree-optimization/80928
5946 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
5947 (vect_analyze_loop_operations): Properly guard analysis for
5949 (vect_transform_loop): Likewise.
5950 (vect_analyze_loop_2): Also reset SLP type on PHIs.
5951 (vect_model_induction_cost): Do not cost for pure SLP.
5952 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
5953 of induction in inner loop vectorization.
5954 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
5955 (vect_get_and_check_slp_defs): Handle vect_induction_def.
5956 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
5958 (vect_analyze_slp_cost_1): Cost induction.
5959 (vect_detect_hybrid_slp_stmts): Handle PHIs.
5960 (vect_get_slp_vect_defs): Likewise.
5961 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
5962 (vect_transform_stmt): Handle SLP reductions.
5963 * tree-vectorizer.h (vectorizable_induction): Adjust.
5965 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5967 * config/rs6000/rs6000.c (make_resolver_func): Update
5968 init_lowered_empty_function call.
5970 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
5972 * doc/invoke.texi: Document the -fprofile-abs-path option.
5973 * common.opt (fprofile-abs-path): New option.
5974 * gcov-io.h (gcov_write_filename): Declare.
5975 * gcov-io.c (gcov_write_filename): New function.
5976 * coverage.c (coverage_begin_function): Use gcov_write_filename.
5977 * profile.c (output_location): Likewise.
5979 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5981 * shring-wrap.c: Revert accidental commit.
5983 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
5985 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
5987 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5989 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
5991 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
5992 profile in callgraph edge.
5993 * profile-count.h (apply_probability): If THIS is 0, then result is 0
5994 (apply_scale): Likewise.
5995 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
5996 Also scale profile when inlining function with zero profile.
5997 (initialize_cfun): Update exit block profile even when it is zero.
5998 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
5999 when profile is read.
6001 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6003 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6004 (CLONE_*): New constants to define the processors we can generate
6005 code for with the target_clone attribute.
6006 (rs6000_clone_map): New array to identify which clone processors
6007 the current program is running on.
6008 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6009 target_clone attribute.
6010 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6011 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6012 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6013 (cpu_expand_builtin): Add support for target_clone attribute.
6014 (rs6000_valid_attribute_p): Allow "default" attribute.
6015 (get_decl_name): New debug function to simplify printing the
6016 current function name in debugging statements.
6017 (rs6000_clone_priority): New functions to support the target_clone
6018 attribute, and be able to generate code to switch between ISA 2.05
6019 through ISA 3.0 (power6 through power9).
6020 (rs6000_compare_version_priority): Likewise.
6021 (rs6000_get_function_versions_dispatcher): Likewise.
6022 (make_resolver_func): Likewise.
6023 (add_condition_to_bb): Likewise.
6024 (dispatch_function_versions): Likewise.
6025 (rs6000_generate_version_dispatcher_body): Likewise.
6026 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6027 (fusion_gpr_load_p): Fix a spacing issue.
6028 * doc/extend.texi (Common Function Attributes): Document that the
6029 PowerPC supports the target_clone attribute.
6031 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6033 * config/arm/arm.h: explain F symbol found in description of ARM
6034 register allocation in its legend.
6036 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6038 * config/mips/frame-header-opt.c: Include profile-count.h.
6039 * config/riscv/riscv.c: Include profile-count.h
6041 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6043 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6045 (sm_set_flag_if_changed): Add bbs field.
6046 (execute_sm_if_changed_flag_set): Pass BBS.
6047 (execute_sm): Update.
6049 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6051 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6054 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6056 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6057 (peephole2): New peephole2 to emit the above.
6058 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6060 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6062 * config/aarch64/aarch64.c (define_peephole2 above
6063 *sub_<shift>_<mode>): New peephole.
6065 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6067 * config/i386/i386.c (make_resolver_func): Update.
6068 * Makefile.in: Add profile-count.h and profile-count.o
6069 * auto-profile.c (afdo_indirect_call): Update to new API.
6070 (afdo_set_bb_count): Update.
6071 (afdo_propagate_edge): Update.
6072 (afdo_propagate_circuit): Update.
6073 (afdo_calculate_branch_prob): Update.
6074 (afdo_annotate_cfg): Update.
6075 * basic-block.h: Include profile-count.h
6076 (struct edge_def): Turn count to profile_count.
6077 (struct basic_block_def): Likewie.
6078 (REG_BR_PROB_BASE): Move to profile-count.h
6079 (RDIV): Move to profile-count.h
6080 * bb-reorder.c (max_entry_count): Turn to profile_count.
6081 (find_traces): Update.
6082 (rotate_loop):Update.
6083 (connect_traces):Update.
6084 (sanitize_hot_paths):Update.
6085 * bt-load.c (migrate_btr_defs): Update.
6086 * cfg.c (RDIV): Remove.
6087 (init_flow): Use alloc_block.
6088 (alloc_block): Uninitialize count.
6089 (unchecked_make_edge): Uninitialize count.
6090 (check_bb_profile): Update.
6091 (dump_edge_info): Update.
6092 (dump_bb_info): Update.
6093 (update_bb_profile_for_threading): Update.
6094 (scale_bbs_frequencies_int): Update.
6095 (scale_bbs_frequencies_gcov_type): Update.
6096 (scale_bbs_frequencies_profile_count): New.
6097 * cfg.h (update_bb_profile_for_threading): Update.
6098 (scale_bbs_frequencies_profile_count): Declare.
6099 * cfgbuild.c (compute_outgoing_frequencies): Update.
6100 (find_many_sub_basic_blocks): Update.
6101 * cfgcleanup.c (try_forward_edges): Update.
6102 (try_crossjump_to_edge): Update.
6103 * cfgexpand.c (expand_gimple_tailcall): Update.
6104 (construct_exit_block): Update.
6105 * cfghooks.c (verify_flow_info): Update.
6106 (dump_bb_for_graph): Update.
6107 (split_edge): Update.
6108 (make_forwarder_block): Update.
6109 (duplicate_block): Update.
6110 (account_profile_record): Update.
6111 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6112 (get_estimated_loop_iterations): Update.
6113 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6114 (single_likely_exit): Update.
6115 * cfgloopmanip.c (scale_loop_profile): Update.
6117 (set_zero_probability): Update.
6118 (lv_adjust_loop_entry_edge): Update.
6119 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6120 (purge_dead_edges): Update.
6121 (rtl_account_profile_record): Update.
6122 * cgraph.c (cgraph_node::create): Uninitialize count.
6123 (symbol_table::create_edge): Uninitialize count.
6124 (cgraph_update_edges_for_call_stmt_node): Update.
6125 (cgraph_edge::dump_edge_flags): Update.
6126 (cgraph_node::dump): Update.
6127 (cgraph_edge::maybe_hot_p): Update.
6128 * cgraph.h: Include profile-count.h
6129 (create_clone), create_edge, create_indirect_edge): Update.
6130 (cgraph_node): Turn count to profile_count.
6131 (cgraph_edge0: Likewise.
6132 (make_speculative, clone): Update.
6133 (create_edge): Update.
6134 (init_lowered_empty_function): Update.
6135 * cgraphclones.c (cgraph_edge::clone): Update.
6136 (duplicate_thunk_for_node): Update.
6137 (cgraph_node::create_clone): Update.
6138 * cgraphunit.c (cgraph_node::analyze): Update.
6139 (cgraph_node::expand_thunk): Update.
6140 * final.c (dump_basic_block_info): Update.
6141 * gimple-streamer-in.c (input_bb): Update.
6142 * gimple-streamer-out.c (output_bb): Update.
6143 * graphite.c (print_global_statistics): Update.
6144 (print_graphite_scop_statistics): Update.
6145 * hsa-brig.c: Include basic-block.h.
6146 * hsa-dump.c: Include basic-block.h.
6147 * hsa-gen.c (T sum_slice): Update.
6148 (convert_switch_statements):Update.
6149 * hsa-regalloc.c: Include basic-block.h.
6150 * ipa-chkp.c (chkp_produce_thunks): Update.
6151 * ipa-cp.c (struct caller_statistics): Update.
6152 (init_caller_stats): Update.
6153 (gather_caller_stats): Update.
6154 (ipcp_cloning_candidate_p): Update.
6155 (good_cloning_opportunity_p): Update.
6156 (get_info_about_necessary_edges): Update.
6157 (dump_profile_updates): Update.
6158 (update_profiling_info): Update.
6159 (update_specialized_profile): Update.
6160 (perhaps_add_new_callers): Update.
6161 (decide_about_value): Update.
6162 (ipa_cp_c_finalize): Update.
6163 * ipa-devirt.c (struct odr_type_warn_count): Update.
6164 (struct decl_warn_count): Update.
6165 (struct final_warning_record): Update.
6166 (possible_polymorphic_call_targets): Update.
6167 (ipa_devirt): Update.
6168 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6169 * ipa-icf.c (sem_function::merge): Update.
6170 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6171 * ipa-inline.c (compute_uninlined_call_time): Update.
6172 (compute_inlined_call_time): Update.
6173 (want_inline_small_function_p): Update.
6174 (want_inline_self_recursive_call_p): Update.
6175 (edge_badness): Update.
6176 (lookup_recursive_calls): Update.
6177 (recursive_inlining): Update.
6178 (inline_small_functions): Update.
6179 (dump_overall_stats): Update.
6180 (dump_inline_stats): Update.
6181 * ipa-profile.c (ipa_profile_generate_summary): Update.
6182 (ipa_propagate_frequency): Update.
6183 (ipa_profile): Update.
6184 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6185 * ipa-utils.c (ipa_merge_profiles): Update.
6186 * loop-doloop.c (doloop_modify): Update.
6187 * loop-unroll.c (report_unroll): Update.
6188 (unroll_loop_runtime_iterations): Update.
6189 * lto-cgraph.c (lto_output_edge): Update.
6190 (lto_output_node): Update.
6191 (input_node): Update.
6192 (input_edge): Update.
6193 (merge_profile_summaries): Update.
6194 * lto-streamer-in.c (input_cfg): Update.
6195 * lto-streamer-out.c (output_cfg): Update.
6196 * mcf.c (create_fixup_graph): Update.
6197 (adjust_cfg_counts): Update.
6198 (sum_edge_counts): Update.
6199 * modulo-sched.c (sms_schedule): Update.
6200 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6201 * predict.c (maybe_hot_count_p): Update.
6202 (probably_never_executed): Update.
6203 (dump_prediction): Update.
6204 (combine_predictions_for_bb): Update.
6205 (propagate_freq): Update.
6206 (handle_missing_profiles): Update.
6207 (counts_to_freqs): Update.
6208 (rebuild_frequencies): Update.
6209 (force_edge_cold): Update.
6210 * predict.h: Include profile-count.h
6211 (maybe_hot_count_p, counts_to_freqs): UPdate.
6212 * print-rtl-function.c: Do not include cfg.h
6213 * print-rtl.c: Include basic-block.h
6214 * profile-count.c: New file.
6215 * profile-count.h: New file.
6216 * profile.c (is_edge_inconsistent): Update.
6217 (correct_negative_edge_counts): Update.
6218 (is_inconsistent): Update.
6219 (set_bb_counts): Update.
6220 (read_profile_edge_counts): Update.
6221 (compute_frequency_overlap): Update.
6222 (compute_branch_probabilities): Update; Initialize and deinitialize
6224 (branch_prob): Update.
6225 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6226 (edge_gcov_count): New.
6227 (bb_gcov_count): New.
6228 * shrink-wrap.c (try_shrink_wrapping): Update.
6229 * tracer.c (better_p): Update.
6230 * trans-mem.c (expand_transaction): Update.
6231 (ipa_tm_insert_irr_call): Update.
6232 (ipa_tm_insert_gettmclone_call): Update.
6233 * tree-call-cdce.c: Update.
6234 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6235 (gimple_duplicate_sese_tail): Update.
6236 (gimple_account_profile_record): Update.
6237 (execute_fixup_cfg): Update.
6238 * tree-inline.c (copy_bb): Update.
6239 (copy_edges_for_bb): Update.
6240 (initialize_cfun): Update.
6241 (freqs_to_counts): Update.
6242 (copy_cfg_body): Update.
6243 (expand_call_inline): Update.
6244 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6245 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6246 (try_unroll_loop_completely): Update.
6247 (try_peel_loop): Update.
6248 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6249 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6250 * tree-ssa-loop-split.c (connect_loops): Update.
6251 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6252 * tree-ssa-reassoc.c (branch_fixup): Update.
6253 * tree-ssa-tail-merge.c (replace_block_by): Update.
6254 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6255 (compute_path_counts): Update.
6256 (update_profile): Update.
6257 (recompute_probabilities): Update.
6258 (update_joiner_offpath_counts): Update.
6259 (estimated_freqs_path): Update.
6260 (freqs_to_counts_path): Update.
6261 (clear_counts_path): Update.
6262 (ssa_fix_duplicate_block_edges): Update.
6263 (duplicate_thread_path): Update.
6264 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6265 (struct switch_conv_info): Update.
6266 * tree-tailcall.c (decrease_profile): Update.
6267 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6268 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6269 * value-prof.c (check_counter): Update.
6270 (gimple_divmod_fixed_value): Update.
6271 (gimple_mod_pow2): Update.
6272 (gimple_mod_subtract): Update.
6273 (gimple_ic_transform): Update.
6274 (gimple_stringop_fixed_value): Update.
6275 * value-prof.h (gimple_ic): Update.
6277 2017-06-02 Carl Love <cel@us.ibm.com>
6279 * config/rs6000/rs6000-c: Add support for built-in functions
6280 vector double vec_doublee (vector signed int);
6281 vector double vec_doublee (vector unsigned int);
6282 vector double vec_doublee (vector float);
6283 vector double vec_doubleh (vector signed int);
6284 vector double vec_doubleh (vector unsigned int);
6285 vector double vec_doubleh (vector float);
6286 vector double vec_doublel (vector signed int);
6287 vector double vec_doublel (vector unsigned int);
6288 vector double vec_doublel (vector float);
6289 vector double vec_doubleo (vector signed int);
6290 vector double vec_doubleo (vector unsigned int);
6291 vector double vec_doubleo (vector float);.
6292 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6293 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6295 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6296 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6297 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6299 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6300 vec_doublel, vec_doubleh.
6301 * doc/extend.texi: Update the built-in documentation file for the
6302 new built-in functions.
6304 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6307 * ipa-inline-analysis.c (free_growth_caches): Set
6308 edge_removal_hook_holder to NULL after removing it.
6310 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6312 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6313 comparision with zero.
6315 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6316 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6317 for early expansion of vec_min and vec_max builtins.
6318 (builtin_function_type): Add min/max unsigned variants to those
6319 identified as having unsigned arguments.
6321 2017-06-02 Olivier Hainque <hainque@adacore.com>
6323 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6325 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6327 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6328 Use VALL_F16 iterator rather than VALL.
6330 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6332 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6333 Emit CBNZ inside loop when doing a strong exchange and comparing
6334 against zero. Generate the CC flags after the loop.
6336 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6338 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6339 (dl_section_ref): New.
6340 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6341 On AIX, append an expression to subtract the size of the
6342 section length to dl_section_ref.
6344 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6346 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6347 for early expansion of vector absolute builtins.
6349 2017-06-02 Richard Biener <rguenther@suse.de>
6351 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6352 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6354 2017-06-02 Richard Biener <rguenther@suse.de>
6356 PR tree-optimization/80948
6357 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6360 2017-06-02 Richard Biener <rguenther@suse.de>
6362 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6364 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6365 for inductions as relevant.
6367 2017-06-02 Richard Biener <rguenther@suse.de>
6369 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6370 (vectorizable_induction): ... this. Remove dead code.
6372 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6374 * builtins. (expand_builtin_alloca): Remove second parameter and
6375 infer its value from the first parameter instead.
6376 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6378 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6380 PR rtl-optimization/80903
6381 * loop-doloop.c (add_test): Unshare sequence.
6383 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6385 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6387 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6389 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6391 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6392 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6393 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6394 (xlogue_layout::get_stub_rtx): Make static.
6395 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6396 (xlogue_layout::compute_stub_managed_regs): Rename to...
6397 (xlogue_layout::count_stub_managed_regs): ...this.
6398 (xlogue_layout::is_stub_managed_reg): New function.
6399 (xlogue_layout::m_stub_names): Rename to...
6400 (xlogue_layout::s_stub_names): ...this, make static.
6401 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6402 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6403 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6404 xlogue_layout::s_stub_names): Instantiate statics.
6405 (stub_managed_regs): Remove.
6406 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6407 (disable_call_ms2sysv_xlogues): Rename to...
6408 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6409 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6411 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6412 change after reload_completed.
6413 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6415 (ix86_expand_prologue): Likewise.
6416 (ix86_expand_epilogue): Likewise.
6417 (ix86_expand_split_stack_prologue): Likewise.
6418 (ix86_compute_frame_layout): Remove frame parameter ...
6419 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6420 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6422 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6423 (ix86_frame): Move from here ...
6424 * config/i386/i386.h (ix86_frame): ... to here.
6425 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6426 complete ix86_frame data structure instead. Remove some_ld_name.
6428 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6430 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6431 symbols that hold a DECL_VALUE_EXPR.
6433 2017-06-01 Martin Jambor <mjambor@suse.cz>
6435 PR tree-optimization/80898
6436 * tree-sra.c (process_subtree_disqualification): Removed.
6437 (disqualify_candidate): Do not acll
6438 process_subtree_disqualification.
6439 (subtree_mark_written_and_enqueue): New function.
6440 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6441 RHS has been disqualified and re-queue LHS if necessary. Apart
6442 from that, ignore disqualified RHS.
6444 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6446 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6447 address fetch for z10 or later.
6449 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6451 * config/arc/arc.md (tst_movb): Add guard when splitting.
6453 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6455 * config/arc/arc.c (arc_can_eliminate): Test against
6456 arc_frame_pointer_needed.
6458 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6460 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6461 to prevent store reordering.
6462 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6463 (type): Add block type.
6464 (stack_tie): Define special instruction to be used in
6467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6469 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6470 constraint. It is not valid for the pattern.
6471 (noncommutative_binary_comparison): Likewise.
6473 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6475 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6478 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6480 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6481 be used by the reg-alloc.
6483 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6485 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6486 reg-alloc when having mul64 or mul32x16 instructions.
6487 (mulsidi3): Likewise.
6488 (umulsidi3): Likewise.
6489 (mulsi32x16): New pattern.
6490 (mulsi64): Likewise.
6491 (mulsidi64): Likewise.
6492 (umulsidi64): Likewise.
6493 (MUL32x16_REG): Define.
6494 (mul64_600): Use MUL32x16_REG.
6495 (mac64_600): Likewise.
6496 (umul64_600): Likewise.
6497 (umac64_600): Likewise.
6499 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6501 * config/arc/arc.md (mulsi3_700): Make it commutative.
6503 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6505 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6507 (*sign_extendsidi2_insn): Likewise for movstosw.
6509 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6511 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6512 the type of the input discriminant value. Convert the
6513 discriminant value of signedness vary.
6515 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6517 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6518 Add to -Wall section.
6520 2017-06-01 Richard Biener <rguenther@suse.de>
6523 * fold-const.c (fold_plusminus_mult_expr): If the factored
6524 factor may be zero use a wrapping type for the inner operation.
6525 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6526 and handle moved defs.
6527 (process_assignment): Properly guard the unary op case. Return a
6528 tri-state indicating that moving the stmt before the call may allow
6529 to continue. Pass through to_move.
6530 (find_tail_calls): Handle moving unrelated defs before
6533 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6536 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6537 splitter result in the canonical way.
6539 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6541 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6542 also for 32bit target. Update insn attributes.
6543 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6545 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6547 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6548 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6549 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6550 (_mm_maskz_min_ss): New intrinsics.
6552 2017-05-31 Martin Liska <mliska@suse.cz>
6554 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6555 Change comment style to one we normally use.
6556 (vectorizable_reduction): Likewise.
6557 (vectorizable_induction): Likewise.
6558 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6559 (vectorizable_call): Likewise.
6560 (vectorizable_simd_clone_call): Likewise.
6561 (vectorizable_conversion): Likewise.
6562 (vectorizable_assignment): Likewise.
6563 (vectorizable_shift): Likewise.
6564 (vectorizable_operation): Likewise.
6565 (vectorizable_store): Likewise.
6566 (vectorizable_load): Likewise.
6567 * tree-vectorizer.h: Likewise.
6569 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6571 * passes.c (emergency_dump_function): New.
6572 * tree-pass.h (emergency_dump_function): Declare.
6573 * plugin.c (plugins_internal_error_function): Remove.
6574 * plugin.h (plugins_internal_error_function): Remove declaration.
6575 * toplev.c (internal_error_function): New static function. Use it...
6576 (general_init): ...here.
6578 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6580 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6581 (arc_rtx_costs): Add costs for new patterns.
6582 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6583 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6585 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6587 * tree-ssa-strlen.c (get_next_strinfo): New function.
6588 (get_stridx_plus_constant): Use it.
6589 (zero_length_string): Likewise.
6590 (adjust_related_strinfos): Likewise.
6591 (adjust_last_stmt): Likewise.
6593 2017-05-31 Richard Biener <rguenther@suse.de>
6596 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6597 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6599 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6601 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6602 loop_vinfo argument and use of dependence distance vectors.
6603 Check instead whether the two references differ only in their
6604 initial value and assume that they have the same alignment if the
6605 difference is a multiple of the vector alignment.
6606 (vect_analyze_data_refs_alignment): Update call accordingly.
6608 2017-05-31 Martin Liska <mliska@suse.cz>
6611 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6613 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6615 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6616 (create_intersect_range_checks): Move from ...
6617 * tree-data-ref.c (create_intersect_range_checks_index)
6618 (create_intersect_range_checks): ... to here.
6619 (create_runtime_alias_checks): New function factored from ...
6620 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6621 here. Call above function.
6622 * tree-data-ref.h (create_runtime_alias_checks): New function.
6624 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6626 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6627 segment length for dr_b and compute it in wide_int.
6629 2017-05-31 Richard Biener <rguenther@suse.de>
6631 PR tree-optimization/80906
6632 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6633 and pass through iv_map.
6634 (copy_bb_and_scalar_dependences): Adjust.
6635 (translate_pending_phi_nodes): Likewise.
6636 (copy_loop_close_phi_args): Handle code-generating IVs instead
6639 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6641 * diagnostic-color.c (color_dict): Add "type-diff".
6642 (parse_gcc_colors): Update comment.
6643 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6644 -fdiagnostics-show-template-tree and -fno-elide-type.
6645 (GCC_COLORS): Add type-diff to example.
6647 (-fdiagnostics-show-template-tree): New.
6648 (-fno-elide-type): New.
6649 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6650 the pp_format_decoder callback. Call any m_format_postprocessor's
6652 (pretty_printer::pretty_printer): Initialize
6653 m_format_postprocessor.
6654 (pretty_printer::~pretty_printer): Delete any
6655 m_format_postprocessor.
6656 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6657 (class format_postprocessor): New class.
6658 (struct pretty_printer::format_decoder): Document the new parameters.
6659 (struct pretty_printer::m_format_postprocessor): New field.
6660 * tree-diagnostic.c (default_tree_printer): Update for new
6661 bool and const char ** params.
6662 * tree-diagnostic.h (default_tree_printer): Likewise.
6664 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6666 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6667 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6669 rs6000_gen_cell_microcode code.
6670 (rs6000_final_prescan_insn): Delete.
6671 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6672 "warn-cell-microcode" entries.
6673 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6674 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6675 throughout. Change cc_reg_not_micro_cr0_operand to
6676 cc_reg_not_cr0_operand throughout.
6677 (*extendhi<mode>2_noload): Delete.
6678 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6679 (mwarn-cell-microcode): Delete.
6680 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6681 -mgen-cell-microcode and -mwarn-cell-microcode.
6683 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6686 * config/i386/constraints.md (Yd): New constraint.
6688 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6689 and (?Yd, r) alternatives. Update insn attributes.
6690 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6691 and (?*Yd, r) alternatives. Update insn attributes.
6692 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6694 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6696 * gimplify.c (gimplify_modify_expr): Don't create a
6697 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6700 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6702 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6704 2017-05-30 Richard Biener <rguenther@suse.de>
6706 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6707 and reduc_def fields.
6708 (STMT_VINFO_REDUC_TYPE): New define.
6709 (STMT_VINFO_REDUC_DEF): Likewise.
6710 (vect_force_simple_reduction): Adjust prototype.
6711 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6712 (vect_is_simple_reduction): Remove check_reduction argument.
6713 (vect_force_simple_reduction): Adjust and set
6714 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6715 (vectorizable_reduction): Do not re-do reduction analysis
6716 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6717 * tree-parloops.c (gather_scalar_reductions): Adjust.
6719 2017-05-30 Richard Biener <rguenther@suse.de>
6722 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6725 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6727 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6728 Introduce unknown_misalignment parameter and remove vf.
6729 (vect_peeling_hash_get_lowest_cost):
6730 Pass unknown_misalignment parameter.
6731 (vect_enhance_data_refs_alignment):
6732 Fix unsupportable data ref treatment.
6734 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6736 * tree-vect-data-refs.c (vect_get_data_access_cost):
6737 Workaround for SLP handling.
6738 (vect_enhance_data_refs_alignment):
6739 Compute costs for doing no peeling at all, compare to the best
6740 peeling costs so far and avoid peeling if cheaper.
6742 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6744 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6745 Return peeling info and set costs to zero for unlimited cost
6747 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6748 with unknown misalignment. Compute and costs for unknown
6749 misalignment, compare them to the costs for known misalignment
6750 and choose the cheapest for peeling.
6752 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6754 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6755 (vect_get_peeling_costs_all_drs): Create function.
6756 (vect_peeling_hash_get_lowest_cost):
6757 Use vect_get_peeling_costs_all_drs.
6758 (vect_peeling_supportable): Create function.
6759 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6761 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6763 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6764 DR_HAS_NEGATIVE_STEP.
6765 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6766 (vect_enhance_data_refs_alignment): Use.
6767 (vect_duplicate_ssa_name_ptr_info): Use.
6768 * tree-vectorizer.h (dr_misalignment): Use.
6769 (known_alignment_for_access_p): Use.
6771 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6774 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6776 (has_section_name): New function.
6778 2017-05-30 Martin Liska <mliska@suse.cz>
6781 * auto-profile.c (get_function_decl_from_block): Fix
6784 2017-05-30 Richard Biener <rguenther@suse.de>
6787 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6789 2017-05-30 Martin Liska <mliska@suse.cz>
6791 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6792 * dumpfile.h (struct dump_file_info): Remove ctors.
6794 2017-05-30 Martin Liska <mliska@suse.cz>
6796 * predict.def: Fix GNU coding style.
6798 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6800 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6801 Mark 'to' argument with ATTRIBUTE_UNUSED.
6803 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6805 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6806 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6807 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6810 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
6812 * doc/install.texi (Options specification): Restore entry of
6813 --enable-sjlj-exceptions.
6815 2017-05-27 Michael Eager <eager@eagercon.com>
6818 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
6820 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
6822 * config/microblaze/microblaze.h
6823 (FIXED_REGISTERS): Update in macro.
6824 (CALL_USED_REGISTERS): Update in macro.
6826 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
6828 * doc/install.texi: Add links to macOS binary distributions.
6830 2017-05-27 Jakub Jelinek <jakub@redhat.com>
6834 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6836 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6838 2017-05-26 Martin Liska <mliska@suse.cz>
6840 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
6842 2017-05-26 Martin Liska <mliska@suse.cz>
6844 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
6845 always leading ';; '.
6846 (dump_bb_info): Likewise.
6847 (brief_dump_cfg): Likewise.
6848 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
6849 * dumpfile.c: Remove usage of TDF_VERBOSE.
6850 * dumpfile.h (enum dump_kind): Likewise.
6851 (dump_gimple_bb_header): Do not use TDF_COMMENT.
6852 * print-tree.c (debug_verbose): Remove.
6853 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
6854 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
6855 * tree-diagnostic.c (default_tree_printer): Replace
6856 TDF_DIAGNOSTIC with TDF_SLIM.
6858 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6860 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
6861 in parameter loop, rather than loop_vinfo.
6862 (create_intersect_range_checks): Ditto.
6863 (vect_create_cond_for_alias_checks): Update call to above functions.
6865 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6867 PR tree-optimization/80815
6868 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
6869 for merging runtime alias checks. Handle negative DR_STEPs.
6871 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6873 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
6875 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
6876 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
6877 out code pruning runtime alias checks.
6878 * tree-data-ref.c (prune_runtime_alias_test_list): New function
6879 factored out from above.
6880 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6882 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6884 (prune_runtime_alias_test_list): New decalaration.
6886 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6888 * tree-vect-data-refs.c (compare_tree): Rename and move ...
6889 * tree-data-ref.c (data_ref_compare_tree): ... to here.
6890 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
6891 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
6892 (operator==, comp_dr_with_seg_len_pair): Ditto.
6893 (vect_prune_runtime_alias_test_list): Ditto.
6895 2017-05-26 Martin Liska <mliska@suse.cz>
6898 * params.def: Bound partial-inlining-entry-probability param.
6900 2017-05-26 Marek Polacek <polacek@redhat.com>
6903 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
6906 2017-05-26 Richard Biener <rguenther@suse.de>
6908 PR tree-optimization/80842
6909 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
6912 2017-05-26 Richard Biener <rguenther@suse.de>
6914 PR tree-optimization/80844
6915 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
6917 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
6919 * doc/md.texi (Machine Constraints): Update x86 family
6920 machine constraints section to match 'config/i386/constraints.md'.
6922 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
6924 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
6926 2017-05-25 Nathan Sidwell <nathan@acm.org>
6928 * doc/invoke.texi (--enable-languages): Update documentation.
6930 2017-05-25 Martin Liska <mliska@suse.cz>
6932 * dumpfile.c: Add TDF_FOLDING.
6933 * dumpfile.h (enum dump_kind): Likewise.
6934 * genmatch.c (dt_simplify::gen_1): Use it.
6936 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6938 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
6940 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6942 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6943 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
6945 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6947 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
6948 * match.pd (X == C): Rewrite it here.
6949 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
6950 with_certain_nonzero_bits2): New predicates.
6951 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
6953 2017-05-24 Nathan Sidwell <nathan@acm.org>
6955 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
6958 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
6961 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6963 * config/powerpcspe: New port. Files are copied from the rs6000
6964 port, with "rs6000" in filenames replaced by "powerpcspe".
6966 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
6968 PR rtl-optimization/80754
6969 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
6971 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
6973 * config/sparc/sparc.md (length): Return the correct value for -mflat
6974 sibcalls to match output_sibcall.
6976 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6980 * config/rs6000/rs6000.c (struct machine_function): Add new field
6982 (rs6000_get_separate_components): Init that field, use it.
6983 (rs6000_components_for_bb): Use the field.
6985 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
6987 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
6989 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
6992 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
6994 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6997 * config/s390/s390.c (s390_check_qrst_address): Check incoming
6998 address against address_operand predicate.
6999 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7001 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7003 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7004 parameters passed indirectly.
7006 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7008 * config/i386/i386.md (*movdi_internal): Remove SSE4
7009 alternative 18 (?r, *v). Update insn attributes.
7010 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7011 Update insn attributes.
7012 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7013 Update insn attributes.
7014 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7015 alternative 1 (r, v). Remove isa attribute.
7016 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7017 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7018 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7020 2017-05-23 Tom de Vries <tom@codesourcery.com>
7022 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7025 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7027 * cgraphunit.c (symbol_table::process_new_functions): Update.
7028 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7029 (inline_generate_summary): Rename to ...
7030 (ipa_fn_summary_generate): ... this one.
7031 (inline_read_summary): Rename to ...
7032 (ipa_fn_summary_read): ... this one.
7033 (inline_write_summary): Rename to ...
7034 (ipa_fn_summary_write): ... this one.
7035 (inline_free_summary): Rename to ...
7036 (ipa_free_fn_summary): ... this one.
7037 (pass_data_local_fn_summary, pass_local_fn_summary,
7038 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7039 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7040 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7041 make_pass_ipa_fn_summary): New.
7042 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7043 inline_write_summary, inline_free_summary): Remove.
7044 (ipa_free_fn_summary) : New.
7045 * ipa-inline.c (ipa_inline): Update.
7046 (pass_ipa_inline): Do not generate summaries.
7047 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7049 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7050 and add pass_ipa_fn_summary.
7051 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7053 (make_pass_inline_parameters): Remove.
7055 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7057 * omp-low.c (struct omp_context): Remove "default_kind" member.
7060 * omp-offload.c (execute_oacc_device_lower): Remove the
7061 parallelism dimensions function attributes for unparallelized
7062 OpenACC kernels constructs.
7064 2017-05-23 Martin Liska <mliska@suse.cz>
7066 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7068 (cgraph_edge::make_speculative): Likewise.
7069 (cgraph_edge::resolve_speculation): Likewise.
7070 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7071 (cgraph_node::dump): Likewise.
7072 * cgraph.h: Likewise.
7073 * cgraphunit.c (analyze_functions): Likewise.
7074 (symbol_table::compile): Likewise.
7075 * ipa-cp.c (print_all_lattices): Likewise.
7076 (determine_versionability): Likewise.
7077 (initialize_node_lattices): Likewise.
7078 (ipcp_verify_propagated_values): Likewise.
7079 (estimate_local_effects): Likewise.
7080 (update_profiling_info): Likewise.
7081 (create_specialized_node): Likewise.
7082 (perhaps_add_new_callers): Likewise.
7083 (decide_about_value): Likewise.
7084 (decide_whether_version_node): Likewise.
7085 (identify_dead_nodes): Likewise.
7086 (ipcp_store_bits_results): Likewise.
7087 * ipa-devirt.c (dump_targets): Likewise.
7088 (ipa_devirt): Likewise.
7089 * ipa-icf.c (sem_item::dump): Likewise.
7090 (sem_function::equals): Likewise.
7091 (sem_variable::equals): Likewise.
7092 (sem_item_optimizer::read_section): Likewise.
7093 (sem_item_optimizer::execute): Likewise.
7094 (congruence_class::dump): Likewise.
7095 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7096 (dump_inline_summary): Likewise.
7097 (estimate_node_size_and_time): Likewise.
7098 (inline_analyze_function): Likewise.
7099 * ipa-inline-transform.c (inline_call): Likewise.
7100 * ipa-inline.c (report_inline_failed_reason): Likewise.
7101 (want_early_inline_function_p): Likewise.
7102 (edge_badness): Likewise.
7103 (update_edge_key): Likewise.
7104 (inline_small_functions): Likewise.
7105 * ipa-profile.c (ipa_profile): Likewise.
7106 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7107 (ipa_make_edge_direct_to_target): Likewise.
7108 (remove_described_reference): Likewise.
7109 (ipa_impossible_devirt_target): Likewise.
7110 (propagate_controlled_uses): Likewise.
7111 (ipa_print_node_params): Likewise.
7112 (ipcp_transform_function): Likewise.
7113 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7114 (propagate_pure_const): Likewise.
7115 * ipa-reference.c (generate_summary): Likewise.
7116 (read_write_all_from_decl): Likewise.
7117 (propagate): Likewise.
7118 (ipa_reference_read_optimization_summary): Likewise.
7119 * ipa-utils.c (ipa_merge_profiles): Likewise.
7120 * ipa.c (walk_polymorphic_call_targets): Likewise.
7121 (symbol_table::remove_unreachable_nodes): Likewise.
7122 (ipa_single_use): Likewise.
7123 * passes.c (execute_todo): Likewise.
7124 * predict.c (drop_profile): Likewise.
7125 * symtab.c (symtab_node::get_dump_name): New function.
7126 (symtab_node::dump_name): Likewise.
7127 (symtab_node::dump_asm_name): Likewise.
7128 (symtab_node::dump_references): Likewise.
7129 (symtab_node::dump_referring): Likewise.
7130 (symtab_node::dump_base): Likewise.
7131 (symtab_node::debug_symtab): Likewise.
7132 * tree-sra.c (convert_callers_for_node): Likewise.
7133 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7134 * value-prof.c (init_node_map): Likewise.
7136 2017-05-23 Martin Liska <mliska@suse.cz>
7138 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7139 and symtab_node::debug_symtab to symbol_table::debug.
7140 * cgraphunit.c (analyze_functions): Use the renamed function.
7141 (symbol_table::compile): Likewise.
7142 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7143 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7144 * passes.c (execute_todo): Likewise.
7145 * symtab.c (symbol_table::dump): New function.
7146 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7148 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7150 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7151 that nonconst implies exec.
7153 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7155 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7156 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7157 (inline_edge_summary_vec): Turn into ...
7158 (ipa_call_summaries): ... this one.
7159 (redirect_to_unreachable, edge_set_predicate,
7160 evaluate_properties_for_edge, inline_summary_alloc,
7161 reset_ipa_call_summary, reset_inline_summary,
7162 inline_summary_t::duplicate): Update.
7163 (inline_edge_duplication_hook): Turn to ...
7164 (ipa_call_summary_t::duplicate): ... this one.
7165 (inline_edge_removal_hook): Turn to ...
7166 (ipa_call_summary_t::remove): ... this one.
7167 (dump_inline_edge_summary): Turn to ...
7168 (dump_ipa_call_summary): ... this one.
7169 (estimate_function_body_sizes): Update.
7170 (inline_update_callee_summaries): Update.
7171 (remap_edge_change_prob): Update.
7172 (remap_edge_summaries): Update.
7173 (inline_merge_summary): Update.
7174 (do_estimate_edge_time): Update.
7175 (inline_generate_summary): Update.
7176 (inline_read_section): Update.
7177 (inline_read_summary): Update.
7178 (inline_free_summary): Update.
7179 * ipa-inline.c (can_inline_edge_p): Update.
7180 (compute_inlined_call_time): Update.
7181 (want_inline_small_function_p): Update.
7182 (edge_badness): Update.
7183 (early_inliner): Update.
7184 * ipa-inline.h (inline_edge_summary): Turn to ...
7185 (ipa_call_summary): ... this one.
7186 (ipa_call_summary_t): New class.
7187 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7188 (ipa_call_summaries): New.
7189 (inline_edge_summary): Remove.
7190 (estimate_edge_growth): Update.
7191 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7192 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7193 * ipa-split.c (execute_split_functions): Update.
7194 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7196 2017-05-23 Tom de Vries <tom@codesourcery.com>
7198 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7199 attributes): Document rdrand effective target.
7201 2017-05-23 Tom de Vries <tom@codesourcery.com>
7203 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7204 attributes): Sort alphabetically.
7206 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7208 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7210 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7213 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7214 V2DF/V2DI splat into two separate patterns, one that handles
7215 registers, and the other that only handles memory. Drop support
7216 for splatting from a GPR on ISA 2.07 and then splitting the
7217 splat into direct move and splat.
7218 (vsx_splat_<mode>_reg): Likewise.
7219 (vsx_splat_<mode>_mem): Likewise.
7221 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7223 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7225 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7228 * omp-low.c (finish_taskreg_remap): New function.
7229 (finish_taskreg_scan): If unit size of ctx->record_type
7230 is non-constant, unshare the size expression and replace
7231 decls in it with possible outer var refs.
7234 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7235 GOVD_SHARED rather than GOVD_PRIVATE with it.
7236 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7237 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7240 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7241 as last argument to build_outer_var_ref for pointer bases of array
7244 2017-05-19 Martin Sebor <msebor@redhat.com>
7246 * print-tree.c (print_node): Print DECL_READ_P flag.
7248 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7250 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7251 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7252 * cgraph.c: Likewise.
7253 * cgraphunit.c: Likewise.
7254 * gengtype.c: Likewise.
7255 * ipa-cp.c: Likewise.
7256 * ipa-devirt.c: Likewise.
7257 * ipa-icf.c: Likewise.
7258 * ipa-predicate.c: Likewise.
7259 * ipa-profile.c: Likewise.
7260 * ipa-prop.c: Likewise.
7261 * ipa-split.c: Likewise.
7263 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7264 edge_predicate_pool, dump_inline_hints,
7265 inline_summary::account_size_time, redirect_to_unreachable,
7266 edge_set_predicate, set_hint_predicate,
7267 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7268 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7269 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7270 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7271 ipa_call_summary_t::remove, initialize_growth_caches,
7272 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7273 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7274 mark_modified, unmodified_parm_1, unmodified_parm,
7275 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7276 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7277 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7278 will_be_nonconstant_predicate, record_modified_bb_info,
7279 get_minimal_bb, record_modified, param_change_prob,
7280 phi_result_unknown_predicate, predicate_for_phi_result,
7281 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7282 estimate_function_body_sizes, compute_inline_parameters,
7283 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7284 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7285 inline_update_callee_summaries, remap_edge_change_prob,
7286 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7287 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7288 inline_analyze_function, inline_summary_t::insert,
7289 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7290 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7291 inline_free_summary): Move to ipa-fnsummary.h
7292 (predicate_t): Remove.
7293 * ipa-fnsummary.c: New file.
7294 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7295 (enum inline_hints_vals, inline_hints, agg_position_info,
7296 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7297 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7298 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7299 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7300 inline_read_summary, inline_write_summary, inline_free_summary,
7301 inline_analyze_function, initialize_inline_failed,
7302 inline_merge_summary, inline_update_overall_summary,
7303 compute_inline_parameters): Move to ipa-fnsummary.h
7304 * ipa-fnsummary.h: New file.
7305 * ipa-inline-transform.h: Include ipa-inline.h.
7306 * ipa-inline.c: LIkewise.
7308 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7310 * ipa-inline.c (edge_badness): Use inlined_time instead of
7311 inline_summaries->get.
7313 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7315 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7317 2017-05-22 Nathan Sidwell <nathan@acm.org>
7319 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7320 (fdump-lang): Document 'raw' option.
7321 * dumpfile.h (TDI_tu): Delete.
7322 * dumpfile.c (dump_files): Remove translation-unit.
7323 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7325 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7327 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7328 command option from $(AWK) call.
7329 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7330 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7331 [FORMAT]: Remove handling of variable.
7332 * config/avr/t-multilib: Regenerate.
7334 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7336 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7338 (dump_inline_summary): Do not print self_time.
7339 (estimate_function_body_sizes): Do not set self_time.
7340 (compute_inline_parameters): Likewise.
7341 (inline_read_section, inline_write_summary): Do not stream self_time.
7342 * ipa-inline.h (inline_summary): Drop self_time.
7344 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7346 * ipa-inline-analysis.c (account_size_time): Rename to ...
7347 (inline_summary::account_size_time): ... this one.
7348 (reset_ipa_call_summary): Turn to ...
7349 (ipa_call_summary::reset): ... this one.
7350 (reset_inline_summary): Turn to ...
7351 (inline_summary::reset): ... this one.
7352 (inline_summary_t::remove): Update.
7353 (inline_summary_t::duplicate): Update.
7354 (ipa_call_summary_t::remove): Update.
7355 (dump_inline_summary): Update.
7356 (estimate_function_body_sizes): Update.
7357 (compute_inline_parameters): Update.
7358 (estimate_node_size_and_time): Update.
7359 (inline_merge_summary): Update.
7360 (inline_update_overall_summary): Update.
7361 (inline_read_section): Update.
7362 (inline_write_summary): Update.
7363 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7364 add account_size_time and reset member functions.
7365 (ipa_call_summary): Add reset function.
7366 * ipa-predicate.h (predicate::operator &): Constify.
7368 2017-05-22 Richard Biener <rguenther@suse.de>
7370 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7372 2017-05-19 Jason Merrill <jason@redhat.com>
7374 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7376 2017-05-19 Marek Polacek <polacek@redhat.com>
7379 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7380 TYPE_OVERFLOW_WRAPS checks.
7382 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7384 * tree-core.h (enum omp_clause_default_kind): Add
7385 "OMP_CLAUSE_DEFAULT_PRESENT".
7386 * tree-pretty-print.c (dump_omp_clause): Handle it.
7387 * gimplify.c (enum gimplify_omp_var_data): Add
7388 "GOVD_MAP_FORCE_PRESENT".
7389 (gimplify_adjust_omp_clauses_1): Map it to
7390 "GOMP_MAP_FORCE_PRESENT".
7391 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7393 * gimplify.c (oacc_default_clause): Clarify.
7395 2017-05-19 Nathan Sidwell <nathan@acm.org>
7397 LANG_HOOK_REGISTER_DUMPS
7398 * toplev.c (general_init): Call register dump lang hook.
7399 * doc/invoke.texi: Document -fdump-lang option family.
7400 * dumpfile.c (dump_files): Remove class dump here.
7401 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7402 * dumpfile.h (tree_dump_index): Remove TDI_class.
7403 * langhooks-def.h (lhd_register_dumps): Declare.
7404 (LANG_HOOKS_REGISTER_DUMPS): Define.
7405 (LANG_HOOKS_INITIALIZER): Add it.
7406 * langhooks.c (lhd_register_dumps): Define.
7407 * langhooks.h (struct lang_hooks): Add register_dumps.
7409 2017-05-19 Nathan Sidwell <nathan@acm.org>
7411 * context.h (context::set_passes): New.
7412 * context.c (context::context): Do not create pass manager.
7413 * toplev.c (general_init): Create pass manager here.
7415 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7417 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7418 use this splitter if two add or or instructions would also work for
7419 the constant we want to generate.
7421 2017-05-19 Richard Biener <rguenther@suse.de>
7424 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7425 predicate evaluation.
7427 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7429 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7431 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7434 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7436 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7437 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7438 (inline_edge_summary_vec): Turn into ...
7439 (ipa_call_summaries): ... this one.
7440 (redirect_to_unreachable, edge_set_predicate,
7441 evaluate_properties_for_edge, inline_summary_alloc,
7442 reset_ipa_call_summary, reset_inline_summary,
7443 inline_summary_t::duplicate): Update.
7444 (inline_edge_duplication_hook): Turn to ...
7445 (ipa_call_summary_t::duplicate): ... this one.
7446 (inline_edge_removal_hook): Turn to ...
7447 (ipa_call_summary_t::remove): ... this one.
7448 (dump_inline_edge_summary): Turn to ...
7449 (dump_ipa_call_summary): ... this one.
7450 (estimate_function_body_sizes): Update.
7451 (inline_update_callee_summaries): Update.
7452 (remap_edge_change_prob): Update.
7453 (remap_edge_summaries): Update.
7454 (inline_merge_summary): Update.
7455 (do_estimate_edge_time): Update.
7456 (inline_generate_summary): Update.
7457 (inline_read_section): Update.
7458 (inline_read_summary): Update.
7459 (inline_free_summary): Update.
7460 * ipa-inline.c (can_inline_edge_p): Update.
7461 (compute_inlined_call_time): Update.
7462 (want_inline_small_function_p): Update.
7463 (edge_badness): Update.
7464 (early_inliner): Update.
7465 * ipa-inline.h (inline_edge_summary): Turn to ...
7466 (ipa_call_summary): ... this one.
7467 (ipa_call_summary_t): New class.
7468 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7469 (ipa_call_summaries): New.
7470 (inline_edge_summary): Remove.
7471 (estimate_edge_growth): Update.
7472 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7473 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7474 * ipa-split.c (execute_split_functions): Update.
7475 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7477 2017-05-19 Richard Biener <rguenther@suse.de>
7480 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7482 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7484 * config/rs6000/rs6000.c (struct machine_function): Add field
7485 fpr_is_wrapped_separately.
7486 (rs6000_get_separate_components): Use 64 components. Handle the
7488 (rs6000_components_for_bb): Handle the FPR components.
7489 (rs6000_emit_prologue_components): Handle the FPR components.
7490 (rs6000_emit_epilogue_components): Handle the FPR components.
7491 (rs6000_set_handled_components): Handle the FPR components.
7492 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7493 that are already separately shrink-wrapped.
7494 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7495 that are already separately shrink-wrapped.
7497 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7500 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7503 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7504 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7505 cases where the register allocator uses a move and an offsettable
7506 memory operation to/from a FPR register on ISA 2.06/2.07.
7507 (define_peephole2 for Altivec d-form store): Likewise.
7509 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7512 * config/i386/mmx.md (*mov<mode>_internal): Enable
7513 alternatives 11, 12, 13 and 14 also for 32bit targets.
7514 Remove alternatives 15, 16, 17 and 18.
7515 * config/i386/sse.md (vec_concatv2di): Change
7516 alternative (!x, *y) to (x, ?!*Yn).
7518 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7520 * dumpfile.h (enum dump_kind): Remove stray comma.
7522 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7524 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7525 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7526 predicate::num_conditions
7527 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7528 (CHANGED): turn into predicate::changed.
7529 (agg_position_info): Move to ipa-predicate.h
7530 (add_condition, predicate::add_clause, predicate::operator &=,
7531 predicate::or_with, predicate::evaluate, predicate::probability,
7532 dump_condition, dump_clause, predicate::dump,
7533 predicate::remap_after_duplication, predicate::remap_after_inlining,
7534 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7535 (evaluate_conditions_for_known_args): Update.
7536 (set_cond_stmt_execution_predicate): Update.
7537 * ipa-inline.h: Include ipa-predicate.h
7538 (condition, inline_param_summary, conditions, agg_position_info,
7539 predicate): Move to ipa-predicate.h
7540 * ipa-predicate.c: New file.
7541 * ipa-predicate.h: New file.
7543 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7545 * final.c (leaf_function_p): Check we are not in a sequence.
7547 2017-05-18 Martin Liska <mliska@suse.cz>
7549 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7550 * dumpfile.c (dump_register): Use new enum dump_kind.
7551 (get_dump_file_name): Likewise.
7552 (dump_enable_all): Likewise.
7553 (dump_switch_p_1): Likewise.
7554 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7555 * dumpfile.h (enum dump_kind): New enum type.
7556 (struct dump_file_info): Create constructor and
7557 format fields and comments.
7558 * passes.c (pass_manager::register_one_dump_file):
7560 * statistics.c (statistics_early_init): Likewise.
7561 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7562 TDF_TREE with TDF_SLIM.
7563 (gather_memory_references_ref): Likewise.
7565 2017-05-18 Martin Liska <mliska@suse.cz>
7567 * vec.h (struct vnull): Use it.
7569 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7571 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7572 (true_predicate, false_predicate, true_predicate_p,
7573 false_predicate_p): Remove.
7574 (single_cond_predicate, not_inlined_predicate): Turn to member function
7576 (add_condition): Update.
7577 (add_clause): Turn to...
7578 (predicate::add_clause): ... this one; update; allow passing NULL
7580 (and_predicates): Turn to ...
7581 (predicate::operator &=): ... this one.
7582 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7583 (or_predicates): Turn to ...
7584 (predicate::or_with): ... this one.
7585 (evaluate_predicate): Turn to ...
7586 (predicate::evaluate): ... this one.
7587 (predicate_probability): Turn to ...
7588 (predicate::probability): ... this one.
7589 (dump_condition): Update.
7590 (dump_predicate): Turn to ...
7591 (predicate::dump): ... this one.
7592 (account_size_time): Update.
7593 (edge_set_predicate): Update.
7594 (set_hint_predicate): UPdate.
7595 (evaluate_conditions_for_known_args): Update.
7596 (evaluate_properties_for_edge): Update.
7597 (remap_predicate_after_duplication): Turn to...
7598 (predicate::remap_after_duplication): ... this one.
7599 (remap_hint_predicate_after_duplication): Update.
7600 (inline_summary_t::duplicate): UPdate.
7601 (dump_inline_edge_summary): Update.
7602 (dump_inline_summary): Update.
7603 (set_cond_stmt_execution_predicate): Update.
7604 (set_switch_stmt_execution_predicate): Update.
7605 (compute_bb_predicates): Update.
7606 (will_be_nonconstant_expr_predicate): Update.
7607 (will_be_nonconstant_predicate): Update.
7608 (phi_result_unknown_predicate): Update.
7609 (predicate_for_phi_result): Update.
7610 (array_index_predicate): Update.
7611 (estimate_function_body_sizes): Update.
7612 (estimate_node_size_and_time): Update.
7613 (estimate_ipcp_clone_size_and_time): Update.
7614 (remap_predicate): Rename to ...
7615 (predicate::remap_after_inlining): ... this one.
7616 (remap_hint_predicate): Update.
7617 (inline_merge_summary): Update.
7618 (inline_update_overall_summary): Update.
7619 (estimate_size_after_inlining): Update.
7620 (read_predicate): Rename to ...
7621 (predicate::stream_in): ... this one.
7622 (read_inline_edge_summary): Update.
7623 (write_predicate): Rename to ...
7624 (predicate::stream_out): ... this one.
7625 (write_inline_edge_summary): Update.
7626 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7627 (clause_t): Turn to uint32_t
7628 (predicate): Turn to class; implement constructor and operators
7630 (size_time_entry): Update.
7631 (inline_summary): Update.
7632 (inline_edge_summary): Update.
7634 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7636 * fold-const.c (fold_binary_loc): Move transformation...
7637 * match.pd (C - X CMP X): ... here.
7639 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7641 * config/sparc/sparc.c (sparc_option_override): Set function
7642 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7643 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7645 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7647 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7649 2017-05-18 Marek Polacek <polacek@redhat.com>
7652 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7653 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7654 gimple_assign_load_p.
7656 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7659 * real.c (do_compare): Give decimal_do_compare preference over
7660 comparing just the signs.
7662 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7664 * doc/md.texi (Canonicalization of Instructions): Describe the
7665 canonical form of instructions that inherently set a condition
7668 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7671 * tree-cfg.c: Move deletion of unreachable case statements to after
7672 the merging of consecutive case labels.
7674 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7676 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7677 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7678 restoring of callee-saved registers.
7680 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7682 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7683 * config/visium/visium.c (single_set_and_flags): Likewise.
7684 * config/visium/visium.md (Substitutions): Likewise.
7686 2017-05-17 Martin Liska <mliska@suse.cz>
7688 * cfg.c: Introduce dump_flags_t type and
7689 use it instead of int type.
7691 * cfghooks.c: Likewise.
7692 * cfghooks.h (struct cfg_hooks): Likewise.
7693 * cfgrtl.c: Likewise.
7694 * cfgrtl.h: Likewise.
7695 * cgraph.c (cgraph_node::get_body): Likewise.
7696 * coretypes.h: Likewise.
7697 * domwalk.c: Likewise.
7698 * domwalk.h: Likewise.
7699 * dumpfile.c (struct dump_option_value_info): Likewise.
7700 (dump_enable_all): Likewise.
7701 (dump_switch_p_1): Likewise.
7702 (opt_info_switch_p): Likewise.
7703 * dumpfile.h (enum tree_dump_index): Likewise.
7704 (struct dump_file_info): Likewise.
7705 * genemit.c: Likewise.
7706 * generic-match-head.c: Likewise.
7707 * gengtype.c (open_base_files): Likewise.
7708 * gimple-pretty-print.c: Likewise.
7709 * gimple-pretty-print.h: Likewise.
7710 * graph.c (print_graph_cfg): Likewise.
7711 * graphite-scop-detection.c (dot_all_sese): Likewise.
7712 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7713 * loop-unroll.c (report_unroll): Likewise.
7714 * passes.c (pass_manager::register_one_dump_file): Likewise.
7715 * print-tree.c: Likewise.
7716 * statistics.c: Likewise.
7717 * tree-cfg.c: Likewise.
7718 * tree-cfg.h: Likewise.
7719 * tree-dfa.c: Likewise.
7720 * tree-dfa.h: Likewise.
7721 * tree-dump.c (dump_function): Likewise.
7722 * tree-dump.h (struct dump_info): Likewise.
7723 * tree-pretty-print.c: Likewise.
7724 * tree-pretty-print.h: Likewise.
7725 * tree-ssa-live.c: Likewise.
7726 * tree-ssa-live.h: Likewise.
7727 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7728 * tree-vect-loop.c: Likewise.
7729 * tree-vect-slp.c: Likewise.
7731 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7732 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7734 PR tree-optimization/80457
7735 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7736 of all arguments to a statement as scalar_to_vec operations.
7737 (vectorizable_call): Adjust call to vect_model_simple_cost for
7739 (vectorizable_conversion): Likewise.
7740 (vectorizable_assignment): Likewise.
7741 (vectorizable_shift): Likewise.
7742 (vectorizable_operation): Likewise.
7743 (vectorizable_comparison): Likewise.
7744 (vect_is_simple_cond): Record the def types for operands.
7745 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7746 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7747 for statement argument count.
7749 2017-05-16 Carl Love <cel@us.ibm.com>
7751 * config/rs6000/rs6000-c: Add support for built-in functions
7752 vector unsigned long long vec_bperm (vector unsigned long long,
7753 vector unsigned char)
7754 vector signed long long vec_mule (vector signed int,
7756 vector unsigned long long vec_mule (vector unsigned int,
7757 vector unsigned int)
7758 vector signed long long vec_mulo (vector signed int,
7760 vector unsigned long long vec_mulo (vector unsigned int,
7761 vector unsigned int)
7762 vector signed char vec_sldw (vector signed char,
7765 vector unsigned char vec_sldw (vector unsigned char,
7766 vector unsigned char,
7768 vector signed short vec_sldw (vector signed short,
7769 vector signed short,
7771 vector unsigned short vec_sldw (vector unsigned short,
7772 vector unsigned short,
7774 vector signed int vec_sldw (vector signed int,
7777 vector unsigned int vec_sldw (vector unsigned int,
7778 vector unsigned int,
7780 vector signed long long vec_sldw (vector signed long long,
7781 vector signed long long,
7783 vector unsigned long long vec_sldw (vector unsigned long long,
7784 vector unsigned long long,
7786 * config/rs6000/rs6000-c: Add support for built-in functions
7787 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7788 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7789 * doc/extend.texi: Update the built-in documentation for the
7790 new built-in functions.
7792 2017-05-16 Marek Polacek <polacek@redhat.com>
7796 * tree.c (save_expr): Don't fold the expression.
7798 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7800 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7801 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7802 and (?*y,m). Update insn attributes.
7804 2017-05-16 Martin Liska <mliska@suse.cz>
7806 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7807 flags argument of print_gimple_stmt, print_gimple_expr,
7808 print_generic_stmt and print_generic_expr.
7809 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7810 * coretypes.h: Likewise.
7811 * except.c (dump_eh_tree): Likewise.
7812 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7813 * gimple-pretty-print.h: Likewise.
7814 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
7815 (backprop::push_to_worklist): Likewise.
7816 (backprop::pop_from_worklist): Likewise.
7817 (backprop::process_use): Likewise.
7818 (backprop::intersect_uses): Likewise.
7819 (note_replacement): Likewise.
7820 * gimple-ssa-store-merging.c
7821 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
7822 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7823 (pass_store_merging::execute): Likewise.
7824 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
7825 (ssa_base_cand_dump_callback): Likewise.
7826 (dump_incr_vec): Likewise.
7827 (replace_refs): Likewise.
7828 (replace_mult_candidate): Likewise.
7829 (create_add_on_incoming_edge): Likewise.
7830 (create_phi_basis): Likewise.
7831 (insert_initializers): Likewise.
7832 (all_phi_incrs_profitable): Likewise.
7833 (introduce_cast_before_cand): Likewise.
7834 (replace_one_candidate): Likewise.
7835 * gimplify.c (gimplify_expr): Likewise.
7836 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
7837 (set_rename): Likewise.
7838 (rename_uses): Likewise.
7839 (copy_loop_phi_nodes): Likewise.
7840 (add_close_phis_to_merge_points): Likewise.
7841 (copy_loop_close_phi_args): Likewise.
7842 (copy_cond_phi_args): Likewise.
7843 (graphite_copy_stmts_from_block): Likewise.
7844 (translate_pending_phi_nodes): Likewise.
7845 * graphite-poly.c (print_pdr): Likewise.
7846 (dump_gbb_cases): Likewise.
7847 (dump_gbb_conditions): Likewise.
7848 (print_scop_params): Likewise.
7849 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
7850 (build_cross_bb_scalars_use): Likewise.
7851 (gather_bbs::before_dom_children): Likewise.
7852 * hsa-dump.c (dump_hsa_immed): Likewise.
7853 * ipa-cp.c (print_ipcp_constant_value): Likewise.
7854 (get_replacement_map): Likewise.
7855 * ipa-inline-analysis.c (dump_condition): Likewise.
7856 (estimate_function_body_sizes): Likewise.
7857 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
7858 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
7859 * ipa-prop.c (ipa_dump_param): Likewise.
7860 (ipa_print_node_jump_functions_for_edge): Likewise.
7861 (ipa_modify_call_arguments): Likewise.
7862 (ipa_modify_expr): Likewise.
7863 (ipa_dump_param_adjustments): Likewise.
7864 (ipa_dump_agg_replacement_values): Likewise.
7865 (ipcp_modif_dom_walker::before_dom_children): Likewise.
7866 * ipa-pure-const.c (check_stmt): Likewise.
7867 (pass_nothrow::execute): Likewise.
7868 * ipa-split.c (execute_split_functions): Likewise.
7869 * omp-offload.c (dump_oacc_loop_part): Likewise.
7870 (dump_oacc_loop): Likewise.
7871 * trans-mem.c (tm_log_emit): Likewise.
7872 (tm_memopt_accumulate_memops): Likewise.
7873 (dump_tm_memopt_set): Likewise.
7874 (dump_tm_memopt_transform): Likewise.
7875 * tree-cfg.c (gimple_verify_flow_info): Likewise.
7876 (print_loop): Likewise.
7877 * tree-chkp-opt.c (chkp_print_addr): Likewise.
7878 (chkp_gather_checks_info): Likewise.
7879 (chkp_get_check_result): Likewise.
7880 (chkp_remove_check_if_pass): Likewise.
7881 (chkp_use_outer_bounds_if_possible): Likewise.
7882 (chkp_reduce_bounds_lifetime): Likewise.
7883 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
7884 (chkp_mark_completed_bounds): Likewise.
7885 (chkp_register_incomplete_bounds): Likewise.
7886 (chkp_mark_invalid_bounds): Likewise.
7887 (chkp_maybe_copy_and_register_bounds): Likewise.
7888 (chkp_build_returned_bound): Likewise.
7889 (chkp_get_bound_for_parm): Likewise.
7890 (chkp_build_bndldx): Likewise.
7891 (chkp_get_bounds_by_definition): Likewise.
7892 (chkp_generate_extern_var_bounds): Likewise.
7893 (chkp_get_bounds_for_decl_addr): Likewise.
7894 * tree-chrec.c (chrec_apply): Likewise.
7895 * tree-data-ref.c (dump_data_reference): Likewise.
7896 (dump_subscript): Likewise.
7897 (dump_data_dependence_relation): Likewise.
7898 (analyze_overlapping_iterations): Likewise.
7899 * tree-inline.c (expand_call_inline): Likewise.
7900 (tree_function_versioning): Likewise.
7901 * tree-into-ssa.c (dump_defs_stack): Likewise.
7902 (dump_currdefs): Likewise.
7903 (dump_names_replaced_by): Likewise.
7904 (dump_update_ssa): Likewise.
7905 (update_ssa): Likewise.
7906 * tree-object-size.c (pass_object_sizes::execute): Likewise.
7907 * tree-parloops.c (build_new_reduction): Likewise.
7908 (try_create_reduction_list): Likewise.
7909 (ref_conflicts_with_region): Likewise.
7910 (oacc_entry_exit_ok_1): Likewise.
7911 (oacc_entry_exit_single_gang): Likewise.
7912 * tree-pretty-print.h: Likewise.
7913 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
7914 (get_scalar_evolution): Likewise.
7915 (add_to_evolution): Likewise.
7916 (get_loop_exit_condition): Likewise.
7917 (analyze_evolution_in_loop): Likewise.
7918 (analyze_initial_condition): Likewise.
7919 (analyze_scalar_evolution): Likewise.
7920 (instantiate_scev): Likewise.
7921 (number_of_latch_executions): Likewise.
7922 (gather_chrec_stats): Likewise.
7923 (final_value_replacement_loop): Likewise.
7924 (scev_const_prop): Likewise.
7925 * tree-sra.c (dump_access): Likewise.
7926 (disqualify_candidate): Likewise.
7927 (create_access): Likewise.
7929 (maybe_add_sra_candidate): Likewise.
7930 (create_access_replacement): Likewise.
7931 (analyze_access_subtree): Likewise.
7932 (analyze_all_variable_accesses): Likewise.
7933 (sra_modify_assign): Likewise.
7934 (initialize_constant_pool_replacements): Likewise.
7935 (find_param_candidates): Likewise.
7936 (decide_one_param_reduction): Likewise.
7937 (replace_removed_params_ssa_names): Likewise.
7938 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
7939 * tree-ssa-copy.c (dump_copy_of): Likewise.
7940 (copy_prop_visit_cond_stmt): Likewise.
7941 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
7942 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
7943 (record_equivalences_from_stmt): Likewise.
7944 * tree-ssa-dse.c (compute_trims): Likewise.
7945 (delete_dead_call): Likewise.
7946 (delete_dead_assignment): Likewise.
7947 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
7948 (forward_propagate_into_cond): Likewise.
7949 (pass_forwprop::execute): Likewise.
7950 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7951 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
7953 (move_computations_worker): Likewise.
7954 (execute_sm): Likewise.
7955 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
7956 (remove_exits_and_undefined_stmts): Likewise.
7957 (remove_redundant_iv_tests): Likewise.
7958 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
7959 (adjust_iv_update_pos): Likewise.
7960 * tree-ssa-math-opts.c (bswap_replace): Likewise.
7961 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
7962 (value_replacement): Likewise.
7963 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
7964 * tree-ssa-pre.c (print_pre_expr): Likewise.
7965 (get_representative_for): Likewise.
7966 (create_expression_by_pieces): Likewise.
7967 (insert_into_preds_of_block): Likewise.
7968 (eliminate_insert): Likewise.
7969 (eliminate_dom_walker::before_dom_children): Likewise.
7970 (eliminate): Likewise.
7971 (remove_dead_inserted_code): Likewise.
7972 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
7973 * tree-ssa-reassoc.c (get_rank): Likewise.
7974 (eliminate_duplicate_pair): Likewise.
7975 (eliminate_plus_minus_pair): Likewise.
7976 (eliminate_not_pairs): Likewise.
7977 (undistribute_ops_list): Likewise.
7978 (eliminate_redundant_comparison): Likewise.
7979 (update_range_test): Likewise.
7980 (optimize_range_tests_var_bound): Likewise.
7981 (optimize_vec_cond_expr): Likewise.
7982 (rewrite_expr_tree): Likewise.
7983 (rewrite_expr_tree_parallel): Likewise.
7984 (linearize_expr): Likewise.
7985 (break_up_subtract): Likewise.
7986 (linearize_expr_tree): Likewise.
7987 (attempt_builtin_powi): Likewise.
7988 (attempt_builtin_copysign): Likewise.
7989 (transform_stmt_to_copy): Likewise.
7990 (transform_stmt_to_multiply): Likewise.
7991 (dump_ops_vector): Likewise.
7992 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
7993 (print_scc): Likewise.
7994 (set_ssa_val_to): Likewise.
7995 (visit_reference_op_store): Likewise.
7996 (visit_use): Likewise.
7997 (sccvn_dom_walker::before_dom_children): Likewise.
7998 (run_scc_vn): Likewise.
7999 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8001 (expr_hash_elt::print): Likewise.
8002 (const_and_copies::pop_to_marker): Likewise.
8003 (const_and_copies::record_const_or_copy_raw): Likewise.
8004 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8005 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8006 (dump_predicates): Likewise.
8007 (find_uninit_use): Likewise.
8008 (warn_uninitialized_phi): Likewise.
8009 (pass_late_warn_uninitialized::execute): Likewise.
8010 * tree-ssa.c (verify_vssa): Likewise.
8011 (verify_ssa): Likewise.
8012 (maybe_optimize_var): Likewise.
8013 * tree-vrp.c (dump_value_range): Likewise.
8014 (dump_all_value_ranges): Likewise.
8015 (dump_asserts_for): Likewise.
8016 (register_edge_assert_for_2): Likewise.
8017 (vrp_visit_cond_stmt): Likewise.
8018 (vrp_visit_switch_stmt): Likewise.
8019 (vrp_visit_stmt): Likewise.
8020 (vrp_visit_phi_node): Likewise.
8021 (simplify_cond_using_ranges_1): Likewise.
8022 (fold_predicate_in): Likewise.
8023 (evrp_dom_walker::before_dom_children): Likewise.
8024 (evrp_dom_walker::push_value_range): Likewise.
8025 (evrp_dom_walker::pop_value_range): Likewise.
8026 (execute_early_vrp): Likewise.
8028 2017-05-16 Richard Biener <rguenther@suse.de>
8030 * dwarf2out.c (loc_list_from_tree_1): Do not create
8031 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8033 2017-05-16 Richard Biener <rguenther@suse.de>
8035 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8037 (note_variable_value_in_expr): If we resolved the decl ref
8038 do not push to the stack.
8040 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8042 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8043 operations in fast-math mode.
8044 (vaddq_f16): Likewise.
8045 (vmul_f16): Likewise.
8046 (vmulq_f16): Likewise.
8047 (vsub_f16): Likewise.
8048 (vsubq_f16): Likewise.
8049 * config/arm/neon.md (add<mode>3): New.
8051 (fma:<VH:mode>3): New. Also remove outdated comment.
8054 2017-05-16 Martin Liska <mliska@suse.cz>
8058 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8059 (odr_types_equivalent_p): Likewise.
8061 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8063 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8065 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8068 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8069 non-interunit SSE move alternatives with '?'.
8070 (zero-extendsidi peephole2): New peephole to skip intermediate
8071 general register in SSE zero-extend sequence.
8073 2017-05-15 Jeff Law <law@redhat.com>
8075 * reorg.c (relax_delay_slots): Create a new variable to hold
8076 the temporary target rather than clobbering TARGET_LABEL.
8078 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8079 missing argument to extract_bit_field call.
8080 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8082 2017-05-15 Martin Liska <mliska@suse.cz>
8085 * gcc.c (process_command): Do not allow empty argument of -o option.
8087 2017-05-15 Renlin Li <renlin.li@arm.com>
8089 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8090 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8091 * config/aarch64/constraints.md (Usf): Add long call check.
8092 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8093 (call_value): Likewise.
8094 (sibcall): Likewise.
8095 (sibcall_value): Likewise.
8097 (call_value_insn): New.
8098 (sibcall_insn): Update rtx pattern.
8099 (sibcall_value_insn): Likewise.
8100 (call_internal): Remove.
8101 (call_value_internal): Likewise.
8102 (sibcall_internal): Likewise.
8103 (sibcall_value_internal): Likewise.
8104 (call_reg): Likewise.
8105 (call_symbol): Likewise.
8106 (call_value_reg): Likewise.
8107 (call_value_symbol): Likewise.
8109 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8112 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8114 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8116 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8117 compatible with CCGOCmode and with CCZmode.
8119 2017-05-14 Martin Sebor <msebor@redhat.com>
8122 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8123 (gimple_fold_builtin_snprintf): Same.
8124 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8125 (gimple_fold_builtin_snprintf): Same.
8126 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8128 (is_call_safe): New function.
8129 (try_substitute_return_value): Call it.
8130 (try_simplify_call): New function.
8131 (pass_sprintf_length::handle_gimple_call): Call it.
8133 2017-05-14 Martin Sebor <msebor@redhat.com>
8136 * builtins.c (expand_builtin_stpncpy): Simplify.
8138 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8140 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8141 * config/i386/i386.h
8142 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8143 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8144 (struct machine_function): Add new members call_ms2sysv,
8145 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8146 (struct machine_frame_state): New fields sp_realigned and
8147 sp_realigned_offset.
8148 * config/i386/i386.c
8149 (enum xlogue_stub): New enum.
8150 (enum xlogue_stub_sets): New enum.
8151 (class xlogue_layout): New class.
8152 (struct ix86_frame): New fields stack_realign_allocate_offset,
8153 stack_realign_offset and outlined_save_offset. Modify comments to
8154 detail stack layout when using out-of-line stubs.
8155 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8156 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8157 -mcall-ms2sysv-xlogues.
8158 (stub_managed_regs): New static variable.
8159 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8160 registers managed by out-of-line stub.
8161 (disable_call_ms2sysv_xlogues): New function.
8162 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8163 m->call_ms2sysv when appropriate and compute frame layout for
8165 (sp_valid_at, fp_valid_at): New inline functions.
8166 (choose_basereg): New function.
8167 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8169 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8170 Use align parameter of choose_baseaddr to generated aligned SSE movs
8172 (pro_epilogue_adjust_stack): Modify to track
8173 machine_frame_state::sp_realigned.
8174 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8175 (ix86_nsaved_sseregs): Likewise.
8176 (ix86_emit_save_regs): Likewise.
8177 (ix86_emit_save_regs_using_mov): Likewise.
8178 (ix86_emit_save_sse_regs_using_mov): Likewise.
8179 (get_scratch_register_on_entry): Likewise.
8180 (gen_frame_set): New function.
8181 (gen_frame_load): Likewise.
8182 (gen_frame_store): Likewise.
8183 (emit_outlined_ms2sysv_save): Likewise.
8184 (emit_outlined_ms2sysv_restore): Likewise.
8185 (ix86_expand_prologue): Modify stack re-alignment code and call
8186 emit_outlined_ms2sysv_save when appropriate.
8187 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8188 parameter rtx_insn *insn, which allows the function to be used to only
8190 (ix86_expand_epilogue): Modify validity checks of frame and stack
8191 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8192 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8193 * config/i386/predicates.md
8194 (save_multiple): New predicate.
8195 (restore_multiple): Likewise.
8196 * config/i386/sse.md
8197 (save_multiple<mode>): New pattern.
8198 (save_multiple_realign<mode>): Likewise.
8199 (restore_multiple<mode>): Likewise.
8200 (restore_multiple_and_return<mode>): Likewise.
8201 (restore_multiple_leave_return<mode>): Likewise.
8202 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8204 2017-05-14 Julia Koval <julia.koval@intel.com>
8206 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8207 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8208 (__builtin_ia32_xsetbv): New builtins.
8209 * config/i386/i386.c (ix86_expand_special_args_builtin):
8211 (ix86_expand_builtin): Special expand for new intrinsics.
8212 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8213 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8214 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8216 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8218 * cfganal.c (inverted_post_order_compute): Change argument type
8220 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8221 * df-core.c (rest_of_handle_df_initialize): Adjust.
8222 (rest_of_handle_df_finish): Likewise.
8223 (df_analyze_1): Likewise.
8224 (df_analyze): Likewise.
8225 (loop_inverted_post_order_compute): Change argument to be a vec *.
8226 (df_analyze_loop): Adjust.
8227 (df_get_n_blocks): Likewise.
8228 (df_get_postorder): Likewise.
8229 * df.h (struct df_d): Change field to be a vec.
8230 * lcm.c (compute_laterin): Adjust.
8231 (compute_available): Likewise.
8232 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8233 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8234 * tree-ssa-pre.c (compute_antic): Likewise.
8236 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8238 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8239 (depth_first_search::depth_first_search): Change structure init
8240 function to this constructor.
8241 (depth_first_search::add_bb): Rename function to this member.
8242 (depth_first_search::execute): Likewise.
8243 (flow_dfs_compute_reverse_finish): Adjust.
8245 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8247 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8248 (longest_simple_path): Likewise.
8249 * shrink-wrap.c (spread_components): Likewise.
8250 (disqualify_problematic_components): Likewise.
8251 (emit_common_heads_for_components): Likewise.
8252 (emit_common_tails_for_components): Likewise.
8253 (insert_prologue_epilogue_for_components): Likewise.
8255 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8257 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8260 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8262 * df-core.c (df_set_blocks): Start using auto_bitmap.
8263 (df_compact_blocks): Likewise.
8264 * df-problems.c (df_rd_confluence_n): Likewise.
8265 * df-scan.c (df_insn_rescan_all): Likewise.
8266 (df_process_deferred_rescans): Likewise.
8267 (df_update_entry_block_defs): Likewise.
8268 (df_update_exit_block_uses): Likewise.
8269 (df_entry_block_bitmap_verify): Likewise.
8270 (df_exit_block_bitmap_verify): Likewise.
8271 (df_scan_verify): Likewise.
8272 * lra-constraints.c (lra_constraints): Likewise.
8273 (undo_optional_reloads): Likewise.
8274 (lra_undo_inheritance): Likewise.
8275 * lra-remat.c (calculate_gen_cands): Likewise.
8276 (do_remat): Likewise.
8277 * lra-spills.c (assign_spill_hard_regs): Likewise.
8278 (spill_pseudos): Likewise.
8279 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8280 (bitmap_set_subtract_values): Likewise.
8282 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8284 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8285 management with auto_bitmap.
8286 (fix_inter_tick): Likewise.
8287 (fix_recovery_deps): Likewise.
8288 * ira.c (add_store_equivs): Likewise.
8289 (find_moveable_pseudos): Likewise.
8290 (split_live_ranges_for_shrink_wrap): Likewise.
8291 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8292 (rtx_reuse_manager::seen_def_p): Likewise.
8293 (rtx_reuse_manager::set_seen_def): Likewise.
8294 * print-rtl.h (class rtx_reuse_manager): Likewise.
8296 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8298 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8300 (migrate_btr_def): Likewise.
8301 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8302 * df-core.c (loop_post_order_compute): Likewise.
8303 (loop_inverted_post_order_compute): Likewise.
8304 * hsa-common.h: Likewise.
8305 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8306 * init-regs.c (initialize_uninitialized_regs): Likewise.
8307 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8308 (inline_small_functions): Likewise.
8309 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8310 * ira.c (combine_and_move_insns): Likewise.
8311 (build_insn_chain): Likewise.
8312 * loop-invariant.c (find_invariants): Likewise.
8313 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8314 * predict.c (tree_predict_by_opcode): Likewise.
8315 (predict_paths_leading_to): Likewise.
8316 (predict_paths_leading_to_edge): Likewise.
8317 (estimate_loops_at_level): Likewise.
8318 (estimate_loops): Likewise.
8319 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8320 (spread_components): Likewise.
8321 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8322 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8323 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8324 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8325 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8326 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8327 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8328 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8329 (mark_threaded_blocks): Likewise.
8330 (thread_through_all_blocks): Likewise.
8331 * tree-ssa.c (verify_ssa): Likewise.
8332 (execute_update_addresses_taken): Likewise.
8333 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8335 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8337 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8339 (post_order_compute): Likewise.
8340 (inverted_post_order_compute): Likewise.
8341 (pre_and_rev_post_order_compute_fn): Likewise.
8343 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8345 * genrecog.c (int_set::int_set): Explicitly construct our
8346 auto_vec base class.
8347 * vec.h (auto_vec::auto_vec): New constructor.
8349 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8351 * bitmap.h (class auto_bitmap): New constructor taking
8352 bitmap_obstack * argument.
8354 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8356 * bitmap.h (class auto_bitmap): Change type of m_bits to
8357 bitmap_head, and adjust ctor / dtor and member operators.
8359 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8361 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8362 when returned register mode doesn't match original mode.
8364 2017-05-12 Jeff Law <law@redhat.com>
8365 Jakub Jelinek <jakub@redhat.com>
8367 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8368 we look for cc setter after the compare-elim changes.
8369 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8370 within the vector to match what compare-elim now expects.
8371 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8372 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8374 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8375 after the compare-elim changes.
8376 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8377 the vector to match what compare-elim now expects.
8378 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8379 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8380 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8381 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8382 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8384 * config/visium/visium.c (single_set_and_flags): Fix where
8385 we look for cc setter after the compare-elim changes.
8386 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8387 with the vector to match what compare-elim now expects.
8388 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8389 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8390 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8391 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8392 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8393 (neg<mode>2_insn_set_overflow): Likewise.
8395 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8398 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8399 maybe_expand_insn call, set ops[0].target. If still set after call,
8400 set alt_rtl. Add extra arg to recursive calls.
8401 (extract_bit_field): Add alt_rtl argument. Pass to
8403 * expmed.h (extract_bit_field): Fix prototype.
8404 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8405 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8406 to extract_bit_field_calls.
8407 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8408 Pass alt_rtl to extract_bit_field calls.
8409 * calls.c (store_unaligned_arguments_into_psuedos)
8410 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8411 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8413 * optabs.h (struct expand_operand): Add target bitfield.
8415 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8417 * compare-elim.c (try_eliminate_compare): Canonicalize
8418 operation with embedded compare to
8419 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8420 (set (reg) (operation)].
8422 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8424 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8427 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8428 cost of adding a carry flag for ADC instruction.
8429 [case MINUS]: Ignore the cost of subtracting a carry flag
8430 for SBB instruction.
8432 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8434 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8436 * config/rs6000/bmiintrin.h: New file.
8437 * config/rs6000/bmi2intrin.h: New file.
8438 * config/rs6000/x86intrin.h: New file.
8440 2017-05-12 Jeff Law <law@redhat.com>
8442 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8445 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8448 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8450 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8452 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8453 early expansion of vector divide builtins.
8454 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8455 builtins identified as having unsigned arguments.
8457 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8459 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8460 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8461 expansion of vector logical operations (and, andc, or, xor,
8464 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8466 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8467 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8469 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8471 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8472 early GIMPLE expansion of vector multiplies.
8474 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8476 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8477 TARGET_HAVE_MOVT conditional.
8478 (movt splitter): Likewise.
8480 2017-05-12 Richard Biener <rguenther@suse.de>
8482 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8483 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8484 Fold all stmts not inplace.
8486 2017-05-12 Richard Biener <rguenther@suse.de>
8488 PR tree-optimization/80713
8489 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8490 inserted_exprs bit for not removed stmts.
8492 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8495 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8496 parallelized" attribute for parallelized OpenACC kernels.
8497 * omp-offload.c (execute_oacc_device_lower): Use it.
8499 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8500 Set "oacc kernels" attribute.
8501 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8502 parameter. Adjust all users.
8503 (oacc_fn_attrib_kernels_p): Remove function.
8504 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8505 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8506 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8507 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8508 assert "oacc kernels" attribute is set.
8510 2017-05-11 Carl Love <cel@us.ibm.com>
8512 * config/rs6000/rs6000-c: Add support for built-in functions
8513 vector unsigned char vec_popcnt (vector signed char)
8514 vector unsigned char vec_popcnt (vector unsigned char)
8515 vector unsigned short vec_popcnt (vector signed short)
8516 vector unsigned short vec_popcnt (vector unsigned short)
8517 vector unsigned int vec_popcnt (vector signed int)
8518 vector unsigned int vec_popcnt (vector unsigned int)
8519 vector unsigned long long vec_popcnt (vector signed long long)
8520 vector unsigned long long vec_popcnt (vector unsigned long long)
8521 vector signed long long vec_slo (vector signed long long,
8523 vector signed long long vec_slo (vector signed long long,
8524 vector unsigned char)
8525 vector unsigned long long vec_slo (vector unsigned long long,
8527 vector unsigned long long vec_slo (vector unsigned long long,
8528 vector unsigned char)
8529 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8530 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8531 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8532 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8533 * doc/extend.texi: Update the built-in documentation file for the
8534 new built-in functions.
8536 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8538 * attribs.h (sorted_attr_string): Move machine independent
8539 functions for target clone support from the i386 port to common
8540 code. Rename ix86_function_versions to common_function_versions.
8541 Rename make_name to make_unique_name.
8542 (common_function_versions): Likewise.
8543 (make_unique_name): Likewise.
8544 (make_dispatcher_decl): Likewise.
8545 (is_function_default_version): Likewise.
8546 * attribs.c (attr_strcmp): Likewise.
8547 (sorted_attr_string): Likewise.
8548 (common_function_versions): Likewise.
8549 (make_unique_name): Likewise.
8550 (make_dispatcher_decl): Likewise.
8551 (is_function_default_version): Likewise.
8552 * config/i386/i386.c (attr_strcmp): Likewise.
8553 (sorted_attr_string): Likewise.
8554 (ix86_function_versions): Likewise.
8555 (make_name): Likewise.
8556 (make_dispatcher_decl): Likewise.
8557 (is_function_default_version): Likewise.
8558 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8560 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8563 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8564 Account for direct move costs for vec_construct of integer
8567 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8570 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8571 (UNSPEC_STX_ATOMIC): Ditto.
8572 (loaddi_via_sse): New insn.
8573 (storedi_via_sse): Ditto.
8574 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8575 Update corresponding peephole2 patterns.
8576 (atomic_storedi_fpu): Ditto.
8578 2017-05-11 Julia Koval <julia.koval@intel.com>
8580 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8581 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8583 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8584 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8585 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8587 2017-05-11 Nathan Sidwell <nathan@acm.org>
8589 * graphite-poly.c: Include dumpfile.h.
8591 * dumpfle.h (dump_function): Declare here ...
8592 * tree-dump.h (dump_function): ... not here.
8593 * dumpfile.c: #include tree-cfg.h.
8594 (dump_function): Move here from ...
8595 * tree-dump.c (dump_function): ... here.
8596 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8597 * graphite-poly.c: Don't include tree-dump.h.
8598 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8599 * print-tree.c: Likewise.
8600 * stor-layout.c: Likewise.
8601 * tree-nested.c: Likewise.
8603 * dumpfile.c (dump_start): Use TDF_FLAGS.
8604 (dump_enable_all): Fix TDF_KIND check thinko.
8606 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8608 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8609 array entries to represent two legal parameterizations of the
8610 overloaded __builtin_cmpb function, as represented by the
8611 P6_OV_BUILTIN_CMPB constant.
8612 (altivec_resolve_overloaded_builtin): Add special case handling
8613 for the __builtin_cmpb function, as represented by the
8614 P6_OV_BUILTIN_CMPB constant.
8615 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8616 (BU_P6_64BIT_2): New macro.
8617 (BU_P6_OVERLOAD_2): New macro
8618 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8619 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8620 (CMPB): Add overload support to represent both 32-bit and 64-bit
8621 compare-bytes function.
8622 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8623 support for TARGET_CMPB.
8624 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8625 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8626 documentation of the __builtin_cmpb overloaded built-in function.
8628 2017-05-11 Richard Biener <rguenther@suse.de>
8630 PR tree-optimization/80705
8631 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8632 bases are not vectorizable.
8634 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8636 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8637 when counting register pressure.
8639 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8641 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8642 (iv_ca_more_deps): Renamed to ...
8643 (iv_ca_compare_deps): ... this.
8644 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8646 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8648 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8650 (determine_group_iv_costs): ... here.
8651 (find_inv_vars_cb): Record inv var if it's not recorded before.
8653 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8655 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8656 (get_shiftadd_cost): Ditto.
8658 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8660 * tree-ssa-address.c: Include header file.
8661 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8663 (add_to_parts): Refactor.
8664 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8665 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8668 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8670 PR tree-optimization/53090
8671 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8673 (extract_cond_operands): Detect condition with IV on both sides
8674 and return COMP_IV_EXPR_2.
8675 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8676 (rewrite_use_compare): Simplify by removing call to function
8677 extract_cond_operands.
8679 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8681 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8682 (extract_cond_operands): Detect condition comparing against non-
8683 invariant bound and return appropriate enum value.
8684 (find_interesting_uses_cond): Update use of extract_cond_operands.
8685 Handle its return value accordingly.
8686 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8688 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8690 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8691 nonlinear iv_use computation in loop invariant sensitive way.
8693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8695 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8696 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8698 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8700 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8701 (dump_cand): Support iv_cand.inv_exprs.
8702 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8704 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8707 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8709 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8711 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8712 as local function. Include necessary header files.
8713 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8715 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8717 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8721 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8722 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8723 RSHIFT_EXPR and BIT_NOT_EXPR.
8725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8727 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8728 (adjust_setup_cost): New parameter supporting round up adjustment.
8729 (struct address_cost_data): Delete.
8730 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8731 (split_address_cost, ptr_difference_cost): Delete.
8732 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8733 (struct ainc_cost_data): New struct.
8734 (get_address_cost_ainc): New function.
8735 (get_address_cost, get_computation_cost): Reimplement.
8736 (determine_group_iv_cost_address): Record inv_expr for all uses of
8738 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8739 (iv_ca_has_deps): Reimplemented to ...
8740 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8742 (iv_ca_extend): Call iv_ca_more_deps.
8744 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8746 * tree-ssa-address.c (struct mem_address): Move to header file.
8747 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8748 * tree-ssa-address.h (struct mem_address): Move from C file.
8749 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8751 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8753 * tree-affine.h (aff_combination_type): New interface.
8754 (aff_combination_zero_p): Remove static.
8755 (aff_combination_const_p): New interface.
8756 (aff_combination_singleton_var_p): New interfaces.
8758 2017-05-11 Richard Biener <rguenther@suse.de>
8760 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8761 Skip unreachable blocks and destinations.
8762 (eliminate): Move stmt removal and fixup ...
8763 (fini_eliminate): ... here. Skip inserted exprs.
8764 (pass_pre::execute): Move fini_pre after fini_eliminate.
8765 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8766 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8767 PRE to get rid of dead code that has invalid SSA form and
8768 split critical edges again.
8770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8772 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8774 2017-05-11 Richard Biener <rguenther@suse.de>
8776 * passes.c (execute_function_todo): Verify loops if they are
8777 said to be up-to-date.
8778 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8779 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8781 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8784 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8785 handle calling assemble_external ourself.
8788 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8789 modes with zero size. Enhance comment.
8791 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8793 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8794 built-ins for vec_xl and vec_xst with short and char pointer
8797 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8799 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8800 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8801 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8802 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8803 (_mm_maskz_min_round_ss): New intrinsics.
8804 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8805 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8806 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8807 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8808 (__builtin_ia32_minss_mask_round): New builtins.
8809 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8810 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8811 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
8813 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
8814 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
8816 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
8819 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8821 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
8822 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
8823 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
8824 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
8825 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
8826 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
8827 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
8828 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8829 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8830 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
8831 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
8832 (__builtin_ia32_mulss_mask_round): New builtins.
8833 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8834 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8835 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
8837 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
8838 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8840 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8843 2017-05-10 Julia Koval <julia.koval@intel.com>
8845 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
8846 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
8847 (_mm256_setr_m128i): New intrinsics.
8849 2017-05-10 Julia Koval <julia.koval@intel.com>
8851 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
8852 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
8853 (_mm_maskz_rcp14_ss): New intrinsics.
8854 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
8855 (__builtin_ia32_rcp14ss_mask): New builtins.
8856 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
8858 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
8860 PR tree-optimization/51513
8861 * tree-cfg.c (gimple_seq_unreachable_p): New function.
8862 (assert_unreachable_fallthru_edge_p): Use it.
8863 (group_case_labels_stmt): Likewise.
8864 * tree-cfg.h: Prototype it.
8865 * stmt.c: Include cfghooks.h and tree-cfg.h.
8866 (emit_case_dispatch_table) <gap_label>: New local variable.
8867 Use it to fill dispatch table gaps.
8868 Test for default_label before updating probabilities.
8869 (expand_case) <default_label>: Remove unneeded initialization.
8870 Test for unreachable default case statement and remove its edge.
8871 Set default_label accordingly.
8872 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
8874 2017-05-10 Carl Love <cel@us.ibm.com>
8876 * config/rs6000/rs6000-c: Add support for built-in functions
8877 vector signed char vec_neg (vector signed char)
8878 vector signed short int vec_neg (vector short int)
8879 vector signed int vec_neg (vector signed int)
8880 vector signed long long vec_neg (vector signed long long)
8881 vector float vec_neg (vector float)
8882 vector double vec_neg (vector double)
8883 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
8885 * config/rs6000/altivec.h: Add define for vec_neg
8886 * doc/extend.texi: Update the built-in documentation for the
8887 new built-in functions.
8889 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8891 PR tree-optimization/77644
8892 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
8894 2017-05-10 Nathan Sidwell <nathan@acm.org>
8896 * dumpfile.h (TDI_lang_all): New.
8897 (TDF_KIND): New. Renumber others
8898 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
8900 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
8902 (get_dump_file_name): Adjust suffix generation.
8903 (dump_enable_all): Use TDF_KIND.
8904 * doc/invoke.texi (-fdump-lang-all): Document.
8906 * dumpfile.h: Tabify.
8908 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
8911 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
8912 Move member access before delete.
8914 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
8916 * tree-inline.c (expand_call_inline): Split block at stmt
8919 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8922 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
8923 are now unused after splitting mov{sf,sd}_hardfloat.
8924 (f32_lr2): Likewise.
8926 (f32_lm2): Likewise.
8928 (f32_li2): Likewise.
8931 (f32_sr2): Likewise.
8933 (f32_sm2): Likewise.
8935 (f32_si2): Likewise.
8938 (f32_vsx): Likewise.
8940 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
8941 For movsf, order stores so the VSX stores occur before the GPR
8942 store which encourages the register allocator to use a traditional
8943 FPR instead of a GPR. For movsd, order the stores so that the GPR
8944 store comes before the VSX stores to allow the power6 to work.
8945 This is due to the power6 not having a 32-bit integer store
8946 instruction from a FPR.
8947 (movsf_hardfloat): Likewise.
8948 (movsd_hardfloat): Likewise.
8950 2017-05-09 Martin Sebor <msebor@redhat.com>
8952 PR translation/80280
8953 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
8956 PR translation/80280
8957 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
8958 data member added in r247778.
8959 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
8961 2017-05-09 Nathan Sidwell <nathan@acm.org>
8963 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
8965 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
8968 2017-05-09 Marek Polacek <polacek@redhat.com>
8970 * doc/invoke.texi: Fix typo.
8972 2017-05-09 Richard Biener <rguenther@suse.de>
8974 * tree-vrp.c (vrp_val_is_max): Adjust comment.
8975 (vrp_val_is_min): Likewise.
8976 (set_value_range_to_value): Likewise.
8977 (set_value_range_to_nonnegative): Likewise.
8978 (gimple_assign_nonzero_p): Likewise.
8979 (gimple_stmt_nonzero_p): Likewise.
8980 (vrp_int_const_binop): Likewise. Remove unreachable case.
8981 (adjust_range_with_scev): Adjust comments.
8982 (compare_range_with_value): Likewise.
8983 (extract_range_from_phi_node): Likewise.
8984 (test_for_singularity): Likewise.
8986 2017-05-09 Richard Biener <rguenther@suse.de>
8988 * tree-vrp.c (get_single_symbol): Add assert that we don't
8989 get overflowed constants as invariant part.
8990 (compare_values_warnv): Add comment before the TREE_NO_WARNING
8991 checks. Use wi::cmp instead of recursing for integer constants.
8992 (compare_values): Just ignore whether we assumed undefined
8993 overflow instead of failing the compare.
8994 (extract_range_for_var_from_comparison_expr): Add comment before the
8995 TREE_NO_WARNING sets.
8996 (test_for_singularity): Likewise.
8997 (extract_range_from_comparison): Do not disable optimization
8998 when we assumed undefined overflow.
8999 (extract_range_basic): Remove init of unused var.
9001 2017-05-09 Richard Biener <rguenther@suse.de>
9003 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9004 (extract_range_from_multiplicative_op_1): Adjust.
9005 (extract_range_from_binary_expr_1): Use int_const_binop.
9007 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9010 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9011 rs6000_store_data_bypass_p in seven define_bypass directives and
9012 in several comments.
9013 * config/rs6000/rs6000-protos.h: Add prototype for
9014 rs6000_store_data_bypass_p function.
9015 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9016 function implements slightly different (rs6000-specific) semantics
9017 than store_data_bypass_p, returning false rather than aborting
9018 with assertion error when arguments do not satisfy the
9019 requirements of store data bypass.
9020 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9021 rs6000_store_data_bypass_p.
9023 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9025 * config/xtensa/xtensa-protos.h
9026 (xtensa_initial_elimination_offset): New declaration.
9027 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9028 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9029 macro definition, add case for FRAME_POINTER_REGNUM when
9030 FRAME_GROWS_DOWNWARD.
9031 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9032 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9033 xtensa_initial_elimination_offset.
9035 2017-05-08 Nathan Sidwell <nathan@acm.org>
9037 * doc/invoke.texi: Alphabetize -fdump options.
9039 2017-05-08 Martin Sebor <msebor@redhat.com>
9041 PR translation/80280
9042 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9044 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9046 * target.def (compute_frame_layout): New optional target hook.
9047 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9048 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9049 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9051 * reload1.c (verify_initial_elim_offsets): Likewise.
9052 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9053 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9054 (arm_get_frame_offsets): Split up into this ...
9055 (arm_compute_frame_layout): ... and this function.
9057 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9059 * config/aarch64/constraints.md (Usa): New constraint.
9060 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9062 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9064 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9065 with_multilib_list after it has been checked.
9067 2017-05-08 Richard Biener <rguenther@suse.de>
9069 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9070 (bitmap_set_subtract_values): Likewise.
9072 2017-05-08 Richard Biener <rguenther@suse.de>
9074 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9075 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9077 (gimple_stmt_nonzero_warnv_p): Rename to ...
9078 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9080 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9081 (extract_range_basic): Adjust, do not disable propagation on
9082 strict overflow sensitive simplification.
9083 (vrp_visit_cond_stmt): Likewise.
9085 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9087 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9088 body size unconditionally.
9090 2017-05-07 Jeff Law <law@redhat.com>
9093 2017-05-06 Jeff Law <law@redhat.com>
9094 PR tree-optimization/78496
9095 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9098 PR tree-optimization/78496
9099 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9100 (simplify_stmt_using_ranges): Call it.
9101 (vrp_dom_walker::before_dom_children): Extract equivalences
9102 from an ASSERT_EXPR with an equality comparison against a
9105 2017-05-06 Jeff Law <law@redhat.com>
9107 PR tree-optimization/78496
9108 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9111 PR tree-optimization/78496
9112 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9113 (simplify_stmt_using_ranges): Call it.
9114 (vrp_dom_walker::before_dom_children): Extract equivalences
9115 from an ASSERT_EXPR with an equality comparison against a
9118 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9120 * lra-constraints.c (lra_copy_reg_equiv): New function.
9121 (split_reg): Use it to copy equivalence information from the
9122 original register to the spill register.
9124 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9126 PR rtl-optimization/75964
9127 * simplify-rtx.c (simplify_const_relational_operation): Remove
9128 invalid handling of comparisons of integer ABS.
9130 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9132 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9134 (init_regs): Remove declaration.
9135 (function_arg_advance_32): Initialize error_p as boolean variable.
9137 2017-05-05 Nathan Sidwell <nathan@acm.org>
9139 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9140 lines. Use for (;;).
9142 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9144 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9145 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9146 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9147 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9148 VF=2 that require versioning.
9150 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9152 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9155 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9157 * diagnostic.h (diagnostic_override_option_index): Convert from
9158 macro to inline function.
9160 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9162 * diagnostic.c (last_module_changed_p): New function.
9163 (set_last_module): New function.
9164 (diagnostic_report_current_module): Convert macro usage to
9165 the above functions.
9166 * diagnostic.h (diagnostic_context::last_module): Strengthen
9167 from const line_map * to const line_map_ordinary *.
9168 (diagnostic_last_module_changed): Delete macro.
9169 (diagnostic_set_last_module): Delete macro.
9171 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9173 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9174 with diagnostic_report_diagnostic.
9175 (diagnostic_n_impl_richloc): Likewise.
9176 * diagnostic.h (report_diagnostic): Delete macro.
9177 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9178 with diagnostic_report_diagnostic.
9179 * substring-locations.c (format_warning_va): Likewise.
9181 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9183 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9184 save/restor of format_spec. Move option-printing code to...
9185 (print_option_information): ...this new function, and
9186 reimplement by simply printing to the pretty_printer,
9187 rather than appending to the format string.
9189 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9191 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9192 handling logic into...
9193 (update_effective_level_from_pragmas): ...this new function.
9195 2017-05-04 Andrew Waterman <andrew@sifive.com>
9197 * config/riscv/riscv.opt (mstrict-align): New option.
9198 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9199 (SLOW_UNALIGNED_ACCESS): Define.
9200 (riscv_slow_unaligned_access): Declare.
9201 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9203 (riscv_slow_unaligned_access): New variable.
9204 (rocket_tune_info): Set slow_unaligned_access to true.
9205 (optimize_size_tune_info): Set slow_unaligned_access to false.
9206 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9207 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9208 (riscv_option_override): Set riscv_slow_unaligned_access.
9209 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9211 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9213 * config/riscv/riscv.md: Unify indentation.
9215 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9220 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9222 (extendsi<mode>2): Add support for doing sign extension via
9223 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9224 don't have ISA 3.0 instructions.
9225 (extendsi<mode>2 splitter): Likewise.
9226 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9227 generate the normal insns since SImode can now go in vector
9228 registers. Disallow the special UNSPECs needed for previous
9229 machines to hide SImode being used. Add new insns
9230 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9231 (fix_trunc<mode>si2_stfiwx): Likewise.
9232 (fix_trunc<mode>si2_internal): Likewise.
9233 (fixuns_trunc<mode>si2): Likewise.
9234 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9235 (fctiw<u>z_<mode>_smallint): Likewise.
9236 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9237 of floating point to 32-bit integer from doing a direct move to
9238 the GPR registers to do a store.
9239 (fctiwz_<mode>): Break long line.
9241 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9243 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9244 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9245 (addr_list, addr_offset_valid_p): New.
9246 (split_address_groups): Check offset validity with above function.
9247 (gt-tree-ssa-loop-ivopts.h): Include header file.
9249 2017-05-05 Nathan Sidwell <nathan@acm.org>
9251 * config.gcc (arm*-*-*): Add missing 'fi'.
9253 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9255 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9256 included in -fopt-info does not matter.
9257 * doc/optinfo.texi (-fopt-info): Fix description of default
9258 behavour. Explicitly say order of options included in -fopt-info
9261 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9263 * config.gcc: Allow combinations of aprofile and rmprofile values for
9264 --with-multilib-list.
9265 * config/arm/t-multilib: New file.
9266 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9267 variables. Remove setting of ISA and floating-point ABI in
9268 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9269 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9270 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9271 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9273 * config/arm/t-rmprofile: Likewise except for the matches changes.
9274 * doc/install.texi (--with-multilib-list): Document the combination of
9275 aprofile and rmprofile values and warn about pitfalls in doing that.
9277 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9279 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9280 (movdi_aarch64): Likewise.
9282 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9284 PR tree-optimization/80632
9285 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9287 (build_arrays): Initialize it for virtual phis.
9288 (fix_phi_nodes): Use it for virtual phis.
9290 PR tree-optimization/80558
9291 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9292 [x, y] op z into [x op, y op z] for op & or | if conditions
9295 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9296 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9299 * config/arm/arm.md (use_literal_pool): Remove.
9300 (64-bit immediate split): No longer takes cost into consideration
9301 if arm_disable_literal_pool is enabled.
9302 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9303 used when arm_disable_literal_pool is enabled.
9304 (arm_max_const_double_inline_cost): Remove use of
9305 arm_disable_literal_pool.
9306 (push_minipool_fix): Add assert.
9307 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9308 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9309 (no_literal_pool_sf_immediate): New.
9311 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9313 PR tree-optimization/80613
9314 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9315 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9317 2017-05-05 Richard Biener <rguenther@suse.de>
9319 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9321 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9323 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9324 of this flag from insn conditions due to removal from r247495.
9326 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9328 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9330 (arm_early_store_addr_dep_ptr): Likewise.
9331 * config/arm/aarch-common-protos.h
9332 (arm_early_load_addr_dep_ptr): Add prototype.
9333 (arm_early_store_addr_dep_ptr): Likewise.
9334 * config/arm/cortex-a53.md: Add new bypasses.
9336 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9338 * tree.c (next_type_uid): Change type to unsigned.
9339 (type_hash_canon): Decrement back next_type_uid if
9340 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9341 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9344 2017-05-04 Martin Sebor <msebor@redhat.com>
9346 * builtins.c: Fix a trivial typo in a comment.
9349 * builtins.c (check_sizes): Adjust to handle reading past the end.
9350 Avoid printing excessive upper bound of ranges. Use %E to print
9351 tree nodes instead of converting them to %wu.
9352 (expand_builtin_memchr): New function.
9353 (compute_dest_size): Rename...
9354 (compute_objsize): ...to this.
9355 (expand_builtin_memcpy): Adjust.
9356 (expand_builtin_mempcpy): Adjust.
9357 (expand_builtin_strcat): Adjust.
9358 (expand_builtin_strcpy): Adjust.
9359 (check_strncat_sizes): Adjust.
9360 (expand_builtin_strncat): Adjust.
9361 (expand_builtin_strncpy): Adjust and simplify.
9362 (expand_builtin_memset): Adjust.
9363 (expand_builtin_bzero): Adjust.
9364 (expand_builtin_memcmp): Adjust.
9365 (expand_builtin): Handle memcmp.
9366 (maybe_emit_chk_warning): Check strncat just once.
9368 2017-05-04 Martin Sebor <msebor@redhat.com>
9370 PR preprocessor/79214
9373 * builtins.c (check_sizes): Add inlining context and issue
9374 warnings even when -Wno-system-headers is set.
9375 (check_strncat_sizes): Same.
9376 (expand_builtin_strncat): Same.
9377 (expand_builtin_memmove): New function.
9378 (expand_builtin_stpncpy): Same.
9379 (expand_builtin): Handle memmove and stpncpy.
9381 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9383 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9384 which is not used any more.
9386 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9388 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9390 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9392 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9393 (cortexa53_tunings): Likewise.
9394 (cortexa57_tunings): Likewise.
9395 (cortexa72_tunings): Likewise.
9396 (cortexa73_tunings): Likewise.
9398 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9400 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9401 Set loop alignment to 8.
9403 2017-05-04 Martin Sebor <msebor@redhat.com>
9405 PR translation/80280
9406 * builtins.c (expand_builtin_object_size): Add missing quoting to
9407 %D and like directives.
9408 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9409 (hsa_type_for_tree_type): Same.
9410 (verify_function_arguments): Same.
9411 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9412 * varasm.c (get_section): Same.
9415 2017-05-04 Martin Sebor <msebor@redhat.com>
9417 PR translation/80280
9418 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9420 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9422 * config/aarch64/aarch64.c (generic_addrcost_table):
9423 Change HI/TI mode setting.
9425 2017-05-04 Martin Jambor <mjambor@suse.cz>
9427 PR tree-optimization/80622
9428 * tree-sra.c (comes_initialized_p): New function.
9429 (build_accesses_from_assign): Only set write lazily when
9430 comes_initialized_p is false.
9431 (analyze_access_subtree): Use comes_initialized_p.
9432 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9433 instead of testing for PARM_DECL.
9435 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9437 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9438 constraint on operand 0 to allow more general addressing modes.
9439 Adjust output template.
9440 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9442 * config/aarch64/aarch64-protos.h
9443 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9444 * config/aarch64/constraints.md (Dp): New address constraint.
9445 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9448 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9450 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9451 update use of estimate_ipcp_clone_size_and_time.
9452 (estimate_local_effects): Update use of
9453 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9454 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9455 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9456 Return nonspecialized time.
9458 2017-05-04 Richard Biener <rguenther@suse.de>
9460 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9461 for the last VUSE which def dominates the PHI. Directly call
9463 (get_continuation_for_phi_1): Remove.
9465 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9467 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9468 to explain the use of truncating division. Cap the number of
9469 iterations to the maximum given by nb_iterations_upper_bound,
9472 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9474 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9475 * configure: Regenerate.
9476 * config.in: Regenerate.
9477 * config/i386/driver-mingw32.c: new file.
9478 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9479 * config.host: Link driver-mingw32.o on MinGW host.
9480 * doc/install.texi: Document new --enable-mingw-wildcard configure
9483 2017-05-04 Marek Polacek <polacek@redhat.com>
9485 PR tree-optimization/80612
9486 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9488 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9489 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9491 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9492 (movt splitter): Likewise.
9493 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9494 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9495 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9496 block for Thumb-1 with MOVT.
9497 (thumb2_legitimate_address_p): Move code block ...
9498 (can_avoid_literal_pool_for_label_p): ... into this new function.
9499 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9501 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9502 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9503 "M-profile targets with the MOVT instruction".
9505 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9507 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9508 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9509 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9511 2017-05-04 Martin Liska <mliska@suse.cz>
9513 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9516 2017-05-04 Richard Biener <rguenther@suse.de>
9518 * tree.c (array_at_struct_end_p): Handle arrays at struct
9519 end with flexarrays more conservatively. Refactor and treat
9520 arrays of arrays or aggregates more strict. Fix
9521 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9522 * tree.c (array_at_struct_end_p): Adjust prototype.
9523 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9524 * gimple-fold.c (get_range_strlen): Likewise.
9525 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9527 2017-05-04 Richard Biener <rguenther@suse.de>
9529 PR tree-optimization/31130
9530 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9532 (supports_overflow_infinity): Likewise.
9533 (is_negative_overflow_infinity): Likewise.
9534 (is_positive_overflow_infinity): Likewise.
9535 (is_overflow_infinity): Likewise.
9536 (stmt_overflow_infinity): Likewise.
9537 (overflow_infinity_range_p): Likewise.
9538 (usable_range_p): Remove as always returning true.
9539 (make_overflow_infinity): Remove.
9540 (negative_overflow_infinity): Likewise.
9541 (positive_overflow_infinity): Likewise.
9542 (avoid_overflow_infinity): Likewise.
9543 (set_value_range): Adjust accordingly.
9544 (set_value_range_to_nonnegative): Likewise, remove now unused
9545 overflow_infinity arg.
9546 (vrp_operand_equal_p): Adjust.
9547 (update_value_range): Likewise.
9548 (range_int_cst_singleton_p): Likewise.
9549 (operand_less_p): Likewise.
9550 (compare_values_warnv): Likewise.
9551 (extract_range_for_var_from_comparison_expr): Likewise.
9552 (vrp_int_const_binop): Likewise.
9553 (zero_nonzero_bits_from_vr): Likewise.
9554 (extract_range_from_multiplicative_op_1): Likewise.
9555 (extract_range_from_binary_expr_1): Likewise.
9556 (extract_range_from_unary_expr): Likewise.
9557 (extract_range_from_comparison): Likewise.
9558 (extract_range_basic): Likewise.
9559 (adjust_range_with_scev): Likewise.
9560 (compare_ranges): Likewise.
9561 (compare_range_with_value): Likewise.
9562 (dump_value_range): Likewise.
9563 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9565 (simplify_cond_using_ranges): Adjust.
9567 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9569 * brig-builtins.def: Added a builtin for class_f64.
9570 * builtin-types.def: Added a builtin type needed by class_f64.
9572 2017-05-03 Jason Merrill <jason@redhat.com>
9574 * timevar.def: Add TV_CONSTEXPR.
9576 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9578 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9580 2017-05-03 Martin Jambor <mjambor@suse.cz>
9582 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9583 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9584 * ipa-cp.c (ipcp_propagate_stage): Do not call
9585 ipa_update_after_lto_read.
9586 * ipa-inline.c (ipa_inline): Likewise.
9588 2017-05-03 Martin Jambor <mjambor@suse.cz>
9590 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9591 tag. Added a default constructor and a destructor.
9592 (ipa_edge_args_sum_t): New class;
9593 (ipa_edge_args_sum): Declare.
9594 (ipa_edge_args_vector): Remove declaration.
9595 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9596 (ipa_free_edge_args_substructures): Remove declaration.
9597 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9598 (ipa_edge_args_info_available_for_edge_p): Likewise.
9599 * ipa-prop.c (ipa_edge_args_vector): Removed.
9600 (edge_removal_hook_holder): Likewise.
9601 (edge_duplication_hook_holder): Likewise.
9602 (ipa_edge_args_sum): New variable.
9603 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9604 ipa_edge_args_vector.
9605 (ipa_free_edge_args_substructures): Likewise.
9606 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9607 ipa_edge_args_vector.
9608 (ipa_edge_removal_hook): Turned into method
9609 ipa_edge_args_sum_t::remove.
9610 (ipa_edge_duplication_hook): Turned into method
9611 ipa_edge_args_sum_t::duplicate.
9612 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9613 registering edge hooks.
9614 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9615 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9616 ipa_edge_args_sum instead of ipa_edge_args_vector.
9617 * ipa-profile.c (ipa_profile): Likewise.
9619 2017-05-03 Martin Jambor <mjambor@suse.cz>
9621 * symbol-summary.h (function_summary): New method exists.
9622 (function_summary::symtab_removal): Deallocate through release.
9623 (call_summary): New class.
9624 (gt_ggc_mx): New overload.
9625 (gt_pch_nx): Likewise.
9626 (gt_pch_nx): Likewise.
9628 2017-05-03 Jeff Law <law@redhat.com>
9630 PR tree-optimization/78496
9631 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9632 from simplify_cond_using_ranges. Split off code to walk
9633 backwards through casts into ...
9634 (simplify_cond_using_ranges_2): New function.
9635 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9636 (execute_vrp): After identifying jump threads, call
9637 simplify_cond_using_ranges_2.
9639 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9642 * ipa-inline.h (inline_summary): Add ctor.
9643 (create_ggc): Do not use ggc_cleared_alloc.
9645 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9646 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9648 * gcc.c (handle_braces): Support escaping in switch matching
9650 * doc/invoke.texi (Spec Files): Document it.
9651 Remove superfluous @code markup in items.
9653 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9655 * diagnostic-show-locus.c (struct column_range): New struct.
9656 (get_affected_columns): New function.
9657 (get_printed_columns): New function.
9658 (struct correction): New struct.
9659 (correction::ensure_capacity): New function.
9660 (correction::ensure_terminated): New function.
9661 (struct line_corrections): New struct.
9662 (line_corrections::~line_corrections): New dtor.
9663 (line_corrections::add_hint): New function.
9664 (layout::print_trailing_fixits): Reimplement in terms of the new
9666 (selftest::test_overlapped_fixit_printing): New function.
9667 (selftest::diagnostic_show_locus_c_tests): Call it.
9669 2017-05-03 Nathan Sidwell <nathan@acm.org>
9671 Canonicalize canonical type hashing
9672 * tree.h (type_hash_canon_hash): Declare.
9673 * tree.c (type_hash_list, attribute_hash_list): Move into
9674 type_hash_canon_hash.
9675 (build_type_attribute_qual_variant): Break out hash code calc into
9676 type_hash_canon_hash.
9677 (type_hash_canon_hash): New. Generic type hash computation.
9678 (build_range_type_1, build_array_type_1, build_function_type,
9679 build_method_type_directly, build_offset_type, build_complex_type,
9680 make_vector_type): Call it.
9682 2017-05-03 Richard Biener <rguenther@suse.de>
9684 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9685 When all DRs have unknown misaligned do not always peel
9686 when there is a store but apply the same costing model as if
9687 there were only loads.
9689 2017-05-03 Richard Biener <rguenther@suse.de>
9692 PR tree-optimization/80492
9693 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9694 compare_base_decls returning dont-know properly.
9696 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9698 * config/arm/iterators.md (CCSI): New mode iterator.
9699 (arch): New mode attribute.
9700 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9701 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9702 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9703 code iterator for success result mode.
9704 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9705 the corresponding new insn generators.
9707 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9710 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9711 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9713 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9715 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9716 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9717 (DDR_A): Wrap DDR argument in brackets.
9718 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9719 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9720 (DDR_REVERSED_P): Likewise.
9722 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9724 PR tree-optimization/79472
9725 * tree-switch-conversion.c (struct switch_conv_info): Add
9726 contiguous_range and default_case_nonstandard fields.
9727 (collect_switch_conv_info): Compute contiguous_range and
9728 default_case_nonstandard fields, don't clear final_bb if
9729 contiguous_range and only the default case doesn't have the required
9731 (check_all_empty_except_final): Set default_case_nonstandard instead
9732 of failing if contiguous_range and the default case doesn't have empty
9734 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9735 and only the default case doesn't have the required constants. Skip
9737 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9738 if default_case_nonstandard.
9739 (build_constructors): Build constant 1 just once. Assert that default
9740 values aren't inserted in between cases if contiguous_range. Skip
9742 (build_arrays): Skip virtual phis.
9743 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9744 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9745 Handle virtual phis.
9746 (gen_inbound_check): Handle default_case_nonstandard case.
9747 (process_switch): Adjust check_final_bb caller. Call
9748 gather_default_values with the first non-default case instead of
9749 default case if default_case_nonstandard.
9751 2017-05-02 Nathan Sidwell <nathan@acm.org>
9753 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9754 check. Fix formatting.
9756 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9758 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9759 errors when comparing specialized and unspecialized times.
9761 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9763 * diagnostic-show-locus.c
9764 (layout::should_print_annotation_line_p): Make private.
9765 (layout::print_annotation_line): Make private.
9766 (layout::annotation_line_showed_range_p): Make private.
9767 (layout::show_ruler): Make private.
9768 (layout::print_source_line): Make private. Pass in line and
9769 line_width, rather than calling location_get_source_line. Drop
9771 (layout::print_leading_fixits): New method.
9772 (layout::print_any_fixits): Rename to...
9773 (layout::print_trailing_fixits): ...this, and make private.
9774 Don't print newline fixits.
9775 (diagnostic_show_locus): Move logic for printing one row into...
9776 (layout::print_line): ...this new function. Move the
9777 location_get_source_line call and error-handling from
9778 print_source_line to here. Call print_leading_fixits, and rename
9779 print_any_fixits to print_trailing_fixits.
9780 (selftest::test_fixit_insert_containing_newline): Update now that
9781 newlines are partially supported.
9782 (selftest::test_fixit_insert_containing_newline_2): New test.
9783 (selftest::test_fixit_replace_containing_newline): Update comments.
9784 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9785 * edit-context.c (class added_line): New class.
9786 (class edited_line): Describe newline handling in comment.
9787 (edited_line::actually_edited_p): New method.
9788 (edited_line::print_content): Delete redundant decl.
9789 (edited_line::m_predecessors): New field.
9790 (edited_file::print_content): Call edited_line::print_content.
9791 (edited_file::print_diff): Update to support newlines.
9792 (edited_file::print_diff_hunk): Likewise.
9793 (edited_file::print_run_of_changed_lines): New function.
9794 (edited_file::print_diff_line): Convert to...
9795 (print_diff_line): ...this.
9796 (edited_file::get_effective_line_count): New function.
9797 (edited_line::edited_line): Initialize new field m_predecessors.
9798 (edited_line::~edited_line): Clean up m_predecessors.
9799 (edited_line::apply_fixit): Handle newlines.
9800 (edited_line::get_effective_line_count): New function.
9801 (edited_line::print_content): New function.
9802 (edited_line::print_diff_lines): New function.
9803 (selftest::test_applying_fixits_insert_containing_newline): New
9805 (selftest::test_applying_fixits_replace_containing_newline): New
9807 (selftest::insert_line): New function.
9808 (selftest::test_applying_fixits_multiple_lines): Add example of
9810 (selftest::edit_context_c_tests): Call the new tests.
9812 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9814 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
9815 parameter cand. Update dump information.
9816 (get_computation_cost): Update uses.
9818 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9820 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
9821 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
9822 (get_computation_at, rewrite_use_address): Update use of
9823 get_computation_aff.
9825 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9827 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
9828 (get_computation): Delete.
9829 (get_computation_cost): Implement like get_computation_cost_at.
9830 Use get_computation_at.
9831 (get_computation_cost_at): Delete.
9832 (rewrite_use_nonlinear_expr): Use get_computation_at.
9833 (rewrite_use_compare, remove_unused_ivs): Ditto.
9835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9837 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
9839 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9841 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
9842 (ivopts_global_cost_for_size): Rename parameter and update uses.
9843 (iv_ca_recount_cost): Update uses.
9844 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
9845 candidates seperately in n_invs and n_cands.
9846 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
9848 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9850 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
9851 (find_inv_vars_cb): New.
9852 (find_depends): Renamed to ...
9853 (find_inv_vars): ... this.
9854 (add_candidate_1, force_var_cost): Call find_inv_vars.
9855 (split_address_cost, determine_group_iv_cost_cond): Ditto.
9857 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9859 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
9860 inv_vars. Add inv_exprs.
9861 (struct iv_cand): Rename depends_on to inv_vars.
9862 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
9863 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
9864 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
9865 (dump_cand): Dump inv_vars.
9866 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
9867 (record_invariant, find_depends, add_candidate_1): Ditto.
9868 (set_group_iv_cost, force_var_cost): Ditto.
9869 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
9870 (get_computation_cost_at, get_computation_cost): Ditto.
9871 (determine_group_iv_cost_generic): Ditto.
9872 (determine_group_iv_cost_address): Ditto.
9873 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
9874 (determine_group_iv_costs): Ditto.
9875 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
9876 (iv_ca_set_remove_invariants): Renamed to ...
9877 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
9878 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
9879 (iv_ca_set_add_invariants): Renamed to ...
9880 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
9881 (iv_ca_set_cp): Use iv_ca_set_add_invs.
9882 (iv_ca_has_deps): Support inv_vars and inv_exprs.
9883 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
9884 (create_new_ivs): Remove useless dump.
9886 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9888 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
9890 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
9891 (iv_ca_set_no_cp, create_new_iv): Ditto.
9893 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9895 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9897 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9899 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
9900 function tree_check2.
9902 2017-05-02 Martin Liska <mliska@suse.cz>
9904 * doc/gcov.texi: Add missing preposition.
9905 * gcov.c (function_info::function_info): Properly fill up
9906 all member variables.
9908 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9910 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
9912 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9914 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
9916 2017-05-02 Martin Liska <mliska@suse.cz>
9919 * lto-streamer-in.c (lto_read_tree_1): Remove
9921 * lto-streamer.c (struct tree_hash_entry): Likewise.
9922 (struct tree_entry_hasher): Likewise.
9923 (tree_entry_hasher::hash): Likewise.
9924 (tree_entry_hasher::equal): Likewise.
9925 (lto_streamer_init): Likewise.
9926 (lto_orig_address_map): Likewise.
9927 (lto_orig_address_get): Likewise.
9928 (lto_orig_address_remove): Likewise.
9929 * lto-streamer.h: Likewise.
9930 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
9931 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
9933 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
9935 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
9936 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
9937 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
9938 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
9939 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
9940 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
9941 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
9942 (mm_maskz_sub_ss): New intrinsics.
9943 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9944 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9945 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
9946 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
9947 (__builtin_ia32_subss_mask_round): New builtins.
9948 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9949 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9950 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
9952 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
9953 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9955 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9958 2017-05-02 Martin Jambor <mjambor@suse.cz>
9960 PR tree-optimization/78687
9961 * tree-sra.c (access): New field parent.
9962 (process_subtree_disqualification): New function.
9963 (disqualify_candidate): Call it.
9964 (build_accesses_from_assign): Reset write flag if creating an
9966 (build_access_subtree): Fill in parent field and also prpagate
9967 down grp_write flag.
9968 (create_artificial_child_access): New parameter set_grp_write, set
9969 grp_write to its value.
9970 (propagate_subaccesses_across_link): Also propagate grp_write flag
9972 (propagate_all_subaccesses): Push the closest parent back to work
9973 queue if add_access_to_work_queue returned true.
9975 2017-05-02 Richard Biener <rguenther@suse.de>
9977 * common.opt (fstrict-overflow): Alias negative to fwrapv.
9978 * doc/invoke.texi (fstrict-overflow): Remove all traces of
9979 -fstrict-overflow documentation.
9980 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
9981 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
9982 flag_strict_overflow.
9983 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
9984 * lto-opts.c (lto_write_options): Do not stream it.
9985 * lto-wrapper.c (merge_and_complain): Do not handle it.
9986 * opts.c (default_options_table): Do not set -fstrict-overflow.
9987 (finish_options): Likewise do not clear it when sanitizing.
9988 * simplify-rtx.c (simplify_const_relational_operation): Do not
9989 test flag_strict_overflow.
9991 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9993 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
9994 using enabled attribute.
9995 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
9996 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
9997 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
9998 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
9999 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10000 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10001 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10002 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10003 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10004 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10006 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10008 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10010 2017-05-02 Richard Biener <rguenther@suse.de>
10012 PR tree-optimization/80591
10014 2017-04-10 Richard Biener <rguenther@suse.de>
10016 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10019 2017-05-02 Richard Biener <rguenther@suse.de>
10021 PR tree-optimization/80549
10022 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10023 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10024 headers if they do not have a preheader.
10026 2017-05-02 Martin Liska <mliska@suse.cz>
10029 * common.opt: Fix typo.
10030 * doc/invoke.texi: Likewise.
10032 2017-05-01 Jan Beulich <jbeulich@suse.com>
10034 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10035 swapping, add (x,x,m,x,n) alternative.
10037 2017-05-01 Nathan Sidwell <nathan@acm.org>
10039 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10040 unnecessary unadjusted_alignment check.
10042 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10045 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10047 * gimplify.c (gimplify_cilk_detach): New function.
10048 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10049 * tree-core.h: Document EXPR_CILK_SPAWN.
10050 * tree.h (EXPR_CILK_SPAWN): Define.
10052 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10054 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10055 to use new fixit_hint representation, using the "replace" logic.
10056 (get_line_span_for_fixit_hint): Likewise.
10057 (layout::print_any_fixits): Likewise.
10058 (selftest::test_one_liner_many_fixits): Rename to...
10059 (selftest::test_one_liner_many_fixits_1): ...this, and update
10060 comment and expected output to reflect that the multiple fix-it
10061 hints are now consolidated into one insertion.
10062 (selftest::test_one_liner_many_fixits_2): New test.
10063 (selftest::test_diagnostic_show_locus_one_liner): Update for
10065 (selftest::test_fixit_consolidation): Update for fix-it API
10067 * diagnostic.c (print_parseable_fixits): Likewise.
10068 * edit-context.c (edited_line::m_line_events): Convert from
10069 auto_vec <line_event *> to auto_vec <line_event>.
10070 (class line_event): Convert from abstract base class to a concrete
10071 class, taking over the role of replace_event.
10072 (class insert_event): Delete.
10073 (class replace_event): Rename to class line_event. Convert to
10075 (edit_context::add_fixits): Reimplement.
10076 (edit_context::apply_insert): Delete.
10077 (edit_context::apply_replace): Rename to...
10078 (edit_context::apply_fixit): ...this. Convert to half-open range.
10079 (edited_file::apply_insert): Delete.
10080 (edited_file::apply_replace): Rename to...
10081 (edited_file::apply_fixit): ...this.
10082 (edited_line::~edited_line): Drop deletion of events.
10083 (edited_line::apply_insert): Delete.
10084 (edited_line::apply_replace): Rename to...
10085 (edited_line::apply_fixit): ...this. Convert to half-open range.
10086 Update for change to type of m_line_events.
10087 * edit-context.h (edit_context::apply_insert): Delete.
10088 (edit_context::apply_replace): Rename to...
10089 (edit_context::apply_fixit): ...this.
10091 2017-05-01 Martin Sebor <msebor@redhat.com>
10093 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10096 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10099 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10100 __get_cpuid_max returns 0.
10101 (__get_cpuid_count): Ditto.
10103 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10105 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10106 replacement expression is another instance of one of its arguments.
10108 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10111 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10112 check for stack push/pop autoinc.
10113 * config/i386/i386.c (ix86_agi_dependent): Return false
10114 if the only reason why modified_in_p returned true is that
10115 addr is SP based and set_insn is a push or pop.
10117 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10119 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10122 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10125 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10126 (account_size_time): Use two predicates - exec_pred and
10128 (evaluate_conditions_for_known_args): Compute both clause and
10130 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10131 (inline_summary_t::duplicate): Update.
10132 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10134 (compute_inline_parameters): Likewise.
10135 (estimate_edge_size_and_time): Update caluclation of time.
10136 (estimate_node_size_and_time): Compute both time and nonspecialized
10138 (estimate_ipcp_clone_size_and_time): Update.
10139 (inline_merge_summary): Update.
10140 (do_estimate_edge_time): Update.
10141 (do_estimate_edge_size): Update.
10142 (do_estimate_edge_hints): Update.
10143 (inline_read_section, inline_write_summary): Stream both new predicates.
10144 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10146 (compute_inlined_call_time): Cleanup.
10147 (big_speedup_p): Update.
10148 (edge_badness): Update.
10149 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10150 (size_time_entry): Replace predicate by exec_predicate and
10151 nonconst_predicate.
10152 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10153 (estimate_edge_time): Return also nonspec_time.
10154 (reset_edge_growth_cache): Update.
10156 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10158 PR rtl-optimization/80491
10159 * ifcvt.c (noce_process_if_block): When looking for x setter
10160 with missing else_bb, don't check only the insn right before
10161 cond_earliest, but look for the last insn that x is modified in
10162 within the same bb.
10164 PR rtl-optimization/80491
10165 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10167 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10169 PR tree-optimization/80487
10170 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10172 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10174 PR tree-optimization/79697
10175 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10176 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10177 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10179 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10180 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10182 2017-04-28 Martin Sebor <msebor@redhat.com>
10184 PR tree-optimization/80523
10185 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10186 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10188 (maybe_warn, format_directive, parse_directive): Use new functions.
10189 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10191 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10193 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10195 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10197 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10198 target_header_dir): Set correctly.
10199 * configure: Regenerated.
10200 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10201 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10202 instead of SYSTEM_HEADER_DIR.
10204 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10206 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10207 (estimate_local_effects): Likewise.
10208 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10209 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10210 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10211 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10212 do_estimate_edge_time, estimate_edge_time): Likewise.
10213 * ipa-inline-analysis.c (estimate_node_size_and_time,
10214 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10215 (estimate_time_after_inlining): Remove.
10217 2017-04-28 Martin Liska <mliska@suse.cz>
10219 * doc/gcov.texi: Enhance documentation of gcov.
10221 2017-04-28 Martin Liska <mliska@suse.cz>
10223 * doc/gcov.texi: Sort options in alphabetic order.
10224 * doc/gcov-dump.texi: Likewise.
10225 * doc/gcov-tool.texi: Likewise.
10226 * gcov.c (print_usage): Likewise.
10227 * gcov-dump.c (print_usage): Likewise.
10228 * gcov-tool.c (print_merge_usage_message): Likewise.
10229 (print_rewrite_usage_message): Likewise.
10230 (print_overlap_usage_message): Likewise.
10232 2017-04-28 Martin Liska <mliska@suse.cz>
10234 PR gcov-profile/53915
10235 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10237 2017-04-28 Martin Liska <mliska@suse.cz>
10239 PR gcov-profile/79891
10240 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10241 is marked by compiler as living on a line.
10242 (get_cycles_count): Remove usage of the union.
10243 (output_intermediate_file): Likewise.
10244 (find_source): Fix GNU coding style.
10245 (accumulate_line_counts): Remove old non-all block mode.
10246 (output_lines): Remove usage of the union.
10247 * profile.c (output_location): Include all BBs, even if
10248 belonging to a same line (and file) as a previous BB.
10250 2017-04-28 Martin Liska <mliska@suse.cz>
10252 * gcov.c (process_args): Handle new argument 'w'.
10253 (read_graph_file): Assign ID to BBs.
10254 (output_branch_count): Display BB # if verbose flag is set.
10255 (output_lines): Likewise for arcs.
10256 (print_usage): Add '--verbose' option help.
10257 * doc/gcov.texi: Document --verbose (-w) option.
10259 2017-04-28 Martin Liska <mliska@suse.cz>
10261 * gcov.c (struct block_location_info): New struct.
10262 (process_file): Fill up the new structure.
10263 (read_graph_file): Replace usage of encoding by the newly added
10265 (add_line_counts): Likewise.
10266 (accumulate_line_counts): Remove usage of the union.
10267 (function_info::function_info): New function.
10268 (function_info::~function_info): Likewise.
10269 (process_file): Call delete instead of release_function.
10270 (release_function): Release the function.
10271 (release_structures): Call delete instead of release_function.
10272 (solve_flow_graph): Replace usage of num_blocks.
10273 (find_exception_blocks): Likewise.
10274 (output_lines): Fix GNU coding style.
10276 2017-04-28 Martin Liska <mliska@suse.cz>
10279 * coverage.c (coverage_remove_note_file): New function.
10280 * coverage.h: Declare the function.
10281 * toplev.c (finalize): Clean if an error has been seen.
10283 2017-04-28 Martin Liska <mliska@suse.cz>
10285 PR gcov-profile/80031
10286 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10287 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10288 * gcov.c (read_graph_file): Read just number of blocks.
10289 * profile.c (branch_prob): Do not stream 0 flags per a basic
10292 2017-04-28 Martin Liska <mliska@suse.cz>
10294 * gcov-dump.c (tag_*): Add new argument to declarations.
10295 (dump_gcov_file): Likewise.
10296 (tag_blocks): Add and use new argument depth.
10297 (tag_arcs): Likewise.
10298 (tag_lines): Likewise.
10299 (tag_counters): Likewise.
10300 (tag_summary): Likewise.
10301 (dump_working_sets): Use depth to do a proper indentation.
10303 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10306 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10307 * symtab.c (symtab_node::debug_symtab): Move definition here.
10309 2017-04-28 Richard Biener <rguenther@suse.de>
10311 * lto-streamer.h (LTO_major_version): Bump to 7.
10313 2017-04-28 Richard Biener <rguenther@suse.de>
10315 * tree-vrp.c (assert_info): New struct.
10316 (add_assert_info): New helper.
10317 (register_edge_assert_for_2): Refactor to add asserts to a vector
10319 (register_edge_assert_for_1): Likewise.
10320 (register_edge_assert_for): Likewise.
10321 (finish_register_edge_assert_for): New helper actually registering
10322 asserts where live on edge.
10323 (find_conditional_asserts): Adjust.
10324 (find_switch_asserts): Likewise.
10325 (evrp_dom_walker::try_find_new_range): Generalize.
10326 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10328 2017-04-27 Marek Polacek <polacek@redhat.com>
10331 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10332 arg10 and arg11 to itype.
10334 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10336 * doc/extend.texi (Object Size Checking): Improve grammar.
10338 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10341 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10342 that the logic for permitting reciprocal estimates matches that
10345 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10348 * tree.c (type_cache_hasher::equal): Only compare
10349 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10350 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10351 non-aggregate element types.
10352 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10353 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10355 2017-04-27 Richard Biener <rguenther@suse.de>
10357 PR middle-end/80533
10358 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10359 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10360 keeping a reference to a trailing array.
10362 2017-04-27 Richard Biener <rguenther@suse.de>
10364 PR middle-end/80539
10365 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10366 being in loop-closed SSA form conservatively.
10367 (chrec_fold_multiply_poly_poly): Likewise.
10369 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10371 PR middle-end/79665
10372 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10373 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10375 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10378 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10379 (aarch64_function_arg_alignment): Return unsigned int again, but still
10380 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10381 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10382 Don't emit -Wpsabi note.
10383 (aarch64_function_arg_boundary): Likewise.
10384 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10387 2017-04-26 Nathan Sidwell <nathan@acm.org>
10389 * tree.h (crc32_unsigned_n): Declare.
10390 (crc32_unsigned, crc32_unsigned): Make inline.
10391 * tree.c (crc32_unsigned_bits): Replace with ...
10392 (crc32_unsigned_n): ... this.
10393 (crc32_unsigned, crc32_byte): Remove.
10394 (crc32_string): Remove unnecessary braces.
10396 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10398 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10399 * ipa-inline-analysis.c (MAX_TIME): Remove.
10400 (account_size_time): Use sreal for time.
10401 (dump_inline_summary): Update.
10402 (estimate_function_body_sizes): Update.
10403 (estimate_edge_size_and_time): Update.
10404 (estimate_calls_size_and_time): Update.
10405 (estimate_node_size_and_time): Update.
10406 (inline_merge_summary): Update.
10407 (inline_update_overall_summary): Update.
10408 (estimate_time_after_inlining): Update.
10409 (inline_read_section): Update.
10410 (inline_write_summary): Update.
10411 * ipa-inline.c (compute_uninlined_call_time): Update.
10412 (compute_inlined_call_time): Update.
10413 (recursive_inlining): Update.
10414 (inline_small_functions): Update.
10415 (dump_overall_stats): Update.
10416 * ipa-inline.h: Include sreal.h.
10417 (size_time_entry): Turn time to sreal.
10418 (inline_summary): Turn self_time nad time to sreal.
10420 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10422 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10424 (sreal::stream_out, sreal::stream_in): New.
10425 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10427 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10429 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10432 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10435 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10437 (dimode_scalar_chain::compute_convert_gain): Ditto.
10438 (dimode_scalar_chain::make_vector_copies): Ditto.
10439 (dimode_scalar_chain::convert_reg): Ditto.
10440 (dimode_scalar_chain::convert_insn): Ditto.
10441 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10442 (VI248_AVX512BW_1): New mode iterator.
10443 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10444 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10447 2017-04-25 Martin Sebor <msebor@redhat.com>
10449 PR tree-optimization/80497
10450 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10451 constants are representable in HOST_WIDE_INT.
10452 (parse_directive): Ditto.
10454 2017-04-25 Martin Sebor <msebor@redhat.com>
10457 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10458 (new_zero_array): Adjust signature.
10459 (dom_info::dom_init): Used unsigned rather that size_t.
10460 (dom_info::dom_info): Same.
10462 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10463 Jakub Jelinek <jakub@redhat.com>
10466 * config/arm/arm.c: Include gimple.h.
10467 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10468 returns negative, increment ncrn only if it returned positive.
10469 (arm_needs_doubleword_align): Return int instead of bool,
10470 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10471 members, but if there is any such non-FIELD_DECL
10472 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10473 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10474 returns negative, increment nregs only if it returned positive.
10475 (arm_setup_incoming_varargs): Likewise.
10476 (arm_function_arg_boundary): Emit -Wpsabi note if
10477 arm_needs_doubleword_align returns negative, return
10478 DOUBLEWORD_ALIGNMENT only if it returned positive.
10480 2017-04-25 Marek Polacek <polacek@redhat.com>
10483 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10484 first argument to type.
10486 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10489 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10490 type checks to test for compatibility instead of equality.
10492 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10493 Jakub Jelinek <jakub@redhat.com>
10496 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10498 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10499 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10500 the alignment computation, but return their maximum in warn_alignment.
10501 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10502 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10504 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10505 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10508 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10510 * config/arc/simdext.md (dmpyh): Fix typo.
10512 2017-04-25 Richard Biener <rguenther@suse.de>
10514 PR tree-optimization/80492
10515 * alias.c (compare_base_decls): Handle registers with asm
10516 specification conservatively.
10517 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10518 compare_base_decls returning dont-know properly.
10520 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10522 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10523 (legitimate_offset_address_p): New function.
10524 (arc_legitimate_address_p): Use above function.
10526 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10528 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10530 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10532 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10533 ACCH registers whenever they are available.
10535 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10537 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10538 double regs fix when not used.
10540 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10542 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10544 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10545 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10547 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10549 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10550 when emitting short ADD instructions.
10552 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10554 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10556 (cmpsi_cc_c_insn): Likewise.
10557 (cbranchsi4_scratch): Compute proper instruction length using
10558 compact_hreg_operand.
10559 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10561 2017-04-25 Richard Biener <rguenther@suse.de>
10563 PR middle-end/80509
10564 * passes.c (pass_manager::pass_manager): Initialize
10565 m_name_to_pass_map.
10567 2017-04-25 Richard Biener <rguenther@suse.de>
10569 PR tree-optimization/79201
10570 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10572 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10575 * config/s390/vector.md: Split MEM->GPR vector moves for
10576 non-s_operand addresses.
10578 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10581 * config/s390/predicates.md (reload_const_wide_int_operand): New
10583 * config/s390/s390.md ("movti"): Remove d/P alternative.
10584 ("movti_bigconst"): New pattern definition.
10586 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10589 * s390-protos.h (s390_expand_cs_hqi): Removed.
10590 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10591 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10592 modes as well as CCZ1mode and CCZmode.
10593 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10594 signature of s390_emit_compare_and_swap.
10595 (s390_expand_cs_hqi): Likewise, make static.
10596 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10597 compare-and-swap, in some cases.
10598 (s390_expand_cs): Wrapper function.
10599 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10601 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10602 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10603 patterns for small and large integers. Forbid symref memory operands.
10604 Move expander to s390.c. Require cc register.
10605 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10606 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10607 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10608 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10609 symref memory operands. Remove CC mode and call s390_match_ccmode
10611 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10613 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10615 * config/s390/s390.md (define_peephole2): New peephole to help
10616 combining the load-and-test pattern with volatile memory.
10618 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10620 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10621 with CCZmode for TARGET_Z196.
10623 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10625 PR rtl-optimization/80501
10626 * combine.c (make_compound_operation_int): Set subreg_code to SET
10627 even for AND with mask of the sign bit of mode.
10629 PR rtl-optimization/80500
10630 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10631 sum's initial value.
10633 2017-04-25 Julian Brown <julian@codesourcery.com>
10634 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10636 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10638 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10640 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10642 2017-04-25 Julian Brown <julian@codesourcery.com>
10643 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10645 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10646 (thunderx2t99_sha): New Reservation.
10648 2017-04-25 Julian Brown <julian@codesourcery.com>
10649 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10651 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10652 type for 1-element load.
10654 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10656 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10658 2017-04-24 Martin Jambor <mjambor@suse.cz>
10660 PR tree-optimization/80293
10661 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10662 char arrays not totally scalarizable if it is false.
10663 (analyze_all_variable_accesses): Pass correct value in the new
10664 parameter. Add a statistics counter.
10666 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10668 PR middle-end/79931
10669 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10671 2017-04-24 Richard Biener <rguenther@suse.de>
10673 PR tree-optimization/80494
10674 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10675 out for complex types.
10677 2017-04-24 Richard Biener <rguenther@suse.de>
10679 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10680 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10681 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10682 (DFS): Adjust and never fail.
10683 (sccvn_dom_walker::fail): Remove.
10684 (sccvn_dom_walker::before_dom_children): Adjust.
10685 (run_scc_vn): Likewise and never fail.
10686 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10687 (pass_fre::execute): Likewise.
10689 2017-04-24 Richard Biener <rguenther@suse.de>
10691 PR tree-optimization/79725
10692 * tree-ssa-sink.c (statement_sink_location): Return whether
10693 failure reason was zero uses. Move that check later.
10694 (sink_code_in_bb): Deal with zero uses by removing the stmt
10697 2017-04-24 Richard Biener <rguenther@suse.de>
10700 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10701 pointer-based references.
10703 2017-04-24 Richard Biener <rguenther@suse.de>
10706 * pass_manager.h (pass_manager::operator new): Remove.
10707 (pass_manager::operator delete): Likewise.
10708 * passes.c (pass_manager::operator new): Remove.
10709 (pass_manager::operator delete): Likewise.
10710 (pass_manager::pass_manager): Zero individual pass members.
10712 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10715 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10716 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10717 Check "XEXP (src, 1)" operand here.
10718 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10719 Check "XEXP (src, 1)" operand here.
10720 (dimode_scalar_chain::make_vector_copies): Detect count register
10721 of a shift instruction. Zero extend count register from QImode
10722 to DImode to satisfy vector shift pattern count operand predicate.
10723 Substitute vector shift count operand with a DImode copy.
10724 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10727 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10729 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10730 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10731 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10732 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10733 (UNSPEC_NOREX_MEM): Remove definition.
10735 2017-04-21 Richard Biener <rguenther@suse.de>
10737 PR tree-optimization/79547
10738 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10739 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10740 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10741 without any constraints.
10743 2017-04-21 Richard Biener <rguenther@suse.de>
10745 PR tree-optimization/78847
10746 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10748 2017-04-21 Richard Biener <rguenther@suse.de>
10750 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10751 (build_distinct_type_copy): Likewise.
10752 (build_variant_type_copy): Likewise.
10753 * tree.c (build_qualified_type): Pass down mem-stat info.
10754 (build_distinct_type_copy): Likewise.
10755 (build_variant_type_copy): Likewise.
10757 2017-04-21 Richard Biener <rguenther@suse.de>
10759 PR tree-optimization/80237
10760 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10762 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10763 for a simplified result.
10765 2016-04-21 Richard Biener <rguenther@suse.de>
10767 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10768 sth as strict as a simple_iv but a chrec without symbols and an
10769 operand defined in the loop we are peeling (and not some subloop).
10770 (propagate_constants_for_unrolling): Propagate all constants.
10772 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10775 * config/i386/i386.c (print_reg): Remove assert for disalowed
10776 regno values, call output_operand_lossage instead.
10778 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10781 * config/i386/constraints.md (Yc): New register constraint.
10782 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10783 Use Yc constraint for alternative 2 of operand 0. Remove
10784 preferred_for_speed attribute.
10786 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10788 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10789 lastprivate clauses in SIMT case.
10791 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10793 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10795 2017-04-20 Richard Biener <rguenther@suse.de>
10797 PR tree-optimization/57796
10798 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10799 as N scalar stores.
10800 (vect_model_load_cost): Cost gathers as N scalar loads.
10802 2017-04-20 Richard Biener <rguenther@suse.de>
10804 * ggc-page.c (ggc_allocated_p): Rename to ...
10805 (safe_lookup_page_table_entry): ... this and return the lookup
10807 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10809 2017-04-20 Richard Biener <rguenther@suse.de>
10811 PR tree-optimization/80453
10812 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
10813 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
10814 from the conditions.
10815 (vn_phi_eq): Pass them down.
10816 (vn_phi_lookup): Record them.
10817 (vn_phi_insert): Likewise.
10819 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
10821 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
10822 uninitialized variable warning to avoid buffer overrun.
10824 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10827 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
10828 is suppressed for '{ 0 }' in C.
10830 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10832 * BASE-VER: Set to 8.0.0.
10834 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
10836 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
10837 priority .init_array and .fini_array section with SECTION_NOTYPE
10840 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10842 PR middle-end/80423
10843 * tree.h (build_array_type): Add typeless_storage default argument.
10844 * tree.c (type_cache_hasher::equal): Also compare
10845 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
10846 (build_array_type): Add typeless_storage argument, set
10847 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
10849 (build_nonshared_array_type): Adjust build_array_type_1 caller.
10850 (build_array_type): Likewise. Add typeless_storage argument.
10852 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10853 Jakub Jelinek <jakub@redhat.com>
10855 PR tree-optimization/80426
10856 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
10857 operation on symbolic operands, also compute the overflow for the
10858 invariant part when the operation degenerates into a negation.
10860 2017-04-19 Jakub Jelinek <jakub@redhat.com>
10863 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
10864 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
10867 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
10869 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
10872 * config/avr/avr.c (tree.h): Include it.
10873 (cgraph.h): Include it.
10874 (avr_encode_section_info): Don't warn for uninitialized progmem
10875 variable if it's just an alias.
10877 2017-04-19 Richard Biener <rguenther@suse.de>
10880 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
10881 when needed by AutoPGO.
10883 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
10886 * doc/lto.texi: Remove an extra 'that'.
10888 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
10890 PR rtl-optimization/80429
10891 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
10892 are only used in debug insns.
10894 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10895 Vladimir Makarov <vmakarov@redhat.com>
10897 * config/sparc/predicates.md (input_operand): Add comment. Return
10898 true for any memory operand when LRA is in progress.
10899 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
10901 2017-04-18 Jeff Law <law@redhat.com>
10904 * mips.md ({return,simple_return}_internal): Do not overwrite
10907 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10909 PR tree-optimization/80443
10910 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
10911 instead of adding 1, subtract -1 and similarly instead of subtracting
10914 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
10916 PR rtl-optimization/80357
10917 * haifa-sched.c (tmp_bitmap): New variable.
10918 (model_recompute): Handle duplicate use records.
10919 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
10920 (free_global_sched_pressure_data): Free it.
10922 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
10925 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10926 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10927 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10928 instead of SYSTEM_HEADER_DIR.
10930 2017-04-18 Jeff Law <law@redhat.com>
10932 PR middle-end/80422
10933 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
10934 predecessors after walking up the insn chain.
10936 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10939 * dwarf2out.c (modified_type_die): Try harder not to emit internal
10940 sizetype type into debug info.
10942 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
10945 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
10946 unneeded test for TARGET_UPPER_REGS_SF.
10947 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
10949 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10952 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
10953 instead of gsi_after_labels.
10955 2017-04-18 Jeff Law <law@redhat.com>
10957 * regcprop.c (maybe_mode_change): Avoid creating copies of the
10961 2017-04-13 Jeff Law <law@redhat.com>
10962 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10963 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10965 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
10968 * config/avr/avr.c (intl.h): Include it.
10969 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
10971 2017-04-18 Martin Liska <mliska@suse.cz>
10973 PR gcov-profile/78783
10974 * gcov-tool.c (gcov_output_files): Validate that destination
10975 file is either removed by the tool or by a user.
10977 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10978 Guy Benyei <guybe@mellanox.com>
10980 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
10981 block, and do not negate it, the stored id is already negative.
10983 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10985 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
10987 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10990 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
10991 masks of options that should be turned off if the VSX vector
10992 options are turned off.
10993 (OTHER_P8_VECTOR_MASKS): Likewise.
10994 (OTHER_VSX_VECTOR_MASKS): Likewise.
10995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
10996 rs6000_disable_incompatible_switches to validate no type switches
10998 (rs6000_incompatible_switch): New function to disallow turning on
10999 other vector options if -mno-vsx, -mno-power8-vector, or
11000 -mno-power9-vector are specified.
11002 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11004 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11006 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11008 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11009 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11010 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11011 (ARG_POINTER_CFA_OFFSET): Likewise.
11013 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11015 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11016 conditions to take advantage of various optimizations.
11018 2017-04-13 Jeff Law <law@redhat.com>
11020 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11021 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11022 (zero_extendsidi2_dext): Likewise.
11024 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11027 * fold-const.c (fold_ternary_loc): Revert
11028 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11031 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11033 PR rtl-optimization/80343
11034 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11035 new scratch pseudo.
11037 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11040 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11041 to ubsan_encode_value.
11043 2017-04-13 Jeff Law <law@redhat.com>
11045 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11046 appearing in DEBUG_INSNs.
11048 2017-04-13 Martin Liska <mliska@suse.cz>
11050 PR gcov-profile/80413
11051 * gcov-io.c (gcov_write_string): Copy to buffer just when
11052 allocated size is greater than zero.
11054 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11057 * dwarf2out.c (decls_for_scope): Ignore declarations of
11058 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11060 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11063 * ipa-visibility.c (non_local_p): Fix typos.
11064 (localize_node): When localizing symbol in same comdat group,
11065 dissolve the group only when we know external symbols are going
11067 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11069 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11071 PR tree-optimization/79390
11072 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11073 order does not result in usable sequence, retry with reversed operand
11079 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11080 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11081 op0 instead of fold_convert_loc (loc, type, arg0).
11083 2017-04-12 Jeff Law <law@redhat.com>
11085 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11086 has a delay slot in the generated code.
11088 * config/cris/cris.md (cris_preferred_reload_class): Return
11089 GENNONACR_REGS rather than GENERAL_REGS.
11091 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11094 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11095 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11096 signedness of the result type.
11098 2017-04-12 Richard Biener <rguenther@suse.de>
11099 Jeff Law <law@redhat.com>
11101 PR tree-optimization/80359
11102 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11103 trim stores to TARGET_MEM_REFs.
11105 2017-04-12 Richard Biener <rguenther@suse.de>
11107 PR tree-optimization/79390
11108 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11109 threading case even more.
11111 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11114 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11115 for quad_address_p for TImode, instead of just not indexed_address.
11117 2017-04-12 Richard Biener <rguenther@suse.de>
11118 Bernd Edlinger <bernd.edlinger@hotmail.de>
11120 PR middle-end/79671
11121 * alias.c (component_uses_parent_alias_set_from): Handle
11122 TYPE_TYPELESS_STORAGE.
11123 (get_alias_set): Likewise.
11124 * tree-core.h (tree_type_common): Add typeless_storage flag.
11125 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11126 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11127 for types containing members with TYPE_TYPELESS_STORAGE.
11128 (place_field): Likewise.
11129 (layout_type): Likewise for ARRAY_TYPE.
11130 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11131 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11132 TYPE_TYPELESS_STORAGE.
11133 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11135 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11138 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11139 first argument to type.
11141 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11145 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11146 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11147 (rs6000_expand_binop_builtin): Likewise.
11148 (rs6000_expand_ternop_builtin): Likewise; also add missing
11149 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11150 vshasigma built-ins.
11151 * doc/extend.texi: Document that vec_xxpermdi's third argument
11152 must be a constant.
11154 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11156 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11157 Use shift_const cost parameter when calculating gain of STV shifts.
11159 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11161 PR rtl-optimization/70478
11162 * lra-constraints.c (process_alt_operands): Check memory for
11163 disfavoring memory insn operand.
11165 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11167 PR middle-end/80100
11168 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11169 left shift in unsigned HOST_WIDE_INT type.
11171 PR rtl-optimization/80385
11172 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11173 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11176 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11177 if they have any depend clauses.
11179 2017-04-11 Martin Liska <mliska@suse.cz>
11182 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11183 * ipa-split.c (split_function): Create a local comdat symbol
11184 if caller is in a comdat group.
11186 2017-04-11 Martin Liska <mliska@suse.cz>
11189 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11192 2017-04-11 Martin Sebor <msebor@redhat.com>
11194 PR middle-end/80364
11195 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11196 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11198 (directive::set_width, directive::set_precision, format_character):
11200 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11203 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11206 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11207 conflict, set target->arch_name instead of target->cpu_name.
11209 2017-04-11 Richard Biener <rguenther@suse.de>
11211 PR tree-optimization/80374
11212 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11213 build_zero_cst, remove fold_convertible_p check again.
11215 2017-04-11 Martin Liska <mliska@suse.cz>
11218 * ubsan.c (instrument_object_size): Do not instrument register
11221 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11224 * config/i386/i386-builtin-types.def
11225 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11226 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11227 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11228 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11229 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11230 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11231 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11232 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11233 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11234 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11235 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11236 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11237 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11238 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11239 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11240 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11241 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11242 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11243 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11244 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11245 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11246 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11247 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11248 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11249 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11250 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11251 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11252 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11253 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11254 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11255 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11256 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11257 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11258 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11259 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11260 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11261 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11262 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11263 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11264 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11265 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11266 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11267 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11268 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11269 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11270 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11272 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11273 flag to second_arg_count, handle 4 argument function type _COUNT
11274 aliases, handle second_arg_count on second argument rather than last.
11276 2017-04-10 Jeff Law <law@redhat.com>
11278 PR tree-optimization/80374
11279 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11280 record anything if we can not convert integer_zero_node to the
11283 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11286 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11287 Enhance special handling given to the TARGET_P9_MINMAX option in
11288 relation to certain other options.
11290 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11292 PR tree-optimization/80153
11293 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11294 remove POINTER_PLUS_EXPR's base part directly, rather than through
11297 2017-04-10 Richard Biener <rguenther@suse.de>
11298 Bin Cheng <bin.cheng@arm.com>
11300 PR tree-optimization/80153
11301 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11302 the first element of pointer type aff_tree. Build result expr in
11304 (add_elt_to_tree): Convert to type unconditionally. Remove other
11305 fold_convert calls.
11306 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11307 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11309 2017-04-10 Richard Biener <rguenther@suse.de>
11311 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11314 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11316 PR rtl-optimization/70478
11317 * lra-constraints.c (curr_small_class_check): New.
11318 (update_and_check_small_class_inputs): New.
11319 (process_alt_operands): Update curr_small_class_check. Disfavor
11320 alternative insn memory operands. Check available regs for small
11323 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11326 * config/mips/mips.opt (-mvirt): Update description.
11327 * doc/invoke.texi (-mvirt): Likewise.
11329 2017-04-10 Richard Biener <rguenther@suse.de>
11331 PR middle-end/80362
11332 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11333 looking for NEGATE_EXPR in -A / -B to A / B folding.
11335 2017-04-10 Martin Liska <mliska@suse.cz>
11337 PR gcov-profile/80224
11338 * gcov.c (print_usage): Fix usage string.
11339 (get_gcov_intermediate_filename): Remove.
11340 (output_gcov_file): Use both for normal and intermediate format.
11341 (generate_results): Do not initialize special file for
11342 intermediate format.
11344 2017-04-10 Richard Biener <rguenther@suse.de>
11346 PR tree-optimization/80304
11347 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11350 2017-04-10 Nathan Sidwell <nathan@acm.org>
11353 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11354 (rs6000_init_builtins): Use it.
11356 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11358 * config/arm/arm.md (<mrc>): Add mode to SET source.
11359 (<mrrc>): Likewise.
11361 2017-04-10 Richard Biener <rguenther@suse.de>
11363 PR middle-end/80344
11364 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11366 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11369 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11370 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11371 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11372 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11373 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11374 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11375 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11376 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11377 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11378 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11379 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11380 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11381 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11382 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11383 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11384 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11385 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11386 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11387 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11388 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11389 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11390 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11391 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11393 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11395 PR rtl-optimization/70478
11396 * lra-constraints.c: Reverse the last patch.
11398 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11400 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11401 Add comment for WCHAR_T.
11403 2017-04-08 Martin Liska <mliska@suse.cz>
11406 2017-04-07 Martin Liska <mliska@suse.cz>
11409 * ipa-split.c (split_function): Add function part to a same comdat
11412 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11415 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11417 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11419 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11420 (vec_pairing): ...to this.
11421 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11422 (rs6000_sched_init): Adjust for name change.
11423 (struct rs6000_sched_context): Likewise.
11424 (rs6000_init_sched_context): Likewise.
11425 (rs6000_set_sched_context): Likewise.
11427 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11433 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11435 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11436 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11437 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11439 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11441 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11443 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11445 PR rtl-optimization/70703
11446 * ira-color.c (update_conflict_hard_regno_costs): Use
11447 int64_t instead of HOST_WIDE_INT.
11449 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11451 PR rtl-optimization/70478
11452 * lra-constraints.c (process_alt_operands): Disfavor alternative
11453 insn memory operands.
11455 2017-04-07 Jeff Law <law@redhat.com>
11457 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11458 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11460 2017-04-07 Martin Liska <mliska@suse.cz>
11463 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11464 Show error message instead of an ICE.
11466 2017-04-07 Martin Liska <mliska@suse.cz>
11469 * ipa-split.c (split_function): Add function part to a same comdat
11472 2017-04-07 Richard Biener <rguenther@suse.de>
11474 PR middle-end/80341
11475 * tree.c (get_unwidened): Also handle ! for_type case for
11477 * convert.c (do_narrow): Split out from ...
11478 (convert_to_integer_1): ... here. Do not pass final truncation
11479 type to get_unwidened for TRUNC_DIV_EXPR.
11481 2017-04-07 Richard Biener <rguenther@suse.de>
11483 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11485 (aff_combination_const): Adjust.
11486 (aff_combination_scale): Likewise.
11487 (aff_combination_add_elt): Likewise.
11488 (aff_combination_add_cst): Likewise.
11489 (aff_combination_convert): Likewise.
11490 (add_elt_to_tree): Likewise. Remove unused argument.
11491 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11493 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11495 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11497 * config/arm/arm.c (arm_default_short_enums): Use
11498 ARM_DEFAULT_SHORT_ENUMS.
11499 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11501 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11504 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11505 members with redundant out-of-class redeclaration.
11507 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11510 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11511 * config/i386/i386.md (*zero_extendsidi2):
11512 Add (?*x,*x) and (?*v,*v) alternatives.
11514 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11517 * config/i386/i386.c (ix86_expand_builtin)
11518 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11519 mode from insn data. Convert operands to insn operand mode.
11520 Copy operands that don't satisfy insn predicate to a register.
11522 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11524 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11527 2017-04-06 Richard Biener <rguenther@suse.de>
11529 PR tree-optimization/80334
11530 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11531 preserve alignment of accesses.
11533 2017-04-06 Richard Biener <rguenther@suse.de>
11535 PR tree-optimization/80262
11536 * tree-sra.c (build_ref_for_offset): Preserve address-space
11538 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11539 Drop useless address-space information on MEM_REF offsets.
11541 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11543 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11545 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11547 PR rtl-optimization/70703
11548 * ira-color.c (update_conflict_hard_regno_costs): Use
11549 HOST_WIDE_INT instead of long.
11551 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11554 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11555 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11557 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11558 for x86_64 target. Handle -m3dnowa option.
11560 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11562 PR rtl-optimization/70703
11563 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11564 (update_conflict_hard_regno_costs): Use long instead of unsigned
11565 arithmetic for cost calculation.
11567 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11568 Bernd Edlinger <bernd.edlinger@hotmail.de>
11571 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11574 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11577 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11578 ptr_mode with Pmode throughout.
11579 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11580 into probe_stack_range and use DImode.
11582 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11585 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11586 call_eh_return is true.
11588 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11590 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11591 Initialize last_match_fntype_index.
11593 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11596 * tree-nvr.c: Include internal-fn.h.
11597 (pass_return_slot::execute): Ignore internal calls without
11600 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11601 Richard Biener <rguenther@suse.de>
11604 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11605 captures used multiple times, except for the last use.
11606 * generic-match-head.c: Include gimplify.h.
11608 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11610 PR tree-optimization/79390
11611 * target.h (struct noce_if_info): Declare.
11612 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11613 * target.def (noce_conversion_profitable_p): New target hook.
11614 * ifcvt.h (struct noce_if_info): New type, moved from ...
11615 * ifcvt.c (struct noce_if_info): ... here.
11616 (noce_conversion_profitable_p): Renamed to ...
11617 (default_noce_conversion_profitable_p): ... this. No longer
11619 (noce_try_store_flag_constants, noce_try_addcc,
11620 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11621 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11622 instead of noce_conversion_profitable_p.
11623 * config/i386/i386.c: Include ifcvt.h.
11624 (ix86_option_override_internal): Don't override
11625 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11626 (ix86_noce_conversion_profitable_p): New function.
11627 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11628 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11629 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11630 * doc/tm.texi: Regenerated.
11632 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11634 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11637 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11640 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11641 instructions for small multiply cores.
11643 2017-04-04 Jeff Law <law@redhat.com>
11645 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11647 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11648 that are not set by the loop over the elements.
11650 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11653 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11654 int mode, convert_modes it to mode as unsigned, otherwise use
11655 lowpart_subreg to mode rather than SImode.
11656 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11657 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11658 Use DImode instead of SImode for the shift count operand.
11659 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11662 2017-04-04 Richard Biener <rguenther@suse.de>
11664 PR middle-end/80281
11665 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11666 arithmetic done for the negate or the plus. Simplify.
11667 (A - (-B) -> A + B): Likewise.
11668 * fold-const.c (split_tree): Make sure to not negate pointers.
11670 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11672 PR rtl-optimization/60818
11673 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11674 a compare of comparisons with the thing compared if this results
11675 in a different machine mode.
11677 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11679 * alias.c (base_alias_check): Fix typo in comment.
11680 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11681 * cgraphunit.c (symbol_table::compile): Likewise.
11682 * collect2.c (maybe_run_lto_and_relink): Likewise.
11683 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11684 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11685 * config/avr/avr.c (avr_map_op_t): Likewise.
11686 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11687 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11688 * config/epiphany/epiphany.md (movcc): Likewise.
11689 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11690 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11692 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11693 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11694 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11695 * config/sh/sh.c (sh_rtx_costs): Likewise.
11696 * fold-const.c (fold_truth_andor): Likewise.
11697 * genautomata.c (collapse_flag): Likewise.
11698 * gengtype.h (struct type::u::s): Likewise.
11699 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11700 * input.c (FORMAT_AMOUNT): Likewise.
11701 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11702 (known_aggs_to_agg_replacement_list): Likewise.
11703 * ipa-inline-analysis.c: Likewise.
11704 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11705 * ipa-polymorphic-call.c
11706 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11707 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11708 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11710 * modulo-sched.c (apply_reg_moves): Likewise.
11711 * omp-expand.c (build_omp_regions_1): Likewise.
11712 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11713 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11714 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11715 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11716 * value-prof.c: Likewise.
11717 * var-tracking.c (val_reset): Likewise.
11719 2017-04-03 Richard Biener <rguenther@suse.de>
11721 PR tree-optimization/80275
11722 * fold-const.c (split_address_to_core_and_offset): Handle
11725 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11727 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11728 descriptors is at least equal to that of functions.
11730 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11732 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11734 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11737 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11738 (mov<IMOD4:mode>): New expander.
11739 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11741 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11743 PR rtl-optimization/79405
11744 * fwprop.c (propagations_left): New variable.
11745 (forward_propagate_into): Decrement it.
11746 (fwprop_init): Initialize it.
11747 (fw_prop): If the variable has reached zero, stop propagating.
11748 (fwprop_addr): Ditto.
11750 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11753 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11754 a FUNCTION_DECL, pass it as decl instead of origin to
11757 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11759 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11762 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11765 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11766 TARGET_VSX_SMALL_INTEGER.
11768 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11770 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11771 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11773 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11775 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11776 extraction from odd-numbered MSA register.
11778 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11780 PR middle-end/80173
11781 * expmed.c (store_bit_field_1): Don't attempt to create
11782 a word subreg out of hard registers wider than word if they
11783 have HARD_REGNO_NREGS of 1 for their mode.
11785 PR middle-end/80163
11786 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11787 conversions to integer types wider than word and pointer.
11790 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11791 (rtx_equal_for_cselib_p): Pass 0 to it.
11792 * cselib.c (cselib_hasher::equal): Likewise.
11793 (rtx_equal_for_cselib_1): Add depth argument. If depth
11794 is 128, don't look up VALUE locs and punt. Increment
11795 depth in recursive calls when walking VALUE locs.
11797 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11799 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11800 (make_gcov_file_name): Use the canonical path name for generating
11802 (read_line): Fix handling of files with ascii null bytes.
11804 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11806 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11807 to initialise a vector register instead
11808 of using a const_int.
11810 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11812 PR translation/80189
11813 * gimplify.c (omp_default_clause): Use %qs instead of %s in
11814 diagnostic messages.
11816 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
11819 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
11820 (dfp_diex_<mode>): Update mode of operand 1.
11821 * doc/extend.texi (dxex, dxexq): Document change to return type.
11822 (diex, diexq): Document change to argument type.
11824 2017-03-30 Martin Jambor <mjambor@suse.cz>
11827 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
11828 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
11829 it reflects the signature changes performed at the callee side.
11830 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
11831 to cgraph_build_function_type_skip_args.
11832 (build_function_decl_skip_args): Adjust call to the above function.
11834 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11837 * config/i386/sse.md
11838 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
11839 register as dest whenever it is a MEM not rtx_equal_p to the
11840 corresponding dup operand, and when forcing into reg move the
11841 reg into the memory afterwards.
11842 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
11843 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
11844 for the force_reg mode.
11845 (avx512vl_vextractf128<mode>): Use register as dest either
11846 always when a MEM, or when it is a MEM not rtx_equal_p to the
11847 corresponding dup operand, or even not when it is a CONST_VECTOR
11848 depending on the mode and lo vs. hi.
11849 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
11851 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
11852 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
11853 Likewise. Require that operands[2] is even.
11854 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
11855 Remove extraneous parens. Require that operands[2] is a multiple
11857 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
11858 operands[0] is a MEM if <mask_applied>, the predicates/constraints
11859 disallow memory then.
11861 2017-03-30 Richard Biener <rguenther@suse.de>
11863 PR tree-optimization/77498
11864 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
11865 to non-constants over backedges.
11867 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
11869 PR rtl-optimization/80233
11870 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
11871 as last_combined_insn. Do not test for BARRIER_P separately.
11873 2017-03-29 Andreas Schwab <schwab@suse.de>
11876 * calls.c (prepare_call_address): Convert funexp to Pmode before
11877 copying to temp reg.
11879 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11881 PR tree-optimization/80158
11882 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
11883 Handle possible future case of more than one alternate
11885 (replace_rhs_if_not_dup): Likewise.
11886 (replace_one_candidate): Likewise.
11888 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
11890 PR rtl-optimization/80193
11891 * ira.c (ira): Do not check allocation for LRA.
11893 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
11895 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
11896 (nvptx_output_simt_exit): Declare.
11897 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
11898 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
11899 (init_softstack_frame): Move initialization of crtl->is_leaf to...
11900 (nvptx_declare_function_name): ...here. Emit declaration of local
11901 memory space buffer for omp_simt_enter insn.
11902 (nvptx_output_unisimt_switch): New.
11903 (nvptx_output_softstack_switch): New.
11904 (nvptx_output_simt_enter): New.
11905 (nvptx_output_simt_exit): New.
11906 * config/nvptx/nvptx.h (struct machine_function): New fields
11907 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
11908 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
11909 (UNSPECV_SIMT_EXIT): Ditto.
11910 (omp_simt_enter_insn): New insn.
11911 (omp_simt_enter): New expansion.
11912 (omp_simt_exit): New insn.
11913 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
11915 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
11916 (expand_GOMP_SIMT_ENTER_ALLOC): New.
11917 (expand_GOMP_SIMT_EXIT): New.
11918 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
11919 (GOMP_SIMT_ENTER_ALLOC): Ditto.
11920 (GOMP_SIMT_EXIT): Ditto.
11921 * target-insns.def (omp_simt_enter): New insn.
11922 (omp_simt_exit): Ditto.
11923 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
11925 (lower_rec_simd_input_clauses): Implement SIMT privatization.
11926 (lower_rec_input_clauses): Likewise.
11927 (lower_lastprivate_clauses): Handle SIMT privatization.
11929 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
11930 (ompdevlow_adjust_simt_enter): New.
11931 (find_simtpriv_var_op): New.
11932 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
11933 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
11935 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
11936 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
11937 (copy_decl_for_dup_finish): Ditto.
11939 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
11941 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
11944 * config/i386/i386.c (ix86_option_override_internal): Always
11945 allow -mpreferred-stack-boundary=3 for 64-bit targets.
11947 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11949 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
11951 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11953 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
11954 mark new edge's irreducible flag accordign to it.
11955 (vect_do_peeling): Check loop preheader edge's irreducible flag
11956 and pass it to function slpeel_add_loop_guard.
11958 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
11960 PR tree-optimization/80218
11961 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
11962 Update block frequencies and counts.
11964 2017-03-28 Richard Biener <rguenther@suse.de>
11966 PR tree-optimization/78644
11967 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
11968 of a simplification result we may not use it at all.
11970 2017-03-28 Richard Biener <rguenther@suse.de>
11973 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
11974 without arguments, generate default definition of a SSA name.
11976 2017-03-28 Richard Biener <rguenther@suse.de>
11978 PR middle-end/80222
11979 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
11980 TYPE_REF_CAN_ALIAS_ALL references.
11981 * fold-const.c (fold_indirect_ref_1): Likewise.
11983 2017-03-28 Martin Liska <mliska@suse.cz>
11986 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
11987 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
11989 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11990 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
11992 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
11993 (EXTRA_SPECS): Define.
11994 (SUBTARGET_EXTRA_SPECS): Likewise.
11995 (SUBTARGET_CPP_SPEC): Likewise.
11996 * config/arc/elf.h (EXTRA_SPECS): Renamed to
11997 SUBTARGET_EXTRA_SPECS.
11998 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12000 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12002 * config/arc/simdext.md (vst64_insn): Update pattern.
12003 (vld32wh_insn): Likewise.
12004 (vld32wl_insn): Likewise.
12005 (vld64_insn): Likewise.
12006 (vld32_insn): Likewise.
12008 2017-03-28 Marek Polacek <polacek@redhat.com>
12011 * fold-const.c (fold_comparison): Use protected_set_expr_location
12012 instead of SET_EXPR_LOCATION.
12014 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12016 * tree.c (add_expr): Avoid name lookup warning.
12018 2017-03-27 Jeff Law <law@redhat.com>
12020 PR tree-optimization/80216
12021 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12022 function name. Limit recursion depth.
12023 (record_temporary_equivalences): Corresponding changes.
12025 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12027 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12030 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12033 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12035 * cfgcleanup.c (reg_note_cfa_p): New array.
12036 (insns_have_identical_cfa_notes): New function.
12037 (old_insns_match_p): Don't cross-jump in between /f
12038 and non-/f instructions. If both i1 and i2 are frame related,
12039 verify all CFA notes, their order and content.
12041 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12044 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12045 HImode and SImode with zero extend to DImode to one insn.
12046 (bswap<mode>2_extenddi): Likewise.
12047 (bswapsi2_extenddi): Likewise.
12048 (bswaphi2_extendsi): Likewise.
12049 (bswaphi2): Combine bswap HImode and SImode into one insn.
12050 Separate memory insns from swapping register.
12051 (bswapsi2): Likewise.
12052 (bswap<mode>2): Likewise.
12053 (bswaphi2_internal): Delete, no longer used.
12054 (bswapsi2_internal): Likewise.
12055 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12056 store, and gpr<-gpr swap insns.
12057 (bswap<mode>2_store): Likewise.
12058 (bswaphi2_reg): Register only splitter, combine with the splitter.
12059 (bswaphi2 splitter): Likewise.
12060 (bswapsi2_reg): Likewise.
12061 (bswapsi2 splitter): Likewise.
12062 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12063 the insns into load, store, and register/register insns.
12064 (bswapdi2_ldbrx): Likewise.
12065 (bswapdi2_load): Likewise.
12066 (bswapdi2_store): Likewise.
12067 (bswapdi2_reg): Likewise.
12069 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12071 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12072 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12074 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12077 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12079 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12080 special handling for target option conflicts between dform
12081 options (-mpower9-dform, -mpower9-dform-vector,
12082 -mpower9-dform-scalar) and -mno-direct-move.
12084 2017-03-27 Richard Biener <rguenther@suse.de>
12086 PR tree-optimization/80181
12087 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12089 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12091 * config/arc/predicates.md (move_double_src_operand): Replace the
12092 call to move_double_src_operand with a call to address_operand.
12094 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12096 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12097 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12098 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12100 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12102 * config/arc/predicates.md (long_immediate_loadstore_operand):
12103 Consider scaled addresses cases.
12105 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12107 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12108 restored when in interrupt.
12109 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12110 doesn't have delay slot.
12112 2017-03-27 Richard Biener <rguenther@suse.de>
12115 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12116 inlined thunk clones.
12118 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12121 * asan.c (instrument_derefs): Copy over last operand from
12122 original COMPONENT_REF to the new COMPONENT_REF with
12123 DECL_BIT_FIELD_REPRESENTATIVE.
12124 * ubsan.c (instrument_object_size): Likewise.
12126 2017-03-27 Richard Biener <rguenther@suse.de>
12128 PR tree-optimization/80170
12129 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12130 sure DR/SCEV didnt fold in constants we do not see when looking
12131 at the reference base alignment.
12133 2017-03-27 Richard Biener <rguenther@suse.de>
12135 PR middle-end/80171
12136 * gimple-fold.c (fold_ctor_reference): Properly guard against
12137 NULL return value from canonicalize_constructor_val.
12139 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12142 * config/i386/i386.c (ix86_expand_builtin)
12143 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12144 flags reg setting and flags reg using instructions.
12145 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12146 clobbering instructions to zero extend op2.
12148 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12150 * doc/install.texi (Configuration) <--with-aix-soname>:
12151 Update link to AIX ld.
12153 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12155 PR rtl-optimization/80160
12156 PR rtl-optimization/80159
12157 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12158 reg_alternate_class into account.
12160 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12163 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12164 to consider in curr_insn_transform.
12166 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12168 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12169 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12170 and emit_mode_inner.
12172 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12174 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12175 argument to the overloaded builtin variants. Use the new flag to
12176 deprecate certain builtin variants.
12177 * config/s390/s390-builtin-types.def: Add new builtin types.
12178 * config/s390/s390-builtins.h: Support new flags field for
12179 overloaded builtins.
12180 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12181 (s390_macro_to_expand): Enable vector float data type.
12182 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12183 builtins by incrementing the __VEC__ version number.
12184 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12186 (s390_resolve_overloaded_builtin): Emit error messages depending
12187 on the builtin flags.
12188 * config/s390/s390.c (s390_expand_builtin): Support additional
12189 flags argument. Change error message to match the messages
12190 emitted in s390-c.c.
12191 * config/s390/s390.md: New UNSPEC_* constants.
12192 (op_type): Add new instruction types.
12193 * config/s390/vecintrin.h: Add new builtins and test data class
12195 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12196 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12197 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12198 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12199 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12200 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12202 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12203 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12204 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12205 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12207 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12208 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12209 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12210 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12212 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12213 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12214 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12215 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12216 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12217 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12218 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12220 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12221 ("vec_scatter_element<V_HW_4:mode>_DI")
12222 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12223 ("vec_fpint<mode>", "vflls")
12224 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12225 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12226 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12227 ("*vec_cmphe<mode>_cc"): ... these.
12229 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12230 mode constant instead of magic value.
12232 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12234 * config/s390/s390.c (s390_expand_vec_compare): Support other
12235 vector floating point modes than just V2DF.
12236 (s390_expand_vcond): Likewise.
12237 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12238 (s390_cannot_change_mode_class): Prevent mode changes between TF
12239 and V1TF in vector registers.
12240 * config/s390/s390.md (DF, SF): New mode attributes.
12241 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12242 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12243 SFmode support for VRs.
12244 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12246 (VFT, VF_HW): New mode iterators.
12247 (vw, sdx): New mode attributes.
12248 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12249 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12250 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12251 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12252 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12253 also the new vector floating point modes. Renaming to ...
12255 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12256 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12257 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12258 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12259 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12260 ("vec_unordered<mode>"): ... these.
12262 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12263 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12264 ("*vec_extendv2df"): New insn definitions.
12266 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12268 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12269 ("mulditi3_2", "*muldi3_sign"): New patterns.
12270 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12271 rename the pattern definition.
12273 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12275 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12277 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12279 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12281 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12282 instruction if possible.
12283 * config/s390/vector.md (vec_halfnumelts): New mode
12285 ("*vec_vllezlf<mode>"): New pattern.
12287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12289 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12290 ("popcountv4si2", "popcountv2di2"): Rename to ...
12291 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12292 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12294 ("popcount<mode>2_vxe"): New pattern.
12296 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12298 * common/config/s390/s390-common.c (processor_flags_table): Add
12300 * config.gcc: Add arch12.
12301 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12302 Default to arch12 for unknown CPU model numbers.
12303 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12304 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12305 PROCESSOR_max sanity check.
12306 * config/s390/s390-opts.h (enum processor_type): Add
12308 * config/s390/s390.c (processor_table): Add arch12.
12309 (s390_expand_builtin): Add check for B_VXE flag.
12310 (s390_issue_rate): Add PROCESSOR_ARCH12.
12311 (s390_get_sched_attrmask): Likewise.
12312 (s390_get_unit_mask): Likewise.
12313 (s390_sched_score): Enable z13 scheduling for arch12.
12314 (s390_sched_reorder): Likewise.
12315 (s390_sched_variable_issue): Likewise.
12316 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12318 (s390_tune_attr): Use z13 scheduling also for arch12.
12319 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12320 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12321 (TARGET_VXE_P): New macros.
12322 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12323 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12324 * config/s390/s390.opt: Add arch12 as processor_type.
12326 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12328 * config/s390/s390.md
12329 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12330 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12331 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12333 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12334 Rename expanders to ...
12336 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12337 ("fixuns_truncdddi2_emu"): ... these.
12339 ("fixuns_trunc<mode>si2_emu"): New expander.
12341 ("*fixuns_truncdfdi2_z13"): Rename to ...
12342 ("*fixuns_truncdfdi2_vx"): ... this.
12344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12346 * config/s390/2964.md: Remove the single element vector compare
12347 instructions which are no longer used.
12348 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12350 (s390_canonicalize_comparison): Remove handling of DFmode
12352 (s390_expand_vec_compare_scalar): Remove function.
12353 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12354 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12356 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12358 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12360 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12362 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12363 will anyway by matched by mov<mode>_64dfp.
12365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12367 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12368 vlef/vstef. Add missing operand to vleif.
12370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12372 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12373 pair for all vector types with 64 bit elements.
12374 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12375 * config/s390/vector.md (V_HW_64): ... here.
12376 (V_128_NOSINGLE): New mode iterator.
12377 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12378 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12379 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12380 ("*vec_load_pairv2di"): Change to ...
12381 ("*vec_load_pair<mode>"): ... this one.
12383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12385 * config/s390/constraints.md: Add comments.
12386 (jKK): Reject element sizes > 8 bytes.
12387 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12389 * config/s390/s390.md: Add the s_operand checks formerly in
12390 s390_split_ok_p to various splitters where they are still
12392 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12393 for 128 bit vectors. Plus two splitters.
12395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12397 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12400 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12403 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12404 error if the boundary argument is not constant.
12406 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12408 PR rtl-optimization/80112
12409 * loop-doloop.c (doloop_condition_get): Don't check condition
12410 if cmp isn't SET with IF_THEN_ELSE src.
12412 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12414 PR tree-optimization/80158
12415 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12416 replacing a candidate statement, also replace it for the
12417 candidate's alternate interpretation.
12418 (replace_rhs_if_not_dup): Likewise.
12419 (replace_one_candidate): Likewise.
12421 2017-03-24 Richard Biener <rguenther@suse.de>
12423 PR tree-optimization/80167
12424 * graphite-isl-ast-to-gimple.c
12425 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12427 (translate_isl_ast_to_gimple::get_rename): Likewise.
12429 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12431 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12432 handling of certain combinations of target options, including the
12433 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12434 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12436 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12439 * config/arm/arm.md (*load_multiple): Add reload_completed to
12440 matching condition.
12442 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12443 Richard Biener <rguenth@suse.de>
12445 PR tree-optimization/79908
12446 PR tree-optimization/80136
12447 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12448 been cast away, gimplify_and_add suffices.
12450 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12452 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12454 2017-03-23 Richard Biener <rguenther@suse.de>
12456 PR tree-optimization/80032
12457 * gimplify.c (gimple_push_cleanup): Forced unconditional
12458 cleanups still have to go to the conditional_cleanups
12461 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12463 PR tree-optimization/80072
12464 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12466 (next_operand_entry_id): Change type to unsigned int.
12467 (sort_by_operand_rank): Make sure to return the right return value
12468 even if unsigned fields are bigger than INT_MAX.
12469 (struct oecount): Change cnt and id type to unsigned int.
12470 (oecount_hasher::equal): Formatting fix.
12471 (oecount_cmp): Make sure to return the right return value
12472 even if unsigned fields are bigger than INT_MAX.
12473 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12476 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12477 TREE_READONLY on result if writing it more than once.
12480 * doc/invoke.texi (-fsanitize=thread): Document that with
12481 -fnon-call-exceptions atomics are not able to throw
12485 * tsan.c: Include tree-eh.h.
12486 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12487 maybe_clean_or_replace_eh_stmt where needed.
12488 (instrument_memory_accesses): Add cfg_changed argument.
12489 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12490 if it returned true.
12491 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12493 PR rtl-optimization/63191
12494 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12495 wrapper function, moved the whole old content into ...
12496 (ix86_delegitimize_address_1): ... this. New inline function.
12497 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12498 true as last argument instead of ix86_delegitimize_address.
12500 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12502 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12503 cortexa57_branch_cost.
12505 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12507 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12509 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12512 * doc/md.texi (Constraints): Document wA constraint.
12513 * config/rs6000/constraints.md (wA): New.
12514 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12515 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12516 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12517 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12519 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12522 * gimplify.c (is_oacc_declared): New function.
12523 (oacc_default_clause): Use it to set default flags for acc declared
12524 variables inside parallel regions.
12525 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12526 declared variables.
12527 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12528 declare attribute to any decl as necessary.
12530 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12533 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12534 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12535 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12536 (arm_arch_lpae): This.
12537 * config/arm/arm.c (arm_arch7ve): Rename into ...
12538 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12539 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12542 2017-03-22 Martin Liska <mliska@suse.cz>
12545 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12546 error message instead of an ICE.
12548 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12550 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12552 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12554 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12556 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12559 2017-03-21 Martin Sebor <msebor@redhat.com>
12561 * doc/extend.texi: Use "cannot" instead of "can't."
12562 * doc/hostconfig.texi: Same.
12563 * doc/install.texi: Same.
12564 * doc/invoke.texi: Same.
12565 * doc/loop.texi: Same.
12566 * doc/md.texi: Same.
12567 * doc/objc.texi: Same.
12568 * doc/rtl.texi: Same.
12569 * doc/tm.texi: Same.
12570 * doc/tm.texi.in: Same.
12571 * doc/trouble.texi: Same.
12573 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12576 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12577 (collect_checksum_attributes): Set it.
12578 (die_checksum_ordered): Use it.
12580 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12582 PR tree-optimization/79908
12583 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12584 change: For a VA_ARG whose LHS has been cast away, use
12585 force_gimple_operand to construct the side effects.
12587 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12589 PR translation/80001
12590 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12591 more amenable to translation.
12592 (oacc_loop_auto_partitions): Likewise.
12594 2017-03-21 Marek Polacek <polacek@redhat.com>
12595 Martin Sebor <msebor@redhat.com>
12597 PR tree-optimization/80109
12598 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12599 on INTEGRAL_TYPE_P.
12601 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12602 Segher Boessenkool <segher@kernel.crashing.org>
12605 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12606 check reg_used_between_p between insn and one of succ or succ2
12607 depending on if succ is artificial insn not inserted into insn
12610 2017-03-21 Martin Liska <mliska@suse.cz>
12612 PR gcov-profile/80081
12613 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12614 * doc/gcc.texi: Include gcov-dump stuff.
12615 * doc/gcov-dump.texi: New file.
12617 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12619 PR rtl-optimization/79150
12620 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12621 conditional jump, if the jump is the last insn of the loop.
12623 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12624 Richard Biener <rguenth@suse.de>
12626 PR tree-optimization/79908
12627 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12628 been cast away, use force_gimple_operand to construct the side
12631 2017-03-21 Martin Liska <mliska@suse.cz>
12633 PR libfortran/79956
12634 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12637 2017-03-21 Brad Spengler <spender@grsecurity.net>
12640 * plugin.c (htab_hash_plugin): New function.
12641 (add_new_plugin): Use it and adjust.
12642 (parse_plugin_arg_opt): Adjust.
12643 (init_one_plugin): Likewise.
12645 2017-03-21 Richard Biener <rguenther@suse.de>
12647 PR tree-optimization/80032
12648 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12649 if set force the cleanup to happen unconditionally.
12650 (gimplify_target_expr): Push inserted clobbers with force_uncond
12651 to avoid them being removed by control-dependent DCE.
12653 2017-03-21 Richard Biener <rguenther@suse.de>
12655 PR tree-optimization/80122
12656 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12657 va_arg_pack_len when the inlined call stmt requires pack
12659 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12661 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12664 * tsan.c (instrument_builtin_call): If the memory model argument
12665 is not a constant, assume it is valid.
12668 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12671 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12673 PR rtl-optimization/79910
12674 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12675 if its dest is used by an insn before I2 (other than the combined
12676 insns themselves, which are properly handled already).
12678 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12681 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12683 * combine.c (record_used_regs): New static function.
12684 (try_combine): Handle situations where there is an additional
12685 instruction between I2 and I3 which needs to have a LOG_LINK
12689 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12691 * combine.c (try_combine): Delete redundant i1 test. Call
12692 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12694 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12697 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12698 alternatives 13/14.
12700 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12702 PR tree-optimization/80054
12703 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12704 the optimization if a PHI or any of its arguments is not dominated
12705 by the candidate's basis. Use gphi* rather than gimple* as
12707 (replace_profitable_candidates): Clean up a gimple* variable that
12708 should be a gphi* variable.
12710 2017-03-20 Martin Sebor <msebor@redhat.com>
12713 * doc/extend.texi (attribute constructor): Document present limitation.
12715 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12718 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12719 __POWER9_VECTOR__ #ifdef control, change template definition to
12720 use Power9-specific built-in function.
12721 (vec_any_eq): Likewise.
12722 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12723 to control outcomes from this test.
12724 (vector_ae_<mode>p): For VEC_F modes, likewise.
12726 2017-03-20 Ian Lance Taylor <iant@google.com>
12728 * config/i386/i386.c (ix86_function_regparm): Save an extra
12729 register for -fsplit-stack with DECL_STATIC_CHAIN.
12731 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12734 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12735 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12737 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12739 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12741 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12744 2017-03-20 Marek Polacek <polacek@redhat.com>
12747 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12749 2017-03-20 Richard Biener <rguenther@suse.de>
12751 PR tree-optimization/80113
12752 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12753 allocate extra SSA name for PHI def.
12754 (add_close_phis_to_outer_loops): Likewise.
12755 (add_close_phis_to_merge_points): Likewise.
12756 (copy_loop_close_phi_args): Likewise.
12757 (copy_cond_phi_nodes): Likewise.
12759 2017-03-20 Martin Liska <mliska@suse.cz>
12761 PR middle-end/79753
12762 * tree-chkp.c (chkp_build_returned_bound): Do not build
12763 returned bounds for a LHS that's not a BOUNDED_P type.
12765 2017-03-20 Martin Liska <mliska@suse.cz>
12769 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12770 COMPLEX_CST and VECTOR_CST.
12772 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12775 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12776 target operand. A new splitter adds the clobber statement in case
12777 the target operand is dead anyway.
12779 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12781 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12782 to age-old versions of binutils and glibc.
12784 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12786 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12788 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12790 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12792 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12794 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12795 requirement for binutils 2.13.
12797 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12799 * combine.c (try_combine): Delete redundant i1 test. Call
12800 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12802 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12804 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12805 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12807 <riscv64-*-elf>: Re-arrange section
12808 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12809 <riscv32-*-linux>: Likewise.
12810 <riscv64-*-elf>: Likewise
12811 <riscv64-*-linux>: Likewise.
12813 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
12816 * aarch64.opt(verbose-cost-dump): Fix typo.
12818 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
12821 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
12822 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
12824 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12826 * reload.c (find_reloads): When reloading a nonoffsettable address,
12827 use RELOAD_OTHER for it and its address reloads.
12829 PR rtl-optimization/79910
12830 * combine.c (record_used_regs): New static function.
12831 (try_combine): Handle situations where there is an additional
12832 instruction between I2 and I3 which needs to have a LOG_LINK
12835 2017-03-17 Jeff Law <law@redhat.com>
12837 PR tree-optimization/71437
12838 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
12839 conditional in the hash table first.
12840 (vrp_dom_walker::before_dom_children): Extract condition from
12841 ASSERT_EXPR. Record condition, its inverion and any implied
12842 conditions as well.
12844 2017-03-17 Marek Polacek <polacek@redhat.com>
12845 Markus Trippelsdorf <markus@trippelsdorf.de>
12847 PR tree-optimization/80079
12848 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
12851 2017-03-17 Richard Biener <rguenther@suse.de>
12853 PR middle-end/80075
12854 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
12855 Properly verify the LHS before the RHS possibly claims to be
12857 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
12860 2017-03-17 Martin Jambor <mjambor@suse.cz>
12862 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
12863 (List of -O2 options): Likewise.
12864 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
12867 2017-03-17 Tom de Vries <tom@codesourcery.com>
12869 * gcov-dump.c (print_usage): Print bug_report_url.
12871 2017-03-17 Richard Biener <rguenther@suse.de>
12873 PR middle-end/80050
12874 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
12875 (parser::peek): Likewise.
12877 2017-03-17 Richard Biener <rguenther@suse.de>
12879 PR tree-optimization/80048
12880 * sese.c (free_sese_info): Properly release rename_map and
12881 copied_bb_map elements.
12883 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
12885 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
12886 Add linked-list forward and backlinks. Insert on
12887 construction, remove on destruction.
12888 (class pass_store_merging): Add m_stores_head field.
12889 (pass_store_merging::terminate_and_process_all_chains):
12890 Iterate over m_stores_head list.
12891 (pass_store_merging::terminate_all_aliasing_chains):
12893 (pass_store_merging::execute): Check for debug stmts first.
12894 Push new chains onto the m_stores_head stack.
12896 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
12899 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
12900 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
12901 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
12903 2017-03-16 Jeff Law <law@redhat.com>
12905 PR tree-optimization/71437
12906 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
12907 member function. Implementation moved into after_dom_children
12908 member function and into the threader's thread_outgoing_edges
12910 (dom_opt_dom_walker::after_dom_children): Simplify by moving
12911 some code into new thread_outgoing_edges.
12912 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
12913 definition. Simplify marker handling (do it here). Assume we always
12914 have the available expression and the const/copies tables.
12915 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
12917 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
12918 * tree-vrp.c (equiv_stack): No longer file scoped.
12919 (vrp_dom_walker): New class.
12920 (vrp_dom_walker::before_dom_children): New member function.
12921 (vrp_dom_walker::after_dom_children): Likewise.
12922 (identify_jump_threads): Setup domwalker. Use it rather than
12923 walking edges in a random order by hand. Simplify setup/finalization.
12924 (finalize_jump_threads): Remove.
12925 (vrp_finalize): Do not call identify_jump_threads here.
12926 (execute_vrp): Do it here instead and call thread_through_all_blocks
12929 PR tree-optimization/71437
12930 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
12932 (simplify_stmt_for_jump_threading): Add basic_block argument. All
12934 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
12935 (dom_opt_dom_walker::thread_across_edge): Remove
12936 handle_dominating_asserts argument. All callers changed.
12937 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
12938 changes. Remove calls to lhs_of_dominating_assert. Other
12939 uses of handle_dominating_asserts turn into unconditional code
12940 (simplify_control_stmt_condition_1): Likewise.
12941 (simplify_control_stmt_condition): Likewise.
12942 (thread_through_normal_block, thread_across_edge): Likewise.
12943 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
12944 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
12945 object if it is not an SSA_NAME.
12946 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
12947 before calling into the VRP specific simplifiers.
12948 (identify_jump_threads): Remove handle_dominating_asserts
12951 2017-03-16 Jakub Jelinek <jakub@redhat.com>
12954 * tree-diagnostic.c (default_tree_printer): No longer static.
12955 * tree-diagnostic.h (default_tree_printer): New prototype.
12957 2017-03-16 Tamar Christina <tamar.christina@arm.com>
12959 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
12960 Change ins into fmov.
12962 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12964 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
12965 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
12966 Use h_con constraint for operand 1.
12967 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
12968 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
12970 2017-03-15 Jeff Law <law@redhat.com>
12972 PR tree-optimization/71437
12973 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
12974 (record_temporary_equivalences): Use it.
12976 PR tree-optimization/71437
12977 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
12978 tree-ssa-scopedtables.
12979 (lookup_avail_expr, build_and_record_new_cond): Likewise.
12980 (record_conditions, record_cond, vuse_eq): Likewise.
12981 (record_edge_info): Adjust to API tweak of record_conditions.
12982 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
12983 (record_temporary_equivalences, optimize_stmt): Likewise.
12984 (eliminate_redundant_computations): Likewise.
12985 (record_equivalences_from_stmt): Likewise.
12986 * tree-ssa-scopedtables.c: Include options.h and params.h.
12987 (vuse_eq): New function, moved from tree-ssa-dom.c
12988 (build_and_record_new_cond): Likewise.
12989 (record_conditions): Likewise. Accept vector of conditions rather
12990 than edge_equivalence structure for first argument.
12991 for the first argument.
12992 (avail_exprs_stack::lookup_avail_expr): New member function, moved
12993 from tree-ssa-dom.c.
12994 (avail_exprs_stack::record_cond): Likewise.
12995 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
12996 from tree-ssa-dom.c.
12997 (avail_exprs_stack): Add new member functions lookup_avail_expr
12999 (record_conditions): Declare.
13001 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13004 * lra-constraints.c (process_alt_operands): Increase reject for
13005 reloading an input/output operand.
13007 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13010 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13011 insns to convert from signed/unsigned char/short to IEEE 128-bit
13013 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13015 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13018 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13019 subreg of inner mode for values already in registers.
13021 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13023 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13024 iteration reg is used after the loop.
13026 2017-03-14 Martin Sebor <msebor@redhat.com>
13028 PR tree-optimization/79800
13029 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13030 precision in negative-positive range.
13031 (format_floating): Call non-const overload with adjusted precision.
13033 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13036 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13039 2017-03-14 Martin Sebor <msebor@redhat.com>
13041 PR middle-end/80020
13042 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13043 * builtins.def (aligned_alloc): Use it.
13046 * Makefile.in (GTFILES): Add calls.c.
13047 * calls.c: Include "gt-calls.h".
13049 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13051 PR rtl-optimization/79728
13052 * regs.h (struct target_regs): New field
13053 x_contains_allocatable_regs_of_mode.
13054 (contains_allocatable_regs_of_mode): New macro.
13055 * reginfo.c (init_reg_sets_1): Initialize it, and change
13056 contains_reg_of_mode so it includes global regs as well.
13057 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13058 rather than contains_regs_of_mode.
13060 2017-03-14 Martin Liska <mliska@suse.cz>
13062 * doc/invoke.texi: Document options that can't be combined with
13063 -fcheck-pointer-bounds.
13065 2017-03-14 Martin Liska <mliska@suse.cz>
13067 PR middle-end/79831
13068 * doc/invoke.texi (-Wchkp): Document the option.
13070 2017-03-14 Martin Liska <mliska@suse.cz>
13072 * Makefile.in: Install gcov-dump.
13074 2017-03-14 Martin Liska <mliska@suse.cz>
13076 * multiple_target.c (expand_target_clones): Bail out for
13077 an invalid attribute.
13079 2017-03-14 Richard Biener <rguenther@suse.de>
13081 * alias.c (struct alias_set_entry): Pack properly.
13082 * cfgloop.h (struct loop): Likewise.
13083 * cse.c (struct set): Likewise.
13084 * ipa-utils.c (struct searchc_env): Likewise.
13085 * loop-invariant.c (struct invariant): Likewise.
13086 * lra-remat.c (struct cand): Likewise.
13087 * recog.c (struct change_t): Likewise.
13088 * rtl.h (struct address_info): Likewise.
13089 * symbol-summary.h (function_summary): Likewise.
13090 * tree-loop-distribution.c (struct partition): Likewise.
13091 * tree-object-size.c (struct object_size_info): Likewise.
13092 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13093 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13094 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13095 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13096 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13097 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13098 (struct _stmt_vec_info): Likewise.
13100 2017-03-14 Martin Liska <mliska@suse.cz>
13103 * multiple_target.c (create_dispatcher_calls): Check that
13104 a target can create a function dispatcher.
13106 2017-03-14 Martin Liska <mliska@suse.cz>
13109 * multiple_target.c (expand_target_clones): Drop local.local
13110 flag for default implementation.
13112 2017-03-14 Richard Biener <rguenther@suse.de>
13114 PR tree-optimization/80030
13115 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13117 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13119 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13120 gcc_fallthrough() instead of __attribute__((fallthrough));
13122 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13124 * doc/gcc.texi: Remove "up" link to (DIR).
13125 * doc/gccint.texi: Ditto.
13127 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13129 * doc/install.texi (Specific) <avr>: Remove reference to
13132 2017-03-13 Jeff Law <law@redhat.com>
13134 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13135 attribute rather than comments.
13137 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13138 match_scratch operand is highest.
13140 2017-03-13 Martin Liska <mliska@suse.cz>
13142 PR middle-end/78339
13143 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13144 is a CHKP clone, use original declaration.
13146 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13148 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13149 (arc_conditional_register_usage): Use a different allocation order
13150 when optimizing for size.
13151 * common/config/arc/arc-common.c (arc_option_optimization_table):
13152 Section anchors default on when optimizing for size.
13154 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13156 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13158 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13160 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13161 * config/arc/arc.md (cpu_facility): Add cd variant.
13162 (*movqi_insn): Add code density variant.
13163 (*movhi_insn): Likewise.
13164 (*movqi_insn): Likewise.
13165 (*addsi3_mixed): Likewise.
13166 (subsi3_insn): Likewise.
13168 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13170 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13172 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13174 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13175 expressions with MINUS and UNARY ops.
13177 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13180 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13182 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13183 between vec_select and vector argument.
13184 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13185 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13186 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13187 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13188 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13189 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13191 2017-03-13 Richard Biener <rguenther@suse.de>
13194 * params.def (vect-max-peeling-for-alignment): Fix typo.
13196 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13198 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13199 issue that only occurred with binutils below 2.18.
13201 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13203 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13204 refer to binutils 2.11/2.12 minimum.
13206 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13208 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13209 ftp.kernel.org and simplify binutils requirement.
13211 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13213 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13215 (Optimize Options): Ditto. Also remove redundancy.
13217 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13219 PR translation/79848
13220 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13222 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13223 to G_ to avoid double translation.
13225 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13227 PR translation/79923
13228 * auto-profile.c (get_combined_location): Convert leading
13229 character of diagnostics to lower case and remove trailing period.
13230 (read_profile): Likewise for various diagnostics.
13231 * config/arm/arm.c (arm_option_override): Remove trailing period
13232 from various diagnostics.
13233 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13234 (msp430_expand_delay_cycles): Likewise.
13236 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13239 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13240 full command-line argument, rather than just "str".
13241 (aarch64_validate_march): Likewise.
13242 (aarch64_validate_mtune): Likewise.
13244 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13246 PR rtl-optimization/78911
13247 * lra-assigns.c (must_not_spill_p): New function.
13248 (spill_for): Use it.
13250 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13252 PR tree-optimization/79981
13253 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13254 ATOMIC_COMPARE_EXCHANGE ifn result.
13255 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13256 IFN_ATOMIC_COMPARE_EXCHANGE.
13258 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13261 * opts.c (parse_sanitizer_options): Add missing question mark to
13262 "did you mean" message.
13264 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13266 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13268 (VMULEUH_UNS): Likewise.
13269 (VMULOUB_UNS): Likewise.
13270 (VMULOUH_UNS): Likewise.
13271 * config/rs6000/rs6000.c (builtin_function_type): Remove
13272 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13274 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13277 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13278 x with result of extra_parsing_for_operand_code_0.
13279 (function_reader::extra_parsing_for_operand_code_0): Convert
13280 return type from void to rtx, returning x. When reading
13281 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13282 larger size containing struct block_symbol.
13284 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13286 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13287 -mfloat128-hardware without -m64.
13289 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13292 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13293 entries to the case statement that marks unsigned arguments to
13294 overloaded functions.
13296 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13298 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13299 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13301 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13304 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13305 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13307 2017-03-10 Martin Liska <mliska@suse.cz>
13311 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13312 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13315 2017-03-10 Olivier Hainque <hainque@adacore.com>
13317 * tree-switch-conversion (array_value_type): Start by resetting
13318 candidate type to it's main variant.
13320 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13322 PR rtl-optimization/79909
13323 * combine.c (try_combine): Use simplify_replace_rtx on individual
13324 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13325 of the whole CALL_INSN_FUNCTION_USAGE.
13327 PR tree-optimization/79972
13328 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13329 get_range_info on SSA_NAMEs. Formatting fixes.
13331 2017-03-10 Richard Biener <rguenther@suse.de>
13332 Jakub Jelinek <jakub@redhat.com>
13334 PR tree-optimization/77975
13335 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13336 edge to be constant.
13337 (get_val_for): For constant x return it. Formatting fix.
13338 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13339 would use the same bases as the current one.
13341 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13343 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13344 instead of vec_select for V1TImode.
13345 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13347 (VSX_LE_128): Add V1TI to this mode iterator.
13348 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13349 (*vsx_le_perm_store_<mode>): Likewise.
13350 (pre-reload splitter for VSX stores): Likewise.
13351 (post-reload splitter for VSX stores): Likewise.
13352 (*vsx_xxpermdi2_le_<mode>): Likewise.
13353 (*vsx_lxvd2x2_le_<mode>): Likewise.
13354 (*vsx_stxvd2x2_le_<mode>): Likewise.
13356 2017-03-09 Michael Eager <eager@eagercon.com>
13358 Correct failures with --enable-checking=yes,rtl.
13360 * config/microblaze/microblaze.c (microblaze_expand_shift):
13361 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13362 test for const0_rtx.
13363 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13364 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13366 2017-03-09 Richard Biener <rguenther@suse.de>
13368 PR tree-optimization/79977
13369 * graphite-scop-detection.c (scop_detection::merge_sese):
13370 Handle the case of extra exits to blocks dominating the entry.
13372 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13374 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13377 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13379 PR rtl-optimization/79949
13380 * lra-constraints.c (process_alt_operands): Check memory when
13381 trying to predict a cycle. Print about the overall increase.
13383 2017-03-09 Richard Biener <rguenther@suse.de>
13385 PR middle-end/79971
13386 * gimple-expr.c (useless_type_conversion_p): Preserve
13387 TYPE_SATURATING for fixed-point types.
13389 2017-03-09 Richard Biener <rguenther@suse.de>
13392 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13393 alignment of BLKmode params.
13395 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13398 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13399 (VALL_NO_V2Q): Likewise.
13401 * config/aarch64/aarch64-simd.md
13402 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13404 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13405 VALL_NO_V2Q mode iterator.
13406 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13408 2017-03-09 Martin Liska <mliska@suse.cz>
13410 PR tree-optimization/79631
13411 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13412 tree_int_cst_sign_bit just for INTEGER constants.
13414 2017-03-09 Martin Liska <mliska@suse.cz>
13418 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13421 2017-03-09 Marek Polacek <polacek@redhat.com>
13424 * tree.c (inchash::add_expr): Handle TREE_VEC.
13426 2017-03-09 Martin Liska <mliska@suse.cz>
13429 (chkp_narrow_size_and_offset): New function.
13430 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13431 (void chkp_parse_bit_field_ref): New function.
13432 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13433 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13435 2017-03-09 Martin Liska <mliska@suse.cz>
13438 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13439 (chkp_find_bounds_1): Remove gcc_unreachable.
13441 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13444 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13445 BUILT_IN_SYNC*, determine the access type from the size suffix and
13446 always build a MEM_REF with that type. Handle forgotten
13447 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13450 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13451 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13452 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13453 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13454 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13455 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13456 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13457 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13458 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13459 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13460 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13461 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13462 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13463 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13464 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13465 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13466 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13467 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13468 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13469 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13470 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13471 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13472 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13473 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13474 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13475 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13476 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13477 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13478 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13479 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13480 definitions outside of __OPTIMIZE__ guarded section.
13483 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13484 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13485 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13486 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13489 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13491 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13492 ("vfenez<mode>"): Add missing constraints.
13494 2017-03-08 Martin Sebor <msebor@redhat.com>
13497 * config/nds32/nds32.c (nds32_option_override):
13498 Fix misspelled diagnostic.
13500 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13503 * gimplify.c (gimplify_omp_for): Replace index var in outer
13504 taskloop statement with an artificial variable and add
13505 OMP_CLAUSE_PRIVATE clause for it.
13507 2017-03-08 Richard Biener <rguenther@suse.de>
13509 PR tree-optimization/79955
13510 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13511 for accesses that are completely outside of the variable.
13513 2017-03-08 Andrew Haley <aph@redhat.com>
13515 PR tree-optimization/79943
13516 * tree-ssa-loop-split.c (compute_new_first_bound): When
13517 calculating the new upper bound, (END-BEG) should be added, not
13520 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13522 * config/avr/avr.md (setmemhi): Make sure match_dup
13523 operand number comes before match_scratch.
13525 2017-03-08 Richard Biener <rguenther@suse.de>
13527 PR tree-optimization/79920
13528 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13529 with ncopies == 1 to ...
13530 (vect_transform_slp_perm_load): ... here. Properly compute
13531 all element loads by iterating VF times over the group. Do
13532 not handle ncopies (computed in a broken way) in
13533 vect_create_mask_and_perm.
13535 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13538 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13539 is a uniform vector, use uniform_vector_p return value instead of
13540 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13542 2017-03-07 Marek Polacek <polacek@redhat.com>
13544 PR middle-end/79809
13545 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13546 (alloca_call_type): Likewise.
13548 2017-03-07 Martin Liska <mliska@suse.cz>
13550 * gcov.c (process_args): Put comment to correct location.
13552 2017-03-07 Martin Liska <mliska@suse.cz>
13554 PR middle-end/68270
13555 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13556 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13557 (chkp_narrow_bounds_for_field): Likewise.
13558 (chkp_parse_array_and_component_ref): Pass one more argument to
13561 2017-03-07 Richard Biener <rguenther@suse.de>
13563 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13566 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13568 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13569 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13571 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13574 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13575 to end of description.
13576 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13578 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13580 PR rtl-optimization/79901
13581 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13583 (*avx512f_<code><mode>3<mask_name>): ... this.
13584 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13585 iterator instead of VI8_AVX2_AVX512BW.
13587 PR rtl-optimization/79901
13588 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13589 min/max expander, expand it using expand_vec_cond_expr.
13592 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13595 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13598 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13599 to void * for PCH reasons.
13600 * dwarf2out.c (output_loc_operands, output_die): Cast
13601 v.val_vec.array to unsigned char *.
13603 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13606 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13609 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13611 PR rtl-optimization/79571
13612 * lra-constraints.c (process_alt_operands): Calculate static
13613 reject and subtract it from overall when only addresses will be
13616 2017-03-06 Julia Koval <julia.koval@intel.com>
13619 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13620 incoming stack boundary to 128 for 64-bit targets.
13622 2017-03-06 Richard Biener <rguenther@suse.de>
13624 PR tree-optimization/79894
13625 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13626 to NULL after folding it.
13628 2017-03-06 Richard Biener <rguenther@suse.de>
13630 PR tree-optimization/79824
13631 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13632 check disabling peeling for gaps.
13634 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13636 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13637 attributes): Document gettimeofday.
13639 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13641 * config/s390/s390.c (s390_option_override_internal): Set
13642 PARAM_MIN_VECT_LOOP_BOUND
13644 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13646 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13647 * config/s390/s390.md: Likewise.
13649 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13652 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13653 (<avx2_avx512>_perm<mode>): Rename to ...
13654 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13656 (<avx512>_perm<mode>_mask): Rename to ...
13657 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13659 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13660 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13661 instead of VI8F_256_512.
13662 (avx512f_perm<mode>): New define_expand.
13663 (avx512f_perm<mode>_mask): Likewise.
13664 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13665 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13667 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13669 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13670 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13672 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13674 2017-03-06 Martin Liska <mliska@suse.cz>
13677 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13678 when having a SSA NAME w/o VAR_DECL assigned to it.
13680 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13682 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13683 msa_dpsub_<su>_d): Fix MODE for vec_select.
13685 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13687 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13689 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13691 2017-03-06 Richard Biener <rguenther@suse.de>
13693 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13694 * plugin.c (register_plugin_info): Likewise.
13695 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13697 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13699 * config/i386/sse.md (sse_storehps, sse_storelps,
13700 avx_<castmode><avxsizesuffix>_<castmode>,
13701 avx512f_<castmode><avxsizesuffix>_<castmode>,
13702 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13703 in condition that at least one operand is not a MEM.
13705 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13707 PR middle-end/79805
13708 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13709 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13711 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13712 gimple_call_nothrow_p flag based on whether original builtin can throw.
13713 If it can, emit following stmts on the fallthrough edge.
13714 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13715 don't create new bb if inserting just debug stmts on the edge, try to
13716 insert them on the fallthru bb or just reset debug stmts.
13718 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13721 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13722 restore recog_data (including the operand rtxes inside it) around
13723 the call to get_insn_template.
13725 2017-03-03 Martin Sebor <msebor@redhat.com>
13727 PR tree-optimization/79699
13728 * context.c (context::~context): Free MPFR caches to avoid
13729 a memory leak on program exit.
13731 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13733 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13734 Use wide_int::ulow () instead of .elt (0).
13736 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13738 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13739 (*pushxf): Limit oF constraint to 32bit targets and add oC
13740 constraint for 64bit targets.
13741 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13742 (*pushdf): Change rmF constraint to rmC.
13744 2017-03-03 Martin Liska <mliska@suse.cz>
13746 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13747 Remove unused variable.
13749 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13752 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13753 is a memory operand, increase num_memory.
13754 (ix86_expand_args_builtin): Likewise.
13756 2017-03-03 Jan Hubicka <jh@suse.cz>
13759 * ipa-devirt.c (maybe_record_node): Properly handle
13760 __cxa_pure_virtual visibility.
13762 2017-03-03 Martin Liska <mliska@suse.cz>
13764 PR tree-optimization/79803
13765 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13767 (pass_loop_prefetch::execute): Disabled optimization if an
13768 assumption about L1 cache size is not met.
13770 2017-03-03 Martin Liska <mliska@suse.cz>
13772 PR rtl-optimization/79574
13773 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13774 (hash_scan_set): Likewise.
13775 (dump_hash_table): Likewise.
13776 (hoist_code): Likewise.
13778 2017-03-03 Richard Biener <rguenther@suse.de>
13780 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13781 in place of uhigh ().
13782 (fixed_convert_from_real): Likewise.
13784 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13787 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13789 2017-03-03 Richard Biener <rguenther@suse.de>
13791 PR middle-end/79818
13792 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13793 TYPE_OVERFLOW_UNDEFINED check.
13795 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13797 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13799 (vector_ae_<mode>_p): Likewise.
13800 (vector_nez_<mode>_p): Likewise.
13801 (vector_ne_v2di_p): Likewise.
13802 (vector_ae_v2di_p): Likewise.
13803 (vector_ne_<mode>_p): Likewise.
13804 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13806 (vsx_tsqrt<mode>2_fe): Likewise.
13808 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13811 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
13813 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13815 PR rtl-optimization/79780
13816 * cprop.c (one_cprop_pass): When second and further conditional trap
13817 in a single basic block is turned into an unconditional trap, turn it
13818 into a deleted note to avoid RTL verification failures.
13820 2017-03-02 Richard Biener <rguenther@suse.de>
13822 * fold-const.c (const_binop): Use ulow () instead of elt (0).
13824 2017-03-02 Richard Biener <rguenther@suse.de>
13826 PR tree-optimization/79345
13828 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
13829 param and abort the walk, returning -1 if it is hit.
13830 (walk_aliased_vdefs): Take a limit param and pass it on.
13831 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
13832 defaulting to 0 and return a signed int.
13833 * tree-ssa-uninit.c (struct check_defs_data): New struct.
13834 (check_defs): New helper.
13835 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
13836 about uninitialized memory.
13837 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
13838 bogus uninitialized warning.
13839 (fixed_convert_from_real): Likewise.
13841 2017-03-02 Bin Cheng <bin.cheng@arm.com>
13843 PR tree-optimization/66768
13844 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
13845 iv_use if base object can't be determined.
13847 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13849 PR tree-optimization/79345
13850 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
13851 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
13852 (get_pattern_stats): Initialize it.
13853 * genemit.c (gen_expand): Verify match_scratch numbers come after
13854 match_operand/match_dup numbers.
13855 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
13856 match_scratch numbers.
13857 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
13859 * config/s390/s390.md (trunctdsd2): Likewise.
13861 2017-03-02 Richard Biener <rguenther@suse.de>
13863 * wide-int.h (wide_int_storage::operator=): Implement in terms
13866 2017-03-02 Richard Biener <rguenther@suse.de>
13868 PR tree-optimization/79777
13869 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
13870 the to insert expression to sth existing.
13872 2017-03-01 Martin Sebor <msebor@redhat.com>
13874 PR middle-end/79692
13875 * gimple-ssa-sprintf.c
13876 (directive::known_width_and_precision): New function.
13877 (format_integer): Use it.
13878 (get_mpfr_format_length): Consider the full range of precision
13879 when computing %g output with the # flag. Set the likely byte
13880 count to 3 rather than 1 when precision is indeterminate.
13881 (format_floating): Correct the lower bound of precision.
13883 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13885 * doc/invoke.texi: Document default code model for 64-bit Linux.
13887 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13890 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
13891 udiv rather than div since input pattern is unsigned.
13893 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
13895 * config/i386/i386.c (print_reg): Warn for values of
13896 unsupported size in integer register.
13898 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
13901 * config/rs6000/predicates.md (current_file_function_operand): Do
13902 not allow self calls to be local if the function is replaceable.
13904 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
13907 * config/rs6000/altivec.h (vec_ctz and others): Change the
13908 preprocessor macro that controls conditional compilation from
13909 _ARCH_PWR9 to __POWER9_VECTOR__.
13910 (vec_all_ne): Change parameterization of __altivec_scalar_pred
13911 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
13912 control (instead of _ARCH_PWR9 control) so that template
13913 definition uses power9-specific function.
13914 (vec_any_eq): Likewise.
13915 (vec_all_ne): Change macro definition to use a power9-specific
13916 expansion under #ifdef __POWER9_VECTOR__ control (instead of
13917 _ARCH_PWR9 control).
13918 (vec_any_eq) Likewise.
13919 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
13920 expansion for CMPNEF to remove support for xvcmpnesp instruction.
13921 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
13922 support for xvcmpnedp instruction.
13923 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
13924 macro expansion so that Power9 implementation of vec_all_ne does
13925 not use the AltiVec predicate framework.
13926 (VCMPNEH_P): Likewise.
13927 (VCMPNEW_P): Likewise.
13928 (VCMPNED_P): Likewise.
13929 (VCMPNEFP_P): Likewise.
13930 (VCMPNEDP_P): Likewise.
13931 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
13932 implementation of vec_any_eq to not use AltiVec predicate
13934 (VCMPAEH_P): Likewise.
13935 (VCMPAEW_P): Likewise.
13936 (VCMPAED_P): Likewise.
13937 (VCMPAEFP_P): Likewise.
13938 (VCMPAEDP_P): Likewise.
13939 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
13940 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
13941 not use the AltiVec predicate framework.
13942 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
13943 of vec_any_eq to not use AltiVec predicate framework.
13944 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13945 support for predefined __POWER9_VECTOR__ macro to indicate that
13946 Power9 instruction selection is enabled.
13947 (altivec_overloaded_builtins): Remove extraneous
13948 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
13949 function argument types RS6000_BTI_bool_V16QI and
13950 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
13951 entry for overloaded function argument types RS6000_BTI_bool_V4SI
13952 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
13953 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
13954 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
13955 Power9 for implementations of vec_cmpne. Change the signature for
13956 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
13957 (representing vec_all_ne) to remove the previously described first
13958 argument of type RS6000_BTI_INTSI, as this was an artifact of
13959 reliance on the AltiVec predicate framework, which is no longer
13960 used in the implementation of these functions. Add
13961 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
13962 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
13963 since, unlike the AltiVec predicate framework implementation, we
13964 do not share function descriptors between vec_alle and vec_anyeq.
13965 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
13966 set of modes that receive special treatment even when
13967 TARGET_P9_VECTOR is true. The special treatment emits code that
13968 does not depend on Power9 instructions.
13969 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
13970 define_expand to not rely on AltiVec predicate framework.
13971 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13973 (vector_ne_v2di_p): Change this define_expand to not rely on
13974 AltiVec predicate framework.
13975 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
13977 (vector_ne_<mode>_p): Change this define_expand to not rely on
13978 AltiVec predicate framework.
13979 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13981 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
13982 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
13983 define_insn pattern.
13984 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
13985 define_insn pattern because the xvcmpne<VSs>. instruction is not
13987 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
13988 instruction is not supported.
13990 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13992 * config/nvptx/nvptx.c: Include intl.h.
13994 2017-03-01 Martin Jambor <mjambor@suse.cz>
13997 * ipa-prop.h (ipa_bits): Removed field known.
13998 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
13999 to pointers. Adjusted their comments to warn about their sharing.
14000 (ipcp_transformation_summary): Change bits to a vector of pointers.
14001 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14002 (ipa_get_ipa_bits_for_value): Declare.
14003 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14004 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14005 (ipa_bits_hash_table): Likewise.
14006 (ipa_vr_ggc_hash_traits): Likewise.
14007 (ipa_vr_hash_table): Likewise.
14008 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14009 being pointers and vr_known being removed.
14010 (ipa_set_jf_unknown): Likewise.
14011 (ipa_get_ipa_bits_for_value): New function.
14012 (ipa_set_jfunc_bits): Likewise.
14013 (ipa_get_value_range): New overloaded functions.
14014 (ipa_set_jfunc_vr): Likewise.
14015 (ipa_compute_jump_functions_for_edge): Use the above functions to
14016 construct bits and vr parts of jump functions.
14017 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14018 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14020 (ipcp_grow_transformations_if_necessary): Also allocate
14021 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14023 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14024 them. Fix too long lines.
14025 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14026 vr_known being removed.
14027 (ipa_read_jump_function): Use new setter functions to construct bits
14028 and vr parts of jump functions or set them to NULL.
14029 (write_ipcp_transformation_info): Adjust for bits being pointers.
14030 (read_ipcp_transformation_info): Likewise.
14031 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14033 Include gt-ipa-prop.h.
14034 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14036 (ipcp_store_bits_results): Likewise.
14037 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14038 Do not write to existing jump functions but use a temporary instead.
14040 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14043 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14044 attempt to use its first operand as BIT_FIELD_REF base.
14046 2017-03-01 Richard Biener <rguenther@suse.de>
14048 PR middle-end/79721
14049 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14050 interpolating formula in wrapping arithmetic.
14051 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14053 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14055 PR tree-optimization/79734
14056 * tree-vect-generic.c (expand_vector_condition): Optimize
14057 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14058 Handle VEC_COND_EXPR where comparison has different inner width from
14059 type's inner width.
14061 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14063 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14064 markup, and similar issues. Remove @opindex entries for things
14065 that aren't options. Add missing -mmpy-option entries.
14067 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14069 PR tree-optimization/79737
14070 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14071 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14072 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14073 instead of byte_size. Formatting fix.
14074 (shift_bytes_in_array_right): Formatting fix.
14076 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14079 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14080 condition on optimize for the leaf function test.
14082 2017-02-28 Martin Liska <mliska@suse.cz>
14085 * read-rtl-function.c (function_reader::handle_unknown_directive):
14086 Bail out when one uses -flto.
14088 2017-02-28 Martin Liska <mliska@suse.cz>
14090 * common.opt: Replace space with tabular for options of <number>
14092 * config/i386/i386.opt: Show <number> value for
14093 -mlarge-data-threshold.
14094 * opts.c (print_filtered_help): Do not display number in hexadecimal
14097 2017-02-28 Martin Liska <mliska@suse.cz>
14099 * common.opt: Fix --help=option -Q for options which are of
14102 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14104 * config/i386/i386.c (print_reg): Error out for values
14105 of 8-bit size in invalid integer register.
14107 2017-02-28 Martin Sebor <msebor@redhat.com>
14109 PR tree-optimization/79691
14110 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14112 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14115 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14116 gcc_unreachable with output_operand_lossage.
14118 2017-02-28 Richard Biener <rguenther@suse.de>
14120 PR tree-optimization/79740
14121 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14123 (visit_nary_op): Insert the nary into the hashtable if we
14124 pattern-matched sth.
14125 * tree-ssa-pre.c (eliminate_insert): Robustify.
14127 2017-02-28 Richard Biener <rguenther@suse.de>
14129 PR middle-end/79731
14130 * fold-const.c (decode_field_reference): Reject out-of-bound
14133 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14135 * config/i386/i386.c: Include intl.h.
14136 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14137 instead of just cond ? "..." : "...".
14138 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14139 * coverage.c (read_counts_file): Likewise.
14140 * omp-offload.c: Include intl.h.
14141 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14142 of just cond ? "..." : "...".
14143 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14144 of just cond ? "..." : "...".
14146 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14149 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14151 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14152 'tune for' CPU name.
14153 * config/arm/arm-cpu-data.h: Regenerated.
14155 2017-02-28 Richard Biener <rguenther@suse.de>
14157 PR tree-optimization/79732
14158 * tree-inline.c (expand_call_inline): Do not shadow var.
14160 2017-02-28 Richard Biener <rguenther@suse.de>
14162 PR tree-optimization/79723
14163 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14164 address-space properly.
14166 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14168 * doc/optinfo.texi (Optimization groups): Fix option used for
14170 * doc/invoke.texi (-fopt-info): Document "omp".
14171 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14172 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14173 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14174 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14175 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14177 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14179 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14180 OPTGROUP_OMP with "omp".
14182 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14185 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14186 for arithmetic shift of unsigned V2DI.
14188 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14190 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14191 arc/linux.h headers.
14192 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14193 (LINK_SPEC): Likewise.
14194 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14195 (EXTRA_SPECS): Likewise.
14196 (STARTFILE_SPEC): Likewise.
14197 (ENDFILE_SPEC): Likewise.
14198 (LIB_SPEC): Likewise.
14199 (TARGET_SDATA_DEFAULT): Likewise.
14200 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14201 (MULTILIB_DEFAULTS): Likewise.
14202 (DWARF2_UNWIND_INFO): Likewise.
14203 * config/arc/big.h: New file.
14204 * config/arc/elf.h: Likewise.
14205 * config/arc/linux.h: Likewise.
14206 * config/arc/t-uClibc: Remove.
14208 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14210 PR tree-optimization/77536
14211 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14212 (tree_transform_and_unroll_loop): Use above function to compute the
14213 estimated niter of unrolled loop and use it when scaling profile.
14214 Also use count info rather than frequency if it's non-zero.
14215 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14216 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14217 (vect_transform_loop): Call above function.
14219 2017-02-27 Richard Biener <rguenther@suse.de>
14221 PR tree-optimization/45397
14222 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14223 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14224 (visit_nary_op): Add pattern matching for CSEing sign-changed
14225 or truncated operations with wider ones.
14227 2017-02-27 Richard Biener <rguenther@suse.de>
14229 PR tree-optimization/79690
14230 * tree-vect-stmts.c (vectorizable_store): Use vector type
14231 built from the DR with address-space.
14233 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14235 * doc/invoke.texi (Optimize Options): Refine the description
14236 of asan-use-after-return.
14238 2017-02-25 Alan Modra <amodra@gmail.com>
14240 PR rtl-optimization/79584
14241 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14242 base, not ad->base_term, the reg within base. Remove assertion
14243 that ad->base == ad->base_term. Replace gen_int_mode using
14244 bogus mode with const0_rtx.
14246 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14248 PR middle-end/79396
14249 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14250 FMA_EXPR like tcc_binary or tcc_unary.
14252 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14255 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14257 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14258 (output_loc_operands): Handle DW_OP_call_ref and
14259 DW_OP_GNU_variable_value.
14260 (struct variable_value_struct): New type.
14261 (struct variable_value_hasher): Likewise.
14262 (variable_value_hash): New variable.
14263 (string_types): Remove.
14264 (copy_loc_descr): New function.
14265 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14266 (prepend_loc_descr_to_each): New function.
14267 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14268 instead of add_loc_descr_to_each if the first argument is single
14269 location list and the second has multiple.
14270 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14271 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14272 when looking for variable value which doesn't have other location info.
14273 (loc_list_from_tree): Formatting fix.
14274 (gen_array_type_die): Simplify DW_AT_string_length handling.
14275 (adjust_string_types): Remove.
14276 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14277 string_types. Call resolve_variable_values.
14278 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14279 (resolve_addr_in_expr): Likewise. Add A argument.
14280 (copy_deref_exprloc): Remove deref argument. Adjust for the
14281 original expression being DW_OP_GNU_variable_value with optionally
14282 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14283 optionally after it.
14284 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14285 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14286 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14287 (variable_value_hasher::hash, variable_value_hasher::equal): New
14289 (resolve_variable_value_in_expr, resolve_variable_value,
14290 resolve_variable_values, note_variable_value_in_expr,
14291 note_variable_value): New functions.
14292 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14295 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14298 * opts.h (handle_generated_option): Add GENERATED_P argument.
14299 * opts-common.c (handle_option): Adjust function comment.
14300 (handle_generated_option): Add GENERATED_P argument, pass it to
14302 (control_warning_option): Pass false to handle_generated_option
14304 * opts.c (maybe_default_option): Pass true to handle_generated_option
14306 * optc-gen.awk: Likewise.
14308 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14310 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14311 a REG, look at the REG it is a SUBREG of.
14312 (splitter for cmpeqsi_t): Ditto.
14314 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14316 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14317 the special USEs with the pattern of the insn, not the insn itself.
14319 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14322 * doc/invoke.texi: Document -mload-store-pairs.
14324 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14325 Sandra Loosemore <sandra@codesourcery.com>
14327 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14328 argument isn't a CONST_INT.
14329 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14330 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14331 (nios2_validate_compare): Bypass alternate compare logic if *op2
14332 is not a CONST_INT.
14333 (ldstwm_operation_p): Return false if first_base is not a REG or
14334 if first_offset is not a CONST_INT.
14336 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14338 * config/cris/cris.md: Use correct operand in a define_peephole2.
14340 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14342 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14344 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14346 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14347 this_insn if it is an INSN or JUMP_INSN.
14348 (force_offsettable): Look at base, not at addr.
14349 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14350 on things that aren't necessarily CONST_INTs.
14352 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14354 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14355 -mfpmath=sse is the default also for x86-32 targets with SSE2
14356 instruction set when @option{-ffast-math} is enabled
14358 2017-02-24 Jeff Law <law@redhat.com>
14360 PR rtl-optimizatoin/79286
14361 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14364 2017-02-24 Richard Biener <rguenther@suse.de>
14366 PR tree-optimization/79389
14367 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14370 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14372 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14373 function comment to reflect reality.
14374 (loop_exits_before_overflow): Fix typo in function description.
14376 2017-02-24 Richard Biener <rguenther@suse.de>
14378 PR tree-optimization/79389
14379 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14380 properly that a threading opportunity exists. Detect conditional
14381 copy/constant propagation opportunities.
14383 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14385 * config/visium/visium.md (type): Add trap.
14386 (b): New mode attribute.
14387 (*btst): Rename into...
14388 (*btst<mode>): ...this and adjust.
14389 (*cbranchsi4_btst_insn): Rename into...
14390 (*cbranch<mode>4_btst_insn): ...this and adjust.
14391 (trap): New define_insn.
14393 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14395 PR tree-optimization/79389
14396 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14397 (noce_reversed_cond_code): New function.
14398 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14399 reversed_comparison_code. Formatting fix.
14400 (noce_try_store_flag): Test rev_cond != NULL in addition to
14401 reversed_comparison_code.
14402 (noce_try_store_flag_constants): Likewise.
14403 (noce_try_store_flag_mask): Likewise.
14404 (noce_try_addcc): Use rev_cond if non-NULL instead of
14405 reversed_comparison_code.
14406 (noce_try_cmove_arith): Likewise. Formatting fixes.
14407 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14408 (noce_find_if_block): Initialize rev_cond.
14409 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14410 instead of false as last argument never attempt to reverse it
14413 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14415 PR tree-optimization/79663
14416 * tree-predcom.c (combine_chains): Process refs in reverse order
14417 only for ZERO length chains, and add explaining comment.
14419 2017-02-23 Jeff Law <law@redhat.com>
14421 PR tree-optimization/79578
14422 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14423 in call to operand_equal_p.
14425 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14428 * config/i386/cpuid.h: Fix another undefined behavior.
14430 2017-02-23 Richard Biener <rguenther@suse.de>
14432 PR tree-optimization/79683
14433 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14434 vector types for data-refs.
14436 2017-02-23 Martin Liska <mliska@suse.cz>
14438 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14440 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14442 PR middle-end/79665
14443 * internal-fn.c (get_range_pos_neg): Moved to ...
14444 * tree.c (get_range_pos_neg): ... here. No longer static.
14445 * tree.h (get_range_pos_neg): New prototype.
14446 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14447 are known to be in between 0 and signed maximum inclusive, try to
14448 expand both unsigned and signed divmod and use the cheaper one from
14451 2017-02-22 Jeff Law <law@redhat.com>
14453 PR tree-optimization/79578
14454 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14455 to compare base operands.
14457 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14460 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14461 gpc_reg_operand instead of fpr_reg_operand.
14463 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14465 * config/mips/mips.c (mips_return_in_memory): Force FP
14466 vector types to be returned in memory for o32 ABI.
14468 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14470 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14471 instead of DW_TAG_member for static data member declarations and don't
14472 set no_linkage_name for static inline data members.
14473 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14476 2017-02-22 Martin Liska <mliska@suse.cz>
14478 * doc/invoke.texi: Replace inequality signs with square brackets
14481 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14483 PR tree-optimization/68644
14484 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14486 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14489 * lra-constraints.c (simplify_operand_subreg): Handle
14490 WORD_REGISTER_OPERATIONS targets.
14492 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14495 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14496 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14497 elimination by swapping fld*.
14499 2017-02-22 Richard Biener <rguenther@suse.de>
14501 PR tree-optimization/79673
14502 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14503 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14504 irrelevant address-space qualifiers and avoiding a
14505 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14507 2017-02-22 Richard Biener <rguenther@suse.de>
14509 PR tree-optimization/79666
14510 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14511 to not symbolically negate if that may introduce undefined
14514 2017-02-22 Martin Liska <mliska@suse.cz>
14517 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14518 * data-streamer-out.c (streamer_write_gcov_count_stream):
14520 * value-prof.c (stream_out_histogram_value): Make assert more
14521 precise based on type of counter.
14523 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14526 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14527 Use nonimmediate_operand instead of memory_operand for operand 1.
14528 (float-extend standard_x87sse_constant_load splitter): Ditto.
14530 2017-02-21 Jeff Law <law@redhat.com>
14532 PR tree-optimization/79621
14533 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14534 blocks with edges to themselves.
14536 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14539 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14540 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14541 Use gimple_call_builtin_p.
14544 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14545 on temporarily removed DEBUG_INSNs.
14547 PR tree-optimization/79649
14548 * tree-loop-distribution.c (classify_partition): Give up on
14549 non-generic address space loads/stores.
14551 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14553 * doc/loop.texi (Loop manipulation): Remove nonexistent
14554 tree_ssa_loop_version from the documentation.
14555 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14557 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14560 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14561 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14562 * config/rs6000/rs6000.c: Include except.h.
14563 (rs6000_expand_split_stack_prologue): Call
14564 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14566 2017-02-21 Martin Jambor <mjambor@suse.cz>
14569 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14570 have been analyzed.
14572 2017-02-21 Martin Jambor <mjambor@suse.cz>
14574 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14575 for backward compatibility only.
14576 * doc/invoke.texi (Option Summary): Remove all references to
14577 -fipa-cp-alignment.
14579 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14583 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14585 * lra-constraints.c (curr_insn_transform): Handle
14586 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14588 2017-02-21 Martin Liska <mliska@suse.cz>
14590 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14592 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14594 PR translation/79638
14595 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14597 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14600 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14601 (arm_function_ok_for_sibcall): Return false for an indirect call by
14602 descriptor if all the argument registers are used.
14603 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14604 alignment of the function.
14606 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14608 PR tree-optimization/61441
14609 * simplify-rtx.c (simplify_const_unary_operation): For
14610 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14611 the sNaN unmodified.
14613 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14615 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14616 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14617 instead of SYSTEM_HEADER_DIR.
14619 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14620 Martin Liška <mliska@suse.cz>
14622 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14623 Fix typos and grammar, use active voice, and clarify.
14625 2017-02-20 Marek Polacek <polacek@redhat.com>
14627 PR middle-end/79537
14628 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14631 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14633 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14636 * config/i386/i386.c (ix86_expand_builtin): Handle
14637 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14638 ix86_builtins_isa[fcode].isa as a requirement of those
14639 flags and any other flag in the bitmask.
14640 (ix86_init_mmx_sse_builtins): Use 0 instead of
14641 ~OPTION_MASK_ISA_64BIT as mask.
14642 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14643 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14644 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14645 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14647 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14650 * lra-constraints.c (split_reg): Check requested split mode
14651 is supported by the register.
14653 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14655 * lra-constraints.c (simplify_operand_subreg): Remove early
14658 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14661 * lra-constraints.c (curr_insn_transform): Tighten condition
14662 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14664 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14667 * lra-constraints.c (curr_insn_transform): Handle
14668 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14670 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14673 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14675 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14677 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14680 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14683 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14685 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14688 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14690 * final.c (last_columnnum, override_columnnum): New variables.
14691 (final_start_function): Set last_columnnum, pass it to begin_prologue
14692 hook and pass 0 to dwarf2out_begin_prologue.
14693 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14694 to source_line debug hook.
14695 (notice_source_line): Compute last_columnnum and for debug_column_info
14696 return true on column changes.
14697 * debug.h (struct gcc_debug_hooks): Add column argument to
14698 source_line and begin_prologue hooks.
14699 (debug_nothing_int_charstar_int_bool): Remove prototype.
14700 (debug_nothing_int_int_charstar,
14701 debug_nothing_int_int_charstar_int_bool): New prototypes.
14702 (dwarf2out_begin_prologue): Add column argument.
14703 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14704 begin_prologue hooks.
14705 (debug_nothing_int_charstar_int_bool): Remove.
14706 (debug_nothing_int_int_charstar,
14707 debug_nothing_int_int_charstar_int_bool): New functions.
14708 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14709 through to dwarf2out_source_line.
14710 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14711 (dwarf2out_source_line): Add column argument, emit it if requested.
14712 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14714 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14715 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14716 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14717 through to dwarf2out_begin_prologue.
14718 (vmsdbgout_source_line): Add column argument, pass it through to
14719 dwarf2out_source_line.
14720 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14721 dbxout_source_line caller.
14722 (dbxout_source_line): Add column argument.
14724 * common.opt (gno-column-info, gcolumn-info): New options.
14725 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14726 (check_die): Also test for multiple DW_AT_decl_column attributes.
14727 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14728 DW_AT_decl_column if requested.
14729 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14731 (gen_variable_die): Likewise.
14732 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14733 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14736 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14737 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14738 (ix86_handle_option): Handle OPT_m3dnowa.
14739 * doc/invoke.texi (-m3dnowa): Document.
14740 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14741 -m3dnowa instead of -m3dnow -march=athlon.
14744 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14745 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14747 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14750 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14751 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14752 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14753 generator for vsx_xxpermdi_<mode>_be.
14754 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14755 force big-endian semantics.
14756 (vsx_xxpermdi_<mode>_be): New define_expand with same
14757 implementation as previous version of vsx_xxpermdi_<mode>.
14759 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14761 PR tree-optimization/79327
14762 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14763 variable, its initialization and use.
14765 2017-02-17 Julia Koval <julia.koval@intel.com>
14767 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14768 (OPTION_MASK_ISA_PKU_UNSET): New.
14769 (ix86_handle_option): Handle -mrdpid.
14770 * config/i386/cpuid.h (bit_RDPID): New.
14771 * config/i386/driver-i386.c (host_detect_local_cpu):
14772 Detect RDPID feature.
14773 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14774 * config/i386/i386-c.c (ix86_target_macros_internal):
14776 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14777 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14778 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14779 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14780 * config/i386/i386.md (define_insn "rdpid"): New.
14781 * config/i386/i386.opt Add -mrdpid.
14782 * config/i386/immintrin.h (_rdpid_u32): New.
14784 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14786 PR rtl-optimization/79541
14787 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14788 instead of transforming it into USE.
14790 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14792 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14793 If HONOR_SNANS (SFmode) force the input to a register.
14794 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14795 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14796 an frsp or similar insn.
14798 2017-02-17 Martin Liska <mliska@suse.cz>
14800 PR rtl-optimization/79577
14801 * params.def (selsched-max-sched-times): Increase minimum to 1.
14803 2017-02-17 Martin Liska <mliska@suse.cz>
14805 PR rtl-optimization/79574
14806 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14808 2017-02-17 Martin Liska <mliska@suse.cz>
14810 PR tree-optimization/79529
14811 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
14812 ssa_defined_default_def_p to handle cases which are implicitly
14814 * tree-ssa.c (ssa_defined_default_def_p): New function.
14815 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
14816 which are implicitly defined.
14817 * tree-ssa.h (ssa_defined_default_def_p): Declare.
14819 2017-02-17 Richard Biener <rguenther@suse.de>
14821 PR middle-end/79576
14822 * params.def (max-ssa-name-query-depth): Limit to 10.
14824 2017-02-17 Richard Biener <rguenther@suse.de>
14826 PR tree-optimization/79552
14827 * tree-ssa-structalias.c (visit_loadstore): Properly verify
14830 2017-02-17 Richard Biener <rguenther@suse.de>
14833 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
14835 2017-02-17 Marek Polacek <polacek@redhat.com>
14837 PR middle-end/79536
14838 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
14839 (fold_negate_expr): New wrapper.
14841 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
14843 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
14844 Correct terminology and de-emphasize pre-standard behavior.
14846 2017-02-16 Alan Modra <amodra@gmail.com>
14848 PR rtl-optimization/79286
14849 * ira.c (def_dominates_uses): New function.
14850 (update_equiv_regs): Don't create an equivalence for insns that
14851 may trap where the register def does not dominate the use.
14853 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
14855 PR rtl-optimization/78127
14856 * lra.c (lra): Call lra_eliminate before finish the loop after
14859 2017-02-16 Richard Biener <rguenther@suse.de>
14861 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
14863 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
14864 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
14865 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
14866 (isl_val_int_from_wi): New function.
14867 (extract_affine_gmp): Rename to ...
14868 (extract_affine_wi): ... this, take a widest_int.
14869 (extract_affine_int): Just wrap extract_affine_wi.
14870 (add_param_constraints): Use isl_val_int_from_wi.
14871 (add_loop_constraints): Likewise, and extract_affine_wi.
14873 2017-02-15 Jeff Law <law@redhat.com>
14875 PR middle-end/79521
14876 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
14877 ira_init_register_move_cost_if_necessary.
14879 2017-02-15 Martin Sebor <msebor@redhat.com>
14881 PR middle-end/32003
14882 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
14883 removed in a prior commit.
14885 2017-02-15 Bin Cheng <bin.cheng@arm.com>
14887 PR tree-optimization/79347
14888 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
14889 counters during peeling.
14891 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
14893 * Makefile.in (site.exp): Remove "set ISLVER".
14895 2017-02-15 Jakub Jelinek <jakub@redhat.com>
14898 * real.c (real_from_integer): Call real_convert even for decimal.
14900 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
14903 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
14905 2017-02-14 Andrew Pinski <apinski@cavium.com>
14907 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
14908 cores and change the partno/implementer to be correct.
14909 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
14910 the 'B" as the implementer.
14911 * config/aarch64/aarch64-tune.md: Regenerate.
14913 2017-02-14 Carl Love <cel@us.ibm.com>
14915 * config/rs6000/rs6000.c: Add case statement entry to make the
14916 xvcvuxdsp built-in argument unsigned.
14917 * config/rs6000/vsx.md: Fix the source and return operand types so they
14918 match the instruction definitions from the ISA document. Fix typo
14919 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
14922 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
14925 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
14926 member early_clobber_alts.
14927 * lra-lives.c (reg_early_clobber_p): New.
14928 (process_bb_lives): Use it.
14929 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
14930 (debug_operand_data): Initialize early_clobber_alts.
14931 (setup_operand_alternative): Set up early_clobber_alts.
14932 (collect_non_operand_hard_regs): Ditto. Pass early clobber
14933 alternatives to new_insn_reg.
14934 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
14936 (lra_update_insn_regno_info): Pass the new arg.
14938 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14940 PR middle-end/79505
14941 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
14942 (new_oacc_loop_raw): Don't clear already cleared fields.
14945 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
14946 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
14947 _mm512_prefetch_i64gather_ps): New inline functions and macros.
14949 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
14952 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
14954 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
14957 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
14958 the extra instruction to the right place to store 128-bit constant
14961 2017-02-14 Martin Sebor <msebor@redhat.com>
14963 PR middle-end/79448
14964 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
14965 warning for strings of unknown length.
14967 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
14969 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
14971 2017-02-14 Jeff Law <law@redhat.com>
14974 * ira-costs.c (scan_one_insn): Initialize register move costs
14975 for pseudos seen in USE/CLOBBER insns.
14977 PR tree-optimization/79095
14978 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
14979 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
14980 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
14981 if the operands are known to be not equal, then the resulting range
14983 (intersect_ranges): If the new range is ~[0,0] and the old range is
14984 wide, then prefer ~[0,0].
14985 * tree-vrp.c (overflow_comparison_p_1): New function.
14986 (overflow_comparison_p): New function.
14987 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
14988 if NAME is used in an overflow test.
14989 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
14990 overflow check that can be expressed as an equality test, then adjust
14991 ops to be that equality test.
14993 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14995 * config/s390/s390-builtin-types.def: Remove flags argument.
14996 * config/s390/s390.c (s390_init_builtins): Likewise.
14998 2017-02-14 Martin Liska <mliska@suse.cz>
15000 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15001 vector. Fix trailing white spaces.
15003 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15005 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15008 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15010 PR rtl-optimization/68664
15011 * config/arm/arm.c (arm_sched_can_speculate_insn):
15012 New function. Declare prototype.
15013 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15015 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15017 PR rtl-optimization/68664
15018 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15020 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15022 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15024 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15025 max skip bytes for function, loop and jump.
15027 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15029 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15030 ABS_EXPR for gimple dump.
15032 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15035 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15037 PR tree-optimization/79408
15038 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15039 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15040 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15041 also if rhs1 is INTEGER_CST.
15043 2017-02-14 Richard Biener <rguenther@suse.de>
15045 PR middle-end/79432
15046 * tree-into-ssa.c (insert_phi_nodes): When the function can
15047 have abnormal edges rewrite SSA names with broken use-def
15048 dominance out of SSA and register them for PHI insertion.
15050 2017-02-13 Martin Sebor <msebor@redhat.com>
15052 PR middle-end/79496
15053 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15054 clearing info.nowrite flag when snprintf size argument is a range.
15056 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15058 * cprop.c (cprop_jump): Add missing space in string literal.
15059 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15060 (get_constraint_for_component_ref): Likewise.
15061 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15062 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15063 * lra-constraints.c (process_alt_operands): Likewise.
15064 * ipa-inline.c (inline_small_functions): Likewise.
15065 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15066 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15067 * trans-mem.c (diagnose_tm_1_op): Likewise.
15068 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15069 (grid_parallel_clauses_gridifiable): Likewise.
15071 * config/nvptx/mkoffload.c (process): Add space in between
15074 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15075 "MOD4_SSE_REGS" and "ALL_REGS".
15077 * spellcheck.c (test_data): Add , in between "foo" and "food".
15079 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15082 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15083 boundary crossing check and subsequent code generation agree.
15085 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15087 * config/aarch64/aarch64.c (has_memory_op): Delete.
15088 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15091 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15093 PR rtl-optimization/79388
15094 PR rtl-optimization/79450
15095 * combine.c (distribute_notes): When removing TEM_INSN for which
15096 corresponding dest has last value recorded, invalidate that last
15099 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15101 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15102 of explicit '@'. Add missing assembly comment marker on branch costs
15105 2017-02-13 Nathan Sidwell <nathan@acm.org>
15107 * gengtype-lex.l (<in_struct>): Add '/'.
15109 2017-02-13 Martin Liska <mliska@suse.cz>
15112 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15114 2017-02-13 Richard Biener <rguenther@suse.de>
15116 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15118 * configure: Re-generate.
15119 * config.in: Likewise.
15120 * graphite-dependences.c: Simplify as if
15121 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15122 * graphite-isl-ast-to-gimple.c: Likewise.
15123 * graphite-optimize-isl.c: Likewise.
15124 * graphite-poly.c: Likewise.
15125 * graphite-sese-to-poly.c: Likewise.
15126 * graphite.h: Likewise.
15127 * toplev.c: Include isl/version.h and use isl_version () for
15128 printing the ISL version.
15129 * doc/install.texi: Update ISL requirement.
15131 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15133 * doc/standards.texi (Standards): Update reference to
15136 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15138 * doc/extend.texi (Named Address Spaces): sourceware.org now
15140 * doc/install.texi (Binaries): Ditto.
15143 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15145 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15146 standard terminology "stringize"/"stringizing" throughout.
15147 * doc/cppinternals.texi: Likewise.
15149 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15151 * doc/extend.texi: Fix some spelling mistakes and typos.
15152 * doc/invoke.texi: Likewise.
15154 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15157 * params.def (inline-min-speedup) Change from 10 to 8.
15159 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15161 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15164 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15167 * ipa-inline-analysis.c (get_minimal_bb): New function.
15168 (record_modified): Use it.
15169 (remap_edge_change_prob): Handle also ancestor functions.
15171 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15173 * doc/contrib.texi (Contributors): Remove broken link into
15174 the Mauve CVS repository.
15176 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15178 PR middle-end/79454
15179 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15180 result computation whenever lhs doesn't have vector mode, not
15181 just when it has BLKmode.
15183 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15185 * doc/makefile.texi (profiledbootstrap): Refer to the
15186 installation instructions only in textual form.
15188 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15191 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15193 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15195 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15196 (Specific): Update mingw-w64 reference.
15198 (Specific): Remove broken link to Renesas RX processor.
15200 2017-02-10 Richard Biener <rguenther@suse.de>
15202 * toplev.c (process_options): Do not mention obsolete graphite
15203 options when printing sorry message about missing graphite support.
15204 Mention -floop-nest-optimize.
15206 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15208 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15209 (vtst_p16): Likewise.
15210 (vtstq_p8): Likewise.
15211 (vtstq_p16): Likewise.
15213 (vtstq_p64): Likewise.
15214 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15215 (vset_lane_p64): New.
15216 (vsetq_lane_p64): New.
15218 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15220 PR tree-optimization/79411
15221 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15222 stmt operands are SSA_NAMEs used in abnormal phis.
15223 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15226 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15229 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15232 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15234 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15236 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15238 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15242 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15243 not arbitrary TREE_CONSTANT.
15246 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15247 "omp declare target link" attribute unless is_global_var.
15248 * omp-offload.c (find_link_var_op): Likewise.
15250 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15251 Chung-Lin Tang <cltang@codesourcery.com>
15253 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15255 (gimplify_adjust_omp_clauses): Don't delete TILE.
15256 (gimplify_omp_for): Deal with TILE.
15257 * internal-fn.c (expand_GOACC_TILE): New function.
15258 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15260 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15261 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15263 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15264 avoid DIV for outermost collapse var.
15265 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15266 Remove out of date comments, fix whitespace.
15267 * omp-general.c (omp_extract_for_data): Deal with tiling.
15268 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15269 adjust OLF_DIM_BASE value.
15270 (struct omp_for_data): Add tiling field.
15271 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15272 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15273 for auto loops. Remove default auto determining, moved to
15274 oacc_loop_fixed_partitions.
15275 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15276 stmts, add e_mask field.
15277 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15278 (oacc_thread_numbers): Use oacc_dim_call.
15279 (oacc_xform_tile): New.
15280 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15281 (finish_oacc_loop): Adjust for ifns vector.
15282 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15283 add case for GOACC_TILE fns.
15284 (oacc_loop_xform_loop): Delete.
15285 (oacc_loop_process): Iterate over call list directly, and add
15286 handling for GOACC_TILE fns.
15287 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15289 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15290 vector partitioning to outer loops. Assign 2 partitions to loops
15291 when available. Add TILE handling.
15292 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15293 (execite_oacc_device_lower): Process GOACC_TILE fns,
15294 ignore unknown specs.
15295 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15296 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15297 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15299 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15301 * configure.ac (ACX_BUGURL): Update.
15302 * configure: Regenerate.
15304 2017-02-09 Richard Biener <rguenther@suse.de>
15306 PR tree-optimization/69823
15307 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15308 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15310 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15312 * config/arc/arc-c.def: Add __NPS400__ definition.
15313 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15314 (TARGET_NPS400): Define.
15316 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15318 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15320 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15321 pointer, arch_info.
15322 (arc_cpu_types): Fill the arch_info field with a pointer into the
15323 arc_arch_types table.
15324 (arc_selected_cpu): Declare.
15325 * config/arc/arc.c (arc_selected_cpu): Make global.
15326 (arc_selected_arch): Delete.
15327 (arc_base_cpu): Delete.
15328 (arc_override_options): Remove references to deleted variables,
15329 update access to arch information.
15330 (ARC_OPT): Update access to arch information.
15331 (ARC_OPTX): Likewise.
15332 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15333 (TARGET_ARC600): Update access to arch information.
15334 (TARGET_ARC601): Likewise.
15335 (TARGET_ARC700): Likewise.
15336 (TARGET_EM): Likewise.
15337 (TARGET_HS): Likewise.
15338 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15341 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15344 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15345 condition/operands for integer GE/LE/GEU/LEU operations.
15347 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15349 PR translation/79397
15350 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15353 2017-02-08 Martin Jambor <mjambor@suse.cz>
15356 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15357 whether allocation happened.
15358 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15359 nothing was allocated.
15361 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15363 PR tree-optimization/79408
15364 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15365 constant, but SSA_NAME with a known integer range, use the minimum
15366 of that range instead of op1 to determine if modulo can be replaced
15367 with its first operand.
15369 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15371 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15373 2017-02-08 Richard Biener <rguenther@suse.de>
15375 PR tree-optimization/71824
15376 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15377 Check all loops contained in the merged region.
15379 2017-02-07 Andrew Pinski <apinski@cavium.com>
15381 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15383 2017-02-07 Andrew Pinski <apinski@cavium.com>
15385 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15386 (thunderxt88): Likewise.
15387 (thunderxt81): Disable LSE and change v8.1 to v8.
15388 (thunderxt83): Likewise.
15390 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15391 Richard Biener <rguenther@suse.de>
15393 PR middle-end/79399
15394 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15395 type from int to size_t.
15396 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15398 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15400 PR rtl-optimization/79386
15401 * cprop.c (bypass_conditional_jumps): Initialize
15402 bypass_last_basic_block already before splitting bbs after
15403 unconditional traps...
15404 (bypass_conditional_jumps): ... rather than here.
15407 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15408 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15409 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15410 fix -masm=intel patterns.
15412 2017-02-07 Richard Biener <rguenther@suse.de>
15414 PR tree-optimization/79256
15415 PR middle-end/79278
15416 * builtins.c (get_object_alignment_2): Use min_align_of_type
15417 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15418 and ADJUST_FIELD_ALIGN.
15420 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15422 * doc/tm.texi: Regenerate.
15423 * stor-layout.c (layout_decl): Adjust.
15424 (update_alignment_for_field): Likewise.
15425 (place_field): Likewise.
15426 (min_align_of_type): Likewise.
15427 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15428 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15429 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15430 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15431 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15432 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15433 * config/i386/i386.c (x86_field_alignment): Likewise.
15434 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15435 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15436 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15437 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15438 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15439 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15443 2017-01-30 Richard Biener <rguenther@suse.de>
15445 PR tree-optimization/79256
15446 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15447 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15450 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15452 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15453 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15454 builtins to SImode and emit a zero-extend, if necessary.
15456 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15458 * docs/invoke.texi (RISC-V Options): Alphabetize.
15460 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15462 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15465 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15467 * config/riscv/riscv.c: New file.
15468 * common/config/riscv/riscv-common.c: Likewise.
15469 * config.gcc: Likewise.
15470 * config/riscv/constraints.md: Likewise.
15471 * config/riscv/elf.h: Likewise.
15472 * config/riscv/generic.md: Likewise.
15473 * config/riscv/linux.h: Likewise.
15474 * config/riscv/multilib-generator: Likewise.
15475 * config/riscv/peephole.md: Likewise.
15476 * config/riscv/pic.md: Likewise.
15477 * config/riscv/predicates.md: Likewise.
15478 * config/riscv/riscv-builtins.c: Likewise.
15479 * config/riscv/riscv-c.c: Likewise.
15480 * config/riscv/riscv-ftypes.def: Likewise.
15481 * config/riscv/riscv-modes.def: Likewise.
15482 * config/riscv/riscv-opts.h: Likewise.
15483 * config/riscv/riscv-protos.h: Likewise.
15484 * config/riscv/riscv.h: Likewise.
15485 * config/riscv/riscv.md: Likewise.
15486 * config/riscv/riscv.opt: Likewise.
15487 * config/riscv/sync.md: Likewise.
15488 * config/riscv/t-elf-multilib: Likewise.
15489 * config/riscv/t-linux: Likewise.
15490 * config/riscv/t-linux-multilib: Likewise.
15491 * config/riscv/t-riscv: Likewise.
15492 * configure.ac: Likewise.
15493 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15494 Waterman as RISC-V maintainers.
15495 * doc/install.texi: Add RISC-V entries.
15496 * doc/invoke.texi: Add RISC-V options section.
15497 * doc/md.texi: Add RISC-V constraints section.
15498 * configure: Regenerated.
15500 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15503 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15504 false values to be constant vectors with all 0 or all 1 bits set.
15505 (vcondu<mode><mode>): Likewise.
15506 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15508 (fpmask_comparison_operator): Update comment.
15509 (vecint_comparison_operator): New predicate.
15510 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15511 vector conditionals when the true and false values are constant
15512 vectors with all 0 bits or all 1 bits set.
15514 2017-02-06 Martin Sebor <msebor@redhat.com>
15516 PR tree-optimization/79376
15517 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15519 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15521 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15522 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15523 to simplify split condition.
15525 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15527 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15528 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15531 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15533 PR rtl-optimization/68664
15534 * target.def (can_speculate_insn): New hook.
15535 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15536 * doc/tm.texi: Regenerate.
15537 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15538 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15539 (rs6000_sched_can_speculate_insn): New function.
15541 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15543 PR tree-optimization/79284
15544 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15545 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15546 vectorizable_mask_load_store, vectorizable_operation,
15547 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15548 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15549 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15550 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15551 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15552 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15553 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15554 is_gimple_assign (stmt). Replace another such test with
15555 is_gimple_assign (stmt).
15557 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15560 * config/avr/avr.c (rtl-iter.h): Include it.
15561 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15562 (avr_legitimate_combined_insn): ...and implementation.
15564 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15566 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15567 * config/s390/s390.c (s390_const_operand_ok)
15568 (s390_canonicalize_comparison, s390_extract_part)
15569 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15570 (s390_contiguous_bitmask_p, s390_rtx_costs)
15571 (legitimize_pic_address): Likewise.
15572 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15573 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15574 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15575 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15576 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15578 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15580 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15581 REGNO($0) == REGNO($1).
15583 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15585 * config/s390/linux.h(SIZE_TYPE): Add comment.
15587 2017-02-06 Julian Brown <julian@codesourcery.com>
15588 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15589 Virendra Pathak <virendra.pathak@broadcom.com>
15591 * config/aarch64/aarch64-cores.def: Change the scheduler
15593 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15594 * config/aarch64/thunderx2t99.md: New file.
15596 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15598 * doc/standards.texi (Go Language): Update link to language
15601 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15603 * tree-eh.c (lower_resx): Sanitize profile.
15604 (cleanup_empty_eh_move_lp): Likewise.
15606 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15609 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15611 * cfgloopmanip.h (loop_version): Update prototype.
15612 * modulo-sched.c (sms_schedule): Update call of loop_version.
15613 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15614 * tree-parloops.c (gen_parallel_loop): Likewise.
15615 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15616 * tree-ssa-loop-split.c (split_loop): Likewise.
15617 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15618 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15620 2017-02-05 Martin Liska <mliska@suse.cz>
15623 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15625 (print_operand_address): Initialize a struct to zero.
15627 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15629 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15630 garbage collector only in textual form.
15632 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15634 * doc/extend.texi (x86 specific memory model extensions for
15635 transactional memory): Simplify a phrase.
15637 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15640 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15641 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15642 (atomic_storedi_1): Likewise.
15644 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15646 PR tree-optimization/79338
15647 * tree-parloops.c (gather_scalar_reductions): Don't call
15648 vect_analyze_loop_form for loop->inner before destroying loop's
15651 2017-02-03 Martin Sebor <msebor@redhat.com>
15653 PR tree-optimization/79327
15654 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15655 when precision has resulted in leading zeros.
15656 (format_integer): Adjust the likely counter to assume an unknown
15657 argument that may be zero is non-zero.
15659 2017-02-03 Jason Merrill <jason@redhat.com>
15662 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15663 avoid copying non-taken branch.
15665 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15667 PR tree-optimization/79340
15668 * tree-vect-loop.c (vectorizable_reduction): Release
15669 vec_defs elements after safe_splicing them into other vectors.
15672 PR tree-optimization/79327
15673 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15674 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15676 (format_integer): Use wide_int_to_tree instead of build_int_cst
15677 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15678 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15679 of shortest and longest sequence.
15681 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15683 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15684 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15686 2017-02-03 Walter Lee <walt@tilera.com>
15689 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15690 after initial stackframe link reg save.
15691 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15693 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15696 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15697 wu for stxssp alternative.
15699 2017-02-03 Martin Sebor <msebor@redhat.com>
15701 PR tree-optimization/79352
15702 * gimple-fold.c (get_range_strlen): Add argument.
15703 (get_range_strlen): Change return type to bool.
15704 (get_maxval_strlen): Pass in a dummy argument.
15705 * gimple-fold.h (get_range_strlen): Change return type to bool.
15706 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15707 * tree.h (array_at_struct_end_p): Add argument.
15708 * tree.c (array_at_struct_end_p): Handle it.
15710 2017-02-03 Martin Liska <mliska@suse.cz>
15713 * multiple_target.c (create_dispatcher_calls): Redirect edge
15714 from a caller of a dispatcher.
15715 (expand_target_clones): Make the clones local.
15716 (ipa_target_clone): Do both target clones and resolvers.
15717 (ipa_dispatcher_calls): Remove the pass.
15718 (pass_dispatcher_calls::gate): Likewise.
15719 (make_pass_dispatcher_calls): Likewise.
15720 * passes.def (pass_target_clone): Put as very first IPA early
15723 2017-02-03 Martin Liska <mliska@suse.cz>
15725 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15726 in case of a function with ifunc attribute.
15728 2017-02-03 Martin Liska <mliska@suse.cz>
15730 * cgraph.c (cgraph_node::dump): Dump function version info.
15731 * symtab.c (symtab_node::dump_base): Add missing new line.
15733 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15735 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15736 (ifcombine_ifandif): Use it.
15738 2017-02-03 Martin Liska <mliska@suse.cz>
15740 * doc/invoke.texi: Document default value for
15741 use-after-scope-direct-emission-threshold.
15743 2017-02-03 Martin Liska <mliska@suse.cz>
15745 PR tree-optimization/79339
15746 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15747 (format_floating): Likewise.
15749 2017-02-03 Martin Liska <mliska@suse.cz>
15752 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15754 (ipa_node_params_t::remove): Likewise.
15755 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15756 initialization from removed ipa_node_params_t::insert.
15757 (ipa_node_params::~ipa_node_params): Move from removed
15758 ipa_node_params_t::release.
15759 * symbol-summary.h (symbol_summary::m_released): New member.
15760 Do not release a summary twice. Do not allow to call finalizer
15761 for types of a summary that live in GGC memory.
15763 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15765 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15768 2017-02-02 Martin Sebor <msebor@redhat.com>
15770 PR middle-end/79275
15771 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15772 (format_string): Tighten up the range of output for non-constant
15773 strings and correct the expected range for wide non-constant strings.
15775 2017-02-02 Martin Sebor <msebor@redhat.com>
15777 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15779 PR middle-end/32003
15780 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15782 (-fdump-tree-@var): Add to index and document how to come up
15783 with pass-specific option and dump file names.
15784 (-fdump-passes): Clarify where to look for output.
15786 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15788 PR middle-end/77445
15789 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15790 statistics of the analyzed path; allow threading for speed when
15791 any of BBs along the path are optimized for speed.
15793 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15795 PR middle-end/78468
15796 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15797 settings of the virtual registers.
15800 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15802 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15803 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15806 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15808 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15809 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15811 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15813 * config/s390/s390.md: Add missing comments with the expanded
15815 * config/s390/vector.md: Likewise.
15816 * config/s390/vx-builtins.md: Likewise.
15818 2017-02-02 Jakub Jelinek <jakub@redhat.com>
15821 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
15822 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
15823 conditions on a single line.
15825 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15827 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15828 __S390_VX__ to __VX__.
15830 2017-02-01 Andrew Pinski <apinski@cavium.com>
15832 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
15833 stmt_info to record_stmt_cost.
15834 (vect_get_known_peeling_cost): Pass stmt_info if known to
15836 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
15837 cpu_vector_cost field into
15838 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
15839 field into vec_int_stmt_cost and vec_fp_stmt_cost.
15840 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
15841 splitting of scalar_stmt_cost and vec_stmt_cost.
15842 (thunderx_vector_cost): Likewise.
15843 (cortexa57_vector_cost): LIkewise.
15844 (exynosm1_vector_cost): Likewise.
15845 (xgene1_vector_cost): Likewise.
15846 (thunderx2t99_vector_cost): Improve after the splitting of the two
15848 (aarch64_builtin_vectorization_cost): Update for the splitting of
15849 scalar_stmt_cost and vec_stmt_cost.
15851 2017-02-01 Torvald Riegel <triegel@redhat.com>
15852 Richard Henderson <rth@redhat.com>
15854 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
15855 conditional on existance of a fast atomic load.
15856 * optabs-query.c (can_atomic_load_p): New function.
15857 * optabs-query.h (can_atomic_load_p): Declare it.
15858 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
15859 no fast atomic load is available for the particular size of access.
15860 (expand_atomic_compare_and_swap): Likewise.
15861 (expand_atomic_load): Likewise.
15862 (expand_atomic_store): Likewise.
15863 (expand_atomic_fetch_op): Likewise.
15864 * testsuite/lib/target-supports.exp
15865 (check_effective_target_sync_int_128): Remove x86 because it provides
15866 no fast atomic load.
15867 (check_effective_target_sync_int_128_runtime): Likewise.
15869 2017-02-01 Richard Biener <rguenther@suse.de>
15871 * graphite.c: Include tree-vectorizer.h for find_loop_location.
15872 (graphite_transform_loops): Provide opt-info for optimized nests.
15873 * tree-parloop.c (parallelize_loops): Provide opt-info for
15874 parallelized loops.
15876 2017-02-01 Richard Biener <rguenther@suse.de>
15878 PR middle-end/79315
15879 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
15880 was not set before.
15882 2017-02-01 Richard Biener <rguenther@suse.de>
15884 PR tree-optimization/71824
15885 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15886 Verify the loops are valid in the merged SESE region.
15887 (scop_detection::can_represent_loop_1): Check analyzing the
15888 evolution of the number of iterations in the region succeeds.
15890 2017-01-31 Ian Lance Taylor <iant@golang.org>
15892 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
15893 REG_ARGS_SIZE note to 32-bit push insns and call insn.
15895 2017-01-31 David Malcolm <dmalcolm@redhat.com>
15897 PR preprocessor/79210
15898 * input.c (get_substring_ranges_for_loc): Replace line_width
15899 assertion with error-handling.
15901 2017-01-31 Richard Biener <rguenther@suse.de>
15903 PR tree-optimization/77318
15904 * graphite-sese-to-poly.c (extract_affine): Fix assert.
15905 (create_pw_aff_from_tree): Take loop parameter.
15906 (add_condition_to_pbb): Pass loop of the condition to
15907 create_pw_aff_from_tree.
15909 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15911 * config/s390/s390.c (s390_asan_shadow_offset): New function.
15912 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
15914 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
15918 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
15920 (convert_int_to_float128): Likewise.
15921 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
15922 (convert_int_to_float128): Likewise.
15923 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
15924 (UNSPEC_IEEE128_CONVERT): Likewise.
15925 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
15926 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
15927 Use local variables for IBM extended format.
15928 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
15929 (fix_trunc<mode>si2_fprs): Likewise.
15930 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
15931 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
15932 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
15933 to know that we can now have integers of all sizes in vector
15935 (fix<uns>_<mode>di2_hw): Likewise.
15936 (float<uns>_<mode>si2_hw): Likewise.
15937 (fix_<mode>si2_hw): Likewise.
15938 (fixuns_<mode>si2_hw): Likewise.
15939 (float<uns>_<mode>di2_hw): Likewise.
15940 (float_<mode>di2_hw): Likewise.
15941 (float_<mode>si2_hw): Likewise.
15942 (floatuns_<mode>di2_hw): Likewise.
15943 (floatuns_<mode>si2_hw): Likewise.
15944 (xscvqp<su>wz_<mode>): Delete, no longer used.
15945 (xscvqp<su>dz_<mode>): Likewise.
15946 (xscv<su>dqp_<mode>): Likewise.
15947 (ieee128_mfvsrd_64bit): Likewise.
15948 (ieee128_mfvsrd_32bit): Likewise.
15949 (ieee128_mfvsrwz): Likewise.
15950 (ieee128_mtvsrw): Likewise.
15951 (ieee128_mtvsrd_64bit): Likewise.
15952 (ieee128_mtvsrd_32bit): Likewise.
15954 2017-01-31 Martin Liska <mliska@suse.cz>
15957 * ipa-prop.c (ipa_free_all_node_params): Call release method
15958 instead of ~sumbol_summary to not to trigger double times
15961 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
15963 PR tree-optimization/71691
15964 * bitmap.h (class auto_bitmap): New.
15965 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
15966 is_maybe_undefined instead of ssa_undefined_value_p.
15968 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15970 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15971 __S390_ARCH_LEVEL__ to __ARCH__.
15973 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15975 PR tree-optimization/79267
15976 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
15977 if should_remove_lhs_p is true.
15979 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
15982 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
15983 (add_alignment_attribute): New.
15984 (base_type_die): Add alignment attribute.
15985 (subrange_type_die): Likewise.
15986 (modified_type_die): Likewise.
15987 (gen_array_type_die): Likewise.
15988 (gen_descr_array_type_die: Likewise.
15989 (gen_enumeration_type_die): Likewise.
15990 (gen_subprogram_die): Likewise.
15991 (gen_variable_die): Likewise.
15992 (gen_field_die): Likewise.
15993 (gen_ptr_to_mbr_type_die): Likewise.
15994 (gen_struct_or_union_type_die): Likewise.
15995 (gen_subroutine_type_die): Likewise.
15996 (gen_typedef_die): Likewise.
15997 (base_type_cmp): Compare alignment attribute.
15999 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16002 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16003 (setb_unsigned) New pattern for setb with CCUNS.
16004 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16005 subfc./subfe sequence to avoid overflow problems. Generate a
16006 shorter sequence with cmpld/setb for power9.
16007 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16008 for generating subfc. instruction.
16009 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16010 now uses this instruction.
16012 2017-01-30 Ian Lance Taylor <iant@google.com>
16015 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16016 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16018 2017-01-30 Martin Sebor <msebor@redhat.com>
16020 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16021 Move constant to the right of a relational operator.
16022 (get_mpfr_format_length, format_character, format_string): Ditto.
16023 (should_warn_p, maybe_warn): Same.
16025 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16027 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16030 * asan.c (get_translation_unit_decl): Remove function.
16031 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16033 2017-01-30 Martin Liska <mliska@suse.cz>
16035 PR gcov-profile/79259
16036 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16037 -fprofile-generate.
16039 2017-01-30 Martin Liska <mliska@suse.cz>
16042 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16043 Initialize variables with NULL value.
16045 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16048 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16050 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16052 2017-01-30 Richard Biener <rguenther@suse.de>
16054 PR tree-optimization/79276
16055 * tree-vrp.c (process_assert_insertions): Properly adjust common
16056 when removing a duplicate.
16058 * gcc.dg/torture/pr79276.c: New testcase.
16060 2017-01-30 Richard Biener <rguenther@suse.de>
16062 PR tree-optimization/79256
16063 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16064 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16066 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16068 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16071 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16072 ("*r<noxa>sbg_<mode>_sll_bitmask")
16073 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16074 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16075 Use contiguous_bitmask_nowrap_operand.
16077 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16080 * config/rs6000/altivec.h (vec_xl): Revise #define.
16081 (vec_xst): Likewise.
16083 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16085 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16087 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16089 PR rtl-optimization/79194
16090 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16091 traps before call to bypass_conditional_jumps.
16093 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16095 PR tree-optimization/71374
16096 * lra-constraints.c (check_conflict_input_operands): New.
16097 (match_reload): Use it.
16099 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16102 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16103 account to calculate conflict_set.
16105 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16107 PR rtl-optimization/78559
16108 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16109 other_insn in combine.
16111 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16113 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16114 uint16_type_node for BT_UINT16.
16116 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16118 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16119 "RTL Tests" to menu.
16120 (GIMPLE Tests): New node.
16121 (RTL Tests): New node.
16123 2017-01-27 Richard Biener <rguenther@suse.de>
16125 PR tree-optimization/79245
16126 * tree-loop-distribution.c (distribute_loop): Apply cost
16127 modeling also to detected patterns.
16129 2017-01-27 Richard Biener <rguenther@suse.de>
16131 PR tree-optimization/71433
16132 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16133 (compare_assert_loc): New function.
16134 (process_assert_insertions): Sort and optimize assert locations
16135 to remove duplicates and push down identical assertions on
16136 edges to their destination block.
16138 2017-01-27 Richard Biener <rguenther@suse.de>
16140 PR tree-optimization/79244
16141 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16142 out SSA names even if abnormal.
16144 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16146 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16147 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16148 instead of MPFR_RNDN.
16150 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16153 * arm.c (arm_option_override): Don't call build_target_option_node
16154 until after doing all option overrides.
16155 (arm_valid_target_attribute_tree): Likewise.
16157 2017-01-27 Martin Liska <mliska@suse.cz>
16159 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16160 for {cd}tors and C++ {cd}tors.
16162 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16164 * config/s390/s390.md ("*setmem_long_and")
16165 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16167 2017-01-26 Martin Sebor <msebor@redhat.com>
16169 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16172 2017-01-26 Martin Sebor <msebor@redhat.com>
16174 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16175 HAVE_DFmode before using XFmode or DFmode.
16176 (parse_directive): Avoid using the z length modifier to avoid
16177 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16179 PR middle-end/78703
16180 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16181 to accept adjustment as an array.
16182 (get_int_range): New function.
16183 (struct directive): Make width and prec arrays.
16184 (directive::set_width, directive::set_precision): Call get_int_range.
16185 (format_integer, format_floating): Handle width and precision ranges.
16186 (format_string, parse_directive): Same.
16188 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16191 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16192 comdat_type_p set, just clone them, but keep the children in the
16196 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16197 which have direct callers with -fvar-tracking-assignments enabled
16199 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16200 inside of type units.
16202 2017-01-26 Martin Sebor <msebor@redhat.com>
16204 PR middle-end/78703
16205 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16207 (struct format_result): Replace number_chars, number_chars_min,
16208 and number_chars_max with a single member of struct result_range.
16210 (format_result::operator+=): Adjust.
16211 (struct fmtresult): Remove bounded. Handle likely and unlikely
16213 (fmtresult::adjust_for_width_or_precision): New function.
16214 (fmtresult:type_max_digits): New function.
16215 (bytes_remaining): Handle likely and unlikely counters.
16216 (min_bytes_remaining): Remove.
16217 (format_percent): Simplify.
16218 (format_integer, format_floating): Set likely and unlikely counters.
16219 (get_string_length, format_character, format_string): Same.
16220 (format_plain, should_warn_p): New function.
16221 (maybe_warn): Call should_warn_p. Update diagnostic messages
16222 and handle those for all directives, including plain strings.
16223 (format_directive): Handle likely and unlikely counters.
16224 Remove unnecessary quoting from diagnostics. Add an informational
16226 (add_bytes): Remove.
16227 (pass_sprintf_length::compute_format_length): Simplify.
16228 (try_substitute_return_value): Handle likely and unlikely counters.
16230 2017-01-26 Carl Love <cel@us.ibm.com>
16232 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16233 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16235 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16238 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16239 endianess for subregs into account.
16240 * lra-constraints.c (lra_constraints): Do risky transformations
16241 always on the first iteration.
16242 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16243 last_call_used_reg_set.
16244 (process_bb_lives): Define and use last_call_used_reg_set.
16245 * lra.c (lra): Always continue after lra_constraints on the first
16248 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16250 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16252 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16254 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16256 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16257 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16258 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16259 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16260 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16261 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16262 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16263 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16264 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16266 2017-01-26 Marek Polacek <polacek@redhat.com>
16269 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16270 for the third operand.
16272 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16274 PR middle-end/79236
16275 * omp-low.c (struct omp_context): Add simt_stmt field.
16276 (scan_omp_for): Return omp_context *.
16277 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16278 context to the _simt_ SIMD stmt.
16279 (lower_omp_for): For combined SIMD with sibling _simt_
16280 SIMD, make sure to use the same decls in _looptemp_
16281 clauses as in the sibling.
16283 2017-01-26 David Sherwood <david.sherwood@arm.com>
16285 PR middle-end/79212
16286 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16289 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16292 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16293 emit fld b; fld a; if possible.
16295 * brig-builtins.def: Update copyright years.
16296 * config/arm/arm_acle_builtins.def: Update copyright years.
16298 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16301 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16302 constraint instead of o for the stxsd instruction.
16304 2017-01-25 Carl Love <cel@us.ibm.com>
16306 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16307 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16309 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16311 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16313 2017-01-25 Richard Biener <rguenther@suse.de>
16315 PR tree-optimization/69264
16316 * target.def (vector_alignment_reachable): Improve documentation.
16317 * doc/tm.texi: Regenerate.
16318 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16320 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16321 earlier changes with respect to TYPE_USER_ALIGN.
16322 (vector_alignment_reachable_p): Likewise. Improve dumping.
16324 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16327 * config/arm/arm.md (xordi3): Force constant operand into a register
16330 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16332 * doc/invoke.texi (-fstore-merging): Correct default optimization
16333 levels at which it is enabled.
16334 (-O): Move -fstore-merging from list to...
16337 2017-01-25 Richard Biener <rguenther@suse.de>
16340 * omp-expand.c: Include debug.h.
16341 (expand_omp_taskreg): Make sure to generate early debug before
16342 outlining anything from a function.
16343 (expand_omp_target): Likewise.
16344 (grid_expand_target_grid_body): Likewise.
16346 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16349 * asan.c (get_translation_unit_decl): New function.
16350 (asan_add_global): Extract modules file name from globals
16351 TRANSLATION_UNIT_DECL name.
16353 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16356 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16357 for long calls with APCS frame and VFP.
16359 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16361 * cfg.c (original_copy_tables_initialized_p): New function.
16362 * cfg.h (original_copy_tables_initialized_p): New decl.
16363 * cfgrtl.c (relink_block_chain): Guard the call to
16364 free_original_copy_tables with a call to
16365 original_copy_tables_initialized_p.
16366 * cgraph.h (symtab_node::native_rtl_p): New decl.
16367 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16368 (symtab_node::needed_p): Don't assert for early assembly output
16369 for __RTL functions.
16370 (cgraph_node::finalize_function): Set "force_output" for __RTL
16372 (cgraph_node::analyze): Bail out early for __RTL functions.
16373 (analyze_functions): Update assertion to support __RTL functions.
16374 (cgraph_node::expand): Bail out early for __RTL functions.
16375 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16377 * function.h (struct function): Update comment for field
16379 * gimple-expr.c: Include "tree-pass.h".
16380 (gimple_has_body_p): Return false for __RTL functions.
16381 * Makefile.in (OBJS): Add run-rtl-passes.o.
16382 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16384 (gcc::pass_manager::get_clean_slate): New accessor.
16385 * passes.c: Include "insn-addr.h".
16386 (should_skip_pass_p): Add logging. Update logic for running
16387 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16388 property-provider override so it is only done for gimple passes.
16390 (skip_pass): New function.
16391 (execute_one_pass): Call skip_pass when skipping passes.
16392 * read-md.c (md_reader::read_char): Support filtering
16393 the input to a subset of line numbers.
16394 (md_reader::md_reader): Initialize fields
16395 m_first_line and m_last_line.
16396 (md_reader::read_file_fragment): New function.
16397 * read-md.h (md_reader::read_file_fragment): New decl.
16398 (md_reader::m_first_line): New field.
16399 (md_reader::m_last_line): New field.
16400 * read-rtl-function.c (function_reader::create_function): Only
16401 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16402 curr_properties. Set DECL_INITIAL to a dummy block.
16403 (read_rtl_function_body_from_file_range): New function.
16404 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16406 * run-rtl-passes.c: New file.
16407 * run-rtl-passes.h: New file.
16409 2017-01-24 Jeff Law <law@redhat.com>
16411 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16414 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16416 PR tree-optimization/79159
16417 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16418 (record_nonwrapping_iv): Improve boundary using above function if no
16419 value range information.
16421 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16422 Martin Jambor <mjambor@suse.cz>
16424 * brig-builtins.def: New file.
16425 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16426 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16427 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16428 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16429 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16430 * builtin-types.def (BT_INT8): New.
16431 (BT_INT16): Likewise.
16432 (BT_UINT8): Likewise.
16433 (BT_UINT16): Likewise.
16434 (BT_FN_ULONG): Likewise.
16435 (BT_FN_UINT_INT): Likewise.
16436 (BT_FN_UINT_ULONG): Likewise.
16437 (BT_FN_UINT_LONG): Likewise.
16438 (BT_FN_UINT_PTR): Likewise.
16439 (BT_FN_ULONG_PTR): Likewise.
16440 (BT_FN_INT8_FLOAT): Likewise.
16441 (BT_FN_INT16_FLOAT): Likewise.
16442 (BT_FN_UINT32_FLOAT): Likewise.
16443 (BT_FN_UINT16_FLOAT): Likewise.
16444 (BT_FN_UINT8_FLOAT): Likewise.
16445 (BT_FN_UINT64_FLOAT): Likewise.
16446 (BT_FN_UINT16_UINT32): Likewise.
16447 (BT_FN_UINT32_UINT16): Likewise.
16448 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16449 (BT_FN_INT_PTR_INT): Likewise.
16450 (BT_FN_UINT_PTR_UINT): Likewise.
16451 (BT_FN_LONG_PTR_LONG): Likewise.
16452 (BT_FN_ULONG_PTR_ULONG): Likewise.
16453 (BT_FN_VOID_UINT64_UINT64): Likewise.
16454 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16455 (BT_FN_INT8_INT8_INT8): Likewise.
16456 (BT_FN_INT16_INT16_INT16): Likewise.
16457 (BT_FN_INT_INT_INT): Likewise.
16458 (BT_FN_UINT_FLOAT_UINT): Likewise.
16459 (BT_FN_FLOAT_UINT_UINT): Likewise.
16460 (BT_FN_ULONG_UINT_UINT): Likewise.
16461 (BT_FN_ULONG_UINT_PTR): Likewise.
16462 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16463 (BT_FN_UINT_UINT_UINT): Likewise.
16464 (BT_FN_VOID_UINT_PTR): Likewise.
16465 (BT_FN_UINT_UINT_PTR: Likewise.
16466 (BT_FN_UINT32_UINT64_PTR): Likewise.
16467 (BT_FN_INT_INT_UINT_UINT): Likewise.
16468 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16469 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16470 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16471 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16472 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16473 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16474 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16475 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16476 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16477 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16478 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16479 * doc/frontends.texi: List BRIG FE.
16480 * doc/install.texi (Testing): Add BRIG tesring requirements.
16481 * doc/invoke.texi (Overall Options): Mention BRIG.
16482 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16484 2017-01-24 Richard Biener <rguenther@suse.de>
16486 PR translation/79208
16487 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16489 2017-01-24 Martin Jambor <mjambor@suse.cz>
16492 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16493 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16494 and known_contexts.
16496 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16498 PR middle-end/79123
16499 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16500 casts from signed to unsigned really don't have a range.
16502 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16504 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16505 GMP_RNDx for compatiblity.
16507 2017-01-24 Martin Liska <mliska@suse.cz>
16510 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16511 that would prevent us to call alloca with -1 as argument.
16513 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16515 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16516 Avoid -Wformat-security warning.
16518 2017-01-23 Andrew Pinski <apinski@cavium.com>
16520 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16523 2017-01-23 Martin Sebor <msebor@redhat.com>
16525 PR middle-end/78703
16526 * gimple-ssa-sprintf.c (warn_level): New global.
16527 (format_integer): Use it here and throughout the rest of the file.
16528 Use the same switch to compute sign as base.
16529 (maybe_warn): New function.
16530 (format_directive): Factor out warnings into maybe_warn.
16531 Add debugging output. Use warn_level.
16532 (add_bytes): Use warn_level.
16533 (pass_sprintf_length::compute_format_length): Add debugging output.
16534 (try_substitute_return_value): Same.
16535 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16537 PR middle-end/78703
16538 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16539 (struct fmtresult, format_integer, format_floating): Adjust.
16540 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16541 (get_string_length, format_string,format_directive): Same.
16542 (pass_sprintf_length::compute_format_length): Same.
16543 (try_substitute_return_value): Simplify slightly.
16545 PR middle-end/78703
16546 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16547 (fmtresult::operator+=): Outlined.
16548 (struct fmtresult): Add ctors.
16549 (struct conversion_spec): Rename...
16550 (struct directive): ...to this. Add and remove data members.
16551 (directive::set_width, directive::set_precision): New functions.
16552 (format_percent): Use fmtresult ctor.
16553 (get_width_and_precision): Remove.
16554 (format_integer): Make naming changes. Avoid computing width and
16556 (format_floating): Same. Adjust indentation.
16557 (format_character, format_none): New functions.
16558 (format_string): Moved character handling to format_character.
16559 (format_directive): Remove arguments, change return type.
16560 (parse_directive): New function.
16561 (pass_sprintf_length::compute_format_length): Move directive
16562 parsing to parse_directive.
16564 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16566 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16567 (assign_assembler_name_if_needed): ... this.
16568 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16569 (assign_assembler_name_if_needed): ... this.
16570 (free_lang_data_in_cgraph): Adjust callers.
16571 * cgraphunit.c (cgraph_node::analyze): Likewise.
16572 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16575 2017-01-23 Richard Biener <rguenther@suse.de>
16577 PR tree-optimization/79088
16578 PR tree-optimization/79188
16579 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16580 resetting loop bounds after last path deletion. Reset loop
16581 bounds of the target loop, make code match the comments.
16582 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16583 Make sure loops need no fixups.
16585 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16587 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16588 exponent support with double type for first argument.
16589 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16590 type returned by __builtin_vec_extract_sig,
16591 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16592 functions from "vector int" to "vector unsigned int" or from
16593 "vector long long int" to "vector unsigned long long int".
16594 Changed type returned by __builtin_vec_extract_exp,
16595 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16596 functions from "vector int" to "vector unsigned int" or from
16597 "vector long long int" to "vector unsigned long long int".
16598 Changed return type of __builtin_vec_test_data_class,
16599 __builtin_vec_test_data_class_sp, and
16600 __builtin_vec_test_data_class_dp from "vector int" to
16601 "vector bool int" or from "vector long long int" to "vector bool
16602 long long int" and changed second argument type from "unsigned
16603 int" to "int". Added new overloaded function forms "vector float
16604 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16605 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16606 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16607 double, vector unsigned long long int)" and "vector double
16608 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16609 long int)". Changed return type of
16610 __builtin_scalar_test_data_class and
16611 __builtin_scalar_test_data_class_sp and
16612 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16613 int" and changed second argument from "unsigned int" to "int".
16614 Changed type returned by __builtin_scalar_test_neg,
16615 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16616 from "int" to "bool int". Added new overloaded function form
16617 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16618 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16619 exponent double-precision with floating point first argument.
16620 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16621 documentation of scalar_test_data_class, scalar_test_neg,
16622 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16623 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16624 vec_test_data_class built-in functions to reflect refinements in
16625 their type signatures.
16627 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16629 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16631 (aarch64_elf_asm_destructor): Likewise.
16633 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16635 PR rtl-optimization/78634
16636 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16637 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16638 * ifcvt.c (noce_try_cmove): Add missing cost check.
16640 PR rtl-optimization/71724
16641 * combine.c (if_then_else_cond): Look for situations where it is
16642 beneficial to undo the work of one of the recursive calls.
16644 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16646 PR tree-optimization/70754
16647 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16648 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16649 combined stmt before it if not NULL.
16650 (combine_chains): Process refs reversely and compute dominance point
16653 2017-01-23 Martin Liska <mliska@suse.cz>
16655 PR tree-optimization/79196
16656 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16657 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16659 (strlen_optimize_stmt): Call the renamed function.
16661 2017-01-23 Michael Matz <matz@suse.de>
16663 PR tree-optimization/78384
16664 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16666 2017-01-23 Richard Biener <rguenther@suse.de>
16668 PR tree-optimization/79186
16669 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16670 both incoming edges before moving an assert.
16672 2017-01-23 Martin Jambor <mjambor@suse.cz>
16674 * ipa-prop.c (load_from_param_1): Removed.
16675 (load_from_unmodified_param): Bits from load_from_param_1 put back
16677 (load_from_param): Removed.
16678 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16679 with stmt. Reverted back to use of load_from_unmodified_param.
16681 2017-01-23 Martin Jambor <mjambor@suse.cz>
16684 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16685 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16686 field a pointer to garbage collected vector, mark lattices and
16687 ipcp_orig_node with GTY((skip)).
16688 (ipa_get_param_count): Adjust to descriptors being a pointer.
16689 (ipa_get_param): Likewise.
16690 (ipa_get_type): Likewise.
16691 (ipa_get_param_move_cost): Likewise.
16692 (ipa_set_param_used): Likewise.
16693 (ipa_get_controlled_uses): Likewise.
16694 (ipa_set_controlled_uses): Likewise.
16695 (ipa_is_param_used): Likewise.
16696 (ipa_node_params_t): Move into garbage collector. New methods insert
16698 (ipa_node_params_sum): Annotate wth GTY(()).
16699 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16701 (ipa_load_from_parm_agg): Adjust declaration.
16702 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16703 * ipa-profile.c (ipa_profile): Likewise.
16704 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16705 (ipa_populate_param_decls): Make descriptors parameter garbage
16707 (ipa_dump_param): Adjust to descriptors being a pointer.
16708 (ipa_alloc_node_params): Likewise.
16709 (ipa_initialize_node_params): Likewise.
16710 (load_from_param_1): Make descriptors parameter garbage collected.
16711 (load_from_unmodified_param): Likewise.
16712 (load_from_param): Likewise.
16713 (ipa_load_from_parm_agg): Likewise.
16714 (ipa_node_params::~ipa_node_params): Removed.
16715 (ipa_free_all_node_params): Remove call to delete operator.
16716 (ipa_node_params_t::insert): New.
16717 (ipa_node_params_t::remove): Likewise.
16718 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16719 copy known_csts and known_contexts vectors.
16720 (ipa_read_node_info): Adjust to descriptors being a pointer.
16721 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16723 (ipcp_transform_function): Make descriptors variable garbage
16726 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16728 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16729 * config/i386/avx512dqintrin.h: Ditto.
16730 * config/i386/avx512fintrin.h: Ditto.
16731 * config/i386/i386.c: Handle new builtins.
16732 * config/i386/i386-builtin.def: Add new builtins.
16733 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16734 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16736 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16737 Martin Liska <mliska@suse.cz>
16739 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16740 * asan.c (asan_expand_poison_ifn): Support stores and use
16741 appropriate ASAN report function.
16742 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16743 * internal-fn.def (ASAN_POISON_USE): Declare.
16744 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16745 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16746 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16747 ASAN_POISON calls w/o LHS.
16748 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16749 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16750 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16751 * gimplify.c (asan_poison_variables): Add attribute
16752 use_after_scope_memory to variables that really needs to live
16754 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16755 having the attribute.
16757 2017-01-23 Martin Liska <mliska@suse.cz>
16759 * asan.c (create_asan_shadow_var): New function.
16760 (asan_expand_poison_ifn): Likewise.
16761 * asan.h (asan_expand_poison_ifn): New declaration.
16762 * internal-fn.c (expand_ASAN_POISON): Likewise.
16763 * internal-fn.def (ASAN_POISON): New builtin.
16764 * sanopt.c (pass_sanopt::execute): Expand
16765 asan_expand_poison_ifn.
16766 * tree-inline.c (copy_decl_for_dup_finish): Make function
16768 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16769 * tree-ssa.c (is_asan_mark_p): New function.
16770 (execute_update_addresses_taken): Rewrite local variables
16771 (identified just by use-after-scope as addressable) into SSA.
16773 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16775 * doc/install.texi (Specific): opensource.apple.com uses https
16776 now. Remove trailing slash.
16778 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16780 * README.Portability: Remove note on an Irix compatibility issue.
16782 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16784 * gcov.c (INCLUDE_ALGORITHM): Define.
16785 (INCLUDE_VECTOR): Define.
16786 No longer include <vector> and <algorithm> directly.
16788 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16790 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16792 * doc/invoke.texi (Code Gen Options): Ditto.
16794 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16797 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16799 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16801 rtl-optimization/79125
16802 * cprop.c (local_cprop_pass): Handle cases where we make an
16803 unconditional trap.
16805 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16809 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16810 read from, for big endian.
16812 2017-01-20 Jiong Wang <jiong.wang@arm.com>
16814 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
16815 register pauth builtins for LP64 only.
16817 2017-01-20 Marek Polacek <polacek@redhat.com>
16820 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
16823 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16825 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
16827 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
16828 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
16829 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
16831 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16834 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
16835 in big-endian mode when they are not a single duplicated value.
16837 2017-01-20 Richard Biener <rguenther@suse.de>
16839 * BASE-VER: Bump to 7.0.1.
16841 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16843 * omp-low.c (omplow_simd_context): New struct. Use it...
16844 (lower_rec_simd_input_clauses): ...here and...
16845 (lower_rec_input_clauses): ...here to hold common data. Adjust all
16846 references to idx, lane, max_vf, is_simt.
16848 2017-01-20 Graham Markall <graham.markall@embecosm.com>
16850 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
16853 2017-01-20 Martin Jambor <mjambor@suse.cz>
16855 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
16856 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
16858 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
16859 (GTFILES): Rename hsa.c to hsa-common.c.
16860 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
16861 * hsa-dump.c: Likewise.
16862 * hsa-gen.c: Likewise.
16863 * hsa-regalloc.c: Likewise.
16864 * ipa-hsa.c: Likewise.
16865 * omp-expand.c: Likewise.
16866 * omp-low.c: Likewise.
16867 * toplev.c: Likewise.
16869 2017-01-20 Marek Polacek <polacek@redhat.com>
16872 * doc/invoke.texi: Document -Wduplicated-branches.
16873 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
16874 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
16875 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
16876 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
16877 return 0 only when not OEP_LEXICOGRAPHIC.
16878 (fold_build_cleanup_point_expr): Use the expression
16879 location when building CLEANUP_POINT_EXPR.
16880 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
16881 * tree.c (add_expr): Handle error_mark_node.
16883 2017-01-20 Martin Liska <mliska@suse.cz>
16886 * tree-profile.c (init_ic_make_global_vars): Do not call
16888 (gimple_init_gcov_profiler): Likewise.
16890 2017-01-20 Martin Liska <mliska@suse.cz>
16893 * cgraph.h (maybe_create_reference): Remove argument and
16895 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
16897 * ipa-cp.c (create_specialized_node): Likewise.
16898 * symtab.c (symtab_node::maybe_create_reference): Handle
16899 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
16901 2017-01-20 Martin Liska <mliska@suse.cz>
16903 * read-rtl-function.c (function_reader::create_function): Use
16904 build_decl instread of build_decl_stat.
16906 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
16908 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
16909 * config/i386/avx512dqintrin.h: Ditto.
16910 * config/i386/avx512fintrin.h: Ditto.
16911 * config/i386/i386-builtin-types.def: Add new types.
16912 * config/i386/i386.c: Handle new types.
16913 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
16914 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
16915 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
16916 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
16917 (__builtin_ia32_kshiftridi): New.
16918 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
16920 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
16924 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
16925 define to rs6000_init_stack_protect_guard.
16926 (rs6000_init_stack_protect_guard): New function.
16928 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16929 Yunqiang Su <yunqiang.su@imgtec.com>
16931 * config.gcc (supported_defaults): Add madd4.
16932 (with_madd4): Add validation.
16933 (all_defaults): Add madd4.
16934 * config/mips/mips.opt (mmadd4): New option.
16935 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16937 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
16939 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
16940 (ISA_HAS_FUSED_MADD4): Likewise.
16941 * doc/invoke.texi (-mmadd4): Document the new option.
16942 * doc/install.texi (--with-madd4): Document the new option.
16944 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16946 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
16947 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
16948 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
16949 (aarch64_init_pauth_hint_builtins): New.
16950 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
16951 (aarch64_expand_builtin): Expand new builtins.
16953 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16955 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
16956 * combine-stack-adj.c (no_unhandled_cfa): Handle
16957 REG_CFA_TOGGLE_RA_MANGLE.
16958 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
16959 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
16960 info for return address signing.
16961 (aarch64_expand_epilogue): Likewise.
16963 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16965 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
16966 * config/aarch64/aarch64-protos.h
16967 (aarch64_return_address_signing_enabled): New declaration.
16968 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
16970 (aarch64_expand_prologue): Sign return address before it's pushed onto
16972 (aarch64_expand_epilogue): Authenticate return address fetched from
16974 (aarch64_override_options): Sanity check for ILP32 and ISA level.
16975 (aarch64_attributes): New function attributes for "sign-return-address".
16976 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
16977 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
16978 ("*do_return"): Generate combined instructions according to key index.
16979 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
16980 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
16982 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
16983 * config/aarch64/aarch64.opt (msign-return-address=): New.
16984 * doc/extend.texi (AArch64 Function Attributes): Documents
16985 "sign-return-address=".
16986 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
16988 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16990 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
16991 overall option summary.
16993 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16995 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
16996 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
16997 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
16998 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17000 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17002 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17003 -mpower9-minmax by default for -mcpu=power9.
17004 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17005 128-bit floating point.
17007 2017-01-20 Alan Modra <amodra@gmail.com>
17009 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17010 optimizing for size.
17012 2017-01-20 Alan Modra <amodra@gmail.com>
17015 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17016 for strcmp and strncmp from corresponding builtin decl.
17018 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17020 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17021 instead of i386/rtems-64.h.
17022 * config/i386/rtems-64.h: Remove.
17024 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17028 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17030 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17032 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17034 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17035 Change int to HOST_WIDE_INT.
17036 * config/aarch64/aarch64-protos.h
17037 (aarch64_simd_gen_const_vector_dup): Likewise.
17038 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17040 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17042 * langhooks-def.h (lhd_type_for_size): New decl.
17043 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17044 * langhooks.c (lhd_type_for_size): New function, taken from
17047 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17049 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17050 define_bypass for CR latency.
17051 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17052 (power9-alu2): Add define_bypass for CR latency.
17054 (power9-mul): Update insn latency.
17055 (power9-mul-compare): Update insn latency, bypass latency and remove
17058 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17060 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17062 * config/aarch64/aarch64.md
17063 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17064 aarch64_nopcrelative_literal_loads.
17065 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17067 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17069 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17070 TARGET_LOONGSON_3A.
17071 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17073 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17076 * config.gcc (supported_defaults): Add lxc1-sxc1.
17077 (with_lxc1_sxc1): Add validation.
17078 (all_defaults): Add lxc1-sxc1.
17079 * config/mips/mips.opt (mlxc1-sxc1): New option.
17080 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17082 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17083 __mips_no_lxc1_sxc1.
17084 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17085 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17086 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17088 2017-01-19 Richard Biener <rguenther@suse.de>
17090 PR tree-optimization/72488
17091 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17092 sure to restore SSA info.
17093 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17095 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17097 PR rtl-optimization/79121
17098 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17099 of the inner type when shifting an extended value.
17101 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17104 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17105 interposable aliases.
17107 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17110 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17111 Use the evmergelohi instruction.
17112 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17113 (mov_sitf_e500_subreg8_2_be): Likewise.
17114 (mov_sitf_e500_subreg12_2_le): Likewise.
17115 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17116 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17117 (mov_sitf_e500_subreg8_2_le): Likewise.
17118 (mov_sitf_e500_subreg12_2_be): Likewise.
17120 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17122 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17123 attribute from vecsimple to vecperm.
17124 (altivec_vbpermq2): Likewise.
17126 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17129 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17131 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17132 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17133 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17134 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17135 case where N arg is SIZE_MAX.
17136 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17137 (cmpstrsi): Add pattern.
17139 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17141 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17142 __builtin_vec_revb builtins.
17143 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17144 built-in functions to support generation of the ISA 3.0 XXBR<x>
17145 vector byte reverse instructions.
17146 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17147 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17148 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17149 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17150 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17151 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17152 (P9V_BUILTIN_VEC_REVB): Likewise.
17153 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17154 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17155 (p9_xxbrq_v16qi): Likewise.
17156 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17157 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17158 (p9_xxbrh_v8hi): Likewise.
17159 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17160 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17161 vec_revb built-in functions.
17163 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17165 PR rtl-optimization/78952
17166 * config/i386/i386.md (any_extract): New code iterator.
17167 (*insvqi_2): Use any_extract for source operand.
17168 (*insvqi_3): Use any_shiftrt for source operand.
17170 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17172 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17174 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17176 2017-01-18 Matthias Klose <doko@ubuntu.com>
17178 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17180 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17182 * config/rs6000/altivec.h (vec_bperm): Change #define.
17183 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17184 (altivec_vbpermq2): New define_insn.
17185 (altivec_vbpermd): Likewise.
17186 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17187 function interface.
17188 (VBPERMD): Likewise.
17189 (VBPERM): New polymorphic function interface.
17190 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17191 Add entries for P9V_BUILTIN_VEC_VBPERM.
17192 * doc/extend.texi: Add interfaces for vec_bperm.
17194 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17196 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17197 first letter of error messages.
17198 (s390_resolve_overloaded_builtin): Likewise.
17199 * config/s390/s390.c (s390_expand_builtin): Likewise.
17200 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17201 (s390_valid_target_attribute_inner_p): Likewise.
17202 * config/s390/s390.md ("tabort"): Likewise.
17204 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17206 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17207 (ISA_AVOID_DIV_HILO): New macro.
17208 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17209 (ISA_HAS_DDIV): Likewise.
17211 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17213 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17215 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17217 * doc/invoke.texi (fabi-version): Spelling fix.
17219 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17222 * doc/invoke.texi (fabi-version): Mention mangling fix for
17225 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17228 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17230 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17233 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17234 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17236 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17238 (stack_protect_test): Ditto.
17239 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17240 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17242 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17243 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17244 -mstack-protector-guard-offset=.
17245 (RS/6000 and PowerPC Options): Ditto.
17247 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17249 * config/i386/i386.h (MASK_CLASS_P): New define.
17250 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17251 there are no registers from different register sets also when
17252 mask registers are used. Update function comment.
17253 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17254 to (*k/*r) and (*k/*km) alternatives.
17256 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17258 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17259 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17260 (EH_RETURN_HANDLER_RTX): New define.
17261 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17262 Force frame pointer in EH return functions.
17263 (aarch64_expand_epilogue): Add barrier for eh_return.
17264 (aarch64_final_eh_return_addr): Remove.
17265 (aarch64_eh_return_handler_rtx): New function.
17266 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17268 (aarch64_eh_return_handler_rtx): New prototype.
17270 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17272 * config/rs6000/altivec.h (vec_rlmi): New #define.
17273 (vec_vrlnm): Likewise.
17274 (vec_rlnm): Likewise.
17275 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17276 (UNSPEC_VRLNM): Likewise.
17277 (VIlong): New mode iterator.
17278 (altivec_vrl<VI_char>mi): New define_insn.
17279 (altivec_vrl<VI_char>nm): Likewise.
17280 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17282 (VRLDNM): Likewise.
17283 (RLNM): New polymorphic function entry.
17284 (VRLWMI): New monomorphic function entry.
17285 (VRLDMI): Likewise.
17286 (RLMI): New polymorphic function entry.
17287 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17288 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17289 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17292 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17295 * dwarf2out.c (field_byte_offset): Restore the
17296 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17297 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17298 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17301 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17304 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17306 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17309 * dwarf2out.c (add_data_member_location_attribute): For constant
17310 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17311 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17312 DW_AT_byte_size attributes.
17314 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17316 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17317 after forcing to constant memory when the code model is medium.
17319 2017-01-17 Julia Koval <julia.koval@intel.com>
17322 * config/i386/avx512fintrin.h
17323 (_mm512_i32gather_ps): Change __addr type to void const*.
17324 (_mm512_mask_i32gather_ps): Ditto.
17325 (_mm512_i32gather_pd): Ditto.
17326 (_mm512_mask_i32gather_pd): Ditto.
17327 (_mm512_i64gather_ps): Ditto.
17328 (_mm512_mask_i64gather_ps): Ditto.
17329 (_mm512_i64gather_pd): Ditto.
17330 (_mm512_mask_i64gather_pd): Ditto.
17331 (_mm512_i32gather_epi32): Ditto.
17332 (_mm512_mask_i32gather_epi32): Ditto.
17333 (_mm512_i32gather_epi64): Ditto.
17334 (_mm512_mask_i32gather_epi64): Ditto.
17335 (_mm512_i64gather_epi32): Ditto.
17336 (_mm512_mask_i64gather_epi32): Ditto.
17337 (_mm512_i64gather_epi64): Ditto.
17338 (_mm512_mask_i64gather_epi64): Ditto.
17339 (_mm512_i32scatter_ps): Change __addr type to void*.
17340 (_mm512_mask_i32scatter_ps): Ditto.
17341 (_mm512_i32scatter_pd): Ditto.
17342 (_mm512_mask_i32scatter_pd): Ditto.
17343 (_mm512_i64scatter_ps): Ditto.
17344 (_mm512_mask_i64scatter_ps): Ditto.
17345 (_mm512_i64scatter_pd): Ditto.
17346 (_mm512_mask_i64scatter_pd): Ditto.
17347 (_mm512_i32scatter_epi32): Ditto.
17348 (_mm512_mask_i32scatter_epi32): Ditto.
17349 (_mm512_i32scatter_epi64): Ditto.
17350 (_mm512_mask_i32scatter_epi64): Ditto.
17351 (_mm512_i64scatter_epi32): Ditto.
17352 (_mm512_mask_i64scatter_epi32): Ditto.
17353 (_mm512_i64scatter_epi64): Ditto.
17354 (_mm512_mask_i64scatter_epi64): Ditto.
17355 * config/i386/avx512pfintrin.h
17356 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17357 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17358 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17359 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17360 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17361 (_mm512_prefetch_i32scatter_ps): Ditto.
17362 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17363 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17364 (_mm512_prefetch_i64scatter_pd): Ditto.
17365 (_mm512_prefetch_i64scatter_ps): Ditto.
17366 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17367 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17368 * config/i386/avx512vlintrin.h
17369 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17370 (_mm_mmask_i32gather_ps): Ditto.
17371 (_mm256_mmask_i32gather_pd): Ditto.
17372 (_mm_mmask_i32gather_pd): Ditto.
17373 (_mm256_mmask_i64gather_ps): Ditto.
17374 (_mm_mmask_i64gather_ps): Ditto.
17375 (_mm256_mmask_i64gather_pd): Ditto.
17376 (_mm_mmask_i64gather_pd): Ditto.
17377 (_mm256_mmask_i32gather_epi32): Ditto.
17378 (_mm_mmask_i32gather_epi32): Ditto.
17379 (_mm256_mmask_i32gather_epi64): Ditto.
17380 (_mm_mmask_i32gather_epi64): Ditto.
17381 (_mm256_mmask_i64gather_epi32): Ditto.
17382 (_mm_mmask_i64gather_epi32): Ditto.
17383 (_mm256_mmask_i64gather_epi64): Ditto.
17384 (_mm_mmask_i64gather_epi64): Ditto.
17385 (_mm256_i32scatter_ps): Change __addr type to void*.
17386 (_mm256_mask_i32scatter_ps): Ditto.
17387 (_mm_i32scatter_ps): Ditto.
17388 (_mm_mask_i32scatter_ps): Ditto.
17389 (_mm256_i32scatter_pd): Ditto.
17390 (_mm256_mask_i32scatter_pd): Ditto.
17391 (_mm_i32scatter_pd): Ditto.
17392 (_mm_mask_i32scatter_pd): Ditto.
17393 (_mm256_i64scatter_ps): Ditto.
17394 (_mm256_mask_i64scatter_ps): Ditto.
17395 (_mm_i64scatter_ps): Ditto.
17396 (_mm_mask_i64scatter_ps): Ditto.
17397 (_mm256_i64scatter_pd): Ditto.
17398 (_mm256_mask_i64scatter_pd): Ditto.
17399 (_mm_i64scatter_pd): Ditto.
17400 (_mm_mask_i64scatter_pd): Ditto.
17401 (_mm256_i32scatter_epi32): Ditto.
17402 (_mm256_mask_i32scatter_epi32): Ditto.
17403 (_mm_i32scatter_epi32): Ditto.
17404 (_mm_mask_i32scatter_epi32): Ditto.
17405 (_mm256_i32scatter_epi64): Ditto.
17406 (_mm256_mask_i32scatter_epi64): Ditto.
17407 (_mm_i32scatter_epi64): Ditto.
17408 (_mm_mask_i32scatter_epi64): Ditto.
17409 (_mm256_i64scatter_epi32): Ditto.
17410 (_mm256_mask_i64scatter_epi32): Ditto.
17411 (_mm_i64scatter_epi32): Ditto.
17412 (_mm_mask_i64scatter_epi32): Ditto.
17413 (_mm256_i64scatter_epi64): Ditto.
17414 (_mm256_mask_i64scatter_epi64): Ditto.
17415 (_mm_i64scatter_epi64): Ditto.
17416 (_mm_mask_i64scatter_epi64): Ditto.
17417 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17418 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17419 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17420 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17421 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17422 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17423 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17424 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17425 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17426 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17427 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17428 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17429 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17430 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17431 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17432 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17433 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17434 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17435 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17436 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17437 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17438 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17439 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17440 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17441 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17442 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17443 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17444 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17445 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17446 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17447 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17448 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17449 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17450 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17451 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17452 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17453 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17454 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17455 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17456 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17457 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17458 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17459 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17460 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17461 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17462 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17463 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17464 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17465 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17466 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17467 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17468 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17469 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17470 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17471 definitions accordingly.
17473 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17474 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17477 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17480 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17483 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17484 hard regs for inner regno.
17486 2017-01-17 Martin Liska <mliska@suse.cz>
17489 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17490 assumption and add comment.
17492 2017-01-17 Nathan Sidwell <nathan@acm.org>
17494 * ipa-visibility.c (localize_node): New function, broken out of ...
17495 (function_and_variable_visibility): ... here. Call it.
17497 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17499 PR middle-end/77445
17500 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17501 correctly set frequency of oudgoing edge.
17502 (duplicate_thread_path): Fix profile updating.
17504 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17507 * configure.ac: Add GCC_BASE_VER.
17508 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17509 version from BASE-VER file.
17510 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17511 (gcc.o): Depend on $(BASEVER).
17512 * common.opt (dumpfullversion): New option.
17513 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17514 * doc/invoke.texi: Document -dumpfullversion.
17515 * doc/install.texi: Document --with-gcc-major-version-only.
17516 * configure: Regenerated.
17518 2017-01-17 Richard Biener <rguenther@suse.de>
17520 PR tree-optimization/71433
17521 * tree-vrp.c (register_new_assert_for): Merge same asserts
17522 on all incoming edges.
17523 (process_assert_insertions_for): Handle insertions at the
17526 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17528 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17529 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17531 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17534 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17537 2017-01-17 Alan Modra <amodra@gmail.com>
17540 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17541 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17542 symbolic stack limit when pic.
17544 2017-01-16 Martin Sebor <msebor@redhat.com>
17546 PR tree-optimization/78608
17547 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17549 2017-01-16 Jeff Law <law@redhat.com>
17552 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17553 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17554 for several include directories that may be relative to sysroot.
17555 * config/i386/x-mingw32 (gplus_includedir): Define.
17556 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17557 (native_system_includedir): Likewise.
17558 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17559 override if TARGET_SYSTEM_ROOT is defined.
17560 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17562 PR tree-optimization/79090
17563 PR tree-optimization/33562
17564 PR tree-optimization/61912
17565 PR tree-optimization/77485
17566 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17567 and computed trims into the dump file.
17569 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17571 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17573 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17576 * gimplify.c (gimplify_init_constructor): If want_value and
17577 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17581 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17582 sequence. Formatting fixes.
17583 (doloop_optimize): Formatting fixes.
17586 * gcc.c (debug_level_greater_than_spec_func): New function.
17587 (static_spec_functions): Add debug-level-gt spec function.
17588 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17590 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17591 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17592 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17593 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17594 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17595 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17597 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17599 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17600 QImode fixups to general and mask registers only.
17602 2017-01-16 Carl Love <cel@us.ibm.com>
17604 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17605 for built-in functions
17606 vector signed char vec_nabs (vector signed char)
17607 vector signed short vec_nabs (vector signed short)
17608 vector signed int vec_nabs (vector signed int)
17609 vector signed long long vec_nabs (vector signed long long)
17610 vector float vec_nabs (vector float)
17611 vector double vec_nabs (vector double)
17612 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17614 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17615 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17616 * doc/extend.texi: Update the documentation file for the new built-in
17619 2017-01-16 Martin Sebor <msebor@redhat.com>
17621 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17624 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17626 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17627 UNSPEC_VSX__XXSPLTD to require special splat handling.
17629 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17632 * system.h: Poison strndup.
17634 2017-01-16 Alan Modra <amodra@gmail.com>
17637 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17640 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17642 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17644 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17646 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17647 call recog here. Assert that INSN_CODE (insn) is non-negative.
17649 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17652 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17654 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17655 in the currently scheduled RTL fragment.
17657 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17659 PR rtl-optimization/78751
17660 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17663 2017-01-14 Jeff Law <law@redhat.com>
17665 PR tree-optimization/79090
17666 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17667 variable length stores.
17668 (compute_trims): Delete dead assignment to *trim_tail.
17669 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17672 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17674 PR rtl-optimization/78626
17675 PR rtl-optimization/78727
17676 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17677 of a block, and split such blocks after everything else is finished.
17679 2017-01-14 Alan Modra <amodra@gmail.com>
17682 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17683 target legitimate_combined_insn.
17684 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17685 (rs6000_legitimate_combined_insn): New function.
17686 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17688 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17689 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17690 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17692 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17694 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17696 2017-01-13 Jeff Law <law@redhat.com>
17698 PR tree-optimization/33562
17699 PR tree-optimization/61912
17700 PR tree-optimization/77485
17701 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17703 (delete_dead_assignment): Likewise.
17704 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17705 statement to delete_dead_call and delete_dead_assignment.
17707 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17710 * substring-locations.c (format_warning_va): Strengthen case 1 so
17711 that both endpoints of the substring must be within the format
17712 range for just the substring to be printed.
17714 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17716 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17717 * config/i386/i386.c (ix86_target_string): Add missing options
17718 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17719 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17720 flags_other and ix86_target_other to flags2_other. Display unknown
17722 (ix86_valid_target_attribute_inner_p): Add missing options and
17723 reorder options by implied ISAs, as in ix86_target_string.
17725 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17727 * hash-table.h (hash_table::too_empty_p): New function.
17728 (hash_table::expand): Use it.
17729 (hash_table::traverse): Likewise.
17730 (hash_table::empty_slot): Use sizeof (value_type) instead of
17731 sizeof (PTR) to convert bytes to elements. Shrink the table
17732 if the current size is excessive for the current number of
17735 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17737 * ira-costs.c (record_reg_classes): Break from the inner loop
17738 early once alt_fail is known to be true. Update outer loop
17739 handling accordingly.
17741 2017-01-13 Jeff Law <law@redhat.com>
17743 * tree-ssa-dse.c (decrement_count): New function.
17744 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17745 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17746 when we know the partially dead statement is a mem* function.
17748 PR tree-optimization/61912
17749 PR tree-optimization/77485
17750 * tree-ssa-dse.c: Include expr.h.
17751 (maybe_trim_constructor_store): New function.
17752 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17754 PR tree-optimization/33562
17755 PR tree-optimization/61912
17756 PR tree-optimization/77485
17757 * doc/invoke.texi: Document new dse-max-object-size param.
17758 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17759 * tree-ssa-dse.c: Include params.h.
17760 (dse_store_status): New enum.
17761 (initialize_ao_ref_for_dse): New, partially extracted from
17763 (valid_ao_ref_for_dse, normalize_ref): New.
17764 (setup_live_bytes_from_ref, compute_trims): Likewise.
17765 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17766 (maybe_trim_partially_dead_store): Likewise.
17767 (maybe_trim_complex_store): Likewise.
17768 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17769 Track what bytes live from the original store. Return tri-state
17770 for dead, partially dead or live.
17771 (dse_dom_walker): Add constructor, destructor and new private members.
17772 (delete_dead_call, delete_dead_assignment): New extracted from
17774 (dse_optimize_stmt): Make a member of dse_dom_walker.
17775 Use initialize_ao_ref_for_dse.
17777 PR tree-optimization/33562
17778 PR tree-optimization/61912
17779 PR tree-optimization/77485
17780 * sbitmap.h (bitmap_count_bits): Prototype.
17781 (bitmap_clear_range, bitmap_set_range): Likewise.
17782 * sbitmap.c (bitmap_clear_range): New function.
17783 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17785 2017-01-13 Martin Liska <mliska@suse.cz>
17788 * function.c (set_cfun): Add new argument force.
17789 * function.h (set_cfun): Likewise.
17790 * ipa-inline-transform.c (inline_call): Use the function when
17791 strict alising from is dropped for function we inline to.
17793 2017-01-13 Richard Biener <rguenther@suse.de>
17795 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17796 for dumping GIMPLE INTEGER_CSTs.
17798 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17800 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17801 to 201112L since C++17.
17803 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17806 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17807 if -fsanitize=kernel-address is present.
17809 2017-01-13 Richard Biener <rguenther@suse.de>
17811 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
17812 as _Literal ( type ) number in case usual suffixes do not
17813 preserve all information.
17815 2017-01-13 Richard Biener <rguenther@suse.de>
17817 PR tree-optimization/77283
17818 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
17819 and ssa-iterators.h.
17820 (is_feasible_trace): Implement a cost model based on joiner
17823 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17826 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
17827 char or short to __float128/_Float128 directly.
17829 2017-01-12 Martin Sebor <msebor@redhat.com>
17831 to -Wformat-overflow.
17832 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
17833 (min_bytes_remaining): Same.
17834 (get_string_length): Same.
17835 (format_string): Same.
17836 (format_directive): Same.
17838 (pass_sprintf_length::handle_gimple_call): Same.
17840 2017-01-12 Jakub Jelinek <jakub@redhat.com>
17842 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
17843 info.nowrite calls with no lhs that can't throw. Return bool
17844 whether gsi_remove has been called or not.
17845 (pass_sprintf_length::handle_gimple_call): Return bool whether
17846 try_substitute_return_value called gsi_remove. Formatting fix.
17847 (pass_sprintf_length::execute): Don't use gsi_remove if
17848 handle_gimple_call returned true.
17851 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
17852 be removed due to side-effects, don't remove following barrier nor
17853 turn the successor edge into fallthru edge.
17855 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17858 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
17859 element-reversing loads and stores as not swappable.
17861 2017-01-12 Nathan Sidwell <nathan@acm.org>
17862 Nicolai Stange <nicstange@gmail.com>
17864 * combine.c (try_combine): Don't ignore result of overlap checking
17865 loop. Combine overlap & asm check into single loop.
17867 2017-01-12 Richard Biener <rguenther@suse.de>
17869 * tree-pretty-print.c (dump_generic_node): Provide -gimple
17870 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
17872 2017-01-12 Richard Biener <rguenther@suse.de>
17874 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
17875 and TS_TARGET_OPTION directly derive from TS_BASE.
17876 * tree-core.h (tree_optimization_option): Derive from tree_base.
17877 (tree_target_option): Likewise.
17879 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
17881 * config/i386/i386.c (memory_address_length): Increase len
17882 only when rip_relative_addr_p returns false.
17884 2017-01-11 Julia Koval <julia.koval@intel.com>
17886 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
17887 (OPTION_MASK_ISA_SGX_SET): New.
17888 (ix86_handle_option): Handle OPT_msgx.
17889 * config.gcc: Added sgxintrin.h.
17890 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
17891 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
17892 * config/i386/i386.c (ix86_target_string): Add -msgx.
17894 (ix86_option_override_internal): Handle new options.
17895 (ix86_valid_target_attribute_inner_p): Add sgx.
17896 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
17897 * config/i386/i386.opt: Add msgx.
17898 * config/i386/sgxintrin.h: New file.
17899 * config/i386/x86intrin.h: Add sgxintrin.h.
17901 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17904 * fold-const.c (maybe_nonzero_address): Return 1 for function
17906 (tree_single_nonzero_warnv_p): Don't handle function local objects
17910 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
17913 2017-01-11 David Malcolm <dmalcolm@redhat.com>
17916 * opts.c: Include "spellcheck.h"
17917 (struct string_fragment): New struct.
17918 (struct edit_distance_traits<const string_fragment &>): New
17920 (get_closest_sanitizer_option): New function.
17921 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
17923 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17925 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
17927 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
17928 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
17929 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
17930 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
17931 for initial die_offset if dwarf_split_debug_info.
17932 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
17933 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
17935 (output_skeleton_debug_sections): Formatting fix. Use
17936 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
17937 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
17939 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
17941 * config/arm/cortex-a53.md: Add bypasses for
17942 cortex_a53_r2f_cvt.
17943 (cortex_a53_r2f): Only use for transfers.
17944 (cortex_a53_f2r): Likewise.
17945 (cortex_a53_r2f_cvt): Add reservation for conversions.
17946 (cortex_a53_f2r_cvt): Likewise.
17948 2017-01-11 Tamar Christina <tamar.christina@arm.com>
17950 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
17951 to all inlined functions, change static to extern.
17953 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
17956 * config/arm/arm.c (legitimize_pic_address): Handle reference to
17958 (arm_assemble_integer): Likewise.
17960 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
17962 * config.gcc: Use new awk script to check CPU, FPU and architecture
17963 parameters for --with-... options.
17964 * config/arm/parsecpu.awk: New file
17965 * config/arm/arm-cpus.in: New file.
17966 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
17968 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
17970 * config/arm/t-arm: Update dependency rules.
17971 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
17972 of processing .def files.
17973 * config/arm/genopt.sh: Deleted.
17974 * config/arm/gentune.sh: Deleted.
17975 * config/arm/arm-cores.def: Deleted.
17976 * config/arm/arm-arches.def: Deleted.
17977 * config/arm/arm-fpus.def: Deleted.
17978 * config/arm/arm-tune.md: Regenerated.
17979 * config/arm/arm-tables.opt: Regenerated.
17980 * config/arm/arm-cpu.h: New generated file.
17981 * config/arm/arm-cpu-data.h: New generated file.
17982 * config/arm/arm-cpu-cdata.h: New generated file.
17984 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
17987 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
17989 (input_varpool_node): Unpack dynamically_initialized bit.
17991 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
17993 PR rtl-optimization/79032
17994 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
17995 the alignment of the adjusted memory reference against that of MODE,
17996 instead of the alignment of the original memory reference.
17998 2017-01-11 Martin Jambor <mjambor@suse.cz>
18000 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18002 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18003 decorated functions.
18005 2017-01-11 Richard Biener <rguenther@suse.de>
18007 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18008 set range/nonnull info for PHI results. Do not set it on
18009 stmts marked for removal.
18011 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18013 * expr.c (store_field): In the bitfield case, fetch the return value
18014 from the registers before applying a single big-endian adjustment.
18015 Always do a final load for a BLKmode value not larger than a word.
18017 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18020 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18021 that we correctly handle column numbers greater than
18022 LINE_MAP_MAX_COLUMN_NUMBER.
18024 2017-01-10 Martin Sebor <msebor@redhat.com>
18026 PR middle-end/78245
18027 * gimple-ssa-sprintf.c (get_destination_size): Call
18028 {init,fini}object_sizes.
18029 * tree-object-size.c (addr_object_size): Adjust.
18030 (pass_through_call): Adjust.
18031 (pass_object_sizes::execute): Adjust.
18032 * tree-object-size.h (fini_object_sizes): Declare.
18034 2017-01-10 Martin Sebor <msebor@redhat.com>
18036 PR tree-optimization/78775
18037 * builtins.c (get_size_range): Move...
18038 * calls.c: ...to here.
18039 (alloc_max_size): Accept zero argument.
18040 (operand_signed_p): Remove.
18041 (maybe_warn_alloc_args_overflow): Call get_size_range.
18042 * calls.h (get_size_range): Declare.
18044 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18046 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18047 from TI's devices.csv file as of September 2016.
18048 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18050 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18052 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18053 * doc/invoke.texi: Likewise.
18054 * doc/md.texi: Likewise.
18055 * doc/objc.texi: Likewise.
18057 2017-01-10 Joshua Conner <joshconner@google.com>
18059 * config/arm/fuchsia-elf.h: New file.
18060 * config/fuchsia.h: New file.
18061 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18062 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18064 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18066 2016-01-10 Richard Biener <rguenther@suse.de>
18068 PR tree-optimization/79034
18069 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18070 Propagate out degenerate PHIs in the joiner.
18072 2017-01-10 Martin Liska <mliska@suse.cz>
18074 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18075 (sort_congruence_classes_by_decl_uid): Likewise.
18076 (sort_congruence_class_groups_by_decl_uid): Likewise.
18077 (sem_item_optimizer::merge_classes): Sort class, groups in these
18078 classes and members in the groups by DECL_UID of declarations.
18079 This would make merge operations stable.
18081 2017-01-10 Martin Liska <mliska@suse.cz>
18083 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18084 usage of m_classes_vec.
18085 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18086 (sem_item_optimizer::get_group_by_hash): Likewise.
18087 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18088 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18089 (sem_item_optimizer::verify_classes): Likewise.
18090 (sem_item_optimizer::process_cong_reduction): Likewise.
18091 (sem_item_optimizer::dump_cong_classes): Likewise.
18092 (sem_item_optimizer::merge_classes): Likewise.
18093 * ipa-icf.h (congruence_class_hash): Rename from
18094 congruence_class_group_hash. Remove declaration of m_classes_vec.
18096 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18098 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18099 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18100 * config.gcc: Add avx512vpopcntdqintrin.h.
18101 * config/i386/avx512vpopcntdqintrin.h: New.
18102 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18103 * config/i386/i386-builtin-types.def: Add new types.
18104 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18105 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18106 __builtin_ia32_vpopcountq_v8di_mask): New.
18107 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18108 __AVX512VPOPCNTDQ__.
18109 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18110 (PTA_AVX512VPOPCNTDQ): Define.
18111 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18112 TARGET_AVX512VPOPCNTDQ_P): Define.
18113 * config/i386/i386.opt: Add mavx512vpopcntdq.
18114 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18115 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18117 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18119 PR middle-end/77484
18120 * predict.def (PRED_CALL): Set to 67.
18122 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18124 * expr.c (store_field): In the bitfield case, if the value comes from
18125 a function call and is of an aggregate type returned in registers, do
18126 not modify the field mode; extract the value in all cases if the mode
18127 is BLKmode and the size is not larger than a word.
18129 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18132 * config/i386/cpuid.h: Fix undefined behavior.
18134 2017-01-04 Jeff Law <law@redhat.com>
18136 PR tree-optimization/79007
18137 PR tree-optimization/67955
18138 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18139 conservative for pt.null when flag_non_call_exceptions is on.
18141 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18143 PR translation/79019
18144 PR translation/79020
18145 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18146 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18147 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18149 * config/avr/avr.opt (maccumulate-args): Likewise.
18150 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18151 * common.opt (freport-bug): Likewise.
18152 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18153 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18154 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18155 translatable string.
18156 * config/i386/i386.c (function_value_32): Likewise.
18157 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18158 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18160 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18161 * common/config/msp430/msp430-common.c (msp430_handle_option):
18163 * symtab.c (symtab_node::verify_base): Likewise.
18164 * opts.c (set_debug_level): Likewise.
18165 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18166 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18167 missing whitespace to translatable strings.
18168 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18169 * config/sh/superh.h: Likewise.
18170 * config/i386/xopintrin.h: Likewise.
18171 * config/i386/znver1.md: Likewise.
18172 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18173 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18174 * double-int.h (struct double_int): Likewise.
18175 * double-int.c (div_and_round_double): Likewise.
18176 * wide-int.cc: Likewise.
18177 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18178 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18179 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18180 (crossjumps_occurred): ... this.
18181 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18184 PR tree-optimization/78899
18185 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18186 returning bool return struct loop *, NULL for failure and the new
18188 (versionable_outer_loop_p): Don't version outer loop if it has
18189 dont_vectorized bit set.
18190 (tree_if_conversion): When versioning outer loop, ensure
18191 tree_if_conversion is performed also on the inner loop of the
18192 non-vectorizable outer loop copy.
18193 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18194 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18195 prevent vectorization of it.
18196 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18197 the outer loop vectorization of the non-scalar version is attempted
18198 before vectorization of the inner loop in scalar version. If
18199 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18200 vectorization of its inner loop.
18201 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18202 has 2 inner loops, rename also on edges from bb whose single pred
18203 is outer_loop->header. Fix typo in function comment.
18205 2017-01-09 Martin Sebor <msebor@redhat.com>
18208 * asan.c (asan_emit_stack_protection): Increase local buffer size
18209 to avoid snprintf truncation warning.
18211 2017-01-09 Andrew Pinski <apinski@cavium.com>
18213 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18214 to reference thunderx2t99 for the tuning structure
18215 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18217 (thunderx2t99_extra_costs): This.
18218 * config/aarch64/aarch64-tune.md: Regenerate.
18219 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18220 (vulcan_addrcost_table): This.
18221 (vulcan_regmove_cost): Rename to ...
18222 (thunderx2t99_regmove_cost): This.
18223 (vulcan_vector_cost): Rename to ...
18224 (thunderx2t99_vector_cost): this.
18225 (vulcan_branch_cost): Rename to ...
18226 (thunderx2t99_branch_cost): This.
18227 (vulcan_tunings): Rename to ...
18228 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18229 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18231 2017-01-09 Martin Jambor <mjambor@suse.cz>
18235 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18236 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18237 (propagate_vr_accross_jump_function): Use the above function for all
18238 value range computations for pass-through jump functions and type
18239 converasion from explicit value range values.
18240 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18241 parameters from TYPE_ARG_TYPES.
18242 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18243 (ipa_write_node_info): Stream type of the actual argument.
18244 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18246 2017-01-09 Martin Liska <mliska@suse.cz>
18249 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18250 (lookup_compiler): Do not show error message with have_E.
18252 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18254 PR tree-optimization/78938
18255 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18256 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18257 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18258 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18261 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18263 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18266 2017-01-09 Richard Biener <rguenther@suse.de>
18268 PR tree-optimization/78997
18269 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18270 name condition properly.
18272 2017-01-09 Richard Biener <rguenther@suse.de>
18275 * dwarf2out.c (is_cxx): New overload with context.
18276 (is_naming_typedef_decl): Use it.
18278 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18280 * invoke.texi (Option Summary): Correct spacing in option lists
18281 and add line breaks to fix over-long lines.
18283 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18285 PR middle-end/17660
18287 * extend.texi (Common Variable Attributes): Add xref to GCC
18288 Internals manual to explain mode attribute keywords.
18290 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18293 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18294 and Preprocessor Options.
18295 (Options for Linking): Document -pthread here....
18296 (RS/6000 and PowerPC Options): ...not here.
18297 (Solaris 2 Options): ...or here.
18298 * doc/cppopts.texi: Document -pthread.
18300 2017-01-08 Martin Sebor <msebor@redhat.com>
18302 PR middle-end/77708
18303 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18304 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18305 New member functions.
18306 (format_directive): Used them.
18308 (pass_sprintf_length::handle_gimple_call): Same.
18309 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18310 to avoid truncation for any argument.
18311 (extract_affine_mul): Same.
18312 * tree.c (get_file_function_name): Same.
18314 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18316 PR middle-end/77484
18317 * predict.def (PRED_INDIR_CALL): Set to 86.
18319 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18321 PR preprocessor/54124
18322 * doc/cppopts.texi: Reformat -d subtable to list the full name
18323 of the options. Add cross-reference to the docs for the general
18324 compiler -d options.
18325 * doc/invoke.texi (Developer Options): Add cross-reference to the
18326 preprocessor-specific -d option documentation.
18328 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18330 PR preprocessor/13498
18331 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18332 redudant material, and reflect new command-line options.
18333 (System Headers): Likewise.
18335 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18337 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18338 -isystem, and -idirafter. Copy-edit.
18339 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18340 default for -ftrack-macro-expansion. Delete obsolete and
18341 badly-formatted implementation details about -fdebug-cpp output.
18342 * doc/cppwarnopts.texi: Copy-edit.
18344 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18347 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18348 that the transition from a max line width >= 1<<10 to narrower
18349 lines works correctly.
18351 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18353 * doc/options.texi (PerFunction): New.
18354 * opt-functions.awk (switch_flags): Map both Optimization and
18355 PerFunction to CL_OPTIMIZATION.
18356 * opth-gen.awk: Test for PerFunction flag along with
18358 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18359 it only when the latter is present. Skip those that don't in
18360 the hash function generator.
18361 * common.opt (fvar-tracking): Mark as PerFunction instead of
18363 (fvar-tracking-assignments): Likewise.
18364 (fvar-tracking-assignments-toggle): Likewise.
18365 (fvar-tracking-uninit): Likewise.
18367 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18369 PR translation/79018
18370 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18373 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18376 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18377 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18378 TARGET_LONG_JUMP_TABLE_OFFSETS.
18379 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18380 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18381 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18382 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18383 * config/m68k/m68k.md (tablejump expander): Likewise.
18384 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18385 TARGET_LONG_JUMP_TABLE_OFFSETS.
18386 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18387 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18389 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18390 David Holsgrove <david.holsgrove@xilinx.com>
18392 * common/config/microblaze/microblaze-common.c
18393 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18394 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18396 * config/microblaze/microblaze.c (microblaze_must_save_register)
18397 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18399 (microblaze_eh_return): New function.
18400 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18401 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18402 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18403 * config/microblaze/microblaze.md (eh_return): New pattern.
18405 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18407 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18408 GCC_DIAGNOSTIC_STRINGIFY): Define.
18410 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18412 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18414 * config/arm/arm.md (<mcrr>): New.
18416 * config/arm/arm.c (arm_arch5te): New.
18417 (arm_option_override): Set arm_arch5te.
18418 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18420 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18421 (arm_mcrr_qualifiers): ... this. New.
18422 (MRRC_QUALIFIERS): Define to...
18423 (arm_mrrc_qualifiers): ... this. New.
18424 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18426 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18427 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18428 (MRRCI, mrrc, MRRC): New.
18429 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18430 VUNSPEC_MRRC2): New.
18432 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18434 * config/arm/arm.md (<mcr>): New.
18436 * config/arm/arm.c (arm_coproc_builtin_available): Add
18437 support for mcr, mrc, mcr2 and mrc2.
18438 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18439 (arm_mcr_qualifiers): ... this. New.
18440 (MRC_QUALIFIERS): Define to ...
18441 (arm_mrc_qualifiers): ... this. New.
18442 (MCR_QUALIFIERS): Define to ...
18443 (arm_mcr_qualifiers): ... this. New.
18444 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18446 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18447 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18448 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18449 VUNSPEC_MRC2): New.
18451 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18453 * config/arm/arm.md (*ldc): New.
18457 * config/arm/arm.c (arm_coproc_builtin_available): Add
18458 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18459 (arm_coproc_ldc_stc_legitimate_address): New.
18460 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18461 'qualifier_const_pointer'.
18462 (LDC_QUALIFIERS): Define to...
18463 (arm_ldc_qualifiers): ... this. New.
18464 (STC_QUALIFIERS): Define to...
18465 (arm_stc_qualifiers): ... this. New.
18466 * config/arm/arm-protos.h
18467 (arm_coproc_ldc_stc_legitimate_address): New.
18468 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18469 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18470 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18471 stc2, stcl, stc2l): New.
18472 * config/arm/constraints.md (Uz): New.
18473 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18474 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18475 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18476 VUNSPEC_STC2L): New.
18478 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18480 * config/arm/arm.md (<cdp>): New.
18481 * config/arm/arm.c (neon_const_bounds): Rename this ...
18482 (arm_const_bounds): ... this.
18483 (arm_coproc_builtin_available): New.
18484 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18485 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18486 (CDP_QUALIFIERS): Define to...
18487 (arm_cdp_qualifiers): ... this. New.
18489 (arm_expand_builtin_args): Add case for 6 arguments.
18490 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18491 (arm_const_bounds): ... this.
18492 (arm_coproc_builtin_available): New.
18493 * config/arm/arm_acle.h (__arm_cdp): New.
18495 * config/arm/arm_acle_builtins.def (cdp): New.
18497 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18498 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18499 'arm_const_bounds'.
18500 * config/arm/types.md (coproc): New.
18501 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18502 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18503 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18504 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18506 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18508 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18509 (UBINOP_QUALIFIERS): New.
18511 (acle_builtin_data): New. Change comment.
18512 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18513 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18514 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18515 arm_acle_builtins.def.
18516 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18517 (arm_init_acle_builtins): New.
18518 (CRC32_BUILTIN): Remove.
18519 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18520 crc32cb, crc32ch and crc32cw.
18521 (arm_init_crc32_builtins): Remove.
18522 (arm_init_builtins): Use arm_init_acle_builtins rather
18523 than arm_init_crc32_builtins.
18524 (arm_expand_acle_builtin): New.
18525 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18526 * config/arm/arm_acle_builtins.def: New.
18528 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18530 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18531 (arm_builtin_datum): ... this.
18532 (arm_init_neon_builtin): Rename to ...
18533 (arm_init_builtin): ... this. Add a new parameters PREFIX
18534 and USE_SIG_IN_NAME.
18535 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18536 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18537 'arm_builtin_datum'.
18538 (arm_init_vfp_builtins): Likewise.
18539 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18540 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18541 (arm_expand_neon_args): Rename to ...
18542 (arm_expand_builtin_args): ... this. Rename builtin_arg
18543 enum values and differentiate between ARG_BUILTIN_MEMORY
18544 and ARG_BUILTIN_NEON_MEMORY.
18545 (arm_expand_neon_builtin_1): Rename to ...
18546 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18547 values, arm_expand_builtin_args and add bool parameter NEON.
18548 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18549 (arm_expand_vfp_builtin): Likewise.
18550 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18552 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18554 PR middle-end/77484
18555 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18556 * predict.c (tree_estimate_probability_bb): Reverse direction of
18557 polymorphic call predictor.
18559 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18561 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18562 (determine_pass_name_match): ...this new function and...
18563 (should_skip_pass_p): ...this new function.
18565 2017-01-06 Nathan Sidwell <nathan@acm.org>
18567 * ipa-visibility.c (function_and_variable_visibility): Reformat
18568 comments and long lines. Remove extrneous if.
18569 * symtab.c (symtab_node::make_decl_local): Fix code format.
18570 (symtab_node::set_section_for_node): Fix comment typo.
18572 2017-01-06 Martin Liska <mliska@suse.cz>
18575 * lra-constraints.c: Rename invariant to lra_invariant.
18576 * predict.c (set_even_probabilities): Initialize e to NULL.
18578 2017-01-05 Martin Sebor <msebor@redhat.com>
18580 PR tree-optimization/78910
18581 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18582 (format_integer): Correct off-by-one error in the handling
18583 of precision with negative numbers in signed conversions..
18585 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18587 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18589 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18591 PR tree-optimization/71016
18592 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18593 factor_out_conditional_conversion. Formatting fix.
18594 (factor_out_conditional_conversion): Add cond_stmt argument.
18595 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18596 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18599 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18601 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18602 read-rtl-function.o, and selftest-rtl.o.
18603 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18604 (selftest::aarch64_test_loading_full_dump): New function.
18605 (selftest::aarch64_run_selftests): New function.
18606 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18607 selftest::aarch64_run_selftests.
18608 * config/i386/i386.c
18609 (selftest::ix86_test_loading_dump_fragment_1): New function.
18610 (selftest::ix86_test_loading_call_insn): New function.
18611 (selftest::ix86_test_loading_full_dump): New function.
18612 (selftest::ix86_test_loading_unspec): New function.
18613 (selftest::ix86_run_selftests): Call the new functions.
18614 * emit-rtl.c (maybe_set_max_label_num): New function.
18615 * emit-rtl.h (maybe_set_max_label_num): New decl.
18616 * function.c (instantiate_decls): Guard call to
18617 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18618 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18620 * gensupport.c (gen_reader::gen_reader): Pass "false"
18621 for new "compact" param of rtx_reader.
18622 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18623 rather than an empty string for NULL strings.
18624 * read-md.c: Potentially include config.h rather than bconfig.h.
18625 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18626 (have_error): New global, copied from errors.c.
18627 (md_reader::read_name): Rename to...
18628 (md_reader::read_name_1): ...this, adding "out_loc" param,
18629 and converting "missing name or number" to returning false, rather
18631 (md_reader::read_name): Reimplement in terms of read_name_1.
18632 (md_reader::read_name_or_nil): New function.
18633 (md_reader::read_string): Handle "(nil)" by returning NULL.
18634 (md_reader::md_reader): Add new param "compact".
18635 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18636 (md_reader::read_file): New method.
18637 * read-md.h (md_reader::md_reader): Add new param "compact".
18638 (md_reader::read_file): New method.
18639 (md_reader::is_compact): New accessor.
18640 (md_reader::read_name): Convert return type from void to file_location.
18641 (md_reader::read_name_or_nil): New decl.
18642 (md_reader::read_name_1): New decl.
18643 (md_reader::m_compact): New field.
18644 (noop_reader::noop_reader): Pass "false" for new "compact" param
18646 (rtx_reader::rtx_reader): Add new "compact" param.
18647 (rtx_reader::read_rtx_operand): Make virtual and convert return
18648 type from void to rtx.
18649 (rtx_reader::read_until): New decl.
18650 (rtx_reader::handle_any_trailing_information): New virtual function.
18651 (rtx_reader::postprocess): New virtual function.
18652 (rtx_reader::finalize_string): New virtual function.
18653 (rtx_reader::m_in_call_function_usage): New field.
18654 (rtx_reader::m_reuse_rtx_by_id): New field.
18655 * read-rtl-function.c: New file.
18656 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18657 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18658 (selftest::verify_three_block_rtl_cfg): New decl.
18659 * read-rtl-function.h: New file.
18660 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18661 For host, include function.h, memmodel.h, and emit-rtl.h.
18662 (one_time_initialization): New function.
18663 (struct compact_insn_name): New struct.
18664 (compact_insn_names): New array.
18665 (find_code): Handle insn codes in compact dumps.
18666 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18667 (bind_subst_iter_and_attr): Likewise.
18668 (add_condition_to_string): Likewise.
18669 (add_condition_to_rtx): Likewise.
18670 (apply_attribute_uses): Likewise.
18671 (add_current_iterators): Likewise.
18672 (apply_iterators): Likewise.
18673 (initialize_iterators): Guard usage of apply_subst_iterator with
18674 #ifdef GENERATOR_FILE.
18675 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18676 (md_reader::read_mapping): Likewise.
18677 (add_define_attr_for_define_subst): Likewise.
18678 (add_define_subst_attr): Likewise.
18679 (read_subst_mapping): Likewise.
18680 (check_code_iterator): Likewise.
18681 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18683 (one_time_initialization): New function.
18684 (rtx_reader::read_until): New method.
18685 (read_flags): New function.
18686 (parse_reg_note_name): New function.
18687 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18688 Handle reuse_rtx ids.
18689 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18690 Add parsing support for RTL dumps, mirroring the special-cases in
18691 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18692 values, and calling handle_any_trailing_information.
18693 (rtx_reader::read_rtx_operand): Convert return type from void
18694 to rtx, returning return_rtx. Handle case 'e'. Call
18695 finalize_string on XSTR and XTMPL fields.
18696 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18697 "(nil)" values were omitted. Call the postprocess vfunc on the
18699 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18700 class ctor. Initialize m_in_call_function_usage. Call
18701 one_time_initialization.
18702 * rtl-tests.c (selftest::test_uncond_jump): Call
18703 set_new_first_and_last_insn.
18704 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18705 * selftest-rtl.c: New file.
18706 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18707 (selftest::get_insn_by_uid): New decl.
18708 * selftest-run-tests.c (selftest::run_tests): Call
18709 read_rtl_function_c_tests.
18710 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18711 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18714 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18716 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18717 operands in a special way. Assert that pos+len <= mode precision.
18719 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18721 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18722 3 argument Alias with unlimited for the negative form.
18723 (fno-vect-cost-model): Removed.
18725 2017-01-05 Martin Liska <mliska@suse.cz>
18727 * hsa-gen.c (gen_hsa_divmod): New function.
18728 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18730 2017-01-05 Martin Liska <mliska@suse.cz>
18733 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18736 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18738 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18739 small constant length operands.
18741 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18743 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18744 between loop iterations.
18746 2017-01-05 Martin Liska <mliska@suse.cz>
18749 * gimplify.c (gimplify_decl_expr): Compare to
18750 asan_poisoned_variables instread of checking flags.
18751 (gimplify_target_expr): Likewise.
18752 (gimplify_expr): Likewise.
18753 (gimplify_function_tree): Conditionally initialize
18754 asan_poisoned_variables.
18756 2017-01-04 Jeff Law <law@redhat.com>
18758 PR tree-optimizatin/78812
18759 * rtl.h (contains_mem_rtx_p): Prototype.
18760 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18761 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18762 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18763 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18764 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18766 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18768 * input.c (assert_char_at_range): Default-initialize actual_range.
18770 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18772 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18773 to match the caller.
18775 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18777 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18778 insns after final jump in test to emit dummy move.
18780 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18782 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18783 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18785 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18787 * multiple_target.c (create_dispatcher_calls): Init e_next.
18788 * tree-ssa-loop-split.c (split_loop): Init border.
18789 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18792 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18797 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18798 (altivec_register_operand): Do not return true if the operand
18799 contains a SUBREG mixing SImode and SFmode.
18800 (vsx_register_operand): Likewise.
18801 (vsx_reg_sfsubreg_ok): New predicate.
18802 (vfloat_operand): Do not return true if the operand contains a
18803 SUBREG mixing SImode and SFmode.
18804 (vint_operand): Likewise.
18805 (vlogical_operand): Likewise.
18806 (gpc_reg_operand): Likewise.
18807 (int_reg_operand): Likewise.
18808 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18809 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18810 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
18812 (rs6000_emit_move_si_sf_subreg): New helper function.
18813 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
18814 fixup SUBREGs involving SImode and SFmode.
18815 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
18816 numbers for the new peephole2 optimization.
18817 (peephole2 for SFmode unions): New peephole2 to optimize cases in
18818 the GLIBC math library that do AND/IOR/XOR operations on single
18819 precision floating point.
18820 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
18821 target macros to say whether we need to avoid SUBREGs mixing
18823 (TARGET_ALLOW_SF_SUBREG): Likewise.
18824 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
18825 (UNSPEC_SI_FROM_SF): Likewise.
18826 (iorxor): Change spacing.
18827 (and_ior_xor): New iterator for AND, IOR, and XOR.
18828 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
18829 (movdi_from_sf_zero_ext): Likewise.
18830 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
18831 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
18832 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
18833 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
18834 (fms<mode>4): Likewise.
18835 (fnma<mode>4): Likewise.
18836 (fnms<mode>4): Likewise.
18837 (nfma<mode>4): Likewise.
18838 (nfms<mode>4): Likewise.
18840 2017-01-04 Marek Polacek <polacek@redhat.com>
18843 * doc/invoke.texi: Document -Wpointer-compare.
18845 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18847 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
18850 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
18851 descriptions for -gdwarf-5 and emit them as uleb128 instead of
18854 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
18857 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
18858 documentation of the powerpc_popcntb_ok attribute.
18859 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18860 code to issue warning messages if a requested CPU configuration is
18861 not supported by the binary (assembler and loader) toolchain.
18862 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
18863 made to define a built-in function that has been disabled.
18864 (paired_init_builtins): Add assertion to prevent ICE if attempt is
18865 made to define a built-in function that has been disabled.
18866 (altivec_init_builtins): Add comment explaining why definition
18867 of the DST built-in functions is not preceded by an assertion
18868 check. Add assertions to prevent ICE if attempts are made to
18869 define an altivec predicate or an abs* built-in function that has
18871 (htm_init_builtins): Add comment explaining why definition of the
18872 htm built-in functions is not preceded by an assertion check.
18874 2017-01-04 Jeff Law <law@redhat.com>
18876 PR tree-optimizatin/67955
18877 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
18878 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
18879 the points-to solution does not include pt_null. Use DECL_PT_UID
18882 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
18884 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
18885 Use gen_int_mode instead of gen_lopwart for const_int operands.
18887 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18889 PR tree-optimization/71563
18890 * match.pd: Simplify X << Y into X if Y is known to be 0 or
18891 out of range value - has low bits known to be zero.
18893 2017-01-04 Alan Modra <amodra@gmail.com>
18895 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
18896 * configure: Regenerate.
18897 * config.in: Regenerate.
18899 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18902 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
18903 a substring of the message, but strcmp with the whole message. Ifdef
18904 ENABLE_NLS, translate the message first using dgettext.
18906 2017-01-03 Jeff Law <law@redhat.com>
18908 PR tree-optimizatin/78856
18909 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
18910 (mark_threaded_blocks): Remove code to truncate thread paths that
18911 cross multiple loop headers. Instead invalidate the cached loop
18912 iteration information and handle case of a thread path walking
18913 into an irreducible region.
18915 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
18918 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
18919 assertions. Add support for doing the signbit if the IEEE 128-bit
18920 floating point value is in a GPR.
18921 * config/rs6000/rs6000.md (Fsignbit): Delete.
18922 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
18923 Update the length attribute if the value is in a GPR.
18924 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
18925 the sign or zero extension instruction, since the value is always 0/1.
18926 (signbit<mode>2_dm2): Delete using <Fsignbit>.
18929 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
18930 extracting SImode to a GPR register so that we can generate a
18931 store, limit the vector to be in a traditional Altivec register
18932 for the vextuwrx instruction.
18934 2017-01-03 Ian Lance Taylor <iant@google.com>
18936 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
18938 2017-01-03 Martin Sebor <msebor@redhat.com>
18940 PR tree-optimization/78696
18941 * gimple-ssa-sprintf.c (format_floating): Correct handling of
18942 precision. Use MPFR for %f for greater fidelity. Correct handling
18944 (pass_sprintf_length::compute_format_length): Set width and precision
18945 specified by asrerisk to void_node for vararg functions.
18946 (try_substitute_return_value): Adjust dump output.
18948 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
18950 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
18952 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
18954 * doc/invoke.texi (SPARC options): Document -mlra as the default.
18955 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
18956 -mlra/-mno-lra was passed to the compiler.
18958 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
18960 PR rtl-optimization/65618
18961 * emit-rtl.c (try_split): Move initialization of "before" and
18962 "after" to just before the call to emit_insn_after_setloc.
18964 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
18966 * doc/md.texi (Standard Names): Remove reference to Java frontend.
18968 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
18970 * dwarf2out.c (gen_enumeration_type_die): When
18971 -gno-strict-dwarf, add a DW_AT_encoding attribute.
18973 2017-01-03 Jakub Jelinek <jakub@redhat.com>
18975 PR tree-optimization/78965
18976 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
18977 Change first argument from const call_info & to call_info &. For %n
18978 set info.nowrite to false.
18980 PR middle-end/78901
18981 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
18982 possibly throwing calls.
18984 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
18985 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
18986 and fns handling, rather than in a separate case SSA_NAME.
18988 2017-01-02 Jeff Law <law@redhat.com>
18990 * config/darwin-driver.c (darwin_driver_init): Const-correctness
18991 fixes for first_period and second_period variables.
18993 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
18996 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
18997 (*insvqi_1): New insn pattern.
18998 (*insvqi_1_mem_rex64): Ditto.
18999 (*insvqi_2): Ditto.
19000 (*insvqi_3): Rename from *insvqi.
19002 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19004 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19006 * doc/cfg.texi (Edges): Remove reference to Java.
19007 (Maintaining the CFG): Ditto.
19009 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19011 PR middle-end/77674
19012 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19013 transparent aliases.
19015 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19017 PR middle-end/77484
19018 * predict.def (PRED_CALL): Update hitrate.
19019 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19020 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19021 into direct/indirect/polymorphic variants.
19023 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19025 Update copyright years.
19027 * gcc.c (process_command): Update copyright notice dates.
19028 * gcov-dump.c (print_version): Ditto.
19029 * gcov.c (print_version): Ditto.
19030 * gcov-tool.c (print_version): Ditto.
19031 * gengtype.c (create_file): Ditto.
19032 * doc/cpp.texi: Bump @copying's copyright year.
19033 * doc/cppinternals.texi: Ditto.
19034 * doc/gcc.texi: Ditto.
19035 * doc/gccint.texi: Ditto.
19036 * doc/gcov.texi: Ditto.
19037 * doc/install.texi: Ditto.
19038 * doc/invoke.texi: Ditto.
19040 Copyright (C) 2017 Free Software Foundation, Inc.
19042 Copying and distribution of this file, with or without modification,
19043 are permitted in any medium without royalty provided the copyright
19044 notice and this notice are preserved.