1 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9 * config/i386/gstabs.h: Delete.
10 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14 * config/i386/i386.c (ix86_decompose_address): Do not check for
15 register RTX when looking at index_reg or base_reg.
16 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21 to update EH info here.
23 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
25 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
27 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
29 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
31 2017-07-25 Torsten Duwe <duwe@suse.de>
33 * common.opt: Introduce -fpatchable-function-entry
34 command line option, and its variables function_entry_patch_area_size
35 and function_entry_patch_area_start.
36 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
37 including a two-value parser.
38 * target.def (print_patchable_function_entry): New target hook.
39 * targhooks.h (default_print_patchable_function_entry): New function.
40 * targhooks.c (default_print_patchable_function_entry): Likewise.
41 * toplev.c (process_options): Switch off IPA-RA if
42 patchable function entries are being generated.
43 * varasm.c (assemble_start_function): Look at the
44 patchable-function-entry command line switch and current
45 function attributes and maybe generate NOP instructions by
46 calling the print_patchable_function_entry hook.
47 * doc/extend.texi: Document patchable_function_entry attribute.
48 * doc/invoke.texi: Document -fpatchable_function_entry
50 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
52 * doc/tm.texi: Re-generate.
54 2017-07-25 Jakub Jelinek <jakub@redhat.com>
57 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
58 TARGET_AVX512DQ rather than TARGET_AVX512BW.
60 2017-07-25 Tamar Christina <tamar.christina@arm.com>
62 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
64 2017-07-25 Richard Biener <rguenther@suse.de>
66 PR tree-optimization/81455
67 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
68 not walk in cycles when looking for guards.
70 2017-07-25 Richard Biener <rguenther@suse.de>
72 PR tree-optimization/81529
73 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
74 when optimizing backedge uses.
76 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
78 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
80 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
81 to dl_section_ref. On AIX, append an expression to subtract
82 the size of the section length to dl_section_ref.
84 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
86 * configure.ac: If any of the config.* scripts fail, exit 1.
87 * configure: Regenerate.
89 2017-07-25 Richard Biener <rguenther@suse.de>
92 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
93 of immediate uses, be more verbose on errors.
95 2017-07-25 Richard Biener <rguenther@suse.de>
97 PR tree-optimization/81510
98 * tree-vect-loop.c (vect_is_simple_reduction): When the
99 reduction stmt is not inside the loop bail out.
101 2017-07-25 Richard Biener <rguenther@suse.de>
103 PR tree-optimization/81303
104 * tree-vect-loop-manip.c (vect_loop_versioning): Build
105 profitability check against LOOP_VINFO_NITERSM1.
107 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
109 * domwalk.c (cmp_bb_postorder): Simplify.
110 (sort_bbs_postorder): New function. Use it...
111 (dom_walker::walk): ...here to optimize common cases.
113 2017-07-25 Martin Liska <mliska@suse.cz>
116 * ipa-visibility.c (function_and_variable_visibility): Make the
117 redirection just on target that supports aliasing.
118 Fix GNU coding style.
120 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
123 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
125 * config/arm/rtems.h: Likewise.
126 * config/bfin/rtems.h: Likewise.
127 * config/i386/rtemself.h: Likewise.
128 * config/lm32/rtems.h: Likewise.
129 * config/m32c/rtems.h: Likewise.
130 * config/m68k/rtemself.h: Likewise.
131 * config/microblaze/rtems.h: Likewise.
132 * config/mips/rtems.h: Likewise.
133 * config/moxie/rtems.h: Likewise.
134 * config/nios2/rtems.h: Likewise.
135 * config/powerpcspe/rtems.h: Likewise.
136 * config/rs6000/rtems.h: Likewise.
137 * config/rtems.h: Likewise.
138 * config/sh/rtems.h: Likewise.
139 * config/sh/rtemself.h: Likewise.
140 * config/sparc/rtemself.h: Likewise.
142 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
145 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
146 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
147 * tree-ssa-structalias.c (alias_get_name): Same.
149 2017-07-25 Bin Cheng <bin.cheng@arm.com>
152 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
153 instructions if no du chain is found.
155 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
157 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
159 2017-07-25 Richard Biener <rguenther@suse.de>
162 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
165 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
167 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
169 (ISA_2_7_MASKS_SERVER): Likewise.
170 (ISA_3_0_MASKS_IEEE): Likewise.
171 (OTHER_P8_VECTOR_MASKS): Likewise.
172 (OTHER_VSX_VECTOR_MASKS): Likewise.
173 (POWERPC_MASKS): Likewise.
174 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
175 duplicate list of options.
176 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
177 explicit -mupper-regs options.
178 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
179 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
180 alias for -mupper-regs-df.
181 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
182 (rs6000_init_hard_regno_mode_ok): Likewise.
183 (rs6000_option_override_internal): Likewise.
184 (rs6000_opt_masks): Likewise.
185 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
186 options in terms of whether -mvsx or -mpower8-vector was used.
187 (TARGET_UPPER_REGS_DI): Likewise.
188 (TARGET_UPPER_REGS_SF): Likewise.
189 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
190 -mupper-regs-* options.
192 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
194 * passes.c (emergency_dump_function): Print some empty lines and a
195 header before the RTL dump.
197 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
199 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
201 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
204 * config/aarch64/aarch64.c (aarch64_classify_symbol):
205 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
207 2017-07-24 Carl Love <cel@us.ibm.com>
209 * config/rs6000/rs6000-c.c: Add support for built-in functions
210 vector float vec_extract_fp32_from_shorth (vector unsigned short);
211 vector float vec_extract_fp32_from_shortl (vector unsigned short);
212 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
213 vec_extract_fp_from_shortl): Add defines for the two builtins.
214 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
215 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
217 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
218 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
219 * doc/extend.texi: Update the built-in documentation file for the
220 new built-in function.
222 2017-07-24 Jakub Jelinek <jakub@redhat.com>
225 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
227 * doc/generic.texi: Likewise.
228 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
229 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
231 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
233 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
234 (aarch64_mls_elt_merge<mode>): Likewise.
236 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
238 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
241 2017-07-23 Michael Collison <michael.collison@arm.com>
243 * config/arm/arm.c (arm_option_override): Deprecate
244 use of -mstructure-size-boundary.
245 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
246 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
248 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
251 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
252 Reduce cost estimate for direct moves.
254 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
257 * config/i386/i386.c (ix86_option_override_internal): Disable
258 BMI, BMI2 and TBM instructions for -m16.
260 2017-07-21 Carl Love <cel@us.ibm.com>
262 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
263 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
264 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
266 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
267 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
268 VMULOSW): New enum "unspec" values.
269 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
270 altivec_vmulosw): New patterns.
271 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
272 VMULOSW): Add definitions.
274 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
276 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
278 * config/aarch64/aarch64-options-extensions.def (rdma); New.
279 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
280 (AARCH64_FL_V8_1): Renumber.
281 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
282 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
283 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
284 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
285 rdma to feature modifiers list.
287 2017-07-21 Yury Gribov <tetra2005@gmail.com>
290 * ipa-visibility (function_and_variable_visibility): Convert
291 recursive PLT call to direct call if appropriate.
293 2017-07-21 Andrew Pinski <apinski@cavium.com>
295 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
296 operand 1 to see if the types precision matches.
297 * fold-const.c (operand_equal_p): Likewise.
299 2017-07-21 Richard Biener <rguenther@suse.de>
301 PR tree-optimization/81303
302 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
303 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
304 (vect_peeling_hash_get_lowest_cost): Adjust.
305 (vect_enhance_data_refs_alignment): Likewise. Use
306 vect_get_peeling_costs_all_drs to compute the penalty for no
307 peeling to match up costs.
309 2017-07-21 Richard Biener <rguenther@suse.de>
311 PR tree-optimization/81500
312 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
313 we didn't identify a reduction path.
315 2017-07-21 Tom de Vries <tom@codesourcery.com>
316 Cesar Philippidis <cesar@codesourcery.com>
318 PR gcov-profile/81442
319 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
322 2017-07-21 Tom de Vries <tom@codesourcery.com>
325 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
327 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
328 nvptx_override_options_after_change.
330 2017-07-21 Ulrich Drepper <drepper@redhat.com>
332 * dwarf2out.c (output_file_names): Avoid double testing for
335 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
337 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
339 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
341 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
343 (try_crossjump_to_edge): Do not punt on partitioned functions.
345 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
347 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
348 Put all BBs reachable only via paths crossing cold region to cold
350 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
352 2016-07-21 Richard Biener <rguenther@suse.de>
354 PR tree-optimization/81303
355 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
356 into account prologue and epilogue iterations when raising
357 min_profitable_iters to sth at least covering one vector iteration.
359 2017-07-21 Tamar Christina <tamar.christina@arm.com>
361 * config/arm/arm.c (arm_test_cpu_arch_dat):
364 2017-07-20 Nathan Sidwell <nathan@acm.org>
367 * tree.h (TYPE_METHODS): Delete.
368 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
369 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
370 (dbxout_type_methods): Scan TYPE_FIELDS.
371 (dbxout_type): Don't check TYPE_METHODS here.
372 * function.c (use_register_for_decl): Always ignore register for
373 class types when not optimizing.
374 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
375 * tree.c (free_lang_data_in_type): Stitch out member functions and
376 templates from TYPE_FIELDS.
377 (build_distinct_type_copy, verify_type_variant,
378 verify_type): Member fns are on TYPE_FIELDS.
379 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
380 * tree-pretty-print.c (dump_generic_node): Likewise.
382 2017-07-20 Jakub Jelinek <jakub@redhat.com>
385 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
386 V2TImode and V4TImode.
387 (ix86_expand_vector_extract): Likewise.
388 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
389 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
390 (ssescalarmode): Handle V4TImode and V2TImode.
391 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
392 (*vec_extractv2ti, *vec_extractv4ti): New insns.
393 (VEXTRACTI128_MODE): New mode iterator.
394 (splitter for *vec_extractv?ti first element): New.
395 (VEC_INIT_MODE): New mode iterator.
396 (vec_init<mode>): Consolidate 3 expanders into one using
397 VEC_INIT_MODE mode iterator.
399 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
401 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
402 non_spilled_static_chain_regno_p.
404 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
406 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
408 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
410 * bb-reorder.c (connect_traces): Allow copying of blocks within
413 2017-07-20 Richard Biener <rguenther@suse.de>
415 * gimple.h (gimple_phi_result): Add gphi * overload.
416 (gimple_phi_result_ptr): Likewise.
417 (gimple_phi_arg): Likewise. Adjust index assert to only
418 allow actual argument accesses rather than all slots available
420 (gimple_phi_arg_def): Add gphi * overload.
421 * tree-phinodes.c (make_phi_node): Initialize only actual
423 (resize_phi_node): Clear memory not covered by old node,
424 do not initialize excess argument slots.
425 (reserve_phi_args_for_new_edge): Initialize new argument slot
428 2017-07-20 Bin Cheng <bin.cheng@arm.com>
430 PR tree-optimization/81388
432 2016-07-21 Bin Cheng <bin.cheng@arm.com>
434 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
435 by removing computation of may_be_zero.
437 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
438 Tom de Vries <tom@codesourcery.com>
441 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
442 when gimple level profile disagrees with what RTL expander did.
444 2017-07-20 Richard Biener <rguenther@suse.de>
446 PR tree-optimization/61171
447 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
448 (vect_analyze_stmt): Add slp instance parameter.
449 (vectorizable_reduction): Likewise.
450 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
451 (vect_is_simple_reduction): Deal with chains not detected
452 as SLP reduction chain, specifically not properly associated
453 chains containing a mix of plus/minus.
454 (get_reduction_op): Remove.
455 (get_initial_defs_for_reduction): Simplify, pass in whether
456 this is a reduction chain, pass in the SLP node for the PHIs.
457 (vect_create_epilog_for_reduction): Get the SLP instance as
459 (vectorizable_reduction): Get the SLP instance as arg.
460 During analysis remember the SLP node with the PHIs in the
461 instance. Simplify getting at the vectorized reduction PHIs.
462 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
463 through SLP instance.
464 (vect_slp_analyze_operations): Likewise.
465 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
466 (vect_transform_stmt): Likewise.
468 2017-07-20 Tom de Vries <tom@codesourcery.com>
470 PR tree-optimization/81489
471 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
472 read of phi arg location to before loop that modifies phi.
474 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
476 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
479 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
482 * except.c (execute): Fix ordering issue.
484 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
486 PR rtl-optimization/81423
487 * combine.c (make_compound_operation_int): Don't try to optimize
488 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
490 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
492 PR rtl-optimization/81423
493 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
494 with a constant that is -1 in the truncated to mode.
496 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
498 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
499 (determine_unlikely_bbs): ... here.
500 * predict.h (propagate_unlikely_bbs_forward): Declare.
501 * cfgexpand.c (pass_expand::execute): Use it.
502 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
504 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
505 propagate_unlikely_bbs_forward.
507 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
510 * except.c (maybe_add_nop_after_section_switch): New function.
513 2017-07-19 Tom de Vries <tom@codesourcery.com>
515 * gimple.h (gimple_phi_set_arg): Make assert more strict.
517 2017-07-19 Tom de Vries <tom@codesourcery.com>
519 * gimple.h (gimple_phi_arg): Make assert more strict.
521 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
523 * config.gcc (powerpc*-*-*): Add mmintrin.h.
524 * config/rs6000/mmintrin.h: New file.
525 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
527 2017-07-19 Jakub Jelinek <jakub@redhat.com>
529 PR tree-optimization/81346
530 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
532 2017-07-19 Tom de Vries <tom@codesourcery.com>
534 * config/nvptx/nvptx.md (VECIM): Add V2DI.
536 2017-07-19 Tom de Vries <tom@codesourcery.com>
538 * config/nvptx/nvptx-modes.def: Add V2DImode.
539 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
540 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
541 (nvptx_output_mov_insn): Handle lack of mov.b128.
542 (nvptx_print_operand): Handle 'H' and 'L' codes.
543 (nvptx_vector_mode_supported): Allow V2DImode.
544 (nvptx_preferred_simd_mode): New function.
545 (nvptx_data_alignment): New function.
546 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
547 nvptx_preferred_simd_mode.
548 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
550 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
552 2017-07-19 Tom de Vries <tom@codesourcery.com>
554 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
555 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
556 (nvptx_vector_mode_supported): New function. Allow V2SImode.
557 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
558 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
559 (mov<VECIM>_insn): New define_insn.
560 (define_expand "mov<VECIM>): New define_expand.
562 2017-07-19 Tom de Vries <tom@codesourcery.com>
564 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
566 2017-07-19 Jakub Jelinek <jakub@redhat.com>
568 PR tree-optimization/81346
569 * fold-const.h (fold_div_compare, range_check_type): Declare.
570 * fold-const.c (range_check_type): New function.
571 (build_range_check): Use range_check_type.
572 (fold_div_compare): No longer static, rewritten into
573 a match.pd helper function.
574 (fold_comparison): Don't call fold_div_compare here.
575 * match.pd (X / C1 op C2): New optimization using fold_div_compare
578 2017-07-19 Nathan Sidwell <nathan@acm.org>
580 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
581 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
582 * tree.c (find_decls_types_r, verify_type): Use
583 TYPE_{MIN,MAX}_VALUE_RAW.
584 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
585 (hash_tree): Likewise.
586 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
588 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
591 2017-07-18 Tom de Vries <tom@codesourcery.com>
594 * omp-expand.c (expand_omp_for_static_chunk): Handle
595 equal-argument loop exit phi.
597 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
600 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
601 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
603 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
605 (ror,rol -> rorx splitters): Use const_int_operand as
608 2017-06-18 Richard Biener <rguenther@suse.de>
610 PR tree-optimization/81410
611 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
612 the gap in the ! slp_perm SLP case after each group.
614 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
617 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
620 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
623 * predict.c (set_even_probabilities): Cleanup; do not affect
624 probabilities that are already known.
625 (combine_predictions_for_bb): Call even when count is set.
627 2017-07-18 Nathan Sidwell <nathan@acm.org>
629 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
632 2017-07-18 Bin Cheng <bin.cheng@arm.com>
635 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
636 optimization for loop niter analysis.
638 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
641 * config/avr/avr.c (avr_optimize_casesi): Don't use
642 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
644 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
646 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
647 body_cost_vec from _vect_peel_extended_info.
648 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
649 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
652 2017-07-18 Bin Cheng <bin.cheng@arm.com>
654 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
656 2017-07-18 Richard Biener <rguenther@suse.de>
658 PR tree-optimization/80620
659 PR tree-optimization/81403
660 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
661 info when re-using a VN table entry.
663 2017-07-18 Richard Biener <rguenther@suse.de>
665 PR tree-optimization/81418
666 * tree-vect-loop.c (vectorizable_reduction): Properly compute
667 vectype_in. Verify that with lane-reducing reduction operations
668 we have a single def-use cycle.
670 2017-07-17 Carl Love <cel@us.ibm.com>
672 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
674 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
675 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
676 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
677 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
678 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
679 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
680 VMULOSW): New enum "unspec" values.
681 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
682 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
683 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
684 altivec_vmulosw): New patterns.
685 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
686 VMULOSW): Add definitions.
687 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
689 * config/alpha/alpha.c: Include predict.h.
691 2017-07-17 Yury Gribov <tetra2005@gmail.com>
693 * tree-vrp.c (compare_assert_loc): Fix comparison function
694 to return predictable results.
696 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
698 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
701 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
702 * doc/invoke.texi (mexpand-adddi): Update text.
704 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
706 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
707 that also clobbers the CC register. The old expand code is moved
709 (*arc_clzsi2): ... here.
710 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
711 the CC register. The old expand code is moved to ...
712 (arc_ctzsi2): ... here.
714 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
716 * config/arc/arc.opt (mindexed-loads): Use initial value
717 TARGET_INDEXED_LOADS_DEFAULT.
718 (mauto-modify-reg): Use initial value
719 TARGET_AUTO_MODIFY_REG_DEFAULT.
720 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
721 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
722 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
723 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
725 2017-07-17 Martin Liska <mliska@suse.cz>
728 * opts.c (finish_options): Do not allow -fgnu-tm
729 w/ -fsanitize={kernel-,}address. Say sorry.
731 2017-07-17 Bin Cheng <bin.cheng@arm.com>
734 * tree-loop-distribution.c (classify_partition): Only assert on
736 (merge_dep_scc_partitions): Delete prameter. Update function call.
737 (distribute_loop): Remove code handling loop with unknown niters.
738 (pass_loop_distribution::execute): Skip loop with unknown niters.
740 2017-07-17 Bin Cheng <bin.cheng@arm.com>
743 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
744 function sort_partitions_by_post_order.
746 2017-07-17 Bin Cheng <bin.cheng@arm.com>
748 PR tree-optimization/81374
749 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
750 the max index of basic blocks, rather than number of basic blocks.
752 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
754 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
756 (arc_legitimate_pic_operand_p): Likewise.
757 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
759 (arc_needs_pcl_p): Likewise.
760 (arc_legitimate_pc_offset_p): Likewise.
761 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
762 function is also used in constrains.md.
763 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
764 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
765 PLUS. Only return true/false in known cases, otherwise assert.
766 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
767 is already called in arc_legitimate_constant_p.
768 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
770 (LEGITIMATE_PIC_OPERAND_P): Use
771 arc_raw_symbolic_reference_mentioned_p function.
772 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
777 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
778 Andrew Burgess <andrew.burgess@embecosm.com>
780 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
781 (arc_return_address_register): New function.
782 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
783 (arc_handle_fndecl_attribute): Add naked attribute.
784 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
785 (TARGET_WARN_FUNC_RETURN): Likewise.
786 (arc_allocate_stack_slots_for_args): New function.
787 (arc_warn_func_return): Likewise.
788 (machine_function): Change type fn_type.
789 (arc_compute_function_type): Consider new naked function type,
790 change function return type.
791 (arc_must_save_register): Adapt to handle new
792 arc_compute_function_type's return type.
793 (arc_expand_prologue): Likewise.
794 (arc_expand_epilogue): Likewise.
795 (arc_return_address_regs): Delete.
796 (arc_return_address_register): New function.
797 (arc_epilogue_uses): Use above function.
798 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
799 (arc_function_type): Change encoding, add naked type.
800 (ARC_INTERRUPT_P): Change to handle the new encoding.
801 (ARC_FAST_INTERRUPT_P): Likewise.
802 (ARC_NORMAL_P): Define.
803 (ARC_NAKED_P): Likewise.
804 (arc_compute_function_type): Delete prototype.
805 * config/arc/arc.md (in_ret_delay_slot): Use
806 arc_return_address_register function.
807 (simple_return): Likewise.
808 (p_return_i): Likewise.
810 2017-07-17 Jakub Jelinek <jakub@redhat.com>
812 PR tree-optimization/81428
813 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
814 can't be built for those types.
816 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
818 Remove stuff dead since r239246.
820 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
821 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
822 (avr_inform_devices): Remove dead stuff.
824 2017-07-17 Tamar Christina <tamar.christina@arm.com>
826 * config/arm/arm_neon.h: Fix softp typo.
828 2017-07-17 Jakub Jelinek <jakub@redhat.com>
830 PR tree-optimization/81365
831 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
832 aggregate moves onto bb predecessor edges, make sure there are no
833 loads that could alias the lhs in between the start of bb and the
836 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
839 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
840 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
841 [LSHIFTRT, outer_code = TRUNCATE]: Same.
843 2017-07-17 Jakub Jelinek <jakub@redhat.com>
845 PR tree-optimization/81396
846 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
847 (init_symbolic_number): Initialize it to 1.
848 (perform_symbolic_merge): Add n_ops from both operands into the new
850 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
851 without base_addr as useless if they need more than one operation.
852 (bswap_replace): Handle !bswap case for NULL base_addr.
854 2017-07-17 Tom de Vries <tom@codesourcery.com>
857 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
860 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
862 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
863 conditional builtin define __FIX_LEON3FT_B2BST.
865 2017-07-17 Daniel Cederman <cederman@gaisler.com>
867 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
868 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
871 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
873 PR rtl-optimization/81424
874 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
875 to remove potential trapping from operands if -fnon-call-exceptions.
877 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
879 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
880 profile_proability for scalling.
881 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
883 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
885 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
887 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
889 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
890 fixpoint arithmetics.
892 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
894 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
895 fixpoint arithmetics.
897 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
899 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
900 fixpoint arithmetics.
902 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
904 * profile-count.h (profile_probability::from_reg_br_prob_note,
905 profile_probability::to_reg_br_prob_note): New functions.
906 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
907 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
908 * predict.c (probability_reliable_p): Update.
909 (edge_probability_reliable_p): Update.
910 (br_prob_note_reliable_p): Update.
911 (invert_br_probabilities): Update.
912 (add_reg_br_prob_note): New function.
913 (combine_predictions_for_insn): Update.
914 * asan.c (asan_clear_shadow): Update.
915 * cfgbuild.c (compute_outgoing_frequencies): Update.
916 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
917 (update_br_prob_note): Update.
918 (rtl_verify_edges): Update.
919 (purge_dead_edges): Update.
920 (fixup_reorder_chain): Update.
921 * emit-rtl.c (try_split): Update.
922 * ifcvt.c (cond_exec_process_insns): Update.
923 (cond_exec_process_if_block): Update.
924 (dead_or_predicable): Update.
925 * internal-fn.c (expand_addsub_overflow): Update.
926 (expand_neg_overflow): Update.
927 (expand_mul_overflow): Update.
928 * loop-doloop.c (doloop_modify): Update.
929 * loop-unroll.c (compare_and_jump_seq): Update.
930 * optabs.c (emit_cmp_and_jump_insn_1): Update.
932 * reorg.c (mostly_true_jump): Update.
934 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
935 * config/alpha/alpha.c (emit_unlikely_jump): Update.
936 * config/arc/arc.c: (emit_unlikely_jump): Update.
937 * config/arm/arm.c: (emit_unlikely_jump): Update.
938 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
939 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
940 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
941 (ix86_print_operand): Update.
942 (ix86_split_fp_branch): Update.
943 (predict_jump): Update.
944 * config/ia64/ia64.c (ia64_print_operand): Update.
945 * config/mmix/mmix.c (mmix_print_operand): Update.
946 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
947 (rs6000_expand_split_stack_prologue): Update.
948 * config/rs6000/rs6000.c: Update.
949 * config/s390/s390.c (s390_expand_vec_strlen): Update.
950 (s390_expand_vec_movstr): Update.
951 (s390_expand_cs_tdsi): Update.
952 (s390_expand_split_stack_prologue): Update.
953 * config/sh/sh.c (sh_print_operand): Update.
954 (expand_cbranchsi4): Update.
955 (expand_cbranchdi4): Update.
956 * config/sparc/sparc.c (output_v9branch): Update.
957 * config/spu/spu.c (get_branch_target): Update.
958 (ea_load_store_inline): Update.
959 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
960 * config/tilepro/tilepro.c: Update.
962 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
964 * gimplify.c (mostly_copy_tree_r): Revert latest change.
965 (gimplify_save_expr): Likewise.
967 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
969 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
971 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
973 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
975 * timevar.def (TV_IPA_FNSUMMARY): Define.
977 2017-07-16 Daniel Cederman <cederman@gaisler.com>
979 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
980 to back store errata sensitive sequence from being generated.
981 (sqrtdf2_fix): Likewise.
983 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
985 * tree-ssa-threadupdate.c (compute_path_counts,
986 update_joiner_offpath_counts): Use profile_probability.
988 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
991 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
993 * config/arm/arm-c.c (arm_cpu_builtins): Define
994 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
996 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
998 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
999 array entries to represent __ieee128 versions of the
1000 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1001 scalar_extract_sig, and scalar_insert_exp built-in functions.
1002 (altivec_resolve_overloaded_builtin): Add special case handling
1003 for the __builtin_scalar_insert_exp function, as represented by
1004 the P9V_BUILTIN_VEC_VSIEDP constant.
1005 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1006 exponent support for __ieee128 argument.
1007 (VSESQP): Add scalar extract signature support for __ieee128
1009 (VSTDCNQP): Add scalar test negative support for __ieee128
1011 (VSIEQP): Add scalar insert exponent support for __int128 argument
1012 with __ieee128 result.
1013 (VSIEQPF): Add scalar insert exponent support for __ieee128
1014 argument with __ieee128 result.
1015 (VSTDCQP): Add scalar test data class support for __ieee128
1017 (VSTDCNQP): Add overload support for scalar test negative with
1019 (VSTDCQP): Add overload support for scalar test data class
1021 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1023 (UNSPEC_VSX_SIEXPQP): New constant.
1024 (xsxexpqp): New insn for VSX scalar extract exponent quad
1026 (xsxsigqp): New insn for VSX scalar extract significand quad
1028 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1029 precision with floating point argument.
1030 (xststdcqp): New expand for VSX scalar test data class quad
1032 (xststdcnegqp): New expand for VSX scalar test negative quad
1034 (xststdcqp): New insn to match expansions for VSX scalar test data
1035 class quad precision and VSX scalar test negative quad precision.
1036 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1037 special case operand checking to enforce that second operand of
1038 VSX scalar test data class with quad precision argument is a 7-bit
1040 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1041 prototypes and descriptions of __ieee128 versions of
1042 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1043 scalar_test_data_class, and scalar_test_neg built-in functions.
1045 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1047 PR tree-optimization/81162
1048 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1049 replace a negate with an add.
1051 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1053 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1055 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1057 * config/arm/arm-c.c (arm_cpu_builtins): Define
1058 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1060 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1062 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1063 (armv8-r): Set ARM Cortex-R52 as default CPU.
1064 * config/arm/arm-tables.opt: Regenerate.
1065 * config/arm/arm-tune.md: Regenerate.
1066 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1068 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1069 extension for -mcpu=cortex-r52.
1071 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1073 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1074 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1075 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1076 (fp-armv8): Define it as FP_ARMv8 only.
1077 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1078 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1080 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1081 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1082 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1083 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1084 than TARGET_FPU_ARMV8.
1085 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1086 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1087 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1089 * config/arm/neon.md (neon_vrint): Likewise.
1090 (neon_vcvt): Likewise.
1091 (neon_<fmaxmin_op><mode>): Likewise.
1092 (<fmaxmin><mode>3): Likewise.
1093 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1094 * config/arm/predicates.md (arm_cond_move_operator): Check against
1095 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1097 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1099 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1102 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1104 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1105 loop in comment with memset.
1107 2017-07-14 Martin Liska <mliska@suse.cz>
1109 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1110 * dwarf2out.c (is_java): Remove the function.
1111 (output_pubname): Remove usage of the function.
1112 (lower_bound_default): Remove usage of DW_LANG_Java.
1113 (gen_compile_unit_die): Likewise.
1114 * gcc.c: Remove compiler defaults for .java and .zip files.
1115 * gimple-expr.c (remove_suffix): Change as there's no longer
1116 extension than 4-letter one.
1117 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1118 (gimplify_save_expr): Likewise.
1119 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1120 as it's possible even for other languages than Java.
1121 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1122 * lto-opts.c (lto_write_options): Remove reference to Java.
1123 * opts.c (strip_off_ending): Update file extension handling.
1124 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1125 * tree-eh.c (lower_resx): Likewise.
1126 * tree.c (free_lang_data_in_type): Remove dead code.
1127 (find_decls_types_r): Likewise.
1128 (build_common_builtin_nodes): Remove Java from a comment.
1129 (verify_type): Remove dead code.
1130 * varasm.c (assemble_external): Remove Java from a comment.
1132 2017-07-14 Martin Liska <mliska@suse.cz>
1134 * opts.c (finish_options): Add quotes.
1135 (common_handle_option): Likewise.
1137 2017-07-14 Martin Liska <mliska@suse.cz>
1139 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1140 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1142 * dwarf2out.c (lower_bound_default): Do not handle
1144 (gen_compile_unit_die): Likewise.
1145 * gcc.c: Remove default extension binding for GNU Pascal.
1146 * stmt.c: Remove Pascal language from a comment.
1147 * xcoffout.c: Likewise.
1149 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1152 * diagnostic-show-locus.c (fixit_cmp): New function.
1153 (layout::layout): Sort m_fixit_hints.
1154 (column_range::column_range): Assert that the values are valid.
1155 (struct char_span): New struct.
1156 (correction::overwrite): New method.
1157 (struct source_line): New struct.
1158 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1159 calls in terms of classes source_line and char_span, and
1160 correction::overwrite.
1161 (selftest::test_overlapped_fixit_printing_2): New function.
1162 (selftest::diagnostic_show_locus_c_tests): Call it.
1164 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1166 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1167 early if there is no lhs.
1169 2017-07-13 Martin Liska <mliska@suse.cz>
1171 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1172 (gen_reference_type_die): Likewise.
1173 * stor-layout.c: Remove Pascal-related comment.
1175 2017-07-13 Martin Liska <mliska@suse.cz>
1177 * opts.c (finish_options): Add quotes to error messages.
1178 (parse_sanitizer_options): Likewise.
1180 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1182 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1184 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1186 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1188 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1190 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1192 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1194 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1197 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1198 provides the hardware capability bits, define the macro
1199 __BUILTIN_CPU_SUPPORTS__.
1200 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1201 if GLIBC does not provide the hardware capability bits. Add a
1202 gcc_unreachable call if the built-in cpu function is neither
1203 __builtin_cpu_is nor __builtin_cpu_supports.
1204 (rs6000_get_function_versions_dispatcher): Change the warning
1205 that an old GLIBC is used which does not export the capability
1206 bits to be an error.
1207 * doc/extend.texi (target_clones attribute): Document the
1208 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1209 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1210 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1211 the macros defined by GCC if the newer GLIBC is available.
1213 2017-07-12 Jeff Law <law@redhat.com>
1215 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1216 remaining includes slightly.
1217 * config/riscv/riscv-builtins.c: Include profile-count.h.
1219 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1222 * config/avr/avr.c (avr_set_current_function): In diagnostic
1223 messages: Quote keywords and (parts of) identifiers.
1224 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1227 2017-07-12 Carl Love <cel@us.ibm.com>
1229 * config/rs6000/rs6000-c.c: Add support for built-in functions
1230 vector bool char vec_revb (vector bool char);
1231 vector bool short vec_revb (vector short char);
1232 vector bool int vec_revb (vector bool int);
1233 vector bool long long vec_revb (vector bool long long);
1234 * doc/extend.texi: Update the built-in documentation file for the
1235 new built-in functions.
1237 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1239 * config/s390/s390.md: Remove movcc splitter.
1241 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1243 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1244 load/store on condition.
1246 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1249 * config/avr/avr.c (avr_encode_section_info)
1250 [progmem && !TREE_READONLY]: Error if progmem object needs
1253 2017-07-11 Michael Collison <michael.collison@arm.com>
1255 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1258 2017-07-11 Carl Love <cel@us.ibm.com>
1260 * config/rs6000/rs6000-c.c: Add support for builtins
1261 vector unsigned int vec_parity_lsbb (vector signed int);
1262 vector unsigned int vec_parity_lsbb (vector unsigned int);
1263 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1264 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1265 vector unsigned long long vec_parity_lsbb (vector signed long long);
1266 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1267 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1268 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1269 * doc/extend.texi: Update the built-in documentation file for the
1270 new built-in functions.
1272 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1274 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1275 (layout::m_primary_loc): New field.
1276 (layout::layout): Initialize new field. Move location filtering
1277 logic from here to...
1278 (layout::maybe_add_location_range): ...this new method. Add
1279 support for filtering to just the lines already specified by other
1281 (layout::will_show_line_p): New method.
1282 (gcc_rich_location::add_location_if_nearby): New method.
1283 (selftest::test_add_location_if_nearby): New test function.
1284 (selftest::diagnostic_show_locus_c_tests): Call it.
1285 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1288 2017-07-11 Tom de Vries <tom@codesourcery.com>
1290 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1291 (bb_first_real_insn): New function.
1292 (nvptx_single): Add extra initialization of broadcasted condition
1295 2017-07-11 Nathan Sidwell <nathan@acm.org>
1297 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1299 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1301 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1302 Binutils doc as TEXI will mess them up.
1303 * doc/invoke.texi (AVR Options): Same here.
1305 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1307 * config/sparc/sparc.opt (mfix-ut700): New option.
1308 (mfix-gr712rc): Likewise.
1309 (sparc_fix_b2bst): New variable.
1310 * doc/invoke.texi (SPARC options): Document them.
1311 (ARM options): Fix warnings.
1312 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1313 instructions to prevent sequences that can trigger the store-store
1314 errata for certain LEON3FT processors.
1315 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1316 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1317 * config/sparc/sparc.md (fix_b2bst): New attribute.
1318 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1320 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1323 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1325 (*rsqrtsf2_sse): Ditto.
1327 (div<mode>3): Macroize insn from divdf3 and divsf3
1328 using MODEF mode iterator.
1330 2017-07-10 Martin Sebor <msebor@redhat.com>
1332 PR tree-optimization/80397
1333 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1334 instead of testing for equality to INTEGER_TYPE.
1336 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1338 * config.gcc: Remove uclibc from arc target spec.
1340 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1342 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1344 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1347 * lto-wrapper.c (remove_option): New function.
1348 (merge_and_complain): Merge PIC/PIE options more realistically.
1350 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1352 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1356 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1357 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1358 * config.in: Regenerate.
1359 * configure: Regenerate.
1360 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1361 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1362 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1363 (TARGET_GASISR_PROLOGUES): ...target mask.
1364 * common/config/avr/avr-common.c
1365 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1366 Set -mgas-isr-prologues.
1367 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1368 INSERT_PASS_BEFORE for it.
1369 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1370 * config/avr/avr.c (avr_option_override)
1371 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1372 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1373 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1374 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1375 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1377 (make_avr_pass_pre_proep): New function.
1378 (emit_push_sfr) <treg>: Add argument to function and use it
1380 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1381 and set machine->gasisr.yes.
1382 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1383 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1384 __gcc_isr.n_pushed to .L__stack_usage.
1385 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1386 (avr_asm_final_postscan_insn): ...this new static function.
1387 * config/avr/avr.h (machine_function)
1388 <is_no_gccisr, use_L__stack_usage>: New fields.
1389 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1390 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1391 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1392 (gasisr, *gasisr): New expander and insn.
1393 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1394 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1395 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1397 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1399 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1402 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1404 Move jump-tables out of .text again.
1407 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1408 (ASM_OUTPUT_ADDR_VEC): New function.
1409 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1410 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1411 INSN_ADDRESSes as asm comment.
1412 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1413 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1414 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1415 * config/avr/avr.md (*tablejump): Adjust comment.
1416 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1417 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1419 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1420 (avr_output_addr_vec): New proto.
1421 (avr_log_t) <insn_addresses>: New field.
1423 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1426 * config/i386/i386.c (ix86_function_arg_advance): Set
1427 outgoing_args_on_stack to true if there are outgoing arguments
1429 (ix86_function_arg): Likewise.
1430 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1431 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1432 * config/i386/i386.h (machine_function): Add
1433 outgoing_args_on_stack.
1435 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1437 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1439 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1441 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1443 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1444 (REAL_H): Remove $(MACHMODE_H).
1445 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1447 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1448 $(MACHMODE_H) and double-int.h.
1449 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1451 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1454 2017-07-07 Andrew Pinski <apinski@cavium.com>
1456 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1457 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1459 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1461 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1462 Add warning if GCC was not configured to link against a GLIBC that
1463 exports the hardware capability bits.
1464 (make_resolver_func): Make resolver function private and not a
1465 COMDAT function. Create the name with clone_function_name instead
1466 of make_unique_name.
1469 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1470 correct operand in doing the split.
1472 2017-07-07 Carl Love <cel@us.ibm.com>
1474 * config/rs6000/rs6000-c: Add support for built-in function
1475 vector unsigned short vec_pack_to_short_fp32 (vector float,
1477 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1478 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1479 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1480 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1481 (convert_4f32_8i16): Add define_expand.
1482 * doc/extend.texi: Update the built-in documentation file for the
1483 new built-in function.
1485 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1487 * config/sparc/m8.md: New file.
1488 * config/sparc/sparc.md: Include m8.md.
1490 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1492 * config/sparc/sparc.opt: New option -mvis4b.
1493 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1494 (sparc_option_override): Handle VIS4B.
1495 (enum sparc_builtins): Define
1496 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1497 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1498 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1499 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1500 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1501 (check_constant_argument): New function.
1502 (sparc_vis_init_builtins): Define builtins
1503 __builtin_vis_dictunpack{8,16,32},
1504 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1505 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1506 __builtin_vis_fpcmpde{8,16,32}shl and
1507 __builtin_vis_fpcmpur{8,16,32}shl.
1508 (sparc_expand_builtin): Check that the constant operands to
1509 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1510 constant and in range.
1511 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1513 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1514 (SPARC_IMM5_P): Likewise.
1515 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1516 (enabled): Handle vis4b.
1517 (UNSPEC_DICTUNPACK): New unspec.
1518 (UNSPEC_FPCMPSHL): Likewise.
1519 (UNSPEC_FPUCMPSHL): Likewise.
1520 (UNSPEC_FPCMPDESHL): Likewise.
1521 (UNSPEC_FPCMPURSHL): Likewise.
1522 (cpu_feature): New CPU feature `vis4b'.
1523 (dictunpack{8,16,32}): New insns.
1524 (FPCSMODE): New mode iterator.
1525 (fpcscond): New code iterator.
1526 (fpcsucond): Likewise.
1527 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1528 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1529 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1530 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1531 * config/sparc/constraints.md: Define constraints `q' for unsigned
1532 2-bit integer constants and `t' for unsigned 5-bit integer
1534 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1536 (imm5_operand_dictunpack16): Likewise.
1537 (imm5_operand_dictunpack32): Likewise.
1538 (imm2_operand): Likewise.
1539 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1540 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1541 ditunpack* and fpcmp*shl builtins.
1543 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1545 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1546 * config.in: Add HAVE_AS_SPARC6 define.
1547 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1549 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1551 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1552 (CPP_CPU_SPEC): Handle m8.
1553 (ASM_CPU_SPEC): Likewise.
1554 * config/sparc/sparc-opts.h (enum processor_type): Add
1556 * config/sparc/sparc.c (m8_costs): New struct.
1557 (sparc_option_override): Handle TARGET_CPU_m8.
1558 (sparc32_initialize_trampoline): Likewise.
1559 (sparc64_initialize_trampoline): Likewise.
1560 (sparc_issue_rate): Likewise.
1561 (sparc_register_move_cost): Likewise.
1562 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1563 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1564 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1565 (CPP_CPU_SPEC): Handle M8.
1566 (ASM_CPU_SPEC): Likewise.
1567 (AS_M8_FLAG): Define.
1568 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1569 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1570 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1572 * configure: Regenerate.
1573 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1576 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1578 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1580 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1581 ("*movdi_insn_sp32"): Do not set v3pipe.
1582 ("*movsi_insn"): Likewise.
1583 ("*movdi_insn_sp64"): Likewise.
1584 ("*movsf_insn"): Likewise.
1585 ("*movdf_insn_sp32"): Likewise.
1586 ("*movdf_insn_sp64"): Likewise.
1587 ("*zero_extendsidi2_insn_sp64"): Likewise.
1588 ("*sign_extendsidi2_insn"): Likewise.
1589 ("*mov<VM32:mode>_insn"): Likewise.
1590 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1591 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1592 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1593 ("<vlop:code><VL:mode>3"): Likewise.
1594 ("*not_<vlop:code><VL:mode>3"): Likewise.
1595 ("*nand<VL:mode>_vis"): Likewise.
1596 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1597 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1598 ("one_cmpl<VL:mode>2"): Likewise.
1599 ("faligndata<VM64:mode>_vis"): Likewise.
1600 ("alignaddrsi_vis"): Likewise.
1601 ("alignaddrdi_vis"): Likweise.
1602 ("alignaddrlsi_vis"): Likewise.
1603 ("alignaddrldi_vis"): Likewise.
1604 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1605 ("bmaskdi_vis"): Likewise.
1606 ("bmasksi_vis"): Likewise.
1607 ("bshuffle<VM64:mode>_vis"): Likewise.
1608 ("cmask8<P:mode>_vis"): Likewise.
1609 ("cmask16<P:mode>_vis"): Likewise.
1610 ("cmask32<P:mode>_vis"): Likewise.
1611 ("pdistn<P:mode>_vis"): Likewise.
1612 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1614 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1616 * config/sparc/sparc.md ("subtype"): New insn attribute.
1617 ("*wrgsr_sp64"): Set insn subtype.
1618 ("*rdgsr_sp64"): Likewise.
1619 ("alignaddrsi_vis"): Likewise.
1620 ("alignaddrdi_vis"): Likewise.
1621 ("alignaddrlsi_vis"): Likewise.
1622 ("alignaddrldi_vis"): Likewise.
1623 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1624 ("fexpand_vis"): Likewise.
1625 ("fpmerge_vis"): Likewise.
1626 ("faligndata<VM64:mode>_vis"): Likewise.
1627 ("bshuffle<VM64:mode>_vis"): Likewise.
1628 ("cmask8<P:mode>_vis"): Likewise.
1629 ("cmask16<P:mode>_vis"): Likewise.
1630 ("cmask32<P:mode>_vis"): Likewise.
1631 ("fchksm16_vis"): Likewise.
1632 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1633 ("fmean16_vis"): Likewise.
1634 ("fp<plusminus_insn>64_vis"): Likewise.
1635 ("<plusminus_insn>v8qi3"): Likewise.
1636 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1637 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1638 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1639 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1640 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1641 ("*movqi_insn"): Likewise.
1642 ("*movhi_insn"): Likewise.
1643 ("*movsi_insn"): Likewise.
1644 ("movsi_pic_gotdata_op"): Likewise.
1645 ("*movdi_insn_sp32"): Likewise.
1646 ("*movdi_insn_sp64"): Likewise.
1647 ("movdi_pic_gotdata_op"): Likewise.
1648 ("*movsf_insn"): Likewise.
1649 ("*movdf_insn_sp32"): Likewise.
1650 ("*movdf_insn_sp64"): Likewise.
1651 ("*zero_extendhisi2_insn"): Likewise.
1652 ("*zero_extendqihi2_insn"): Likewise.
1653 ("*zero_extendqisi2_insn"): Likewise.
1654 ("*zero_extendqidi2_insn"): Likewise.
1655 ("*zero_extendhidi2_insn"): Likewise.
1656 ("*zero_extendsidi2_insn_sp64"): Likewise.
1657 ("ldfsr"): Likewise.
1658 ("prefetch_64"): Likewise.
1659 ("prefetch_32"): Likewise.
1660 ("tie_ld32"): Likewise.
1661 ("tie_ld64"): Likewise.
1662 ("*tldo_ldub_sp32"): Likewise.
1663 ("*tldo_ldub1_sp32"): Likewise.
1664 ("*tldo_ldub2_sp32"): Likewise.
1665 ("*tldo_ldub_sp64"): Likewise.
1666 ("*tldo_ldub1_sp64"): Likewise.
1667 ("*tldo_ldub2_sp64"): Likewise.
1668 ("*tldo_ldub3_sp64"): Likewise.
1669 ("*tldo_lduh_sp32"): Likewise.
1670 ("*tldo_lduh1_sp32"): Likewise.
1671 ("*tldo_lduh_sp64"): Likewise.
1672 ("*tldo_lduh1_sp64"): Likewise.
1673 ("*tldo_lduh2_sp64"): Likewise.
1674 ("*tldo_lduw_sp32"): Likewise.
1675 ("*tldo_lduw_sp64"): Likewise.
1676 ("*tldo_lduw1_sp64"): Likewise.
1677 ("*tldo_ldx_sp64"): Likewise.
1678 ("*mov<VM32:mode>_insn"): Likewise.
1679 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1680 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1682 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1684 * config/sparc/sparc.md ("type"): New insn type viscmp.
1685 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1687 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1688 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1689 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1690 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1692 ("n7_vis_logical_11cycle"): Likewise.
1693 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1694 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1695 * config/sparc/niagara.md ("niag_vis"): Likewise.
1696 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1697 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1699 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1701 * config/sparc/sparc.md: New instruction type `bmask'.
1702 (bmaskdi_vis): Use the `bmask' type.
1703 (bmasksi_vis): Likewise.
1704 * config/sparc/ultra3.md (us3_array): Likewise.
1705 * config/sparc/niagara7.md (n7_array): Likewise.
1706 * config/sparc/niagara4.md (n4_array): Likewise.
1707 * config/sparc/niagara2.md (niag2_vis): Likewise.
1708 (niag3_vis): Likewise.
1709 * config/sparc/niagara.md (niag_vis): Likewise.
1711 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1713 * ipa-comdats.c: Remove optimize check from gate.
1714 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1715 for functions not optimized.
1716 (ipa_fn_summary_read): Skip optimize check.
1717 (ipa_fn_summary_write): Likewise.
1718 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1720 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1722 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1723 (check_callers): Give up on caller which is not optimized.
1724 (inline_small_functions): Likewise.
1725 (ipa_inline): Do not give up when not optimizing.
1726 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1727 away unoptimizes cdtors.
1728 (whole_program_function_and_variable_visibility): Do
1729 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1730 * ipa.c (process_references): Do not check optimize.
1731 (symbol_table::remove_unreachable_nodes): Update optimize check.
1732 (set_writeonly_bit): Update optimize check.
1733 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1734 (pass_ipa_single_use::gate): Remove.
1736 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1738 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1739 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1740 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1741 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1742 permute_load, permute_store, adjust_extract, adjust_splat,
1743 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1744 replace_swap_with_copy, dump_swap_insn_table,
1745 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1746 recombine_lvx_pattern, recombine_stvx_pattern,
1747 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1748 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1749 to file rs6000-p8swap.c.
1750 * config/rs6000/rs6000-p8swap.c: New file.
1751 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1752 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1753 and rs6000*-*-* targets.
1755 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1757 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1759 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1761 * lto-wrapper.c (merge_and_complain): Do not merge
1762 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
1763 fsigned_zeros, ftrapping_math, fwrapv.
1764 (append_compiler_options): Do not track these options.
1765 (append_linker_options): Likewie
1767 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1769 * cgraphunit.c (cgraph_node::finalize_function): When
1770 !flag_toplevel_reorde set no_reorder flag.
1771 (varpool_node::finalize_decl): Likewise.
1772 (symbol_table::compile): Drop no toplevel reorder path.
1774 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1776 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
1777 edges; zero probability is not better than uninitialized.
1779 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1781 * asan.h (asan_sanitize_allocas_p): Declare.
1782 * asan.c (asan_sanitize_allocas_p): New function.
1783 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
1784 (handle_builtin_alloca): Likewise.
1785 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
1786 if !asan_sanitize_allocas_p.
1787 * params.def (asan-instrument-allocas): Add new option.
1788 * params.h (ASAN_PROTECT_ALLOCAS): Define.
1789 * opts.c (common_handle_option): Disable allocas sanitization for
1792 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1794 * asan.c: Include gimple-fold.h.
1795 (get_last_alloca_addr): New function.
1796 (handle_builtin_stackrestore): Likewise.
1797 (handle_builtin_alloca): Likewise.
1798 (asan_emit_allocas_unpoison): Likewise.
1799 (get_mem_refs_of_builtin_call): Add new parameter, remove const
1800 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
1801 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
1802 (instrument_builtin_call): Pass gimple iterator to
1803 get_mem_refs_of_builtin_call.
1804 (last_alloca_addr): New global.
1805 * asan.h (asan_emit_allocas_unpoison): Declare.
1806 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
1807 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
1808 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
1809 if function calls alloca.
1810 * gimple-fold.c (replace_call_with_value): Remove static keyword.
1811 * gimple-fold.h (replace_call_with_value): Declare.
1812 * internal-fn.c: Include asan.h.
1813 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
1814 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
1816 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1818 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
1819 (C_SELFTEST_FLAGS): New.
1820 (CPP_SELFTEST_FLAGS): New.
1821 (SELFTEST_DEPS): New, from deps of s-selftest.
1822 (C_SELFTEST_DEPS): New, from deps of s-selftest.
1823 (CPP_SELFTEST_DEPS): New.
1824 (selftest): Add dependency on s-selftest-c++.
1825 (s-selftest): Rename to...
1826 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
1827 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
1828 than SELFTEST_FLAGS.
1829 (selftest-gdb): Rename to...
1830 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
1832 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
1833 (selftest-valgrind): Rename to...
1834 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
1836 (selftest-valgrind): Reintroduce as an alias for
1837 selftest-c-valgrind.
1838 (s-selftest-c++): New.
1839 (selftest-c++-gdb): New.
1840 (selftest-c++-valgrind): New.
1842 2017-07-06 Olivier Hainque <hainque@adacore.com>
1844 * gcc.c (process_command): When deciding if undefined variables
1845 should be ignored when processing specs, accept "gcc -v" as well.
1847 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1849 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
1850 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
1852 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1854 * config/arm/arm-cpus.in (armv8-r): Add new entry.
1855 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
1856 * config/arm/arm-tables.opt: Regenerate.
1857 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
1859 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
1861 2017-07-06 Carl Love <cel@us.ibm.com>
1863 * ChangeLog: Clean up from mid air collision
1865 2017-07-06 Carl Love <cel@us.ibm.com>
1867 * config/rs6000/rs6000-c.c: Add support for built-in functions
1868 vector signed int vec_subc (vector signed int, vector signed int);
1869 vector signed __int128 vec_subc (vector signed __int128,
1870 vector signed __int128);
1871 vector unsigned __int128 vec_subc (vector unsigned __int128,
1872 vector unsigned __int128);
1873 vector signed int vec_sube (vector signed int, vector signed int,
1875 vector unsigned int vec_sube (vector unsigned int,
1876 vector unsigned int,
1877 vector unsigned int);
1878 vector signed __int128 vec_sube (vector signed __int128,
1879 vector signed __int128,
1880 vector signed__int128);
1881 vector unsigned __int128 vec_sube (vector unsigned __int128,
1882 vector unsigned __int128,
1883 vector unsigned __int128);
1884 vector signed int vec_subec (vector signed int, vector signed int,
1886 vector unsigned int vec_subec (vector unsigned int,
1887 vector unsigned int,
1888 vector unsigned int);
1889 vector signed __int128 vec_subec (vector signed __int128,
1890 vector signed __int128,
1891 vector signed__int128);
1892 vector unsigned __int128 vec_subec (vector unsigned __int128,
1893 vector unsigned __int128,
1894 vector unsigned __int128);
1895 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
1896 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
1897 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
1898 BU_ALTIVEC_OVERLOAD_X definitions.
1899 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
1900 * doc/extend.texi: Update the built-in documentation file for the new
1903 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1906 * diagnostic-show-locus.c (layout::layout): Use start and finish
1907 spelling location for the start and finish of each range.
1908 * genmatch.c (linemap_client_expand_location_to_spelling_point):
1909 Add unused aspect param.
1910 * input.c (expand_location_1): Add "aspect" param, and use it
1911 to access the correct part of the location.
1912 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
1914 (expand_location_to_spelling_point): Likewise.
1915 (linemap_client_expand_location_to_spelling_point): Add "aspect"
1916 param, and pass it to expand_location_1.
1918 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
1920 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
1921 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
1922 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
1923 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
1924 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
1925 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
1926 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
1927 _mm_maskz_getmant_ss): New intrinsics.
1928 (__builtin_ia32_getexpss128_mask): Changed to ...
1929 __builtin_ia32_getexpss128_round ... this.
1930 (__builtin_ia32_getexpsd128_mask): Changed to ...
1931 __builtin_ia32_getexpsd128_round ... this.
1932 * config/i386/i386-builtin-types.def
1933 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
1934 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
1935 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
1936 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
1937 __builtin_ia32_getmantss_mask_round): New builtins.
1938 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
1939 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
1940 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
1941 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
1942 * config/i386/sse.md
1943 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
1944 avx512f_sgetexp<mode><mask_scalar_name>
1945 <round_saeonly_scalar_name> ... this.
1946 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
1947 %0, %1, %2<round_saeonly_op3>}): Changed to ...
1948 vgetexp<ssescalarmodesuffix>
1949 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1950 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
1951 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
1952 avx512f_vgetmant<mode><mask_scalar_name>
1953 <round_saeonly_scalar_name> ... this.
1954 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
1955 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
1956 vgetmant<ssescalarmodesuffix>
1957 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
1958 %0<mask_scalar_operand4>, %1, %2
1959 <round_saeonly_scalar_mask_op4>, %3} ... this.
1960 * config/i386/subst.md (mask_scalar_operand4,
1961 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
1962 round_saeonly_scalar_nimm_predicate): New subst attributes.
1964 2017-07-06 Julia Koval <julia.koval@intel.com>
1966 * config/i386/i386.c (ix86_erase_embedded_rounding):
1967 Remove code for old rounding pattern.
1969 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
1971 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
1973 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
1975 * doc/sourcebuild.texi (Test Directives, Variants of
1976 dg-require-support): Add documentation for dg-require-stack-check.
1978 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
1980 * config/i386/subst.md (mask_scalar, round_scalar,
1981 round_saeonly_scalar): New meta-templates.
1982 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
1983 round_scalar_mask_operand3, round_scalar_mask_op3,
1984 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
1985 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
1986 round_saeonly_scalar_constraint,
1987 round_saeonly_scalar_prefix): New subst attribute.
1988 * config/i386/sse.md
1989 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
1990 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
1991 <round_scalar_name> ... this.
1992 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
1993 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
1994 <round_scalar_name> ... this.
1995 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
1996 <sse>_vm<code><mode>3<mask_scalar_name>
1997 <round_saeonly_scalar_name> ... this.
1998 (v<plusminus_mnemonic><ssescalarmodesuffix>
1999 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2000 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2001 v<plusminus_mnemonic><ssescalarmodesuffix>
2002 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2003 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2004 (v<multdiv_mnemonic><ssescalarmodesuffix>
2005 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2006 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2007 v<multdiv_mnemonic><ssescalarmodesuffix>
2008 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2009 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2010 (v<maxmin_float><ssescalarmodesuffix>
2011 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2012 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2013 v<maxmin_float><ssescalarmodesuffix>
2014 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2015 %0<mask_scalar_operand3>, %1, %<iptr>2
2016 <round_saeonly_scalar_mask_op3>} ... this.
2018 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2020 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2021 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2023 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2024 Alan Hayward <alan.hayward@arm.com>
2025 David Sherwood <david.sherwood@arm.com>
2027 * combine.c (simplify_if_then_else): Remove "enum" before
2029 * compare-elim.c (can_eliminate_compare): Likewise.
2030 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2032 (aarch64_lookup_simd_builtin_type): Likewise.
2033 (aarch64_simd_builtin_type): Likewise.
2034 (aarch64_init_simd_builtin_types): Likewise.
2035 (aarch64_simd_expand_args): Likewise.
2036 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2038 (aarch64_reverse_mask): Likewise.
2039 (aarch64_simd_emit_reg_reg_move): Likewise.
2040 (aarch64_gen_adjusted_ldpstp): Likewise.
2041 (aarch64_ccmp_mode_to_code): Likewise.
2042 (aarch64_operands_ok_for_ldpstp): Likewise.
2043 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2044 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2046 (aarch64_min_divisions_for_recip_mul): Likewise.
2047 (aarch64_reassociation_width): Likewise.
2048 (aarch64_get_condition_code_1): Likewise.
2049 (aarch64_simd_emit_reg_reg_move): Likewise.
2050 (aarch64_simd_attr_length_rglist): Likewise.
2051 (aarch64_reverse_mask): Likewise.
2052 (aarch64_operands_ok_for_ldpstp): Likewise.
2053 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2054 (aarch64_gen_adjusted_ldpstp): Likewise.
2055 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2057 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2058 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2059 (arm_lookup_simd_builtin_type): Likewise.
2060 (arm_simd_builtin_type): Likewise.
2061 (arm_init_simd_builtin_types): Likewise.
2062 (arm_expand_builtin_args): Likewise.
2063 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2064 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2065 (ft32_setup_incoming_varargs): Likewise.
2066 (ft32_function_arg): Likewise.
2067 (ft32_function_arg_advance): Likewise.
2068 (ft32_pass_by_reference): Likewise.
2069 (ft32_arg_partial_bytes): Likewise.
2070 (ft32_valid_pointer_mode): Likewise.
2071 (ft32_addr_space_pointer_mode): Likewise.
2072 (ft32_addr_space_legitimate_address_p): Likewise.
2073 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2075 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2076 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2077 (iamcu_alignment): Likewise.
2078 (canonicalize_vector_int_perm): Likewise.
2079 (ix86_noce_conversion_profitable_p): Likewise.
2080 (ix86_mpx_bound_mode): Likewise.
2081 (ix86_operands_ok_for_move_multiple): Likewise.
2082 * config/microblaze/microblaze-protos.h
2083 (microblaze_expand_conditional_branch_reg): Likewise.
2084 * config/microblaze/microblaze.c
2085 (microblaze_expand_conditional_branch_reg): Likewise.
2086 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2088 (rs6000_reassociation_width): Likewise.
2089 (rs6000_invalid_binary_op): Likewise.
2090 (fusion_p9_p): Likewise.
2091 (emit_fusion_p9_load): Likewise.
2092 (emit_fusion_p9_store): Likewise.
2093 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2095 (riscv_hard_regno_mode_ok_p): Likewise.
2096 (riscv_address_insns): Likewise.
2097 (riscv_split_symbol): Likewise.
2098 (riscv_legitimize_move): Likewise.
2099 (riscv_function_value): Likewise.
2100 (riscv_hard_regno_nregs): Likewise.
2101 (riscv_expand_builtin): Likewise.
2102 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2103 (riscv_build_integer): Likewise.
2104 (riscv_split_integer): Likewise.
2105 (riscv_legitimate_constant_p): Likewise.
2106 (riscv_cannot_force_const_mem): Likewise.
2107 (riscv_regno_mode_ok_for_base_p): Likewise.
2108 (riscv_valid_base_register_p): Likewise.
2109 (riscv_valid_offset_p): Likewise.
2110 (riscv_valid_lo_sum_p): Likewise.
2111 (riscv_classify_address): Likewise.
2112 (riscv_legitimate_address_p): Likewise.
2113 (riscv_address_insns): Likewise.
2114 (riscv_load_store_insns): Likewise.
2115 (riscv_force_binary): Likewise.
2116 (riscv_split_symbol): Likewise.
2117 (riscv_force_address): Likewise.
2118 (riscv_legitimize_address): Likewise.
2119 (riscv_move_integer): Likewise.
2120 (riscv_legitimize_const_move): Likewise.
2121 (riscv_legitimize_move): Likewise.
2122 (riscv_address_cost): Likewise.
2123 (riscv_subword): Likewise.
2124 (riscv_output_move): Likewise.
2125 (riscv_canonicalize_int_order_test): Likewise.
2126 (riscv_emit_int_order_test): Likewise.
2127 (riscv_function_arg_boundary): Likewise.
2128 (riscv_pass_mode_in_fpr_p): Likewise.
2129 (riscv_pass_fpr_single): Likewise.
2130 (riscv_pass_fpr_pair): Likewise.
2131 (riscv_get_arg_info): Likewise.
2132 (riscv_function_arg): Likewise.
2133 (riscv_function_arg_advance): Likewise.
2134 (riscv_arg_partial_bytes): Likewise.
2135 (riscv_function_value): Likewise.
2136 (riscv_pass_by_reference): Likewise.
2137 (riscv_setup_incoming_varargs): Likewise.
2138 (riscv_print_operand): Likewise.
2139 (riscv_elf_select_rtx_section): Likewise.
2140 (riscv_save_restore_reg): Likewise.
2141 (riscv_for_each_saved_reg): Likewise.
2142 (riscv_register_move_cost): Likewise.
2143 (riscv_hard_regno_mode_ok_p): Likewise.
2144 (riscv_hard_regno_nregs): Likewise.
2145 (riscv_class_max_nregs): Likewise.
2146 (riscv_memory_move_cost): Likewise.
2147 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2148 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2149 (rl78_addr_space_address_mode): Likewise.
2150 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2152 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2153 (rs6000_reassociation_width): Likewise.
2154 (rs6000_invalid_binary_op): Likewise.
2155 (fusion_p9_p): Likewise.
2156 (emit_fusion_p9_load): Likewise.
2157 (emit_fusion_p9_store): Likewise.
2158 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2159 (ok_for_simple_move_operands): Likewise.
2160 (ok_for_simple_move_strict_operands): Likewise.
2161 (ok_for_simple_arith_logic_operands): Likewise.
2162 (visium_legitimize_reload_address): Likewise.
2163 (visium_select_cc_mode): Likewise.
2164 (output_cbranch): Likewise.
2165 (visium_split_double_move): Likewise.
2166 (visium_expand_copysign): Likewise.
2167 (visium_expand_int_cstore): Likewise.
2168 (visium_expand_fp_cstore): Likewise.
2169 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2170 (visium_function_arg): Likewise.
2171 (visium_function_arg_advance): Likewise.
2172 (visium_libcall_value): Likewise.
2173 (visium_setup_incoming_varargs): Likewise.
2174 (visium_legitimate_constant_p): Likewise.
2175 (visium_legitimate_address_p): Likewise.
2176 (visium_legitimize_address): Likewise.
2177 (visium_secondary_reload): Likewise.
2178 (visium_register_move_cost): Likewise.
2179 (visium_memory_move_cost): Likewise.
2180 (prepare_move_operands): Likewise.
2181 (ok_for_simple_move_operands): Likewise.
2182 (ok_for_simple_move_strict_operands): Likewise.
2183 (ok_for_simple_arith_logic_operands): Likewise.
2184 (visium_function_value_1): Likewise.
2185 (rtx_ok_for_offset_p): Likewise.
2186 (visium_legitimize_reload_address): Likewise.
2187 (visium_split_double_move): Likewise.
2188 (visium_expand_copysign): Likewise.
2189 (visium_expand_int_cstore): Likewise.
2190 (visium_expand_fp_cstore): Likewise.
2191 (visium_split_cstore): Likewise.
2192 (visium_select_cc_mode): Likewise.
2193 (visium_split_cbranch): Likewise.
2194 (output_cbranch): Likewise.
2195 (visium_print_operand_address): Likewise.
2196 * expmed.c (flip_storage_order): Likewise.
2197 * expmed.h (emit_cstore): Likewise.
2198 (flip_storage_order): Likewise.
2199 * genrecog.c (validate_pattern): Likewise.
2200 * hsa-gen.c (gen_hsa_addr): Likewise.
2201 * internal-fn.c (expand_arith_overflow): Likewise.
2202 * ira-color.c (allocno_copy_cost_saving): Likewise.
2203 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2204 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2205 (process_invariant_for_inheritance): Likewise.
2206 * lra-eliminations.c (move_plus_up): Likewise.
2207 * omp-low.c (lower_oacc_reductions): Likewise.
2208 * simplify-rtx.c (simplify_subreg): Likewise.
2209 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2210 (TARGET_CHKP_BOUND_MODE): Likewise..
2211 * targhooks.c (default_chkp_bound_mode): Likewise.
2212 (default_setup_incoming_vararg_bounds): Likewise.
2213 * targhooks.h (default_chkp_bound_mode): Likewise.
2214 (default_setup_incoming_vararg_bounds): Likewise.
2215 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2216 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2217 (have_whole_vector_shift): Likewise.
2218 * tree-vect-stmts.c (vectorizable_load): Likewise.
2219 * doc/tm.texi: Regenerate.
2221 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2223 Graceful degrade if Binutils PR21472 is not available.
2226 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2227 .rodata in flash test fails.
2228 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2229 * confgure: Regenerate.
2230 * config.in: Regenerate.
2231 * config/avr/avr.c (avr_asm_named_section)
2232 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2233 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2234 (avr_asm_init_sections): Same.
2236 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2238 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2239 (fma<VH:mode>4_intrinsic): Likewise.
2240 (*fmsub<VCVTF:mode>4): Likewise.
2241 (*fmsub<VH:mode>4_intrinsic): Likewise.
2243 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2246 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2247 Don't depend on "optimize > 0".
2248 (out_movhi_r_mr, out_movqi_mr_r): Same.
2249 (out_movhi_mr_r, out_movqi_r_mr): Same.
2250 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2251 io_address_operand on "optimize > 0".
2253 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2255 * tree-loop-distribution.c: Add general explanantion on the pass.
2256 (generate_loops_for_partition): Mark distributed loop.
2257 (pg_add_dependence_edges): New parameter. Handle alias data
2258 dependence specially and record it in the parameter if asked.
2259 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2260 (init_partition_graph_vertices, add_partition_graph_edge): New.
2261 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2262 (free_partition_graph_vdata, build_partition_graph): New.
2263 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2264 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2265 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2266 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2267 (version_for_distribution_p, finalize_partitions): New.
2268 (distribute_loop): Handle alias data dependence specially. Factor
2269 out loop fusion code as functions and call these functions.
2271 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2273 * tree-loop-distribution.c (classify_partition): New parameter and
2274 better handle reduction statement.
2275 (rdg_build_partitions): Revise comment.
2276 (distribute_loop): Compute statements in all partitions and pass it
2277 to classify_partition.
2279 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2281 * tree-loop-distribution.c (enum partition_type): New.
2282 (struct partition): New field type.
2283 (partition_merge_into): Add parameter. Update partition type.
2284 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2285 (build_rdg_partition_for_vertex): Compute partition type.
2286 (rdg_build_partitions): Dump partition type.
2287 (distribute_loop): Update calls to partition_merge_into.
2289 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2291 * tree-loop-distribution.c (struct ddr_hasher): New.
2292 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2294 (classify_partition): Call get_data_dependence.
2295 (pg_add_dependence_edges): Ditto.
2296 (distribute_loop): Release data dependence hash table.
2298 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2300 * tree-loop-distribution.c (ref_base_address): Delete.
2301 (similar_memory_accesses): Rename ...
2302 (share_memory_accesses): ... to this. Check if partitions access
2303 the same memory reference.
2304 (distribute_loop): Call share_memory_accesses.
2306 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2308 * tree-loop-distribution.c (struct partition): New field recording
2310 (partition_alloc, partition_free): Init and release data refs.
2311 (partition_merge_into): Merge data refs.
2312 (build_rdg_partition_for_vertex): Collect data refs for partition.
2313 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2315 (distribute_loop): Remve data refs from vertice data of partition
2318 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2320 * tree-loop-distribution.c (params.h): Include header file.
2321 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2322 (datarefs_vec): New global var.
2323 (create_rdg_vertices): Use datarefs_vec directly.
2324 (free_rdg): Don't free data references.
2325 (build_rdg): Update use. Don't free data references.
2326 (distribute_loop): Compute global variable for data references.
2327 Bail out if there are too many data references.
2329 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2331 * tree-loop-distribution.c (loop_nest): New global var.
2332 (build_rdg): Use loop directly, rather than loop nest.
2333 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2335 (distribute_loop): Compute global variable loop nest. Update use.
2337 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2339 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2340 (partition_merge_into): New parameter. Dump reason for fusion.
2341 (distribute_loop): Update use of partition_merge_into.
2343 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2345 * tree-loop-distribution.c (bb_top_order_index): New.
2346 (bb_top_order_index_size, bb_top_order_cmp): New.
2347 (stmts_from_loop): Use topological order.
2348 (pass_loop_distribution::execute): Compute and release topological
2349 order for basic blocks.
2351 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2353 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2358 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2359 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2360 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2361 * internal-fn.def (LOOP_DIST_ALIAS): New.
2362 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2363 (fold_loop_internal_call): ... this.
2364 (vect_loop_dist_alias_call): New function.
2365 (set_uid_loop_bbs): Call fold_loop_internal_call.
2366 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2369 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2372 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2373 Require dead FLAGS_REG at the beginning of a peephole.
2375 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2378 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2379 arguments in the call to __builtin_ia32_sbb_u32.
2380 (_subborrow_u64): Swap _X and _Y arguments in the call to
2381 __builtin_ia32_sbb_u64.
2383 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2386 * tree-vrp.c (compare_assert_loc): Turn into a function template
2387 with stable template parameter. Only test if a->e is NULL,
2388 !a->e == !b->e has been verified already. Use e == NULL or
2389 e != NULL instead of e or ! e tests. If stable is true, don't use
2390 iterative_hash_expr, on the other side allow a or b or both NULL
2391 and sort the NULLs last.
2392 (process_assert_insertions): Sort using compare_assert_loc<false>
2393 instead of compare_assert_loc, later sort using
2394 compare_assert_loc<true> before calling process_assert_insertions_for
2395 in a loop. Use break instead of continue once seen NULL pointer.
2397 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2399 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2400 Cortex-R7 and Cortex-R8 processors.
2402 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2404 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2405 uninitialized while src is not.
2407 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2409 * common/config/arm/arm-common.c: Adjust include path for
2411 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2412 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2413 (arm-cpu-data.h): Likewise.
2414 (arm-cpu-cdata.h): Likewise.
2415 * config/arm/arm-cpu.h: Delete.
2416 * config/arm/arm-cpu-cdata.h: Delete.
2417 * config/arm/arm-cpu-data.h: Delete.
2419 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2421 * config/arm/arm-cpus.in (cortex-a55): New.
2422 (cortex-a75): Likewise.
2423 (cortex-a75.cortex-a55): Likewise.
2424 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2426 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2427 * config/arm/arm-cpu-cdata.h: Regenerate.
2428 * config/arm/arm-cpu-data.h: Regenerate.
2429 * config/arm/arm-cpu.h: Regenerate.
2430 * config/arm/arm-tables.opt: Regenerate.
2431 * config/arm/arm-tune.md: Regenerate.
2433 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2435 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2437 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2439 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2442 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2444 PR tree-optimization/81292
2445 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2446 full_string_p, also call adjust_related_strinfos if the adjustment
2447 is simple, otherwise invalidate related strinfos.
2449 2017-07-04 Martin Liska <mliska@suse.cz>
2452 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2453 newly created variable as DECL_IGNORED_P.
2455 2017-07-04 Martin Liska <mliska@suse.cz>
2458 * ipa-inline.c (inline_small_functions):
2459 Use xstrdup_for_dump.
2461 2017-07-04 Tom de Vries <tom@codesourcery.com>
2463 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2465 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2468 * config/darwin.c (darwin_function_switched_text_sections):
2471 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2473 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2475 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2477 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2479 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2481 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2482 min_profitable_iters, and th as inclusive lower bounds.
2483 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2484 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2485 for min_profitable_iters and min_profitable_estimate.
2486 (vect_transform_loop): Treat th as an inclusive lower bound.
2487 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2489 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2492 * config/darwin.c (darwin_function_switched_text_sections):
2493 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2494 in two pieces, and suppress the use of buf.
2496 2017-07-03 Nathan Sidwell <nathan@acm.org>
2498 * hash-table.h (hash_table_mod1): Fix indentation.
2500 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2503 * predict.c (force_edge_cold): Be more careful about propagation
2505 * profile-count.h (profile_probability::guessed,
2506 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2508 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2510 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2512 * doc/invoke.texi (rcpc architecture extension): Document it.
2514 2017-07-03 Richard Biener <rguenther@suse.de>
2516 PR tree-optimization/60510
2517 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2518 the scalar reduction PHI and use it.
2519 (vectorizable_reduction): Properly guard the single_defuse_cycle
2520 path for non-SLP reduction chains where we cannot use it.
2521 Rework reduc_def/index and vector type deduction. Rework
2522 vector operand gathering during reduction op code-gen.
2523 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2524 chains dissolve the chain and leave it to non-SLP reduction
2527 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2529 * tree-data-ref.h (dr_alignment): Declare.
2530 * tree-data-ref.c (dr_alignment): New function.
2531 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2532 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2534 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2536 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2538 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2539 and base_misalignment fields.
2540 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2541 * tree-data-ref.c: Include builtins.h.
2542 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2543 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2544 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2545 * tree-vect-data-refs.c: Include tree-cfg.h.
2546 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2547 fields instead of calculating an alignment here.
2548 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2549 innermost_loop_behavior fields.
2551 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2553 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2555 (DR_STEP_ALIGNMENT): New macro.
2556 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2557 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2558 (create_data_ref): Print it.
2559 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2560 to tell whether the step preserves vector (mis)alignment.
2561 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2562 Move the check for an integer step and generalise to all INTEGER_CST.
2563 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2564 Print the outer step alignment.
2566 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2568 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2569 with offset_alignment.
2570 (DR_ALIGNED_TO): Delete.
2571 (DR_OFFSET_ALIGNMENT): New macro.
2572 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2573 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2574 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2575 (create_data_ref): Likewise.
2576 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2577 (vect_analyze_data_refs): Likewise.
2578 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2579 creating dummy innermost behavior.
2581 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2583 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2584 with a "innermost_loop_behavior *" and refeence tree.
2585 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2586 (create_data_ref): Update call accordingly.
2587 * tree-predcom.c (find_looparound_phi): Likewise.
2589 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2591 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2592 fields with dr_wrt_vec_loop.
2593 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2594 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2595 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2596 (vect_dr_behavior): New function.
2597 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2598 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2599 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2600 track whether the step preserves the misalignment.
2601 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2602 Use vect_dr_behavior.
2603 (vect_setup_realignment): Update call accordingly.
2604 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2605 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2606 call to vect_create_addr_base_for_vector_ref.
2607 (vect_create_cond_for_align_checks): Likewise.
2608 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2609 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2610 (vect_recog_mask_conversion_pattern): Likewise.
2611 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2612 (new_stmt_vec_info): Remove redundant zeroing.
2614 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2616 * common/config/arm/arm-common.c (arm_be8_option): New function.
2617 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2618 (ISA_ARMv6): Add isa_bit_be8.
2619 * config/arm/arm.h (arm_be8_option): Add prototype.
2620 (BE8_SPEC_FUNCTION): New define.
2621 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2622 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2623 (mlittle-endian): Similarly.
2624 (mbe8, mbe32): New options.
2625 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2626 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2628 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2630 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2632 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2634 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2635 (cleanup_tree_cfg_bb): Use it.
2636 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2638 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2640 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2643 * loop-doloop.c (add_test): Update profile.
2645 2017-07-03 Martin Liska <mliska@suse.cz>
2648 * sanopt.c (rewrite_usage_of_param): New function.
2649 (sanitize_rewrite_addressable_params): Likewise.
2650 (pass_sanopt::execute): Call rewrite_usage_of_param.
2652 2017-07-03 Richard Biener <rguenther@suse.de>
2654 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2655 back to using VIEW_CONVERT_EXPR.
2657 2017-07-03 Martin Liska <mliska@suse.cz>
2660 * doc/extend.texi: Document when a resolver function is
2661 generated for target_clones.
2663 2017-07-03 Martin Liska <mliska@suse.cz>
2665 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2666 and shadow memory of auto variables which are subject of
2667 use-after-scope sanitization.
2668 (asan_expand_mark_ifn): Add do set only when is_poison.
2670 2016-07-03 Richard Biener <rguenther@suse.de>
2672 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2674 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2675 (vectorizable_reduction): Perform reduction PHI creation when
2676 visiting a reduction PHI and adjust and simplify code generation
2677 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2678 (vect_transform_loop): Visit reduction PHIs.
2679 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2680 defs into the SLP tree.
2681 (vect_build_slp_tree): Reduction defs terminate the recursion.
2682 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2684 (vect_get_vec_defs_for_stmt_copy): Export.
2685 (vect_get_vec_defs): Likewise.
2686 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2688 (vect_get_vec_defs_for_stmt_copy): Declare.
2689 (vect_get_vec_defs): Likewise.
2691 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2693 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2694 parameter with a "loop" parameter and use it instead of the
2695 loop containing DR_STMT. Don't check simple_iv when doing
2696 BB analysis. Describe the two analysis modes in the comment.
2698 2017-07-03 Tom de Vries <tom@codesourcery.com>
2700 PR tree-optimization/69468
2701 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2702 (find_same_succ_bb): Handle ignore_edge_flags.
2704 2017-07-03 Tom de Vries <tom@codesourcery.com>
2706 PR tree-optimization/81192
2707 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2709 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2711 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2713 2017-07-03 Tom de Vries <tom@codesourcery.com>
2715 PR tree-optimization/81192
2716 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2717 BB_SAME_SUCC (bb) == NULL.
2719 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2721 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2724 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2726 * dumpfile.c: Include profile-count.h
2727 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2729 (insert_cond_bb): Update profile.
2730 * tree-cfg.h (insert_cond_bb): Update prototype.
2731 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2732 * tree-dump.c: Do not include tree-cfg.
2734 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2736 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2738 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2740 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2743 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2745 * tree-complex.c (expand_complex_div_wide): update profile.
2747 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2752 (CORETYPES_H): New define.
2753 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2754 (insn-modes-inline.h, s-modes-inline-h): New rules.
2755 (generated_files): Add insn-modes-inline.h.
2756 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2757 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2758 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2759 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2760 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2761 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2762 (build/gencodes.o, build/genconditions.o): Likewise.
2763 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
2764 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
2765 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
2766 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
2767 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
2768 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
2769 * coretypes.h: Include everything up to real.h for generators.
2770 Include insn-modes.h first. Include wide-int-print.h after
2771 wide-int.h. Include insn-modes-inline.h and then machmode.h.
2772 * machmode.h: Don't include insn-modes.h here.
2773 * function-tests.c: Remove includes of signop.h, machmode.h,
2774 double-int.h and wide-int.h.
2776 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
2778 * optc-save-gen.awk: Likewise.
2779 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
2780 * godump.c: Remove include of wide-int-print.h.
2781 * pretty-print.h: Likewise.
2782 * wide-int-print.cc: Likewise.
2783 * wide-int.cc: Likewise.
2784 * hash-map-tests.c: Remove include of signop.h.
2785 * hash-set-tests.c: Likewise.
2786 * rtl-tests.c: Likewise.
2787 * mkconfig.sh: Remove include of machmode.h.
2788 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
2790 (emit_insn_modes_inline_h): ...this new function. Emit the code
2791 into an insn-modes-inline.h header file, adding appropriate
2792 include guards and end comments.
2793 (emit_insn_modes_c_header): Remove include of machmode.h.
2794 (emit_min_insn_modes_c_header): Include coretypes.h rather than
2796 (main): Handle -i flag and call emit_insn_modes_inline_h when
2799 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2801 * tree-ssa-strlen.c (strinfo): Rename the length field to
2802 nonzero_chars. Add a full_string_p field.
2803 (compare_nonzero_chars, zero_length_string_p): New functions.
2804 (get_addr_stridx): Add an offset_out parameter.
2805 Use compare_nonzero_chars.
2806 (get_stridx): Update accordingly. Use compare_nonzero_chars.
2807 (new_strinfo): Update after above changes to strinfo.
2808 (set_endptr_and_length): Set full_string_p.
2809 (get_string_length): Update after above changes to strinfo.
2810 (unshare_strinfo): Update call to new_strinfo.
2811 (maybe_invalidate): Likewise.
2812 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
2813 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
2814 as a uhwi instead of an shwi. Update after above changes to
2815 strinfo and new_strinfo.
2816 (zero_length_string): Assert that chainsi contains full strings.
2817 Use zero_length_string_p. Update call to new_strinfo.
2818 (adjust_related_strinfos): Update after above changes to strinfo.
2819 Copy full_string_p from origsi.
2820 (adjust_last_stmt): Use zero_length_string_p.
2821 (handle_builtin_strlen): Update after above changes to strinfo and
2822 new_strinfo. Install the lhs as the string length if the previous
2823 entry didn't describe a full string.
2824 (handle_builtin_strchr): Update after above changes to strinfo
2826 (handle_builtin_strcpy): Likewise.
2827 (handle_builtin_strcat): Likewise.
2828 (handle_builtin_malloc): Likewise.
2829 (handle_pointer_plus): Likewise.
2830 (handle_builtin_memcpy): Likewise. Track nonzero characters
2831 that aren't necessarily followed by a nul terminator.
2832 (handle_char_store): Likewise.
2834 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2836 PR tree-optimization/80769
2837 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
2838 for malloc and calloc. Document the new invariant that all related
2839 strinfos have delayed lengths or none do.
2840 (verify_related_strinfos): Move earlier in file.
2841 (set_endptr_and_length): New function, split out from...
2842 (get_string_length): ...here. Also set the lengths of related
2844 (zero_length_string): Assert that chainsi has known (rather than
2846 (adjust_related_strinfos): Likewise.
2848 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2850 PR tree-optimization/81136
2851 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
2852 assert that two references with the same misalignment have the same
2853 compile-time misalignment if those compile-time misalignments
2856 2017-07-01 Andi Kleen <ak@linux.intel.com>
2858 * print-tree.c (print_node): Print all attributes.
2860 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2862 * cfg.c (scale_bbs_frequencies): New function.
2863 * cfg.h (scale_bbs_frequencies): Declare it.
2864 * cfgloopanal.c (single_likely_exit): Cleanup.
2865 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
2867 (scale_loop_profile): Likewise.
2868 (loop_version): Likewise.
2869 (create_empty_loop_on_edge): Update.
2870 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
2871 scale_loop_frequencies, scale_loop_profile, loopify,
2872 loop_version): Update prototypes.
2873 * modulo-sched.c (sms_schedule): Update.
2874 * predict.c (unlikely_executed_edge_p): Also check probability.
2875 (probably_never_executed_edge_p): Fix typo.
2876 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2877 * tree-parloops.c (gen_parallel_loop): Update.
2878 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
2879 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2880 * tree-ssa-loop-split.c (split_loop): Update.
2881 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2882 * tree-vect-loop-manip.c (vect_do_peeling): Update.
2883 (vect_loop_versioning): Update.
2884 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2886 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2888 * trans-mem.c (split_bb_make_tm_edge): Update profile.
2890 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2892 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
2893 to keep profile consistent.
2895 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2897 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
2898 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
2899 * profile-count.h (max_safe_multiplier): Make unsigned.
2900 (profile_count::guessed_zero): New.
2902 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2904 * bb-reorder.c (fix_up_crossing_landing_pad,
2905 fix_crossing_conditional_branches): Use make_single_succ_edge
2906 to keep profile consistent.
2908 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2910 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
2913 2017-07-01 Jakub Jelinek <jakub@redhat.com>
2916 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
2917 the right scopes, make sure cond_jump isn't preserved between multiple
2918 iterations. Search for fallthru edge whenever there are 3+ edges and
2919 use find_fallthru_edge for it.
2921 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2923 Patch by Alexander Monakov <amonakov@ispras.ru>
2924 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
2925 probabilities consistently.
2927 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2929 * pa.c (pa_expand_compare_and_swap_loop): Update call of
2930 emit_cmp_and_jump_insns.
2932 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2935 * tree-inline.c (expand_call_inline): Combine profile statuses.
2937 2017-06-30 Andrew Pinski <apinski@cavium.com>
2939 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
2940 fold_stmt returned true.
2942 2017-06-30 Nathan Sidwell <nathan@acm.org>
2944 * ggc.h (empty_string): Delete.
2945 * cfgexpand.c (expand_asm_stmt): Use plain "".
2946 * optabs.c (expand_asm_memory_barrier): Likewise.
2947 * stringpool.c (empty_string): Delete.
2948 (digit_vector, digit_string): Delete.
2949 (ggc_alloc_string): Use plain "", don't optimize single digit
2950 strings. Use ggc_alloc_atomic.
2952 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
2954 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
2955 comparison set and one other set, use the cost of the non-comparison
2958 2017-06-30 Nathan Sidwell <nathan@acm.org>
2960 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
2963 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
2965 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
2966 loops. Remove now unneeded calls to gimple_switch_set_label() that
2967 just set removed labels to NULL_TREE.
2969 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
2971 * tree-ssanames.c (set_range_info_raw): Abstract from ...
2972 (set_range_info): ...here. Only call set_range_info_raw if domain
2974 (set_nonzero_bits): Call set_range_info_raw.
2975 * tree-ssanames.h (set_range_info_raw): New.
2977 2017-06-30 Jakub Jelinek <jakub@redhat.com>
2980 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
2981 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
2982 of nonimmediate_operand and <store_mask_constraint> instead of m
2983 for the input operand. For V8FI iterator, always split if input
2984 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
2985 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
2986 <store_mask_predicate> instead of register_operand and
2987 <store_mask_constraint> instead of v for the input operand. Make
2988 sure both operands aren't MEMs for if not <mask_applied>.
2990 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
2992 * lto-wrapper.c (copy_file) Close both file descriptors before
2995 2017-06-30 Martin Liska <mliska@suse.cz>
2998 * multiple_target.c (create_dispatcher_calls): Make ifunc
2999 also for function that don't have calls or are not referenced.
3001 2017-06-30 Richard Biener <rguenther@suse.de>
3003 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3004 analyze the first scalar stmt. Move vector type computation
3005 for the BB case here from ...
3006 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3007 live operation processing in the SLP case properly.
3009 2017-06-30 Richard Biener <rguenther@suse.de>
3011 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3013 2017-06-30 Martin Liska <mliska@suse.cz>
3016 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3017 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3019 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3021 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3024 2017-06-30 Richard Biener <rguenther@suse.de>
3026 PR tree-optimization/81249
3027 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3028 condition reduction result to original scalar type.
3030 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3032 * profile-count.h (enum profile_quality): Fix typos and whitespace
3035 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3037 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3038 type for branch probabilities.
3040 2017-06-29 Julian Brown <julian@codesourcery.com>
3041 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3043 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3044 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3045 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3046 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3048 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3050 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3051 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3052 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3053 CC usage from generic code to here.
3054 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3055 CC usage into the target macros.
3057 2017-06-29 Maya Rashish <coypu@sdf.org>
3059 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3062 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3064 * arm/arm-builtins.c: Include profile-count.h
3065 * except.c (sjlj_emit_function_enter): Use
3066 profile_probability::unlikely.
3068 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3070 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3071 and tocrel_offset be pointer args rather than implicitly using
3073 (legitimate_constant_pool_address_p, rs6000_emit_move,
3074 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3075 tocrel_offset and use in toc_relative_expr_p call.
3076 (print_operand, print_operand_address): Use static tocrel_base_oac
3077 and tocrel_offset_oac.
3078 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3081 2017-06-29 Maya Rashish <coypu@sdf.org>
3083 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3085 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3087 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3088 objects, take into account only the alignment of 'op0' and 'mode1' if
3091 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3093 * ccmp.c (ccmp_tree_comparison_p): New function.
3094 (ccmp_candidate_p): Update to use above function.
3095 (get_compare_parts): New function.
3096 (expand_ccmp_next): Update to use new functions.
3097 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3099 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3100 take mode as argument.
3101 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3102 * expr.c (expand_expr_real_1): Pass mode as argument.
3104 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3106 * combine.c (combine_instructions): Print insns to dump_file, together
3109 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3111 * asan.c (asan_emit_stack_protection): Update.
3112 (create_cond_insert_point): Update.
3113 * auto-profile.c (afdo_propagate_circuit): Update.
3114 * basic-block.h (struct edge_def): Turn probability to
3115 profile_probability.
3116 (EDGE_FREQUENCY): Update.
3117 * bb-reorder.c (find_traces_1_round): Update.
3118 (better_edge_p): Update.
3119 (sanitize_hot_paths): Update.
3120 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3121 (make_single_succ_edge): Update.
3122 (check_bb_profile): Update.
3123 (dump_edge_info): Update.
3124 (update_bb_profile_for_threading): Update.
3125 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3126 probabilitycount to 0.
3127 * cfgbuild.c (compute_outgoing_frequencies): Update.
3128 * cfgcleanup.c (try_forward_edges): Update.
3129 (outgoing_edges_match): Update.
3130 (try_crossjump_to_edge): Update.
3131 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3132 (expand_gimple_tailcall): Update.
3133 (construct_init_block): Use make_single_succ_edge.
3134 (construct_exit_block): Use make_single_succ_edge.
3135 * cfghooks.c (verify_flow_info): Update.
3136 (redirect_edge_succ_nodup): Update.
3137 (split_edge): Update.
3138 (account_profile_record): Update.
3139 * cfgloopanal.c (single_likely_exit): Update.
3140 * cfgloopmanip.c (scale_loop_profile): Update.
3141 (set_zero_probability): Remove.
3142 (duplicate_loop_to_header_edge): Update.
3143 * cfgloopmanip.h (loop_version): Update prototype.
3144 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3145 (force_nonfallthru_and_redirect): Update.
3146 (update_br_prob_note): Update.
3147 (rtl_verify_edges): Update.
3148 (purge_dead_edges): Update.
3149 (rtl_lv_add_condition_to_bb): Update.
3150 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3151 * cgraphunit.c (init_lowered_empty_function): Update.
3152 (cgraph_node::expand_thunk): Update.
3153 * cilk-common.c: Include profile-count.h
3154 * dojump.c (inv): Remove.
3155 (jumpifnot): Update.
3156 (jumpifnot_1): Update.
3157 (do_jump_1): Update.
3159 (do_jump_by_parts_greater_rtx): Update.
3160 (do_compare_rtx_and_jump): Update.
3161 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3162 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3163 * dwarf2cfi.c: Include profile-count.h
3164 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3165 (sjlj_emit_dispatch_table): Likewise.
3166 * explow.c: Include profile-count.h
3167 * expmed.c (emit_store_flag_force): Update.
3168 (do_cmp_and_jump): Update.
3169 * expr.c (compare_by_pieces_d::generate): Update.
3170 (compare_by_pieces_d::finish_mode): Update.
3171 (emit_block_move_via_loop): Update.
3172 (store_expr_with_bounds): Update.
3173 (store_constructor): Update.
3174 (expand_expr_real_2): Update.
3175 (expand_expr_real_1): Update.
3176 * expr.h (try_casesi, try_tablejump): Update prototypes.
3177 * gimple-pretty-print.c (dump_probability): Update.
3178 (dump_profile): New.
3179 (dump_gimple_label): Update.
3180 (dump_gimple_bb_header): Update.
3181 * graph.c (draw_cfg_node_succ_edges): Update.
3182 * hsa-gen.c (convert_switch_statements): Update.
3183 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3184 (find_if_case_1): Update.
3185 (find_if_case_2): Update.
3186 * internal-fn.c (expand_arith_overflow_result_store): Update.
3187 (expand_addsub_overflow): Update.
3188 (expand_neg_overflow): Update.
3189 (expand_mul_overflow): Update.
3190 (expand_vector_ubsan_overflow): Update.
3191 * ipa-cp.c (good_cloning_opportunity_p): Update.
3192 * ipa-split.c (split_function): Use make_single_succ_edge.
3193 * ipa-utils.c (ipa_merge_profiles): Update.
3194 * loop-doloop.c (add_test): Update.
3195 (doloop_modify): Update.
3196 * loop-unroll.c (compare_and_jump_seq): Update.
3197 (unroll_loop_runtime_iterations): Update.
3198 * lra-constraints.c (lra_inheritance): Update.
3199 * lto-streamer-in.c (input_cfg): Update.
3200 * lto-streamer-out.c (output_cfg): Update.
3201 * mcf.c (adjust_cfg_counts): Update.
3202 * modulo-sched.c (sms_schedule): Update.
3203 * omp-expand.c (expand_omp_for_init_counts): Update.
3204 (extract_omp_for_update_vars): Update.
3205 (expand_omp_ordered_sink): Update.
3206 (expand_omp_for_ordered_loops): Update.
3207 (expand_omp_for_generic): Update.
3208 (expand_omp_for_static_nochunk): Update.
3209 (expand_omp_for_static_chunk): Update.
3210 (expand_cilk_for): Update.
3211 (expand_omp_simd): Update.
3212 (expand_omp_taskloop_for_outer): Update.
3213 (expand_omp_taskloop_for_inner): Update.
3214 * omp-simd-clone.c (simd_clone_adjust): Update.
3215 * optabs.c (expand_doubleword_shift): Update.
3216 (expand_abs): Update.
3217 (emit_cmp_and_jump_insn_1): Update.
3218 (expand_compare_and_swap_loop): Update.
3219 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3220 * predict.c (predictable_edge_p): Update.
3221 (edge_probability_reliable_p): Update.
3222 (set_even_probabilities): Update.
3223 (combine_predictions_for_insn): Update.
3224 (combine_predictions_for_bb): Update.
3225 (propagate_freq): Update.
3226 (estimate_bb_frequencies): Update.
3227 (force_edge_cold): Update.
3228 * profile-count.c (profile_count::dump): Add missing space into dump.
3229 (profile_count::debug): Add newline.
3230 (profile_count::differs_from_p): Explicitly convert to unsigned.
3231 (profile_count::stream_in): Update.
3232 (profile_probability::dump): New member function.
3233 (profile_probability::debug): New member function.
3234 (profile_probability::differs_from_p): New member function.
3235 (profile_probability::differs_lot_from_p): New member function.
3236 (profile_probability::stream_in): New member function.
3237 (profile_probability::stream_out): New member function.
3238 * profile-count.h (profile_count_quality): Rename to ...
3239 (profile_quality): ... this one.
3240 (profile_probability): New.
3241 (profile_count): Update.
3242 * profile.c (compute_branch_probabilities): Update.
3243 * recog.c (peep2_attempt): Update.
3244 * sched-ebb.c (schedule_ebbs): Update.
3245 * sched-rgn.c (find_single_block_region): Update.
3246 (compute_dom_prob_ps): Update.
3247 (schedule_region): Update.
3248 * sel-sched-ir.c (compute_succs_info): Update.
3249 * stmt.c (struct case_node): Update.
3250 (do_jump_if_equal): Update.
3251 (get_outgoing_edge_probs): Update.
3252 (conditional_probability): Update.
3253 (emit_case_dispatch_table): Update.
3254 (expand_case): Update.
3255 (expand_sjlj_dispatch_table): Update.
3256 (emit_case_nodes): Update.
3257 * targhooks.c: Update.
3258 * tracer.c (better_p): Update.
3259 (find_best_successor): Update.
3260 * trans-mem.c (expand_transaction): Update.
3261 * tree-call-cdce.c: Update.
3262 * tree-cfg.c (gimple_split_edge): Upate.
3263 (move_sese_region_to_fn): Upate.
3264 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3265 * tree-eh.c (lower_resx): Upate.
3266 (cleanup_empty_eh_move_lp): Upate.
3267 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3268 * tree-inline.c (copy_edges_for_bb): Update.
3269 (copy_cfg_body): Update.
3270 * tree-parloops.c (gen_parallel_loop): Update.
3271 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3272 (gimple_gen_time_profiler): Update.
3273 * tree-ssa-dce.c (remove_dead_stmt): Update.
3274 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3275 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3276 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3277 (unloop_loops): Update.
3278 (try_peel_loop): Update.
3279 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3280 * tree-ssa-loop-split.c (connect_loops): Update.
3281 (split_loop): Update.
3282 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3283 (hoist_guard): Update.
3284 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3285 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3286 (value_replacement): Update.
3287 * tree-ssa-reassoc.c (branch_fixup): Update.
3288 * tree-ssa-tail-merge.c (replace_block_by): Update.
3289 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3290 (create_edge_and_update_destination_phis): Update.
3291 (compute_path_counts): Update.
3292 (recompute_probabilities): Update.
3293 (update_joiner_offpath_counts): Update.
3294 (freqs_to_counts_path): Update.
3295 (duplicate_thread_path): Update.
3296 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3297 (struct switch_conv_info): Update.
3298 (gen_inbound_check): Update.
3299 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3300 (vect_do_peeling): Update.
3301 (vect_loop_versioning): Update.
3302 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3303 (optimize_mask_stores): Update.
3304 * ubsan.c (ubsan_expand_null_ifn): Update.
3305 * value-prof.c (gimple_divmod_fixed_value): Update.
3306 (gimple_divmod_fixed_value_transform): Update.
3307 (gimple_mod_pow2): Update.
3308 (gimple_mod_pow2_value_transform): Update.
3309 (gimple_mod_subtract): Update.
3310 (gimple_mod_subtract_transform): Update.
3311 (gimple_ic): Update.
3312 (gimple_stringop_fixed_value): Update.
3313 (gimple_stringops_transform): Update.
3314 * value-prof.h: Update.
3316 2017-06-29 Carl Love <cel@us.ibm.com>
3318 * config/rs6000/rs6000-c.c: Add support for built-in functions
3319 vector signed int vec_signed (vector float);
3320 vector signed long long vec_signed (vector double);
3321 vector signed int vec_signed2 (vector double, vector double);
3322 vector signed int vec_signede (vector double);
3323 vector signed int vec_signedo (vector double);
3324 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3325 instruction generator.
3326 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3327 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3328 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3330 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3331 vunsignede_v2df): Add define_expands.
3332 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3333 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3334 VEC_UNSIGNEDO): Add definitions.
3335 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3336 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3337 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3338 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3339 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3340 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3341 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3342 vec_unsignede, vec_unsignedo): Add builtin defines.
3343 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3345 * doc/extend.texi: Update the built-in documentation file for the
3346 new built-in functions.
3348 2017-06-29 Richard Biener <rguenther@suse.de>
3350 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3351 reduction chains to LOOP_VINFO_REDUCTIONS.
3352 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3353 SLP reductions after processing reduction chains.
3355 2017-06-29 Nathan Sidwell <nathan@acm.org>
3357 * builtins.c (fold_builtin_FUNCTION): Use
3358 lang_hooks.decl_printable_name.
3360 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3363 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3364 with only one label.
3365 * stmt.c (expand_case): Assert NCASES is greater than one.
3367 2017-06-29 Richard Biener <rguenther@suse.de>
3369 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3371 (group_case_labels): Likewise.
3372 (find_taken_edge): Push sanity checking on val to workers...
3373 (find_taken_edge_cond_expr): ... here
3374 (find_taken_edge_switch_expr): ... and here, handle cases
3375 with just a default label.
3376 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3377 (group_case_labels): Likewise.
3378 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3379 group_case_labels does anything cleanup the CFG again.
3381 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3383 PR tree-optimization/81196
3384 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3385 exit condition comparing two IVs.
3387 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3389 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3390 profile to the dummy entry at the end of the list of architectures.
3391 * config/arm/arm-cpu-cdata.h: Regenerated.
3393 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3394 Michael Collison <michael.collison@arm.com>
3397 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3399 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3400 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3401 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3402 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3403 of shift when the shift amount is masked with constant equal to
3404 the size of the mode.
3405 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3408 2017-06-29 Martin Liska <mliska@suse.cz>
3410 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3412 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3415 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3418 (edge_growth_cache_entry::edge_growth_cache_entry): New
3420 (reset_edge_growth_cache): Update to use constructor.
3422 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3424 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3425 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3426 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3428 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3430 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3431 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3433 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3435 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3436 (*-linux-uclibc*): Add t-uclibc tmake_file.
3437 * config/t-musl: New.
3438 * config/t-uclibc: New.
3440 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3442 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3444 (gen_comm_data): Emit architectural setting of arch_prof.
3445 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3447 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3448 (armv8-m.base, armv8-m.main): Likewise.
3449 * arm-protos.h (arm_build_target): Add profile field.
3450 (arch_option): Likewise.
3451 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3453 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3454 arm_active_target.profile.
3456 2017-06-28 Richard Biener <rguenther@suse.de>
3459 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3460 TYPE_OVERFLOW_WRAPS.
3461 * match.pd (negate_expr_p): Likewise.
3462 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3463 fold_build2, not fold_binary.
3465 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3467 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3468 Convert memory address to Pmode.
3469 (aarch64_print_operand): Assert MEM operands are always Pmode.
3471 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3474 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3475 Remove redundant if.
3476 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3477 * config/arm/aarch-common-protos.h
3478 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3479 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3481 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3484 * multiple_target.c (create_dispatcher_calls): Set the default
3485 clone to be static, not public.
3487 2017-06-28 Richard Biener <rguenther@suse.de>
3489 * tree-vect-loop.c (vectorizable_reduction): Move special
3490 cond reduction IV var creation ...
3491 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3492 parameter. Use STMT_VINFO_VECTYPE.
3493 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3496 2017-06-28 Martin Liska <mliska@suse.cz>
3499 * ipa-visibility.c (non_local_p): Handle visibility.
3501 2017-06-28 Martin Liska <mliska@suse.cz>
3504 * common.opt: Add IntegerRange to various options.
3505 * opt-functions.awk (integer_range_info): New function.
3506 * optc-gen.awk: Add integer_range_info to cl_options struct.
3507 * opts-common.c (decode_cmdline_option): Handle
3508 CL_ERR_INT_RANGE_ARG.
3509 (cmdline_handle_error): Likewise.
3510 * opts.c (print_filtered_help): Show valid interval in
3511 when --help is provided.
3512 * opts.h (struct cl_option): Add range_min and range_max fields.
3513 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3515 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3517 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3518 (x * C EQ/NE y * C): New transformation.
3520 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3522 * genmultilib (combination_space): Accept '+' in option names.
3524 2017-06-28 Martin Liska <mliska@suse.cz>
3527 * asan.c (instrument_derefs): Bail out inner references
3528 that are hard register variables.
3530 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3533 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3534 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3536 2017-06-28 Richard Biener <rguenther@suse.de>
3538 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3539 (vect_get_slp_defs): Adjust.
3540 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3542 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3544 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3545 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3546 (vectorizable_reduction): Adjust.
3547 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3549 (vect_get_slp_defs): Likewise.
3550 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3551 (vectorizable_bswap): Adjust.
3552 (vectorizable_call): Likewise.
3553 (vectorizable_conversion): Likewise.
3554 (vectorizable_assignment): Likewise.
3555 (vectorizable_shift): Likewise.
3556 (vectorizable_operation): Likewise.
3557 (vectorizable_store): Likewise.
3558 (vectorizable_condition): Likewise.
3559 (vectorizable_comparison): Likewise.
3561 2017-06-28 Michael Collison <michael.collison@arm.com>
3564 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3566 (arm_gen_movmemqi): Removed unused variable 'i'.
3567 Convert 'for' loop into 'while' loop.
3568 (arm_expand_prologue): Remove last unnecessary set of insn.
3569 (thumb_pop): Remove unused variable 'pushed_words'.
3570 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3572 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3574 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3575 * config/s390/s390-protos.h: Add prototype of
3576 s390_rel_address_ok_p.
3577 * config/s390/s390.c (s390_got_symbol): New function.
3578 (s390_rel_address_ok_p): New function.
3579 (legitimize_pic_address): Use s390_rel_address_ok_p.
3580 (s390_load_got): Use s390_got_symbol.
3581 (s390_option_override): Issue error if
3582 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3583 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3585 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3587 2017-06-27 Andrew Pinski <apinski@cavium.com>
3589 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3590 (X * copysign (1.0, X)): New pattern.
3591 (X * copysign (1.0, -X)): New pattern.
3592 (copysign (-1.0, CST)): New pattern.
3594 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3596 * genmultilib (combination_space): Remove variable.
3597 Validate reuse rules against regular expression for any sequence
3598 of multilib options in any order.
3600 2017-06-27 Michael Collison <michael.collison@arm.com>
3602 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3603 call aarch64_split_simd_combine.
3604 * (aarch64_combine_internal<mode>): Delete pattern.
3605 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3606 Allow register and subreg operands.
3608 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3610 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3611 specific need, just fallback on defaults.
3612 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3614 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3615 Olivier Hainque <hainque@adacore.com>
3617 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3619 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3620 targets. Pick a default if no particular attempt applied.
3621 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3624 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3626 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3627 (x86_64-wrs-vxworks7): Likewise.
3629 2017-06-27 Marek Polacek <polacek@redhat.com>
3632 * ubsan.c (instrument_null): Check get_base_address's result for null.
3634 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3636 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3638 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3640 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3641 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3642 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3643 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3645 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3646 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3647 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3648 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3649 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3650 BUILT_IN_FEUPDATEENV): New builtins.
3651 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3652 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3653 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3654 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3656 (builtin_structptr_types): Adjust size.
3657 * tree.c (builtin_structptr_types): Add four entries.
3659 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3660 Olivier Hainque <hainque@adacore.com>
3662 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3663 (TLS_SYM): New local macro, forcing reference to __tls__ on
3664 link command lines for VxWorks 7 RTPs, triggering initialization
3666 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3667 OS features TLS support, true for RTPs on VxWorks 7.
3668 * config/vxworks.c (vxworks_override_options): Setup emutls
3671 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3673 * predict.c (test_prediction_value_range): Use -1U instead of -1
3674 to avoid narrowing conversion warning.
3675 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3676 to avoid narrowing conversion warning.
3677 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3679 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3681 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3683 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3684 64bit configurations.
3685 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3686 (SIZE_TYPE): Likewise.
3687 * config/vxworks.c (vxworks_emutls_var_fields): Use
3688 long_unsigned_type_node instead of unsigned_type_node as the offset
3689 field type, which is "pointer" mode in emutls.c.
3691 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3694 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3697 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3698 gimple_vuse copying separately from gimple_vdef copying.
3700 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3702 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3705 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3706 Olivier Hainque <hainque@adacore.com>
3708 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3709 for all vxworks7 targets.
3710 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3711 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3712 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3713 variations for VX6/VX7 and 32/64bits later on in ...
3714 (VXWORKS_LIB_SPEC): Leverage new macros.
3715 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3716 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3718 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3720 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3721 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3723 2017-06-26 Carl Love <cel@us.ibm.com>
3725 * config/rs6000/rs6000-c.c: Add support for built-in functions
3726 vector bool char vec_reve (vector bool char);
3727 vector signed char vec_reve (vector signed char);
3728 vector unsigned char vec_reve (vector unsigned char);
3729 vector bool int vec_reve (vector bool int);
3730 vector signed int vec_reve (vector signed int);
3731 vector unsigned int vec_reve (vector unsigned int);
3732 vector bool long long vec_reve (vector bool long long);
3733 vector signed long long vec_reve (vector signed long long);
3734 vector unsigned long long vec_reve (vector unsigned long long);
3735 vector bool short vec_reve (vector bool short);
3736 vector signed short vec_reve (vector signed short);
3737 vector double vec_reve (vector double);
3738 vector float vec_reve (vector float);
3739 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3740 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3741 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3742 (altivec_vreve): New pattern.
3743 * config/rs6000/altivec.h (vec_reve): New define.
3744 * doc/extend.texi (vec_rev): Update the built-in documentation file
3745 for the new built-in functions.
3747 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3749 PR tree-optimization/71815
3750 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3752 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3754 (slsr_process_phi): Likewise.
3755 (replace_uncond_cands_and_profitable_phis): Don't replace a
3756 multiply candidate with a stride of 1 (copy or cast).
3757 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3759 (lowest_cost_path): Likewise.
3760 (total_savings): Likewise.
3762 2017-06-26 Richard Biener <rguenther@suse.de>
3765 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
3766 Use def_builtin_pure for all gather builtins.
3768 2017-06-26 Richard Biener <rguenther@suse.de>
3770 PR tree-optimization/81203
3771 * tree-tailcall.c (find_tail_calls): Do not move stmts into
3774 2017-06-26 Marek Polacek <polacek@redhat.com>
3777 * doc/invoke.texi: Document -Wmultistatement-macros.
3779 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
3781 * doc/sourcebuild.texi (ARM-specific attributes): Document new
3782 arm_neon_ok_no_float_abi effective target.
3784 2017-06-26 Richard Biener <rguenther@suse.de>
3786 PR tree-optimization/80928
3787 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
3788 (copy_bbs): Set BB_DUPLICATED flag early.
3789 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
3791 (execute_on_shrinking_pred): Likewise.
3792 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
3793 BB_DUPLICATED blocks.
3794 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
3795 iterate over all PHIs considering removal of *gsi.
3797 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
3799 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
3802 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3804 * config/rs6000/rs6000-string.c: (expand_block_clear,
3805 do_load_for_compare, select_block_compare_mode,
3806 compute_current_alignment, expand_block_compare,
3807 expand_strncmp_align_check, expand_strn_compare,
3808 expand_block_move, rs6000_output_load_multiple)
3809 Move functions related to string/block move/compare
3811 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
3812 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
3813 for this function which is now used in two files.
3814 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
3815 * config.gcc: Add rs6000-string.o to extra_objs for
3816 targets powerpc*-*-* and rs6000*-*-*.
3818 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3821 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
3822 32-bit, since indexed is not valid for DImode.
3823 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
3824 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
3825 (define_peephole2 for Altivec d-form load): Add 32-bit support.
3826 (define_peephole2 for Altivec d-form store): Likewise.
3829 * multiple_target.c (create_dispatcher_calls): Only create the
3830 dispatcher call if the function is the default clone of a
3833 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
3836 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
3837 a call, force the call to not be a tail call.
3839 2017-06-23 Jeff Law <law@redhat.com>
3841 * doc/contrib.texi: Add entry for Steven Pemberton's work on
3844 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
3846 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
3847 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
3848 handling for early expansion of vector shifts (sl,sr,sra,rl).
3849 (builtin_function_type): Add vector shift right instructions
3850 to the unsigned argument list.
3852 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3854 rtl-optimizatoin/79286
3855 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
3856 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
3857 trap. PIC register plus a const unspec without offset can never trap.
3859 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
3861 * tree.h (builtin_structptr_type): New type.
3862 (builtin_structptr_types): Declare new array.
3863 * tree.c (builtin_structptr_types): New array.
3864 (free_lang_data, build_common_tree_nodes): Use it.
3866 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
3869 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
3872 2017-06-22 Matt Turner <mattst88@gmail.com>
3874 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
3875 Lake models to skylake case. Assume skylake for unknown
3876 models with clflushopt.
3878 2017-06-22 Jeff Law <law@redhat.com>
3880 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
3881 frame sizes that do not satisfy aarch64_uimm12_shift.
3883 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
3885 * profile-count.h (apply_probability,
3886 apply_scale, probability_in): Fix checks for zero.
3888 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3890 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
3891 * doc/cppdiropts.texi (-I @var{dir}): Document it.
3893 2016-06-22 Richard Biener <rguenther@suse.de>
3895 * tree-vect-loop.c (vect_model_reduction_cost): Handle
3896 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
3897 REDUC_MAX_EXPR support.
3898 (vectorizable_reduction): Likewise.
3899 (vect_create_epilog_for_reduction): Likewise.
3901 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
3903 * match.pd (A / (1 << B) -> A >> B): New.
3904 * generic-match-head.c: Include optabs-tree.h.
3905 * gimple-match-head.c: Likewise.
3906 * optabs-tree.h (target_supports_op_p): New.
3907 * optabs-tree.c (target_supports_op_p): New.
3909 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3911 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
3912 $gcc_cv_ld --help output.
3913 (gcc_cv_ld_demangle): Likewise.
3914 (gcc_cv_ld_eh_frame_hdr): Likewise.
3915 (gcc_cv_ld_pie): Likewise.
3916 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
3917 (gcc_cv_ld_buildid): Likewise.
3918 (gcc_cv_ld_sysroot): Likewise.
3919 (ld_bndplt_support): Likewise.
3920 (ld_pushpopstate_support): Likewise.
3921 * configure: Regenerate.
3922 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
3924 2017-06-21 Jakub Jelinek <jakub@redhat.com>
3927 * config/i386/sse.md (round<mode>2): Renumber match_dup and
3928 operands indexes to avoid gap between operands and match_dups.
3930 2017-06-21 Andrew Pinski <apinski@cavium.com>
3932 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
3933 Increment Arith_shift and Arith_shift_reg by 1.
3934 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
3936 * config/aarch64/aarch64.c (thunderx_tunings): Enable
3937 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
3938 (aarch64_strip_extend): Add new argument and test for it.
3939 (aarch64_cheap_mult_shift_p): New function.
3940 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
3941 add a cost if it is true.
3942 Update calls to aarch64_strip_extend.
3943 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
3945 2017-06-21 Andrew Pinski <apinski@cavium.com>
3947 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
3949 (thunderxt88): Likewise.
3950 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
3951 (thunderx_prefetch_tune): New variable.
3952 (thunderx2t99_prefetch_tune): Update for the correct values.
3953 (thunderxt88_tunings): New variable.
3954 (thunderx_tunings): Use thunderx_prefetch_tune instead of
3955 generic_prefetch_tune.
3956 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
3958 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3960 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
3961 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
3962 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
3963 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
3964 (aarch64_atomic_cas<mode>, GPI): Likewise.
3966 2017-06-21 Martin Liska <mliska@suse.cz>
3968 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
3969 statements on cold and hot labels.
3970 * predict.c (tree_estimate_probability_bb): Remove the
3971 prediction from this place.
3973 2017-06-21 Martin Liska <mliska@suse.cz>
3975 PR tree-optimization/79489
3976 * gimplify.c (maybe_add_early_return_predict_stmt): New
3978 (gimplify_return_expr): Call the function.
3979 * predict.c (tree_estimate_probability_bb): Remove handling
3981 * predict.def: Update comment about early return predictor.
3982 * gimple-predict.h (is_gimple_predict): New function.
3983 * predict.def: Change default value of early return to 66.
3984 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
3986 * passes.def: Put pass_strip_predict_hints to the beginning of
3989 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
3991 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
3992 FUNCTION_DECL declarations.
3993 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
3995 (dwaf2out_decl): Likewise.
3996 * godump.c (go_early_global_decl): Skip call to the real debug hook
3997 for FUNCTION_DECL declarations.
3998 * passes.c (rest_of_decl_compilation): Skip call to the
3999 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4000 -fdump-go-spec is passed.
4002 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4004 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4005 Reorder for compactness.
4006 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4007 (def_builtin_pure, def_builtin_pure2): New functions.
4008 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4010 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4012 * match.pd (nop_convert): New predicate.
4013 ((A +- CST1) +- CST2): Allow some NOP conversions.
4015 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4018 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4019 with ctors/dtors if GOVD_SHARED is set.
4021 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4023 * config/aarch64/aarch64.md (movti_aarch64):
4024 Emit mov rather than orr.
4025 (movtf_aarch64): Likewise.
4026 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4027 Emit mov rather than orr.
4029 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4031 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4032 Swap alternatives, make integer dup more expensive.
4034 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4036 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4037 Return true for non-tls symbols.
4039 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4041 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4042 (cortex-a75): Likewise.
4043 (cortex-a75.cortex-a55): Likewise.
4044 * config/aarch64/aarch64-tune.md: Regenerate.
4045 * doc/invoke.texi (-mtune): Document new values for -mtune.
4047 2017-06-21 Tom de Vries <tom@codesourcery.com>
4049 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4051 (Effective-Target Keywords, Other attributes): Suggest using
4052 dg-add-options stack_size feature to get stack limit in stack_size
4053 effective target documentation.
4055 2017-06-21 Julian Brown <julian@codesourcery.com>
4056 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4058 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4059 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4060 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4062 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4063 attribute type list for neon_multiply.
4064 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4065 attribute type list for neon_multiply.
4066 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4067 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4068 attribute type list for neon_multiply.
4069 * config/arm/types.md (crypto_pmull): Add.
4070 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4071 attribute type list.
4073 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4075 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4078 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4080 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4081 to make sure not to dereference a NULL cost_classes_ptr pointer.
4083 2017-06-20 Carl Love <cel@us.ibm.com>
4085 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4086 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4087 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4088 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4089 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4090 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4091 VMULOSW): New enum "unspec" values.
4092 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4093 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4094 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4095 altivec_vmulosw): New patterns.
4096 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4097 VMULOSW): Add definitions.
4099 2017-06-20 Julia Koval <julia.koval@intel.com>
4101 * config/i386/i386.c: Fix rounding expand for new pattern.
4102 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4104 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4106 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4107 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4109 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4111 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4114 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4116 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4117 architecture, then by implementer ID.
4118 * config/aarch64/aarch64-tune.md: Regenerate.
4120 2017-06-20 Richard Biener <rguenther@suse.de>
4123 * fold-const.c (split_tree): Fold to type before negating.
4125 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4127 * diagnostic-show-locus.c
4128 (selftest::test_fixit_deletion_affecting_newline): New function.
4129 (selftest::diagnostic_show_locus_c_tests): Call it.
4131 2017-06-20 Andreas Schwab <schwab@suse.de>
4134 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4137 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4139 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4140 __ARM_FEATURE_COPROC according to support.
4142 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4144 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4145 Rewritten to avoid overflow for > 32-bit pointers.
4148 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4149 by removing enum keyword.
4150 (ubsan_type_descriptor): Likewise. Formatting fix.
4153 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4154 splitter): Require TARGET_SSE2 in the condition.
4156 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4159 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4160 for doing vector set of SFmode on ISA 3.0.
4161 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4162 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4164 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4165 SFmode value into a V4SF variable that was extracted from another
4166 V4SF variable without converting the element to double precision
4167 and back to single precision vector format.
4168 (vsx_insert_extract_v4sf_p9_2): Likewise.
4170 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4172 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4173 in UWHI to avoid undefined overflow.
4176 * ubsan.h (enum ubsan_encode_value_phase): New.
4177 (ubsan_encode_value): Change second argument to
4178 enum ubsan_encode_value_phase with default value of
4179 UBSAN_ENCODE_VALUE_GENERIC.
4180 * ubsan.c (ubsan_encode_value): Change second argument to
4181 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4182 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4183 create_tmp_var_raw instead of create_tmp_var and use a
4185 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4186 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4187 ubsan_encode_value callers.
4190 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4191 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4192 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4194 2017-06-19 Richard Biener <rguenther@suse.de>
4197 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4198 estimates if we changed anything.
4200 2017-06-19 Richard Biener <rguenther@suse.de>
4202 PR tree-optimization/80887
4203 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4204 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4205 simplified lookups, then reset mprts_hook.
4206 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4208 (try_to_simplify): Likewise.
4210 2017-06-19 Martin Liska <mliska@suse.cz>
4213 * gimplify.c (gimplify_switch_expr):
4214 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4216 2017-06-19 Martin Liska <mliska@suse.cz>
4218 * doc/install.texi: Document that PGO runs in 4 stages.
4220 2017-06-19 Martin Liska <mliska@suse.cz>
4223 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4224 to dispatcher function name.
4225 * multiple_target.c (replace_function_decl): New function.
4226 (create_dispatcher_calls): Redirect both edges and references.
4228 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4230 * profile-count.c (profile_count::dump): Dump quality.
4231 (profile_count::differs_from_p): Update for unsigned val.
4232 * profile-count.h (profile_count_quality): New enum.
4233 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4235 2017-06-19 Richard Biener <rguenther@suse.de>
4237 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4238 struct function as arg.
4239 (estimate_numbers_of_iterations): Export overload with loop arg.
4240 (free_numbers_of_iterations_estimates_loop): Use an overload of
4241 free_numbers_of_iterations_estimates instead.
4242 * tree-cfg.c (remove_bb): Adjust.
4243 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4244 * tree-parloops.c (gen_parallel_loop): Likewise.
4245 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4247 (tree_unroll_loops_completely): Likewise.
4248 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4249 Use an overload instead and export.
4250 (estimated_loop_iterations): Adjust.
4251 (max_loop_iterations): Likewise.
4252 (likely_max_loop_iterations): Likewise.
4253 (estimate_numbers_of_iterations): Take struct function as arg
4255 (loop_exits_before_overflow): Adjust.
4256 (free_numbers_of_iterations_estimates_loop): Use an overload.
4257 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4258 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4260 2017-06-19 Richard Biener <rguenther@suse.de>
4263 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4264 RANGE_EXPR conservatively.
4266 2017-06-16 Carl Love <cel@us.ibm.com>
4268 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4269 definitions for vec_float, vec_float2, vec_floato,
4270 vec_floate built-ins.
4271 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4272 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4274 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4275 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4276 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4277 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4278 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4279 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4280 vec_floato): Add builtin defines.
4281 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4282 Update the built-in documentation file for the new built-in
4285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4287 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4288 (mthumb): Mark as the negative of -marm.
4290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4292 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4294 (ARM Options, -mtune): Document that this accepts the same
4295 extension options as -mcpu.
4296 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4298 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4300 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4301 permitted extensions.
4303 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4305 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4306 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4307 (armv8-m.main): Add option +nodsp.
4308 * config/arm/arm-cpu-cdata.h: Regenerated.
4310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4312 * config/arm/t-fuchsia: New file.
4313 * config.gcc (arm*-*-fuchsia*): Use it.
4315 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4317 * config/arm/t-symbian: Rewrite for new option infrastructure.
4319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4321 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4322 (MULTILIB_REQUIRED): Likewise.
4324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4326 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4327 (MULTILIB_RESUE): Likewise.
4328 (MULTILIB_MATCHES): Likewise.
4329 (MULTLIB_REQUIRED): Likewise.
4331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4333 * config/arm/t-rtems: Rewrite for new option framework.
4335 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4337 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4338 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4339 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4340 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4341 * config/arm/t-multilib: ... here.
4342 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4343 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4344 armv7-a and armv8*-a when A-profile libraries have not been built.
4345 * config/arm/t-rmprofile: Rewrite.
4347 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4349 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4350 with a backslash. Remove the backslash after substituting unescaped
4352 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4356 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4357 the driver to pass through the default setting of -mfloat-abi.
4358 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4360 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4361 (all_feat_combs): New rule.
4362 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4364 * config/arm/t-aprofile: Rewrite.
4366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4368 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4369 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4370 fpu is not specified by the user/command-line.
4371 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4372 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4373 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4374 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4375 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4376 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4378 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4380 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4381 * config/arm/t-arm-elf: Rewritten.
4383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4385 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4386 have some floating-point instructions.
4387 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4388 (TARGET_MAYBE_HARD_FLOAT): New macro.
4389 * config/arm/arm-builtins.c (arm_init_builtins): Use
4390 TARGET_MAYBE_HARD_FLOAT.
4391 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4395 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4396 (configargs.h): Include it.
4397 (arm_print_hint_for_fpu_option): New function.
4398 (arm_parse_fpu_option): New function.
4399 (candidate_extension): New class.
4400 (arm_canon_for_multilib): New function.
4401 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4402 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4403 (ARCH_CANONICAL_SPECS): New macro.
4404 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4408 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4409 are set after handling multilib fragments. Set target_cpu_default2
4412 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4414 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4416 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4417 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4419 (check_arch): Likewise.
4420 * config/arm/arm.c (arm_configure_build_target): Handle
4421 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4422 options in the default.
4424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4426 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4427 when an option is an alias of another.
4428 * config/arm/parsecpu.awk (optalias): New parser token.
4429 (gen_comm_data): Mark non-alias options as such. Emit entries
4430 for extension aliases.
4431 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4432 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4433 (armv6kz, armv6zk, armv6t2): Likewise.
4434 (armv7): Make vfpv3-d16 an alias.
4435 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4437 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4438 Sort in canonical order.
4439 (armv8-a): Sort in canonical order.
4440 (armv8.1-a, armv8.2-a): Likewise.
4441 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4443 (cortex-a9): Sort in canonical order.
4444 * config/arm/arm.c (selftests.h): Include it.
4445 (arm_test_cpu_arch_data): New function.
4446 (arm_run_self_tests): New function.
4447 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4448 (targetm): Move declaration to the end of the file.
4449 * arm-cpu-cdata.h: Regenerated.
4451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4453 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4454 call to target_mode_check describing the type of option passed.
4455 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4456 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4457 arm_parse_cpu_option_name to match parameters against list of
4459 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4460 arm_arch_core_flags data structure.
4461 * config/arm/arm-cpu_cdata.h: Regenerated.
4463 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4465 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4467 (arm_print_hint_for_cpu_option): Likewise.
4468 (arm_print_hint_for_arch_option): Likewise.
4469 (arm_parse_cpu_option_name): Likewise.
4470 (arm_parse_arch_option_name): Likewise.
4471 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4472 of entries in the all_fpus list.
4473 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4474 (arm_parse_cpu_option_name): Declare.
4475 (arm_parse_arch_option_name): Declare.
4476 (arm_parse_option_features): Declare.
4477 (arm_intialize_isa): Declare.
4478 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4480 (gen_comm_data): ... here. Make definitions non-static.
4481 * config/arm/arm-cpu-data.h: Regenerated.
4482 * config/arm/arm-cpu-cdata.h: Regenerated.
4484 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4486 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4487 (cpu_arch_extension): New structure.
4488 (cpu_arch_option, arch_option, cpu_option): New structures.
4489 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4491 (gen_data): Generate new format data tables.
4492 * config/arm/arm.c (cpu_tune): New structure.
4493 (cpu_option, processors): Delete.
4494 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4495 (arm_print_hint_for_cpu_option): ... this and ...
4496 (arm_print_hint_for_arch_option): ... this.
4497 (arm_parse_arch_cpu_name): Delete. Replace with ...
4498 (arm_parse_cpu_option_name): ... this and ...
4499 (arm_parse_arch_option_name): ... this.
4500 (arm_unrecognized_feature): Change type of target parameter to
4502 (arm_parse_arch_cpu_features): Delete. Replace with ...
4503 (arm_parse_option_features): ... this.
4504 (arm_configure_build_target): Rework to use new configuration data
4506 (arm_print_tune_info): Rework for new configuration data tables.
4507 * config/arm/arm-cpu-data.h: Regenerated.
4508 * config/arm/arm-cpu.h: Regenerated.
4510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4512 * Makefile.in (OBJS): Move sbitmap.o from here ...
4513 (OBJS-libcommon): ... to here.
4515 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4517 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4518 (ISA_ALL_CRYPTO): New macro.
4519 (ISA_ALL_SIMD): New macro
4520 (ISA_ALL_FP): New macro.
4521 * config/arm/arm.c (fpu_bitlist): Update initializer.
4522 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4524 (arm9e): Add fpu. Add option for nofp
4525 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4526 (arm926ej-s, arm1026ej-s): Likewise.
4527 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4528 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4529 neon-fp16, neon-vfpv4, nofp and nosimd.
4530 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4531 (cortex-a8): Add fpu. Add option for nofp.
4532 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4533 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4534 (cortex-r4f): Add fpu.
4535 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4536 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4538 (cortex-r8): Likewise.
4539 (cortex-m4): Add fpu. Add option for nofp.
4540 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4541 (cortex-a17.cortex-a7): Likewise.
4542 (cortex-a32): Add fpu. Add options for crypto and nofp.
4543 (cortex-a35, cortex-a53): Likewise.
4544 (cortex-a57): Add fpu. Add option for crypto.
4545 (cortex-a72, cortex-a73): Likewise.
4546 (exynos-m1): Likewise.
4547 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4548 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4549 (cortex-m33): Add fpu. Add option for nofp.
4550 * config/arm/arm-cpu-cdata.h: Regenerated
4551 * config/arm/arm-cpu-data.h: Regenerated.
4553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4555 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4556 (armv5te, armv5tej): Likewise.
4557 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4558 (armv7): Add options fp and vfpv3-d16.
4559 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4560 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4562 (armv7ve): Likewise.
4563 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4564 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4565 (armv8-a): Add nocrypto option.
4566 (armv8.1-a, armv8.2-a): Likewise.
4567 (armv8-m.main): add options fp, fp.dp and nofp.
4569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4571 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4573 (armv8-a+crc): Delete.
4574 (armv8.1-a): Add options simd, crypto and nofp.
4575 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4576 (armv8.2-a+fp16): Delete.
4577 (armv8-m.main): Add option dsp.
4578 (armv8-m.main+dsp): Delete.
4579 (cortex-a8): Add fpu. Add nofp option.
4580 (cortex-a9): Add fpu. Add nofp and nosimd options.
4581 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4582 link to main cpu and architecture data structures.
4583 (gen_comm_data): Only put isa attributes from the main architecture
4585 (option): New statement for architecture and CPU entries.
4586 * arm.c (struct cpu_option): New structure.
4587 (struct processors): Add entry for options.
4588 (arm_unrecognized_feature): New function.
4589 (arm_parse_arch_cpu_name): Ignore any characters after the first
4591 (arm_parse_arch_cpu_feature): New function.
4592 (arm_configure_build_target): Separate out any CPU and architecture
4593 features and parse separately. Don't error out if -mfpu=auto is
4594 used with only an architecture string.
4595 (arm_print_asm_arch_directives): New function.
4596 (arm_file_start): Call it.
4597 * config/arm/arm-cpu-cdata.h: Regenerated.
4598 * config/arm/arm-cpu-data.h: Likewise.
4599 * config/arm/arm-tables.opt: Likewise.
4601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4603 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4604 assembler when it is not -mfpu=auto.
4606 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4608 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4609 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4610 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4611 (ASM_CPU_SPEC): Rewrite.
4612 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4613 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4614 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4615 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4616 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4617 copied string is NUL-terminated. Also strip any characters prefixed
4619 (arm_rewrite_selected_arch): New function.
4620 (arm_rewrite_march): New function.
4622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4624 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4625 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4626 (march, mcpu, mtune): Convert to string-based options.
4627 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4628 (arm_parse_arch_cpu_name): New function.
4629 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4630 identify selected architecture or CPU.
4631 (arm_option_save): New function.
4632 (TARGET_OPTION_SAVE): Redefine.
4633 (arm_option_restore): Restore string options.
4634 (arm_option_print): Print string options.
4636 2017-06-16 Martin Sebor <msebor@redhat.com>
4638 PR tree-optimization/80933
4639 PR tree-optimization/80934
4640 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4641 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4642 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4643 (gimple_fold_builtin): Call them.
4645 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4647 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4648 as unlikely; update profile.
4650 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4652 * predict.c (force_edge_cold): Handle declaring edges impossible
4655 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4657 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4659 (try_unroll_loop_completely): Fix reporting.
4661 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4663 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4665 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4668 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4669 if given a non-constant argument for an intrinsic which requires a
4672 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4674 * profile.c (compare_freqs): New function.
4675 (branch_prob): Sort edge list.
4676 (find_spanning_tree): Assume that the list is priority sorted.
4678 2017-06-16 Richard Biener <rguenther@suse.de>
4680 PR tree-optimization/81090
4681 * passes.def (pass_record_bounds): Remove.
4682 * tree-pass.h (make_pass_record_bounds): Likewise.
4683 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4684 make_pass_record_bounds): Likewise.
4685 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4686 not free niter estimates at the beginning but at the end.
4687 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4689 2017-06-16 Richard Biener <rguenther@suse.de>
4691 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4692 initializer to workaround ICE in host GCC 4.8.
4694 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4696 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4698 (clone_inlined_nodes): Update.
4700 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4702 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4703 prefetch settings, and enable prefetching by default at -O3.
4705 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4707 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4708 Set flag_prefetch_loop_arrays according to tuning data.
4710 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4712 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4714 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4715 [Unrelated to main purpose of the patch] Place the pointer field last
4716 to enable type checking errors when tune structure are wrongly merged.
4717 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4718 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4719 (thunderx2t99_prefetch_tune): New tune constants.
4720 (tune_params *_tunings): Update all tunings (no functional change).
4721 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4722 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4723 from tunings structures.
4725 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4728 * ubsan.c (instrument_null): Add T argument, use it instead
4729 of computing it based on IS_LHS.
4730 (instrument_object_size): Likewise.
4731 (pass_ubsan::execute): Adjust instrument_null and
4732 instrument_object_size callers to pass gimple_get_lhs or
4733 gimple_assign_rhs1 result to it. Use instrument_null instead of
4734 calling get_base_address and instrument_mem_ref. Handle
4735 aggregate call arguments for object-size sanitization.
4737 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4739 PR tree-optimization/81089
4740 * tree-vrp.c (is_masked_range_test): Validate operands of
4743 2017-06-15 Martin Sebor <msebor@redhat.com>
4746 * dumpfile.c (dump_register): Avoid calling memset to initialize
4747 a class with a default ctor.
4748 * gcc.c (struct compiler): Remove const qualification.
4749 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4750 * hash-table.h: Ditto.
4751 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4753 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4754 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4756 * params.h (struct param_info): Make struct members non-const.
4757 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4758 with default initialization.
4759 * vec.h (vec_copy_construct, vec_default_construct): New helper
4761 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4762 with vec_copy_construct.
4763 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
4764 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
4765 * doc/invoke.texi (-Wclass-memaccess): Document.
4767 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4769 * emit-rtl.h (is_leaf): Update comment about local
4772 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4775 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
4776 for a variable to have a section before checking if the section has a
4778 Set section to.persistent if persistent attribute is set.
4779 Warn if .persistent attribute is used on an automatic variable.
4781 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
4783 PR rtl-optimization/80474
4784 * reorg.c (update_block): Do not ignore instructions in a delay slot.
4786 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
4788 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
4791 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
4793 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
4794 (casesi): Emit bounds checking as RTL.
4795 (casesi_internal_mips16_<mode>): Remove bounds checking.
4797 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
4799 * config/xtensa/xtensa.c (xtensa_option_override): Append
4800 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
4801 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
4802 xtensa_doloop_hooks): Define unconditionally.
4803 (xtensa_reorg_loops): Only call reorg_loops in the presence of
4805 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
4806 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
4807 for it in xtensa_option_override.
4808 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
4809 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
4811 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
4813 * doc/cppopts.texi: Document '-' special value to -MF.
4815 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
4817 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
4818 (cortex_a53_fconst): Likewise.
4819 (cortex_a53_fpmul): Likewise.
4820 (cortex_a53_f_load_64): Likewise.
4821 (cortex_a53_f_load_many): Likewise.
4822 (cortex_a53_advsimd_alu): Likewise.
4823 (cortex_a53_advsimd_alu_q): Likewise.
4824 (cortex_a53_advsimd_mul): Likewise.
4825 (cortex_a53_advsimd_mul_q): Likewise.
4826 (fpmac bypass): Add new bypass for fpmac-fpmac case.
4827 Add missing fmul, r2f_cvt and fconst cases.
4829 2017-06-14 Richard Biener <rguenther@suse.de>
4832 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
4834 (fold_binary_loc): When associating do not treat pre-existing
4835 TREE_OVERFLOW on literal constants as a reason to allow
4836 TREE_OVERFLOW on associated literal constants.
4838 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4840 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
4841 (MASK_FEATURES): New macro.
4842 * config/sparc/sparc.c (sparc_option_override): Remove the special
4843 handling of -mfpu and generalize it to all MASK_FEATURES switches.
4845 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4847 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
4848 a division of 0 if non-call exceptions are enabled.
4850 2017-06-14 Andrew Pinski <apinski@cavium.com>
4851 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4854 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
4855 Improve vector initialization code gen for only variable case.
4857 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4859 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
4861 2017-06-14 Richard Biener <rguenther@suse.de>
4863 PR tree-optimization/81083
4864 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
4867 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4869 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
4870 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
4871 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
4872 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
4873 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
4874 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
4876 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4878 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
4879 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
4881 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4883 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
4885 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4887 * config/rs6000/t-rtems: Don't handle SPE.
4889 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4891 * config/rs6000/t-linux: Don't handle SPE.
4893 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4895 * config/rs6000/eabispe.h: Delete file.
4897 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4899 * config/rs6000/t-spe: Delete file.
4901 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4903 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
4904 (rs6000_legitimate_offset_address_p): Return false for anything in
4905 V2SImode or V2SFmode.
4907 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4909 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
4910 except V2SF and V2SI. Rearrange the vector modes, and add comments.
4911 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
4913 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
4914 (rs6000_legitimate_offset_address_p): Ditto.
4915 (rs6000_emit_move): Ditto.
4916 (rs6000_init_builtins): Remove V4HI_type_node.
4918 2017-06-13 Martin Liska <mliska@suse.cz>
4921 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
4922 (gate_asan): Likewise.
4923 * asan.h (asan_no_sanitize_address_p): Remove the function.
4924 (sanitize_flags_p): New function.
4925 * builtins.def: Fix coding style.
4926 * common.opt: Use renamed enum value.
4927 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
4928 * doc/extend.texi: Document no_sanitize attribute.
4929 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
4930 to SANITIZE_UNDEFINED_NONDEFAULT.
4931 * gcc.c (sanitize_spec_function): Use the renamed enum value.
4932 * gimple-fold.c (optimize_atomic_compare_exchange_p):
4933 Use sanitize_flags_p.
4934 * gimplify.c (gimplify_function_tree): Likewise.
4935 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
4936 * opts.c (parse_no_sanitize_attribute): New function.
4937 (common_handle_option): Use renamed enum value.
4938 * opts.h (parse_no_sanitize_attribute): Declare.
4939 * tree.c (sanitize_flags_p): New function.
4940 * tree.h: Declared here.
4941 * tsan.c: Use sanitize_flags_p.
4942 * ubsan.c (ubsan_expand_null_ifn): Likewise.
4943 (instrument_mem_ref): Likewise.
4944 (instrument_bool_enum_load): Likewise.
4945 (do_ubsan_in_current_function): Remove the function.
4946 (pass_ubsan::execute): Use sanitize_flags_p.
4947 * ubsan.h: Remove do_ubsan_in_current_function
4948 * tree-cfg.c (print_no_sanitize_attr_value): New function.
4949 (dump_function_to_file): Use it here.
4951 2017-06-13 Martin Jambor <mjambor@suse.cz>
4953 PR tree-optimization/80803
4954 PR tree-optimization/81063
4955 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
4956 (propagate_subaccesses_across_link): Enqueue subtree whenever
4957 necessary instead of relying on the caller.
4959 2017-06-13 Martin Jambor <mjambor@suse.cz>
4961 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
4962 that have a first_link.
4963 (sort_and_splice_var_accesses): Do not check first_link before
4965 (subtree_mark_written_and_enqueue): Likewise.
4966 (propagate_all_subaccesses): Likewise and do not stop at first
4967 parent with a first_link.
4969 2017-06-13 Martin Jambor <mjambor@suse.cz>
4971 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
4974 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4976 * match.pd: New pattern.
4978 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4980 * tree-vrp.c (is_masked_range_test): New function.
4981 (register_edge_assert_for): Determine ranges for
4984 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4986 PR tree-optimization/67328
4987 * fold-const.c (maskable_range_p): New function.
4988 (build_range_check): Generate bittests if possible.
4990 2017-06-13 Martin Liska <mliska@suse.cz>
4992 * gimple-pretty-print.c (dump_probability): Add new argument.
4993 (dump_edge_probability): Dump both probability and count.
4994 (dump_gimple_label): Likewise.
4995 (dump_gimple_bb_header): Likewise.
4997 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5000 * config/avr/avr-devices.c: Fix indentation.
5001 * config/avr/gen-avr-mmcu-specs.c: Dito.
5003 2017-06-13 Richard Biener <rguenther@suse.de>
5005 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5006 instead get vector type from stmt_info.
5007 (vectorizable_reduction): Adjust. Remove dead code.
5009 2017-06-13 Richard Biener <rguenther@suse.de>
5012 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5013 case of C * (x * C2 + C3).
5014 (fold_addr_of_array_ref_difference): Properly fold index difference.
5016 2017-06-12 David S. Miller <davem@davemloft.net>
5019 * config/sparc/sparc.md (return expander): Emit frame blockage if
5020 function uses alloca.
5022 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5024 * combine.c (make_field_assignment): Check len rather than the mode
5025 precision when calling force_to_mode.
5027 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5029 Support multilibs and devices that see flash in RAM address range.
5032 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5033 (avr_mcu_t) <flash_pm_offset>: New field.
5034 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5035 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5036 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5037 (AVR_TINY_PM_OFFSET): Remove macro.
5038 * config/avr/avr.opt (-mshort-calls): New option.
5039 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5040 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5041 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5042 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5043 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5044 instead of avr_arch->have_jmp_call.
5045 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5046 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5047 avr_arch->flash_pm_offset to define.
5048 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5049 new field flash_pm_offset. Add entry for avrxmega3.
5050 (avr_texinfo): Add entry for avrxmega3.
5051 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5052 attiny212, attiny214,
5053 attiny412, attiny414, attiny416, attiny417,
5054 attiny814, attiny816, attiny817,
5055 attiny1614, attiny1616, attiny1617,
5056 attiny3214, attiny3216, attiny3217.
5057 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5058 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5059 (avr_print_operand_address) [AVR_TINY]: Same.
5060 (avr_asm_init_sections) <readonly_data_section>: Only patch
5061 callback if avr_arch->flash_pm_offset = 0.
5062 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5063 for rodata if avr_arch->flash_pm_offset != 0.
5064 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5065 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5066 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5067 (m_options): Append opt_rcall.
5068 (m_dirnames): Append dir_rcall.
5069 * config/avr/t-multilib: Regenerate.
5071 * configure.ac [target=avr]: Check whether avrxmega3 default
5072 linker description file works as needed.
5073 * configure: Regenerate.
5074 * doc/avr-mmcu.texi: Regenerate.
5075 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5076 <__AVR_ARCH__>: Document avrxmega3 and 103.
5077 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5078 <__AVR_SHORT_CALLS__>: Document it.
5079 <__AVR_PM_BASE_ADDRESS__>: Document it.
5080 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5081 (AVR Variable Attributes) <progmem>: Document this is
5082 not needed for avrxmega3.
5083 (AVR Named Address Spaces) <__flash>: Dito.
5085 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5087 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5089 2017-06-12 Doug Rupp <rupp@adacore.com>
5091 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5092 Append vxworks-stdint.h to the tm_file list.
5093 * config/vxworks-stdint.h: New file.
5095 2017-06-12 Martin Liska <mliska@suse.cz>
5097 PR tree-optimization/81041
5098 * tree-profile.c (gimple_gen_ic_func_profiler):
5099 Create an extra BB in profile-generate
5100 (gimple_gen_time_profiler): Likewise.
5102 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5104 PR tree-optimization/81003
5105 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5106 (update_range_test): Use it instead of force_gimple_operand_gsi.
5108 2017-06-12 Richard Biener <rguenther@suse.de>
5110 PR tree-optimization/81053
5111 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5112 with backedge value not defined in loop. Simplify def stmt
5115 2017-06-11 Tom de Vries <tom@codesourcery.com>
5118 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5120 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5121 nvptx_cannot_force_const_mem.
5123 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5125 * opts.c (finish_options): Move test for flag_split_stack after
5126 it has been initialized.
5128 2017-06-11 Jason Merrill <jason@redhat.com>
5130 * tree.h (id_equal): New.
5131 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5132 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5133 instead of strcmp of IDENTIFIER_POINTER.
5135 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5137 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5138 (mark_all_inlined_calls_cdtor): Fix formating.
5139 (inline_transform): Rescale profile before inlining.
5141 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5143 * cgraph.h (cgraph_edge::clone): Update prototype.
5144 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5145 (cgraph_node::create_clone): Update.
5146 (cgraph_node::create_version_clone): Update.
5147 * tree-inline.c (copy_bb): Update.
5148 (expand_call_inline): Update.
5150 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5152 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5153 factored out from ...
5154 (rs6000_emit_prologue): ... here.
5156 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5158 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5159 factored out from ...
5160 (rs6000_emit_prologue): ... here.
5162 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5164 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5166 (handle_missing_profiles): Fix computation of tp_first_run.
5167 (counts_to_freqs): Do not touch freqs when count is 0.
5169 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5171 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5174 2017-06-10 Tom de Vries <tom@codesourcery.com>
5176 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5177 attributes): Document signal effective target.
5179 2017-06-10 Tom de Vries <tom@codesourcery.com>
5181 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5182 Document effective target stack_size.
5184 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5186 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5187 to the edit_context if they can be auto-applied.
5189 2017-06-9 Ian Lance Taylor <iant@golang.org>
5191 * opts.c (finish_options): If -fsplit-stack, disable implicit
5192 -forder-blocks-and-partition.
5193 * doc/invoke.texi (Optimize Options): Document that when using
5194 -fsplit-stack -forder-blocks-and-partition is not implicitly
5197 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5199 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5200 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5201 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5202 * builtins.def (abort, trap, unreachable): Declare cold.
5203 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5204 * tree-core.h (ECF_COLD): New.
5205 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5206 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5208 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5210 * predict.c (unlikely_executed_stmt_p): Cleanup.
5212 2017-06-09 Richard Biener <rguenther@suse.de>
5214 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5215 model if the ref is always written to.
5217 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5219 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5221 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5223 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5226 2017-06-09 Tom de Vries <tom@codesourcery.com>
5229 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5230 "target cannot support label values" when encountering LABEL_REF.
5232 2017-06-09 Martin Liska <mliska@suse.cz>
5234 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5235 (gimple_gen_ic_func_profiler): Emit direct comparison
5236 of __gcov_indirect_call_callee with NULL.
5237 (gimple_gen_time_profiler): Change probability from
5238 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5240 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5242 * profile.c (edge_gcov_counts): Turn to pointer.
5243 (compute_branch_probabilities, compute_branch_probabilities): Update.
5244 (branch_prob): Do not clear edge_gcov_count.
5245 * profile.h (edge_gcov_counts): Turn to pointer.
5246 (edge_gcov_count): Update.
5248 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5250 * gimple.h (gimple_check_failed): Mark cold.
5252 2017-06-09 Richard Biener <rguenther@suse.de>
5254 PR tree-optimization/66623
5255 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5256 refactor check_reduction into two parts, properly computing
5257 whether we have to check reduction validity for outer loop
5260 2017-06-09 Richard Biener <rguenther@suse.de>
5262 PR tree-optimization/79483
5263 * graphite-scop-detection.c (order): New global.
5264 (get_order): Compute bb to order mapping that satisfies code
5265 generation constraints.
5266 (cmp_pbbs): New helper.
5267 (build_scops): Start domwalk at entry block, sort generated
5270 2017-06-09 Richard Biener <rguenther@suse.de>
5273 * ipa-polymorphic-call.c
5274 (ipa_polymorphic_call_context::restrict_to_inner_class):
5275 Skip FIELD_DECLs with error_mark_node type.
5276 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5279 2017-06-09 Martin Liska <mliska@suse.cz>
5281 * predict.c (struct branch_predictor): New struct.
5282 (test_prediction_value_range): New test.
5283 (predict_c_tests): New function.
5284 * selftest-run-tests.c (selftest::run_tests): Run the function.
5285 * selftest.h: Declare new tests.
5287 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5290 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5291 gen_add3_insn did not fail.
5292 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5293 r0, construct that number in a temporary reg and add that reg to r0.
5294 If asked to put the result in r0 as well, fail.
5296 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5298 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5299 for early expansion of vec_eqv.
5301 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5304 * ubsan.c (instrument_null): Avoid pointless code temporary.
5305 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5307 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5311 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5314 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5317 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5319 * predict.c (unlikely_executed_edge_p): Move ahead.
5320 (probably_never_executed_edge_p): Use it.
5322 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5325 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5326 gimple_call_builtin_p call.
5328 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5330 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5331 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5332 rtl_check_failed_type2, rtl_check_failed_code1,
5333 rtl_check_failed_code2, rtl_check_failed_code_mode,
5334 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5335 rtvec_check_failed_bounds, rtl_check_failed_flag,
5336 _fatal_insn_not_found, _fatal_insn): Likewise.
5337 * tree.h (tree_contains_struct_check_failed,
5338 tree_check_failed, tree_not_check_failed,
5339 tree_class_check_failed, tree_range_check_failed,
5340 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5341 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5342 tree_operand_check_failed, omp_clause_check_failed,
5343 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5346 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5348 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5349 flag_branch_probabilities.
5350 * ipa-inline.c (edge_badness): Likewise.
5351 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5352 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5353 * predict.c (maybe_hot_frequency_p): Likewise.
5354 (probably_never_executed): Likewise.
5355 * sched-ebb.c (schedule_ebbs): Likewise.
5356 * sched-rgn.c (find_single_block_region): Likewise.
5357 * tracer.c (tail_duplicate): Likewise.
5359 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5361 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5362 longer requires x_flag_profile_use.
5364 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5366 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5367 instead of flag_reorder_blocks_and_partition.
5368 * dbxout.c (dbxout_function_end): Likewise.
5369 * dwarf2out.c (gen_subprogram_die): Likewise.
5370 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5371 * hw-doloop.c (reorg_loops): Likewise.
5372 * varasm.c (assemble_start_function,
5373 assemble_end_function): Likewise.
5374 (decide_function_section): Do not check for
5375 flag_reorder_blocks_and_partition.
5377 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5379 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5381 (chkp_get_hard_register_fake_addr_expr): Ditto.
5382 (chkp_build_addr_expr): Add check for hard reg case.
5383 (chkp_parse_array_and_component_ref): Ditto.
5384 (chkp_find_bounds_1): Ditto.
5385 (chkp_process_stmt): Don't generate bounds store for
5388 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5390 * predict.c (maybe_hot_bb_p): Do not check profile status.
5391 (maybe_hot_edge_p): Likewise.
5392 (probably_never_executed): Check for zero counts even if profile
5394 (unlikely_executed_edge_p): New function.
5395 (unlikely_executed_stmt_p): New function.
5396 (unlikely_executed_bb_p): New function.
5397 (set_even_probabilities): Use unlikely predicates.
5398 (combine_predictions_for_bb): Likewise.
5399 (predict_paths_for_bb): Likewise.
5400 (predict_paths_leading_to_edge): Likewise.
5401 (determine_unlikely_bbs): New function.
5402 (estimate_bb_frequencies): Use it.
5403 (compute_function_frequency): Use zero counts even if profile is
5405 * profile-count.h: Fix typo.
5407 2017-08-08 Julia Koval <julia.koval@intel.com>
5409 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5410 _mm512_mask_cvtsepi16_storeu_epi8,
5411 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5412 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5413 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5414 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5415 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5416 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5417 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5418 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5419 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5420 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5421 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5422 __builtin_ia32_pmovuswb256mem_mask,
5423 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5424 __builtin_ia32_pmovwb512mem_mask): New builtins.
5426 2017-08-08 Julia Koval <julia.koval@intel.com>
5428 PR target/73350,80862
5429 * config/i386/subst.md (round): Fix round pattern.
5430 * config/i386/i386.c (ix86_erase_embedded_rounding):
5431 Fix erasing rounding for the fixed pattern.
5433 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5435 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5437 2017-06-08 Martin Liska <mliska@suse.cz>
5439 PR gcov-profile/80911
5440 * gcov.c (block_info::block_info): New constructor.
5442 2017-06-07 Carl Love <cel@us.ibm.com>
5444 * config/rs6000/rs6000-c: The return type of the following
5445 built-in functions was implemented as int not long long. Fix sign
5446 of return value for the unsigned version of vec_mulo and vec_mule.
5447 vector unsigned long long vec_bperm (vector unsigned long long,
5448 vector unsigned char)
5449 vector signed long long vec_mule (vector signed int,
5451 vector unsigned long long vec_mule (vector unsigned int,
5452 vector unsigned int)
5453 vector signed long long vec_mulo (vector signed int,
5455 vector unsigned long long vec_mulo (vector unsigned int,
5456 vector unsigned int)
5457 * doc/extend.texi: Fix the documentation for the built-in
5460 2017-06-07 Carl Love <cel@us.ibm.com>
5463 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5466 2017-06-07 Carl Love <cel@us.ibm.com>
5468 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5469 support, Generate doublehv for signed int/float for BE case only.
5471 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5473 * doc/invoke.texi (mcx16): Rewrite.
5475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5477 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5478 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5479 *mov<mode>_softfloat, and an anonymous splitter): Use
5480 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5482 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5484 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5486 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5487 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5488 (rs6000_debug_reg_global): Adjust.
5489 (rs6000_init_hard_regno_mode_ok): Adjust.
5490 (rs6000_dbx_register_number): Adjust.
5491 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5492 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5493 Remove SPE_ACC and SPEFSCR.
5494 (REG_ALLOC_ORDER): Ditto.
5495 (FRAME_POINTER_REGNUM): Change to 111.
5496 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5497 (REG_CLASS_NAMES): Ditto.
5498 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5499 (REGISTER_NAMES): Ditto.
5500 (ADDITIONAL_REG_NAMES): Ditto.
5501 (rs6000_reg_names): Ditto.
5502 * config/rs6000/rs6000.md: Renumber some register number
5505 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5507 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5509 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5510 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5512 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5513 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5514 Delete the SPE high registers.
5515 (REG_ALLOC_ORDER): Ditto.
5516 (enum reg_class): Remove SPE_HIGH_REGS.
5517 (REG_CLASS_NAMES): Ditto.
5518 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5519 (REGISTER_NAMES): Ditto.
5520 (rs6000_reg_names): Ditto.
5521 * doc/tm.texi.in: Remove SPE as example.
5522 * doc/tm.texi: Regenerate.
5524 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5526 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5527 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5528 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5529 * config/rs6000/rs6000.md (type): Remove "brinc".
5531 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5533 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5534 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5535 * config/rs6000/linuxspe.h: Delete file.
5536 * config/rs6000/rs6000.md: Don't include spe.md.
5537 * config/rs6000/spe.h: Delete file.
5538 * config/rs6000/spe.md: Delete file.
5539 * config/rs6000/t-rs6000: Remove spe.md.
5541 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5543 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5544 (reg_or_none500mem_operand): Delete.
5545 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5546 instead of reg_or_none500mem_operand.
5548 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5550 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5551 handling of SPE flags.
5552 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5554 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5556 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5558 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5559 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5560 paired_divv2sf3): Similar.
5561 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5562 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5563 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5565 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5566 Rename the paired_* instruction patterns.
5567 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5569 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5570 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5571 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5572 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5573 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5575 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5576 (spe_func_has_64bit_regs_p): Delete.
5577 (spe_expand_predicate_builtin): Delete.
5578 (spe_expand_evsel_builtin): Delete.
5579 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5580 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5581 (invalid_e500_subreg): Delete.
5582 (rs6000_legitimize_address): Always force_reg op2 as well, for
5583 paired single memory accesses.
5584 (rs6000_member_type_forces_blk): Delete.
5585 (rs6000_spe_function_arg): Delete.
5586 (rs6000_expand_unop_builtin): Delete SPE handling.
5587 (rs6000_expand_binop_builtin): Ditto.
5588 (spe_expand_stv_builtin): Delete.
5589 (bdesc_2arg_spe): Delete.
5590 (spe_expand_builtin): Delete.
5591 (spe_expand_predicate_builtin): Delete.
5592 (spe_expand_evsel_builtin): Delete.
5593 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5594 (spe_init_builtins): Delete.
5595 (spe_func_has_64bit_regs_p): Delete.
5596 (savres_routine_name): Delete "info" parameter. Adjust callers.
5597 (rs6000_emit_stack_reset): Ditto.
5598 (rs6000_dwarf_register_span): Delete.
5599 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5600 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5601 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5603 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5605 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5606 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5607 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5608 mulv2sf3, divv2sf3): Delete expanders.
5610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5612 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5614 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5616 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5617 * config/rs6000/rs6000.c: Ditto.
5619 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5621 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5622 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5623 comparison_operator.
5625 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5627 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5628 * config/rs6000/rs6000.opt: Ditto.
5629 * config/rs6000/t-rtems: Ditto.
5631 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5633 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5634 TARGET_E500_SINGLE by 0, simplify.
5635 * config/rs6000/rs6000.c: Ditto.
5636 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5637 (spe_build_register_parallel): Delete.
5638 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5639 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5640 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5641 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5642 (E500_CONVERT): Delete.
5643 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5645 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5647 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5648 * config/rs6000/dfp.md: Ditto.
5649 (negdd2, *negdd2_fpr): Merge.
5650 (absdd2, *absdd2_fpr): Merge.
5651 (negtd2, *negtd2_fpr): Merge.
5652 (abstd2, *abstd2_fpr): Merge.
5653 * config/rs6000/e500.h: Delete file.
5654 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5655 TARGET_FPRS by 1 and simplify.
5656 * config/rs6000/rs6000-c.c: Ditto.
5657 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5659 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5661 * config/rs6000/rs6000.md: Ditto.
5662 (floatdidf2, *floatdidf2_fpr): Merge.
5663 (move_from_CR_gt_bit): Delete.
5664 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5665 (E500_CR_IOR_COMPARE): Delete.
5666 (All patterns that require !TARGET_FPRS): Delete.
5667 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5669 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5671 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5673 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5675 * graphds.c (add_edge): Intitialize edge's attached data.
5676 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5677 pointer parameter. Call pointed function on each edge during
5678 graph traversing. Skip traversing the edge when the function
5680 (graphds_dfs, graphds_scc): Ditto.
5681 (for_each_edge): New parameter. Pass the new parameter to callback
5683 * graphds.h (skip_edge_callback): New function pointer type.
5684 (graphds_dfs, graphds_scc): New function pointer parameter.
5685 (graphds_edge_callback, for_each_edge): New parameter.
5687 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5689 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5690 out code checking if runtime alias check is possible to below ...
5691 Call the new function.
5692 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5693 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5695 2017-06-07 Marek Polacek <polacek@redhat.com>
5698 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5699 TYPE_OVERFLOW_WRAPS check.
5701 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5703 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5704 if versioning is required.
5705 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5706 peeling with the check for versioning.
5708 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5710 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5711 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5712 Set true to new parameter if new ssa variable is defined.
5713 (vect_gen_vector_loop_niters): Refactor. Set range information
5714 for the new vector loop bound variable.
5715 (vect_do_peeling): Ditto.
5717 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5719 * tree-affine.c (ssa.h): Include header file.
5720 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5721 has wrapping overflow behavior.
5723 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5725 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5727 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5729 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5730 (tree_to_aff_combination): ... here.
5732 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5734 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5735 reg_pressure model function.
5736 (ivopts_global_cost_for_size): Delete.
5737 (determine_set_costs, iv_ca_recount_cost): Call new model function
5738 ivopts_estimate_reg_pressure.
5740 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5742 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5743 expensive than udiv. Remove floating point cases from mod.
5745 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5747 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5750 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5752 * config/aarch64/aarch64.md
5753 (copysignsf3): Fix mask generation.
5755 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5757 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5759 (class dump_manager): Add register_dumps method.
5760 * dumpfile.c: Include langhooks.h.
5761 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5762 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
5763 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
5764 (dump_manager::dump_register): Start with 512 entries instead of 32.
5765 (dump_manager::register_dumps): New method.
5766 * toplev.c (general_init): Instead of invoking register_dumps
5767 langhook, invoke register_dumps method on the dump manager.
5768 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
5771 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
5773 * doc/md.texi: Clarify the restrictions on a define_insn condition.
5774 Say that # requires an associated define_split to exist, and that
5775 the define_split must be suitable for use after register allocation.
5777 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5779 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
5780 (compute_outgoing_frequencies): Also initialize zero counts.
5781 (find_many_sub_basic_blocks): Do not produce uninitialized profile
5782 around loops; preserve more of profile when nothing changes.
5784 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
5786 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
5788 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
5789 * config/arm/arm-cpu-cdata.h: Regenerate.
5790 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
5791 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
5792 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
5793 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
5794 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
5796 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
5798 * config/arm/t-rmprofile: Likewise.
5799 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
5801 2017-06-06 David S. Miller <davem@davemloft.net>
5804 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
5805 blockage if function uses alloca.
5807 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5809 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
5810 New "uid" fields to hold pretty-print IDs of group and ref.
5811 Memory references are now identified as <group_id>:<ref_id>
5812 instead of using [random] addresses.
5813 (dump_mem_details): Simplify, no functional change.
5814 (dump_mem_ref): Simplify and make output more concise.
5815 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
5816 (find_or_create_group): Initialize group uid.
5817 (record_ref): Initialize ref uid. Improve debug output.
5818 (prune_group_by_reuse, should_issue_prefetch_p,)
5819 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
5820 (mark_nontemporal_store, determine_loop_nest_reuse):
5821 Improve debug output.
5823 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5825 * dbgcnt.def (prefetch): New debug counter.
5826 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
5827 (schedule_prefetches): Stop issueing prefetches if debug counter
5830 2017-06-06 Tom de Vries <tom@codesourcery.com>
5832 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
5833 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
5835 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5837 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
5838 Use aarch64_reg_or_zero predicate for operand 4.
5839 (aarch64_compare_and_swap<mode> define_insn_and_split):
5840 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
5841 (aarch64_store_exclusive<mode>): Likewise for operand 2.
5843 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5845 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
5846 (arm_compute_save_core_reg_mask): This.
5847 (thumb1_compute_save_reg_mask): Rename into ...
5848 (thumb1_compute_save_core_reg_mask): This.
5849 (arm_compute_save_reg0_reg12_mask): Adapt comment.
5850 (arm_compute_frame_layout): Likewise.
5852 2017-06-06 Richard Biener <rguenther@suse.de>
5854 PR tree-optimization/80974
5855 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
5856 keep or clear leaders SSA info.
5858 2017-06-06 Tom de Vries <tom@codesourcery.com>
5860 * config/nvptx/nvptx.c (split_mode_p): New function.
5861 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
5863 2017-06-06 Tom de Vries <tom@codesourcery.com>
5865 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
5867 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5870 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
5873 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5875 * shrink-wrap.c (handle_simple_exit): Update profile.
5876 (try_shrink_wrapping): Upate profile.
5878 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5880 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
5881 (tree_guess_outgoing_edge_probabilities): New.
5882 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
5883 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
5885 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5887 * ipa-split.c (split_function): Initialize return bb profile.
5889 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5891 * profile.c (compute_branch_probabilities): Also initialize
5894 2017-06-06 Richard Biener <rguenther@suse.de>
5896 PR tree-optimization/80928
5897 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
5898 (vect_analyze_loop_operations): Properly guard analysis for
5900 (vect_transform_loop): Likewise.
5901 (vect_analyze_loop_2): Also reset SLP type on PHIs.
5902 (vect_model_induction_cost): Do not cost for pure SLP.
5903 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
5904 of induction in inner loop vectorization.
5905 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
5906 (vect_get_and_check_slp_defs): Handle vect_induction_def.
5907 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
5909 (vect_analyze_slp_cost_1): Cost induction.
5910 (vect_detect_hybrid_slp_stmts): Handle PHIs.
5911 (vect_get_slp_vect_defs): Likewise.
5912 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
5913 (vect_transform_stmt): Handle SLP reductions.
5914 * tree-vectorizer.h (vectorizable_induction): Adjust.
5916 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5918 * config/rs6000/rs6000.c (make_resolver_func): Update
5919 init_lowered_empty_function call.
5921 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
5923 * doc/invoke.texi: Document the -fprofile-abs-path option.
5924 * common.opt (fprofile-abs-path): New option.
5925 * gcov-io.h (gcov_write_filename): Declare.
5926 * gcov-io.c (gcov_write_filename): New function.
5927 * coverage.c (coverage_begin_function): Use gcov_write_filename.
5928 * profile.c (output_location): Likewise.
5930 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5932 * shring-wrap.c: Revert accidental commit.
5934 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
5936 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
5938 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5940 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
5942 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
5943 profile in callgraph edge.
5944 * profile-count.h (apply_probability): If THIS is 0, then result is 0
5945 (apply_scale): Likewise.
5946 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
5947 Also scale profile when inlining function with zero profile.
5948 (initialize_cfun): Update exit block profile even when it is zero.
5949 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
5950 when profile is read.
5952 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5954 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
5955 (CLONE_*): New constants to define the processors we can generate
5956 code for with the target_clone attribute.
5957 (rs6000_clone_map): New array to identify which clone processors
5958 the current program is running on.
5959 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
5960 target_clone attribute.
5961 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
5962 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
5963 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
5964 (cpu_expand_builtin): Add support for target_clone attribute.
5965 (rs6000_valid_attribute_p): Allow "default" attribute.
5966 (get_decl_name): New debug function to simplify printing the
5967 current function name in debugging statements.
5968 (rs6000_clone_priority): New functions to support the target_clone
5969 attribute, and be able to generate code to switch between ISA 2.05
5970 through ISA 3.0 (power6 through power9).
5971 (rs6000_compare_version_priority): Likewise.
5972 (rs6000_get_function_versions_dispatcher): Likewise.
5973 (make_resolver_func): Likewise.
5974 (add_condition_to_bb): Likewise.
5975 (dispatch_function_versions): Likewise.
5976 (rs6000_generate_version_dispatcher_body): Likewise.
5977 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
5978 (fusion_gpr_load_p): Fix a spacing issue.
5979 * doc/extend.texi (Common Function Attributes): Document that the
5980 PowerPC supports the target_clone attribute.
5982 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
5984 * config/arm/arm.h: explain F symbol found in description of ARM
5985 register allocation in its legend.
5987 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5989 * config/mips/frame-header-opt.c: Include profile-count.h.
5990 * config/riscv/riscv.c: Include profile-count.h
5992 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5994 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
5996 (sm_set_flag_if_changed): Add bbs field.
5997 (execute_sm_if_changed_flag_set): Pass BBS.
5998 (execute_sm): Update.
6000 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6002 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6005 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6007 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6008 (peephole2): New peephole2 to emit the above.
6009 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6011 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6013 * config/aarch64/aarch64.c (define_peephole2 above
6014 *sub_<shift>_<mode>): New peephole.
6016 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6018 * config/i386/i386.c (make_resolver_func): Update.
6019 * Makefile.in: Add profile-count.h and profile-count.o
6020 * auto-profile.c (afdo_indirect_call): Update to new API.
6021 (afdo_set_bb_count): Update.
6022 (afdo_propagate_edge): Update.
6023 (afdo_propagate_circuit): Update.
6024 (afdo_calculate_branch_prob): Update.
6025 (afdo_annotate_cfg): Update.
6026 * basic-block.h: Include profile-count.h
6027 (struct edge_def): Turn count to profile_count.
6028 (struct basic_block_def): Likewie.
6029 (REG_BR_PROB_BASE): Move to profile-count.h
6030 (RDIV): Move to profile-count.h
6031 * bb-reorder.c (max_entry_count): Turn to profile_count.
6032 (find_traces): Update.
6033 (rotate_loop):Update.
6034 (connect_traces):Update.
6035 (sanitize_hot_paths):Update.
6036 * bt-load.c (migrate_btr_defs): Update.
6037 * cfg.c (RDIV): Remove.
6038 (init_flow): Use alloc_block.
6039 (alloc_block): Uninitialize count.
6040 (unchecked_make_edge): Uninitialize count.
6041 (check_bb_profile): Update.
6042 (dump_edge_info): Update.
6043 (dump_bb_info): Update.
6044 (update_bb_profile_for_threading): Update.
6045 (scale_bbs_frequencies_int): Update.
6046 (scale_bbs_frequencies_gcov_type): Update.
6047 (scale_bbs_frequencies_profile_count): New.
6048 * cfg.h (update_bb_profile_for_threading): Update.
6049 (scale_bbs_frequencies_profile_count): Declare.
6050 * cfgbuild.c (compute_outgoing_frequencies): Update.
6051 (find_many_sub_basic_blocks): Update.
6052 * cfgcleanup.c (try_forward_edges): Update.
6053 (try_crossjump_to_edge): Update.
6054 * cfgexpand.c (expand_gimple_tailcall): Update.
6055 (construct_exit_block): Update.
6056 * cfghooks.c (verify_flow_info): Update.
6057 (dump_bb_for_graph): Update.
6058 (split_edge): Update.
6059 (make_forwarder_block): Update.
6060 (duplicate_block): Update.
6061 (account_profile_record): Update.
6062 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6063 (get_estimated_loop_iterations): Update.
6064 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6065 (single_likely_exit): Update.
6066 * cfgloopmanip.c (scale_loop_profile): Update.
6068 (set_zero_probability): Update.
6069 (lv_adjust_loop_entry_edge): Update.
6070 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6071 (purge_dead_edges): Update.
6072 (rtl_account_profile_record): Update.
6073 * cgraph.c (cgraph_node::create): Uninitialize count.
6074 (symbol_table::create_edge): Uninitialize count.
6075 (cgraph_update_edges_for_call_stmt_node): Update.
6076 (cgraph_edge::dump_edge_flags): Update.
6077 (cgraph_node::dump): Update.
6078 (cgraph_edge::maybe_hot_p): Update.
6079 * cgraph.h: Include profile-count.h
6080 (create_clone), create_edge, create_indirect_edge): Update.
6081 (cgraph_node): Turn count to profile_count.
6082 (cgraph_edge0: Likewise.
6083 (make_speculative, clone): Update.
6084 (create_edge): Update.
6085 (init_lowered_empty_function): Update.
6086 * cgraphclones.c (cgraph_edge::clone): Update.
6087 (duplicate_thunk_for_node): Update.
6088 (cgraph_node::create_clone): Update.
6089 * cgraphunit.c (cgraph_node::analyze): Update.
6090 (cgraph_node::expand_thunk): Update.
6091 * final.c (dump_basic_block_info): Update.
6092 * gimple-streamer-in.c (input_bb): Update.
6093 * gimple-streamer-out.c (output_bb): Update.
6094 * graphite.c (print_global_statistics): Update.
6095 (print_graphite_scop_statistics): Update.
6096 * hsa-brig.c: Include basic-block.h.
6097 * hsa-dump.c: Include basic-block.h.
6098 * hsa-gen.c (T sum_slice): Update.
6099 (convert_switch_statements):Update.
6100 * hsa-regalloc.c: Include basic-block.h.
6101 * ipa-chkp.c (chkp_produce_thunks): Update.
6102 * ipa-cp.c (struct caller_statistics): Update.
6103 (init_caller_stats): Update.
6104 (gather_caller_stats): Update.
6105 (ipcp_cloning_candidate_p): Update.
6106 (good_cloning_opportunity_p): Update.
6107 (get_info_about_necessary_edges): Update.
6108 (dump_profile_updates): Update.
6109 (update_profiling_info): Update.
6110 (update_specialized_profile): Update.
6111 (perhaps_add_new_callers): Update.
6112 (decide_about_value): Update.
6113 (ipa_cp_c_finalize): Update.
6114 * ipa-devirt.c (struct odr_type_warn_count): Update.
6115 (struct decl_warn_count): Update.
6116 (struct final_warning_record): Update.
6117 (possible_polymorphic_call_targets): Update.
6118 (ipa_devirt): Update.
6119 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6120 * ipa-icf.c (sem_function::merge): Update.
6121 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6122 * ipa-inline.c (compute_uninlined_call_time): Update.
6123 (compute_inlined_call_time): Update.
6124 (want_inline_small_function_p): Update.
6125 (want_inline_self_recursive_call_p): Update.
6126 (edge_badness): Update.
6127 (lookup_recursive_calls): Update.
6128 (recursive_inlining): Update.
6129 (inline_small_functions): Update.
6130 (dump_overall_stats): Update.
6131 (dump_inline_stats): Update.
6132 * ipa-profile.c (ipa_profile_generate_summary): Update.
6133 (ipa_propagate_frequency): Update.
6134 (ipa_profile): Update.
6135 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6136 * ipa-utils.c (ipa_merge_profiles): Update.
6137 * loop-doloop.c (doloop_modify): Update.
6138 * loop-unroll.c (report_unroll): Update.
6139 (unroll_loop_runtime_iterations): Update.
6140 * lto-cgraph.c (lto_output_edge): Update.
6141 (lto_output_node): Update.
6142 (input_node): Update.
6143 (input_edge): Update.
6144 (merge_profile_summaries): Update.
6145 * lto-streamer-in.c (input_cfg): Update.
6146 * lto-streamer-out.c (output_cfg): Update.
6147 * mcf.c (create_fixup_graph): Update.
6148 (adjust_cfg_counts): Update.
6149 (sum_edge_counts): Update.
6150 * modulo-sched.c (sms_schedule): Update.
6151 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6152 * predict.c (maybe_hot_count_p): Update.
6153 (probably_never_executed): Update.
6154 (dump_prediction): Update.
6155 (combine_predictions_for_bb): Update.
6156 (propagate_freq): Update.
6157 (handle_missing_profiles): Update.
6158 (counts_to_freqs): Update.
6159 (rebuild_frequencies): Update.
6160 (force_edge_cold): Update.
6161 * predict.h: Include profile-count.h
6162 (maybe_hot_count_p, counts_to_freqs): UPdate.
6163 * print-rtl-function.c: Do not include cfg.h
6164 * print-rtl.c: Include basic-block.h
6165 * profile-count.c: New file.
6166 * profile-count.h: New file.
6167 * profile.c (is_edge_inconsistent): Update.
6168 (correct_negative_edge_counts): Update.
6169 (is_inconsistent): Update.
6170 (set_bb_counts): Update.
6171 (read_profile_edge_counts): Update.
6172 (compute_frequency_overlap): Update.
6173 (compute_branch_probabilities): Update; Initialize and deinitialize
6175 (branch_prob): Update.
6176 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6177 (edge_gcov_count): New.
6178 (bb_gcov_count): New.
6179 * shrink-wrap.c (try_shrink_wrapping): Update.
6180 * tracer.c (better_p): Update.
6181 * trans-mem.c (expand_transaction): Update.
6182 (ipa_tm_insert_irr_call): Update.
6183 (ipa_tm_insert_gettmclone_call): Update.
6184 * tree-call-cdce.c: Update.
6185 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6186 (gimple_duplicate_sese_tail): Update.
6187 (gimple_account_profile_record): Update.
6188 (execute_fixup_cfg): Update.
6189 * tree-inline.c (copy_bb): Update.
6190 (copy_edges_for_bb): Update.
6191 (initialize_cfun): Update.
6192 (freqs_to_counts): Update.
6193 (copy_cfg_body): Update.
6194 (expand_call_inline): Update.
6195 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6196 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6197 (try_unroll_loop_completely): Update.
6198 (try_peel_loop): Update.
6199 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6200 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6201 * tree-ssa-loop-split.c (connect_loops): Update.
6202 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6203 * tree-ssa-reassoc.c (branch_fixup): Update.
6204 * tree-ssa-tail-merge.c (replace_block_by): Update.
6205 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6206 (compute_path_counts): Update.
6207 (update_profile): Update.
6208 (recompute_probabilities): Update.
6209 (update_joiner_offpath_counts): Update.
6210 (estimated_freqs_path): Update.
6211 (freqs_to_counts_path): Update.
6212 (clear_counts_path): Update.
6213 (ssa_fix_duplicate_block_edges): Update.
6214 (duplicate_thread_path): Update.
6215 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6216 (struct switch_conv_info): Update.
6217 * tree-tailcall.c (decrease_profile): Update.
6218 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6219 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6220 * value-prof.c (check_counter): Update.
6221 (gimple_divmod_fixed_value): Update.
6222 (gimple_mod_pow2): Update.
6223 (gimple_mod_subtract): Update.
6224 (gimple_ic_transform): Update.
6225 (gimple_stringop_fixed_value): Update.
6226 * value-prof.h (gimple_ic): Update.
6228 2017-06-02 Carl Love <cel@us.ibm.com>
6230 * config/rs6000/rs6000-c: Add support for built-in functions
6231 vector double vec_doublee (vector signed int);
6232 vector double vec_doublee (vector unsigned int);
6233 vector double vec_doublee (vector float);
6234 vector double vec_doubleh (vector signed int);
6235 vector double vec_doubleh (vector unsigned int);
6236 vector double vec_doubleh (vector float);
6237 vector double vec_doublel (vector signed int);
6238 vector double vec_doublel (vector unsigned int);
6239 vector double vec_doublel (vector float);
6240 vector double vec_doubleo (vector signed int);
6241 vector double vec_doubleo (vector unsigned int);
6242 vector double vec_doubleo (vector float);.
6243 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6244 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6246 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6247 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6248 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6250 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6251 vec_doublel, vec_doubleh.
6252 * doc/extend.texi: Update the built-in documentation file for the
6253 new built-in functions.
6255 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6258 * ipa-inline-analysis.c (free_growth_caches): Set
6259 edge_removal_hook_holder to NULL after removing it.
6261 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6263 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6264 comparision with zero.
6266 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6267 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6268 for early expansion of vec_min and vec_max builtins.
6269 (builtin_function_type): Add min/max unsigned variants to those
6270 identified as having unsigned arguments.
6272 2017-06-02 Olivier Hainque <hainque@adacore.com>
6274 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6276 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6278 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6279 Use VALL_F16 iterator rather than VALL.
6281 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6283 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6284 Emit CBNZ inside loop when doing a strong exchange and comparing
6285 against zero. Generate the CC flags after the loop.
6287 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6289 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6290 (dl_section_ref): New.
6291 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6292 On AIX, append an expression to subtract the size of the
6293 section length to dl_section_ref.
6295 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6297 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6298 for early expansion of vector absolute builtins.
6300 2017-06-02 Richard Biener <rguenther@suse.de>
6302 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6303 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6305 2017-06-02 Richard Biener <rguenther@suse.de>
6307 PR tree-optimization/80948
6308 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6311 2017-06-02 Richard Biener <rguenther@suse.de>
6313 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6315 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6316 for inductions as relevant.
6318 2017-06-02 Richard Biener <rguenther@suse.de>
6320 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6321 (vectorizable_induction): ... this. Remove dead code.
6323 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6325 * builtins. (expand_builtin_alloca): Remove second parameter and
6326 infer its value from the first parameter instead.
6327 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6329 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6331 PR rtl-optimization/80903
6332 * loop-doloop.c (add_test): Unshare sequence.
6334 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6336 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6338 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6340 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6342 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6343 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6344 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6345 (xlogue_layout::get_stub_rtx): Make static.
6346 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6347 (xlogue_layout::compute_stub_managed_regs): Rename to...
6348 (xlogue_layout::count_stub_managed_regs): ...this.
6349 (xlogue_layout::is_stub_managed_reg): New function.
6350 (xlogue_layout::m_stub_names): Rename to...
6351 (xlogue_layout::s_stub_names): ...this, make static.
6352 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6353 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6354 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6355 xlogue_layout::s_stub_names): Instantiate statics.
6356 (stub_managed_regs): Remove.
6357 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6358 (disable_call_ms2sysv_xlogues): Rename to...
6359 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6360 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6362 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6363 change after reload_completed.
6364 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6366 (ix86_expand_prologue): Likewise.
6367 (ix86_expand_epilogue): Likewise.
6368 (ix86_expand_split_stack_prologue): Likewise.
6369 (ix86_compute_frame_layout): Remove frame parameter ...
6370 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6371 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6373 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6374 (ix86_frame): Move from here ...
6375 * config/i386/i386.h (ix86_frame): ... to here.
6376 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6377 complete ix86_frame data structure instead. Remove some_ld_name.
6379 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6381 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6382 symbols that hold a DECL_VALUE_EXPR.
6384 2017-06-01 Martin Jambor <mjambor@suse.cz>
6386 PR tree-optimization/80898
6387 * tree-sra.c (process_subtree_disqualification): Removed.
6388 (disqualify_candidate): Do not acll
6389 process_subtree_disqualification.
6390 (subtree_mark_written_and_enqueue): New function.
6391 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6392 RHS has been disqualified and re-queue LHS if necessary. Apart
6393 from that, ignore disqualified RHS.
6395 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6397 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6398 address fetch for z10 or later.
6400 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6402 * config/arc/arc.md (tst_movb): Add guard when splitting.
6404 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6406 * config/arc/arc.c (arc_can_eliminate): Test against
6407 arc_frame_pointer_needed.
6409 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6411 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6412 to prevent store reordering.
6413 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6414 (type): Add block type.
6415 (stack_tie): Define special instruction to be used in
6418 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6420 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6421 constraint. It is not valid for the pattern.
6422 (noncommutative_binary_comparison): Likewise.
6424 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6426 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6429 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6431 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6432 be used by the reg-alloc.
6434 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6436 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6437 reg-alloc when having mul64 or mul32x16 instructions.
6438 (mulsidi3): Likewise.
6439 (umulsidi3): Likewise.
6440 (mulsi32x16): New pattern.
6441 (mulsi64): Likewise.
6442 (mulsidi64): Likewise.
6443 (umulsidi64): Likewise.
6444 (MUL32x16_REG): Define.
6445 (mul64_600): Use MUL32x16_REG.
6446 (mac64_600): Likewise.
6447 (umul64_600): Likewise.
6448 (umac64_600): Likewise.
6450 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6452 * config/arc/arc.md (mulsi3_700): Make it commutative.
6454 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6456 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6458 (*sign_extendsidi2_insn): Likewise for movstosw.
6460 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6462 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6463 the type of the input discriminant value. Convert the
6464 discriminant value of signedness vary.
6466 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6468 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6469 Add to -Wall section.
6471 2017-06-01 Richard Biener <rguenther@suse.de>
6474 * fold-const.c (fold_plusminus_mult_expr): If the factored
6475 factor may be zero use a wrapping type for the inner operation.
6476 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6477 and handle moved defs.
6478 (process_assignment): Properly guard the unary op case. Return a
6479 tri-state indicating that moving the stmt before the call may allow
6480 to continue. Pass through to_move.
6481 (find_tail_calls): Handle moving unrelated defs before
6484 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6487 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6488 splitter result in the canonical way.
6490 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6492 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6493 also for 32bit target. Update insn attributes.
6494 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6496 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6498 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6499 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6500 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6501 (_mm_maskz_min_ss): New intrinsics.
6503 2017-05-31 Martin Liska <mliska@suse.cz>
6505 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6506 Change comment style to one we normally use.
6507 (vectorizable_reduction): Likewise.
6508 (vectorizable_induction): Likewise.
6509 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6510 (vectorizable_call): Likewise.
6511 (vectorizable_simd_clone_call): Likewise.
6512 (vectorizable_conversion): Likewise.
6513 (vectorizable_assignment): Likewise.
6514 (vectorizable_shift): Likewise.
6515 (vectorizable_operation): Likewise.
6516 (vectorizable_store): Likewise.
6517 (vectorizable_load): Likewise.
6518 * tree-vectorizer.h: Likewise.
6520 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6522 * passes.c (emergency_dump_function): New.
6523 * tree-pass.h (emergency_dump_function): Declare.
6524 * plugin.c (plugins_internal_error_function): Remove.
6525 * plugin.h (plugins_internal_error_function): Remove declaration.
6526 * toplev.c (internal_error_function): New static function. Use it...
6527 (general_init): ...here.
6529 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6531 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6532 (arc_rtx_costs): Add costs for new patterns.
6533 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6534 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6536 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6538 * tree-ssa-strlen.c (get_next_strinfo): New function.
6539 (get_stridx_plus_constant): Use it.
6540 (zero_length_string): Likewise.
6541 (adjust_related_strinfos): Likewise.
6542 (adjust_last_stmt): Likewise.
6544 2017-05-31 Richard Biener <rguenther@suse.de>
6547 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6548 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6550 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6552 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6553 loop_vinfo argument and use of dependence distance vectors.
6554 Check instead whether the two references differ only in their
6555 initial value and assume that they have the same alignment if the
6556 difference is a multiple of the vector alignment.
6557 (vect_analyze_data_refs_alignment): Update call accordingly.
6559 2017-05-31 Martin Liska <mliska@suse.cz>
6562 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6564 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6566 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6567 (create_intersect_range_checks): Move from ...
6568 * tree-data-ref.c (create_intersect_range_checks_index)
6569 (create_intersect_range_checks): ... to here.
6570 (create_runtime_alias_checks): New function factored from ...
6571 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6572 here. Call above function.
6573 * tree-data-ref.h (create_runtime_alias_checks): New function.
6575 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6577 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6578 segment length for dr_b and compute it in wide_int.
6580 2017-05-31 Richard Biener <rguenther@suse.de>
6582 PR tree-optimization/80906
6583 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6584 and pass through iv_map.
6585 (copy_bb_and_scalar_dependences): Adjust.
6586 (translate_pending_phi_nodes): Likewise.
6587 (copy_loop_close_phi_args): Handle code-generating IVs instead
6590 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6592 * diagnostic-color.c (color_dict): Add "type-diff".
6593 (parse_gcc_colors): Update comment.
6594 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6595 -fdiagnostics-show-template-tree and -fno-elide-type.
6596 (GCC_COLORS): Add type-diff to example.
6598 (-fdiagnostics-show-template-tree): New.
6599 (-fno-elide-type): New.
6600 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6601 the pp_format_decoder callback. Call any m_format_postprocessor's
6603 (pretty_printer::pretty_printer): Initialize
6604 m_format_postprocessor.
6605 (pretty_printer::~pretty_printer): Delete any
6606 m_format_postprocessor.
6607 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6608 (class format_postprocessor): New class.
6609 (struct pretty_printer::format_decoder): Document the new parameters.
6610 (struct pretty_printer::m_format_postprocessor): New field.
6611 * tree-diagnostic.c (default_tree_printer): Update for new
6612 bool and const char ** params.
6613 * tree-diagnostic.h (default_tree_printer): Likewise.
6615 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6617 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6618 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6620 rs6000_gen_cell_microcode code.
6621 (rs6000_final_prescan_insn): Delete.
6622 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6623 "warn-cell-microcode" entries.
6624 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6625 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6626 throughout. Change cc_reg_not_micro_cr0_operand to
6627 cc_reg_not_cr0_operand throughout.
6628 (*extendhi<mode>2_noload): Delete.
6629 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6630 (mwarn-cell-microcode): Delete.
6631 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6632 -mgen-cell-microcode and -mwarn-cell-microcode.
6634 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6637 * config/i386/constraints.md (Yd): New constraint.
6639 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6640 and (?Yd, r) alternatives. Update insn attributes.
6641 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6642 and (?*Yd, r) alternatives. Update insn attributes.
6643 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6645 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6647 * gimplify.c (gimplify_modify_expr): Don't create a
6648 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6651 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6653 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6655 2017-05-30 Richard Biener <rguenther@suse.de>
6657 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6658 and reduc_def fields.
6659 (STMT_VINFO_REDUC_TYPE): New define.
6660 (STMT_VINFO_REDUC_DEF): Likewise.
6661 (vect_force_simple_reduction): Adjust prototype.
6662 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6663 (vect_is_simple_reduction): Remove check_reduction argument.
6664 (vect_force_simple_reduction): Adjust and set
6665 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6666 (vectorizable_reduction): Do not re-do reduction analysis
6667 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6668 * tree-parloops.c (gather_scalar_reductions): Adjust.
6670 2017-05-30 Richard Biener <rguenther@suse.de>
6673 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6676 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6678 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6679 Introduce unknown_misalignment parameter and remove vf.
6680 (vect_peeling_hash_get_lowest_cost):
6681 Pass unknown_misalignment parameter.
6682 (vect_enhance_data_refs_alignment):
6683 Fix unsupportable data ref treatment.
6685 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6687 * tree-vect-data-refs.c (vect_get_data_access_cost):
6688 Workaround for SLP handling.
6689 (vect_enhance_data_refs_alignment):
6690 Compute costs for doing no peeling at all, compare to the best
6691 peeling costs so far and avoid peeling if cheaper.
6693 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6695 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6696 Return peeling info and set costs to zero for unlimited cost
6698 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6699 with unknown misalignment. Compute and costs for unknown
6700 misalignment, compare them to the costs for known misalignment
6701 and choose the cheapest for peeling.
6703 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6705 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6706 (vect_get_peeling_costs_all_drs): Create function.
6707 (vect_peeling_hash_get_lowest_cost):
6708 Use vect_get_peeling_costs_all_drs.
6709 (vect_peeling_supportable): Create function.
6710 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6712 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6714 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6715 DR_HAS_NEGATIVE_STEP.
6716 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6717 (vect_enhance_data_refs_alignment): Use.
6718 (vect_duplicate_ssa_name_ptr_info): Use.
6719 * tree-vectorizer.h (dr_misalignment): Use.
6720 (known_alignment_for_access_p): Use.
6722 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6725 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6727 (has_section_name): New function.
6729 2017-05-30 Martin Liska <mliska@suse.cz>
6732 * auto-profile.c (get_function_decl_from_block): Fix
6735 2017-05-30 Richard Biener <rguenther@suse.de>
6738 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6740 2017-05-30 Martin Liska <mliska@suse.cz>
6742 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6743 * dumpfile.h (struct dump_file_info): Remove ctors.
6745 2017-05-30 Martin Liska <mliska@suse.cz>
6747 * predict.def: Fix GNU coding style.
6749 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6751 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6752 Mark 'to' argument with ATTRIBUTE_UNUSED.
6754 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6756 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6757 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6758 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6761 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
6763 * doc/install.texi (Options specification): Restore entry of
6764 --enable-sjlj-exceptions.
6766 2017-05-27 Michael Eager <eager@eagercon.com>
6769 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
6771 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
6773 * config/microblaze/microblaze.h
6774 (FIXED_REGISTERS): Update in macro.
6775 (CALL_USED_REGISTERS): Update in macro.
6777 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
6779 * doc/install.texi: Add links to macOS binary distributions.
6781 2017-05-27 Jakub Jelinek <jakub@redhat.com>
6785 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6787 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6789 2017-05-26 Martin Liska <mliska@suse.cz>
6791 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
6793 2017-05-26 Martin Liska <mliska@suse.cz>
6795 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
6796 always leading ';; '.
6797 (dump_bb_info): Likewise.
6798 (brief_dump_cfg): Likewise.
6799 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
6800 * dumpfile.c: Remove usage of TDF_VERBOSE.
6801 * dumpfile.h (enum dump_kind): Likewise.
6802 (dump_gimple_bb_header): Do not use TDF_COMMENT.
6803 * print-tree.c (debug_verbose): Remove.
6804 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
6805 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
6806 * tree-diagnostic.c (default_tree_printer): Replace
6807 TDF_DIAGNOSTIC with TDF_SLIM.
6809 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6811 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
6812 in parameter loop, rather than loop_vinfo.
6813 (create_intersect_range_checks): Ditto.
6814 (vect_create_cond_for_alias_checks): Update call to above functions.
6816 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6818 PR tree-optimization/80815
6819 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
6820 for merging runtime alias checks. Handle negative DR_STEPs.
6822 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6824 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
6826 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
6827 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
6828 out code pruning runtime alias checks.
6829 * tree-data-ref.c (prune_runtime_alias_test_list): New function
6830 factored out from above.
6831 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6833 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6835 (prune_runtime_alias_test_list): New decalaration.
6837 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6839 * tree-vect-data-refs.c (compare_tree): Rename and move ...
6840 * tree-data-ref.c (data_ref_compare_tree): ... to here.
6841 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
6842 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
6843 (operator==, comp_dr_with_seg_len_pair): Ditto.
6844 (vect_prune_runtime_alias_test_list): Ditto.
6846 2017-05-26 Martin Liska <mliska@suse.cz>
6849 * params.def: Bound partial-inlining-entry-probability param.
6851 2017-05-26 Marek Polacek <polacek@redhat.com>
6854 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
6857 2017-05-26 Richard Biener <rguenther@suse.de>
6859 PR tree-optimization/80842
6860 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
6863 2017-05-26 Richard Biener <rguenther@suse.de>
6865 PR tree-optimization/80844
6866 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
6868 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
6870 * doc/md.texi (Machine Constraints): Update x86 family
6871 machine constraints section to match 'config/i386/constraints.md'.
6873 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
6875 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
6877 2017-05-25 Nathan Sidwell <nathan@acm.org>
6879 * doc/invoke.texi (--enable-languages): Update documentation.
6881 2017-05-25 Martin Liska <mliska@suse.cz>
6883 * dumpfile.c: Add TDF_FOLDING.
6884 * dumpfile.h (enum dump_kind): Likewise.
6885 * genmatch.c (dt_simplify::gen_1): Use it.
6887 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6889 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
6891 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6893 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6894 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
6896 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6898 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
6899 * match.pd (X == C): Rewrite it here.
6900 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
6901 with_certain_nonzero_bits2): New predicates.
6902 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
6904 2017-05-24 Nathan Sidwell <nathan@acm.org>
6906 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
6909 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
6912 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6914 * config/powerpcspe: New port. Files are copied from the rs6000
6915 port, with "rs6000" in filenames replaced by "powerpcspe".
6917 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
6919 PR rtl-optimization/80754
6920 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
6922 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
6924 * config/sparc/sparc.md (length): Return the correct value for -mflat
6925 sibcalls to match output_sibcall.
6927 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6931 * config/rs6000/rs6000.c (struct machine_function): Add new field
6933 (rs6000_get_separate_components): Init that field, use it.
6934 (rs6000_components_for_bb): Use the field.
6936 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
6938 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
6940 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
6943 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
6945 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6948 * config/s390/s390.c (s390_check_qrst_address): Check incoming
6949 address against address_operand predicate.
6950 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
6952 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
6954 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
6955 parameters passed indirectly.
6957 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
6959 * config/i386/i386.md (*movdi_internal): Remove SSE4
6960 alternative 18 (?r, *v). Update insn attributes.
6961 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
6962 Update insn attributes.
6963 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
6964 Update insn attributes.
6965 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
6966 alternative 1 (r, v). Remove isa attribute.
6967 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
6968 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
6969 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
6971 2017-05-23 Tom de Vries <tom@codesourcery.com>
6973 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
6976 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6978 * cgraphunit.c (symbol_table::process_new_functions): Update.
6979 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
6980 (inline_generate_summary): Rename to ...
6981 (ipa_fn_summary_generate): ... this one.
6982 (inline_read_summary): Rename to ...
6983 (ipa_fn_summary_read): ... this one.
6984 (inline_write_summary): Rename to ...
6985 (ipa_fn_summary_write): ... this one.
6986 (inline_free_summary): Rename to ...
6987 (ipa_free_fn_summary): ... this one.
6988 (pass_data_local_fn_summary, pass_local_fn_summary,
6989 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
6990 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
6991 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
6992 make_pass_ipa_fn_summary): New.
6993 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
6994 inline_write_summary, inline_free_summary): Remove.
6995 (ipa_free_fn_summary) : New.
6996 * ipa-inline.c (ipa_inline): Update.
6997 (pass_ipa_inline): Do not generate summaries.
6998 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7000 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7001 and add pass_ipa_fn_summary.
7002 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7004 (make_pass_inline_parameters): Remove.
7006 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7008 * omp-low.c (struct omp_context): Remove "default_kind" member.
7011 * omp-offload.c (execute_oacc_device_lower): Remove the
7012 parallelism dimensions function attributes for unparallelized
7013 OpenACC kernels constructs.
7015 2017-05-23 Martin Liska <mliska@suse.cz>
7017 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7019 (cgraph_edge::make_speculative): Likewise.
7020 (cgraph_edge::resolve_speculation): Likewise.
7021 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7022 (cgraph_node::dump): Likewise.
7023 * cgraph.h: Likewise.
7024 * cgraphunit.c (analyze_functions): Likewise.
7025 (symbol_table::compile): Likewise.
7026 * ipa-cp.c (print_all_lattices): Likewise.
7027 (determine_versionability): Likewise.
7028 (initialize_node_lattices): Likewise.
7029 (ipcp_verify_propagated_values): Likewise.
7030 (estimate_local_effects): Likewise.
7031 (update_profiling_info): Likewise.
7032 (create_specialized_node): Likewise.
7033 (perhaps_add_new_callers): Likewise.
7034 (decide_about_value): Likewise.
7035 (decide_whether_version_node): Likewise.
7036 (identify_dead_nodes): Likewise.
7037 (ipcp_store_bits_results): Likewise.
7038 * ipa-devirt.c (dump_targets): Likewise.
7039 (ipa_devirt): Likewise.
7040 * ipa-icf.c (sem_item::dump): Likewise.
7041 (sem_function::equals): Likewise.
7042 (sem_variable::equals): Likewise.
7043 (sem_item_optimizer::read_section): Likewise.
7044 (sem_item_optimizer::execute): Likewise.
7045 (congruence_class::dump): Likewise.
7046 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7047 (dump_inline_summary): Likewise.
7048 (estimate_node_size_and_time): Likewise.
7049 (inline_analyze_function): Likewise.
7050 * ipa-inline-transform.c (inline_call): Likewise.
7051 * ipa-inline.c (report_inline_failed_reason): Likewise.
7052 (want_early_inline_function_p): Likewise.
7053 (edge_badness): Likewise.
7054 (update_edge_key): Likewise.
7055 (inline_small_functions): Likewise.
7056 * ipa-profile.c (ipa_profile): Likewise.
7057 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7058 (ipa_make_edge_direct_to_target): Likewise.
7059 (remove_described_reference): Likewise.
7060 (ipa_impossible_devirt_target): Likewise.
7061 (propagate_controlled_uses): Likewise.
7062 (ipa_print_node_params): Likewise.
7063 (ipcp_transform_function): Likewise.
7064 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7065 (propagate_pure_const): Likewise.
7066 * ipa-reference.c (generate_summary): Likewise.
7067 (read_write_all_from_decl): Likewise.
7068 (propagate): Likewise.
7069 (ipa_reference_read_optimization_summary): Likewise.
7070 * ipa-utils.c (ipa_merge_profiles): Likewise.
7071 * ipa.c (walk_polymorphic_call_targets): Likewise.
7072 (symbol_table::remove_unreachable_nodes): Likewise.
7073 (ipa_single_use): Likewise.
7074 * passes.c (execute_todo): Likewise.
7075 * predict.c (drop_profile): Likewise.
7076 * symtab.c (symtab_node::get_dump_name): New function.
7077 (symtab_node::dump_name): Likewise.
7078 (symtab_node::dump_asm_name): Likewise.
7079 (symtab_node::dump_references): Likewise.
7080 (symtab_node::dump_referring): Likewise.
7081 (symtab_node::dump_base): Likewise.
7082 (symtab_node::debug_symtab): Likewise.
7083 * tree-sra.c (convert_callers_for_node): Likewise.
7084 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7085 * value-prof.c (init_node_map): Likewise.
7087 2017-05-23 Martin Liska <mliska@suse.cz>
7089 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7090 and symtab_node::debug_symtab to symbol_table::debug.
7091 * cgraphunit.c (analyze_functions): Use the renamed function.
7092 (symbol_table::compile): Likewise.
7093 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7094 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7095 * passes.c (execute_todo): Likewise.
7096 * symtab.c (symbol_table::dump): New function.
7097 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7099 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7101 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7102 that nonconst implies exec.
7104 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7106 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7107 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7108 (inline_edge_summary_vec): Turn into ...
7109 (ipa_call_summaries): ... this one.
7110 (redirect_to_unreachable, edge_set_predicate,
7111 evaluate_properties_for_edge, inline_summary_alloc,
7112 reset_ipa_call_summary, reset_inline_summary,
7113 inline_summary_t::duplicate): Update.
7114 (inline_edge_duplication_hook): Turn to ...
7115 (ipa_call_summary_t::duplicate): ... this one.
7116 (inline_edge_removal_hook): Turn to ...
7117 (ipa_call_summary_t::remove): ... this one.
7118 (dump_inline_edge_summary): Turn to ...
7119 (dump_ipa_call_summary): ... this one.
7120 (estimate_function_body_sizes): Update.
7121 (inline_update_callee_summaries): Update.
7122 (remap_edge_change_prob): Update.
7123 (remap_edge_summaries): Update.
7124 (inline_merge_summary): Update.
7125 (do_estimate_edge_time): Update.
7126 (inline_generate_summary): Update.
7127 (inline_read_section): Update.
7128 (inline_read_summary): Update.
7129 (inline_free_summary): Update.
7130 * ipa-inline.c (can_inline_edge_p): Update.
7131 (compute_inlined_call_time): Update.
7132 (want_inline_small_function_p): Update.
7133 (edge_badness): Update.
7134 (early_inliner): Update.
7135 * ipa-inline.h (inline_edge_summary): Turn to ...
7136 (ipa_call_summary): ... this one.
7137 (ipa_call_summary_t): New class.
7138 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7139 (ipa_call_summaries): New.
7140 (inline_edge_summary): Remove.
7141 (estimate_edge_growth): Update.
7142 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7143 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7144 * ipa-split.c (execute_split_functions): Update.
7145 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7147 2017-05-23 Tom de Vries <tom@codesourcery.com>
7149 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7150 attributes): Document rdrand effective target.
7152 2017-05-23 Tom de Vries <tom@codesourcery.com>
7154 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7155 attributes): Sort alphabetically.
7157 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7159 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7161 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7164 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7165 V2DF/V2DI splat into two separate patterns, one that handles
7166 registers, and the other that only handles memory. Drop support
7167 for splatting from a GPR on ISA 2.07 and then splitting the
7168 splat into direct move and splat.
7169 (vsx_splat_<mode>_reg): Likewise.
7170 (vsx_splat_<mode>_mem): Likewise.
7172 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7174 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7176 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7179 * omp-low.c (finish_taskreg_remap): New function.
7180 (finish_taskreg_scan): If unit size of ctx->record_type
7181 is non-constant, unshare the size expression and replace
7182 decls in it with possible outer var refs.
7185 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7186 GOVD_SHARED rather than GOVD_PRIVATE with it.
7187 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7188 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7191 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7192 as last argument to build_outer_var_ref for pointer bases of array
7195 2017-05-19 Martin Sebor <msebor@redhat.com>
7197 * print-tree.c (print_node): Print DECL_READ_P flag.
7199 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7201 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7202 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7203 * cgraph.c: Likewise.
7204 * cgraphunit.c: Likewise.
7205 * gengtype.c: Likewise.
7206 * ipa-cp.c: Likewise.
7207 * ipa-devirt.c: Likewise.
7208 * ipa-icf.c: Likewise.
7209 * ipa-predicate.c: Likewise.
7210 * ipa-profile.c: Likewise.
7211 * ipa-prop.c: Likewise.
7212 * ipa-split.c: Likewise.
7214 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7215 edge_predicate_pool, dump_inline_hints,
7216 inline_summary::account_size_time, redirect_to_unreachable,
7217 edge_set_predicate, set_hint_predicate,
7218 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7219 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7220 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7221 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7222 ipa_call_summary_t::remove, initialize_growth_caches,
7223 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7224 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7225 mark_modified, unmodified_parm_1, unmodified_parm,
7226 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7227 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7228 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7229 will_be_nonconstant_predicate, record_modified_bb_info,
7230 get_minimal_bb, record_modified, param_change_prob,
7231 phi_result_unknown_predicate, predicate_for_phi_result,
7232 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7233 estimate_function_body_sizes, compute_inline_parameters,
7234 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7235 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7236 inline_update_callee_summaries, remap_edge_change_prob,
7237 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7238 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7239 inline_analyze_function, inline_summary_t::insert,
7240 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7241 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7242 inline_free_summary): Move to ipa-fnsummary.h
7243 (predicate_t): Remove.
7244 * ipa-fnsummary.c: New file.
7245 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7246 (enum inline_hints_vals, inline_hints, agg_position_info,
7247 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7248 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7249 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7250 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7251 inline_read_summary, inline_write_summary, inline_free_summary,
7252 inline_analyze_function, initialize_inline_failed,
7253 inline_merge_summary, inline_update_overall_summary,
7254 compute_inline_parameters): Move to ipa-fnsummary.h
7255 * ipa-fnsummary.h: New file.
7256 * ipa-inline-transform.h: Include ipa-inline.h.
7257 * ipa-inline.c: LIkewise.
7259 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7261 * ipa-inline.c (edge_badness): Use inlined_time instead of
7262 inline_summaries->get.
7264 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7266 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7268 2017-05-22 Nathan Sidwell <nathan@acm.org>
7270 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7271 (fdump-lang): Document 'raw' option.
7272 * dumpfile.h (TDI_tu): Delete.
7273 * dumpfile.c (dump_files): Remove translation-unit.
7274 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7276 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7278 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7279 command option from $(AWK) call.
7280 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7281 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7282 [FORMAT]: Remove handling of variable.
7283 * config/avr/t-multilib: Regenerate.
7285 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7287 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7289 (dump_inline_summary): Do not print self_time.
7290 (estimate_function_body_sizes): Do not set self_time.
7291 (compute_inline_parameters): Likewise.
7292 (inline_read_section, inline_write_summary): Do not stream self_time.
7293 * ipa-inline.h (inline_summary): Drop self_time.
7295 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7297 * ipa-inline-analysis.c (account_size_time): Rename to ...
7298 (inline_summary::account_size_time): ... this one.
7299 (reset_ipa_call_summary): Turn to ...
7300 (ipa_call_summary::reset): ... this one.
7301 (reset_inline_summary): Turn to ...
7302 (inline_summary::reset): ... this one.
7303 (inline_summary_t::remove): Update.
7304 (inline_summary_t::duplicate): Update.
7305 (ipa_call_summary_t::remove): Update.
7306 (dump_inline_summary): Update.
7307 (estimate_function_body_sizes): Update.
7308 (compute_inline_parameters): Update.
7309 (estimate_node_size_and_time): Update.
7310 (inline_merge_summary): Update.
7311 (inline_update_overall_summary): Update.
7312 (inline_read_section): Update.
7313 (inline_write_summary): Update.
7314 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7315 add account_size_time and reset member functions.
7316 (ipa_call_summary): Add reset function.
7317 * ipa-predicate.h (predicate::operator &): Constify.
7319 2017-05-22 Richard Biener <rguenther@suse.de>
7321 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7323 2017-05-19 Jason Merrill <jason@redhat.com>
7325 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7327 2017-05-19 Marek Polacek <polacek@redhat.com>
7330 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7331 TYPE_OVERFLOW_WRAPS checks.
7333 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7335 * tree-core.h (enum omp_clause_default_kind): Add
7336 "OMP_CLAUSE_DEFAULT_PRESENT".
7337 * tree-pretty-print.c (dump_omp_clause): Handle it.
7338 * gimplify.c (enum gimplify_omp_var_data): Add
7339 "GOVD_MAP_FORCE_PRESENT".
7340 (gimplify_adjust_omp_clauses_1): Map it to
7341 "GOMP_MAP_FORCE_PRESENT".
7342 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7344 * gimplify.c (oacc_default_clause): Clarify.
7346 2017-05-19 Nathan Sidwell <nathan@acm.org>
7348 LANG_HOOK_REGISTER_DUMPS
7349 * toplev.c (general_init): Call register dump lang hook.
7350 * doc/invoke.texi: Document -fdump-lang option family.
7351 * dumpfile.c (dump_files): Remove class dump here.
7352 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7353 * dumpfile.h (tree_dump_index): Remove TDI_class.
7354 * langhooks-def.h (lhd_register_dumps): Declare.
7355 (LANG_HOOKS_REGISTER_DUMPS): Define.
7356 (LANG_HOOKS_INITIALIZER): Add it.
7357 * langhooks.c (lhd_register_dumps): Define.
7358 * langhooks.h (struct lang_hooks): Add register_dumps.
7360 2017-05-19 Nathan Sidwell <nathan@acm.org>
7362 * context.h (context::set_passes): New.
7363 * context.c (context::context): Do not create pass manager.
7364 * toplev.c (general_init): Create pass manager here.
7366 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7368 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7369 use this splitter if two add or or instructions would also work for
7370 the constant we want to generate.
7372 2017-05-19 Richard Biener <rguenther@suse.de>
7375 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7376 predicate evaluation.
7378 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7380 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7382 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7385 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7387 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7388 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7389 (inline_edge_summary_vec): Turn into ...
7390 (ipa_call_summaries): ... this one.
7391 (redirect_to_unreachable, edge_set_predicate,
7392 evaluate_properties_for_edge, inline_summary_alloc,
7393 reset_ipa_call_summary, reset_inline_summary,
7394 inline_summary_t::duplicate): Update.
7395 (inline_edge_duplication_hook): Turn to ...
7396 (ipa_call_summary_t::duplicate): ... this one.
7397 (inline_edge_removal_hook): Turn to ...
7398 (ipa_call_summary_t::remove): ... this one.
7399 (dump_inline_edge_summary): Turn to ...
7400 (dump_ipa_call_summary): ... this one.
7401 (estimate_function_body_sizes): Update.
7402 (inline_update_callee_summaries): Update.
7403 (remap_edge_change_prob): Update.
7404 (remap_edge_summaries): Update.
7405 (inline_merge_summary): Update.
7406 (do_estimate_edge_time): Update.
7407 (inline_generate_summary): Update.
7408 (inline_read_section): Update.
7409 (inline_read_summary): Update.
7410 (inline_free_summary): Update.
7411 * ipa-inline.c (can_inline_edge_p): Update.
7412 (compute_inlined_call_time): Update.
7413 (want_inline_small_function_p): Update.
7414 (edge_badness): Update.
7415 (early_inliner): Update.
7416 * ipa-inline.h (inline_edge_summary): Turn to ...
7417 (ipa_call_summary): ... this one.
7418 (ipa_call_summary_t): New class.
7419 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7420 (ipa_call_summaries): New.
7421 (inline_edge_summary): Remove.
7422 (estimate_edge_growth): Update.
7423 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7424 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7425 * ipa-split.c (execute_split_functions): Update.
7426 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7428 2017-05-19 Richard Biener <rguenther@suse.de>
7431 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7433 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7435 * config/rs6000/rs6000.c (struct machine_function): Add field
7436 fpr_is_wrapped_separately.
7437 (rs6000_get_separate_components): Use 64 components. Handle the
7439 (rs6000_components_for_bb): Handle the FPR components.
7440 (rs6000_emit_prologue_components): Handle the FPR components.
7441 (rs6000_emit_epilogue_components): Handle the FPR components.
7442 (rs6000_set_handled_components): Handle the FPR components.
7443 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7444 that are already separately shrink-wrapped.
7445 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7446 that are already separately shrink-wrapped.
7448 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7451 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7454 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7455 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7456 cases where the register allocator uses a move and an offsettable
7457 memory operation to/from a FPR register on ISA 2.06/2.07.
7458 (define_peephole2 for Altivec d-form store): Likewise.
7460 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7463 * config/i386/mmx.md (*mov<mode>_internal): Enable
7464 alternatives 11, 12, 13 and 14 also for 32bit targets.
7465 Remove alternatives 15, 16, 17 and 18.
7466 * config/i386/sse.md (vec_concatv2di): Change
7467 alternative (!x, *y) to (x, ?!*Yn).
7469 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7471 * dumpfile.h (enum dump_kind): Remove stray comma.
7473 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7475 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7476 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7477 predicate::num_conditions
7478 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7479 (CHANGED): turn into predicate::changed.
7480 (agg_position_info): Move to ipa-predicate.h
7481 (add_condition, predicate::add_clause, predicate::operator &=,
7482 predicate::or_with, predicate::evaluate, predicate::probability,
7483 dump_condition, dump_clause, predicate::dump,
7484 predicate::remap_after_duplication, predicate::remap_after_inlining,
7485 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7486 (evaluate_conditions_for_known_args): Update.
7487 (set_cond_stmt_execution_predicate): Update.
7488 * ipa-inline.h: Include ipa-predicate.h
7489 (condition, inline_param_summary, conditions, agg_position_info,
7490 predicate): Move to ipa-predicate.h
7491 * ipa-predicate.c: New file.
7492 * ipa-predicate.h: New file.
7494 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7496 * final.c (leaf_function_p): Check we are not in a sequence.
7498 2017-05-18 Martin Liska <mliska@suse.cz>
7500 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7501 * dumpfile.c (dump_register): Use new enum dump_kind.
7502 (get_dump_file_name): Likewise.
7503 (dump_enable_all): Likewise.
7504 (dump_switch_p_1): Likewise.
7505 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7506 * dumpfile.h (enum dump_kind): New enum type.
7507 (struct dump_file_info): Create constructor and
7508 format fields and comments.
7509 * passes.c (pass_manager::register_one_dump_file):
7511 * statistics.c (statistics_early_init): Likewise.
7512 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7513 TDF_TREE with TDF_SLIM.
7514 (gather_memory_references_ref): Likewise.
7516 2017-05-18 Martin Liska <mliska@suse.cz>
7518 * vec.h (struct vnull): Use it.
7520 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7522 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7523 (true_predicate, false_predicate, true_predicate_p,
7524 false_predicate_p): Remove.
7525 (single_cond_predicate, not_inlined_predicate): Turn to member function
7527 (add_condition): Update.
7528 (add_clause): Turn to...
7529 (predicate::add_clause): ... this one; update; allow passing NULL
7531 (and_predicates): Turn to ...
7532 (predicate::operator &=): ... this one.
7533 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7534 (or_predicates): Turn to ...
7535 (predicate::or_with): ... this one.
7536 (evaluate_predicate): Turn to ...
7537 (predicate::evaluate): ... this one.
7538 (predicate_probability): Turn to ...
7539 (predicate::probability): ... this one.
7540 (dump_condition): Update.
7541 (dump_predicate): Turn to ...
7542 (predicate::dump): ... this one.
7543 (account_size_time): Update.
7544 (edge_set_predicate): Update.
7545 (set_hint_predicate): UPdate.
7546 (evaluate_conditions_for_known_args): Update.
7547 (evaluate_properties_for_edge): Update.
7548 (remap_predicate_after_duplication): Turn to...
7549 (predicate::remap_after_duplication): ... this one.
7550 (remap_hint_predicate_after_duplication): Update.
7551 (inline_summary_t::duplicate): UPdate.
7552 (dump_inline_edge_summary): Update.
7553 (dump_inline_summary): Update.
7554 (set_cond_stmt_execution_predicate): Update.
7555 (set_switch_stmt_execution_predicate): Update.
7556 (compute_bb_predicates): Update.
7557 (will_be_nonconstant_expr_predicate): Update.
7558 (will_be_nonconstant_predicate): Update.
7559 (phi_result_unknown_predicate): Update.
7560 (predicate_for_phi_result): Update.
7561 (array_index_predicate): Update.
7562 (estimate_function_body_sizes): Update.
7563 (estimate_node_size_and_time): Update.
7564 (estimate_ipcp_clone_size_and_time): Update.
7565 (remap_predicate): Rename to ...
7566 (predicate::remap_after_inlining): ... this one.
7567 (remap_hint_predicate): Update.
7568 (inline_merge_summary): Update.
7569 (inline_update_overall_summary): Update.
7570 (estimate_size_after_inlining): Update.
7571 (read_predicate): Rename to ...
7572 (predicate::stream_in): ... this one.
7573 (read_inline_edge_summary): Update.
7574 (write_predicate): Rename to ...
7575 (predicate::stream_out): ... this one.
7576 (write_inline_edge_summary): Update.
7577 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7578 (clause_t): Turn to uint32_t
7579 (predicate): Turn to class; implement constructor and operators
7581 (size_time_entry): Update.
7582 (inline_summary): Update.
7583 (inline_edge_summary): Update.
7585 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7587 * fold-const.c (fold_binary_loc): Move transformation...
7588 * match.pd (C - X CMP X): ... here.
7590 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7592 * config/sparc/sparc.c (sparc_option_override): Set function
7593 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7594 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7596 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7598 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7600 2017-05-18 Marek Polacek <polacek@redhat.com>
7603 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7604 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7605 gimple_assign_load_p.
7607 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7610 * real.c (do_compare): Give decimal_do_compare preference over
7611 comparing just the signs.
7613 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7615 * doc/md.texi (Canonicalization of Instructions): Describe the
7616 canonical form of instructions that inherently set a condition
7619 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7622 * tree-cfg.c: Move deletion of unreachable case statements to after
7623 the merging of consecutive case labels.
7625 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7627 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7628 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7629 restoring of callee-saved registers.
7631 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7633 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7634 * config/visium/visium.c (single_set_and_flags): Likewise.
7635 * config/visium/visium.md (Substitutions): Likewise.
7637 2017-05-17 Martin Liska <mliska@suse.cz>
7639 * cfg.c: Introduce dump_flags_t type and
7640 use it instead of int type.
7642 * cfghooks.c: Likewise.
7643 * cfghooks.h (struct cfg_hooks): Likewise.
7644 * cfgrtl.c: Likewise.
7645 * cfgrtl.h: Likewise.
7646 * cgraph.c (cgraph_node::get_body): Likewise.
7647 * coretypes.h: Likewise.
7648 * domwalk.c: Likewise.
7649 * domwalk.h: Likewise.
7650 * dumpfile.c (struct dump_option_value_info): Likewise.
7651 (dump_enable_all): Likewise.
7652 (dump_switch_p_1): Likewise.
7653 (opt_info_switch_p): Likewise.
7654 * dumpfile.h (enum tree_dump_index): Likewise.
7655 (struct dump_file_info): Likewise.
7656 * genemit.c: Likewise.
7657 * generic-match-head.c: Likewise.
7658 * gengtype.c (open_base_files): Likewise.
7659 * gimple-pretty-print.c: Likewise.
7660 * gimple-pretty-print.h: Likewise.
7661 * graph.c (print_graph_cfg): Likewise.
7662 * graphite-scop-detection.c (dot_all_sese): Likewise.
7663 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7664 * loop-unroll.c (report_unroll): Likewise.
7665 * passes.c (pass_manager::register_one_dump_file): Likewise.
7666 * print-tree.c: Likewise.
7667 * statistics.c: Likewise.
7668 * tree-cfg.c: Likewise.
7669 * tree-cfg.h: Likewise.
7670 * tree-dfa.c: Likewise.
7671 * tree-dfa.h: Likewise.
7672 * tree-dump.c (dump_function): Likewise.
7673 * tree-dump.h (struct dump_info): Likewise.
7674 * tree-pretty-print.c: Likewise.
7675 * tree-pretty-print.h: Likewise.
7676 * tree-ssa-live.c: Likewise.
7677 * tree-ssa-live.h: Likewise.
7678 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7679 * tree-vect-loop.c: Likewise.
7680 * tree-vect-slp.c: Likewise.
7682 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7683 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7685 PR tree-optimization/80457
7686 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7687 of all arguments to a statement as scalar_to_vec operations.
7688 (vectorizable_call): Adjust call to vect_model_simple_cost for
7690 (vectorizable_conversion): Likewise.
7691 (vectorizable_assignment): Likewise.
7692 (vectorizable_shift): Likewise.
7693 (vectorizable_operation): Likewise.
7694 (vectorizable_comparison): Likewise.
7695 (vect_is_simple_cond): Record the def types for operands.
7696 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7697 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7698 for statement argument count.
7700 2017-05-16 Carl Love <cel@us.ibm.com>
7702 * config/rs6000/rs6000-c: Add support for built-in functions
7703 vector unsigned long long vec_bperm (vector unsigned long long,
7704 vector unsigned char)
7705 vector signed long long vec_mule (vector signed int,
7707 vector unsigned long long vec_mule (vector unsigned int,
7708 vector unsigned int)
7709 vector signed long long vec_mulo (vector signed int,
7711 vector unsigned long long vec_mulo (vector unsigned int,
7712 vector unsigned int)
7713 vector signed char vec_sldw (vector signed char,
7716 vector unsigned char vec_sldw (vector unsigned char,
7717 vector unsigned char,
7719 vector signed short vec_sldw (vector signed short,
7720 vector signed short,
7722 vector unsigned short vec_sldw (vector unsigned short,
7723 vector unsigned short,
7725 vector signed int vec_sldw (vector signed int,
7728 vector unsigned int vec_sldw (vector unsigned int,
7729 vector unsigned int,
7731 vector signed long long vec_sldw (vector signed long long,
7732 vector signed long long,
7734 vector unsigned long long vec_sldw (vector unsigned long long,
7735 vector unsigned long long,
7737 * config/rs6000/rs6000-c: Add support for built-in functions
7738 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7739 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7740 * doc/extend.texi: Update the built-in documentation for the
7741 new built-in functions.
7743 2017-05-16 Marek Polacek <polacek@redhat.com>
7747 * tree.c (save_expr): Don't fold the expression.
7749 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7751 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7752 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7753 and (?*y,m). Update insn attributes.
7755 2017-05-16 Martin Liska <mliska@suse.cz>
7757 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7758 flags argument of print_gimple_stmt, print_gimple_expr,
7759 print_generic_stmt and print_generic_expr.
7760 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7761 * coretypes.h: Likewise.
7762 * except.c (dump_eh_tree): Likewise.
7763 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7764 * gimple-pretty-print.h: Likewise.
7765 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
7766 (backprop::push_to_worklist): Likewise.
7767 (backprop::pop_from_worklist): Likewise.
7768 (backprop::process_use): Likewise.
7769 (backprop::intersect_uses): Likewise.
7770 (note_replacement): Likewise.
7771 * gimple-ssa-store-merging.c
7772 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
7773 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7774 (pass_store_merging::execute): Likewise.
7775 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
7776 (ssa_base_cand_dump_callback): Likewise.
7777 (dump_incr_vec): Likewise.
7778 (replace_refs): Likewise.
7779 (replace_mult_candidate): Likewise.
7780 (create_add_on_incoming_edge): Likewise.
7781 (create_phi_basis): Likewise.
7782 (insert_initializers): Likewise.
7783 (all_phi_incrs_profitable): Likewise.
7784 (introduce_cast_before_cand): Likewise.
7785 (replace_one_candidate): Likewise.
7786 * gimplify.c (gimplify_expr): Likewise.
7787 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
7788 (set_rename): Likewise.
7789 (rename_uses): Likewise.
7790 (copy_loop_phi_nodes): Likewise.
7791 (add_close_phis_to_merge_points): Likewise.
7792 (copy_loop_close_phi_args): Likewise.
7793 (copy_cond_phi_args): Likewise.
7794 (graphite_copy_stmts_from_block): Likewise.
7795 (translate_pending_phi_nodes): Likewise.
7796 * graphite-poly.c (print_pdr): Likewise.
7797 (dump_gbb_cases): Likewise.
7798 (dump_gbb_conditions): Likewise.
7799 (print_scop_params): Likewise.
7800 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
7801 (build_cross_bb_scalars_use): Likewise.
7802 (gather_bbs::before_dom_children): Likewise.
7803 * hsa-dump.c (dump_hsa_immed): Likewise.
7804 * ipa-cp.c (print_ipcp_constant_value): Likewise.
7805 (get_replacement_map): Likewise.
7806 * ipa-inline-analysis.c (dump_condition): Likewise.
7807 (estimate_function_body_sizes): Likewise.
7808 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
7809 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
7810 * ipa-prop.c (ipa_dump_param): Likewise.
7811 (ipa_print_node_jump_functions_for_edge): Likewise.
7812 (ipa_modify_call_arguments): Likewise.
7813 (ipa_modify_expr): Likewise.
7814 (ipa_dump_param_adjustments): Likewise.
7815 (ipa_dump_agg_replacement_values): Likewise.
7816 (ipcp_modif_dom_walker::before_dom_children): Likewise.
7817 * ipa-pure-const.c (check_stmt): Likewise.
7818 (pass_nothrow::execute): Likewise.
7819 * ipa-split.c (execute_split_functions): Likewise.
7820 * omp-offload.c (dump_oacc_loop_part): Likewise.
7821 (dump_oacc_loop): Likewise.
7822 * trans-mem.c (tm_log_emit): Likewise.
7823 (tm_memopt_accumulate_memops): Likewise.
7824 (dump_tm_memopt_set): Likewise.
7825 (dump_tm_memopt_transform): Likewise.
7826 * tree-cfg.c (gimple_verify_flow_info): Likewise.
7827 (print_loop): Likewise.
7828 * tree-chkp-opt.c (chkp_print_addr): Likewise.
7829 (chkp_gather_checks_info): Likewise.
7830 (chkp_get_check_result): Likewise.
7831 (chkp_remove_check_if_pass): Likewise.
7832 (chkp_use_outer_bounds_if_possible): Likewise.
7833 (chkp_reduce_bounds_lifetime): Likewise.
7834 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
7835 (chkp_mark_completed_bounds): Likewise.
7836 (chkp_register_incomplete_bounds): Likewise.
7837 (chkp_mark_invalid_bounds): Likewise.
7838 (chkp_maybe_copy_and_register_bounds): Likewise.
7839 (chkp_build_returned_bound): Likewise.
7840 (chkp_get_bound_for_parm): Likewise.
7841 (chkp_build_bndldx): Likewise.
7842 (chkp_get_bounds_by_definition): Likewise.
7843 (chkp_generate_extern_var_bounds): Likewise.
7844 (chkp_get_bounds_for_decl_addr): Likewise.
7845 * tree-chrec.c (chrec_apply): Likewise.
7846 * tree-data-ref.c (dump_data_reference): Likewise.
7847 (dump_subscript): Likewise.
7848 (dump_data_dependence_relation): Likewise.
7849 (analyze_overlapping_iterations): Likewise.
7850 * tree-inline.c (expand_call_inline): Likewise.
7851 (tree_function_versioning): Likewise.
7852 * tree-into-ssa.c (dump_defs_stack): Likewise.
7853 (dump_currdefs): Likewise.
7854 (dump_names_replaced_by): Likewise.
7855 (dump_update_ssa): Likewise.
7856 (update_ssa): Likewise.
7857 * tree-object-size.c (pass_object_sizes::execute): Likewise.
7858 * tree-parloops.c (build_new_reduction): Likewise.
7859 (try_create_reduction_list): Likewise.
7860 (ref_conflicts_with_region): Likewise.
7861 (oacc_entry_exit_ok_1): Likewise.
7862 (oacc_entry_exit_single_gang): Likewise.
7863 * tree-pretty-print.h: Likewise.
7864 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
7865 (get_scalar_evolution): Likewise.
7866 (add_to_evolution): Likewise.
7867 (get_loop_exit_condition): Likewise.
7868 (analyze_evolution_in_loop): Likewise.
7869 (analyze_initial_condition): Likewise.
7870 (analyze_scalar_evolution): Likewise.
7871 (instantiate_scev): Likewise.
7872 (number_of_latch_executions): Likewise.
7873 (gather_chrec_stats): Likewise.
7874 (final_value_replacement_loop): Likewise.
7875 (scev_const_prop): Likewise.
7876 * tree-sra.c (dump_access): Likewise.
7877 (disqualify_candidate): Likewise.
7878 (create_access): Likewise.
7880 (maybe_add_sra_candidate): Likewise.
7881 (create_access_replacement): Likewise.
7882 (analyze_access_subtree): Likewise.
7883 (analyze_all_variable_accesses): Likewise.
7884 (sra_modify_assign): Likewise.
7885 (initialize_constant_pool_replacements): Likewise.
7886 (find_param_candidates): Likewise.
7887 (decide_one_param_reduction): Likewise.
7888 (replace_removed_params_ssa_names): Likewise.
7889 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
7890 * tree-ssa-copy.c (dump_copy_of): Likewise.
7891 (copy_prop_visit_cond_stmt): Likewise.
7892 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
7893 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
7894 (record_equivalences_from_stmt): Likewise.
7895 * tree-ssa-dse.c (compute_trims): Likewise.
7896 (delete_dead_call): Likewise.
7897 (delete_dead_assignment): Likewise.
7898 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
7899 (forward_propagate_into_cond): Likewise.
7900 (pass_forwprop::execute): Likewise.
7901 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7902 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
7904 (move_computations_worker): Likewise.
7905 (execute_sm): Likewise.
7906 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
7907 (remove_exits_and_undefined_stmts): Likewise.
7908 (remove_redundant_iv_tests): Likewise.
7909 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
7910 (adjust_iv_update_pos): Likewise.
7911 * tree-ssa-math-opts.c (bswap_replace): Likewise.
7912 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
7913 (value_replacement): Likewise.
7914 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
7915 * tree-ssa-pre.c (print_pre_expr): Likewise.
7916 (get_representative_for): Likewise.
7917 (create_expression_by_pieces): Likewise.
7918 (insert_into_preds_of_block): Likewise.
7919 (eliminate_insert): Likewise.
7920 (eliminate_dom_walker::before_dom_children): Likewise.
7921 (eliminate): Likewise.
7922 (remove_dead_inserted_code): Likewise.
7923 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
7924 * tree-ssa-reassoc.c (get_rank): Likewise.
7925 (eliminate_duplicate_pair): Likewise.
7926 (eliminate_plus_minus_pair): Likewise.
7927 (eliminate_not_pairs): Likewise.
7928 (undistribute_ops_list): Likewise.
7929 (eliminate_redundant_comparison): Likewise.
7930 (update_range_test): Likewise.
7931 (optimize_range_tests_var_bound): Likewise.
7932 (optimize_vec_cond_expr): Likewise.
7933 (rewrite_expr_tree): Likewise.
7934 (rewrite_expr_tree_parallel): Likewise.
7935 (linearize_expr): Likewise.
7936 (break_up_subtract): Likewise.
7937 (linearize_expr_tree): Likewise.
7938 (attempt_builtin_powi): Likewise.
7939 (attempt_builtin_copysign): Likewise.
7940 (transform_stmt_to_copy): Likewise.
7941 (transform_stmt_to_multiply): Likewise.
7942 (dump_ops_vector): Likewise.
7943 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
7944 (print_scc): Likewise.
7945 (set_ssa_val_to): Likewise.
7946 (visit_reference_op_store): Likewise.
7947 (visit_use): Likewise.
7948 (sccvn_dom_walker::before_dom_children): Likewise.
7949 (run_scc_vn): Likewise.
7950 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7952 (expr_hash_elt::print): Likewise.
7953 (const_and_copies::pop_to_marker): Likewise.
7954 (const_and_copies::record_const_or_copy_raw): Likewise.
7955 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
7956 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
7957 (dump_predicates): Likewise.
7958 (find_uninit_use): Likewise.
7959 (warn_uninitialized_phi): Likewise.
7960 (pass_late_warn_uninitialized::execute): Likewise.
7961 * tree-ssa.c (verify_vssa): Likewise.
7962 (verify_ssa): Likewise.
7963 (maybe_optimize_var): Likewise.
7964 * tree-vrp.c (dump_value_range): Likewise.
7965 (dump_all_value_ranges): Likewise.
7966 (dump_asserts_for): Likewise.
7967 (register_edge_assert_for_2): Likewise.
7968 (vrp_visit_cond_stmt): Likewise.
7969 (vrp_visit_switch_stmt): Likewise.
7970 (vrp_visit_stmt): Likewise.
7971 (vrp_visit_phi_node): Likewise.
7972 (simplify_cond_using_ranges_1): Likewise.
7973 (fold_predicate_in): Likewise.
7974 (evrp_dom_walker::before_dom_children): Likewise.
7975 (evrp_dom_walker::push_value_range): Likewise.
7976 (evrp_dom_walker::pop_value_range): Likewise.
7977 (execute_early_vrp): Likewise.
7979 2017-05-16 Richard Biener <rguenther@suse.de>
7981 * dwarf2out.c (loc_list_from_tree_1): Do not create
7982 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
7984 2017-05-16 Richard Biener <rguenther@suse.de>
7986 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
7988 (note_variable_value_in_expr): If we resolved the decl ref
7989 do not push to the stack.
7991 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
7993 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
7994 operations in fast-math mode.
7995 (vaddq_f16): Likewise.
7996 (vmul_f16): Likewise.
7997 (vmulq_f16): Likewise.
7998 (vsub_f16): Likewise.
7999 (vsubq_f16): Likewise.
8000 * config/arm/neon.md (add<mode>3): New.
8002 (fma:<VH:mode>3): New. Also remove outdated comment.
8005 2017-05-16 Martin Liska <mliska@suse.cz>
8009 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8010 (odr_types_equivalent_p): Likewise.
8012 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8014 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8016 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8019 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8020 non-interunit SSE move alternatives with '?'.
8021 (zero-extendsidi peephole2): New peephole to skip intermediate
8022 general register in SSE zero-extend sequence.
8024 2017-05-15 Jeff Law <law@redhat.com>
8026 * reorg.c (relax_delay_slots): Create a new variable to hold
8027 the temporary target rather than clobbering TARGET_LABEL.
8029 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8030 missing argument to extract_bit_field call.
8031 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8033 2017-05-15 Martin Liska <mliska@suse.cz>
8036 * gcc.c (process_command): Do not allow empty argument of -o option.
8038 2017-05-15 Renlin Li <renlin.li@arm.com>
8040 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8041 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8042 * config/aarch64/constraints.md (Usf): Add long call check.
8043 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8044 (call_value): Likewise.
8045 (sibcall): Likewise.
8046 (sibcall_value): Likewise.
8048 (call_value_insn): New.
8049 (sibcall_insn): Update rtx pattern.
8050 (sibcall_value_insn): Likewise.
8051 (call_internal): Remove.
8052 (call_value_internal): Likewise.
8053 (sibcall_internal): Likewise.
8054 (sibcall_value_internal): Likewise.
8055 (call_reg): Likewise.
8056 (call_symbol): Likewise.
8057 (call_value_reg): Likewise.
8058 (call_value_symbol): Likewise.
8060 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8063 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8065 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8067 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8068 compatible with CCGOCmode and with CCZmode.
8070 2017-05-14 Martin Sebor <msebor@redhat.com>
8073 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8074 (gimple_fold_builtin_snprintf): Same.
8075 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8076 (gimple_fold_builtin_snprintf): Same.
8077 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8079 (is_call_safe): New function.
8080 (try_substitute_return_value): Call it.
8081 (try_simplify_call): New function.
8082 (pass_sprintf_length::handle_gimple_call): Call it.
8084 2017-05-14 Martin Sebor <msebor@redhat.com>
8087 * builtins.c (expand_builtin_stpncpy): Simplify.
8089 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8091 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8092 * config/i386/i386.h
8093 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8094 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8095 (struct machine_function): Add new members call_ms2sysv,
8096 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8097 (struct machine_frame_state): New fields sp_realigned and
8098 sp_realigned_offset.
8099 * config/i386/i386.c
8100 (enum xlogue_stub): New enum.
8101 (enum xlogue_stub_sets): New enum.
8102 (class xlogue_layout): New class.
8103 (struct ix86_frame): New fields stack_realign_allocate_offset,
8104 stack_realign_offset and outlined_save_offset. Modify comments to
8105 detail stack layout when using out-of-line stubs.
8106 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8107 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8108 -mcall-ms2sysv-xlogues.
8109 (stub_managed_regs): New static variable.
8110 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8111 registers managed by out-of-line stub.
8112 (disable_call_ms2sysv_xlogues): New function.
8113 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8114 m->call_ms2sysv when appropriate and compute frame layout for
8116 (sp_valid_at, fp_valid_at): New inline functions.
8117 (choose_basereg): New function.
8118 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8120 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8121 Use align parameter of choose_baseaddr to generated aligned SSE movs
8123 (pro_epilogue_adjust_stack): Modify to track
8124 machine_frame_state::sp_realigned.
8125 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8126 (ix86_nsaved_sseregs): Likewise.
8127 (ix86_emit_save_regs): Likewise.
8128 (ix86_emit_save_regs_using_mov): Likewise.
8129 (ix86_emit_save_sse_regs_using_mov): Likewise.
8130 (get_scratch_register_on_entry): Likewise.
8131 (gen_frame_set): New function.
8132 (gen_frame_load): Likewise.
8133 (gen_frame_store): Likewise.
8134 (emit_outlined_ms2sysv_save): Likewise.
8135 (emit_outlined_ms2sysv_restore): Likewise.
8136 (ix86_expand_prologue): Modify stack re-alignment code and call
8137 emit_outlined_ms2sysv_save when appropriate.
8138 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8139 parameter rtx_insn *insn, which allows the function to be used to only
8141 (ix86_expand_epilogue): Modify validity checks of frame and stack
8142 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8143 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8144 * config/i386/predicates.md
8145 (save_multiple): New predicate.
8146 (restore_multiple): Likewise.
8147 * config/i386/sse.md
8148 (save_multiple<mode>): New pattern.
8149 (save_multiple_realign<mode>): Likewise.
8150 (restore_multiple<mode>): Likewise.
8151 (restore_multiple_and_return<mode>): Likewise.
8152 (restore_multiple_leave_return<mode>): Likewise.
8153 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8155 2017-05-14 Julia Koval <julia.koval@intel.com>
8157 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8158 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8159 (__builtin_ia32_xsetbv): New builtins.
8160 * config/i386/i386.c (ix86_expand_special_args_builtin):
8162 (ix86_expand_builtin): Special expand for new intrinsics.
8163 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8164 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8165 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8167 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8169 * cfganal.c (inverted_post_order_compute): Change argument type
8171 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8172 * df-core.c (rest_of_handle_df_initialize): Adjust.
8173 (rest_of_handle_df_finish): Likewise.
8174 (df_analyze_1): Likewise.
8175 (df_analyze): Likewise.
8176 (loop_inverted_post_order_compute): Change argument to be a vec *.
8177 (df_analyze_loop): Adjust.
8178 (df_get_n_blocks): Likewise.
8179 (df_get_postorder): Likewise.
8180 * df.h (struct df_d): Change field to be a vec.
8181 * lcm.c (compute_laterin): Adjust.
8182 (compute_available): Likewise.
8183 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8184 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8185 * tree-ssa-pre.c (compute_antic): Likewise.
8187 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8189 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8190 (depth_first_search::depth_first_search): Change structure init
8191 function to this constructor.
8192 (depth_first_search::add_bb): Rename function to this member.
8193 (depth_first_search::execute): Likewise.
8194 (flow_dfs_compute_reverse_finish): Adjust.
8196 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8198 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8199 (longest_simple_path): Likewise.
8200 * shrink-wrap.c (spread_components): Likewise.
8201 (disqualify_problematic_components): Likewise.
8202 (emit_common_heads_for_components): Likewise.
8203 (emit_common_tails_for_components): Likewise.
8204 (insert_prologue_epilogue_for_components): Likewise.
8206 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8208 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8211 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8213 * df-core.c (df_set_blocks): Start using auto_bitmap.
8214 (df_compact_blocks): Likewise.
8215 * df-problems.c (df_rd_confluence_n): Likewise.
8216 * df-scan.c (df_insn_rescan_all): Likewise.
8217 (df_process_deferred_rescans): Likewise.
8218 (df_update_entry_block_defs): Likewise.
8219 (df_update_exit_block_uses): Likewise.
8220 (df_entry_block_bitmap_verify): Likewise.
8221 (df_exit_block_bitmap_verify): Likewise.
8222 (df_scan_verify): Likewise.
8223 * lra-constraints.c (lra_constraints): Likewise.
8224 (undo_optional_reloads): Likewise.
8225 (lra_undo_inheritance): Likewise.
8226 * lra-remat.c (calculate_gen_cands): Likewise.
8227 (do_remat): Likewise.
8228 * lra-spills.c (assign_spill_hard_regs): Likewise.
8229 (spill_pseudos): Likewise.
8230 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8231 (bitmap_set_subtract_values): Likewise.
8233 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8235 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8236 management with auto_bitmap.
8237 (fix_inter_tick): Likewise.
8238 (fix_recovery_deps): Likewise.
8239 * ira.c (add_store_equivs): Likewise.
8240 (find_moveable_pseudos): Likewise.
8241 (split_live_ranges_for_shrink_wrap): Likewise.
8242 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8243 (rtx_reuse_manager::seen_def_p): Likewise.
8244 (rtx_reuse_manager::set_seen_def): Likewise.
8245 * print-rtl.h (class rtx_reuse_manager): Likewise.
8247 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8249 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8251 (migrate_btr_def): Likewise.
8252 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8253 * df-core.c (loop_post_order_compute): Likewise.
8254 (loop_inverted_post_order_compute): Likewise.
8255 * hsa-common.h: Likewise.
8256 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8257 * init-regs.c (initialize_uninitialized_regs): Likewise.
8258 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8259 (inline_small_functions): Likewise.
8260 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8261 * ira.c (combine_and_move_insns): Likewise.
8262 (build_insn_chain): Likewise.
8263 * loop-invariant.c (find_invariants): Likewise.
8264 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8265 * predict.c (tree_predict_by_opcode): Likewise.
8266 (predict_paths_leading_to): Likewise.
8267 (predict_paths_leading_to_edge): Likewise.
8268 (estimate_loops_at_level): Likewise.
8269 (estimate_loops): Likewise.
8270 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8271 (spread_components): Likewise.
8272 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8273 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8274 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8275 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8276 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8277 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8278 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8279 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8280 (mark_threaded_blocks): Likewise.
8281 (thread_through_all_blocks): Likewise.
8282 * tree-ssa.c (verify_ssa): Likewise.
8283 (execute_update_addresses_taken): Likewise.
8284 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8286 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8288 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8290 (post_order_compute): Likewise.
8291 (inverted_post_order_compute): Likewise.
8292 (pre_and_rev_post_order_compute_fn): Likewise.
8294 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8296 * genrecog.c (int_set::int_set): Explicitly construct our
8297 auto_vec base class.
8298 * vec.h (auto_vec::auto_vec): New constructor.
8300 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8302 * bitmap.h (class auto_bitmap): New constructor taking
8303 bitmap_obstack * argument.
8305 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8307 * bitmap.h (class auto_bitmap): Change type of m_bits to
8308 bitmap_head, and adjust ctor / dtor and member operators.
8310 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8312 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8313 when returned register mode doesn't match original mode.
8315 2017-05-12 Jeff Law <law@redhat.com>
8316 Jakub Jelinek <jakub@redhat.com>
8318 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8319 we look for cc setter after the compare-elim changes.
8320 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8321 within the vector to match what compare-elim now expects.
8322 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8323 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8325 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8326 after the compare-elim changes.
8327 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8328 the vector to match what compare-elim now expects.
8329 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8330 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8331 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8332 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8333 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8335 * config/visium/visium.c (single_set_and_flags): Fix where
8336 we look for cc setter after the compare-elim changes.
8337 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8338 with the vector to match what compare-elim now expects.
8339 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8340 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8341 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8342 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8343 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8344 (neg<mode>2_insn_set_overflow): Likewise.
8346 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8349 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8350 maybe_expand_insn call, set ops[0].target. If still set after call,
8351 set alt_rtl. Add extra arg to recursive calls.
8352 (extract_bit_field): Add alt_rtl argument. Pass to
8354 * expmed.h (extract_bit_field): Fix prototype.
8355 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8356 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8357 to extract_bit_field_calls.
8358 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8359 Pass alt_rtl to extract_bit_field calls.
8360 * calls.c (store_unaligned_arguments_into_psuedos)
8361 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8362 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8364 * optabs.h (struct expand_operand): Add target bitfield.
8366 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8368 * compare-elim.c (try_eliminate_compare): Canonicalize
8369 operation with embedded compare to
8370 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8371 (set (reg) (operation)].
8373 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8375 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8378 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8379 cost of adding a carry flag for ADC instruction.
8380 [case MINUS]: Ignore the cost of subtracting a carry flag
8381 for SBB instruction.
8383 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8385 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8387 * config/rs6000/bmiintrin.h: New file.
8388 * config/rs6000/bmi2intrin.h: New file.
8389 * config/rs6000/x86intrin.h: New file.
8391 2017-05-12 Jeff Law <law@redhat.com>
8393 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8396 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8399 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8401 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8403 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8404 early expansion of vector divide builtins.
8405 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8406 builtins identified as having unsigned arguments.
8408 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8410 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8411 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8412 expansion of vector logical operations (and, andc, or, xor,
8415 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8417 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8418 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8420 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8422 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8423 early GIMPLE expansion of vector multiplies.
8425 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8427 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8428 TARGET_HAVE_MOVT conditional.
8429 (movt splitter): Likewise.
8431 2017-05-12 Richard Biener <rguenther@suse.de>
8433 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8434 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8435 Fold all stmts not inplace.
8437 2017-05-12 Richard Biener <rguenther@suse.de>
8439 PR tree-optimization/80713
8440 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8441 inserted_exprs bit for not removed stmts.
8443 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8446 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8447 parallelized" attribute for parallelized OpenACC kernels.
8448 * omp-offload.c (execute_oacc_device_lower): Use it.
8450 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8451 Set "oacc kernels" attribute.
8452 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8453 parameter. Adjust all users.
8454 (oacc_fn_attrib_kernels_p): Remove function.
8455 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8456 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8457 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8458 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8459 assert "oacc kernels" attribute is set.
8461 2017-05-11 Carl Love <cel@us.ibm.com>
8463 * config/rs6000/rs6000-c: Add support for built-in functions
8464 vector unsigned char vec_popcnt (vector signed char)
8465 vector unsigned char vec_popcnt (vector unsigned char)
8466 vector unsigned short vec_popcnt (vector signed short)
8467 vector unsigned short vec_popcnt (vector unsigned short)
8468 vector unsigned int vec_popcnt (vector signed int)
8469 vector unsigned int vec_popcnt (vector unsigned int)
8470 vector unsigned long long vec_popcnt (vector signed long long)
8471 vector unsigned long long vec_popcnt (vector unsigned long long)
8472 vector signed long long vec_slo (vector signed long long,
8474 vector signed long long vec_slo (vector signed long long,
8475 vector unsigned char)
8476 vector unsigned long long vec_slo (vector unsigned long long,
8478 vector unsigned long long vec_slo (vector unsigned long long,
8479 vector unsigned char)
8480 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8481 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8482 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8483 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8484 * doc/extend.texi: Update the built-in documentation file for the
8485 new built-in functions.
8487 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8489 * attribs.h (sorted_attr_string): Move machine independent
8490 functions for target clone support from the i386 port to common
8491 code. Rename ix86_function_versions to common_function_versions.
8492 Rename make_name to make_unique_name.
8493 (common_function_versions): Likewise.
8494 (make_unique_name): Likewise.
8495 (make_dispatcher_decl): Likewise.
8496 (is_function_default_version): Likewise.
8497 * attribs.c (attr_strcmp): Likewise.
8498 (sorted_attr_string): Likewise.
8499 (common_function_versions): Likewise.
8500 (make_unique_name): Likewise.
8501 (make_dispatcher_decl): Likewise.
8502 (is_function_default_version): Likewise.
8503 * config/i386/i386.c (attr_strcmp): Likewise.
8504 (sorted_attr_string): Likewise.
8505 (ix86_function_versions): Likewise.
8506 (make_name): Likewise.
8507 (make_dispatcher_decl): Likewise.
8508 (is_function_default_version): Likewise.
8509 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8511 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8514 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8515 Account for direct move costs for vec_construct of integer
8518 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8521 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8522 (UNSPEC_STX_ATOMIC): Ditto.
8523 (loaddi_via_sse): New insn.
8524 (storedi_via_sse): Ditto.
8525 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8526 Update corresponding peephole2 patterns.
8527 (atomic_storedi_fpu): Ditto.
8529 2017-05-11 Julia Koval <julia.koval@intel.com>
8531 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8532 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8534 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8535 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8536 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8538 2017-05-11 Nathan Sidwell <nathan@acm.org>
8540 * graphite-poly.c: Include dumpfile.h.
8542 * dumpfle.h (dump_function): Declare here ...
8543 * tree-dump.h (dump_function): ... not here.
8544 * dumpfile.c: #include tree-cfg.h.
8545 (dump_function): Move here from ...
8546 * tree-dump.c (dump_function): ... here.
8547 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8548 * graphite-poly.c: Don't include tree-dump.h.
8549 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8550 * print-tree.c: Likewise.
8551 * stor-layout.c: Likewise.
8552 * tree-nested.c: Likewise.
8554 * dumpfile.c (dump_start): Use TDF_FLAGS.
8555 (dump_enable_all): Fix TDF_KIND check thinko.
8557 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8559 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8560 array entries to represent two legal parameterizations of the
8561 overloaded __builtin_cmpb function, as represented by the
8562 P6_OV_BUILTIN_CMPB constant.
8563 (altivec_resolve_overloaded_builtin): Add special case handling
8564 for the __builtin_cmpb function, as represented by the
8565 P6_OV_BUILTIN_CMPB constant.
8566 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8567 (BU_P6_64BIT_2): New macro.
8568 (BU_P6_OVERLOAD_2): New macro
8569 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8570 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8571 (CMPB): Add overload support to represent both 32-bit and 64-bit
8572 compare-bytes function.
8573 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8574 support for TARGET_CMPB.
8575 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8576 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8577 documentation of the __builtin_cmpb overloaded built-in function.
8579 2017-05-11 Richard Biener <rguenther@suse.de>
8581 PR tree-optimization/80705
8582 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8583 bases are not vectorizable.
8585 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8587 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8588 when counting register pressure.
8590 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8592 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8593 (iv_ca_more_deps): Renamed to ...
8594 (iv_ca_compare_deps): ... this.
8595 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8597 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8599 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8601 (determine_group_iv_costs): ... here.
8602 (find_inv_vars_cb): Record inv var if it's not recorded before.
8604 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8606 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8607 (get_shiftadd_cost): Ditto.
8609 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8611 * tree-ssa-address.c: Include header file.
8612 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8614 (add_to_parts): Refactor.
8615 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8616 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8619 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8621 PR tree-optimization/53090
8622 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8624 (extract_cond_operands): Detect condition with IV on both sides
8625 and return COMP_IV_EXPR_2.
8626 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8627 (rewrite_use_compare): Simplify by removing call to function
8628 extract_cond_operands.
8630 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8632 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8633 (extract_cond_operands): Detect condition comparing against non-
8634 invariant bound and return appropriate enum value.
8635 (find_interesting_uses_cond): Update use of extract_cond_operands.
8636 Handle its return value accordingly.
8637 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8639 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8641 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8642 nonlinear iv_use computation in loop invariant sensitive way.
8644 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8646 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8647 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8649 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8651 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8652 (dump_cand): Support iv_cand.inv_exprs.
8653 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8655 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8658 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8660 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8662 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8663 as local function. Include necessary header files.
8664 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8666 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8668 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8670 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8672 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8673 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8674 RSHIFT_EXPR and BIT_NOT_EXPR.
8676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8678 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8679 (adjust_setup_cost): New parameter supporting round up adjustment.
8680 (struct address_cost_data): Delete.
8681 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8682 (split_address_cost, ptr_difference_cost): Delete.
8683 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8684 (struct ainc_cost_data): New struct.
8685 (get_address_cost_ainc): New function.
8686 (get_address_cost, get_computation_cost): Reimplement.
8687 (determine_group_iv_cost_address): Record inv_expr for all uses of
8689 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8690 (iv_ca_has_deps): Reimplemented to ...
8691 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8693 (iv_ca_extend): Call iv_ca_more_deps.
8695 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8697 * tree-ssa-address.c (struct mem_address): Move to header file.
8698 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8699 * tree-ssa-address.h (struct mem_address): Move from C file.
8700 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8702 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8704 * tree-affine.h (aff_combination_type): New interface.
8705 (aff_combination_zero_p): Remove static.
8706 (aff_combination_const_p): New interface.
8707 (aff_combination_singleton_var_p): New interfaces.
8709 2017-05-11 Richard Biener <rguenther@suse.de>
8711 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8712 Skip unreachable blocks and destinations.
8713 (eliminate): Move stmt removal and fixup ...
8714 (fini_eliminate): ... here. Skip inserted exprs.
8715 (pass_pre::execute): Move fini_pre after fini_eliminate.
8716 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8717 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8718 PRE to get rid of dead code that has invalid SSA form and
8719 split critical edges again.
8721 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8723 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8725 2017-05-11 Richard Biener <rguenther@suse.de>
8727 * passes.c (execute_function_todo): Verify loops if they are
8728 said to be up-to-date.
8729 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8730 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8732 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8735 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8736 handle calling assemble_external ourself.
8739 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8740 modes with zero size. Enhance comment.
8742 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8744 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8745 built-ins for vec_xl and vec_xst with short and char pointer
8748 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8750 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8751 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8752 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8753 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8754 (_mm_maskz_min_round_ss): New intrinsics.
8755 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8756 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8757 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8758 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8759 (__builtin_ia32_minss_mask_round): New builtins.
8760 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8761 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8762 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
8764 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
8765 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
8767 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
8770 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8772 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
8773 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
8774 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
8775 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
8776 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
8777 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
8778 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
8779 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8780 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8781 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
8782 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
8783 (__builtin_ia32_mulss_mask_round): New builtins.
8784 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8785 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8786 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
8788 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
8789 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8791 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8794 2017-05-10 Julia Koval <julia.koval@intel.com>
8796 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
8797 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
8798 (_mm256_setr_m128i): New intrinsics.
8800 2017-05-10 Julia Koval <julia.koval@intel.com>
8802 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
8803 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
8804 (_mm_maskz_rcp14_ss): New intrinsics.
8805 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
8806 (__builtin_ia32_rcp14ss_mask): New builtins.
8807 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
8809 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
8811 PR tree-optimization/51513
8812 * tree-cfg.c (gimple_seq_unreachable_p): New function.
8813 (assert_unreachable_fallthru_edge_p): Use it.
8814 (group_case_labels_stmt): Likewise.
8815 * tree-cfg.h: Prototype it.
8816 * stmt.c: Include cfghooks.h and tree-cfg.h.
8817 (emit_case_dispatch_table) <gap_label>: New local variable.
8818 Use it to fill dispatch table gaps.
8819 Test for default_label before updating probabilities.
8820 (expand_case) <default_label>: Remove unneeded initialization.
8821 Test for unreachable default case statement and remove its edge.
8822 Set default_label accordingly.
8823 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
8825 2017-05-10 Carl Love <cel@us.ibm.com>
8827 * config/rs6000/rs6000-c: Add support for built-in functions
8828 vector signed char vec_neg (vector signed char)
8829 vector signed short int vec_neg (vector short int)
8830 vector signed int vec_neg (vector signed int)
8831 vector signed long long vec_neg (vector signed long long)
8832 vector float vec_neg (vector float)
8833 vector double vec_neg (vector double)
8834 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
8836 * config/rs6000/altivec.h: Add define for vec_neg
8837 * doc/extend.texi: Update the built-in documentation for the
8838 new built-in functions.
8840 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8842 PR tree-optimization/77644
8843 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
8845 2017-05-10 Nathan Sidwell <nathan@acm.org>
8847 * dumpfile.h (TDI_lang_all): New.
8848 (TDF_KIND): New. Renumber others
8849 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
8851 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
8853 (get_dump_file_name): Adjust suffix generation.
8854 (dump_enable_all): Use TDF_KIND.
8855 * doc/invoke.texi (-fdump-lang-all): Document.
8857 * dumpfile.h: Tabify.
8859 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
8862 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
8863 Move member access before delete.
8865 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
8867 * tree-inline.c (expand_call_inline): Split block at stmt
8870 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8873 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
8874 are now unused after splitting mov{sf,sd}_hardfloat.
8875 (f32_lr2): Likewise.
8877 (f32_lm2): Likewise.
8879 (f32_li2): Likewise.
8882 (f32_sr2): Likewise.
8884 (f32_sm2): Likewise.
8886 (f32_si2): Likewise.
8889 (f32_vsx): Likewise.
8891 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
8892 For movsf, order stores so the VSX stores occur before the GPR
8893 store which encourages the register allocator to use a traditional
8894 FPR instead of a GPR. For movsd, order the stores so that the GPR
8895 store comes before the VSX stores to allow the power6 to work.
8896 This is due to the power6 not having a 32-bit integer store
8897 instruction from a FPR.
8898 (movsf_hardfloat): Likewise.
8899 (movsd_hardfloat): Likewise.
8901 2017-05-09 Martin Sebor <msebor@redhat.com>
8903 PR translation/80280
8904 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
8907 PR translation/80280
8908 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
8909 data member added in r247778.
8910 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
8912 2017-05-09 Nathan Sidwell <nathan@acm.org>
8914 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
8916 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
8919 2017-05-09 Marek Polacek <polacek@redhat.com>
8921 * doc/invoke.texi: Fix typo.
8923 2017-05-09 Richard Biener <rguenther@suse.de>
8925 * tree-vrp.c (vrp_val_is_max): Adjust comment.
8926 (vrp_val_is_min): Likewise.
8927 (set_value_range_to_value): Likewise.
8928 (set_value_range_to_nonnegative): Likewise.
8929 (gimple_assign_nonzero_p): Likewise.
8930 (gimple_stmt_nonzero_p): Likewise.
8931 (vrp_int_const_binop): Likewise. Remove unreachable case.
8932 (adjust_range_with_scev): Adjust comments.
8933 (compare_range_with_value): Likewise.
8934 (extract_range_from_phi_node): Likewise.
8935 (test_for_singularity): Likewise.
8937 2017-05-09 Richard Biener <rguenther@suse.de>
8939 * tree-vrp.c (get_single_symbol): Add assert that we don't
8940 get overflowed constants as invariant part.
8941 (compare_values_warnv): Add comment before the TREE_NO_WARNING
8942 checks. Use wi::cmp instead of recursing for integer constants.
8943 (compare_values): Just ignore whether we assumed undefined
8944 overflow instead of failing the compare.
8945 (extract_range_for_var_from_comparison_expr): Add comment before the
8946 TREE_NO_WARNING sets.
8947 (test_for_singularity): Likewise.
8948 (extract_range_from_comparison): Do not disable optimization
8949 when we assumed undefined overflow.
8950 (extract_range_basic): Remove init of unused var.
8952 2017-05-09 Richard Biener <rguenther@suse.de>
8954 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
8955 (extract_range_from_multiplicative_op_1): Adjust.
8956 (extract_range_from_binary_expr_1): Use int_const_binop.
8958 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
8961 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
8962 rs6000_store_data_bypass_p in seven define_bypass directives and
8963 in several comments.
8964 * config/rs6000/rs6000-protos.h: Add prototype for
8965 rs6000_store_data_bypass_p function.
8966 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
8967 function implements slightly different (rs6000-specific) semantics
8968 than store_data_bypass_p, returning false rather than aborting
8969 with assertion error when arguments do not satisfy the
8970 requirements of store data bypass.
8971 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
8972 rs6000_store_data_bypass_p.
8974 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
8976 * config/xtensa/xtensa-protos.h
8977 (xtensa_initial_elimination_offset): New declaration.
8978 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
8979 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
8980 macro definition, add case for FRAME_POINTER_REGNUM when
8981 FRAME_GROWS_DOWNWARD.
8982 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
8983 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
8984 xtensa_initial_elimination_offset.
8986 2017-05-08 Nathan Sidwell <nathan@acm.org>
8988 * doc/invoke.texi: Alphabetize -fdump options.
8990 2017-05-08 Martin Sebor <msebor@redhat.com>
8992 PR translation/80280
8993 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
8995 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
8997 * target.def (compute_frame_layout): New optional target hook.
8998 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
8999 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9000 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9002 * reload1.c (verify_initial_elim_offsets): Likewise.
9003 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9004 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9005 (arm_get_frame_offsets): Split up into this ...
9006 (arm_compute_frame_layout): ... and this function.
9008 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9010 * config/aarch64/constraints.md (Usa): New constraint.
9011 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9013 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9015 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9016 with_multilib_list after it has been checked.
9018 2017-05-08 Richard Biener <rguenther@suse.de>
9020 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9021 (bitmap_set_subtract_values): Likewise.
9023 2017-05-08 Richard Biener <rguenther@suse.de>
9025 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9026 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9028 (gimple_stmt_nonzero_warnv_p): Rename to ...
9029 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9031 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9032 (extract_range_basic): Adjust, do not disable propagation on
9033 strict overflow sensitive simplification.
9034 (vrp_visit_cond_stmt): Likewise.
9036 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9038 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9039 body size unconditionally.
9041 2017-05-07 Jeff Law <law@redhat.com>
9044 2017-05-06 Jeff Law <law@redhat.com>
9045 PR tree-optimization/78496
9046 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9049 PR tree-optimization/78496
9050 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9051 (simplify_stmt_using_ranges): Call it.
9052 (vrp_dom_walker::before_dom_children): Extract equivalences
9053 from an ASSERT_EXPR with an equality comparison against a
9056 2017-05-06 Jeff Law <law@redhat.com>
9058 PR tree-optimization/78496
9059 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9062 PR tree-optimization/78496
9063 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9064 (simplify_stmt_using_ranges): Call it.
9065 (vrp_dom_walker::before_dom_children): Extract equivalences
9066 from an ASSERT_EXPR with an equality comparison against a
9069 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9071 * lra-constraints.c (lra_copy_reg_equiv): New function.
9072 (split_reg): Use it to copy equivalence information from the
9073 original register to the spill register.
9075 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9077 PR rtl-optimization/75964
9078 * simplify-rtx.c (simplify_const_relational_operation): Remove
9079 invalid handling of comparisons of integer ABS.
9081 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9083 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9085 (init_regs): Remove declaration.
9086 (function_arg_advance_32): Initialize error_p as boolean variable.
9088 2017-05-05 Nathan Sidwell <nathan@acm.org>
9090 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9091 lines. Use for (;;).
9093 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9095 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9096 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9097 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9098 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9099 VF=2 that require versioning.
9101 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9103 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9106 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9108 * diagnostic.h (diagnostic_override_option_index): Convert from
9109 macro to inline function.
9111 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9113 * diagnostic.c (last_module_changed_p): New function.
9114 (set_last_module): New function.
9115 (diagnostic_report_current_module): Convert macro usage to
9116 the above functions.
9117 * diagnostic.h (diagnostic_context::last_module): Strengthen
9118 from const line_map * to const line_map_ordinary *.
9119 (diagnostic_last_module_changed): Delete macro.
9120 (diagnostic_set_last_module): Delete macro.
9122 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9124 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9125 with diagnostic_report_diagnostic.
9126 (diagnostic_n_impl_richloc): Likewise.
9127 * diagnostic.h (report_diagnostic): Delete macro.
9128 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9129 with diagnostic_report_diagnostic.
9130 * substring-locations.c (format_warning_va): Likewise.
9132 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9134 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9135 save/restor of format_spec. Move option-printing code to...
9136 (print_option_information): ...this new function, and
9137 reimplement by simply printing to the pretty_printer,
9138 rather than appending to the format string.
9140 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9142 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9143 handling logic into...
9144 (update_effective_level_from_pragmas): ...this new function.
9146 2017-05-04 Andrew Waterman <andrew@sifive.com>
9148 * config/riscv/riscv.opt (mstrict-align): New option.
9149 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9150 (SLOW_UNALIGNED_ACCESS): Define.
9151 (riscv_slow_unaligned_access): Declare.
9152 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9154 (riscv_slow_unaligned_access): New variable.
9155 (rocket_tune_info): Set slow_unaligned_access to true.
9156 (optimize_size_tune_info): Set slow_unaligned_access to false.
9157 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9158 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9159 (riscv_option_override): Set riscv_slow_unaligned_access.
9160 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9162 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9164 * config/riscv/riscv.md: Unify indentation.
9166 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9171 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9173 (extendsi<mode>2): Add support for doing sign extension via
9174 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9175 don't have ISA 3.0 instructions.
9176 (extendsi<mode>2 splitter): Likewise.
9177 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9178 generate the normal insns since SImode can now go in vector
9179 registers. Disallow the special UNSPECs needed for previous
9180 machines to hide SImode being used. Add new insns
9181 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9182 (fix_trunc<mode>si2_stfiwx): Likewise.
9183 (fix_trunc<mode>si2_internal): Likewise.
9184 (fixuns_trunc<mode>si2): Likewise.
9185 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9186 (fctiw<u>z_<mode>_smallint): Likewise.
9187 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9188 of floating point to 32-bit integer from doing a direct move to
9189 the GPR registers to do a store.
9190 (fctiwz_<mode>): Break long line.
9192 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9194 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9195 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9196 (addr_list, addr_offset_valid_p): New.
9197 (split_address_groups): Check offset validity with above function.
9198 (gt-tree-ssa-loop-ivopts.h): Include header file.
9200 2017-05-05 Nathan Sidwell <nathan@acm.org>
9202 * config.gcc (arm*-*-*): Add missing 'fi'.
9204 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9206 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9207 included in -fopt-info does not matter.
9208 * doc/optinfo.texi (-fopt-info): Fix description of default
9209 behavour. Explicitly say order of options included in -fopt-info
9212 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9214 * config.gcc: Allow combinations of aprofile and rmprofile values for
9215 --with-multilib-list.
9216 * config/arm/t-multilib: New file.
9217 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9218 variables. Remove setting of ISA and floating-point ABI in
9219 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9220 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9221 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9222 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9224 * config/arm/t-rmprofile: Likewise except for the matches changes.
9225 * doc/install.texi (--with-multilib-list): Document the combination of
9226 aprofile and rmprofile values and warn about pitfalls in doing that.
9228 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9230 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9231 (movdi_aarch64): Likewise.
9233 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9235 PR tree-optimization/80632
9236 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9238 (build_arrays): Initialize it for virtual phis.
9239 (fix_phi_nodes): Use it for virtual phis.
9241 PR tree-optimization/80558
9242 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9243 [x, y] op z into [x op, y op z] for op & or | if conditions
9246 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9247 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9250 * config/arm/arm.md (use_literal_pool): Remove.
9251 (64-bit immediate split): No longer takes cost into consideration
9252 if arm_disable_literal_pool is enabled.
9253 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9254 used when arm_disable_literal_pool is enabled.
9255 (arm_max_const_double_inline_cost): Remove use of
9256 arm_disable_literal_pool.
9257 (push_minipool_fix): Add assert.
9258 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9259 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9260 (no_literal_pool_sf_immediate): New.
9262 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9264 PR tree-optimization/80613
9265 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9266 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9268 2017-05-05 Richard Biener <rguenther@suse.de>
9270 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9272 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9274 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9275 of this flag from insn conditions due to removal from r247495.
9277 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9279 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9281 (arm_early_store_addr_dep_ptr): Likewise.
9282 * config/arm/aarch-common-protos.h
9283 (arm_early_load_addr_dep_ptr): Add prototype.
9284 (arm_early_store_addr_dep_ptr): Likewise.
9285 * config/arm/cortex-a53.md: Add new bypasses.
9287 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9289 * tree.c (next_type_uid): Change type to unsigned.
9290 (type_hash_canon): Decrement back next_type_uid if
9291 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9292 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9295 2017-05-04 Martin Sebor <msebor@redhat.com>
9297 * builtins.c: Fix a trivial typo in a comment.
9300 * builtins.c (check_sizes): Adjust to handle reading past the end.
9301 Avoid printing excessive upper bound of ranges. Use %E to print
9302 tree nodes instead of converting them to %wu.
9303 (expand_builtin_memchr): New function.
9304 (compute_dest_size): Rename...
9305 (compute_objsize): ...to this.
9306 (expand_builtin_memcpy): Adjust.
9307 (expand_builtin_mempcpy): Adjust.
9308 (expand_builtin_strcat): Adjust.
9309 (expand_builtin_strcpy): Adjust.
9310 (check_strncat_sizes): Adjust.
9311 (expand_builtin_strncat): Adjust.
9312 (expand_builtin_strncpy): Adjust and simplify.
9313 (expand_builtin_memset): Adjust.
9314 (expand_builtin_bzero): Adjust.
9315 (expand_builtin_memcmp): Adjust.
9316 (expand_builtin): Handle memcmp.
9317 (maybe_emit_chk_warning): Check strncat just once.
9319 2017-05-04 Martin Sebor <msebor@redhat.com>
9321 PR preprocessor/79214
9324 * builtins.c (check_sizes): Add inlining context and issue
9325 warnings even when -Wno-system-headers is set.
9326 (check_strncat_sizes): Same.
9327 (expand_builtin_strncat): Same.
9328 (expand_builtin_memmove): New function.
9329 (expand_builtin_stpncpy): Same.
9330 (expand_builtin): Handle memmove and stpncpy.
9332 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9334 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9335 which is not used any more.
9337 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9339 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9341 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9343 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9344 (cortexa53_tunings): Likewise.
9345 (cortexa57_tunings): Likewise.
9346 (cortexa72_tunings): Likewise.
9347 (cortexa73_tunings): Likewise.
9349 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9351 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9352 Set loop alignment to 8.
9354 2017-05-04 Martin Sebor <msebor@redhat.com>
9356 PR translation/80280
9357 * builtins.c (expand_builtin_object_size): Add missing quoting to
9358 %D and like directives.
9359 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9360 (hsa_type_for_tree_type): Same.
9361 (verify_function_arguments): Same.
9362 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9363 * varasm.c (get_section): Same.
9366 2017-05-04 Martin Sebor <msebor@redhat.com>
9368 PR translation/80280
9369 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9371 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9373 * config/aarch64/aarch64.c (generic_addrcost_table):
9374 Change HI/TI mode setting.
9376 2017-05-04 Martin Jambor <mjambor@suse.cz>
9378 PR tree-optimization/80622
9379 * tree-sra.c (comes_initialized_p): New function.
9380 (build_accesses_from_assign): Only set write lazily when
9381 comes_initialized_p is false.
9382 (analyze_access_subtree): Use comes_initialized_p.
9383 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9384 instead of testing for PARM_DECL.
9386 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9388 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9389 constraint on operand 0 to allow more general addressing modes.
9390 Adjust output template.
9391 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9393 * config/aarch64/aarch64-protos.h
9394 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9395 * config/aarch64/constraints.md (Dp): New address constraint.
9396 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9399 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9401 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9402 update use of estimate_ipcp_clone_size_and_time.
9403 (estimate_local_effects): Update use of
9404 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9405 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9406 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9407 Return nonspecialized time.
9409 2017-05-04 Richard Biener <rguenther@suse.de>
9411 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9412 for the last VUSE which def dominates the PHI. Directly call
9414 (get_continuation_for_phi_1): Remove.
9416 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9418 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9419 to explain the use of truncating division. Cap the number of
9420 iterations to the maximum given by nb_iterations_upper_bound,
9423 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9425 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9426 * configure: Regenerate.
9427 * config.in: Regenerate.
9428 * config/i386/driver-mingw32.c: new file.
9429 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9430 * config.host: Link driver-mingw32.o on MinGW host.
9431 * doc/install.texi: Document new --enable-mingw-wildcard configure
9434 2017-05-04 Marek Polacek <polacek@redhat.com>
9436 PR tree-optimization/80612
9437 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9439 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9440 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9442 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9443 (movt splitter): Likewise.
9444 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9445 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9446 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9447 block for Thumb-1 with MOVT.
9448 (thumb2_legitimate_address_p): Move code block ...
9449 (can_avoid_literal_pool_for_label_p): ... into this new function.
9450 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9452 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9453 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9454 "M-profile targets with the MOVT instruction".
9456 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9458 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9459 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9460 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9462 2017-05-04 Martin Liska <mliska@suse.cz>
9464 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9467 2017-05-04 Richard Biener <rguenther@suse.de>
9469 * tree.c (array_at_struct_end_p): Handle arrays at struct
9470 end with flexarrays more conservatively. Refactor and treat
9471 arrays of arrays or aggregates more strict. Fix
9472 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9473 * tree.c (array_at_struct_end_p): Adjust prototype.
9474 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9475 * gimple-fold.c (get_range_strlen): Likewise.
9476 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9478 2017-05-04 Richard Biener <rguenther@suse.de>
9480 PR tree-optimization/31130
9481 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9483 (supports_overflow_infinity): Likewise.
9484 (is_negative_overflow_infinity): Likewise.
9485 (is_positive_overflow_infinity): Likewise.
9486 (is_overflow_infinity): Likewise.
9487 (stmt_overflow_infinity): Likewise.
9488 (overflow_infinity_range_p): Likewise.
9489 (usable_range_p): Remove as always returning true.
9490 (make_overflow_infinity): Remove.
9491 (negative_overflow_infinity): Likewise.
9492 (positive_overflow_infinity): Likewise.
9493 (avoid_overflow_infinity): Likewise.
9494 (set_value_range): Adjust accordingly.
9495 (set_value_range_to_nonnegative): Likewise, remove now unused
9496 overflow_infinity arg.
9497 (vrp_operand_equal_p): Adjust.
9498 (update_value_range): Likewise.
9499 (range_int_cst_singleton_p): Likewise.
9500 (operand_less_p): Likewise.
9501 (compare_values_warnv): Likewise.
9502 (extract_range_for_var_from_comparison_expr): Likewise.
9503 (vrp_int_const_binop): Likewise.
9504 (zero_nonzero_bits_from_vr): Likewise.
9505 (extract_range_from_multiplicative_op_1): Likewise.
9506 (extract_range_from_binary_expr_1): Likewise.
9507 (extract_range_from_unary_expr): Likewise.
9508 (extract_range_from_comparison): Likewise.
9509 (extract_range_basic): Likewise.
9510 (adjust_range_with_scev): Likewise.
9511 (compare_ranges): Likewise.
9512 (compare_range_with_value): Likewise.
9513 (dump_value_range): Likewise.
9514 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9516 (simplify_cond_using_ranges): Adjust.
9518 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9520 * brig-builtins.def: Added a builtin for class_f64.
9521 * builtin-types.def: Added a builtin type needed by class_f64.
9523 2017-05-03 Jason Merrill <jason@redhat.com>
9525 * timevar.def: Add TV_CONSTEXPR.
9527 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9529 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9531 2017-05-03 Martin Jambor <mjambor@suse.cz>
9533 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9534 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9535 * ipa-cp.c (ipcp_propagate_stage): Do not call
9536 ipa_update_after_lto_read.
9537 * ipa-inline.c (ipa_inline): Likewise.
9539 2017-05-03 Martin Jambor <mjambor@suse.cz>
9541 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9542 tag. Added a default constructor and a destructor.
9543 (ipa_edge_args_sum_t): New class;
9544 (ipa_edge_args_sum): Declare.
9545 (ipa_edge_args_vector): Remove declaration.
9546 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9547 (ipa_free_edge_args_substructures): Remove declaration.
9548 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9549 (ipa_edge_args_info_available_for_edge_p): Likewise.
9550 * ipa-prop.c (ipa_edge_args_vector): Removed.
9551 (edge_removal_hook_holder): Likewise.
9552 (edge_duplication_hook_holder): Likewise.
9553 (ipa_edge_args_sum): New variable.
9554 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9555 ipa_edge_args_vector.
9556 (ipa_free_edge_args_substructures): Likewise.
9557 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9558 ipa_edge_args_vector.
9559 (ipa_edge_removal_hook): Turned into method
9560 ipa_edge_args_sum_t::remove.
9561 (ipa_edge_duplication_hook): Turned into method
9562 ipa_edge_args_sum_t::duplicate.
9563 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9564 registering edge hooks.
9565 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9566 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9567 ipa_edge_args_sum instead of ipa_edge_args_vector.
9568 * ipa-profile.c (ipa_profile): Likewise.
9570 2017-05-03 Martin Jambor <mjambor@suse.cz>
9572 * symbol-summary.h (function_summary): New method exists.
9573 (function_summary::symtab_removal): Deallocate through release.
9574 (call_summary): New class.
9575 (gt_ggc_mx): New overload.
9576 (gt_pch_nx): Likewise.
9577 (gt_pch_nx): Likewise.
9579 2017-05-03 Jeff Law <law@redhat.com>
9581 PR tree-optimization/78496
9582 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9583 from simplify_cond_using_ranges. Split off code to walk
9584 backwards through casts into ...
9585 (simplify_cond_using_ranges_2): New function.
9586 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9587 (execute_vrp): After identifying jump threads, call
9588 simplify_cond_using_ranges_2.
9590 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9593 * ipa-inline.h (inline_summary): Add ctor.
9594 (create_ggc): Do not use ggc_cleared_alloc.
9596 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9597 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9599 * gcc.c (handle_braces): Support escaping in switch matching
9601 * doc/invoke.texi (Spec Files): Document it.
9602 Remove superfluous @code markup in items.
9604 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9606 * diagnostic-show-locus.c (struct column_range): New struct.
9607 (get_affected_columns): New function.
9608 (get_printed_columns): New function.
9609 (struct correction): New struct.
9610 (correction::ensure_capacity): New function.
9611 (correction::ensure_terminated): New function.
9612 (struct line_corrections): New struct.
9613 (line_corrections::~line_corrections): New dtor.
9614 (line_corrections::add_hint): New function.
9615 (layout::print_trailing_fixits): Reimplement in terms of the new
9617 (selftest::test_overlapped_fixit_printing): New function.
9618 (selftest::diagnostic_show_locus_c_tests): Call it.
9620 2017-05-03 Nathan Sidwell <nathan@acm.org>
9622 Canonicalize canonical type hashing
9623 * tree.h (type_hash_canon_hash): Declare.
9624 * tree.c (type_hash_list, attribute_hash_list): Move into
9625 type_hash_canon_hash.
9626 (build_type_attribute_qual_variant): Break out hash code calc into
9627 type_hash_canon_hash.
9628 (type_hash_canon_hash): New. Generic type hash computation.
9629 (build_range_type_1, build_array_type_1, build_function_type,
9630 build_method_type_directly, build_offset_type, build_complex_type,
9631 make_vector_type): Call it.
9633 2017-05-03 Richard Biener <rguenther@suse.de>
9635 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9636 When all DRs have unknown misaligned do not always peel
9637 when there is a store but apply the same costing model as if
9638 there were only loads.
9640 2017-05-03 Richard Biener <rguenther@suse.de>
9643 PR tree-optimization/80492
9644 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9645 compare_base_decls returning dont-know properly.
9647 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9649 * config/arm/iterators.md (CCSI): New mode iterator.
9650 (arch): New mode attribute.
9651 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9652 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9653 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9654 code iterator for success result mode.
9655 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9656 the corresponding new insn generators.
9658 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9661 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9662 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9664 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9666 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9667 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9668 (DDR_A): Wrap DDR argument in brackets.
9669 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9670 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9671 (DDR_REVERSED_P): Likewise.
9673 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9675 PR tree-optimization/79472
9676 * tree-switch-conversion.c (struct switch_conv_info): Add
9677 contiguous_range and default_case_nonstandard fields.
9678 (collect_switch_conv_info): Compute contiguous_range and
9679 default_case_nonstandard fields, don't clear final_bb if
9680 contiguous_range and only the default case doesn't have the required
9682 (check_all_empty_except_final): Set default_case_nonstandard instead
9683 of failing if contiguous_range and the default case doesn't have empty
9685 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9686 and only the default case doesn't have the required constants. Skip
9688 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9689 if default_case_nonstandard.
9690 (build_constructors): Build constant 1 just once. Assert that default
9691 values aren't inserted in between cases if contiguous_range. Skip
9693 (build_arrays): Skip virtual phis.
9694 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9695 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9696 Handle virtual phis.
9697 (gen_inbound_check): Handle default_case_nonstandard case.
9698 (process_switch): Adjust check_final_bb caller. Call
9699 gather_default_values with the first non-default case instead of
9700 default case if default_case_nonstandard.
9702 2017-05-02 Nathan Sidwell <nathan@acm.org>
9704 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9705 check. Fix formatting.
9707 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9709 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9710 errors when comparing specialized and unspecialized times.
9712 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9714 * diagnostic-show-locus.c
9715 (layout::should_print_annotation_line_p): Make private.
9716 (layout::print_annotation_line): Make private.
9717 (layout::annotation_line_showed_range_p): Make private.
9718 (layout::show_ruler): Make private.
9719 (layout::print_source_line): Make private. Pass in line and
9720 line_width, rather than calling location_get_source_line. Drop
9722 (layout::print_leading_fixits): New method.
9723 (layout::print_any_fixits): Rename to...
9724 (layout::print_trailing_fixits): ...this, and make private.
9725 Don't print newline fixits.
9726 (diagnostic_show_locus): Move logic for printing one row into...
9727 (layout::print_line): ...this new function. Move the
9728 location_get_source_line call and error-handling from
9729 print_source_line to here. Call print_leading_fixits, and rename
9730 print_any_fixits to print_trailing_fixits.
9731 (selftest::test_fixit_insert_containing_newline): Update now that
9732 newlines are partially supported.
9733 (selftest::test_fixit_insert_containing_newline_2): New test.
9734 (selftest::test_fixit_replace_containing_newline): Update comments.
9735 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9736 * edit-context.c (class added_line): New class.
9737 (class edited_line): Describe newline handling in comment.
9738 (edited_line::actually_edited_p): New method.
9739 (edited_line::print_content): Delete redundant decl.
9740 (edited_line::m_predecessors): New field.
9741 (edited_file::print_content): Call edited_line::print_content.
9742 (edited_file::print_diff): Update to support newlines.
9743 (edited_file::print_diff_hunk): Likewise.
9744 (edited_file::print_run_of_changed_lines): New function.
9745 (edited_file::print_diff_line): Convert to...
9746 (print_diff_line): ...this.
9747 (edited_file::get_effective_line_count): New function.
9748 (edited_line::edited_line): Initialize new field m_predecessors.
9749 (edited_line::~edited_line): Clean up m_predecessors.
9750 (edited_line::apply_fixit): Handle newlines.
9751 (edited_line::get_effective_line_count): New function.
9752 (edited_line::print_content): New function.
9753 (edited_line::print_diff_lines): New function.
9754 (selftest::test_applying_fixits_insert_containing_newline): New
9756 (selftest::test_applying_fixits_replace_containing_newline): New
9758 (selftest::insert_line): New function.
9759 (selftest::test_applying_fixits_multiple_lines): Add example of
9761 (selftest::edit_context_c_tests): Call the new tests.
9763 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9765 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
9766 parameter cand. Update dump information.
9767 (get_computation_cost): Update uses.
9769 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9771 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
9772 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
9773 (get_computation_at, rewrite_use_address): Update use of
9774 get_computation_aff.
9776 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9778 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
9779 (get_computation): Delete.
9780 (get_computation_cost): Implement like get_computation_cost_at.
9781 Use get_computation_at.
9782 (get_computation_cost_at): Delete.
9783 (rewrite_use_nonlinear_expr): Use get_computation_at.
9784 (rewrite_use_compare, remove_unused_ivs): Ditto.
9786 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9788 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
9790 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9792 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
9793 (ivopts_global_cost_for_size): Rename parameter and update uses.
9794 (iv_ca_recount_cost): Update uses.
9795 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
9796 candidates seperately in n_invs and n_cands.
9797 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
9799 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9801 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
9802 (find_inv_vars_cb): New.
9803 (find_depends): Renamed to ...
9804 (find_inv_vars): ... this.
9805 (add_candidate_1, force_var_cost): Call find_inv_vars.
9806 (split_address_cost, determine_group_iv_cost_cond): Ditto.
9808 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9810 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
9811 inv_vars. Add inv_exprs.
9812 (struct iv_cand): Rename depends_on to inv_vars.
9813 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
9814 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
9815 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
9816 (dump_cand): Dump inv_vars.
9817 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
9818 (record_invariant, find_depends, add_candidate_1): Ditto.
9819 (set_group_iv_cost, force_var_cost): Ditto.
9820 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
9821 (get_computation_cost_at, get_computation_cost): Ditto.
9822 (determine_group_iv_cost_generic): Ditto.
9823 (determine_group_iv_cost_address): Ditto.
9824 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
9825 (determine_group_iv_costs): Ditto.
9826 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
9827 (iv_ca_set_remove_invariants): Renamed to ...
9828 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
9829 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
9830 (iv_ca_set_add_invariants): Renamed to ...
9831 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
9832 (iv_ca_set_cp): Use iv_ca_set_add_invs.
9833 (iv_ca_has_deps): Support inv_vars and inv_exprs.
9834 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
9835 (create_new_ivs): Remove useless dump.
9837 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9839 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
9841 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
9842 (iv_ca_set_no_cp, create_new_iv): Ditto.
9844 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9846 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9848 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9850 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
9851 function tree_check2.
9853 2017-05-02 Martin Liska <mliska@suse.cz>
9855 * doc/gcov.texi: Add missing preposition.
9856 * gcov.c (function_info::function_info): Properly fill up
9857 all member variables.
9859 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9861 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
9863 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9865 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
9867 2017-05-02 Martin Liska <mliska@suse.cz>
9870 * lto-streamer-in.c (lto_read_tree_1): Remove
9872 * lto-streamer.c (struct tree_hash_entry): Likewise.
9873 (struct tree_entry_hasher): Likewise.
9874 (tree_entry_hasher::hash): Likewise.
9875 (tree_entry_hasher::equal): Likewise.
9876 (lto_streamer_init): Likewise.
9877 (lto_orig_address_map): Likewise.
9878 (lto_orig_address_get): Likewise.
9879 (lto_orig_address_remove): Likewise.
9880 * lto-streamer.h: Likewise.
9881 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
9882 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
9884 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
9886 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
9887 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
9888 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
9889 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
9890 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
9891 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
9892 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
9893 (mm_maskz_sub_ss): New intrinsics.
9894 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9895 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9896 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
9897 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
9898 (__builtin_ia32_subss_mask_round): New builtins.
9899 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9900 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9901 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
9903 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
9904 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9906 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9909 2017-05-02 Martin Jambor <mjambor@suse.cz>
9911 PR tree-optimization/78687
9912 * tree-sra.c (access): New field parent.
9913 (process_subtree_disqualification): New function.
9914 (disqualify_candidate): Call it.
9915 (build_accesses_from_assign): Reset write flag if creating an
9917 (build_access_subtree): Fill in parent field and also prpagate
9918 down grp_write flag.
9919 (create_artificial_child_access): New parameter set_grp_write, set
9920 grp_write to its value.
9921 (propagate_subaccesses_across_link): Also propagate grp_write flag
9923 (propagate_all_subaccesses): Push the closest parent back to work
9924 queue if add_access_to_work_queue returned true.
9926 2017-05-02 Richard Biener <rguenther@suse.de>
9928 * common.opt (fstrict-overflow): Alias negative to fwrapv.
9929 * doc/invoke.texi (fstrict-overflow): Remove all traces of
9930 -fstrict-overflow documentation.
9931 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
9932 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
9933 flag_strict_overflow.
9934 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
9935 * lto-opts.c (lto_write_options): Do not stream it.
9936 * lto-wrapper.c (merge_and_complain): Do not handle it.
9937 * opts.c (default_options_table): Do not set -fstrict-overflow.
9938 (finish_options): Likewise do not clear it when sanitizing.
9939 * simplify-rtx.c (simplify_const_relational_operation): Do not
9940 test flag_strict_overflow.
9942 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9944 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
9945 using enabled attribute.
9946 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
9947 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
9948 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
9949 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
9950 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
9951 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
9952 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
9953 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
9954 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
9955 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
9957 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9959 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
9961 2017-05-02 Richard Biener <rguenther@suse.de>
9963 PR tree-optimization/80591
9965 2017-04-10 Richard Biener <rguenther@suse.de>
9967 * tree-ssa-structalias.c (find_func_aliases): Properly handle
9970 2017-05-02 Richard Biener <rguenther@suse.de>
9972 PR tree-optimization/80549
9973 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
9974 (cleanup_tree_cfg_noloop): Create forwarders to known loop
9975 headers if they do not have a preheader.
9977 2017-05-02 Martin Liska <mliska@suse.cz>
9980 * common.opt: Fix typo.
9981 * doc/invoke.texi: Likewise.
9983 2017-05-01 Jan Beulich <jbeulich@suse.com>
9985 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
9986 swapping, add (x,x,m,x,n) alternative.
9988 2017-05-01 Nathan Sidwell <nathan@acm.org>
9990 * calls.c (combine_pending_stack_adjustment_and_call): Remove
9991 unnecessary unadjusted_alignment check.
9993 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9996 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
9998 * gimplify.c (gimplify_cilk_detach): New function.
9999 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10000 * tree-core.h: Document EXPR_CILK_SPAWN.
10001 * tree.h (EXPR_CILK_SPAWN): Define.
10003 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10005 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10006 to use new fixit_hint representation, using the "replace" logic.
10007 (get_line_span_for_fixit_hint): Likewise.
10008 (layout::print_any_fixits): Likewise.
10009 (selftest::test_one_liner_many_fixits): Rename to...
10010 (selftest::test_one_liner_many_fixits_1): ...this, and update
10011 comment and expected output to reflect that the multiple fix-it
10012 hints are now consolidated into one insertion.
10013 (selftest::test_one_liner_many_fixits_2): New test.
10014 (selftest::test_diagnostic_show_locus_one_liner): Update for
10016 (selftest::test_fixit_consolidation): Update for fix-it API
10018 * diagnostic.c (print_parseable_fixits): Likewise.
10019 * edit-context.c (edited_line::m_line_events): Convert from
10020 auto_vec <line_event *> to auto_vec <line_event>.
10021 (class line_event): Convert from abstract base class to a concrete
10022 class, taking over the role of replace_event.
10023 (class insert_event): Delete.
10024 (class replace_event): Rename to class line_event. Convert to
10026 (edit_context::add_fixits): Reimplement.
10027 (edit_context::apply_insert): Delete.
10028 (edit_context::apply_replace): Rename to...
10029 (edit_context::apply_fixit): ...this. Convert to half-open range.
10030 (edited_file::apply_insert): Delete.
10031 (edited_file::apply_replace): Rename to...
10032 (edited_file::apply_fixit): ...this.
10033 (edited_line::~edited_line): Drop deletion of events.
10034 (edited_line::apply_insert): Delete.
10035 (edited_line::apply_replace): Rename to...
10036 (edited_line::apply_fixit): ...this. Convert to half-open range.
10037 Update for change to type of m_line_events.
10038 * edit-context.h (edit_context::apply_insert): Delete.
10039 (edit_context::apply_replace): Rename to...
10040 (edit_context::apply_fixit): ...this.
10042 2017-05-01 Martin Sebor <msebor@redhat.com>
10044 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10047 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10050 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10051 __get_cpuid_max returns 0.
10052 (__get_cpuid_count): Ditto.
10054 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10056 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10057 replacement expression is another instance of one of its arguments.
10059 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10062 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10063 check for stack push/pop autoinc.
10064 * config/i386/i386.c (ix86_agi_dependent): Return false
10065 if the only reason why modified_in_p returned true is that
10066 addr is SP based and set_insn is a push or pop.
10068 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10070 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10073 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10076 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10077 (account_size_time): Use two predicates - exec_pred and
10079 (evaluate_conditions_for_known_args): Compute both clause and
10081 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10082 (inline_summary_t::duplicate): Update.
10083 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10085 (compute_inline_parameters): Likewise.
10086 (estimate_edge_size_and_time): Update caluclation of time.
10087 (estimate_node_size_and_time): Compute both time and nonspecialized
10089 (estimate_ipcp_clone_size_and_time): Update.
10090 (inline_merge_summary): Update.
10091 (do_estimate_edge_time): Update.
10092 (do_estimate_edge_size): Update.
10093 (do_estimate_edge_hints): Update.
10094 (inline_read_section, inline_write_summary): Stream both new predicates.
10095 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10097 (compute_inlined_call_time): Cleanup.
10098 (big_speedup_p): Update.
10099 (edge_badness): Update.
10100 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10101 (size_time_entry): Replace predicate by exec_predicate and
10102 nonconst_predicate.
10103 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10104 (estimate_edge_time): Return also nonspec_time.
10105 (reset_edge_growth_cache): Update.
10107 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10109 PR rtl-optimization/80491
10110 * ifcvt.c (noce_process_if_block): When looking for x setter
10111 with missing else_bb, don't check only the insn right before
10112 cond_earliest, but look for the last insn that x is modified in
10113 within the same bb.
10115 PR rtl-optimization/80491
10116 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10118 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10120 PR tree-optimization/80487
10121 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10123 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10125 PR tree-optimization/79697
10126 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10127 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10128 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10130 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10131 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10133 2017-04-28 Martin Sebor <msebor@redhat.com>
10135 PR tree-optimization/80523
10136 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10137 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10139 (maybe_warn, format_directive, parse_directive): Use new functions.
10140 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10142 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10144 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10146 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10148 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10149 target_header_dir): Set correctly.
10150 * configure: Regenerated.
10151 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10152 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10153 instead of SYSTEM_HEADER_DIR.
10155 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10157 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10158 (estimate_local_effects): Likewise.
10159 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10160 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10161 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10162 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10163 do_estimate_edge_time, estimate_edge_time): Likewise.
10164 * ipa-inline-analysis.c (estimate_node_size_and_time,
10165 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10166 (estimate_time_after_inlining): Remove.
10168 2017-04-28 Martin Liska <mliska@suse.cz>
10170 * doc/gcov.texi: Enhance documentation of gcov.
10172 2017-04-28 Martin Liska <mliska@suse.cz>
10174 * doc/gcov.texi: Sort options in alphabetic order.
10175 * doc/gcov-dump.texi: Likewise.
10176 * doc/gcov-tool.texi: Likewise.
10177 * gcov.c (print_usage): Likewise.
10178 * gcov-dump.c (print_usage): Likewise.
10179 * gcov-tool.c (print_merge_usage_message): Likewise.
10180 (print_rewrite_usage_message): Likewise.
10181 (print_overlap_usage_message): Likewise.
10183 2017-04-28 Martin Liska <mliska@suse.cz>
10185 PR gcov-profile/53915
10186 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10188 2017-04-28 Martin Liska <mliska@suse.cz>
10190 PR gcov-profile/79891
10191 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10192 is marked by compiler as living on a line.
10193 (get_cycles_count): Remove usage of the union.
10194 (output_intermediate_file): Likewise.
10195 (find_source): Fix GNU coding style.
10196 (accumulate_line_counts): Remove old non-all block mode.
10197 (output_lines): Remove usage of the union.
10198 * profile.c (output_location): Include all BBs, even if
10199 belonging to a same line (and file) as a previous BB.
10201 2017-04-28 Martin Liska <mliska@suse.cz>
10203 * gcov.c (process_args): Handle new argument 'w'.
10204 (read_graph_file): Assign ID to BBs.
10205 (output_branch_count): Display BB # if verbose flag is set.
10206 (output_lines): Likewise for arcs.
10207 (print_usage): Add '--verbose' option help.
10208 * doc/gcov.texi: Document --verbose (-w) option.
10210 2017-04-28 Martin Liska <mliska@suse.cz>
10212 * gcov.c (struct block_location_info): New struct.
10213 (process_file): Fill up the new structure.
10214 (read_graph_file): Replace usage of encoding by the newly added
10216 (add_line_counts): Likewise.
10217 (accumulate_line_counts): Remove usage of the union.
10218 (function_info::function_info): New function.
10219 (function_info::~function_info): Likewise.
10220 (process_file): Call delete instead of release_function.
10221 (release_function): Release the function.
10222 (release_structures): Call delete instead of release_function.
10223 (solve_flow_graph): Replace usage of num_blocks.
10224 (find_exception_blocks): Likewise.
10225 (output_lines): Fix GNU coding style.
10227 2017-04-28 Martin Liska <mliska@suse.cz>
10230 * coverage.c (coverage_remove_note_file): New function.
10231 * coverage.h: Declare the function.
10232 * toplev.c (finalize): Clean if an error has been seen.
10234 2017-04-28 Martin Liska <mliska@suse.cz>
10236 PR gcov-profile/80031
10237 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10238 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10239 * gcov.c (read_graph_file): Read just number of blocks.
10240 * profile.c (branch_prob): Do not stream 0 flags per a basic
10243 2017-04-28 Martin Liska <mliska@suse.cz>
10245 * gcov-dump.c (tag_*): Add new argument to declarations.
10246 (dump_gcov_file): Likewise.
10247 (tag_blocks): Add and use new argument depth.
10248 (tag_arcs): Likewise.
10249 (tag_lines): Likewise.
10250 (tag_counters): Likewise.
10251 (tag_summary): Likewise.
10252 (dump_working_sets): Use depth to do a proper indentation.
10254 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10257 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10258 * symtab.c (symtab_node::debug_symtab): Move definition here.
10260 2017-04-28 Richard Biener <rguenther@suse.de>
10262 * lto-streamer.h (LTO_major_version): Bump to 7.
10264 2017-04-28 Richard Biener <rguenther@suse.de>
10266 * tree-vrp.c (assert_info): New struct.
10267 (add_assert_info): New helper.
10268 (register_edge_assert_for_2): Refactor to add asserts to a vector
10270 (register_edge_assert_for_1): Likewise.
10271 (register_edge_assert_for): Likewise.
10272 (finish_register_edge_assert_for): New helper actually registering
10273 asserts where live on edge.
10274 (find_conditional_asserts): Adjust.
10275 (find_switch_asserts): Likewise.
10276 (evrp_dom_walker::try_find_new_range): Generalize.
10277 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10279 2017-04-27 Marek Polacek <polacek@redhat.com>
10282 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10283 arg10 and arg11 to itype.
10285 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10287 * doc/extend.texi (Object Size Checking): Improve grammar.
10289 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10292 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10293 that the logic for permitting reciprocal estimates matches that
10296 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10299 * tree.c (type_cache_hasher::equal): Only compare
10300 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10301 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10302 non-aggregate element types.
10303 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10304 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10306 2017-04-27 Richard Biener <rguenther@suse.de>
10308 PR middle-end/80533
10309 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10310 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10311 keeping a reference to a trailing array.
10313 2017-04-27 Richard Biener <rguenther@suse.de>
10315 PR middle-end/80539
10316 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10317 being in loop-closed SSA form conservatively.
10318 (chrec_fold_multiply_poly_poly): Likewise.
10320 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10322 PR middle-end/79665
10323 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10324 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10326 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10329 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10330 (aarch64_function_arg_alignment): Return unsigned int again, but still
10331 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10332 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10333 Don't emit -Wpsabi note.
10334 (aarch64_function_arg_boundary): Likewise.
10335 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10338 2017-04-26 Nathan Sidwell <nathan@acm.org>
10340 * tree.h (crc32_unsigned_n): Declare.
10341 (crc32_unsigned, crc32_unsigned): Make inline.
10342 * tree.c (crc32_unsigned_bits): Replace with ...
10343 (crc32_unsigned_n): ... this.
10344 (crc32_unsigned, crc32_byte): Remove.
10345 (crc32_string): Remove unnecessary braces.
10347 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10349 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10350 * ipa-inline-analysis.c (MAX_TIME): Remove.
10351 (account_size_time): Use sreal for time.
10352 (dump_inline_summary): Update.
10353 (estimate_function_body_sizes): Update.
10354 (estimate_edge_size_and_time): Update.
10355 (estimate_calls_size_and_time): Update.
10356 (estimate_node_size_and_time): Update.
10357 (inline_merge_summary): Update.
10358 (inline_update_overall_summary): Update.
10359 (estimate_time_after_inlining): Update.
10360 (inline_read_section): Update.
10361 (inline_write_summary): Update.
10362 * ipa-inline.c (compute_uninlined_call_time): Update.
10363 (compute_inlined_call_time): Update.
10364 (recursive_inlining): Update.
10365 (inline_small_functions): Update.
10366 (dump_overall_stats): Update.
10367 * ipa-inline.h: Include sreal.h.
10368 (size_time_entry): Turn time to sreal.
10369 (inline_summary): Turn self_time nad time to sreal.
10371 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10373 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10375 (sreal::stream_out, sreal::stream_in): New.
10376 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10378 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10380 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10383 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10386 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10388 (dimode_scalar_chain::compute_convert_gain): Ditto.
10389 (dimode_scalar_chain::make_vector_copies): Ditto.
10390 (dimode_scalar_chain::convert_reg): Ditto.
10391 (dimode_scalar_chain::convert_insn): Ditto.
10392 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10393 (VI248_AVX512BW_1): New mode iterator.
10394 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10395 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10398 2017-04-25 Martin Sebor <msebor@redhat.com>
10400 PR tree-optimization/80497
10401 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10402 constants are representable in HOST_WIDE_INT.
10403 (parse_directive): Ditto.
10405 2017-04-25 Martin Sebor <msebor@redhat.com>
10408 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10409 (new_zero_array): Adjust signature.
10410 (dom_info::dom_init): Used unsigned rather that size_t.
10411 (dom_info::dom_info): Same.
10413 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10414 Jakub Jelinek <jakub@redhat.com>
10417 * config/arm/arm.c: Include gimple.h.
10418 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10419 returns negative, increment ncrn only if it returned positive.
10420 (arm_needs_doubleword_align): Return int instead of bool,
10421 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10422 members, but if there is any such non-FIELD_DECL
10423 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10424 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10425 returns negative, increment nregs only if it returned positive.
10426 (arm_setup_incoming_varargs): Likewise.
10427 (arm_function_arg_boundary): Emit -Wpsabi note if
10428 arm_needs_doubleword_align returns negative, return
10429 DOUBLEWORD_ALIGNMENT only if it returned positive.
10431 2017-04-25 Marek Polacek <polacek@redhat.com>
10434 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10435 first argument to type.
10437 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10440 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10441 type checks to test for compatibility instead of equality.
10443 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10444 Jakub Jelinek <jakub@redhat.com>
10447 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10449 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10450 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10451 the alignment computation, but return their maximum in warn_alignment.
10452 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10453 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10455 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10456 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10459 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10461 * config/arc/simdext.md (dmpyh): Fix typo.
10463 2017-04-25 Richard Biener <rguenther@suse.de>
10465 PR tree-optimization/80492
10466 * alias.c (compare_base_decls): Handle registers with asm
10467 specification conservatively.
10468 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10469 compare_base_decls returning dont-know properly.
10471 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10473 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10474 (legitimate_offset_address_p): New function.
10475 (arc_legitimate_address_p): Use above function.
10477 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10479 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10481 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10483 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10484 ACCH registers whenever they are available.
10486 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10488 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10489 double regs fix when not used.
10491 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10493 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10495 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10496 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10498 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10500 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10501 when emitting short ADD instructions.
10503 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10505 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10507 (cmpsi_cc_c_insn): Likewise.
10508 (cbranchsi4_scratch): Compute proper instruction length using
10509 compact_hreg_operand.
10510 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10512 2017-04-25 Richard Biener <rguenther@suse.de>
10514 PR middle-end/80509
10515 * passes.c (pass_manager::pass_manager): Initialize
10516 m_name_to_pass_map.
10518 2017-04-25 Richard Biener <rguenther@suse.de>
10520 PR tree-optimization/79201
10521 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10523 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10526 * config/s390/vector.md: Split MEM->GPR vector moves for
10527 non-s_operand addresses.
10529 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10532 * config/s390/predicates.md (reload_const_wide_int_operand): New
10534 * config/s390/s390.md ("movti"): Remove d/P alternative.
10535 ("movti_bigconst"): New pattern definition.
10537 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10540 * s390-protos.h (s390_expand_cs_hqi): Removed.
10541 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10542 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10543 modes as well as CCZ1mode and CCZmode.
10544 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10545 signature of s390_emit_compare_and_swap.
10546 (s390_expand_cs_hqi): Likewise, make static.
10547 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10548 compare-and-swap, in some cases.
10549 (s390_expand_cs): Wrapper function.
10550 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10552 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10553 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10554 patterns for small and large integers. Forbid symref memory operands.
10555 Move expander to s390.c. Require cc register.
10556 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10557 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10558 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10559 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10560 symref memory operands. Remove CC mode and call s390_match_ccmode
10562 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10564 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10566 * config/s390/s390.md (define_peephole2): New peephole to help
10567 combining the load-and-test pattern with volatile memory.
10569 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10571 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10572 with CCZmode for TARGET_Z196.
10574 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10576 PR rtl-optimization/80501
10577 * combine.c (make_compound_operation_int): Set subreg_code to SET
10578 even for AND with mask of the sign bit of mode.
10580 PR rtl-optimization/80500
10581 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10582 sum's initial value.
10584 2017-04-25 Julian Brown <julian@codesourcery.com>
10585 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10587 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10589 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10591 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10593 2017-04-25 Julian Brown <julian@codesourcery.com>
10594 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10596 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10597 (thunderx2t99_sha): New Reservation.
10599 2017-04-25 Julian Brown <julian@codesourcery.com>
10600 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10602 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10603 type for 1-element load.
10605 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10607 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10609 2017-04-24 Martin Jambor <mjambor@suse.cz>
10611 PR tree-optimization/80293
10612 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10613 char arrays not totally scalarizable if it is false.
10614 (analyze_all_variable_accesses): Pass correct value in the new
10615 parameter. Add a statistics counter.
10617 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10619 PR middle-end/79931
10620 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10622 2017-04-24 Richard Biener <rguenther@suse.de>
10624 PR tree-optimization/80494
10625 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10626 out for complex types.
10628 2017-04-24 Richard Biener <rguenther@suse.de>
10630 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10631 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10632 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10633 (DFS): Adjust and never fail.
10634 (sccvn_dom_walker::fail): Remove.
10635 (sccvn_dom_walker::before_dom_children): Adjust.
10636 (run_scc_vn): Likewise and never fail.
10637 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10638 (pass_fre::execute): Likewise.
10640 2017-04-24 Richard Biener <rguenther@suse.de>
10642 PR tree-optimization/79725
10643 * tree-ssa-sink.c (statement_sink_location): Return whether
10644 failure reason was zero uses. Move that check later.
10645 (sink_code_in_bb): Deal with zero uses by removing the stmt
10648 2017-04-24 Richard Biener <rguenther@suse.de>
10651 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10652 pointer-based references.
10654 2017-04-24 Richard Biener <rguenther@suse.de>
10657 * pass_manager.h (pass_manager::operator new): Remove.
10658 (pass_manager::operator delete): Likewise.
10659 * passes.c (pass_manager::operator new): Remove.
10660 (pass_manager::operator delete): Likewise.
10661 (pass_manager::pass_manager): Zero individual pass members.
10663 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10666 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10667 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10668 Check "XEXP (src, 1)" operand here.
10669 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10670 Check "XEXP (src, 1)" operand here.
10671 (dimode_scalar_chain::make_vector_copies): Detect count register
10672 of a shift instruction. Zero extend count register from QImode
10673 to DImode to satisfy vector shift pattern count operand predicate.
10674 Substitute vector shift count operand with a DImode copy.
10675 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10678 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10680 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10681 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10682 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10683 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10684 (UNSPEC_NOREX_MEM): Remove definition.
10686 2017-04-21 Richard Biener <rguenther@suse.de>
10688 PR tree-optimization/79547
10689 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10690 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10691 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10692 without any constraints.
10694 2017-04-21 Richard Biener <rguenther@suse.de>
10696 PR tree-optimization/78847
10697 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10699 2017-04-21 Richard Biener <rguenther@suse.de>
10701 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10702 (build_distinct_type_copy): Likewise.
10703 (build_variant_type_copy): Likewise.
10704 * tree.c (build_qualified_type): Pass down mem-stat info.
10705 (build_distinct_type_copy): Likewise.
10706 (build_variant_type_copy): Likewise.
10708 2017-04-21 Richard Biener <rguenther@suse.de>
10710 PR tree-optimization/80237
10711 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10713 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10714 for a simplified result.
10716 2016-04-21 Richard Biener <rguenther@suse.de>
10718 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10719 sth as strict as a simple_iv but a chrec without symbols and an
10720 operand defined in the loop we are peeling (and not some subloop).
10721 (propagate_constants_for_unrolling): Propagate all constants.
10723 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10726 * config/i386/i386.c (print_reg): Remove assert for disalowed
10727 regno values, call output_operand_lossage instead.
10729 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10732 * config/i386/constraints.md (Yc): New register constraint.
10733 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10734 Use Yc constraint for alternative 2 of operand 0. Remove
10735 preferred_for_speed attribute.
10737 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10739 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10740 lastprivate clauses in SIMT case.
10742 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10744 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10746 2017-04-20 Richard Biener <rguenther@suse.de>
10748 PR tree-optimization/57796
10749 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10750 as N scalar stores.
10751 (vect_model_load_cost): Cost gathers as N scalar loads.
10753 2017-04-20 Richard Biener <rguenther@suse.de>
10755 * ggc-page.c (ggc_allocated_p): Rename to ...
10756 (safe_lookup_page_table_entry): ... this and return the lookup
10758 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10760 2017-04-20 Richard Biener <rguenther@suse.de>
10762 PR tree-optimization/80453
10763 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
10764 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
10765 from the conditions.
10766 (vn_phi_eq): Pass them down.
10767 (vn_phi_lookup): Record them.
10768 (vn_phi_insert): Likewise.
10770 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
10772 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
10773 uninitialized variable warning to avoid buffer overrun.
10775 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10778 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
10779 is suppressed for '{ 0 }' in C.
10781 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10783 * BASE-VER: Set to 8.0.0.
10785 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
10787 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
10788 priority .init_array and .fini_array section with SECTION_NOTYPE
10791 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10793 PR middle-end/80423
10794 * tree.h (build_array_type): Add typeless_storage default argument.
10795 * tree.c (type_cache_hasher::equal): Also compare
10796 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
10797 (build_array_type): Add typeless_storage argument, set
10798 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
10800 (build_nonshared_array_type): Adjust build_array_type_1 caller.
10801 (build_array_type): Likewise. Add typeless_storage argument.
10803 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10804 Jakub Jelinek <jakub@redhat.com>
10806 PR tree-optimization/80426
10807 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
10808 operation on symbolic operands, also compute the overflow for the
10809 invariant part when the operation degenerates into a negation.
10811 2017-04-19 Jakub Jelinek <jakub@redhat.com>
10814 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
10815 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
10818 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
10820 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
10823 * config/avr/avr.c (tree.h): Include it.
10824 (cgraph.h): Include it.
10825 (avr_encode_section_info): Don't warn for uninitialized progmem
10826 variable if it's just an alias.
10828 2017-04-19 Richard Biener <rguenther@suse.de>
10831 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
10832 when needed by AutoPGO.
10834 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
10837 * doc/lto.texi: Remove an extra 'that'.
10839 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
10841 PR rtl-optimization/80429
10842 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
10843 are only used in debug insns.
10845 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10846 Vladimir Makarov <vmakarov@redhat.com>
10848 * config/sparc/predicates.md (input_operand): Add comment. Return
10849 true for any memory operand when LRA is in progress.
10850 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
10852 2017-04-18 Jeff Law <law@redhat.com>
10855 * mips.md ({return,simple_return}_internal): Do not overwrite
10858 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10860 PR tree-optimization/80443
10861 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
10862 instead of adding 1, subtract -1 and similarly instead of subtracting
10865 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
10867 PR rtl-optimization/80357
10868 * haifa-sched.c (tmp_bitmap): New variable.
10869 (model_recompute): Handle duplicate use records.
10870 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
10871 (free_global_sched_pressure_data): Free it.
10873 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
10876 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10877 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10878 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10879 instead of SYSTEM_HEADER_DIR.
10881 2017-04-18 Jeff Law <law@redhat.com>
10883 PR middle-end/80422
10884 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
10885 predecessors after walking up the insn chain.
10887 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10890 * dwarf2out.c (modified_type_die): Try harder not to emit internal
10891 sizetype type into debug info.
10893 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
10896 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
10897 unneeded test for TARGET_UPPER_REGS_SF.
10898 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
10900 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10903 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
10904 instead of gsi_after_labels.
10906 2017-04-18 Jeff Law <law@redhat.com>
10908 * regcprop.c (maybe_mode_change): Avoid creating copies of the
10912 2017-04-13 Jeff Law <law@redhat.com>
10913 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10914 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10916 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
10919 * config/avr/avr.c (intl.h): Include it.
10920 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
10922 2017-04-18 Martin Liska <mliska@suse.cz>
10924 PR gcov-profile/78783
10925 * gcov-tool.c (gcov_output_files): Validate that destination
10926 file is either removed by the tool or by a user.
10928 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10929 Guy Benyei <guybe@mellanox.com>
10931 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
10932 block, and do not negate it, the stored id is already negative.
10934 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10936 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
10938 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10941 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
10942 masks of options that should be turned off if the VSX vector
10943 options are turned off.
10944 (OTHER_P8_VECTOR_MASKS): Likewise.
10945 (OTHER_VSX_VECTOR_MASKS): Likewise.
10946 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
10947 rs6000_disable_incompatible_switches to validate no type switches
10949 (rs6000_incompatible_switch): New function to disallow turning on
10950 other vector options if -mno-vsx, -mno-power8-vector, or
10951 -mno-power9-vector are specified.
10953 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10955 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
10957 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10959 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
10960 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
10961 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
10962 (ARG_POINTER_CFA_OFFSET): Likewise.
10964 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10966 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
10967 conditions to take advantage of various optimizations.
10969 2017-04-13 Jeff Law <law@redhat.com>
10971 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10972 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10973 (zero_extendsidi2_dext): Likewise.
10975 2017-04-13 Jakub Jelinek <jakub@redhat.com>
10978 * fold-const.c (fold_ternary_loc): Revert
10979 use op0 instead of fold_convert_loc (loc, type, arg0) part of
10982 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
10984 PR rtl-optimization/80343
10985 * lra-remat.c (update_scratch_ops): Assign original hard reg to
10986 new scratch pseudo.
10988 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
10991 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
10992 to ubsan_encode_value.
10994 2017-04-13 Jeff Law <law@redhat.com>
10996 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
10997 appearing in DEBUG_INSNs.
10999 2017-04-13 Martin Liska <mliska@suse.cz>
11001 PR gcov-profile/80413
11002 * gcov-io.c (gcov_write_string): Copy to buffer just when
11003 allocated size is greater than zero.
11005 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11008 * dwarf2out.c (decls_for_scope): Ignore declarations of
11009 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11011 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11014 * ipa-visibility.c (non_local_p): Fix typos.
11015 (localize_node): When localizing symbol in same comdat group,
11016 dissolve the group only when we know external symbols are going
11018 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11020 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11022 PR tree-optimization/79390
11023 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11024 order does not result in usable sequence, retry with reversed operand
11030 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11031 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11032 op0 instead of fold_convert_loc (loc, type, arg0).
11034 2017-04-12 Jeff Law <law@redhat.com>
11036 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11037 has a delay slot in the generated code.
11039 * config/cris/cris.md (cris_preferred_reload_class): Return
11040 GENNONACR_REGS rather than GENERAL_REGS.
11042 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11045 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11046 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11047 signedness of the result type.
11049 2017-04-12 Richard Biener <rguenther@suse.de>
11050 Jeff Law <law@redhat.com>
11052 PR tree-optimization/80359
11053 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11054 trim stores to TARGET_MEM_REFs.
11056 2017-04-12 Richard Biener <rguenther@suse.de>
11058 PR tree-optimization/79390
11059 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11060 threading case even more.
11062 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11065 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11066 for quad_address_p for TImode, instead of just not indexed_address.
11068 2017-04-12 Richard Biener <rguenther@suse.de>
11069 Bernd Edlinger <bernd.edlinger@hotmail.de>
11071 PR middle-end/79671
11072 * alias.c (component_uses_parent_alias_set_from): Handle
11073 TYPE_TYPELESS_STORAGE.
11074 (get_alias_set): Likewise.
11075 * tree-core.h (tree_type_common): Add typeless_storage flag.
11076 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11077 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11078 for types containing members with TYPE_TYPELESS_STORAGE.
11079 (place_field): Likewise.
11080 (layout_type): Likewise for ARRAY_TYPE.
11081 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11082 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11083 TYPE_TYPELESS_STORAGE.
11084 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11086 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11089 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11090 first argument to type.
11092 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11096 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11097 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11098 (rs6000_expand_binop_builtin): Likewise.
11099 (rs6000_expand_ternop_builtin): Likewise; also add missing
11100 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11101 vshasigma built-ins.
11102 * doc/extend.texi: Document that vec_xxpermdi's third argument
11103 must be a constant.
11105 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11107 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11108 Use shift_const cost parameter when calculating gain of STV shifts.
11110 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11112 PR rtl-optimization/70478
11113 * lra-constraints.c (process_alt_operands): Check memory for
11114 disfavoring memory insn operand.
11116 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11118 PR middle-end/80100
11119 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11120 left shift in unsigned HOST_WIDE_INT type.
11122 PR rtl-optimization/80385
11123 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11124 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11127 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11128 if they have any depend clauses.
11130 2017-04-11 Martin Liska <mliska@suse.cz>
11133 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11134 * ipa-split.c (split_function): Create a local comdat symbol
11135 if caller is in a comdat group.
11137 2017-04-11 Martin Liska <mliska@suse.cz>
11140 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11143 2017-04-11 Martin Sebor <msebor@redhat.com>
11145 PR middle-end/80364
11146 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11147 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11149 (directive::set_width, directive::set_precision, format_character):
11151 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11154 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11157 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11158 conflict, set target->arch_name instead of target->cpu_name.
11160 2017-04-11 Richard Biener <rguenther@suse.de>
11162 PR tree-optimization/80374
11163 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11164 build_zero_cst, remove fold_convertible_p check again.
11166 2017-04-11 Martin Liska <mliska@suse.cz>
11169 * ubsan.c (instrument_object_size): Do not instrument register
11172 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11175 * config/i386/i386-builtin-types.def
11176 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11177 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11178 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11179 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11180 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11181 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11182 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11183 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11184 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11185 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11186 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11187 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11188 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11189 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11190 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11191 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11192 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11193 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11194 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11195 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11196 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11197 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11198 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11199 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11200 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11201 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11202 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11203 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11204 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11205 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11206 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11207 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11208 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11209 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11210 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11211 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11212 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11213 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11214 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11215 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11216 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11217 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11218 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11219 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11220 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11221 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11223 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11224 flag to second_arg_count, handle 4 argument function type _COUNT
11225 aliases, handle second_arg_count on second argument rather than last.
11227 2017-04-10 Jeff Law <law@redhat.com>
11229 PR tree-optimization/80374
11230 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11231 record anything if we can not convert integer_zero_node to the
11234 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11237 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11238 Enhance special handling given to the TARGET_P9_MINMAX option in
11239 relation to certain other options.
11241 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11243 PR tree-optimization/80153
11244 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11245 remove POINTER_PLUS_EXPR's base part directly, rather than through
11248 2017-04-10 Richard Biener <rguenther@suse.de>
11249 Bin Cheng <bin.cheng@arm.com>
11251 PR tree-optimization/80153
11252 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11253 the first element of pointer type aff_tree. Build result expr in
11255 (add_elt_to_tree): Convert to type unconditionally. Remove other
11256 fold_convert calls.
11257 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11258 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11260 2017-04-10 Richard Biener <rguenther@suse.de>
11262 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11265 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11267 PR rtl-optimization/70478
11268 * lra-constraints.c (curr_small_class_check): New.
11269 (update_and_check_small_class_inputs): New.
11270 (process_alt_operands): Update curr_small_class_check. Disfavor
11271 alternative insn memory operands. Check available regs for small
11274 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11277 * config/mips/mips.opt (-mvirt): Update description.
11278 * doc/invoke.texi (-mvirt): Likewise.
11280 2017-04-10 Richard Biener <rguenther@suse.de>
11282 PR middle-end/80362
11283 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11284 looking for NEGATE_EXPR in -A / -B to A / B folding.
11286 2017-04-10 Martin Liska <mliska@suse.cz>
11288 PR gcov-profile/80224
11289 * gcov.c (print_usage): Fix usage string.
11290 (get_gcov_intermediate_filename): Remove.
11291 (output_gcov_file): Use both for normal and intermediate format.
11292 (generate_results): Do not initialize special file for
11293 intermediate format.
11295 2017-04-10 Richard Biener <rguenther@suse.de>
11297 PR tree-optimization/80304
11298 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11301 2017-04-10 Nathan Sidwell <nathan@acm.org>
11304 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11305 (rs6000_init_builtins): Use it.
11307 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11309 * config/arm/arm.md (<mrc>): Add mode to SET source.
11310 (<mrrc>): Likewise.
11312 2017-04-10 Richard Biener <rguenther@suse.de>
11314 PR middle-end/80344
11315 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11317 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11320 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11321 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11322 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11323 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11324 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11325 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11326 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11327 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11328 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11329 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11330 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11331 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11332 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11333 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11334 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11335 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11336 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11337 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11338 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11339 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11340 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11341 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11342 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11344 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11346 PR rtl-optimization/70478
11347 * lra-constraints.c: Reverse the last patch.
11349 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11351 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11352 Add comment for WCHAR_T.
11354 2017-04-08 Martin Liska <mliska@suse.cz>
11357 2017-04-07 Martin Liska <mliska@suse.cz>
11360 * ipa-split.c (split_function): Add function part to a same comdat
11363 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11366 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11368 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11370 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11371 (vec_pairing): ...to this.
11372 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11373 (rs6000_sched_init): Adjust for name change.
11374 (struct rs6000_sched_context): Likewise.
11375 (rs6000_init_sched_context): Likewise.
11376 (rs6000_set_sched_context): Likewise.
11378 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11384 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11386 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11387 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11388 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11390 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11392 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11394 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11396 PR rtl-optimization/70703
11397 * ira-color.c (update_conflict_hard_regno_costs): Use
11398 int64_t instead of HOST_WIDE_INT.
11400 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11402 PR rtl-optimization/70478
11403 * lra-constraints.c (process_alt_operands): Disfavor alternative
11404 insn memory operands.
11406 2017-04-07 Jeff Law <law@redhat.com>
11408 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11409 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11411 2017-04-07 Martin Liska <mliska@suse.cz>
11414 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11415 Show error message instead of an ICE.
11417 2017-04-07 Martin Liska <mliska@suse.cz>
11420 * ipa-split.c (split_function): Add function part to a same comdat
11423 2017-04-07 Richard Biener <rguenther@suse.de>
11425 PR middle-end/80341
11426 * tree.c (get_unwidened): Also handle ! for_type case for
11428 * convert.c (do_narrow): Split out from ...
11429 (convert_to_integer_1): ... here. Do not pass final truncation
11430 type to get_unwidened for TRUNC_DIV_EXPR.
11432 2017-04-07 Richard Biener <rguenther@suse.de>
11434 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11436 (aff_combination_const): Adjust.
11437 (aff_combination_scale): Likewise.
11438 (aff_combination_add_elt): Likewise.
11439 (aff_combination_add_cst): Likewise.
11440 (aff_combination_convert): Likewise.
11441 (add_elt_to_tree): Likewise. Remove unused argument.
11442 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11444 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11446 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11448 * config/arm/arm.c (arm_default_short_enums): Use
11449 ARM_DEFAULT_SHORT_ENUMS.
11450 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11452 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11455 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11456 members with redundant out-of-class redeclaration.
11458 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11461 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11462 * config/i386/i386.md (*zero_extendsidi2):
11463 Add (?*x,*x) and (?*v,*v) alternatives.
11465 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11468 * config/i386/i386.c (ix86_expand_builtin)
11469 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11470 mode from insn data. Convert operands to insn operand mode.
11471 Copy operands that don't satisfy insn predicate to a register.
11473 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11475 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11478 2017-04-06 Richard Biener <rguenther@suse.de>
11480 PR tree-optimization/80334
11481 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11482 preserve alignment of accesses.
11484 2017-04-06 Richard Biener <rguenther@suse.de>
11486 PR tree-optimization/80262
11487 * tree-sra.c (build_ref_for_offset): Preserve address-space
11489 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11490 Drop useless address-space information on MEM_REF offsets.
11492 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11494 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11496 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11498 PR rtl-optimization/70703
11499 * ira-color.c (update_conflict_hard_regno_costs): Use
11500 HOST_WIDE_INT instead of long.
11502 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11505 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11506 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11508 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11509 for x86_64 target. Handle -m3dnowa option.
11511 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11513 PR rtl-optimization/70703
11514 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11515 (update_conflict_hard_regno_costs): Use long instead of unsigned
11516 arithmetic for cost calculation.
11518 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11519 Bernd Edlinger <bernd.edlinger@hotmail.de>
11522 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11525 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11528 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11529 ptr_mode with Pmode throughout.
11530 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11531 into probe_stack_range and use DImode.
11533 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11536 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11537 call_eh_return is true.
11539 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11541 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11542 Initialize last_match_fntype_index.
11544 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11547 * tree-nvr.c: Include internal-fn.h.
11548 (pass_return_slot::execute): Ignore internal calls without
11551 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11552 Richard Biener <rguenther@suse.de>
11555 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11556 captures used multiple times, except for the last use.
11557 * generic-match-head.c: Include gimplify.h.
11559 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11561 PR tree-optimization/79390
11562 * target.h (struct noce_if_info): Declare.
11563 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11564 * target.def (noce_conversion_profitable_p): New target hook.
11565 * ifcvt.h (struct noce_if_info): New type, moved from ...
11566 * ifcvt.c (struct noce_if_info): ... here.
11567 (noce_conversion_profitable_p): Renamed to ...
11568 (default_noce_conversion_profitable_p): ... this. No longer
11570 (noce_try_store_flag_constants, noce_try_addcc,
11571 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11572 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11573 instead of noce_conversion_profitable_p.
11574 * config/i386/i386.c: Include ifcvt.h.
11575 (ix86_option_override_internal): Don't override
11576 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11577 (ix86_noce_conversion_profitable_p): New function.
11578 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11579 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11580 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11581 * doc/tm.texi: Regenerated.
11583 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11585 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11588 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11591 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11592 instructions for small multiply cores.
11594 2017-04-04 Jeff Law <law@redhat.com>
11596 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11598 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11599 that are not set by the loop over the elements.
11601 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11604 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11605 int mode, convert_modes it to mode as unsigned, otherwise use
11606 lowpart_subreg to mode rather than SImode.
11607 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11608 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11609 Use DImode instead of SImode for the shift count operand.
11610 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11613 2017-04-04 Richard Biener <rguenther@suse.de>
11615 PR middle-end/80281
11616 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11617 arithmetic done for the negate or the plus. Simplify.
11618 (A - (-B) -> A + B): Likewise.
11619 * fold-const.c (split_tree): Make sure to not negate pointers.
11621 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11623 PR rtl-optimization/60818
11624 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11625 a compare of comparisons with the thing compared if this results
11626 in a different machine mode.
11628 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11630 * alias.c (base_alias_check): Fix typo in comment.
11631 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11632 * cgraphunit.c (symbol_table::compile): Likewise.
11633 * collect2.c (maybe_run_lto_and_relink): Likewise.
11634 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11635 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11636 * config/avr/avr.c (avr_map_op_t): Likewise.
11637 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11638 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11639 * config/epiphany/epiphany.md (movcc): Likewise.
11640 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11641 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11643 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11644 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11645 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11646 * config/sh/sh.c (sh_rtx_costs): Likewise.
11647 * fold-const.c (fold_truth_andor): Likewise.
11648 * genautomata.c (collapse_flag): Likewise.
11649 * gengtype.h (struct type::u::s): Likewise.
11650 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11651 * input.c (FORMAT_AMOUNT): Likewise.
11652 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11653 (known_aggs_to_agg_replacement_list): Likewise.
11654 * ipa-inline-analysis.c: Likewise.
11655 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11656 * ipa-polymorphic-call.c
11657 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11658 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11659 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11661 * modulo-sched.c (apply_reg_moves): Likewise.
11662 * omp-expand.c (build_omp_regions_1): Likewise.
11663 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11664 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11665 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11666 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11667 * value-prof.c: Likewise.
11668 * var-tracking.c (val_reset): Likewise.
11670 2017-04-03 Richard Biener <rguenther@suse.de>
11672 PR tree-optimization/80275
11673 * fold-const.c (split_address_to_core_and_offset): Handle
11676 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11678 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11679 descriptors is at least equal to that of functions.
11681 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11683 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11685 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11688 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11689 (mov<IMOD4:mode>): New expander.
11690 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11692 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11694 PR rtl-optimization/79405
11695 * fwprop.c (propagations_left): New variable.
11696 (forward_propagate_into): Decrement it.
11697 (fwprop_init): Initialize it.
11698 (fw_prop): If the variable has reached zero, stop propagating.
11699 (fwprop_addr): Ditto.
11701 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11704 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11705 a FUNCTION_DECL, pass it as decl instead of origin to
11708 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11710 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11713 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11716 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11717 TARGET_VSX_SMALL_INTEGER.
11719 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11721 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11722 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11724 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11726 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11727 extraction from odd-numbered MSA register.
11729 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11731 PR middle-end/80173
11732 * expmed.c (store_bit_field_1): Don't attempt to create
11733 a word subreg out of hard registers wider than word if they
11734 have HARD_REGNO_NREGS of 1 for their mode.
11736 PR middle-end/80163
11737 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11738 conversions to integer types wider than word and pointer.
11741 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11742 (rtx_equal_for_cselib_p): Pass 0 to it.
11743 * cselib.c (cselib_hasher::equal): Likewise.
11744 (rtx_equal_for_cselib_1): Add depth argument. If depth
11745 is 128, don't look up VALUE locs and punt. Increment
11746 depth in recursive calls when walking VALUE locs.
11748 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11750 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11751 (make_gcov_file_name): Use the canonical path name for generating
11753 (read_line): Fix handling of files with ascii null bytes.
11755 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11757 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11758 to initialise a vector register instead
11759 of using a const_int.
11761 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11763 PR translation/80189
11764 * gimplify.c (omp_default_clause): Use %qs instead of %s in
11765 diagnostic messages.
11767 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
11770 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
11771 (dfp_diex_<mode>): Update mode of operand 1.
11772 * doc/extend.texi (dxex, dxexq): Document change to return type.
11773 (diex, diexq): Document change to argument type.
11775 2017-03-30 Martin Jambor <mjambor@suse.cz>
11778 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
11779 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
11780 it reflects the signature changes performed at the callee side.
11781 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
11782 to cgraph_build_function_type_skip_args.
11783 (build_function_decl_skip_args): Adjust call to the above function.
11785 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11788 * config/i386/sse.md
11789 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
11790 register as dest whenever it is a MEM not rtx_equal_p to the
11791 corresponding dup operand, and when forcing into reg move the
11792 reg into the memory afterwards.
11793 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
11794 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
11795 for the force_reg mode.
11796 (avx512vl_vextractf128<mode>): Use register as dest either
11797 always when a MEM, or when it is a MEM not rtx_equal_p to the
11798 corresponding dup operand, or even not when it is a CONST_VECTOR
11799 depending on the mode and lo vs. hi.
11800 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
11802 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
11803 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
11804 Likewise. Require that operands[2] is even.
11805 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
11806 Remove extraneous parens. Require that operands[2] is a multiple
11808 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
11809 operands[0] is a MEM if <mask_applied>, the predicates/constraints
11810 disallow memory then.
11812 2017-03-30 Richard Biener <rguenther@suse.de>
11814 PR tree-optimization/77498
11815 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
11816 to non-constants over backedges.
11818 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
11820 PR rtl-optimization/80233
11821 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
11822 as last_combined_insn. Do not test for BARRIER_P separately.
11824 2017-03-29 Andreas Schwab <schwab@suse.de>
11827 * calls.c (prepare_call_address): Convert funexp to Pmode before
11828 copying to temp reg.
11830 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11832 PR tree-optimization/80158
11833 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
11834 Handle possible future case of more than one alternate
11836 (replace_rhs_if_not_dup): Likewise.
11837 (replace_one_candidate): Likewise.
11839 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
11841 PR rtl-optimization/80193
11842 * ira.c (ira): Do not check allocation for LRA.
11844 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
11846 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
11847 (nvptx_output_simt_exit): Declare.
11848 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
11849 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
11850 (init_softstack_frame): Move initialization of crtl->is_leaf to...
11851 (nvptx_declare_function_name): ...here. Emit declaration of local
11852 memory space buffer for omp_simt_enter insn.
11853 (nvptx_output_unisimt_switch): New.
11854 (nvptx_output_softstack_switch): New.
11855 (nvptx_output_simt_enter): New.
11856 (nvptx_output_simt_exit): New.
11857 * config/nvptx/nvptx.h (struct machine_function): New fields
11858 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
11859 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
11860 (UNSPECV_SIMT_EXIT): Ditto.
11861 (omp_simt_enter_insn): New insn.
11862 (omp_simt_enter): New expansion.
11863 (omp_simt_exit): New insn.
11864 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
11866 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
11867 (expand_GOMP_SIMT_ENTER_ALLOC): New.
11868 (expand_GOMP_SIMT_EXIT): New.
11869 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
11870 (GOMP_SIMT_ENTER_ALLOC): Ditto.
11871 (GOMP_SIMT_EXIT): Ditto.
11872 * target-insns.def (omp_simt_enter): New insn.
11873 (omp_simt_exit): Ditto.
11874 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
11876 (lower_rec_simd_input_clauses): Implement SIMT privatization.
11877 (lower_rec_input_clauses): Likewise.
11878 (lower_lastprivate_clauses): Handle SIMT privatization.
11880 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
11881 (ompdevlow_adjust_simt_enter): New.
11882 (find_simtpriv_var_op): New.
11883 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
11884 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
11886 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
11887 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
11888 (copy_decl_for_dup_finish): Ditto.
11890 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
11892 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
11895 * config/i386/i386.c (ix86_option_override_internal): Always
11896 allow -mpreferred-stack-boundary=3 for 64-bit targets.
11898 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11900 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
11902 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11904 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
11905 mark new edge's irreducible flag accordign to it.
11906 (vect_do_peeling): Check loop preheader edge's irreducible flag
11907 and pass it to function slpeel_add_loop_guard.
11909 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
11911 PR tree-optimization/80218
11912 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
11913 Update block frequencies and counts.
11915 2017-03-28 Richard Biener <rguenther@suse.de>
11917 PR tree-optimization/78644
11918 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
11919 of a simplification result we may not use it at all.
11921 2017-03-28 Richard Biener <rguenther@suse.de>
11924 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
11925 without arguments, generate default definition of a SSA name.
11927 2017-03-28 Richard Biener <rguenther@suse.de>
11929 PR middle-end/80222
11930 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
11931 TYPE_REF_CAN_ALIAS_ALL references.
11932 * fold-const.c (fold_indirect_ref_1): Likewise.
11934 2017-03-28 Martin Liska <mliska@suse.cz>
11937 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
11938 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
11940 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11941 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
11943 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
11944 (EXTRA_SPECS): Define.
11945 (SUBTARGET_EXTRA_SPECS): Likewise.
11946 (SUBTARGET_CPP_SPEC): Likewise.
11947 * config/arc/elf.h (EXTRA_SPECS): Renamed to
11948 SUBTARGET_EXTRA_SPECS.
11949 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
11951 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11953 * config/arc/simdext.md (vst64_insn): Update pattern.
11954 (vld32wh_insn): Likewise.
11955 (vld32wl_insn): Likewise.
11956 (vld64_insn): Likewise.
11957 (vld32_insn): Likewise.
11959 2017-03-28 Marek Polacek <polacek@redhat.com>
11962 * fold-const.c (fold_comparison): Use protected_set_expr_location
11963 instead of SET_EXPR_LOCATION.
11965 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
11967 * tree.c (add_expr): Avoid name lookup warning.
11969 2017-03-27 Jeff Law <law@redhat.com>
11971 PR tree-optimization/80216
11972 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
11973 function name. Limit recursion depth.
11974 (record_temporary_equivalences): Corresponding changes.
11976 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
11978 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
11981 2017-03-27 Jakub Jelinek <jakub@redhat.com>
11984 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
11986 * cfgcleanup.c (reg_note_cfa_p): New array.
11987 (insns_have_identical_cfa_notes): New function.
11988 (old_insns_match_p): Don't cross-jump in between /f
11989 and non-/f instructions. If both i1 and i2 are frame related,
11990 verify all CFA notes, their order and content.
11992 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
11995 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
11996 HImode and SImode with zero extend to DImode to one insn.
11997 (bswap<mode>2_extenddi): Likewise.
11998 (bswapsi2_extenddi): Likewise.
11999 (bswaphi2_extendsi): Likewise.
12000 (bswaphi2): Combine bswap HImode and SImode into one insn.
12001 Separate memory insns from swapping register.
12002 (bswapsi2): Likewise.
12003 (bswap<mode>2): Likewise.
12004 (bswaphi2_internal): Delete, no longer used.
12005 (bswapsi2_internal): Likewise.
12006 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12007 store, and gpr<-gpr swap insns.
12008 (bswap<mode>2_store): Likewise.
12009 (bswaphi2_reg): Register only splitter, combine with the splitter.
12010 (bswaphi2 splitter): Likewise.
12011 (bswapsi2_reg): Likewise.
12012 (bswapsi2 splitter): Likewise.
12013 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12014 the insns into load, store, and register/register insns.
12015 (bswapdi2_ldbrx): Likewise.
12016 (bswapdi2_load): Likewise.
12017 (bswapdi2_store): Likewise.
12018 (bswapdi2_reg): Likewise.
12020 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12022 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12023 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12025 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12028 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12030 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12031 special handling for target option conflicts between dform
12032 options (-mpower9-dform, -mpower9-dform-vector,
12033 -mpower9-dform-scalar) and -mno-direct-move.
12035 2017-03-27 Richard Biener <rguenther@suse.de>
12037 PR tree-optimization/80181
12038 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12040 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12042 * config/arc/predicates.md (move_double_src_operand): Replace the
12043 call to move_double_src_operand with a call to address_operand.
12045 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12047 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12048 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12049 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12051 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12053 * config/arc/predicates.md (long_immediate_loadstore_operand):
12054 Consider scaled addresses cases.
12056 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12058 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12059 restored when in interrupt.
12060 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12061 doesn't have delay slot.
12063 2017-03-27 Richard Biener <rguenther@suse.de>
12066 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12067 inlined thunk clones.
12069 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12072 * asan.c (instrument_derefs): Copy over last operand from
12073 original COMPONENT_REF to the new COMPONENT_REF with
12074 DECL_BIT_FIELD_REPRESENTATIVE.
12075 * ubsan.c (instrument_object_size): Likewise.
12077 2017-03-27 Richard Biener <rguenther@suse.de>
12079 PR tree-optimization/80170
12080 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12081 sure DR/SCEV didnt fold in constants we do not see when looking
12082 at the reference base alignment.
12084 2017-03-27 Richard Biener <rguenther@suse.de>
12086 PR middle-end/80171
12087 * gimple-fold.c (fold_ctor_reference): Properly guard against
12088 NULL return value from canonicalize_constructor_val.
12090 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12093 * config/i386/i386.c (ix86_expand_builtin)
12094 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12095 flags reg setting and flags reg using instructions.
12096 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12097 clobbering instructions to zero extend op2.
12099 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12101 * doc/install.texi (Configuration) <--with-aix-soname>:
12102 Update link to AIX ld.
12104 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12106 PR rtl-optimization/80160
12107 PR rtl-optimization/80159
12108 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12109 reg_alternate_class into account.
12111 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12114 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12115 to consider in curr_insn_transform.
12117 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12119 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12120 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12121 and emit_mode_inner.
12123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12125 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12126 argument to the overloaded builtin variants. Use the new flag to
12127 deprecate certain builtin variants.
12128 * config/s390/s390-builtin-types.def: Add new builtin types.
12129 * config/s390/s390-builtins.h: Support new flags field for
12130 overloaded builtins.
12131 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12132 (s390_macro_to_expand): Enable vector float data type.
12133 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12134 builtins by incrementing the __VEC__ version number.
12135 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12137 (s390_resolve_overloaded_builtin): Emit error messages depending
12138 on the builtin flags.
12139 * config/s390/s390.c (s390_expand_builtin): Support additional
12140 flags argument. Change error message to match the messages
12141 emitted in s390-c.c.
12142 * config/s390/s390.md: New UNSPEC_* constants.
12143 (op_type): Add new instruction types.
12144 * config/s390/vecintrin.h: Add new builtins and test data class
12146 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12147 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12148 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12149 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12150 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12151 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12153 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12154 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12155 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12156 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12158 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12159 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12160 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12161 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12163 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12164 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12165 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12166 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12167 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12168 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12169 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12171 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12172 ("vec_scatter_element<V_HW_4:mode>_DI")
12173 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12174 ("vec_fpint<mode>", "vflls")
12175 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12176 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12177 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12178 ("*vec_cmphe<mode>_cc"): ... these.
12180 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12181 mode constant instead of magic value.
12183 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12185 * config/s390/s390.c (s390_expand_vec_compare): Support other
12186 vector floating point modes than just V2DF.
12187 (s390_expand_vcond): Likewise.
12188 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12189 (s390_cannot_change_mode_class): Prevent mode changes between TF
12190 and V1TF in vector registers.
12191 * config/s390/s390.md (DF, SF): New mode attributes.
12192 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12193 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12194 SFmode support for VRs.
12195 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12197 (VFT, VF_HW): New mode iterators.
12198 (vw, sdx): New mode attributes.
12199 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12200 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12201 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12202 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12203 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12204 also the new vector floating point modes. Renaming to ...
12206 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12207 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12208 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12209 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12210 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12211 ("vec_unordered<mode>"): ... these.
12213 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12214 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12215 ("*vec_extendv2df"): New insn definitions.
12217 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12219 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12220 ("mulditi3_2", "*muldi3_sign"): New patterns.
12221 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12222 rename the pattern definition.
12224 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12226 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12228 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12230 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12232 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12233 instruction if possible.
12234 * config/s390/vector.md (vec_halfnumelts): New mode
12236 ("*vec_vllezlf<mode>"): New pattern.
12238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12240 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12241 ("popcountv4si2", "popcountv2di2"): Rename to ...
12242 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12243 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12245 ("popcount<mode>2_vxe"): New pattern.
12247 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12249 * common/config/s390/s390-common.c (processor_flags_table): Add
12251 * config.gcc: Add arch12.
12252 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12253 Default to arch12 for unknown CPU model numbers.
12254 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12255 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12256 PROCESSOR_max sanity check.
12257 * config/s390/s390-opts.h (enum processor_type): Add
12259 * config/s390/s390.c (processor_table): Add arch12.
12260 (s390_expand_builtin): Add check for B_VXE flag.
12261 (s390_issue_rate): Add PROCESSOR_ARCH12.
12262 (s390_get_sched_attrmask): Likewise.
12263 (s390_get_unit_mask): Likewise.
12264 (s390_sched_score): Enable z13 scheduling for arch12.
12265 (s390_sched_reorder): Likewise.
12266 (s390_sched_variable_issue): Likewise.
12267 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12269 (s390_tune_attr): Use z13 scheduling also for arch12.
12270 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12271 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12272 (TARGET_VXE_P): New macros.
12273 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12274 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12275 * config/s390/s390.opt: Add arch12 as processor_type.
12277 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12279 * config/s390/s390.md
12280 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12281 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12282 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12284 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12285 Rename expanders to ...
12287 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12288 ("fixuns_truncdddi2_emu"): ... these.
12290 ("fixuns_trunc<mode>si2_emu"): New expander.
12292 ("*fixuns_truncdfdi2_z13"): Rename to ...
12293 ("*fixuns_truncdfdi2_vx"): ... this.
12295 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12297 * config/s390/2964.md: Remove the single element vector compare
12298 instructions which are no longer used.
12299 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12301 (s390_canonicalize_comparison): Remove handling of DFmode
12303 (s390_expand_vec_compare_scalar): Remove function.
12304 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12305 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12307 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12309 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12311 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12313 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12314 will anyway by matched by mov<mode>_64dfp.
12316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12318 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12319 vlef/vstef. Add missing operand to vleif.
12321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12323 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12324 pair for all vector types with 64 bit elements.
12325 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12326 * config/s390/vector.md (V_HW_64): ... here.
12327 (V_128_NOSINGLE): New mode iterator.
12328 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12329 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12330 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12331 ("*vec_load_pairv2di"): Change to ...
12332 ("*vec_load_pair<mode>"): ... this one.
12334 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12336 * config/s390/constraints.md: Add comments.
12337 (jKK): Reject element sizes > 8 bytes.
12338 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12340 * config/s390/s390.md: Add the s_operand checks formerly in
12341 s390_split_ok_p to various splitters where they are still
12343 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12344 for 128 bit vectors. Plus two splitters.
12346 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12348 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12351 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12354 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12355 error if the boundary argument is not constant.
12357 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12359 PR rtl-optimization/80112
12360 * loop-doloop.c (doloop_condition_get): Don't check condition
12361 if cmp isn't SET with IF_THEN_ELSE src.
12363 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12365 PR tree-optimization/80158
12366 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12367 replacing a candidate statement, also replace it for the
12368 candidate's alternate interpretation.
12369 (replace_rhs_if_not_dup): Likewise.
12370 (replace_one_candidate): Likewise.
12372 2017-03-24 Richard Biener <rguenther@suse.de>
12374 PR tree-optimization/80167
12375 * graphite-isl-ast-to-gimple.c
12376 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12378 (translate_isl_ast_to_gimple::get_rename): Likewise.
12380 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12382 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12383 handling of certain combinations of target options, including the
12384 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12385 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12387 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12390 * config/arm/arm.md (*load_multiple): Add reload_completed to
12391 matching condition.
12393 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12394 Richard Biener <rguenth@suse.de>
12396 PR tree-optimization/79908
12397 PR tree-optimization/80136
12398 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12399 been cast away, gimplify_and_add suffices.
12401 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12403 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12405 2017-03-23 Richard Biener <rguenther@suse.de>
12407 PR tree-optimization/80032
12408 * gimplify.c (gimple_push_cleanup): Forced unconditional
12409 cleanups still have to go to the conditional_cleanups
12412 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12414 PR tree-optimization/80072
12415 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12417 (next_operand_entry_id): Change type to unsigned int.
12418 (sort_by_operand_rank): Make sure to return the right return value
12419 even if unsigned fields are bigger than INT_MAX.
12420 (struct oecount): Change cnt and id type to unsigned int.
12421 (oecount_hasher::equal): Formatting fix.
12422 (oecount_cmp): Make sure to return the right return value
12423 even if unsigned fields are bigger than INT_MAX.
12424 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12427 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12428 TREE_READONLY on result if writing it more than once.
12431 * doc/invoke.texi (-fsanitize=thread): Document that with
12432 -fnon-call-exceptions atomics are not able to throw
12436 * tsan.c: Include tree-eh.h.
12437 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12438 maybe_clean_or_replace_eh_stmt where needed.
12439 (instrument_memory_accesses): Add cfg_changed argument.
12440 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12441 if it returned true.
12442 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12444 PR rtl-optimization/63191
12445 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12446 wrapper function, moved the whole old content into ...
12447 (ix86_delegitimize_address_1): ... this. New inline function.
12448 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12449 true as last argument instead of ix86_delegitimize_address.
12451 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12453 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12454 cortexa57_branch_cost.
12456 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12458 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12460 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12463 * doc/md.texi (Constraints): Document wA constraint.
12464 * config/rs6000/constraints.md (wA): New.
12465 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12466 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12467 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12468 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12470 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12473 * gimplify.c (is_oacc_declared): New function.
12474 (oacc_default_clause): Use it to set default flags for acc declared
12475 variables inside parallel regions.
12476 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12477 declared variables.
12478 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12479 declare attribute to any decl as necessary.
12481 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12484 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12485 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12486 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12487 (arm_arch_lpae): This.
12488 * config/arm/arm.c (arm_arch7ve): Rename into ...
12489 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12490 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12493 2017-03-22 Martin Liska <mliska@suse.cz>
12496 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12497 error message instead of an ICE.
12499 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12501 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12503 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12505 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12507 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12510 2017-03-21 Martin Sebor <msebor@redhat.com>
12512 * doc/extend.texi: Use "cannot" instead of "can't."
12513 * doc/hostconfig.texi: Same.
12514 * doc/install.texi: Same.
12515 * doc/invoke.texi: Same.
12516 * doc/loop.texi: Same.
12517 * doc/md.texi: Same.
12518 * doc/objc.texi: Same.
12519 * doc/rtl.texi: Same.
12520 * doc/tm.texi: Same.
12521 * doc/tm.texi.in: Same.
12522 * doc/trouble.texi: Same.
12524 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12527 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12528 (collect_checksum_attributes): Set it.
12529 (die_checksum_ordered): Use it.
12531 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12533 PR tree-optimization/79908
12534 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12535 change: For a VA_ARG whose LHS has been cast away, use
12536 force_gimple_operand to construct the side effects.
12538 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12540 PR translation/80001
12541 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12542 more amenable to translation.
12543 (oacc_loop_auto_partitions): Likewise.
12545 2017-03-21 Marek Polacek <polacek@redhat.com>
12546 Martin Sebor <msebor@redhat.com>
12548 PR tree-optimization/80109
12549 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12550 on INTEGRAL_TYPE_P.
12552 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12553 Segher Boessenkool <segher@kernel.crashing.org>
12556 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12557 check reg_used_between_p between insn and one of succ or succ2
12558 depending on if succ is artificial insn not inserted into insn
12561 2017-03-21 Martin Liska <mliska@suse.cz>
12563 PR gcov-profile/80081
12564 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12565 * doc/gcc.texi: Include gcov-dump stuff.
12566 * doc/gcov-dump.texi: New file.
12568 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12570 PR rtl-optimization/79150
12571 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12572 conditional jump, if the jump is the last insn of the loop.
12574 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12575 Richard Biener <rguenth@suse.de>
12577 PR tree-optimization/79908
12578 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12579 been cast away, use force_gimple_operand to construct the side
12582 2017-03-21 Martin Liska <mliska@suse.cz>
12584 PR libfortran/79956
12585 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12588 2017-03-21 Brad Spengler <spender@grsecurity.net>
12591 * plugin.c (htab_hash_plugin): New function.
12592 (add_new_plugin): Use it and adjust.
12593 (parse_plugin_arg_opt): Adjust.
12594 (init_one_plugin): Likewise.
12596 2017-03-21 Richard Biener <rguenther@suse.de>
12598 PR tree-optimization/80032
12599 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12600 if set force the cleanup to happen unconditionally.
12601 (gimplify_target_expr): Push inserted clobbers with force_uncond
12602 to avoid them being removed by control-dependent DCE.
12604 2017-03-21 Richard Biener <rguenther@suse.de>
12606 PR tree-optimization/80122
12607 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12608 va_arg_pack_len when the inlined call stmt requires pack
12610 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12612 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12615 * tsan.c (instrument_builtin_call): If the memory model argument
12616 is not a constant, assume it is valid.
12619 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12622 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12624 PR rtl-optimization/79910
12625 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12626 if its dest is used by an insn before I2 (other than the combined
12627 insns themselves, which are properly handled already).
12629 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12632 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12634 * combine.c (record_used_regs): New static function.
12635 (try_combine): Handle situations where there is an additional
12636 instruction between I2 and I3 which needs to have a LOG_LINK
12640 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12642 * combine.c (try_combine): Delete redundant i1 test. Call
12643 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12645 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12648 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12649 alternatives 13/14.
12651 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12653 PR tree-optimization/80054
12654 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12655 the optimization if a PHI or any of its arguments is not dominated
12656 by the candidate's basis. Use gphi* rather than gimple* as
12658 (replace_profitable_candidates): Clean up a gimple* variable that
12659 should be a gphi* variable.
12661 2017-03-20 Martin Sebor <msebor@redhat.com>
12664 * doc/extend.texi (attribute constructor): Document present limitation.
12666 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12669 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12670 __POWER9_VECTOR__ #ifdef control, change template definition to
12671 use Power9-specific built-in function.
12672 (vec_any_eq): Likewise.
12673 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12674 to control outcomes from this test.
12675 (vector_ae_<mode>p): For VEC_F modes, likewise.
12677 2017-03-20 Ian Lance Taylor <iant@google.com>
12679 * config/i386/i386.c (ix86_function_regparm): Save an extra
12680 register for -fsplit-stack with DECL_STATIC_CHAIN.
12682 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12685 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12686 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12688 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12690 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12692 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12695 2017-03-20 Marek Polacek <polacek@redhat.com>
12698 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12700 2017-03-20 Richard Biener <rguenther@suse.de>
12702 PR tree-optimization/80113
12703 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12704 allocate extra SSA name for PHI def.
12705 (add_close_phis_to_outer_loops): Likewise.
12706 (add_close_phis_to_merge_points): Likewise.
12707 (copy_loop_close_phi_args): Likewise.
12708 (copy_cond_phi_nodes): Likewise.
12710 2017-03-20 Martin Liska <mliska@suse.cz>
12712 PR middle-end/79753
12713 * tree-chkp.c (chkp_build_returned_bound): Do not build
12714 returned bounds for a LHS that's not a BOUNDED_P type.
12716 2017-03-20 Martin Liska <mliska@suse.cz>
12720 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12721 COMPLEX_CST and VECTOR_CST.
12723 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12726 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12727 target operand. A new splitter adds the clobber statement in case
12728 the target operand is dead anyway.
12730 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12732 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12733 to age-old versions of binutils and glibc.
12735 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12737 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12739 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12741 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12743 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12745 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12746 requirement for binutils 2.13.
12748 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12750 * combine.c (try_combine): Delete redundant i1 test. Call
12751 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12753 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12755 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12756 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12758 <riscv64-*-elf>: Re-arrange section
12759 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12760 <riscv32-*-linux>: Likewise.
12761 <riscv64-*-elf>: Likewise
12762 <riscv64-*-linux>: Likewise.
12764 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
12767 * aarch64.opt(verbose-cost-dump): Fix typo.
12769 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
12772 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
12773 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
12775 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12777 * reload.c (find_reloads): When reloading a nonoffsettable address,
12778 use RELOAD_OTHER for it and its address reloads.
12780 PR rtl-optimization/79910
12781 * combine.c (record_used_regs): New static function.
12782 (try_combine): Handle situations where there is an additional
12783 instruction between I2 and I3 which needs to have a LOG_LINK
12786 2017-03-17 Jeff Law <law@redhat.com>
12788 PR tree-optimization/71437
12789 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
12790 conditional in the hash table first.
12791 (vrp_dom_walker::before_dom_children): Extract condition from
12792 ASSERT_EXPR. Record condition, its inverion and any implied
12793 conditions as well.
12795 2017-03-17 Marek Polacek <polacek@redhat.com>
12796 Markus Trippelsdorf <markus@trippelsdorf.de>
12798 PR tree-optimization/80079
12799 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
12802 2017-03-17 Richard Biener <rguenther@suse.de>
12804 PR middle-end/80075
12805 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
12806 Properly verify the LHS before the RHS possibly claims to be
12808 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
12811 2017-03-17 Martin Jambor <mjambor@suse.cz>
12813 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
12814 (List of -O2 options): Likewise.
12815 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
12818 2017-03-17 Tom de Vries <tom@codesourcery.com>
12820 * gcov-dump.c (print_usage): Print bug_report_url.
12822 2017-03-17 Richard Biener <rguenther@suse.de>
12824 PR middle-end/80050
12825 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
12826 (parser::peek): Likewise.
12828 2017-03-17 Richard Biener <rguenther@suse.de>
12830 PR tree-optimization/80048
12831 * sese.c (free_sese_info): Properly release rename_map and
12832 copied_bb_map elements.
12834 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
12836 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
12837 Add linked-list forward and backlinks. Insert on
12838 construction, remove on destruction.
12839 (class pass_store_merging): Add m_stores_head field.
12840 (pass_store_merging::terminate_and_process_all_chains):
12841 Iterate over m_stores_head list.
12842 (pass_store_merging::terminate_all_aliasing_chains):
12844 (pass_store_merging::execute): Check for debug stmts first.
12845 Push new chains onto the m_stores_head stack.
12847 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
12850 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
12851 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
12852 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
12854 2017-03-16 Jeff Law <law@redhat.com>
12856 PR tree-optimization/71437
12857 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
12858 member function. Implementation moved into after_dom_children
12859 member function and into the threader's thread_outgoing_edges
12861 (dom_opt_dom_walker::after_dom_children): Simplify by moving
12862 some code into new thread_outgoing_edges.
12863 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
12864 definition. Simplify marker handling (do it here). Assume we always
12865 have the available expression and the const/copies tables.
12866 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
12868 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
12869 * tree-vrp.c (equiv_stack): No longer file scoped.
12870 (vrp_dom_walker): New class.
12871 (vrp_dom_walker::before_dom_children): New member function.
12872 (vrp_dom_walker::after_dom_children): Likewise.
12873 (identify_jump_threads): Setup domwalker. Use it rather than
12874 walking edges in a random order by hand. Simplify setup/finalization.
12875 (finalize_jump_threads): Remove.
12876 (vrp_finalize): Do not call identify_jump_threads here.
12877 (execute_vrp): Do it here instead and call thread_through_all_blocks
12880 PR tree-optimization/71437
12881 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
12883 (simplify_stmt_for_jump_threading): Add basic_block argument. All
12885 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
12886 (dom_opt_dom_walker::thread_across_edge): Remove
12887 handle_dominating_asserts argument. All callers changed.
12888 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
12889 changes. Remove calls to lhs_of_dominating_assert. Other
12890 uses of handle_dominating_asserts turn into unconditional code
12891 (simplify_control_stmt_condition_1): Likewise.
12892 (simplify_control_stmt_condition): Likewise.
12893 (thread_through_normal_block, thread_across_edge): Likewise.
12894 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
12895 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
12896 object if it is not an SSA_NAME.
12897 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
12898 before calling into the VRP specific simplifiers.
12899 (identify_jump_threads): Remove handle_dominating_asserts
12902 2017-03-16 Jakub Jelinek <jakub@redhat.com>
12905 * tree-diagnostic.c (default_tree_printer): No longer static.
12906 * tree-diagnostic.h (default_tree_printer): New prototype.
12908 2017-03-16 Tamar Christina <tamar.christina@arm.com>
12910 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
12911 Change ins into fmov.
12913 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12915 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
12916 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
12917 Use h_con constraint for operand 1.
12918 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
12919 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
12921 2017-03-15 Jeff Law <law@redhat.com>
12923 PR tree-optimization/71437
12924 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
12925 (record_temporary_equivalences): Use it.
12927 PR tree-optimization/71437
12928 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
12929 tree-ssa-scopedtables.
12930 (lookup_avail_expr, build_and_record_new_cond): Likewise.
12931 (record_conditions, record_cond, vuse_eq): Likewise.
12932 (record_edge_info): Adjust to API tweak of record_conditions.
12933 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
12934 (record_temporary_equivalences, optimize_stmt): Likewise.
12935 (eliminate_redundant_computations): Likewise.
12936 (record_equivalences_from_stmt): Likewise.
12937 * tree-ssa-scopedtables.c: Include options.h and params.h.
12938 (vuse_eq): New function, moved from tree-ssa-dom.c
12939 (build_and_record_new_cond): Likewise.
12940 (record_conditions): Likewise. Accept vector of conditions rather
12941 than edge_equivalence structure for first argument.
12942 for the first argument.
12943 (avail_exprs_stack::lookup_avail_expr): New member function, moved
12944 from tree-ssa-dom.c.
12945 (avail_exprs_stack::record_cond): Likewise.
12946 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
12947 from tree-ssa-dom.c.
12948 (avail_exprs_stack): Add new member functions lookup_avail_expr
12950 (record_conditions): Declare.
12952 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
12955 * lra-constraints.c (process_alt_operands): Increase reject for
12956 reloading an input/output operand.
12958 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
12961 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
12962 insns to convert from signed/unsigned char/short to IEEE 128-bit
12964 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
12966 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
12969 * config/i386/i386.c (ix86_vector_duplicate_value): Create
12970 subreg of inner mode for values already in registers.
12972 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
12974 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
12975 iteration reg is used after the loop.
12977 2017-03-14 Martin Sebor <msebor@redhat.com>
12979 PR tree-optimization/79800
12980 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
12981 precision in negative-positive range.
12982 (format_floating): Call non-const overload with adjusted precision.
12984 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12987 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
12990 2017-03-14 Martin Sebor <msebor@redhat.com>
12992 PR middle-end/80020
12993 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
12994 * builtins.def (aligned_alloc): Use it.
12997 * Makefile.in (GTFILES): Add calls.c.
12998 * calls.c: Include "gt-calls.h".
13000 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13002 PR rtl-optimization/79728
13003 * regs.h (struct target_regs): New field
13004 x_contains_allocatable_regs_of_mode.
13005 (contains_allocatable_regs_of_mode): New macro.
13006 * reginfo.c (init_reg_sets_1): Initialize it, and change
13007 contains_reg_of_mode so it includes global regs as well.
13008 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13009 rather than contains_regs_of_mode.
13011 2017-03-14 Martin Liska <mliska@suse.cz>
13013 * doc/invoke.texi: Document options that can't be combined with
13014 -fcheck-pointer-bounds.
13016 2017-03-14 Martin Liska <mliska@suse.cz>
13018 PR middle-end/79831
13019 * doc/invoke.texi (-Wchkp): Document the option.
13021 2017-03-14 Martin Liska <mliska@suse.cz>
13023 * Makefile.in: Install gcov-dump.
13025 2017-03-14 Martin Liska <mliska@suse.cz>
13027 * multiple_target.c (expand_target_clones): Bail out for
13028 an invalid attribute.
13030 2017-03-14 Richard Biener <rguenther@suse.de>
13032 * alias.c (struct alias_set_entry): Pack properly.
13033 * cfgloop.h (struct loop): Likewise.
13034 * cse.c (struct set): Likewise.
13035 * ipa-utils.c (struct searchc_env): Likewise.
13036 * loop-invariant.c (struct invariant): Likewise.
13037 * lra-remat.c (struct cand): Likewise.
13038 * recog.c (struct change_t): Likewise.
13039 * rtl.h (struct address_info): Likewise.
13040 * symbol-summary.h (function_summary): Likewise.
13041 * tree-loop-distribution.c (struct partition): Likewise.
13042 * tree-object-size.c (struct object_size_info): Likewise.
13043 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13044 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13045 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13046 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13047 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13048 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13049 (struct _stmt_vec_info): Likewise.
13051 2017-03-14 Martin Liska <mliska@suse.cz>
13054 * multiple_target.c (create_dispatcher_calls): Check that
13055 a target can create a function dispatcher.
13057 2017-03-14 Martin Liska <mliska@suse.cz>
13060 * multiple_target.c (expand_target_clones): Drop local.local
13061 flag for default implementation.
13063 2017-03-14 Richard Biener <rguenther@suse.de>
13065 PR tree-optimization/80030
13066 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13068 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13070 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13071 gcc_fallthrough() instead of __attribute__((fallthrough));
13073 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13075 * doc/gcc.texi: Remove "up" link to (DIR).
13076 * doc/gccint.texi: Ditto.
13078 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13080 * doc/install.texi (Specific) <avr>: Remove reference to
13083 2017-03-13 Jeff Law <law@redhat.com>
13085 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13086 attribute rather than comments.
13088 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13089 match_scratch operand is highest.
13091 2017-03-13 Martin Liska <mliska@suse.cz>
13093 PR middle-end/78339
13094 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13095 is a CHKP clone, use original declaration.
13097 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13099 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13100 (arc_conditional_register_usage): Use a different allocation order
13101 when optimizing for size.
13102 * common/config/arc/arc-common.c (arc_option_optimization_table):
13103 Section anchors default on when optimizing for size.
13105 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13107 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13109 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13111 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13112 * config/arc/arc.md (cpu_facility): Add cd variant.
13113 (*movqi_insn): Add code density variant.
13114 (*movhi_insn): Likewise.
13115 (*movqi_insn): Likewise.
13116 (*addsi3_mixed): Likewise.
13117 (subsi3_insn): Likewise.
13119 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13121 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13123 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13125 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13126 expressions with MINUS and UNARY ops.
13128 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13131 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13133 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13134 between vec_select and vector argument.
13135 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13136 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13137 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13138 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13139 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13140 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13142 2017-03-13 Richard Biener <rguenther@suse.de>
13145 * params.def (vect-max-peeling-for-alignment): Fix typo.
13147 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13149 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13150 issue that only occurred with binutils below 2.18.
13152 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13154 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13155 refer to binutils 2.11/2.12 minimum.
13157 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13159 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13160 ftp.kernel.org and simplify binutils requirement.
13162 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13164 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13166 (Optimize Options): Ditto. Also remove redundancy.
13168 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13170 PR translation/79848
13171 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13173 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13174 to G_ to avoid double translation.
13176 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13178 PR translation/79923
13179 * auto-profile.c (get_combined_location): Convert leading
13180 character of diagnostics to lower case and remove trailing period.
13181 (read_profile): Likewise for various diagnostics.
13182 * config/arm/arm.c (arm_option_override): Remove trailing period
13183 from various diagnostics.
13184 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13185 (msp430_expand_delay_cycles): Likewise.
13187 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13190 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13191 full command-line argument, rather than just "str".
13192 (aarch64_validate_march): Likewise.
13193 (aarch64_validate_mtune): Likewise.
13195 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13197 PR rtl-optimization/78911
13198 * lra-assigns.c (must_not_spill_p): New function.
13199 (spill_for): Use it.
13201 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13203 PR tree-optimization/79981
13204 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13205 ATOMIC_COMPARE_EXCHANGE ifn result.
13206 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13207 IFN_ATOMIC_COMPARE_EXCHANGE.
13209 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13212 * opts.c (parse_sanitizer_options): Add missing question mark to
13213 "did you mean" message.
13215 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13217 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13219 (VMULEUH_UNS): Likewise.
13220 (VMULOUB_UNS): Likewise.
13221 (VMULOUH_UNS): Likewise.
13222 * config/rs6000/rs6000.c (builtin_function_type): Remove
13223 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13225 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13228 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13229 x with result of extra_parsing_for_operand_code_0.
13230 (function_reader::extra_parsing_for_operand_code_0): Convert
13231 return type from void to rtx, returning x. When reading
13232 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13233 larger size containing struct block_symbol.
13235 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13237 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13238 -mfloat128-hardware without -m64.
13240 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13243 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13244 entries to the case statement that marks unsigned arguments to
13245 overloaded functions.
13247 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13249 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13250 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13252 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13255 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13256 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13258 2017-03-10 Martin Liska <mliska@suse.cz>
13262 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13263 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13266 2017-03-10 Olivier Hainque <hainque@adacore.com>
13268 * tree-switch-conversion (array_value_type): Start by resetting
13269 candidate type to it's main variant.
13271 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13273 PR rtl-optimization/79909
13274 * combine.c (try_combine): Use simplify_replace_rtx on individual
13275 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13276 of the whole CALL_INSN_FUNCTION_USAGE.
13278 PR tree-optimization/79972
13279 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13280 get_range_info on SSA_NAMEs. Formatting fixes.
13282 2017-03-10 Richard Biener <rguenther@suse.de>
13283 Jakub Jelinek <jakub@redhat.com>
13285 PR tree-optimization/77975
13286 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13287 edge to be constant.
13288 (get_val_for): For constant x return it. Formatting fix.
13289 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13290 would use the same bases as the current one.
13292 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13294 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13295 instead of vec_select for V1TImode.
13296 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13298 (VSX_LE_128): Add V1TI to this mode iterator.
13299 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13300 (*vsx_le_perm_store_<mode>): Likewise.
13301 (pre-reload splitter for VSX stores): Likewise.
13302 (post-reload splitter for VSX stores): Likewise.
13303 (*vsx_xxpermdi2_le_<mode>): Likewise.
13304 (*vsx_lxvd2x2_le_<mode>): Likewise.
13305 (*vsx_stxvd2x2_le_<mode>): Likewise.
13307 2017-03-09 Michael Eager <eager@eagercon.com>
13309 Correct failures with --enable-checking=yes,rtl.
13311 * config/microblaze/microblaze.c (microblaze_expand_shift):
13312 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13313 test for const0_rtx.
13314 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13315 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13317 2017-03-09 Richard Biener <rguenther@suse.de>
13319 PR tree-optimization/79977
13320 * graphite-scop-detection.c (scop_detection::merge_sese):
13321 Handle the case of extra exits to blocks dominating the entry.
13323 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13325 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13328 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13330 PR rtl-optimization/79949
13331 * lra-constraints.c (process_alt_operands): Check memory when
13332 trying to predict a cycle. Print about the overall increase.
13334 2017-03-09 Richard Biener <rguenther@suse.de>
13336 PR middle-end/79971
13337 * gimple-expr.c (useless_type_conversion_p): Preserve
13338 TYPE_SATURATING for fixed-point types.
13340 2017-03-09 Richard Biener <rguenther@suse.de>
13343 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13344 alignment of BLKmode params.
13346 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13349 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13350 (VALL_NO_V2Q): Likewise.
13352 * config/aarch64/aarch64-simd.md
13353 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13355 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13356 VALL_NO_V2Q mode iterator.
13357 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13359 2017-03-09 Martin Liska <mliska@suse.cz>
13361 PR tree-optimization/79631
13362 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13363 tree_int_cst_sign_bit just for INTEGER constants.
13365 2017-03-09 Martin Liska <mliska@suse.cz>
13369 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13372 2017-03-09 Marek Polacek <polacek@redhat.com>
13375 * tree.c (inchash::add_expr): Handle TREE_VEC.
13377 2017-03-09 Martin Liska <mliska@suse.cz>
13380 (chkp_narrow_size_and_offset): New function.
13381 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13382 (void chkp_parse_bit_field_ref): New function.
13383 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13384 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13386 2017-03-09 Martin Liska <mliska@suse.cz>
13389 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13390 (chkp_find_bounds_1): Remove gcc_unreachable.
13392 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13395 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13396 BUILT_IN_SYNC*, determine the access type from the size suffix and
13397 always build a MEM_REF with that type. Handle forgotten
13398 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13401 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13402 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13403 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13404 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13405 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13406 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13407 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13408 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13409 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13410 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13411 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13412 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13413 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13414 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13415 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13416 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13417 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13418 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13419 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13420 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13421 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13422 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13423 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13424 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13425 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13426 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13427 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13428 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13429 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13430 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13431 definitions outside of __OPTIMIZE__ guarded section.
13434 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13435 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13436 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13437 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13440 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13442 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13443 ("vfenez<mode>"): Add missing constraints.
13445 2017-03-08 Martin Sebor <msebor@redhat.com>
13448 * config/nds32/nds32.c (nds32_option_override):
13449 Fix misspelled diagnostic.
13451 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13454 * gimplify.c (gimplify_omp_for): Replace index var in outer
13455 taskloop statement with an artificial variable and add
13456 OMP_CLAUSE_PRIVATE clause for it.
13458 2017-03-08 Richard Biener <rguenther@suse.de>
13460 PR tree-optimization/79955
13461 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13462 for accesses that are completely outside of the variable.
13464 2017-03-08 Andrew Haley <aph@redhat.com>
13466 PR tree-optimization/79943
13467 * tree-ssa-loop-split.c (compute_new_first_bound): When
13468 calculating the new upper bound, (END-BEG) should be added, not
13471 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13473 * config/avr/avr.md (setmemhi): Make sure match_dup
13474 operand number comes before match_scratch.
13476 2017-03-08 Richard Biener <rguenther@suse.de>
13478 PR tree-optimization/79920
13479 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13480 with ncopies == 1 to ...
13481 (vect_transform_slp_perm_load): ... here. Properly compute
13482 all element loads by iterating VF times over the group. Do
13483 not handle ncopies (computed in a broken way) in
13484 vect_create_mask_and_perm.
13486 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13489 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13490 is a uniform vector, use uniform_vector_p return value instead of
13491 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13493 2017-03-07 Marek Polacek <polacek@redhat.com>
13495 PR middle-end/79809
13496 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13497 (alloca_call_type): Likewise.
13499 2017-03-07 Martin Liska <mliska@suse.cz>
13501 * gcov.c (process_args): Put comment to correct location.
13503 2017-03-07 Martin Liska <mliska@suse.cz>
13505 PR middle-end/68270
13506 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13507 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13508 (chkp_narrow_bounds_for_field): Likewise.
13509 (chkp_parse_array_and_component_ref): Pass one more argument to
13512 2017-03-07 Richard Biener <rguenther@suse.de>
13514 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13517 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13519 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13520 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13522 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13525 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13526 to end of description.
13527 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13529 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13531 PR rtl-optimization/79901
13532 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13534 (*avx512f_<code><mode>3<mask_name>): ... this.
13535 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13536 iterator instead of VI8_AVX2_AVX512BW.
13538 PR rtl-optimization/79901
13539 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13540 min/max expander, expand it using expand_vec_cond_expr.
13543 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13546 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13549 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13550 to void * for PCH reasons.
13551 * dwarf2out.c (output_loc_operands, output_die): Cast
13552 v.val_vec.array to unsigned char *.
13554 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13557 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13560 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13562 PR rtl-optimization/79571
13563 * lra-constraints.c (process_alt_operands): Calculate static
13564 reject and subtract it from overall when only addresses will be
13567 2017-03-06 Julia Koval <julia.koval@intel.com>
13570 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13571 incoming stack boundary to 128 for 64-bit targets.
13573 2017-03-06 Richard Biener <rguenther@suse.de>
13575 PR tree-optimization/79894
13576 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13577 to NULL after folding it.
13579 2017-03-06 Richard Biener <rguenther@suse.de>
13581 PR tree-optimization/79824
13582 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13583 check disabling peeling for gaps.
13585 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13587 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13588 attributes): Document gettimeofday.
13590 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13592 * config/s390/s390.c (s390_option_override_internal): Set
13593 PARAM_MIN_VECT_LOOP_BOUND
13595 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13597 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13598 * config/s390/s390.md: Likewise.
13600 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13603 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13604 (<avx2_avx512>_perm<mode>): Rename to ...
13605 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13607 (<avx512>_perm<mode>_mask): Rename to ...
13608 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13610 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13611 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13612 instead of VI8F_256_512.
13613 (avx512f_perm<mode>): New define_expand.
13614 (avx512f_perm<mode>_mask): Likewise.
13615 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13616 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13618 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13620 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13621 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13623 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13625 2017-03-06 Martin Liska <mliska@suse.cz>
13628 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13629 when having a SSA NAME w/o VAR_DECL assigned to it.
13631 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13633 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13634 msa_dpsub_<su>_d): Fix MODE for vec_select.
13636 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13638 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13640 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13642 2017-03-06 Richard Biener <rguenther@suse.de>
13644 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13645 * plugin.c (register_plugin_info): Likewise.
13646 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13648 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13650 * config/i386/sse.md (sse_storehps, sse_storelps,
13651 avx_<castmode><avxsizesuffix>_<castmode>,
13652 avx512f_<castmode><avxsizesuffix>_<castmode>,
13653 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13654 in condition that at least one operand is not a MEM.
13656 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13658 PR middle-end/79805
13659 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13660 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13662 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13663 gimple_call_nothrow_p flag based on whether original builtin can throw.
13664 If it can, emit following stmts on the fallthrough edge.
13665 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13666 don't create new bb if inserting just debug stmts on the edge, try to
13667 insert them on the fallthru bb or just reset debug stmts.
13669 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13672 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13673 restore recog_data (including the operand rtxes inside it) around
13674 the call to get_insn_template.
13676 2017-03-03 Martin Sebor <msebor@redhat.com>
13678 PR tree-optimization/79699
13679 * context.c (context::~context): Free MPFR caches to avoid
13680 a memory leak on program exit.
13682 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13684 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13685 Use wide_int::ulow () instead of .elt (0).
13687 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13689 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13690 (*pushxf): Limit oF constraint to 32bit targets and add oC
13691 constraint for 64bit targets.
13692 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13693 (*pushdf): Change rmF constraint to rmC.
13695 2017-03-03 Martin Liska <mliska@suse.cz>
13697 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13698 Remove unused variable.
13700 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13703 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13704 is a memory operand, increase num_memory.
13705 (ix86_expand_args_builtin): Likewise.
13707 2017-03-03 Jan Hubicka <jh@suse.cz>
13710 * ipa-devirt.c (maybe_record_node): Properly handle
13711 __cxa_pure_virtual visibility.
13713 2017-03-03 Martin Liska <mliska@suse.cz>
13715 PR tree-optimization/79803
13716 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13718 (pass_loop_prefetch::execute): Disabled optimization if an
13719 assumption about L1 cache size is not met.
13721 2017-03-03 Martin Liska <mliska@suse.cz>
13723 PR rtl-optimization/79574
13724 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13725 (hash_scan_set): Likewise.
13726 (dump_hash_table): Likewise.
13727 (hoist_code): Likewise.
13729 2017-03-03 Richard Biener <rguenther@suse.de>
13731 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13732 in place of uhigh ().
13733 (fixed_convert_from_real): Likewise.
13735 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13738 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13740 2017-03-03 Richard Biener <rguenther@suse.de>
13742 PR middle-end/79818
13743 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13744 TYPE_OVERFLOW_UNDEFINED check.
13746 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13748 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13750 (vector_ae_<mode>_p): Likewise.
13751 (vector_nez_<mode>_p): Likewise.
13752 (vector_ne_v2di_p): Likewise.
13753 (vector_ae_v2di_p): Likewise.
13754 (vector_ne_<mode>_p): Likewise.
13755 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13757 (vsx_tsqrt<mode>2_fe): Likewise.
13759 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13762 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
13764 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13766 PR rtl-optimization/79780
13767 * cprop.c (one_cprop_pass): When second and further conditional trap
13768 in a single basic block is turned into an unconditional trap, turn it
13769 into a deleted note to avoid RTL verification failures.
13771 2017-03-02 Richard Biener <rguenther@suse.de>
13773 * fold-const.c (const_binop): Use ulow () instead of elt (0).
13775 2017-03-02 Richard Biener <rguenther@suse.de>
13777 PR tree-optimization/79345
13779 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
13780 param and abort the walk, returning -1 if it is hit.
13781 (walk_aliased_vdefs): Take a limit param and pass it on.
13782 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
13783 defaulting to 0 and return a signed int.
13784 * tree-ssa-uninit.c (struct check_defs_data): New struct.
13785 (check_defs): New helper.
13786 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
13787 about uninitialized memory.
13788 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
13789 bogus uninitialized warning.
13790 (fixed_convert_from_real): Likewise.
13792 2017-03-02 Bin Cheng <bin.cheng@arm.com>
13794 PR tree-optimization/66768
13795 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
13796 iv_use if base object can't be determined.
13798 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13800 PR tree-optimization/79345
13801 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
13802 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
13803 (get_pattern_stats): Initialize it.
13804 * genemit.c (gen_expand): Verify match_scratch numbers come after
13805 match_operand/match_dup numbers.
13806 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
13807 match_scratch numbers.
13808 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
13810 * config/s390/s390.md (trunctdsd2): Likewise.
13812 2017-03-02 Richard Biener <rguenther@suse.de>
13814 * wide-int.h (wide_int_storage::operator=): Implement in terms
13817 2017-03-02 Richard Biener <rguenther@suse.de>
13819 PR tree-optimization/79777
13820 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
13821 the to insert expression to sth existing.
13823 2017-03-01 Martin Sebor <msebor@redhat.com>
13825 PR middle-end/79692
13826 * gimple-ssa-sprintf.c
13827 (directive::known_width_and_precision): New function.
13828 (format_integer): Use it.
13829 (get_mpfr_format_length): Consider the full range of precision
13830 when computing %g output with the # flag. Set the likely byte
13831 count to 3 rather than 1 when precision is indeterminate.
13832 (format_floating): Correct the lower bound of precision.
13834 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13836 * doc/invoke.texi: Document default code model for 64-bit Linux.
13838 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13841 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
13842 udiv rather than div since input pattern is unsigned.
13844 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
13846 * config/i386/i386.c (print_reg): Warn for values of
13847 unsupported size in integer register.
13849 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
13852 * config/rs6000/predicates.md (current_file_function_operand): Do
13853 not allow self calls to be local if the function is replaceable.
13855 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
13858 * config/rs6000/altivec.h (vec_ctz and others): Change the
13859 preprocessor macro that controls conditional compilation from
13860 _ARCH_PWR9 to __POWER9_VECTOR__.
13861 (vec_all_ne): Change parameterization of __altivec_scalar_pred
13862 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
13863 control (instead of _ARCH_PWR9 control) so that template
13864 definition uses power9-specific function.
13865 (vec_any_eq): Likewise.
13866 (vec_all_ne): Change macro definition to use a power9-specific
13867 expansion under #ifdef __POWER9_VECTOR__ control (instead of
13868 _ARCH_PWR9 control).
13869 (vec_any_eq) Likewise.
13870 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
13871 expansion for CMPNEF to remove support for xvcmpnesp instruction.
13872 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
13873 support for xvcmpnedp instruction.
13874 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
13875 macro expansion so that Power9 implementation of vec_all_ne does
13876 not use the AltiVec predicate framework.
13877 (VCMPNEH_P): Likewise.
13878 (VCMPNEW_P): Likewise.
13879 (VCMPNED_P): Likewise.
13880 (VCMPNEFP_P): Likewise.
13881 (VCMPNEDP_P): Likewise.
13882 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
13883 implementation of vec_any_eq to not use AltiVec predicate
13885 (VCMPAEH_P): Likewise.
13886 (VCMPAEW_P): Likewise.
13887 (VCMPAED_P): Likewise.
13888 (VCMPAEFP_P): Likewise.
13889 (VCMPAEDP_P): Likewise.
13890 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
13891 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
13892 not use the AltiVec predicate framework.
13893 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
13894 of vec_any_eq to not use AltiVec predicate framework.
13895 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13896 support for predefined __POWER9_VECTOR__ macro to indicate that
13897 Power9 instruction selection is enabled.
13898 (altivec_overloaded_builtins): Remove extraneous
13899 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
13900 function argument types RS6000_BTI_bool_V16QI and
13901 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
13902 entry for overloaded function argument types RS6000_BTI_bool_V4SI
13903 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
13904 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
13905 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
13906 Power9 for implementations of vec_cmpne. Change the signature for
13907 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
13908 (representing vec_all_ne) to remove the previously described first
13909 argument of type RS6000_BTI_INTSI, as this was an artifact of
13910 reliance on the AltiVec predicate framework, which is no longer
13911 used in the implementation of these functions. Add
13912 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
13913 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
13914 since, unlike the AltiVec predicate framework implementation, we
13915 do not share function descriptors between vec_alle and vec_anyeq.
13916 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
13917 set of modes that receive special treatment even when
13918 TARGET_P9_VECTOR is true. The special treatment emits code that
13919 does not depend on Power9 instructions.
13920 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
13921 define_expand to not rely on AltiVec predicate framework.
13922 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13924 (vector_ne_v2di_p): Change this define_expand to not rely on
13925 AltiVec predicate framework.
13926 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
13928 (vector_ne_<mode>_p): Change this define_expand to not rely on
13929 AltiVec predicate framework.
13930 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13932 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
13933 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
13934 define_insn pattern.
13935 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
13936 define_insn pattern because the xvcmpne<VSs>. instruction is not
13938 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
13939 instruction is not supported.
13941 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13943 * config/nvptx/nvptx.c: Include intl.h.
13945 2017-03-01 Martin Jambor <mjambor@suse.cz>
13948 * ipa-prop.h (ipa_bits): Removed field known.
13949 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
13950 to pointers. Adjusted their comments to warn about their sharing.
13951 (ipcp_transformation_summary): Change bits to a vector of pointers.
13952 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
13953 (ipa_get_ipa_bits_for_value): Declare.
13954 * tree-vrp.h (value_range): Mark as GTY((for_user)).
13955 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
13956 (ipa_bits_hash_table): Likewise.
13957 (ipa_vr_ggc_hash_traits): Likewise.
13958 (ipa_vr_hash_table): Likewise.
13959 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
13960 being pointers and vr_known being removed.
13961 (ipa_set_jf_unknown): Likewise.
13962 (ipa_get_ipa_bits_for_value): New function.
13963 (ipa_set_jfunc_bits): Likewise.
13964 (ipa_get_value_range): New overloaded functions.
13965 (ipa_set_jfunc_vr): Likewise.
13966 (ipa_compute_jump_functions_for_edge): Use the above functions to
13967 construct bits and vr parts of jump functions.
13968 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
13969 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
13971 (ipcp_grow_transformations_if_necessary): Also allocate
13972 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
13974 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
13975 them. Fix too long lines.
13976 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
13977 vr_known being removed.
13978 (ipa_read_jump_function): Use new setter functions to construct bits
13979 and vr parts of jump functions or set them to NULL.
13980 (write_ipcp_transformation_info): Adjust for bits being pointers.
13981 (read_ipcp_transformation_info): Likewise.
13982 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
13984 Include gt-ipa-prop.h.
13985 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
13987 (ipcp_store_bits_results): Likewise.
13988 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
13989 Do not write to existing jump functions but use a temporary instead.
13991 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13994 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
13995 attempt to use its first operand as BIT_FIELD_REF base.
13997 2017-03-01 Richard Biener <rguenther@suse.de>
13999 PR middle-end/79721
14000 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14001 interpolating formula in wrapping arithmetic.
14002 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14004 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14006 PR tree-optimization/79734
14007 * tree-vect-generic.c (expand_vector_condition): Optimize
14008 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14009 Handle VEC_COND_EXPR where comparison has different inner width from
14010 type's inner width.
14012 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14014 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14015 markup, and similar issues. Remove @opindex entries for things
14016 that aren't options. Add missing -mmpy-option entries.
14018 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14020 PR tree-optimization/79737
14021 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14022 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14023 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14024 instead of byte_size. Formatting fix.
14025 (shift_bytes_in_array_right): Formatting fix.
14027 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14030 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14031 condition on optimize for the leaf function test.
14033 2017-02-28 Martin Liska <mliska@suse.cz>
14036 * read-rtl-function.c (function_reader::handle_unknown_directive):
14037 Bail out when one uses -flto.
14039 2017-02-28 Martin Liska <mliska@suse.cz>
14041 * common.opt: Replace space with tabular for options of <number>
14043 * config/i386/i386.opt: Show <number> value for
14044 -mlarge-data-threshold.
14045 * opts.c (print_filtered_help): Do not display number in hexadecimal
14048 2017-02-28 Martin Liska <mliska@suse.cz>
14050 * common.opt: Fix --help=option -Q for options which are of
14053 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14055 * config/i386/i386.c (print_reg): Error out for values
14056 of 8-bit size in invalid integer register.
14058 2017-02-28 Martin Sebor <msebor@redhat.com>
14060 PR tree-optimization/79691
14061 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14063 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14066 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14067 gcc_unreachable with output_operand_lossage.
14069 2017-02-28 Richard Biener <rguenther@suse.de>
14071 PR tree-optimization/79740
14072 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14074 (visit_nary_op): Insert the nary into the hashtable if we
14075 pattern-matched sth.
14076 * tree-ssa-pre.c (eliminate_insert): Robustify.
14078 2017-02-28 Richard Biener <rguenther@suse.de>
14080 PR middle-end/79731
14081 * fold-const.c (decode_field_reference): Reject out-of-bound
14084 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14086 * config/i386/i386.c: Include intl.h.
14087 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14088 instead of just cond ? "..." : "...".
14089 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14090 * coverage.c (read_counts_file): Likewise.
14091 * omp-offload.c: Include intl.h.
14092 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14093 of just cond ? "..." : "...".
14094 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14095 of just cond ? "..." : "...".
14097 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14100 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14102 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14103 'tune for' CPU name.
14104 * config/arm/arm-cpu-data.h: Regenerated.
14106 2017-02-28 Richard Biener <rguenther@suse.de>
14108 PR tree-optimization/79732
14109 * tree-inline.c (expand_call_inline): Do not shadow var.
14111 2017-02-28 Richard Biener <rguenther@suse.de>
14113 PR tree-optimization/79723
14114 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14115 address-space properly.
14117 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14119 * doc/optinfo.texi (Optimization groups): Fix option used for
14121 * doc/invoke.texi (-fopt-info): Document "omp".
14122 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14123 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14124 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14125 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14126 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14128 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14130 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14131 OPTGROUP_OMP with "omp".
14133 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14136 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14137 for arithmetic shift of unsigned V2DI.
14139 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14141 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14142 arc/linux.h headers.
14143 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14144 (LINK_SPEC): Likewise.
14145 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14146 (EXTRA_SPECS): Likewise.
14147 (STARTFILE_SPEC): Likewise.
14148 (ENDFILE_SPEC): Likewise.
14149 (LIB_SPEC): Likewise.
14150 (TARGET_SDATA_DEFAULT): Likewise.
14151 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14152 (MULTILIB_DEFAULTS): Likewise.
14153 (DWARF2_UNWIND_INFO): Likewise.
14154 * config/arc/big.h: New file.
14155 * config/arc/elf.h: Likewise.
14156 * config/arc/linux.h: Likewise.
14157 * config/arc/t-uClibc: Remove.
14159 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14161 PR tree-optimization/77536
14162 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14163 (tree_transform_and_unroll_loop): Use above function to compute the
14164 estimated niter of unrolled loop and use it when scaling profile.
14165 Also use count info rather than frequency if it's non-zero.
14166 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14167 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14168 (vect_transform_loop): Call above function.
14170 2017-02-27 Richard Biener <rguenther@suse.de>
14172 PR tree-optimization/45397
14173 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14174 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14175 (visit_nary_op): Add pattern matching for CSEing sign-changed
14176 or truncated operations with wider ones.
14178 2017-02-27 Richard Biener <rguenther@suse.de>
14180 PR tree-optimization/79690
14181 * tree-vect-stmts.c (vectorizable_store): Use vector type
14182 built from the DR with address-space.
14184 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14186 * doc/invoke.texi (Optimize Options): Refine the description
14187 of asan-use-after-return.
14189 2017-02-25 Alan Modra <amodra@gmail.com>
14191 PR rtl-optimization/79584
14192 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14193 base, not ad->base_term, the reg within base. Remove assertion
14194 that ad->base == ad->base_term. Replace gen_int_mode using
14195 bogus mode with const0_rtx.
14197 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14199 PR middle-end/79396
14200 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14201 FMA_EXPR like tcc_binary or tcc_unary.
14203 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14206 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14208 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14209 (output_loc_operands): Handle DW_OP_call_ref and
14210 DW_OP_GNU_variable_value.
14211 (struct variable_value_struct): New type.
14212 (struct variable_value_hasher): Likewise.
14213 (variable_value_hash): New variable.
14214 (string_types): Remove.
14215 (copy_loc_descr): New function.
14216 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14217 (prepend_loc_descr_to_each): New function.
14218 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14219 instead of add_loc_descr_to_each if the first argument is single
14220 location list and the second has multiple.
14221 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14222 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14223 when looking for variable value which doesn't have other location info.
14224 (loc_list_from_tree): Formatting fix.
14225 (gen_array_type_die): Simplify DW_AT_string_length handling.
14226 (adjust_string_types): Remove.
14227 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14228 string_types. Call resolve_variable_values.
14229 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14230 (resolve_addr_in_expr): Likewise. Add A argument.
14231 (copy_deref_exprloc): Remove deref argument. Adjust for the
14232 original expression being DW_OP_GNU_variable_value with optionally
14233 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14234 optionally after it.
14235 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14236 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14237 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14238 (variable_value_hasher::hash, variable_value_hasher::equal): New
14240 (resolve_variable_value_in_expr, resolve_variable_value,
14241 resolve_variable_values, note_variable_value_in_expr,
14242 note_variable_value): New functions.
14243 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14246 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14249 * opts.h (handle_generated_option): Add GENERATED_P argument.
14250 * opts-common.c (handle_option): Adjust function comment.
14251 (handle_generated_option): Add GENERATED_P argument, pass it to
14253 (control_warning_option): Pass false to handle_generated_option
14255 * opts.c (maybe_default_option): Pass true to handle_generated_option
14257 * optc-gen.awk: Likewise.
14259 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14261 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14262 a REG, look at the REG it is a SUBREG of.
14263 (splitter for cmpeqsi_t): Ditto.
14265 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14267 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14268 the special USEs with the pattern of the insn, not the insn itself.
14270 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14273 * doc/invoke.texi: Document -mload-store-pairs.
14275 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14276 Sandra Loosemore <sandra@codesourcery.com>
14278 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14279 argument isn't a CONST_INT.
14280 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14281 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14282 (nios2_validate_compare): Bypass alternate compare logic if *op2
14283 is not a CONST_INT.
14284 (ldstwm_operation_p): Return false if first_base is not a REG or
14285 if first_offset is not a CONST_INT.
14287 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14289 * config/cris/cris.md: Use correct operand in a define_peephole2.
14291 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14293 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14295 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14297 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14298 this_insn if it is an INSN or JUMP_INSN.
14299 (force_offsettable): Look at base, not at addr.
14300 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14301 on things that aren't necessarily CONST_INTs.
14303 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14305 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14306 -mfpmath=sse is the default also for x86-32 targets with SSE2
14307 instruction set when @option{-ffast-math} is enabled
14309 2017-02-24 Jeff Law <law@redhat.com>
14311 PR rtl-optimizatoin/79286
14312 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14315 2017-02-24 Richard Biener <rguenther@suse.de>
14317 PR tree-optimization/79389
14318 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14321 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14323 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14324 function comment to reflect reality.
14325 (loop_exits_before_overflow): Fix typo in function description.
14327 2017-02-24 Richard Biener <rguenther@suse.de>
14329 PR tree-optimization/79389
14330 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14331 properly that a threading opportunity exists. Detect conditional
14332 copy/constant propagation opportunities.
14334 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14336 * config/visium/visium.md (type): Add trap.
14337 (b): New mode attribute.
14338 (*btst): Rename into...
14339 (*btst<mode>): ...this and adjust.
14340 (*cbranchsi4_btst_insn): Rename into...
14341 (*cbranch<mode>4_btst_insn): ...this and adjust.
14342 (trap): New define_insn.
14344 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14346 PR tree-optimization/79389
14347 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14348 (noce_reversed_cond_code): New function.
14349 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14350 reversed_comparison_code. Formatting fix.
14351 (noce_try_store_flag): Test rev_cond != NULL in addition to
14352 reversed_comparison_code.
14353 (noce_try_store_flag_constants): Likewise.
14354 (noce_try_store_flag_mask): Likewise.
14355 (noce_try_addcc): Use rev_cond if non-NULL instead of
14356 reversed_comparison_code.
14357 (noce_try_cmove_arith): Likewise. Formatting fixes.
14358 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14359 (noce_find_if_block): Initialize rev_cond.
14360 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14361 instead of false as last argument never attempt to reverse it
14364 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14366 PR tree-optimization/79663
14367 * tree-predcom.c (combine_chains): Process refs in reverse order
14368 only for ZERO length chains, and add explaining comment.
14370 2017-02-23 Jeff Law <law@redhat.com>
14372 PR tree-optimization/79578
14373 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14374 in call to operand_equal_p.
14376 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14379 * config/i386/cpuid.h: Fix another undefined behavior.
14381 2017-02-23 Richard Biener <rguenther@suse.de>
14383 PR tree-optimization/79683
14384 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14385 vector types for data-refs.
14387 2017-02-23 Martin Liska <mliska@suse.cz>
14389 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14391 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14393 PR middle-end/79665
14394 * internal-fn.c (get_range_pos_neg): Moved to ...
14395 * tree.c (get_range_pos_neg): ... here. No longer static.
14396 * tree.h (get_range_pos_neg): New prototype.
14397 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14398 are known to be in between 0 and signed maximum inclusive, try to
14399 expand both unsigned and signed divmod and use the cheaper one from
14402 2017-02-22 Jeff Law <law@redhat.com>
14404 PR tree-optimization/79578
14405 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14406 to compare base operands.
14408 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14411 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14412 gpc_reg_operand instead of fpr_reg_operand.
14414 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14416 * config/mips/mips.c (mips_return_in_memory): Force FP
14417 vector types to be returned in memory for o32 ABI.
14419 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14421 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14422 instead of DW_TAG_member for static data member declarations and don't
14423 set no_linkage_name for static inline data members.
14424 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14427 2017-02-22 Martin Liska <mliska@suse.cz>
14429 * doc/invoke.texi: Replace inequality signs with square brackets
14432 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14434 PR tree-optimization/68644
14435 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14437 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14440 * lra-constraints.c (simplify_operand_subreg): Handle
14441 WORD_REGISTER_OPERATIONS targets.
14443 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14446 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14447 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14448 elimination by swapping fld*.
14450 2017-02-22 Richard Biener <rguenther@suse.de>
14452 PR tree-optimization/79673
14453 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14454 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14455 irrelevant address-space qualifiers and avoiding a
14456 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14458 2017-02-22 Richard Biener <rguenther@suse.de>
14460 PR tree-optimization/79666
14461 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14462 to not symbolically negate if that may introduce undefined
14465 2017-02-22 Martin Liska <mliska@suse.cz>
14468 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14469 * data-streamer-out.c (streamer_write_gcov_count_stream):
14471 * value-prof.c (stream_out_histogram_value): Make assert more
14472 precise based on type of counter.
14474 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14477 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14478 Use nonimmediate_operand instead of memory_operand for operand 1.
14479 (float-extend standard_x87sse_constant_load splitter): Ditto.
14481 2017-02-21 Jeff Law <law@redhat.com>
14483 PR tree-optimization/79621
14484 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14485 blocks with edges to themselves.
14487 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14490 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14491 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14492 Use gimple_call_builtin_p.
14495 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14496 on temporarily removed DEBUG_INSNs.
14498 PR tree-optimization/79649
14499 * tree-loop-distribution.c (classify_partition): Give up on
14500 non-generic address space loads/stores.
14502 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14504 * doc/loop.texi (Loop manipulation): Remove nonexistent
14505 tree_ssa_loop_version from the documentation.
14506 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14508 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14511 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14512 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14513 * config/rs6000/rs6000.c: Include except.h.
14514 (rs6000_expand_split_stack_prologue): Call
14515 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14517 2017-02-21 Martin Jambor <mjambor@suse.cz>
14520 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14521 have been analyzed.
14523 2017-02-21 Martin Jambor <mjambor@suse.cz>
14525 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14526 for backward compatibility only.
14527 * doc/invoke.texi (Option Summary): Remove all references to
14528 -fipa-cp-alignment.
14530 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14534 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14536 * lra-constraints.c (curr_insn_transform): Handle
14537 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14539 2017-02-21 Martin Liska <mliska@suse.cz>
14541 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14543 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14545 PR translation/79638
14546 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14548 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14551 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14552 (arm_function_ok_for_sibcall): Return false for an indirect call by
14553 descriptor if all the argument registers are used.
14554 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14555 alignment of the function.
14557 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14559 PR tree-optimization/61441
14560 * simplify-rtx.c (simplify_const_unary_operation): For
14561 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14562 the sNaN unmodified.
14564 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14566 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14567 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14568 instead of SYSTEM_HEADER_DIR.
14570 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14571 Martin Liška <mliska@suse.cz>
14573 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14574 Fix typos and grammar, use active voice, and clarify.
14576 2017-02-20 Marek Polacek <polacek@redhat.com>
14578 PR middle-end/79537
14579 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14582 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14584 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14587 * config/i386/i386.c (ix86_expand_builtin): Handle
14588 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14589 ix86_builtins_isa[fcode].isa as a requirement of those
14590 flags and any other flag in the bitmask.
14591 (ix86_init_mmx_sse_builtins): Use 0 instead of
14592 ~OPTION_MASK_ISA_64BIT as mask.
14593 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14594 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14595 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14596 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14598 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14601 * lra-constraints.c (split_reg): Check requested split mode
14602 is supported by the register.
14604 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14606 * lra-constraints.c (simplify_operand_subreg): Remove early
14609 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14612 * lra-constraints.c (curr_insn_transform): Tighten condition
14613 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14615 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14618 * lra-constraints.c (curr_insn_transform): Handle
14619 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14621 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14624 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14626 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14628 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14631 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14634 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14636 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14639 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14641 * final.c (last_columnnum, override_columnnum): New variables.
14642 (final_start_function): Set last_columnnum, pass it to begin_prologue
14643 hook and pass 0 to dwarf2out_begin_prologue.
14644 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14645 to source_line debug hook.
14646 (notice_source_line): Compute last_columnnum and for debug_column_info
14647 return true on column changes.
14648 * debug.h (struct gcc_debug_hooks): Add column argument to
14649 source_line and begin_prologue hooks.
14650 (debug_nothing_int_charstar_int_bool): Remove prototype.
14651 (debug_nothing_int_int_charstar,
14652 debug_nothing_int_int_charstar_int_bool): New prototypes.
14653 (dwarf2out_begin_prologue): Add column argument.
14654 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14655 begin_prologue hooks.
14656 (debug_nothing_int_charstar_int_bool): Remove.
14657 (debug_nothing_int_int_charstar,
14658 debug_nothing_int_int_charstar_int_bool): New functions.
14659 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14660 through to dwarf2out_source_line.
14661 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14662 (dwarf2out_source_line): Add column argument, emit it if requested.
14663 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14665 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14666 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14667 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14668 through to dwarf2out_begin_prologue.
14669 (vmsdbgout_source_line): Add column argument, pass it through to
14670 dwarf2out_source_line.
14671 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14672 dbxout_source_line caller.
14673 (dbxout_source_line): Add column argument.
14675 * common.opt (gno-column-info, gcolumn-info): New options.
14676 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14677 (check_die): Also test for multiple DW_AT_decl_column attributes.
14678 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14679 DW_AT_decl_column if requested.
14680 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14682 (gen_variable_die): Likewise.
14683 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14684 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14687 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14688 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14689 (ix86_handle_option): Handle OPT_m3dnowa.
14690 * doc/invoke.texi (-m3dnowa): Document.
14691 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14692 -m3dnowa instead of -m3dnow -march=athlon.
14695 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14696 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14698 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14701 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14702 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14703 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14704 generator for vsx_xxpermdi_<mode>_be.
14705 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14706 force big-endian semantics.
14707 (vsx_xxpermdi_<mode>_be): New define_expand with same
14708 implementation as previous version of vsx_xxpermdi_<mode>.
14710 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14712 PR tree-optimization/79327
14713 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14714 variable, its initialization and use.
14716 2017-02-17 Julia Koval <julia.koval@intel.com>
14718 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14719 (OPTION_MASK_ISA_PKU_UNSET): New.
14720 (ix86_handle_option): Handle -mrdpid.
14721 * config/i386/cpuid.h (bit_RDPID): New.
14722 * config/i386/driver-i386.c (host_detect_local_cpu):
14723 Detect RDPID feature.
14724 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14725 * config/i386/i386-c.c (ix86_target_macros_internal):
14727 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14728 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14729 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14730 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14731 * config/i386/i386.md (define_insn "rdpid"): New.
14732 * config/i386/i386.opt Add -mrdpid.
14733 * config/i386/immintrin.h (_rdpid_u32): New.
14735 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14737 PR rtl-optimization/79541
14738 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14739 instead of transforming it into USE.
14741 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14743 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14744 If HONOR_SNANS (SFmode) force the input to a register.
14745 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14746 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14747 an frsp or similar insn.
14749 2017-02-17 Martin Liska <mliska@suse.cz>
14751 PR rtl-optimization/79577
14752 * params.def (selsched-max-sched-times): Increase minimum to 1.
14754 2017-02-17 Martin Liska <mliska@suse.cz>
14756 PR rtl-optimization/79574
14757 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14759 2017-02-17 Martin Liska <mliska@suse.cz>
14761 PR tree-optimization/79529
14762 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
14763 ssa_defined_default_def_p to handle cases which are implicitly
14765 * tree-ssa.c (ssa_defined_default_def_p): New function.
14766 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
14767 which are implicitly defined.
14768 * tree-ssa.h (ssa_defined_default_def_p): Declare.
14770 2017-02-17 Richard Biener <rguenther@suse.de>
14772 PR middle-end/79576
14773 * params.def (max-ssa-name-query-depth): Limit to 10.
14775 2017-02-17 Richard Biener <rguenther@suse.de>
14777 PR tree-optimization/79552
14778 * tree-ssa-structalias.c (visit_loadstore): Properly verify
14781 2017-02-17 Richard Biener <rguenther@suse.de>
14784 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
14786 2017-02-17 Marek Polacek <polacek@redhat.com>
14788 PR middle-end/79536
14789 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
14790 (fold_negate_expr): New wrapper.
14792 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
14794 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
14795 Correct terminology and de-emphasize pre-standard behavior.
14797 2017-02-16 Alan Modra <amodra@gmail.com>
14799 PR rtl-optimization/79286
14800 * ira.c (def_dominates_uses): New function.
14801 (update_equiv_regs): Don't create an equivalence for insns that
14802 may trap where the register def does not dominate the use.
14804 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
14806 PR rtl-optimization/78127
14807 * lra.c (lra): Call lra_eliminate before finish the loop after
14810 2017-02-16 Richard Biener <rguenther@suse.de>
14812 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
14814 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
14815 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
14816 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
14817 (isl_val_int_from_wi): New function.
14818 (extract_affine_gmp): Rename to ...
14819 (extract_affine_wi): ... this, take a widest_int.
14820 (extract_affine_int): Just wrap extract_affine_wi.
14821 (add_param_constraints): Use isl_val_int_from_wi.
14822 (add_loop_constraints): Likewise, and extract_affine_wi.
14824 2017-02-15 Jeff Law <law@redhat.com>
14826 PR middle-end/79521
14827 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
14828 ira_init_register_move_cost_if_necessary.
14830 2017-02-15 Martin Sebor <msebor@redhat.com>
14832 PR middle-end/32003
14833 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
14834 removed in a prior commit.
14836 2017-02-15 Bin Cheng <bin.cheng@arm.com>
14838 PR tree-optimization/79347
14839 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
14840 counters during peeling.
14842 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
14844 * Makefile.in (site.exp): Remove "set ISLVER".
14846 2017-02-15 Jakub Jelinek <jakub@redhat.com>
14849 * real.c (real_from_integer): Call real_convert even for decimal.
14851 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
14854 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
14856 2017-02-14 Andrew Pinski <apinski@cavium.com>
14858 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
14859 cores and change the partno/implementer to be correct.
14860 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
14861 the 'B" as the implementer.
14862 * config/aarch64/aarch64-tune.md: Regenerate.
14864 2017-02-14 Carl Love <cel@us.ibm.com>
14866 * config/rs6000/rs6000.c: Add case statement entry to make the
14867 xvcvuxdsp built-in argument unsigned.
14868 * config/rs6000/vsx.md: Fix the source and return operand types so they
14869 match the instruction definitions from the ISA document. Fix typo
14870 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
14873 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
14876 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
14877 member early_clobber_alts.
14878 * lra-lives.c (reg_early_clobber_p): New.
14879 (process_bb_lives): Use it.
14880 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
14881 (debug_operand_data): Initialize early_clobber_alts.
14882 (setup_operand_alternative): Set up early_clobber_alts.
14883 (collect_non_operand_hard_regs): Ditto. Pass early clobber
14884 alternatives to new_insn_reg.
14885 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
14887 (lra_update_insn_regno_info): Pass the new arg.
14889 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14891 PR middle-end/79505
14892 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
14893 (new_oacc_loop_raw): Don't clear already cleared fields.
14896 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
14897 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
14898 _mm512_prefetch_i64gather_ps): New inline functions and macros.
14900 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
14903 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
14905 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
14908 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
14909 the extra instruction to the right place to store 128-bit constant
14912 2017-02-14 Martin Sebor <msebor@redhat.com>
14914 PR middle-end/79448
14915 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
14916 warning for strings of unknown length.
14918 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
14920 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
14922 2017-02-14 Jeff Law <law@redhat.com>
14925 * ira-costs.c (scan_one_insn): Initialize register move costs
14926 for pseudos seen in USE/CLOBBER insns.
14928 PR tree-optimization/79095
14929 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
14930 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
14931 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
14932 if the operands are known to be not equal, then the resulting range
14934 (intersect_ranges): If the new range is ~[0,0] and the old range is
14935 wide, then prefer ~[0,0].
14936 * tree-vrp.c (overflow_comparison_p_1): New function.
14937 (overflow_comparison_p): New function.
14938 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
14939 if NAME is used in an overflow test.
14940 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
14941 overflow check that can be expressed as an equality test, then adjust
14942 ops to be that equality test.
14944 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14946 * config/s390/s390-builtin-types.def: Remove flags argument.
14947 * config/s390/s390.c (s390_init_builtins): Likewise.
14949 2017-02-14 Martin Liska <mliska@suse.cz>
14951 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
14952 vector. Fix trailing white spaces.
14954 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
14956 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
14959 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14961 PR rtl-optimization/68664
14962 * config/arm/arm.c (arm_sched_can_speculate_insn):
14963 New function. Declare prototype.
14964 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
14966 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14968 PR rtl-optimization/68664
14969 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
14971 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
14973 2017-02-14 Amit Pawar <amit.pawar@amd.com>
14975 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
14976 max skip bytes for function, loop and jump.
14978 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14980 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
14981 ABS_EXPR for gimple dump.
14983 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14986 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
14988 PR tree-optimization/79408
14989 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
14990 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
14991 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
14992 also if rhs1 is INTEGER_CST.
14994 2017-02-14 Richard Biener <rguenther@suse.de>
14996 PR middle-end/79432
14997 * tree-into-ssa.c (insert_phi_nodes): When the function can
14998 have abnormal edges rewrite SSA names with broken use-def
14999 dominance out of SSA and register them for PHI insertion.
15001 2017-02-13 Martin Sebor <msebor@redhat.com>
15003 PR middle-end/79496
15004 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15005 clearing info.nowrite flag when snprintf size argument is a range.
15007 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15009 * cprop.c (cprop_jump): Add missing space in string literal.
15010 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15011 (get_constraint_for_component_ref): Likewise.
15012 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15013 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15014 * lra-constraints.c (process_alt_operands): Likewise.
15015 * ipa-inline.c (inline_small_functions): Likewise.
15016 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15017 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15018 * trans-mem.c (diagnose_tm_1_op): Likewise.
15019 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15020 (grid_parallel_clauses_gridifiable): Likewise.
15022 * config/nvptx/mkoffload.c (process): Add space in between
15025 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15026 "MOD4_SSE_REGS" and "ALL_REGS".
15028 * spellcheck.c (test_data): Add , in between "foo" and "food".
15030 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15033 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15034 boundary crossing check and subsequent code generation agree.
15036 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15038 * config/aarch64/aarch64.c (has_memory_op): Delete.
15039 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15042 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15044 PR rtl-optimization/79388
15045 PR rtl-optimization/79450
15046 * combine.c (distribute_notes): When removing TEM_INSN for which
15047 corresponding dest has last value recorded, invalidate that last
15050 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15052 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15053 of explicit '@'. Add missing assembly comment marker on branch costs
15056 2017-02-13 Nathan Sidwell <nathan@acm.org>
15058 * gengtype-lex.l (<in_struct>): Add '/'.
15060 2017-02-13 Martin Liska <mliska@suse.cz>
15063 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15065 2017-02-13 Richard Biener <rguenther@suse.de>
15067 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15069 * configure: Re-generate.
15070 * config.in: Likewise.
15071 * graphite-dependences.c: Simplify as if
15072 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15073 * graphite-isl-ast-to-gimple.c: Likewise.
15074 * graphite-optimize-isl.c: Likewise.
15075 * graphite-poly.c: Likewise.
15076 * graphite-sese-to-poly.c: Likewise.
15077 * graphite.h: Likewise.
15078 * toplev.c: Include isl/version.h and use isl_version () for
15079 printing the ISL version.
15080 * doc/install.texi: Update ISL requirement.
15082 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15084 * doc/standards.texi (Standards): Update reference to
15087 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15089 * doc/extend.texi (Named Address Spaces): sourceware.org now
15091 * doc/install.texi (Binaries): Ditto.
15094 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15096 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15097 standard terminology "stringize"/"stringizing" throughout.
15098 * doc/cppinternals.texi: Likewise.
15100 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15102 * doc/extend.texi: Fix some spelling mistakes and typos.
15103 * doc/invoke.texi: Likewise.
15105 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15108 * params.def (inline-min-speedup) Change from 10 to 8.
15110 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15112 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15115 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15118 * ipa-inline-analysis.c (get_minimal_bb): New function.
15119 (record_modified): Use it.
15120 (remap_edge_change_prob): Handle also ancestor functions.
15122 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15124 * doc/contrib.texi (Contributors): Remove broken link into
15125 the Mauve CVS repository.
15127 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15129 PR middle-end/79454
15130 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15131 result computation whenever lhs doesn't have vector mode, not
15132 just when it has BLKmode.
15134 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15136 * doc/makefile.texi (profiledbootstrap): Refer to the
15137 installation instructions only in textual form.
15139 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15142 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15144 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15146 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15147 (Specific): Update mingw-w64 reference.
15149 (Specific): Remove broken link to Renesas RX processor.
15151 2017-02-10 Richard Biener <rguenther@suse.de>
15153 * toplev.c (process_options): Do not mention obsolete graphite
15154 options when printing sorry message about missing graphite support.
15155 Mention -floop-nest-optimize.
15157 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15159 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15160 (vtst_p16): Likewise.
15161 (vtstq_p8): Likewise.
15162 (vtstq_p16): Likewise.
15164 (vtstq_p64): Likewise.
15165 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15166 (vset_lane_p64): New.
15167 (vsetq_lane_p64): New.
15169 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15171 PR tree-optimization/79411
15172 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15173 stmt operands are SSA_NAMEs used in abnormal phis.
15174 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15177 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15180 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15183 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15185 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15187 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15189 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15193 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15194 not arbitrary TREE_CONSTANT.
15197 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15198 "omp declare target link" attribute unless is_global_var.
15199 * omp-offload.c (find_link_var_op): Likewise.
15201 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15202 Chung-Lin Tang <cltang@codesourcery.com>
15204 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15206 (gimplify_adjust_omp_clauses): Don't delete TILE.
15207 (gimplify_omp_for): Deal with TILE.
15208 * internal-fn.c (expand_GOACC_TILE): New function.
15209 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15211 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15212 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15214 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15215 avoid DIV for outermost collapse var.
15216 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15217 Remove out of date comments, fix whitespace.
15218 * omp-general.c (omp_extract_for_data): Deal with tiling.
15219 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15220 adjust OLF_DIM_BASE value.
15221 (struct omp_for_data): Add tiling field.
15222 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15223 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15224 for auto loops. Remove default auto determining, moved to
15225 oacc_loop_fixed_partitions.
15226 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15227 stmts, add e_mask field.
15228 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15229 (oacc_thread_numbers): Use oacc_dim_call.
15230 (oacc_xform_tile): New.
15231 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15232 (finish_oacc_loop): Adjust for ifns vector.
15233 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15234 add case for GOACC_TILE fns.
15235 (oacc_loop_xform_loop): Delete.
15236 (oacc_loop_process): Iterate over call list directly, and add
15237 handling for GOACC_TILE fns.
15238 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15240 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15241 vector partitioning to outer loops. Assign 2 partitions to loops
15242 when available. Add TILE handling.
15243 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15244 (execite_oacc_device_lower): Process GOACC_TILE fns,
15245 ignore unknown specs.
15246 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15247 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15248 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15250 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15252 * configure.ac (ACX_BUGURL): Update.
15253 * configure: Regenerate.
15255 2017-02-09 Richard Biener <rguenther@suse.de>
15257 PR tree-optimization/69823
15258 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15259 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15261 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15263 * config/arc/arc-c.def: Add __NPS400__ definition.
15264 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15265 (TARGET_NPS400): Define.
15267 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15269 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15271 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15272 pointer, arch_info.
15273 (arc_cpu_types): Fill the arch_info field with a pointer into the
15274 arc_arch_types table.
15275 (arc_selected_cpu): Declare.
15276 * config/arc/arc.c (arc_selected_cpu): Make global.
15277 (arc_selected_arch): Delete.
15278 (arc_base_cpu): Delete.
15279 (arc_override_options): Remove references to deleted variables,
15280 update access to arch information.
15281 (ARC_OPT): Update access to arch information.
15282 (ARC_OPTX): Likewise.
15283 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15284 (TARGET_ARC600): Update access to arch information.
15285 (TARGET_ARC601): Likewise.
15286 (TARGET_ARC700): Likewise.
15287 (TARGET_EM): Likewise.
15288 (TARGET_HS): Likewise.
15289 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15292 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15295 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15296 condition/operands for integer GE/LE/GEU/LEU operations.
15298 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15300 PR translation/79397
15301 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15304 2017-02-08 Martin Jambor <mjambor@suse.cz>
15307 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15308 whether allocation happened.
15309 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15310 nothing was allocated.
15312 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15314 PR tree-optimization/79408
15315 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15316 constant, but SSA_NAME with a known integer range, use the minimum
15317 of that range instead of op1 to determine if modulo can be replaced
15318 with its first operand.
15320 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15322 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15324 2017-02-08 Richard Biener <rguenther@suse.de>
15326 PR tree-optimization/71824
15327 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15328 Check all loops contained in the merged region.
15330 2017-02-07 Andrew Pinski <apinski@cavium.com>
15332 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15334 2017-02-07 Andrew Pinski <apinski@cavium.com>
15336 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15337 (thunderxt88): Likewise.
15338 (thunderxt81): Disable LSE and change v8.1 to v8.
15339 (thunderxt83): Likewise.
15341 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15342 Richard Biener <rguenther@suse.de>
15344 PR middle-end/79399
15345 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15346 type from int to size_t.
15347 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15349 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15351 PR rtl-optimization/79386
15352 * cprop.c (bypass_conditional_jumps): Initialize
15353 bypass_last_basic_block already before splitting bbs after
15354 unconditional traps...
15355 (bypass_conditional_jumps): ... rather than here.
15358 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15359 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15360 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15361 fix -masm=intel patterns.
15363 2017-02-07 Richard Biener <rguenther@suse.de>
15365 PR tree-optimization/79256
15366 PR middle-end/79278
15367 * builtins.c (get_object_alignment_2): Use min_align_of_type
15368 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15369 and ADJUST_FIELD_ALIGN.
15371 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15373 * doc/tm.texi: Regenerate.
15374 * stor-layout.c (layout_decl): Adjust.
15375 (update_alignment_for_field): Likewise.
15376 (place_field): Likewise.
15377 (min_align_of_type): Likewise.
15378 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15379 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15380 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15381 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15382 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15383 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15384 * config/i386/i386.c (x86_field_alignment): Likewise.
15385 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15386 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15387 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15388 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15389 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15390 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15394 2017-01-30 Richard Biener <rguenther@suse.de>
15396 PR tree-optimization/79256
15397 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15398 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15401 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15403 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15404 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15405 builtins to SImode and emit a zero-extend, if necessary.
15407 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15409 * docs/invoke.texi (RISC-V Options): Alphabetize.
15411 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15413 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15416 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15418 * config/riscv/riscv.c: New file.
15419 * common/config/riscv/riscv-common.c: Likewise.
15420 * config.gcc: Likewise.
15421 * config/riscv/constraints.md: Likewise.
15422 * config/riscv/elf.h: Likewise.
15423 * config/riscv/generic.md: Likewise.
15424 * config/riscv/linux.h: Likewise.
15425 * config/riscv/multilib-generator: Likewise.
15426 * config/riscv/peephole.md: Likewise.
15427 * config/riscv/pic.md: Likewise.
15428 * config/riscv/predicates.md: Likewise.
15429 * config/riscv/riscv-builtins.c: Likewise.
15430 * config/riscv/riscv-c.c: Likewise.
15431 * config/riscv/riscv-ftypes.def: Likewise.
15432 * config/riscv/riscv-modes.def: Likewise.
15433 * config/riscv/riscv-opts.h: Likewise.
15434 * config/riscv/riscv-protos.h: Likewise.
15435 * config/riscv/riscv.h: Likewise.
15436 * config/riscv/riscv.md: Likewise.
15437 * config/riscv/riscv.opt: Likewise.
15438 * config/riscv/sync.md: Likewise.
15439 * config/riscv/t-elf-multilib: Likewise.
15440 * config/riscv/t-linux: Likewise.
15441 * config/riscv/t-linux-multilib: Likewise.
15442 * config/riscv/t-riscv: Likewise.
15443 * configure.ac: Likewise.
15444 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15445 Waterman as RISC-V maintainers.
15446 * doc/install.texi: Add RISC-V entries.
15447 * doc/invoke.texi: Add RISC-V options section.
15448 * doc/md.texi: Add RISC-V constraints section.
15449 * configure: Regenerated.
15451 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15454 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15455 false values to be constant vectors with all 0 or all 1 bits set.
15456 (vcondu<mode><mode>): Likewise.
15457 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15459 (fpmask_comparison_operator): Update comment.
15460 (vecint_comparison_operator): New predicate.
15461 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15462 vector conditionals when the true and false values are constant
15463 vectors with all 0 bits or all 1 bits set.
15465 2017-02-06 Martin Sebor <msebor@redhat.com>
15467 PR tree-optimization/79376
15468 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15470 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15472 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15473 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15474 to simplify split condition.
15476 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15478 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15479 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15482 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15484 PR rtl-optimization/68664
15485 * target.def (can_speculate_insn): New hook.
15486 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15487 * doc/tm.texi: Regenerate.
15488 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15489 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15490 (rs6000_sched_can_speculate_insn): New function.
15492 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15494 PR tree-optimization/79284
15495 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15496 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15497 vectorizable_mask_load_store, vectorizable_operation,
15498 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15499 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15500 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15501 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15502 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15503 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15504 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15505 is_gimple_assign (stmt). Replace another such test with
15506 is_gimple_assign (stmt).
15508 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15511 * config/avr/avr.c (rtl-iter.h): Include it.
15512 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15513 (avr_legitimate_combined_insn): ...and implementation.
15515 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15517 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15518 * config/s390/s390.c (s390_const_operand_ok)
15519 (s390_canonicalize_comparison, s390_extract_part)
15520 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15521 (s390_contiguous_bitmask_p, s390_rtx_costs)
15522 (legitimize_pic_address): Likewise.
15523 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15524 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15525 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15526 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15527 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15529 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15531 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15532 REGNO($0) == REGNO($1).
15534 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15536 * config/s390/linux.h(SIZE_TYPE): Add comment.
15538 2017-02-06 Julian Brown <julian@codesourcery.com>
15539 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15540 Virendra Pathak <virendra.pathak@broadcom.com>
15542 * config/aarch64/aarch64-cores.def: Change the scheduler
15544 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15545 * config/aarch64/thunderx2t99.md: New file.
15547 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15549 * doc/standards.texi (Go Language): Update link to language
15552 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15554 * tree-eh.c (lower_resx): Sanitize profile.
15555 (cleanup_empty_eh_move_lp): Likewise.
15557 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15560 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15562 * cfgloopmanip.h (loop_version): Update prototype.
15563 * modulo-sched.c (sms_schedule): Update call of loop_version.
15564 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15565 * tree-parloops.c (gen_parallel_loop): Likewise.
15566 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15567 * tree-ssa-loop-split.c (split_loop): Likewise.
15568 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15569 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15571 2017-02-05 Martin Liska <mliska@suse.cz>
15574 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15576 (print_operand_address): Initialize a struct to zero.
15578 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15580 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15581 garbage collector only in textual form.
15583 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15585 * doc/extend.texi (x86 specific memory model extensions for
15586 transactional memory): Simplify a phrase.
15588 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15591 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15592 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15593 (atomic_storedi_1): Likewise.
15595 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15597 PR tree-optimization/79338
15598 * tree-parloops.c (gather_scalar_reductions): Don't call
15599 vect_analyze_loop_form for loop->inner before destroying loop's
15602 2017-02-03 Martin Sebor <msebor@redhat.com>
15604 PR tree-optimization/79327
15605 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15606 when precision has resulted in leading zeros.
15607 (format_integer): Adjust the likely counter to assume an unknown
15608 argument that may be zero is non-zero.
15610 2017-02-03 Jason Merrill <jason@redhat.com>
15613 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15614 avoid copying non-taken branch.
15616 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15618 PR tree-optimization/79340
15619 * tree-vect-loop.c (vectorizable_reduction): Release
15620 vec_defs elements after safe_splicing them into other vectors.
15623 PR tree-optimization/79327
15624 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15625 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15627 (format_integer): Use wide_int_to_tree instead of build_int_cst
15628 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15629 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15630 of shortest and longest sequence.
15632 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15634 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15635 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15637 2017-02-03 Walter Lee <walt@tilera.com>
15640 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15641 after initial stackframe link reg save.
15642 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15644 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15647 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15648 wu for stxssp alternative.
15650 2017-02-03 Martin Sebor <msebor@redhat.com>
15652 PR tree-optimization/79352
15653 * gimple-fold.c (get_range_strlen): Add argument.
15654 (get_range_strlen): Change return type to bool.
15655 (get_maxval_strlen): Pass in a dummy argument.
15656 * gimple-fold.h (get_range_strlen): Change return type to bool.
15657 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15658 * tree.h (array_at_struct_end_p): Add argument.
15659 * tree.c (array_at_struct_end_p): Handle it.
15661 2017-02-03 Martin Liska <mliska@suse.cz>
15664 * multiple_target.c (create_dispatcher_calls): Redirect edge
15665 from a caller of a dispatcher.
15666 (expand_target_clones): Make the clones local.
15667 (ipa_target_clone): Do both target clones and resolvers.
15668 (ipa_dispatcher_calls): Remove the pass.
15669 (pass_dispatcher_calls::gate): Likewise.
15670 (make_pass_dispatcher_calls): Likewise.
15671 * passes.def (pass_target_clone): Put as very first IPA early
15674 2017-02-03 Martin Liska <mliska@suse.cz>
15676 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15677 in case of a function with ifunc attribute.
15679 2017-02-03 Martin Liska <mliska@suse.cz>
15681 * cgraph.c (cgraph_node::dump): Dump function version info.
15682 * symtab.c (symtab_node::dump_base): Add missing new line.
15684 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15686 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15687 (ifcombine_ifandif): Use it.
15689 2017-02-03 Martin Liska <mliska@suse.cz>
15691 * doc/invoke.texi: Document default value for
15692 use-after-scope-direct-emission-threshold.
15694 2017-02-03 Martin Liska <mliska@suse.cz>
15696 PR tree-optimization/79339
15697 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15698 (format_floating): Likewise.
15700 2017-02-03 Martin Liska <mliska@suse.cz>
15703 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15705 (ipa_node_params_t::remove): Likewise.
15706 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15707 initialization from removed ipa_node_params_t::insert.
15708 (ipa_node_params::~ipa_node_params): Move from removed
15709 ipa_node_params_t::release.
15710 * symbol-summary.h (symbol_summary::m_released): New member.
15711 Do not release a summary twice. Do not allow to call finalizer
15712 for types of a summary that live in GGC memory.
15714 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15716 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15719 2017-02-02 Martin Sebor <msebor@redhat.com>
15721 PR middle-end/79275
15722 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15723 (format_string): Tighten up the range of output for non-constant
15724 strings and correct the expected range for wide non-constant strings.
15726 2017-02-02 Martin Sebor <msebor@redhat.com>
15728 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15730 PR middle-end/32003
15731 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15733 (-fdump-tree-@var): Add to index and document how to come up
15734 with pass-specific option and dump file names.
15735 (-fdump-passes): Clarify where to look for output.
15737 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15739 PR middle-end/77445
15740 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15741 statistics of the analyzed path; allow threading for speed when
15742 any of BBs along the path are optimized for speed.
15744 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15746 PR middle-end/78468
15747 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15748 settings of the virtual registers.
15751 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15753 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15754 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15757 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15759 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15760 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15762 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15764 * config/s390/s390.md: Add missing comments with the expanded
15766 * config/s390/vector.md: Likewise.
15767 * config/s390/vx-builtins.md: Likewise.
15769 2017-02-02 Jakub Jelinek <jakub@redhat.com>
15772 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
15773 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
15774 conditions on a single line.
15776 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15778 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15779 __S390_VX__ to __VX__.
15781 2017-02-01 Andrew Pinski <apinski@cavium.com>
15783 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
15784 stmt_info to record_stmt_cost.
15785 (vect_get_known_peeling_cost): Pass stmt_info if known to
15787 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
15788 cpu_vector_cost field into
15789 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
15790 field into vec_int_stmt_cost and vec_fp_stmt_cost.
15791 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
15792 splitting of scalar_stmt_cost and vec_stmt_cost.
15793 (thunderx_vector_cost): Likewise.
15794 (cortexa57_vector_cost): LIkewise.
15795 (exynosm1_vector_cost): Likewise.
15796 (xgene1_vector_cost): Likewise.
15797 (thunderx2t99_vector_cost): Improve after the splitting of the two
15799 (aarch64_builtin_vectorization_cost): Update for the splitting of
15800 scalar_stmt_cost and vec_stmt_cost.
15802 2017-02-01 Torvald Riegel <triegel@redhat.com>
15803 Richard Henderson <rth@redhat.com>
15805 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
15806 conditional on existance of a fast atomic load.
15807 * optabs-query.c (can_atomic_load_p): New function.
15808 * optabs-query.h (can_atomic_load_p): Declare it.
15809 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
15810 no fast atomic load is available for the particular size of access.
15811 (expand_atomic_compare_and_swap): Likewise.
15812 (expand_atomic_load): Likewise.
15813 (expand_atomic_store): Likewise.
15814 (expand_atomic_fetch_op): Likewise.
15815 * testsuite/lib/target-supports.exp
15816 (check_effective_target_sync_int_128): Remove x86 because it provides
15817 no fast atomic load.
15818 (check_effective_target_sync_int_128_runtime): Likewise.
15820 2017-02-01 Richard Biener <rguenther@suse.de>
15822 * graphite.c: Include tree-vectorizer.h for find_loop_location.
15823 (graphite_transform_loops): Provide opt-info for optimized nests.
15824 * tree-parloop.c (parallelize_loops): Provide opt-info for
15825 parallelized loops.
15827 2017-02-01 Richard Biener <rguenther@suse.de>
15829 PR middle-end/79315
15830 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
15831 was not set before.
15833 2017-02-01 Richard Biener <rguenther@suse.de>
15835 PR tree-optimization/71824
15836 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15837 Verify the loops are valid in the merged SESE region.
15838 (scop_detection::can_represent_loop_1): Check analyzing the
15839 evolution of the number of iterations in the region succeeds.
15841 2017-01-31 Ian Lance Taylor <iant@golang.org>
15843 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
15844 REG_ARGS_SIZE note to 32-bit push insns and call insn.
15846 2017-01-31 David Malcolm <dmalcolm@redhat.com>
15848 PR preprocessor/79210
15849 * input.c (get_substring_ranges_for_loc): Replace line_width
15850 assertion with error-handling.
15852 2017-01-31 Richard Biener <rguenther@suse.de>
15854 PR tree-optimization/77318
15855 * graphite-sese-to-poly.c (extract_affine): Fix assert.
15856 (create_pw_aff_from_tree): Take loop parameter.
15857 (add_condition_to_pbb): Pass loop of the condition to
15858 create_pw_aff_from_tree.
15860 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15862 * config/s390/s390.c (s390_asan_shadow_offset): New function.
15863 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
15865 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
15869 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
15871 (convert_int_to_float128): Likewise.
15872 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
15873 (convert_int_to_float128): Likewise.
15874 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
15875 (UNSPEC_IEEE128_CONVERT): Likewise.
15876 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
15877 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
15878 Use local variables for IBM extended format.
15879 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
15880 (fix_trunc<mode>si2_fprs): Likewise.
15881 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
15882 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
15883 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
15884 to know that we can now have integers of all sizes in vector
15886 (fix<uns>_<mode>di2_hw): Likewise.
15887 (float<uns>_<mode>si2_hw): Likewise.
15888 (fix_<mode>si2_hw): Likewise.
15889 (fixuns_<mode>si2_hw): Likewise.
15890 (float<uns>_<mode>di2_hw): Likewise.
15891 (float_<mode>di2_hw): Likewise.
15892 (float_<mode>si2_hw): Likewise.
15893 (floatuns_<mode>di2_hw): Likewise.
15894 (floatuns_<mode>si2_hw): Likewise.
15895 (xscvqp<su>wz_<mode>): Delete, no longer used.
15896 (xscvqp<su>dz_<mode>): Likewise.
15897 (xscv<su>dqp_<mode>): Likewise.
15898 (ieee128_mfvsrd_64bit): Likewise.
15899 (ieee128_mfvsrd_32bit): Likewise.
15900 (ieee128_mfvsrwz): Likewise.
15901 (ieee128_mtvsrw): Likewise.
15902 (ieee128_mtvsrd_64bit): Likewise.
15903 (ieee128_mtvsrd_32bit): Likewise.
15905 2017-01-31 Martin Liska <mliska@suse.cz>
15908 * ipa-prop.c (ipa_free_all_node_params): Call release method
15909 instead of ~sumbol_summary to not to trigger double times
15912 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
15914 PR tree-optimization/71691
15915 * bitmap.h (class auto_bitmap): New.
15916 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
15917 is_maybe_undefined instead of ssa_undefined_value_p.
15919 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15921 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15922 __S390_ARCH_LEVEL__ to __ARCH__.
15924 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15926 PR tree-optimization/79267
15927 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
15928 if should_remove_lhs_p is true.
15930 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
15933 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
15934 (add_alignment_attribute): New.
15935 (base_type_die): Add alignment attribute.
15936 (subrange_type_die): Likewise.
15937 (modified_type_die): Likewise.
15938 (gen_array_type_die): Likewise.
15939 (gen_descr_array_type_die: Likewise.
15940 (gen_enumeration_type_die): Likewise.
15941 (gen_subprogram_die): Likewise.
15942 (gen_variable_die): Likewise.
15943 (gen_field_die): Likewise.
15944 (gen_ptr_to_mbr_type_die): Likewise.
15945 (gen_struct_or_union_type_die): Likewise.
15946 (gen_subroutine_type_die): Likewise.
15947 (gen_typedef_die): Likewise.
15948 (base_type_cmp): Compare alignment attribute.
15950 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15953 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
15954 (setb_unsigned) New pattern for setb with CCUNS.
15955 * config/rs6000/rs6000.c (expand_block_compare): Use a different
15956 subfc./subfe sequence to avoid overflow problems. Generate a
15957 shorter sequence with cmpld/setb for power9.
15958 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
15959 for generating subfc. instruction.
15960 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
15961 now uses this instruction.
15963 2017-01-30 Ian Lance Taylor <iant@google.com>
15966 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
15967 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
15969 2017-01-30 Martin Sebor <msebor@redhat.com>
15971 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
15972 Move constant to the right of a relational operator.
15973 (get_mpfr_format_length, format_character, format_string): Ditto.
15974 (should_warn_p, maybe_warn): Same.
15976 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
15978 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
15981 * asan.c (get_translation_unit_decl): Remove function.
15982 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
15984 2017-01-30 Martin Liska <mliska@suse.cz>
15986 PR gcov-profile/79259
15987 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
15988 -fprofile-generate.
15990 2017-01-30 Martin Liska <mliska@suse.cz>
15993 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
15994 Initialize variables with NULL value.
15996 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
15999 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16001 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16003 2017-01-30 Richard Biener <rguenther@suse.de>
16005 PR tree-optimization/79276
16006 * tree-vrp.c (process_assert_insertions): Properly adjust common
16007 when removing a duplicate.
16009 * gcc.dg/torture/pr79276.c: New testcase.
16011 2017-01-30 Richard Biener <rguenther@suse.de>
16013 PR tree-optimization/79256
16014 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16015 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16017 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16019 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16022 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16023 ("*r<noxa>sbg_<mode>_sll_bitmask")
16024 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16025 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16026 Use contiguous_bitmask_nowrap_operand.
16028 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16031 * config/rs6000/altivec.h (vec_xl): Revise #define.
16032 (vec_xst): Likewise.
16034 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16036 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16038 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16040 PR rtl-optimization/79194
16041 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16042 traps before call to bypass_conditional_jumps.
16044 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16046 PR tree-optimization/71374
16047 * lra-constraints.c (check_conflict_input_operands): New.
16048 (match_reload): Use it.
16050 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16053 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16054 account to calculate conflict_set.
16056 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16058 PR rtl-optimization/78559
16059 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16060 other_insn in combine.
16062 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16064 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16065 uint16_type_node for BT_UINT16.
16067 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16069 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16070 "RTL Tests" to menu.
16071 (GIMPLE Tests): New node.
16072 (RTL Tests): New node.
16074 2017-01-27 Richard Biener <rguenther@suse.de>
16076 PR tree-optimization/79245
16077 * tree-loop-distribution.c (distribute_loop): Apply cost
16078 modeling also to detected patterns.
16080 2017-01-27 Richard Biener <rguenther@suse.de>
16082 PR tree-optimization/71433
16083 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16084 (compare_assert_loc): New function.
16085 (process_assert_insertions): Sort and optimize assert locations
16086 to remove duplicates and push down identical assertions on
16087 edges to their destination block.
16089 2017-01-27 Richard Biener <rguenther@suse.de>
16091 PR tree-optimization/79244
16092 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16093 out SSA names even if abnormal.
16095 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16097 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16098 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16099 instead of MPFR_RNDN.
16101 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16104 * arm.c (arm_option_override): Don't call build_target_option_node
16105 until after doing all option overrides.
16106 (arm_valid_target_attribute_tree): Likewise.
16108 2017-01-27 Martin Liska <mliska@suse.cz>
16110 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16111 for {cd}tors and C++ {cd}tors.
16113 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16115 * config/s390/s390.md ("*setmem_long_and")
16116 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16118 2017-01-26 Martin Sebor <msebor@redhat.com>
16120 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16123 2017-01-26 Martin Sebor <msebor@redhat.com>
16125 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16126 HAVE_DFmode before using XFmode or DFmode.
16127 (parse_directive): Avoid using the z length modifier to avoid
16128 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16130 PR middle-end/78703
16131 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16132 to accept adjustment as an array.
16133 (get_int_range): New function.
16134 (struct directive): Make width and prec arrays.
16135 (directive::set_width, directive::set_precision): Call get_int_range.
16136 (format_integer, format_floating): Handle width and precision ranges.
16137 (format_string, parse_directive): Same.
16139 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16142 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16143 comdat_type_p set, just clone them, but keep the children in the
16147 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16148 which have direct callers with -fvar-tracking-assignments enabled
16150 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16151 inside of type units.
16153 2017-01-26 Martin Sebor <msebor@redhat.com>
16155 PR middle-end/78703
16156 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16158 (struct format_result): Replace number_chars, number_chars_min,
16159 and number_chars_max with a single member of struct result_range.
16161 (format_result::operator+=): Adjust.
16162 (struct fmtresult): Remove bounded. Handle likely and unlikely
16164 (fmtresult::adjust_for_width_or_precision): New function.
16165 (fmtresult:type_max_digits): New function.
16166 (bytes_remaining): Handle likely and unlikely counters.
16167 (min_bytes_remaining): Remove.
16168 (format_percent): Simplify.
16169 (format_integer, format_floating): Set likely and unlikely counters.
16170 (get_string_length, format_character, format_string): Same.
16171 (format_plain, should_warn_p): New function.
16172 (maybe_warn): Call should_warn_p. Update diagnostic messages
16173 and handle those for all directives, including plain strings.
16174 (format_directive): Handle likely and unlikely counters.
16175 Remove unnecessary quoting from diagnostics. Add an informational
16177 (add_bytes): Remove.
16178 (pass_sprintf_length::compute_format_length): Simplify.
16179 (try_substitute_return_value): Handle likely and unlikely counters.
16181 2017-01-26 Carl Love <cel@us.ibm.com>
16183 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16184 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16186 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16189 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16190 endianess for subregs into account.
16191 * lra-constraints.c (lra_constraints): Do risky transformations
16192 always on the first iteration.
16193 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16194 last_call_used_reg_set.
16195 (process_bb_lives): Define and use last_call_used_reg_set.
16196 * lra.c (lra): Always continue after lra_constraints on the first
16199 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16201 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16203 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16205 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16207 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16208 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16209 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16210 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16211 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16212 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16213 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16214 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16215 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16217 2017-01-26 Marek Polacek <polacek@redhat.com>
16220 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16221 for the third operand.
16223 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16225 PR middle-end/79236
16226 * omp-low.c (struct omp_context): Add simt_stmt field.
16227 (scan_omp_for): Return omp_context *.
16228 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16229 context to the _simt_ SIMD stmt.
16230 (lower_omp_for): For combined SIMD with sibling _simt_
16231 SIMD, make sure to use the same decls in _looptemp_
16232 clauses as in the sibling.
16234 2017-01-26 David Sherwood <david.sherwood@arm.com>
16236 PR middle-end/79212
16237 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16240 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16243 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16244 emit fld b; fld a; if possible.
16246 * brig-builtins.def: Update copyright years.
16247 * config/arm/arm_acle_builtins.def: Update copyright years.
16249 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16252 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16253 constraint instead of o for the stxsd instruction.
16255 2017-01-25 Carl Love <cel@us.ibm.com>
16257 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16258 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16260 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16262 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16264 2017-01-25 Richard Biener <rguenther@suse.de>
16266 PR tree-optimization/69264
16267 * target.def (vector_alignment_reachable): Improve documentation.
16268 * doc/tm.texi: Regenerate.
16269 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16271 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16272 earlier changes with respect to TYPE_USER_ALIGN.
16273 (vector_alignment_reachable_p): Likewise. Improve dumping.
16275 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16278 * config/arm/arm.md (xordi3): Force constant operand into a register
16281 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16283 * doc/invoke.texi (-fstore-merging): Correct default optimization
16284 levels at which it is enabled.
16285 (-O): Move -fstore-merging from list to...
16288 2017-01-25 Richard Biener <rguenther@suse.de>
16291 * omp-expand.c: Include debug.h.
16292 (expand_omp_taskreg): Make sure to generate early debug before
16293 outlining anything from a function.
16294 (expand_omp_target): Likewise.
16295 (grid_expand_target_grid_body): Likewise.
16297 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16300 * asan.c (get_translation_unit_decl): New function.
16301 (asan_add_global): Extract modules file name from globals
16302 TRANSLATION_UNIT_DECL name.
16304 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16307 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16308 for long calls with APCS frame and VFP.
16310 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16312 * cfg.c (original_copy_tables_initialized_p): New function.
16313 * cfg.h (original_copy_tables_initialized_p): New decl.
16314 * cfgrtl.c (relink_block_chain): Guard the call to
16315 free_original_copy_tables with a call to
16316 original_copy_tables_initialized_p.
16317 * cgraph.h (symtab_node::native_rtl_p): New decl.
16318 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16319 (symtab_node::needed_p): Don't assert for early assembly output
16320 for __RTL functions.
16321 (cgraph_node::finalize_function): Set "force_output" for __RTL
16323 (cgraph_node::analyze): Bail out early for __RTL functions.
16324 (analyze_functions): Update assertion to support __RTL functions.
16325 (cgraph_node::expand): Bail out early for __RTL functions.
16326 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16328 * function.h (struct function): Update comment for field
16330 * gimple-expr.c: Include "tree-pass.h".
16331 (gimple_has_body_p): Return false for __RTL functions.
16332 * Makefile.in (OBJS): Add run-rtl-passes.o.
16333 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16335 (gcc::pass_manager::get_clean_slate): New accessor.
16336 * passes.c: Include "insn-addr.h".
16337 (should_skip_pass_p): Add logging. Update logic for running
16338 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16339 property-provider override so it is only done for gimple passes.
16341 (skip_pass): New function.
16342 (execute_one_pass): Call skip_pass when skipping passes.
16343 * read-md.c (md_reader::read_char): Support filtering
16344 the input to a subset of line numbers.
16345 (md_reader::md_reader): Initialize fields
16346 m_first_line and m_last_line.
16347 (md_reader::read_file_fragment): New function.
16348 * read-md.h (md_reader::read_file_fragment): New decl.
16349 (md_reader::m_first_line): New field.
16350 (md_reader::m_last_line): New field.
16351 * read-rtl-function.c (function_reader::create_function): Only
16352 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16353 curr_properties. Set DECL_INITIAL to a dummy block.
16354 (read_rtl_function_body_from_file_range): New function.
16355 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16357 * run-rtl-passes.c: New file.
16358 * run-rtl-passes.h: New file.
16360 2017-01-24 Jeff Law <law@redhat.com>
16362 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16365 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16367 PR tree-optimization/79159
16368 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16369 (record_nonwrapping_iv): Improve boundary using above function if no
16370 value range information.
16372 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16373 Martin Jambor <mjambor@suse.cz>
16375 * brig-builtins.def: New file.
16376 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16377 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16378 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16379 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16380 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16381 * builtin-types.def (BT_INT8): New.
16382 (BT_INT16): Likewise.
16383 (BT_UINT8): Likewise.
16384 (BT_UINT16): Likewise.
16385 (BT_FN_ULONG): Likewise.
16386 (BT_FN_UINT_INT): Likewise.
16387 (BT_FN_UINT_ULONG): Likewise.
16388 (BT_FN_UINT_LONG): Likewise.
16389 (BT_FN_UINT_PTR): Likewise.
16390 (BT_FN_ULONG_PTR): Likewise.
16391 (BT_FN_INT8_FLOAT): Likewise.
16392 (BT_FN_INT16_FLOAT): Likewise.
16393 (BT_FN_UINT32_FLOAT): Likewise.
16394 (BT_FN_UINT16_FLOAT): Likewise.
16395 (BT_FN_UINT8_FLOAT): Likewise.
16396 (BT_FN_UINT64_FLOAT): Likewise.
16397 (BT_FN_UINT16_UINT32): Likewise.
16398 (BT_FN_UINT32_UINT16): Likewise.
16399 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16400 (BT_FN_INT_PTR_INT): Likewise.
16401 (BT_FN_UINT_PTR_UINT): Likewise.
16402 (BT_FN_LONG_PTR_LONG): Likewise.
16403 (BT_FN_ULONG_PTR_ULONG): Likewise.
16404 (BT_FN_VOID_UINT64_UINT64): Likewise.
16405 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16406 (BT_FN_INT8_INT8_INT8): Likewise.
16407 (BT_FN_INT16_INT16_INT16): Likewise.
16408 (BT_FN_INT_INT_INT): Likewise.
16409 (BT_FN_UINT_FLOAT_UINT): Likewise.
16410 (BT_FN_FLOAT_UINT_UINT): Likewise.
16411 (BT_FN_ULONG_UINT_UINT): Likewise.
16412 (BT_FN_ULONG_UINT_PTR): Likewise.
16413 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16414 (BT_FN_UINT_UINT_UINT): Likewise.
16415 (BT_FN_VOID_UINT_PTR): Likewise.
16416 (BT_FN_UINT_UINT_PTR: Likewise.
16417 (BT_FN_UINT32_UINT64_PTR): Likewise.
16418 (BT_FN_INT_INT_UINT_UINT): Likewise.
16419 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16420 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16421 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16422 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16423 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16424 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16425 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16426 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16427 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16428 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16429 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16430 * doc/frontends.texi: List BRIG FE.
16431 * doc/install.texi (Testing): Add BRIG tesring requirements.
16432 * doc/invoke.texi (Overall Options): Mention BRIG.
16433 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16435 2017-01-24 Richard Biener <rguenther@suse.de>
16437 PR translation/79208
16438 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16440 2017-01-24 Martin Jambor <mjambor@suse.cz>
16443 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16444 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16445 and known_contexts.
16447 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16449 PR middle-end/79123
16450 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16451 casts from signed to unsigned really don't have a range.
16453 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16455 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16456 GMP_RNDx for compatiblity.
16458 2017-01-24 Martin Liska <mliska@suse.cz>
16461 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16462 that would prevent us to call alloca with -1 as argument.
16464 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16466 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16467 Avoid -Wformat-security warning.
16469 2017-01-23 Andrew Pinski <apinski@cavium.com>
16471 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16474 2017-01-23 Martin Sebor <msebor@redhat.com>
16476 PR middle-end/78703
16477 * gimple-ssa-sprintf.c (warn_level): New global.
16478 (format_integer): Use it here and throughout the rest of the file.
16479 Use the same switch to compute sign as base.
16480 (maybe_warn): New function.
16481 (format_directive): Factor out warnings into maybe_warn.
16482 Add debugging output. Use warn_level.
16483 (add_bytes): Use warn_level.
16484 (pass_sprintf_length::compute_format_length): Add debugging output.
16485 (try_substitute_return_value): Same.
16486 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16488 PR middle-end/78703
16489 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16490 (struct fmtresult, format_integer, format_floating): Adjust.
16491 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16492 (get_string_length, format_string,format_directive): Same.
16493 (pass_sprintf_length::compute_format_length): Same.
16494 (try_substitute_return_value): Simplify slightly.
16496 PR middle-end/78703
16497 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16498 (fmtresult::operator+=): Outlined.
16499 (struct fmtresult): Add ctors.
16500 (struct conversion_spec): Rename...
16501 (struct directive): ...to this. Add and remove data members.
16502 (directive::set_width, directive::set_precision): New functions.
16503 (format_percent): Use fmtresult ctor.
16504 (get_width_and_precision): Remove.
16505 (format_integer): Make naming changes. Avoid computing width and
16507 (format_floating): Same. Adjust indentation.
16508 (format_character, format_none): New functions.
16509 (format_string): Moved character handling to format_character.
16510 (format_directive): Remove arguments, change return type.
16511 (parse_directive): New function.
16512 (pass_sprintf_length::compute_format_length): Move directive
16513 parsing to parse_directive.
16515 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16517 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16518 (assign_assembler_name_if_needed): ... this.
16519 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16520 (assign_assembler_name_if_needed): ... this.
16521 (free_lang_data_in_cgraph): Adjust callers.
16522 * cgraphunit.c (cgraph_node::analyze): Likewise.
16523 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16526 2017-01-23 Richard Biener <rguenther@suse.de>
16528 PR tree-optimization/79088
16529 PR tree-optimization/79188
16530 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16531 resetting loop bounds after last path deletion. Reset loop
16532 bounds of the target loop, make code match the comments.
16533 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16534 Make sure loops need no fixups.
16536 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16538 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16539 exponent support with double type for first argument.
16540 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16541 type returned by __builtin_vec_extract_sig,
16542 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16543 functions from "vector int" to "vector unsigned int" or from
16544 "vector long long int" to "vector unsigned long long int".
16545 Changed type returned by __builtin_vec_extract_exp,
16546 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16547 functions from "vector int" to "vector unsigned int" or from
16548 "vector long long int" to "vector unsigned long long int".
16549 Changed return type of __builtin_vec_test_data_class,
16550 __builtin_vec_test_data_class_sp, and
16551 __builtin_vec_test_data_class_dp from "vector int" to
16552 "vector bool int" or from "vector long long int" to "vector bool
16553 long long int" and changed second argument type from "unsigned
16554 int" to "int". Added new overloaded function forms "vector float
16555 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16556 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16557 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16558 double, vector unsigned long long int)" and "vector double
16559 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16560 long int)". Changed return type of
16561 __builtin_scalar_test_data_class and
16562 __builtin_scalar_test_data_class_sp and
16563 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16564 int" and changed second argument from "unsigned int" to "int".
16565 Changed type returned by __builtin_scalar_test_neg,
16566 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16567 from "int" to "bool int". Added new overloaded function form
16568 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16569 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16570 exponent double-precision with floating point first argument.
16571 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16572 documentation of scalar_test_data_class, scalar_test_neg,
16573 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16574 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16575 vec_test_data_class built-in functions to reflect refinements in
16576 their type signatures.
16578 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16580 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16582 (aarch64_elf_asm_destructor): Likewise.
16584 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16586 PR rtl-optimization/78634
16587 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16588 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16589 * ifcvt.c (noce_try_cmove): Add missing cost check.
16591 PR rtl-optimization/71724
16592 * combine.c (if_then_else_cond): Look for situations where it is
16593 beneficial to undo the work of one of the recursive calls.
16595 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16597 PR tree-optimization/70754
16598 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16599 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16600 combined stmt before it if not NULL.
16601 (combine_chains): Process refs reversely and compute dominance point
16604 2017-01-23 Martin Liska <mliska@suse.cz>
16606 PR tree-optimization/79196
16607 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16608 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16610 (strlen_optimize_stmt): Call the renamed function.
16612 2017-01-23 Michael Matz <matz@suse.de>
16614 PR tree-optimization/78384
16615 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16617 2017-01-23 Richard Biener <rguenther@suse.de>
16619 PR tree-optimization/79186
16620 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16621 both incoming edges before moving an assert.
16623 2017-01-23 Martin Jambor <mjambor@suse.cz>
16625 * ipa-prop.c (load_from_param_1): Removed.
16626 (load_from_unmodified_param): Bits from load_from_param_1 put back
16628 (load_from_param): Removed.
16629 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16630 with stmt. Reverted back to use of load_from_unmodified_param.
16632 2017-01-23 Martin Jambor <mjambor@suse.cz>
16635 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16636 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16637 field a pointer to garbage collected vector, mark lattices and
16638 ipcp_orig_node with GTY((skip)).
16639 (ipa_get_param_count): Adjust to descriptors being a pointer.
16640 (ipa_get_param): Likewise.
16641 (ipa_get_type): Likewise.
16642 (ipa_get_param_move_cost): Likewise.
16643 (ipa_set_param_used): Likewise.
16644 (ipa_get_controlled_uses): Likewise.
16645 (ipa_set_controlled_uses): Likewise.
16646 (ipa_is_param_used): Likewise.
16647 (ipa_node_params_t): Move into garbage collector. New methods insert
16649 (ipa_node_params_sum): Annotate wth GTY(()).
16650 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16652 (ipa_load_from_parm_agg): Adjust declaration.
16653 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16654 * ipa-profile.c (ipa_profile): Likewise.
16655 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16656 (ipa_populate_param_decls): Make descriptors parameter garbage
16658 (ipa_dump_param): Adjust to descriptors being a pointer.
16659 (ipa_alloc_node_params): Likewise.
16660 (ipa_initialize_node_params): Likewise.
16661 (load_from_param_1): Make descriptors parameter garbage collected.
16662 (load_from_unmodified_param): Likewise.
16663 (load_from_param): Likewise.
16664 (ipa_load_from_parm_agg): Likewise.
16665 (ipa_node_params::~ipa_node_params): Removed.
16666 (ipa_free_all_node_params): Remove call to delete operator.
16667 (ipa_node_params_t::insert): New.
16668 (ipa_node_params_t::remove): Likewise.
16669 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16670 copy known_csts and known_contexts vectors.
16671 (ipa_read_node_info): Adjust to descriptors being a pointer.
16672 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16674 (ipcp_transform_function): Make descriptors variable garbage
16677 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16679 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16680 * config/i386/avx512dqintrin.h: Ditto.
16681 * config/i386/avx512fintrin.h: Ditto.
16682 * config/i386/i386.c: Handle new builtins.
16683 * config/i386/i386-builtin.def: Add new builtins.
16684 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16685 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16687 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16688 Martin Liska <mliska@suse.cz>
16690 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16691 * asan.c (asan_expand_poison_ifn): Support stores and use
16692 appropriate ASAN report function.
16693 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16694 * internal-fn.def (ASAN_POISON_USE): Declare.
16695 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16696 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16697 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16698 ASAN_POISON calls w/o LHS.
16699 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16700 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16701 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16702 * gimplify.c (asan_poison_variables): Add attribute
16703 use_after_scope_memory to variables that really needs to live
16705 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16706 having the attribute.
16708 2017-01-23 Martin Liska <mliska@suse.cz>
16710 * asan.c (create_asan_shadow_var): New function.
16711 (asan_expand_poison_ifn): Likewise.
16712 * asan.h (asan_expand_poison_ifn): New declaration.
16713 * internal-fn.c (expand_ASAN_POISON): Likewise.
16714 * internal-fn.def (ASAN_POISON): New builtin.
16715 * sanopt.c (pass_sanopt::execute): Expand
16716 asan_expand_poison_ifn.
16717 * tree-inline.c (copy_decl_for_dup_finish): Make function
16719 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16720 * tree-ssa.c (is_asan_mark_p): New function.
16721 (execute_update_addresses_taken): Rewrite local variables
16722 (identified just by use-after-scope as addressable) into SSA.
16724 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16726 * doc/install.texi (Specific): opensource.apple.com uses https
16727 now. Remove trailing slash.
16729 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16731 * README.Portability: Remove note on an Irix compatibility issue.
16733 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16735 * gcov.c (INCLUDE_ALGORITHM): Define.
16736 (INCLUDE_VECTOR): Define.
16737 No longer include <vector> and <algorithm> directly.
16739 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16741 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16743 * doc/invoke.texi (Code Gen Options): Ditto.
16745 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16748 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16750 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16752 rtl-optimization/79125
16753 * cprop.c (local_cprop_pass): Handle cases where we make an
16754 unconditional trap.
16756 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16760 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16761 read from, for big endian.
16763 2017-01-20 Jiong Wang <jiong.wang@arm.com>
16765 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
16766 register pauth builtins for LP64 only.
16768 2017-01-20 Marek Polacek <polacek@redhat.com>
16771 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
16774 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16776 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
16778 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
16779 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
16780 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
16782 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16785 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
16786 in big-endian mode when they are not a single duplicated value.
16788 2017-01-20 Richard Biener <rguenther@suse.de>
16790 * BASE-VER: Bump to 7.0.1.
16792 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16794 * omp-low.c (omplow_simd_context): New struct. Use it...
16795 (lower_rec_simd_input_clauses): ...here and...
16796 (lower_rec_input_clauses): ...here to hold common data. Adjust all
16797 references to idx, lane, max_vf, is_simt.
16799 2017-01-20 Graham Markall <graham.markall@embecosm.com>
16801 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
16804 2017-01-20 Martin Jambor <mjambor@suse.cz>
16806 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
16807 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
16809 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
16810 (GTFILES): Rename hsa.c to hsa-common.c.
16811 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
16812 * hsa-dump.c: Likewise.
16813 * hsa-gen.c: Likewise.
16814 * hsa-regalloc.c: Likewise.
16815 * ipa-hsa.c: Likewise.
16816 * omp-expand.c: Likewise.
16817 * omp-low.c: Likewise.
16818 * toplev.c: Likewise.
16820 2017-01-20 Marek Polacek <polacek@redhat.com>
16823 * doc/invoke.texi: Document -Wduplicated-branches.
16824 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
16825 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
16826 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
16827 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
16828 return 0 only when not OEP_LEXICOGRAPHIC.
16829 (fold_build_cleanup_point_expr): Use the expression
16830 location when building CLEANUP_POINT_EXPR.
16831 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
16832 * tree.c (add_expr): Handle error_mark_node.
16834 2017-01-20 Martin Liska <mliska@suse.cz>
16837 * tree-profile.c (init_ic_make_global_vars): Do not call
16839 (gimple_init_gcov_profiler): Likewise.
16841 2017-01-20 Martin Liska <mliska@suse.cz>
16844 * cgraph.h (maybe_create_reference): Remove argument and
16846 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
16848 * ipa-cp.c (create_specialized_node): Likewise.
16849 * symtab.c (symtab_node::maybe_create_reference): Handle
16850 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
16852 2017-01-20 Martin Liska <mliska@suse.cz>
16854 * read-rtl-function.c (function_reader::create_function): Use
16855 build_decl instread of build_decl_stat.
16857 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
16859 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
16860 * config/i386/avx512dqintrin.h: Ditto.
16861 * config/i386/avx512fintrin.h: Ditto.
16862 * config/i386/i386-builtin-types.def: Add new types.
16863 * config/i386/i386.c: Handle new types.
16864 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
16865 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
16866 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
16867 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
16868 (__builtin_ia32_kshiftridi): New.
16869 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
16871 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
16875 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
16876 define to rs6000_init_stack_protect_guard.
16877 (rs6000_init_stack_protect_guard): New function.
16879 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16880 Yunqiang Su <yunqiang.su@imgtec.com>
16882 * config.gcc (supported_defaults): Add madd4.
16883 (with_madd4): Add validation.
16884 (all_defaults): Add madd4.
16885 * config/mips/mips.opt (mmadd4): New option.
16886 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16888 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
16890 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
16891 (ISA_HAS_FUSED_MADD4): Likewise.
16892 * doc/invoke.texi (-mmadd4): Document the new option.
16893 * doc/install.texi (--with-madd4): Document the new option.
16895 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16897 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
16898 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
16899 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
16900 (aarch64_init_pauth_hint_builtins): New.
16901 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
16902 (aarch64_expand_builtin): Expand new builtins.
16904 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16906 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
16907 * combine-stack-adj.c (no_unhandled_cfa): Handle
16908 REG_CFA_TOGGLE_RA_MANGLE.
16909 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
16910 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
16911 info for return address signing.
16912 (aarch64_expand_epilogue): Likewise.
16914 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16916 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
16917 * config/aarch64/aarch64-protos.h
16918 (aarch64_return_address_signing_enabled): New declaration.
16919 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
16921 (aarch64_expand_prologue): Sign return address before it's pushed onto
16923 (aarch64_expand_epilogue): Authenticate return address fetched from
16925 (aarch64_override_options): Sanity check for ILP32 and ISA level.
16926 (aarch64_attributes): New function attributes for "sign-return-address".
16927 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
16928 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
16929 ("*do_return"): Generate combined instructions according to key index.
16930 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
16931 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
16933 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
16934 * config/aarch64/aarch64.opt (msign-return-address=): New.
16935 * doc/extend.texi (AArch64 Function Attributes): Documents
16936 "sign-return-address=".
16937 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
16939 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16941 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
16942 overall option summary.
16944 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16946 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
16947 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
16948 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
16949 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
16951 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
16953 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
16954 -mpower9-minmax by default for -mcpu=power9.
16955 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
16956 128-bit floating point.
16958 2017-01-20 Alan Modra <amodra@gmail.com>
16960 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
16961 optimizing for size.
16963 2017-01-20 Alan Modra <amodra@gmail.com>
16966 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
16967 for strcmp and strncmp from corresponding builtin decl.
16969 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
16971 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
16972 instead of i386/rtems-64.h.
16973 * config/i386/rtems-64.h: Remove.
16975 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
16979 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
16981 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
16983 2017-01-19 Tamar Christina <tamar.christina@arm.com>
16985 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
16986 Change int to HOST_WIDE_INT.
16987 * config/aarch64/aarch64-protos.h
16988 (aarch64_simd_gen_const_vector_dup): Likewise.
16989 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
16991 2017-01-19 David Malcolm <dmalcolm@redhat.com>
16993 * langhooks-def.h (lhd_type_for_size): New decl.
16994 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
16995 * langhooks.c (lhd_type_for_size): New function, taken from
16998 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17000 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17001 define_bypass for CR latency.
17002 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17003 (power9-alu2): Add define_bypass for CR latency.
17005 (power9-mul): Update insn latency.
17006 (power9-mul-compare): Update insn latency, bypass latency and remove
17009 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17011 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17013 * config/aarch64/aarch64.md
17014 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17015 aarch64_nopcrelative_literal_loads.
17016 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17018 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17020 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17021 TARGET_LOONGSON_3A.
17022 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17024 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17027 * config.gcc (supported_defaults): Add lxc1-sxc1.
17028 (with_lxc1_sxc1): Add validation.
17029 (all_defaults): Add lxc1-sxc1.
17030 * config/mips/mips.opt (mlxc1-sxc1): New option.
17031 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17033 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17034 __mips_no_lxc1_sxc1.
17035 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17036 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17037 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17039 2017-01-19 Richard Biener <rguenther@suse.de>
17041 PR tree-optimization/72488
17042 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17043 sure to restore SSA info.
17044 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17046 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17048 PR rtl-optimization/79121
17049 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17050 of the inner type when shifting an extended value.
17052 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17055 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17056 interposable aliases.
17058 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17061 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17062 Use the evmergelohi instruction.
17063 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17064 (mov_sitf_e500_subreg8_2_be): Likewise.
17065 (mov_sitf_e500_subreg12_2_le): Likewise.
17066 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17067 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17068 (mov_sitf_e500_subreg8_2_le): Likewise.
17069 (mov_sitf_e500_subreg12_2_be): Likewise.
17071 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17073 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17074 attribute from vecsimple to vecperm.
17075 (altivec_vbpermq2): Likewise.
17077 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17080 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17082 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17083 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17084 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17085 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17086 case where N arg is SIZE_MAX.
17087 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17088 (cmpstrsi): Add pattern.
17090 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17092 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17093 __builtin_vec_revb builtins.
17094 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17095 built-in functions to support generation of the ISA 3.0 XXBR<x>
17096 vector byte reverse instructions.
17097 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17098 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17099 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17100 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17101 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17102 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17103 (P9V_BUILTIN_VEC_REVB): Likewise.
17104 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17105 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17106 (p9_xxbrq_v16qi): Likewise.
17107 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17108 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17109 (p9_xxbrh_v8hi): Likewise.
17110 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17111 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17112 vec_revb built-in functions.
17114 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17116 PR rtl-optimization/78952
17117 * config/i386/i386.md (any_extract): New code iterator.
17118 (*insvqi_2): Use any_extract for source operand.
17119 (*insvqi_3): Use any_shiftrt for source operand.
17121 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17123 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17125 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17127 2017-01-18 Matthias Klose <doko@ubuntu.com>
17129 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17131 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17133 * config/rs6000/altivec.h (vec_bperm): Change #define.
17134 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17135 (altivec_vbpermq2): New define_insn.
17136 (altivec_vbpermd): Likewise.
17137 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17138 function interface.
17139 (VBPERMD): Likewise.
17140 (VBPERM): New polymorphic function interface.
17141 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17142 Add entries for P9V_BUILTIN_VEC_VBPERM.
17143 * doc/extend.texi: Add interfaces for vec_bperm.
17145 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17147 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17148 first letter of error messages.
17149 (s390_resolve_overloaded_builtin): Likewise.
17150 * config/s390/s390.c (s390_expand_builtin): Likewise.
17151 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17152 (s390_valid_target_attribute_inner_p): Likewise.
17153 * config/s390/s390.md ("tabort"): Likewise.
17155 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17157 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17158 (ISA_AVOID_DIV_HILO): New macro.
17159 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17160 (ISA_HAS_DDIV): Likewise.
17162 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17164 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17166 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17168 * doc/invoke.texi (fabi-version): Spelling fix.
17170 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17173 * doc/invoke.texi (fabi-version): Mention mangling fix for
17176 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17179 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17181 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17184 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17185 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17187 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17189 (stack_protect_test): Ditto.
17190 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17191 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17193 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17194 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17195 -mstack-protector-guard-offset=.
17196 (RS/6000 and PowerPC Options): Ditto.
17198 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17200 * config/i386/i386.h (MASK_CLASS_P): New define.
17201 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17202 there are no registers from different register sets also when
17203 mask registers are used. Update function comment.
17204 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17205 to (*k/*r) and (*k/*km) alternatives.
17207 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17209 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17210 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17211 (EH_RETURN_HANDLER_RTX): New define.
17212 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17213 Force frame pointer in EH return functions.
17214 (aarch64_expand_epilogue): Add barrier for eh_return.
17215 (aarch64_final_eh_return_addr): Remove.
17216 (aarch64_eh_return_handler_rtx): New function.
17217 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17219 (aarch64_eh_return_handler_rtx): New prototype.
17221 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17223 * config/rs6000/altivec.h (vec_rlmi): New #define.
17224 (vec_vrlnm): Likewise.
17225 (vec_rlnm): Likewise.
17226 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17227 (UNSPEC_VRLNM): Likewise.
17228 (VIlong): New mode iterator.
17229 (altivec_vrl<VI_char>mi): New define_insn.
17230 (altivec_vrl<VI_char>nm): Likewise.
17231 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17233 (VRLDNM): Likewise.
17234 (RLNM): New polymorphic function entry.
17235 (VRLWMI): New monomorphic function entry.
17236 (VRLDMI): Likewise.
17237 (RLMI): New polymorphic function entry.
17238 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17239 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17240 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17243 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17246 * dwarf2out.c (field_byte_offset): Restore the
17247 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17248 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17249 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17252 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17255 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17257 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17260 * dwarf2out.c (add_data_member_location_attribute): For constant
17261 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17262 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17263 DW_AT_byte_size attributes.
17265 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17267 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17268 after forcing to constant memory when the code model is medium.
17270 2017-01-17 Julia Koval <julia.koval@intel.com>
17273 * config/i386/avx512fintrin.h
17274 (_mm512_i32gather_ps): Change __addr type to void const*.
17275 (_mm512_mask_i32gather_ps): Ditto.
17276 (_mm512_i32gather_pd): Ditto.
17277 (_mm512_mask_i32gather_pd): Ditto.
17278 (_mm512_i64gather_ps): Ditto.
17279 (_mm512_mask_i64gather_ps): Ditto.
17280 (_mm512_i64gather_pd): Ditto.
17281 (_mm512_mask_i64gather_pd): Ditto.
17282 (_mm512_i32gather_epi32): Ditto.
17283 (_mm512_mask_i32gather_epi32): Ditto.
17284 (_mm512_i32gather_epi64): Ditto.
17285 (_mm512_mask_i32gather_epi64): Ditto.
17286 (_mm512_i64gather_epi32): Ditto.
17287 (_mm512_mask_i64gather_epi32): Ditto.
17288 (_mm512_i64gather_epi64): Ditto.
17289 (_mm512_mask_i64gather_epi64): Ditto.
17290 (_mm512_i32scatter_ps): Change __addr type to void*.
17291 (_mm512_mask_i32scatter_ps): Ditto.
17292 (_mm512_i32scatter_pd): Ditto.
17293 (_mm512_mask_i32scatter_pd): Ditto.
17294 (_mm512_i64scatter_ps): Ditto.
17295 (_mm512_mask_i64scatter_ps): Ditto.
17296 (_mm512_i64scatter_pd): Ditto.
17297 (_mm512_mask_i64scatter_pd): Ditto.
17298 (_mm512_i32scatter_epi32): Ditto.
17299 (_mm512_mask_i32scatter_epi32): Ditto.
17300 (_mm512_i32scatter_epi64): Ditto.
17301 (_mm512_mask_i32scatter_epi64): Ditto.
17302 (_mm512_i64scatter_epi32): Ditto.
17303 (_mm512_mask_i64scatter_epi32): Ditto.
17304 (_mm512_i64scatter_epi64): Ditto.
17305 (_mm512_mask_i64scatter_epi64): Ditto.
17306 * config/i386/avx512pfintrin.h
17307 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17308 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17309 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17310 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17311 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17312 (_mm512_prefetch_i32scatter_ps): Ditto.
17313 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17314 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17315 (_mm512_prefetch_i64scatter_pd): Ditto.
17316 (_mm512_prefetch_i64scatter_ps): Ditto.
17317 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17318 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17319 * config/i386/avx512vlintrin.h
17320 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17321 (_mm_mmask_i32gather_ps): Ditto.
17322 (_mm256_mmask_i32gather_pd): Ditto.
17323 (_mm_mmask_i32gather_pd): Ditto.
17324 (_mm256_mmask_i64gather_ps): Ditto.
17325 (_mm_mmask_i64gather_ps): Ditto.
17326 (_mm256_mmask_i64gather_pd): Ditto.
17327 (_mm_mmask_i64gather_pd): Ditto.
17328 (_mm256_mmask_i32gather_epi32): Ditto.
17329 (_mm_mmask_i32gather_epi32): Ditto.
17330 (_mm256_mmask_i32gather_epi64): Ditto.
17331 (_mm_mmask_i32gather_epi64): Ditto.
17332 (_mm256_mmask_i64gather_epi32): Ditto.
17333 (_mm_mmask_i64gather_epi32): Ditto.
17334 (_mm256_mmask_i64gather_epi64): Ditto.
17335 (_mm_mmask_i64gather_epi64): Ditto.
17336 (_mm256_i32scatter_ps): Change __addr type to void*.
17337 (_mm256_mask_i32scatter_ps): Ditto.
17338 (_mm_i32scatter_ps): Ditto.
17339 (_mm_mask_i32scatter_ps): Ditto.
17340 (_mm256_i32scatter_pd): Ditto.
17341 (_mm256_mask_i32scatter_pd): Ditto.
17342 (_mm_i32scatter_pd): Ditto.
17343 (_mm_mask_i32scatter_pd): Ditto.
17344 (_mm256_i64scatter_ps): Ditto.
17345 (_mm256_mask_i64scatter_ps): Ditto.
17346 (_mm_i64scatter_ps): Ditto.
17347 (_mm_mask_i64scatter_ps): Ditto.
17348 (_mm256_i64scatter_pd): Ditto.
17349 (_mm256_mask_i64scatter_pd): Ditto.
17350 (_mm_i64scatter_pd): Ditto.
17351 (_mm_mask_i64scatter_pd): Ditto.
17352 (_mm256_i32scatter_epi32): Ditto.
17353 (_mm256_mask_i32scatter_epi32): Ditto.
17354 (_mm_i32scatter_epi32): Ditto.
17355 (_mm_mask_i32scatter_epi32): Ditto.
17356 (_mm256_i32scatter_epi64): Ditto.
17357 (_mm256_mask_i32scatter_epi64): Ditto.
17358 (_mm_i32scatter_epi64): Ditto.
17359 (_mm_mask_i32scatter_epi64): Ditto.
17360 (_mm256_i64scatter_epi32): Ditto.
17361 (_mm256_mask_i64scatter_epi32): Ditto.
17362 (_mm_i64scatter_epi32): Ditto.
17363 (_mm_mask_i64scatter_epi32): Ditto.
17364 (_mm256_i64scatter_epi64): Ditto.
17365 (_mm256_mask_i64scatter_epi64): Ditto.
17366 (_mm_i64scatter_epi64): Ditto.
17367 (_mm_mask_i64scatter_epi64): Ditto.
17368 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17369 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17370 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17371 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17372 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17373 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17374 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17375 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17376 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17377 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17378 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17379 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17380 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17381 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17382 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17383 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17384 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17385 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17386 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17387 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17388 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17389 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17390 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17391 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17392 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17393 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17394 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17395 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17396 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17397 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17398 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17399 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17400 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17401 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17402 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17403 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17404 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17405 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17406 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17407 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17408 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17409 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17410 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17411 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17412 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17413 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17414 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17415 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17416 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17417 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17418 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17419 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17420 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17421 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17422 definitions accordingly.
17424 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17425 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17428 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17431 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17434 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17435 hard regs for inner regno.
17437 2017-01-17 Martin Liska <mliska@suse.cz>
17440 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17441 assumption and add comment.
17443 2017-01-17 Nathan Sidwell <nathan@acm.org>
17445 * ipa-visibility.c (localize_node): New function, broken out of ...
17446 (function_and_variable_visibility): ... here. Call it.
17448 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17450 PR middle-end/77445
17451 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17452 correctly set frequency of oudgoing edge.
17453 (duplicate_thread_path): Fix profile updating.
17455 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17458 * configure.ac: Add GCC_BASE_VER.
17459 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17460 version from BASE-VER file.
17461 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17462 (gcc.o): Depend on $(BASEVER).
17463 * common.opt (dumpfullversion): New option.
17464 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17465 * doc/invoke.texi: Document -dumpfullversion.
17466 * doc/install.texi: Document --with-gcc-major-version-only.
17467 * configure: Regenerated.
17469 2017-01-17 Richard Biener <rguenther@suse.de>
17471 PR tree-optimization/71433
17472 * tree-vrp.c (register_new_assert_for): Merge same asserts
17473 on all incoming edges.
17474 (process_assert_insertions_for): Handle insertions at the
17477 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17479 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17480 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17482 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17485 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17488 2017-01-17 Alan Modra <amodra@gmail.com>
17491 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17492 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17493 symbolic stack limit when pic.
17495 2017-01-16 Martin Sebor <msebor@redhat.com>
17497 PR tree-optimization/78608
17498 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17500 2017-01-16 Jeff Law <law@redhat.com>
17503 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17504 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17505 for several include directories that may be relative to sysroot.
17506 * config/i386/x-mingw32 (gplus_includedir): Define.
17507 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17508 (native_system_includedir): Likewise.
17509 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17510 override if TARGET_SYSTEM_ROOT is defined.
17511 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17513 PR tree-optimization/79090
17514 PR tree-optimization/33562
17515 PR tree-optimization/61912
17516 PR tree-optimization/77485
17517 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17518 and computed trims into the dump file.
17520 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17522 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17524 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17527 * gimplify.c (gimplify_init_constructor): If want_value and
17528 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17532 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17533 sequence. Formatting fixes.
17534 (doloop_optimize): Formatting fixes.
17537 * gcc.c (debug_level_greater_than_spec_func): New function.
17538 (static_spec_functions): Add debug-level-gt spec function.
17539 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17541 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17542 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17543 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17544 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17545 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17546 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17548 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17550 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17551 QImode fixups to general and mask registers only.
17553 2017-01-16 Carl Love <cel@us.ibm.com>
17555 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17556 for built-in functions
17557 vector signed char vec_nabs (vector signed char)
17558 vector signed short vec_nabs (vector signed short)
17559 vector signed int vec_nabs (vector signed int)
17560 vector signed long long vec_nabs (vector signed long long)
17561 vector float vec_nabs (vector float)
17562 vector double vec_nabs (vector double)
17563 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17565 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17566 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17567 * doc/extend.texi: Update the documentation file for the new built-in
17570 2017-01-16 Martin Sebor <msebor@redhat.com>
17572 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17575 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17577 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17578 UNSPEC_VSX__XXSPLTD to require special splat handling.
17580 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17583 * system.h: Poison strndup.
17585 2017-01-16 Alan Modra <amodra@gmail.com>
17588 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17591 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17593 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17595 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17597 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17598 call recog here. Assert that INSN_CODE (insn) is non-negative.
17600 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17603 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17605 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17606 in the currently scheduled RTL fragment.
17608 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17610 PR rtl-optimization/78751
17611 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17614 2017-01-14 Jeff Law <law@redhat.com>
17616 PR tree-optimization/79090
17617 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17618 variable length stores.
17619 (compute_trims): Delete dead assignment to *trim_tail.
17620 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17623 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17625 PR rtl-optimization/78626
17626 PR rtl-optimization/78727
17627 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17628 of a block, and split such blocks after everything else is finished.
17630 2017-01-14 Alan Modra <amodra@gmail.com>
17633 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17634 target legitimate_combined_insn.
17635 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17636 (rs6000_legitimate_combined_insn): New function.
17637 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17639 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17640 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17641 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17643 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17645 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17647 2017-01-13 Jeff Law <law@redhat.com>
17649 PR tree-optimization/33562
17650 PR tree-optimization/61912
17651 PR tree-optimization/77485
17652 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17654 (delete_dead_assignment): Likewise.
17655 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17656 statement to delete_dead_call and delete_dead_assignment.
17658 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17661 * substring-locations.c (format_warning_va): Strengthen case 1 so
17662 that both endpoints of the substring must be within the format
17663 range for just the substring to be printed.
17665 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17667 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17668 * config/i386/i386.c (ix86_target_string): Add missing options
17669 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17670 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17671 flags_other and ix86_target_other to flags2_other. Display unknown
17673 (ix86_valid_target_attribute_inner_p): Add missing options and
17674 reorder options by implied ISAs, as in ix86_target_string.
17676 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17678 * hash-table.h (hash_table::too_empty_p): New function.
17679 (hash_table::expand): Use it.
17680 (hash_table::traverse): Likewise.
17681 (hash_table::empty_slot): Use sizeof (value_type) instead of
17682 sizeof (PTR) to convert bytes to elements. Shrink the table
17683 if the current size is excessive for the current number of
17686 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17688 * ira-costs.c (record_reg_classes): Break from the inner loop
17689 early once alt_fail is known to be true. Update outer loop
17690 handling accordingly.
17692 2017-01-13 Jeff Law <law@redhat.com>
17694 * tree-ssa-dse.c (decrement_count): New function.
17695 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17696 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17697 when we know the partially dead statement is a mem* function.
17699 PR tree-optimization/61912
17700 PR tree-optimization/77485
17701 * tree-ssa-dse.c: Include expr.h.
17702 (maybe_trim_constructor_store): New function.
17703 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17705 PR tree-optimization/33562
17706 PR tree-optimization/61912
17707 PR tree-optimization/77485
17708 * doc/invoke.texi: Document new dse-max-object-size param.
17709 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17710 * tree-ssa-dse.c: Include params.h.
17711 (dse_store_status): New enum.
17712 (initialize_ao_ref_for_dse): New, partially extracted from
17714 (valid_ao_ref_for_dse, normalize_ref): New.
17715 (setup_live_bytes_from_ref, compute_trims): Likewise.
17716 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17717 (maybe_trim_partially_dead_store): Likewise.
17718 (maybe_trim_complex_store): Likewise.
17719 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17720 Track what bytes live from the original store. Return tri-state
17721 for dead, partially dead or live.
17722 (dse_dom_walker): Add constructor, destructor and new private members.
17723 (delete_dead_call, delete_dead_assignment): New extracted from
17725 (dse_optimize_stmt): Make a member of dse_dom_walker.
17726 Use initialize_ao_ref_for_dse.
17728 PR tree-optimization/33562
17729 PR tree-optimization/61912
17730 PR tree-optimization/77485
17731 * sbitmap.h (bitmap_count_bits): Prototype.
17732 (bitmap_clear_range, bitmap_set_range): Likewise.
17733 * sbitmap.c (bitmap_clear_range): New function.
17734 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17736 2017-01-13 Martin Liska <mliska@suse.cz>
17739 * function.c (set_cfun): Add new argument force.
17740 * function.h (set_cfun): Likewise.
17741 * ipa-inline-transform.c (inline_call): Use the function when
17742 strict alising from is dropped for function we inline to.
17744 2017-01-13 Richard Biener <rguenther@suse.de>
17746 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17747 for dumping GIMPLE INTEGER_CSTs.
17749 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17751 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17752 to 201112L since C++17.
17754 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17757 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17758 if -fsanitize=kernel-address is present.
17760 2017-01-13 Richard Biener <rguenther@suse.de>
17762 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
17763 as _Literal ( type ) number in case usual suffixes do not
17764 preserve all information.
17766 2017-01-13 Richard Biener <rguenther@suse.de>
17768 PR tree-optimization/77283
17769 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
17770 and ssa-iterators.h.
17771 (is_feasible_trace): Implement a cost model based on joiner
17774 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17777 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
17778 char or short to __float128/_Float128 directly.
17780 2017-01-12 Martin Sebor <msebor@redhat.com>
17782 to -Wformat-overflow.
17783 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
17784 (min_bytes_remaining): Same.
17785 (get_string_length): Same.
17786 (format_string): Same.
17787 (format_directive): Same.
17789 (pass_sprintf_length::handle_gimple_call): Same.
17791 2017-01-12 Jakub Jelinek <jakub@redhat.com>
17793 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
17794 info.nowrite calls with no lhs that can't throw. Return bool
17795 whether gsi_remove has been called or not.
17796 (pass_sprintf_length::handle_gimple_call): Return bool whether
17797 try_substitute_return_value called gsi_remove. Formatting fix.
17798 (pass_sprintf_length::execute): Don't use gsi_remove if
17799 handle_gimple_call returned true.
17802 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
17803 be removed due to side-effects, don't remove following barrier nor
17804 turn the successor edge into fallthru edge.
17806 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17809 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
17810 element-reversing loads and stores as not swappable.
17812 2017-01-12 Nathan Sidwell <nathan@acm.org>
17813 Nicolai Stange <nicstange@gmail.com>
17815 * combine.c (try_combine): Don't ignore result of overlap checking
17816 loop. Combine overlap & asm check into single loop.
17818 2017-01-12 Richard Biener <rguenther@suse.de>
17820 * tree-pretty-print.c (dump_generic_node): Provide -gimple
17821 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
17823 2017-01-12 Richard Biener <rguenther@suse.de>
17825 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
17826 and TS_TARGET_OPTION directly derive from TS_BASE.
17827 * tree-core.h (tree_optimization_option): Derive from tree_base.
17828 (tree_target_option): Likewise.
17830 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
17832 * config/i386/i386.c (memory_address_length): Increase len
17833 only when rip_relative_addr_p returns false.
17835 2017-01-11 Julia Koval <julia.koval@intel.com>
17837 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
17838 (OPTION_MASK_ISA_SGX_SET): New.
17839 (ix86_handle_option): Handle OPT_msgx.
17840 * config.gcc: Added sgxintrin.h.
17841 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
17842 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
17843 * config/i386/i386.c (ix86_target_string): Add -msgx.
17845 (ix86_option_override_internal): Handle new options.
17846 (ix86_valid_target_attribute_inner_p): Add sgx.
17847 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
17848 * config/i386/i386.opt: Add msgx.
17849 * config/i386/sgxintrin.h: New file.
17850 * config/i386/x86intrin.h: Add sgxintrin.h.
17852 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17855 * fold-const.c (maybe_nonzero_address): Return 1 for function
17857 (tree_single_nonzero_warnv_p): Don't handle function local objects
17861 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
17864 2017-01-11 David Malcolm <dmalcolm@redhat.com>
17867 * opts.c: Include "spellcheck.h"
17868 (struct string_fragment): New struct.
17869 (struct edit_distance_traits<const string_fragment &>): New
17871 (get_closest_sanitizer_option): New function.
17872 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
17874 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17876 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
17878 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
17879 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
17880 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
17881 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
17882 for initial die_offset if dwarf_split_debug_info.
17883 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
17884 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
17886 (output_skeleton_debug_sections): Formatting fix. Use
17887 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
17888 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
17890 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
17892 * config/arm/cortex-a53.md: Add bypasses for
17893 cortex_a53_r2f_cvt.
17894 (cortex_a53_r2f): Only use for transfers.
17895 (cortex_a53_f2r): Likewise.
17896 (cortex_a53_r2f_cvt): Add reservation for conversions.
17897 (cortex_a53_f2r_cvt): Likewise.
17899 2017-01-11 Tamar Christina <tamar.christina@arm.com>
17901 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
17902 to all inlined functions, change static to extern.
17904 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
17907 * config/arm/arm.c (legitimize_pic_address): Handle reference to
17909 (arm_assemble_integer): Likewise.
17911 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
17913 * config.gcc: Use new awk script to check CPU, FPU and architecture
17914 parameters for --with-... options.
17915 * config/arm/parsecpu.awk: New file
17916 * config/arm/arm-cpus.in: New file.
17917 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
17919 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
17921 * config/arm/t-arm: Update dependency rules.
17922 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
17923 of processing .def files.
17924 * config/arm/genopt.sh: Deleted.
17925 * config/arm/gentune.sh: Deleted.
17926 * config/arm/arm-cores.def: Deleted.
17927 * config/arm/arm-arches.def: Deleted.
17928 * config/arm/arm-fpus.def: Deleted.
17929 * config/arm/arm-tune.md: Regenerated.
17930 * config/arm/arm-tables.opt: Regenerated.
17931 * config/arm/arm-cpu.h: New generated file.
17932 * config/arm/arm-cpu-data.h: New generated file.
17933 * config/arm/arm-cpu-cdata.h: New generated file.
17935 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
17938 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
17940 (input_varpool_node): Unpack dynamically_initialized bit.
17942 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
17944 PR rtl-optimization/79032
17945 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
17946 the alignment of the adjusted memory reference against that of MODE,
17947 instead of the alignment of the original memory reference.
17949 2017-01-11 Martin Jambor <mjambor@suse.cz>
17951 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
17953 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
17954 decorated functions.
17956 2017-01-11 Richard Biener <rguenther@suse.de>
17958 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
17959 set range/nonnull info for PHI results. Do not set it on
17960 stmts marked for removal.
17962 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
17964 * expr.c (store_field): In the bitfield case, fetch the return value
17965 from the registers before applying a single big-endian adjustment.
17966 Always do a final load for a BLKmode value not larger than a word.
17968 2017-01-10 David Malcolm <dmalcolm@redhat.com>
17971 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
17972 that we correctly handle column numbers greater than
17973 LINE_MAP_MAX_COLUMN_NUMBER.
17975 2017-01-10 Martin Sebor <msebor@redhat.com>
17977 PR middle-end/78245
17978 * gimple-ssa-sprintf.c (get_destination_size): Call
17979 {init,fini}object_sizes.
17980 * tree-object-size.c (addr_object_size): Adjust.
17981 (pass_through_call): Adjust.
17982 (pass_object_sizes::execute): Adjust.
17983 * tree-object-size.h (fini_object_sizes): Declare.
17985 2017-01-10 Martin Sebor <msebor@redhat.com>
17987 PR tree-optimization/78775
17988 * builtins.c (get_size_range): Move...
17989 * calls.c: ...to here.
17990 (alloc_max_size): Accept zero argument.
17991 (operand_signed_p): Remove.
17992 (maybe_warn_alloc_args_overflow): Call get_size_range.
17993 * calls.h (get_size_range): Declare.
17995 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
17997 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
17998 from TI's devices.csv file as of September 2016.
17999 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18001 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18003 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18004 * doc/invoke.texi: Likewise.
18005 * doc/md.texi: Likewise.
18006 * doc/objc.texi: Likewise.
18008 2017-01-10 Joshua Conner <joshconner@google.com>
18010 * config/arm/fuchsia-elf.h: New file.
18011 * config/fuchsia.h: New file.
18012 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18013 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18015 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18017 2016-01-10 Richard Biener <rguenther@suse.de>
18019 PR tree-optimization/79034
18020 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18021 Propagate out degenerate PHIs in the joiner.
18023 2017-01-10 Martin Liska <mliska@suse.cz>
18025 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18026 (sort_congruence_classes_by_decl_uid): Likewise.
18027 (sort_congruence_class_groups_by_decl_uid): Likewise.
18028 (sem_item_optimizer::merge_classes): Sort class, groups in these
18029 classes and members in the groups by DECL_UID of declarations.
18030 This would make merge operations stable.
18032 2017-01-10 Martin Liska <mliska@suse.cz>
18034 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18035 usage of m_classes_vec.
18036 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18037 (sem_item_optimizer::get_group_by_hash): Likewise.
18038 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18039 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18040 (sem_item_optimizer::verify_classes): Likewise.
18041 (sem_item_optimizer::process_cong_reduction): Likewise.
18042 (sem_item_optimizer::dump_cong_classes): Likewise.
18043 (sem_item_optimizer::merge_classes): Likewise.
18044 * ipa-icf.h (congruence_class_hash): Rename from
18045 congruence_class_group_hash. Remove declaration of m_classes_vec.
18047 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18049 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18050 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18051 * config.gcc: Add avx512vpopcntdqintrin.h.
18052 * config/i386/avx512vpopcntdqintrin.h: New.
18053 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18054 * config/i386/i386-builtin-types.def: Add new types.
18055 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18056 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18057 __builtin_ia32_vpopcountq_v8di_mask): New.
18058 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18059 __AVX512VPOPCNTDQ__.
18060 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18061 (PTA_AVX512VPOPCNTDQ): Define.
18062 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18063 TARGET_AVX512VPOPCNTDQ_P): Define.
18064 * config/i386/i386.opt: Add mavx512vpopcntdq.
18065 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18066 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18068 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18070 PR middle-end/77484
18071 * predict.def (PRED_CALL): Set to 67.
18073 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18075 * expr.c (store_field): In the bitfield case, if the value comes from
18076 a function call and is of an aggregate type returned in registers, do
18077 not modify the field mode; extract the value in all cases if the mode
18078 is BLKmode and the size is not larger than a word.
18080 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18083 * config/i386/cpuid.h: Fix undefined behavior.
18085 2017-01-04 Jeff Law <law@redhat.com>
18087 PR tree-optimization/79007
18088 PR tree-optimization/67955
18089 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18090 conservative for pt.null when flag_non_call_exceptions is on.
18092 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18094 PR translation/79019
18095 PR translation/79020
18096 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18097 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18098 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18100 * config/avr/avr.opt (maccumulate-args): Likewise.
18101 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18102 * common.opt (freport-bug): Likewise.
18103 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18104 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18105 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18106 translatable string.
18107 * config/i386/i386.c (function_value_32): Likewise.
18108 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18109 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18111 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18112 * common/config/msp430/msp430-common.c (msp430_handle_option):
18114 * symtab.c (symtab_node::verify_base): Likewise.
18115 * opts.c (set_debug_level): Likewise.
18116 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18117 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18118 missing whitespace to translatable strings.
18119 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18120 * config/sh/superh.h: Likewise.
18121 * config/i386/xopintrin.h: Likewise.
18122 * config/i386/znver1.md: Likewise.
18123 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18124 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18125 * double-int.h (struct double_int): Likewise.
18126 * double-int.c (div_and_round_double): Likewise.
18127 * wide-int.cc: Likewise.
18128 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18129 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18130 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18131 (crossjumps_occurred): ... this.
18132 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18135 PR tree-optimization/78899
18136 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18137 returning bool return struct loop *, NULL for failure and the new
18139 (versionable_outer_loop_p): Don't version outer loop if it has
18140 dont_vectorized bit set.
18141 (tree_if_conversion): When versioning outer loop, ensure
18142 tree_if_conversion is performed also on the inner loop of the
18143 non-vectorizable outer loop copy.
18144 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18145 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18146 prevent vectorization of it.
18147 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18148 the outer loop vectorization of the non-scalar version is attempted
18149 before vectorization of the inner loop in scalar version. If
18150 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18151 vectorization of its inner loop.
18152 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18153 has 2 inner loops, rename also on edges from bb whose single pred
18154 is outer_loop->header. Fix typo in function comment.
18156 2017-01-09 Martin Sebor <msebor@redhat.com>
18159 * asan.c (asan_emit_stack_protection): Increase local buffer size
18160 to avoid snprintf truncation warning.
18162 2017-01-09 Andrew Pinski <apinski@cavium.com>
18164 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18165 to reference thunderx2t99 for the tuning structure
18166 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18168 (thunderx2t99_extra_costs): This.
18169 * config/aarch64/aarch64-tune.md: Regenerate.
18170 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18171 (vulcan_addrcost_table): This.
18172 (vulcan_regmove_cost): Rename to ...
18173 (thunderx2t99_regmove_cost): This.
18174 (vulcan_vector_cost): Rename to ...
18175 (thunderx2t99_vector_cost): this.
18176 (vulcan_branch_cost): Rename to ...
18177 (thunderx2t99_branch_cost): This.
18178 (vulcan_tunings): Rename to ...
18179 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18180 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18182 2017-01-09 Martin Jambor <mjambor@suse.cz>
18186 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18187 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18188 (propagate_vr_accross_jump_function): Use the above function for all
18189 value range computations for pass-through jump functions and type
18190 converasion from explicit value range values.
18191 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18192 parameters from TYPE_ARG_TYPES.
18193 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18194 (ipa_write_node_info): Stream type of the actual argument.
18195 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18197 2017-01-09 Martin Liska <mliska@suse.cz>
18200 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18201 (lookup_compiler): Do not show error message with have_E.
18203 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18205 PR tree-optimization/78938
18206 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18207 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18208 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18209 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18212 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18214 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18217 2017-01-09 Richard Biener <rguenther@suse.de>
18219 PR tree-optimization/78997
18220 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18221 name condition properly.
18223 2017-01-09 Richard Biener <rguenther@suse.de>
18226 * dwarf2out.c (is_cxx): New overload with context.
18227 (is_naming_typedef_decl): Use it.
18229 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18231 * invoke.texi (Option Summary): Correct spacing in option lists
18232 and add line breaks to fix over-long lines.
18234 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18236 PR middle-end/17660
18238 * extend.texi (Common Variable Attributes): Add xref to GCC
18239 Internals manual to explain mode attribute keywords.
18241 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18244 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18245 and Preprocessor Options.
18246 (Options for Linking): Document -pthread here....
18247 (RS/6000 and PowerPC Options): ...not here.
18248 (Solaris 2 Options): ...or here.
18249 * doc/cppopts.texi: Document -pthread.
18251 2017-01-08 Martin Sebor <msebor@redhat.com>
18253 PR middle-end/77708
18254 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18255 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18256 New member functions.
18257 (format_directive): Used them.
18259 (pass_sprintf_length::handle_gimple_call): Same.
18260 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18261 to avoid truncation for any argument.
18262 (extract_affine_mul): Same.
18263 * tree.c (get_file_function_name): Same.
18265 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18267 PR middle-end/77484
18268 * predict.def (PRED_INDIR_CALL): Set to 86.
18270 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18272 PR preprocessor/54124
18273 * doc/cppopts.texi: Reformat -d subtable to list the full name
18274 of the options. Add cross-reference to the docs for the general
18275 compiler -d options.
18276 * doc/invoke.texi (Developer Options): Add cross-reference to the
18277 preprocessor-specific -d option documentation.
18279 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18281 PR preprocessor/13498
18282 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18283 redudant material, and reflect new command-line options.
18284 (System Headers): Likewise.
18286 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18288 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18289 -isystem, and -idirafter. Copy-edit.
18290 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18291 default for -ftrack-macro-expansion. Delete obsolete and
18292 badly-formatted implementation details about -fdebug-cpp output.
18293 * doc/cppwarnopts.texi: Copy-edit.
18295 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18298 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18299 that the transition from a max line width >= 1<<10 to narrower
18300 lines works correctly.
18302 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18304 * doc/options.texi (PerFunction): New.
18305 * opt-functions.awk (switch_flags): Map both Optimization and
18306 PerFunction to CL_OPTIMIZATION.
18307 * opth-gen.awk: Test for PerFunction flag along with
18309 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18310 it only when the latter is present. Skip those that don't in
18311 the hash function generator.
18312 * common.opt (fvar-tracking): Mark as PerFunction instead of
18314 (fvar-tracking-assignments): Likewise.
18315 (fvar-tracking-assignments-toggle): Likewise.
18316 (fvar-tracking-uninit): Likewise.
18318 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18320 PR translation/79018
18321 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18324 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18327 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18328 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18329 TARGET_LONG_JUMP_TABLE_OFFSETS.
18330 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18331 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18332 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18333 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18334 * config/m68k/m68k.md (tablejump expander): Likewise.
18335 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18336 TARGET_LONG_JUMP_TABLE_OFFSETS.
18337 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18338 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18340 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18341 David Holsgrove <david.holsgrove@xilinx.com>
18343 * common/config/microblaze/microblaze-common.c
18344 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18345 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18347 * config/microblaze/microblaze.c (microblaze_must_save_register)
18348 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18350 (microblaze_eh_return): New function.
18351 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18352 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18353 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18354 * config/microblaze/microblaze.md (eh_return): New pattern.
18356 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18358 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18359 GCC_DIAGNOSTIC_STRINGIFY): Define.
18361 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18363 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18365 * config/arm/arm.md (<mcrr>): New.
18367 * config/arm/arm.c (arm_arch5te): New.
18368 (arm_option_override): Set arm_arch5te.
18369 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18371 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18372 (arm_mcrr_qualifiers): ... this. New.
18373 (MRRC_QUALIFIERS): Define to...
18374 (arm_mrrc_qualifiers): ... this. New.
18375 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18377 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18378 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18379 (MRRCI, mrrc, MRRC): New.
18380 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18381 VUNSPEC_MRRC2): New.
18383 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18385 * config/arm/arm.md (<mcr>): New.
18387 * config/arm/arm.c (arm_coproc_builtin_available): Add
18388 support for mcr, mrc, mcr2 and mrc2.
18389 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18390 (arm_mcr_qualifiers): ... this. New.
18391 (MRC_QUALIFIERS): Define to ...
18392 (arm_mrc_qualifiers): ... this. New.
18393 (MCR_QUALIFIERS): Define to ...
18394 (arm_mcr_qualifiers): ... this. New.
18395 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18397 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18398 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18399 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18400 VUNSPEC_MRC2): New.
18402 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18404 * config/arm/arm.md (*ldc): New.
18408 * config/arm/arm.c (arm_coproc_builtin_available): Add
18409 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18410 (arm_coproc_ldc_stc_legitimate_address): New.
18411 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18412 'qualifier_const_pointer'.
18413 (LDC_QUALIFIERS): Define to...
18414 (arm_ldc_qualifiers): ... this. New.
18415 (STC_QUALIFIERS): Define to...
18416 (arm_stc_qualifiers): ... this. New.
18417 * config/arm/arm-protos.h
18418 (arm_coproc_ldc_stc_legitimate_address): New.
18419 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18420 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18421 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18422 stc2, stcl, stc2l): New.
18423 * config/arm/constraints.md (Uz): New.
18424 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18425 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18426 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18427 VUNSPEC_STC2L): New.
18429 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18431 * config/arm/arm.md (<cdp>): New.
18432 * config/arm/arm.c (neon_const_bounds): Rename this ...
18433 (arm_const_bounds): ... this.
18434 (arm_coproc_builtin_available): New.
18435 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18436 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18437 (CDP_QUALIFIERS): Define to...
18438 (arm_cdp_qualifiers): ... this. New.
18440 (arm_expand_builtin_args): Add case for 6 arguments.
18441 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18442 (arm_const_bounds): ... this.
18443 (arm_coproc_builtin_available): New.
18444 * config/arm/arm_acle.h (__arm_cdp): New.
18446 * config/arm/arm_acle_builtins.def (cdp): New.
18448 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18449 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18450 'arm_const_bounds'.
18451 * config/arm/types.md (coproc): New.
18452 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18453 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18454 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18455 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18457 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18459 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18460 (UBINOP_QUALIFIERS): New.
18462 (acle_builtin_data): New. Change comment.
18463 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18464 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18465 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18466 arm_acle_builtins.def.
18467 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18468 (arm_init_acle_builtins): New.
18469 (CRC32_BUILTIN): Remove.
18470 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18471 crc32cb, crc32ch and crc32cw.
18472 (arm_init_crc32_builtins): Remove.
18473 (arm_init_builtins): Use arm_init_acle_builtins rather
18474 than arm_init_crc32_builtins.
18475 (arm_expand_acle_builtin): New.
18476 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18477 * config/arm/arm_acle_builtins.def: New.
18479 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18481 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18482 (arm_builtin_datum): ... this.
18483 (arm_init_neon_builtin): Rename to ...
18484 (arm_init_builtin): ... this. Add a new parameters PREFIX
18485 and USE_SIG_IN_NAME.
18486 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18487 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18488 'arm_builtin_datum'.
18489 (arm_init_vfp_builtins): Likewise.
18490 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18491 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18492 (arm_expand_neon_args): Rename to ...
18493 (arm_expand_builtin_args): ... this. Rename builtin_arg
18494 enum values and differentiate between ARG_BUILTIN_MEMORY
18495 and ARG_BUILTIN_NEON_MEMORY.
18496 (arm_expand_neon_builtin_1): Rename to ...
18497 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18498 values, arm_expand_builtin_args and add bool parameter NEON.
18499 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18500 (arm_expand_vfp_builtin): Likewise.
18501 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18503 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18505 PR middle-end/77484
18506 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18507 * predict.c (tree_estimate_probability_bb): Reverse direction of
18508 polymorphic call predictor.
18510 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18512 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18513 (determine_pass_name_match): ...this new function and...
18514 (should_skip_pass_p): ...this new function.
18516 2017-01-06 Nathan Sidwell <nathan@acm.org>
18518 * ipa-visibility.c (function_and_variable_visibility): Reformat
18519 comments and long lines. Remove extrneous if.
18520 * symtab.c (symtab_node::make_decl_local): Fix code format.
18521 (symtab_node::set_section_for_node): Fix comment typo.
18523 2017-01-06 Martin Liska <mliska@suse.cz>
18526 * lra-constraints.c: Rename invariant to lra_invariant.
18527 * predict.c (set_even_probabilities): Initialize e to NULL.
18529 2017-01-05 Martin Sebor <msebor@redhat.com>
18531 PR tree-optimization/78910
18532 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18533 (format_integer): Correct off-by-one error in the handling
18534 of precision with negative numbers in signed conversions..
18536 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18538 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18540 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18542 PR tree-optimization/71016
18543 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18544 factor_out_conditional_conversion. Formatting fix.
18545 (factor_out_conditional_conversion): Add cond_stmt argument.
18546 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18547 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18550 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18552 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18553 read-rtl-function.o, and selftest-rtl.o.
18554 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18555 (selftest::aarch64_test_loading_full_dump): New function.
18556 (selftest::aarch64_run_selftests): New function.
18557 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18558 selftest::aarch64_run_selftests.
18559 * config/i386/i386.c
18560 (selftest::ix86_test_loading_dump_fragment_1): New function.
18561 (selftest::ix86_test_loading_call_insn): New function.
18562 (selftest::ix86_test_loading_full_dump): New function.
18563 (selftest::ix86_test_loading_unspec): New function.
18564 (selftest::ix86_run_selftests): Call the new functions.
18565 * emit-rtl.c (maybe_set_max_label_num): New function.
18566 * emit-rtl.h (maybe_set_max_label_num): New decl.
18567 * function.c (instantiate_decls): Guard call to
18568 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18569 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18571 * gensupport.c (gen_reader::gen_reader): Pass "false"
18572 for new "compact" param of rtx_reader.
18573 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18574 rather than an empty string for NULL strings.
18575 * read-md.c: Potentially include config.h rather than bconfig.h.
18576 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18577 (have_error): New global, copied from errors.c.
18578 (md_reader::read_name): Rename to...
18579 (md_reader::read_name_1): ...this, adding "out_loc" param,
18580 and converting "missing name or number" to returning false, rather
18582 (md_reader::read_name): Reimplement in terms of read_name_1.
18583 (md_reader::read_name_or_nil): New function.
18584 (md_reader::read_string): Handle "(nil)" by returning NULL.
18585 (md_reader::md_reader): Add new param "compact".
18586 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18587 (md_reader::read_file): New method.
18588 * read-md.h (md_reader::md_reader): Add new param "compact".
18589 (md_reader::read_file): New method.
18590 (md_reader::is_compact): New accessor.
18591 (md_reader::read_name): Convert return type from void to file_location.
18592 (md_reader::read_name_or_nil): New decl.
18593 (md_reader::read_name_1): New decl.
18594 (md_reader::m_compact): New field.
18595 (noop_reader::noop_reader): Pass "false" for new "compact" param
18597 (rtx_reader::rtx_reader): Add new "compact" param.
18598 (rtx_reader::read_rtx_operand): Make virtual and convert return
18599 type from void to rtx.
18600 (rtx_reader::read_until): New decl.
18601 (rtx_reader::handle_any_trailing_information): New virtual function.
18602 (rtx_reader::postprocess): New virtual function.
18603 (rtx_reader::finalize_string): New virtual function.
18604 (rtx_reader::m_in_call_function_usage): New field.
18605 (rtx_reader::m_reuse_rtx_by_id): New field.
18606 * read-rtl-function.c: New file.
18607 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18608 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18609 (selftest::verify_three_block_rtl_cfg): New decl.
18610 * read-rtl-function.h: New file.
18611 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18612 For host, include function.h, memmodel.h, and emit-rtl.h.
18613 (one_time_initialization): New function.
18614 (struct compact_insn_name): New struct.
18615 (compact_insn_names): New array.
18616 (find_code): Handle insn codes in compact dumps.
18617 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18618 (bind_subst_iter_and_attr): Likewise.
18619 (add_condition_to_string): Likewise.
18620 (add_condition_to_rtx): Likewise.
18621 (apply_attribute_uses): Likewise.
18622 (add_current_iterators): Likewise.
18623 (apply_iterators): Likewise.
18624 (initialize_iterators): Guard usage of apply_subst_iterator with
18625 #ifdef GENERATOR_FILE.
18626 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18627 (md_reader::read_mapping): Likewise.
18628 (add_define_attr_for_define_subst): Likewise.
18629 (add_define_subst_attr): Likewise.
18630 (read_subst_mapping): Likewise.
18631 (check_code_iterator): Likewise.
18632 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18634 (one_time_initialization): New function.
18635 (rtx_reader::read_until): New method.
18636 (read_flags): New function.
18637 (parse_reg_note_name): New function.
18638 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18639 Handle reuse_rtx ids.
18640 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18641 Add parsing support for RTL dumps, mirroring the special-cases in
18642 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18643 values, and calling handle_any_trailing_information.
18644 (rtx_reader::read_rtx_operand): Convert return type from void
18645 to rtx, returning return_rtx. Handle case 'e'. Call
18646 finalize_string on XSTR and XTMPL fields.
18647 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18648 "(nil)" values were omitted. Call the postprocess vfunc on the
18650 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18651 class ctor. Initialize m_in_call_function_usage. Call
18652 one_time_initialization.
18653 * rtl-tests.c (selftest::test_uncond_jump): Call
18654 set_new_first_and_last_insn.
18655 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18656 * selftest-rtl.c: New file.
18657 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18658 (selftest::get_insn_by_uid): New decl.
18659 * selftest-run-tests.c (selftest::run_tests): Call
18660 read_rtl_function_c_tests.
18661 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18662 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18665 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18667 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18668 operands in a special way. Assert that pos+len <= mode precision.
18670 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18672 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18673 3 argument Alias with unlimited for the negative form.
18674 (fno-vect-cost-model): Removed.
18676 2017-01-05 Martin Liska <mliska@suse.cz>
18678 * hsa-gen.c (gen_hsa_divmod): New function.
18679 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18681 2017-01-05 Martin Liska <mliska@suse.cz>
18684 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18687 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18689 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18690 small constant length operands.
18692 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18694 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18695 between loop iterations.
18697 2017-01-05 Martin Liska <mliska@suse.cz>
18700 * gimplify.c (gimplify_decl_expr): Compare to
18701 asan_poisoned_variables instread of checking flags.
18702 (gimplify_target_expr): Likewise.
18703 (gimplify_expr): Likewise.
18704 (gimplify_function_tree): Conditionally initialize
18705 asan_poisoned_variables.
18707 2017-01-04 Jeff Law <law@redhat.com>
18709 PR tree-optimizatin/78812
18710 * rtl.h (contains_mem_rtx_p): Prototype.
18711 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18712 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18713 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18714 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18715 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18717 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18719 * input.c (assert_char_at_range): Default-initialize actual_range.
18721 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18723 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18724 to match the caller.
18726 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18728 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18729 insns after final jump in test to emit dummy move.
18731 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18733 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18734 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18736 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18738 * multiple_target.c (create_dispatcher_calls): Init e_next.
18739 * tree-ssa-loop-split.c (split_loop): Init border.
18740 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18743 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18748 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18749 (altivec_register_operand): Do not return true if the operand
18750 contains a SUBREG mixing SImode and SFmode.
18751 (vsx_register_operand): Likewise.
18752 (vsx_reg_sfsubreg_ok): New predicate.
18753 (vfloat_operand): Do not return true if the operand contains a
18754 SUBREG mixing SImode and SFmode.
18755 (vint_operand): Likewise.
18756 (vlogical_operand): Likewise.
18757 (gpc_reg_operand): Likewise.
18758 (int_reg_operand): Likewise.
18759 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18760 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18761 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
18763 (rs6000_emit_move_si_sf_subreg): New helper function.
18764 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
18765 fixup SUBREGs involving SImode and SFmode.
18766 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
18767 numbers for the new peephole2 optimization.
18768 (peephole2 for SFmode unions): New peephole2 to optimize cases in
18769 the GLIBC math library that do AND/IOR/XOR operations on single
18770 precision floating point.
18771 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
18772 target macros to say whether we need to avoid SUBREGs mixing
18774 (TARGET_ALLOW_SF_SUBREG): Likewise.
18775 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
18776 (UNSPEC_SI_FROM_SF): Likewise.
18777 (iorxor): Change spacing.
18778 (and_ior_xor): New iterator for AND, IOR, and XOR.
18779 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
18780 (movdi_from_sf_zero_ext): Likewise.
18781 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
18782 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
18783 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
18784 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
18785 (fms<mode>4): Likewise.
18786 (fnma<mode>4): Likewise.
18787 (fnms<mode>4): Likewise.
18788 (nfma<mode>4): Likewise.
18789 (nfms<mode>4): Likewise.
18791 2017-01-04 Marek Polacek <polacek@redhat.com>
18794 * doc/invoke.texi: Document -Wpointer-compare.
18796 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18798 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
18801 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
18802 descriptions for -gdwarf-5 and emit them as uleb128 instead of
18805 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
18808 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
18809 documentation of the powerpc_popcntb_ok attribute.
18810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18811 code to issue warning messages if a requested CPU configuration is
18812 not supported by the binary (assembler and loader) toolchain.
18813 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
18814 made to define a built-in function that has been disabled.
18815 (paired_init_builtins): Add assertion to prevent ICE if attempt is
18816 made to define a built-in function that has been disabled.
18817 (altivec_init_builtins): Add comment explaining why definition
18818 of the DST built-in functions is not preceded by an assertion
18819 check. Add assertions to prevent ICE if attempts are made to
18820 define an altivec predicate or an abs* built-in function that has
18822 (htm_init_builtins): Add comment explaining why definition of the
18823 htm built-in functions is not preceded by an assertion check.
18825 2017-01-04 Jeff Law <law@redhat.com>
18827 PR tree-optimizatin/67955
18828 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
18829 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
18830 the points-to solution does not include pt_null. Use DECL_PT_UID
18833 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
18835 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
18836 Use gen_int_mode instead of gen_lopwart for const_int operands.
18838 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18840 PR tree-optimization/71563
18841 * match.pd: Simplify X << Y into X if Y is known to be 0 or
18842 out of range value - has low bits known to be zero.
18844 2017-01-04 Alan Modra <amodra@gmail.com>
18846 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
18847 * configure: Regenerate.
18848 * config.in: Regenerate.
18850 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18853 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
18854 a substring of the message, but strcmp with the whole message. Ifdef
18855 ENABLE_NLS, translate the message first using dgettext.
18857 2017-01-03 Jeff Law <law@redhat.com>
18859 PR tree-optimizatin/78856
18860 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
18861 (mark_threaded_blocks): Remove code to truncate thread paths that
18862 cross multiple loop headers. Instead invalidate the cached loop
18863 iteration information and handle case of a thread path walking
18864 into an irreducible region.
18866 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
18869 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
18870 assertions. Add support for doing the signbit if the IEEE 128-bit
18871 floating point value is in a GPR.
18872 * config/rs6000/rs6000.md (Fsignbit): Delete.
18873 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
18874 Update the length attribute if the value is in a GPR.
18875 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
18876 the sign or zero extension instruction, since the value is always 0/1.
18877 (signbit<mode>2_dm2): Delete using <Fsignbit>.
18880 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
18881 extracting SImode to a GPR register so that we can generate a
18882 store, limit the vector to be in a traditional Altivec register
18883 for the vextuwrx instruction.
18885 2017-01-03 Ian Lance Taylor <iant@google.com>
18887 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
18889 2017-01-03 Martin Sebor <msebor@redhat.com>
18891 PR tree-optimization/78696
18892 * gimple-ssa-sprintf.c (format_floating): Correct handling of
18893 precision. Use MPFR for %f for greater fidelity. Correct handling
18895 (pass_sprintf_length::compute_format_length): Set width and precision
18896 specified by asrerisk to void_node for vararg functions.
18897 (try_substitute_return_value): Adjust dump output.
18899 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
18901 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
18903 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
18905 * doc/invoke.texi (SPARC options): Document -mlra as the default.
18906 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
18907 -mlra/-mno-lra was passed to the compiler.
18909 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
18911 PR rtl-optimization/65618
18912 * emit-rtl.c (try_split): Move initialization of "before" and
18913 "after" to just before the call to emit_insn_after_setloc.
18915 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
18917 * doc/md.texi (Standard Names): Remove reference to Java frontend.
18919 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
18921 * dwarf2out.c (gen_enumeration_type_die): When
18922 -gno-strict-dwarf, add a DW_AT_encoding attribute.
18924 2017-01-03 Jakub Jelinek <jakub@redhat.com>
18926 PR tree-optimization/78965
18927 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
18928 Change first argument from const call_info & to call_info &. For %n
18929 set info.nowrite to false.
18931 PR middle-end/78901
18932 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
18933 possibly throwing calls.
18935 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
18936 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
18937 and fns handling, rather than in a separate case SSA_NAME.
18939 2017-01-02 Jeff Law <law@redhat.com>
18941 * config/darwin-driver.c (darwin_driver_init): Const-correctness
18942 fixes for first_period and second_period variables.
18944 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
18947 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
18948 (*insvqi_1): New insn pattern.
18949 (*insvqi_1_mem_rex64): Ditto.
18950 (*insvqi_2): Ditto.
18951 (*insvqi_3): Rename from *insvqi.
18953 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
18955 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
18957 * doc/cfg.texi (Edges): Remove reference to Java.
18958 (Maintaining the CFG): Ditto.
18960 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18962 PR middle-end/77674
18963 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
18964 transparent aliases.
18966 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18968 PR middle-end/77484
18969 * predict.def (PRED_CALL): Update hitrate.
18970 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
18971 * predict.c (tree_estimate_probability_bb): Split CALL predictor
18972 into direct/indirect/polymorphic variants.
18974 2017-01-01 Jakub Jelinek <jakub@redhat.com>
18976 Update copyright years.
18978 * gcc.c (process_command): Update copyright notice dates.
18979 * gcov-dump.c (print_version): Ditto.
18980 * gcov.c (print_version): Ditto.
18981 * gcov-tool.c (print_version): Ditto.
18982 * gengtype.c (create_file): Ditto.
18983 * doc/cpp.texi: Bump @copying's copyright year.
18984 * doc/cppinternals.texi: Ditto.
18985 * doc/gcc.texi: Ditto.
18986 * doc/gccint.texi: Ditto.
18987 * doc/gcov.texi: Ditto.
18988 * doc/install.texi: Ditto.
18989 * doc/invoke.texi: Ditto.
18991 Copyright (C) 2017 Free Software Foundation, Inc.
18993 Copying and distribution of this file, with or without modification,
18994 are permitted in any medium without royalty provided the copyright
18995 notice and this notice are preserved.