1 2017-07-25 Andrew Pinski <apinski@cavium.com>
3 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4 accesses where the use is for the first operand of a BIT_INSERT.
6 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14 * config/i386/gstabs.h: Delete.
15 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
17 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19 * config/i386/i386.c (ix86_decompose_address): Do not check for
20 register RTX when looking at index_reg or base_reg.
21 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
25 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
26 to update EH info here.
28 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
30 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
32 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
34 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
36 2017-07-25 Torsten Duwe <duwe@suse.de>
38 * common.opt: Introduce -fpatchable-function-entry
39 command line option, and its variables function_entry_patch_area_size
40 and function_entry_patch_area_start.
41 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
42 including a two-value parser.
43 * target.def (print_patchable_function_entry): New target hook.
44 * targhooks.h (default_print_patchable_function_entry): New function.
45 * targhooks.c (default_print_patchable_function_entry): Likewise.
46 * toplev.c (process_options): Switch off IPA-RA if
47 patchable function entries are being generated.
48 * varasm.c (assemble_start_function): Look at the
49 patchable-function-entry command line switch and current
50 function attributes and maybe generate NOP instructions by
51 calling the print_patchable_function_entry hook.
52 * doc/extend.texi: Document patchable_function_entry attribute.
53 * doc/invoke.texi: Document -fpatchable_function_entry
55 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
57 * doc/tm.texi: Re-generate.
59 2017-07-25 Jakub Jelinek <jakub@redhat.com>
62 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
63 TARGET_AVX512DQ rather than TARGET_AVX512BW.
65 2017-07-25 Tamar Christina <tamar.christina@arm.com>
67 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
69 2017-07-25 Richard Biener <rguenther@suse.de>
71 PR tree-optimization/81455
72 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
73 not walk in cycles when looking for guards.
75 2017-07-25 Richard Biener <rguenther@suse.de>
77 PR tree-optimization/81529
78 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
79 when optimizing backedge uses.
81 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
83 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
85 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
86 to dl_section_ref. On AIX, append an expression to subtract
87 the size of the section length to dl_section_ref.
89 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
91 * configure.ac: If any of the config.* scripts fail, exit 1.
92 * configure: Regenerate.
94 2017-07-25 Richard Biener <rguenther@suse.de>
97 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
98 of immediate uses, be more verbose on errors.
100 2017-07-25 Richard Biener <rguenther@suse.de>
102 PR tree-optimization/81510
103 * tree-vect-loop.c (vect_is_simple_reduction): When the
104 reduction stmt is not inside the loop bail out.
106 2017-07-25 Richard Biener <rguenther@suse.de>
108 PR tree-optimization/81303
109 * tree-vect-loop-manip.c (vect_loop_versioning): Build
110 profitability check against LOOP_VINFO_NITERSM1.
112 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
114 * domwalk.c (cmp_bb_postorder): Simplify.
115 (sort_bbs_postorder): New function. Use it...
116 (dom_walker::walk): ...here to optimize common cases.
118 2017-07-25 Martin Liska <mliska@suse.cz>
121 * ipa-visibility.c (function_and_variable_visibility): Make the
122 redirection just on target that supports aliasing.
123 Fix GNU coding style.
125 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
128 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
130 * config/arm/rtems.h: Likewise.
131 * config/bfin/rtems.h: Likewise.
132 * config/i386/rtemself.h: Likewise.
133 * config/lm32/rtems.h: Likewise.
134 * config/m32c/rtems.h: Likewise.
135 * config/m68k/rtemself.h: Likewise.
136 * config/microblaze/rtems.h: Likewise.
137 * config/mips/rtems.h: Likewise.
138 * config/moxie/rtems.h: Likewise.
139 * config/nios2/rtems.h: Likewise.
140 * config/powerpcspe/rtems.h: Likewise.
141 * config/rs6000/rtems.h: Likewise.
142 * config/rtems.h: Likewise.
143 * config/sh/rtems.h: Likewise.
144 * config/sh/rtemself.h: Likewise.
145 * config/sparc/rtemself.h: Likewise.
147 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
150 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
151 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
152 * tree-ssa-structalias.c (alias_get_name): Same.
154 2017-07-25 Bin Cheng <bin.cheng@arm.com>
157 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
158 instructions if no du chain is found.
160 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
162 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
164 2017-07-25 Richard Biener <rguenther@suse.de>
167 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
170 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
172 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
174 (ISA_2_7_MASKS_SERVER): Likewise.
175 (ISA_3_0_MASKS_IEEE): Likewise.
176 (OTHER_P8_VECTOR_MASKS): Likewise.
177 (OTHER_VSX_VECTOR_MASKS): Likewise.
178 (POWERPC_MASKS): Likewise.
179 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
180 duplicate list of options.
181 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
182 explicit -mupper-regs options.
183 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
184 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
185 alias for -mupper-regs-df.
186 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
187 (rs6000_init_hard_regno_mode_ok): Likewise.
188 (rs6000_option_override_internal): Likewise.
189 (rs6000_opt_masks): Likewise.
190 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
191 options in terms of whether -mvsx or -mpower8-vector was used.
192 (TARGET_UPPER_REGS_DI): Likewise.
193 (TARGET_UPPER_REGS_SF): Likewise.
194 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
195 -mupper-regs-* options.
197 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
199 * passes.c (emergency_dump_function): Print some empty lines and a
200 header before the RTL dump.
202 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
204 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
206 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
209 * config/aarch64/aarch64.c (aarch64_classify_symbol):
210 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
212 2017-07-24 Carl Love <cel@us.ibm.com>
214 * config/rs6000/rs6000-c.c: Add support for built-in functions
215 vector float vec_extract_fp32_from_shorth (vector unsigned short);
216 vector float vec_extract_fp32_from_shortl (vector unsigned short);
217 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
218 vec_extract_fp_from_shortl): Add defines for the two builtins.
219 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
220 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
222 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
223 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
224 * doc/extend.texi: Update the built-in documentation file for the
225 new built-in function.
227 2017-07-24 Jakub Jelinek <jakub@redhat.com>
230 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
232 * doc/generic.texi: Likewise.
233 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
234 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
236 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
238 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
239 (aarch64_mls_elt_merge<mode>): Likewise.
241 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
243 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
246 2017-07-23 Michael Collison <michael.collison@arm.com>
248 * config/arm/arm.c (arm_option_override): Deprecate
249 use of -mstructure-size-boundary.
250 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
251 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
253 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
256 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
257 Reduce cost estimate for direct moves.
259 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
262 * config/i386/i386.c (ix86_option_override_internal): Disable
263 BMI, BMI2 and TBM instructions for -m16.
265 2017-07-21 Carl Love <cel@us.ibm.com>
267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
268 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
269 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
270 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
271 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
272 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
273 VMULOSW): New enum "unspec" values.
274 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
275 altivec_vmulosw): New patterns.
276 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
277 VMULOSW): Add definitions.
279 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
281 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
283 * config/aarch64/aarch64-options-extensions.def (rdma); New.
284 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
285 (AARCH64_FL_V8_1): Renumber.
286 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
287 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
288 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
289 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
290 rdma to feature modifiers list.
292 2017-07-21 Yury Gribov <tetra2005@gmail.com>
295 * ipa-visibility (function_and_variable_visibility): Convert
296 recursive PLT call to direct call if appropriate.
298 2017-07-21 Andrew Pinski <apinski@cavium.com>
300 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
301 operand 1 to see if the types precision matches.
302 * fold-const.c (operand_equal_p): Likewise.
304 2017-07-21 Richard Biener <rguenther@suse.de>
306 PR tree-optimization/81303
307 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
308 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
309 (vect_peeling_hash_get_lowest_cost): Adjust.
310 (vect_enhance_data_refs_alignment): Likewise. Use
311 vect_get_peeling_costs_all_drs to compute the penalty for no
312 peeling to match up costs.
314 2017-07-21 Richard Biener <rguenther@suse.de>
316 PR tree-optimization/81500
317 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
318 we didn't identify a reduction path.
320 2017-07-21 Tom de Vries <tom@codesourcery.com>
321 Cesar Philippidis <cesar@codesourcery.com>
323 PR gcov-profile/81442
324 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
327 2017-07-21 Tom de Vries <tom@codesourcery.com>
330 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
332 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
333 nvptx_override_options_after_change.
335 2017-07-21 Ulrich Drepper <drepper@redhat.com>
337 * dwarf2out.c (output_file_names): Avoid double testing for
340 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
342 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
344 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
346 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
348 (try_crossjump_to_edge): Do not punt on partitioned functions.
350 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
352 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
353 Put all BBs reachable only via paths crossing cold region to cold
355 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
357 2016-07-21 Richard Biener <rguenther@suse.de>
359 PR tree-optimization/81303
360 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
361 into account prologue and epilogue iterations when raising
362 min_profitable_iters to sth at least covering one vector iteration.
364 2017-07-21 Tamar Christina <tamar.christina@arm.com>
366 * config/arm/arm.c (arm_test_cpu_arch_dat):
369 2017-07-20 Nathan Sidwell <nathan@acm.org>
372 * tree.h (TYPE_METHODS): Delete.
373 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
374 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
375 (dbxout_type_methods): Scan TYPE_FIELDS.
376 (dbxout_type): Don't check TYPE_METHODS here.
377 * function.c (use_register_for_decl): Always ignore register for
378 class types when not optimizing.
379 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
380 * tree.c (free_lang_data_in_type): Stitch out member functions and
381 templates from TYPE_FIELDS.
382 (build_distinct_type_copy, verify_type_variant,
383 verify_type): Member fns are on TYPE_FIELDS.
384 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
385 * tree-pretty-print.c (dump_generic_node): Likewise.
387 2017-07-20 Jakub Jelinek <jakub@redhat.com>
390 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
391 V2TImode and V4TImode.
392 (ix86_expand_vector_extract): Likewise.
393 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
394 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
395 (ssescalarmode): Handle V4TImode and V2TImode.
396 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
397 (*vec_extractv2ti, *vec_extractv4ti): New insns.
398 (VEXTRACTI128_MODE): New mode iterator.
399 (splitter for *vec_extractv?ti first element): New.
400 (VEC_INIT_MODE): New mode iterator.
401 (vec_init<mode>): Consolidate 3 expanders into one using
402 VEC_INIT_MODE mode iterator.
404 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
406 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
407 non_spilled_static_chain_regno_p.
409 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
411 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
413 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
415 * bb-reorder.c (connect_traces): Allow copying of blocks within
418 2017-07-20 Richard Biener <rguenther@suse.de>
420 * gimple.h (gimple_phi_result): Add gphi * overload.
421 (gimple_phi_result_ptr): Likewise.
422 (gimple_phi_arg): Likewise. Adjust index assert to only
423 allow actual argument accesses rather than all slots available
425 (gimple_phi_arg_def): Add gphi * overload.
426 * tree-phinodes.c (make_phi_node): Initialize only actual
428 (resize_phi_node): Clear memory not covered by old node,
429 do not initialize excess argument slots.
430 (reserve_phi_args_for_new_edge): Initialize new argument slot
433 2017-07-20 Bin Cheng <bin.cheng@arm.com>
435 PR tree-optimization/81388
437 2016-07-21 Bin Cheng <bin.cheng@arm.com>
439 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
440 by removing computation of may_be_zero.
442 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
443 Tom de Vries <tom@codesourcery.com>
446 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
447 when gimple level profile disagrees with what RTL expander did.
449 2017-07-20 Richard Biener <rguenther@suse.de>
451 PR tree-optimization/61171
452 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
453 (vect_analyze_stmt): Add slp instance parameter.
454 (vectorizable_reduction): Likewise.
455 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
456 (vect_is_simple_reduction): Deal with chains not detected
457 as SLP reduction chain, specifically not properly associated
458 chains containing a mix of plus/minus.
459 (get_reduction_op): Remove.
460 (get_initial_defs_for_reduction): Simplify, pass in whether
461 this is a reduction chain, pass in the SLP node for the PHIs.
462 (vect_create_epilog_for_reduction): Get the SLP instance as
464 (vectorizable_reduction): Get the SLP instance as arg.
465 During analysis remember the SLP node with the PHIs in the
466 instance. Simplify getting at the vectorized reduction PHIs.
467 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
468 through SLP instance.
469 (vect_slp_analyze_operations): Likewise.
470 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
471 (vect_transform_stmt): Likewise.
473 2017-07-20 Tom de Vries <tom@codesourcery.com>
475 PR tree-optimization/81489
476 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
477 read of phi arg location to before loop that modifies phi.
479 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
481 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
484 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
487 * except.c (execute): Fix ordering issue.
489 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
491 PR rtl-optimization/81423
492 * combine.c (make_compound_operation_int): Don't try to optimize
493 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
495 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
497 PR rtl-optimization/81423
498 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
499 with a constant that is -1 in the truncated to mode.
501 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
503 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
504 (determine_unlikely_bbs): ... here.
505 * predict.h (propagate_unlikely_bbs_forward): Declare.
506 * cfgexpand.c (pass_expand::execute): Use it.
507 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
509 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
510 propagate_unlikely_bbs_forward.
512 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
515 * except.c (maybe_add_nop_after_section_switch): New function.
518 2017-07-19 Tom de Vries <tom@codesourcery.com>
520 * gimple.h (gimple_phi_set_arg): Make assert more strict.
522 2017-07-19 Tom de Vries <tom@codesourcery.com>
524 * gimple.h (gimple_phi_arg): Make assert more strict.
526 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
528 * config.gcc (powerpc*-*-*): Add mmintrin.h.
529 * config/rs6000/mmintrin.h: New file.
530 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
532 2017-07-19 Jakub Jelinek <jakub@redhat.com>
534 PR tree-optimization/81346
535 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
537 2017-07-19 Tom de Vries <tom@codesourcery.com>
539 * config/nvptx/nvptx.md (VECIM): Add V2DI.
541 2017-07-19 Tom de Vries <tom@codesourcery.com>
543 * config/nvptx/nvptx-modes.def: Add V2DImode.
544 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
545 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
546 (nvptx_output_mov_insn): Handle lack of mov.b128.
547 (nvptx_print_operand): Handle 'H' and 'L' codes.
548 (nvptx_vector_mode_supported): Allow V2DImode.
549 (nvptx_preferred_simd_mode): New function.
550 (nvptx_data_alignment): New function.
551 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
552 nvptx_preferred_simd_mode.
553 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
555 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
557 2017-07-19 Tom de Vries <tom@codesourcery.com>
559 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
560 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
561 (nvptx_vector_mode_supported): New function. Allow V2SImode.
562 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
563 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
564 (mov<VECIM>_insn): New define_insn.
565 (define_expand "mov<VECIM>): New define_expand.
567 2017-07-19 Tom de Vries <tom@codesourcery.com>
569 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
571 2017-07-19 Jakub Jelinek <jakub@redhat.com>
573 PR tree-optimization/81346
574 * fold-const.h (fold_div_compare, range_check_type): Declare.
575 * fold-const.c (range_check_type): New function.
576 (build_range_check): Use range_check_type.
577 (fold_div_compare): No longer static, rewritten into
578 a match.pd helper function.
579 (fold_comparison): Don't call fold_div_compare here.
580 * match.pd (X / C1 op C2): New optimization using fold_div_compare
583 2017-07-19 Nathan Sidwell <nathan@acm.org>
585 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
586 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
587 * tree.c (find_decls_types_r, verify_type): Use
588 TYPE_{MIN,MAX}_VALUE_RAW.
589 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
590 (hash_tree): Likewise.
591 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
593 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
596 2017-07-18 Tom de Vries <tom@codesourcery.com>
599 * omp-expand.c (expand_omp_for_static_chunk): Handle
600 equal-argument loop exit phi.
602 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
605 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
606 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
608 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
610 (ror,rol -> rorx splitters): Use const_int_operand as
613 2017-06-18 Richard Biener <rguenther@suse.de>
615 PR tree-optimization/81410
616 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
617 the gap in the ! slp_perm SLP case after each group.
619 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
622 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
625 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
628 * predict.c (set_even_probabilities): Cleanup; do not affect
629 probabilities that are already known.
630 (combine_predictions_for_bb): Call even when count is set.
632 2017-07-18 Nathan Sidwell <nathan@acm.org>
634 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
637 2017-07-18 Bin Cheng <bin.cheng@arm.com>
640 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
641 optimization for loop niter analysis.
643 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
646 * config/avr/avr.c (avr_optimize_casesi): Don't use
647 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
649 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
651 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
652 body_cost_vec from _vect_peel_extended_info.
653 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
654 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
657 2017-07-18 Bin Cheng <bin.cheng@arm.com>
659 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
661 2017-07-18 Richard Biener <rguenther@suse.de>
663 PR tree-optimization/80620
664 PR tree-optimization/81403
665 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
666 info when re-using a VN table entry.
668 2017-07-18 Richard Biener <rguenther@suse.de>
670 PR tree-optimization/81418
671 * tree-vect-loop.c (vectorizable_reduction): Properly compute
672 vectype_in. Verify that with lane-reducing reduction operations
673 we have a single def-use cycle.
675 2017-07-17 Carl Love <cel@us.ibm.com>
677 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
680 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
681 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
683 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
684 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
685 VMULOSW): New enum "unspec" values.
686 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
687 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
688 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
689 altivec_vmulosw): New patterns.
690 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
691 VMULOSW): Add definitions.
692 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
694 * config/alpha/alpha.c: Include predict.h.
696 2017-07-17 Yury Gribov <tetra2005@gmail.com>
698 * tree-vrp.c (compare_assert_loc): Fix comparison function
699 to return predictable results.
701 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
703 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
706 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
707 * doc/invoke.texi (mexpand-adddi): Update text.
709 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
711 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
712 that also clobbers the CC register. The old expand code is moved
714 (*arc_clzsi2): ... here.
715 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
716 the CC register. The old expand code is moved to ...
717 (arc_ctzsi2): ... here.
719 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
721 * config/arc/arc.opt (mindexed-loads): Use initial value
722 TARGET_INDEXED_LOADS_DEFAULT.
723 (mauto-modify-reg): Use initial value
724 TARGET_AUTO_MODIFY_REG_DEFAULT.
725 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
726 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
727 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
728 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
730 2017-07-17 Martin Liska <mliska@suse.cz>
733 * opts.c (finish_options): Do not allow -fgnu-tm
734 w/ -fsanitize={kernel-,}address. Say sorry.
736 2017-07-17 Bin Cheng <bin.cheng@arm.com>
739 * tree-loop-distribution.c (classify_partition): Only assert on
741 (merge_dep_scc_partitions): Delete prameter. Update function call.
742 (distribute_loop): Remove code handling loop with unknown niters.
743 (pass_loop_distribution::execute): Skip loop with unknown niters.
745 2017-07-17 Bin Cheng <bin.cheng@arm.com>
748 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
749 function sort_partitions_by_post_order.
751 2017-07-17 Bin Cheng <bin.cheng@arm.com>
753 PR tree-optimization/81374
754 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
755 the max index of basic blocks, rather than number of basic blocks.
757 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
759 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
761 (arc_legitimate_pic_operand_p): Likewise.
762 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
764 (arc_needs_pcl_p): Likewise.
765 (arc_legitimate_pc_offset_p): Likewise.
766 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
767 function is also used in constrains.md.
768 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
769 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
770 PLUS. Only return true/false in known cases, otherwise assert.
771 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
772 is already called in arc_legitimate_constant_p.
773 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
775 (LEGITIMATE_PIC_OPERAND_P): Use
776 arc_raw_symbolic_reference_mentioned_p function.
777 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
782 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
783 Andrew Burgess <andrew.burgess@embecosm.com>
785 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
786 (arc_return_address_register): New function.
787 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
788 (arc_handle_fndecl_attribute): Add naked attribute.
789 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
790 (TARGET_WARN_FUNC_RETURN): Likewise.
791 (arc_allocate_stack_slots_for_args): New function.
792 (arc_warn_func_return): Likewise.
793 (machine_function): Change type fn_type.
794 (arc_compute_function_type): Consider new naked function type,
795 change function return type.
796 (arc_must_save_register): Adapt to handle new
797 arc_compute_function_type's return type.
798 (arc_expand_prologue): Likewise.
799 (arc_expand_epilogue): Likewise.
800 (arc_return_address_regs): Delete.
801 (arc_return_address_register): New function.
802 (arc_epilogue_uses): Use above function.
803 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
804 (arc_function_type): Change encoding, add naked type.
805 (ARC_INTERRUPT_P): Change to handle the new encoding.
806 (ARC_FAST_INTERRUPT_P): Likewise.
807 (ARC_NORMAL_P): Define.
808 (ARC_NAKED_P): Likewise.
809 (arc_compute_function_type): Delete prototype.
810 * config/arc/arc.md (in_ret_delay_slot): Use
811 arc_return_address_register function.
812 (simple_return): Likewise.
813 (p_return_i): Likewise.
815 2017-07-17 Jakub Jelinek <jakub@redhat.com>
817 PR tree-optimization/81428
818 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
819 can't be built for those types.
821 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
823 Remove stuff dead since r239246.
825 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
826 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
827 (avr_inform_devices): Remove dead stuff.
829 2017-07-17 Tamar Christina <tamar.christina@arm.com>
831 * config/arm/arm_neon.h: Fix softp typo.
833 2017-07-17 Jakub Jelinek <jakub@redhat.com>
835 PR tree-optimization/81365
836 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
837 aggregate moves onto bb predecessor edges, make sure there are no
838 loads that could alias the lhs in between the start of bb and the
841 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
844 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
845 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
846 [LSHIFTRT, outer_code = TRUNCATE]: Same.
848 2017-07-17 Jakub Jelinek <jakub@redhat.com>
850 PR tree-optimization/81396
851 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
852 (init_symbolic_number): Initialize it to 1.
853 (perform_symbolic_merge): Add n_ops from both operands into the new
855 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
856 without base_addr as useless if they need more than one operation.
857 (bswap_replace): Handle !bswap case for NULL base_addr.
859 2017-07-17 Tom de Vries <tom@codesourcery.com>
862 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
865 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
867 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
868 conditional builtin define __FIX_LEON3FT_B2BST.
870 2017-07-17 Daniel Cederman <cederman@gaisler.com>
872 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
873 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
876 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
878 PR rtl-optimization/81424
879 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
880 to remove potential trapping from operands if -fnon-call-exceptions.
882 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
884 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
885 profile_proability for scalling.
886 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
888 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
890 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
892 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
894 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
895 fixpoint arithmetics.
897 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
899 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
900 fixpoint arithmetics.
902 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
904 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
905 fixpoint arithmetics.
907 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
909 * profile-count.h (profile_probability::from_reg_br_prob_note,
910 profile_probability::to_reg_br_prob_note): New functions.
911 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
912 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
913 * predict.c (probability_reliable_p): Update.
914 (edge_probability_reliable_p): Update.
915 (br_prob_note_reliable_p): Update.
916 (invert_br_probabilities): Update.
917 (add_reg_br_prob_note): New function.
918 (combine_predictions_for_insn): Update.
919 * asan.c (asan_clear_shadow): Update.
920 * cfgbuild.c (compute_outgoing_frequencies): Update.
921 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
922 (update_br_prob_note): Update.
923 (rtl_verify_edges): Update.
924 (purge_dead_edges): Update.
925 (fixup_reorder_chain): Update.
926 * emit-rtl.c (try_split): Update.
927 * ifcvt.c (cond_exec_process_insns): Update.
928 (cond_exec_process_if_block): Update.
929 (dead_or_predicable): Update.
930 * internal-fn.c (expand_addsub_overflow): Update.
931 (expand_neg_overflow): Update.
932 (expand_mul_overflow): Update.
933 * loop-doloop.c (doloop_modify): Update.
934 * loop-unroll.c (compare_and_jump_seq): Update.
935 * optabs.c (emit_cmp_and_jump_insn_1): Update.
937 * reorg.c (mostly_true_jump): Update.
939 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
940 * config/alpha/alpha.c (emit_unlikely_jump): Update.
941 * config/arc/arc.c: (emit_unlikely_jump): Update.
942 * config/arm/arm.c: (emit_unlikely_jump): Update.
943 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
944 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
945 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
946 (ix86_print_operand): Update.
947 (ix86_split_fp_branch): Update.
948 (predict_jump): Update.
949 * config/ia64/ia64.c (ia64_print_operand): Update.
950 * config/mmix/mmix.c (mmix_print_operand): Update.
951 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
952 (rs6000_expand_split_stack_prologue): Update.
953 * config/rs6000/rs6000.c: Update.
954 * config/s390/s390.c (s390_expand_vec_strlen): Update.
955 (s390_expand_vec_movstr): Update.
956 (s390_expand_cs_tdsi): Update.
957 (s390_expand_split_stack_prologue): Update.
958 * config/sh/sh.c (sh_print_operand): Update.
959 (expand_cbranchsi4): Update.
960 (expand_cbranchdi4): Update.
961 * config/sparc/sparc.c (output_v9branch): Update.
962 * config/spu/spu.c (get_branch_target): Update.
963 (ea_load_store_inline): Update.
964 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
965 * config/tilepro/tilepro.c: Update.
967 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
969 * gimplify.c (mostly_copy_tree_r): Revert latest change.
970 (gimplify_save_expr): Likewise.
972 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
974 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
976 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
978 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
980 * timevar.def (TV_IPA_FNSUMMARY): Define.
982 2017-07-16 Daniel Cederman <cederman@gaisler.com>
984 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
985 to back store errata sensitive sequence from being generated.
986 (sqrtdf2_fix): Likewise.
988 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
990 * tree-ssa-threadupdate.c (compute_path_counts,
991 update_joiner_offpath_counts): Use profile_probability.
993 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
996 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
998 * config/arm/arm-c.c (arm_cpu_builtins): Define
999 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1001 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1003 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1004 array entries to represent __ieee128 versions of the
1005 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1006 scalar_extract_sig, and scalar_insert_exp built-in functions.
1007 (altivec_resolve_overloaded_builtin): Add special case handling
1008 for the __builtin_scalar_insert_exp function, as represented by
1009 the P9V_BUILTIN_VEC_VSIEDP constant.
1010 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1011 exponent support for __ieee128 argument.
1012 (VSESQP): Add scalar extract signature support for __ieee128
1014 (VSTDCNQP): Add scalar test negative support for __ieee128
1016 (VSIEQP): Add scalar insert exponent support for __int128 argument
1017 with __ieee128 result.
1018 (VSIEQPF): Add scalar insert exponent support for __ieee128
1019 argument with __ieee128 result.
1020 (VSTDCQP): Add scalar test data class support for __ieee128
1022 (VSTDCNQP): Add overload support for scalar test negative with
1024 (VSTDCQP): Add overload support for scalar test data class
1026 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1028 (UNSPEC_VSX_SIEXPQP): New constant.
1029 (xsxexpqp): New insn for VSX scalar extract exponent quad
1031 (xsxsigqp): New insn for VSX scalar extract significand quad
1033 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1034 precision with floating point argument.
1035 (xststdcqp): New expand for VSX scalar test data class quad
1037 (xststdcnegqp): New expand for VSX scalar test negative quad
1039 (xststdcqp): New insn to match expansions for VSX scalar test data
1040 class quad precision and VSX scalar test negative quad precision.
1041 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1042 special case operand checking to enforce that second operand of
1043 VSX scalar test data class with quad precision argument is a 7-bit
1045 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1046 prototypes and descriptions of __ieee128 versions of
1047 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1048 scalar_test_data_class, and scalar_test_neg built-in functions.
1050 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1052 PR tree-optimization/81162
1053 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1054 replace a negate with an add.
1056 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1058 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1060 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1062 * config/arm/arm-c.c (arm_cpu_builtins): Define
1063 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1065 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1067 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1068 (armv8-r): Set ARM Cortex-R52 as default CPU.
1069 * config/arm/arm-tables.opt: Regenerate.
1070 * config/arm/arm-tune.md: Regenerate.
1071 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1073 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1074 extension for -mcpu=cortex-r52.
1076 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1078 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1079 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1080 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1081 (fp-armv8): Define it as FP_ARMv8 only.
1082 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1083 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1085 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1086 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1087 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1088 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1089 than TARGET_FPU_ARMV8.
1090 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1091 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1092 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1094 * config/arm/neon.md (neon_vrint): Likewise.
1095 (neon_vcvt): Likewise.
1096 (neon_<fmaxmin_op><mode>): Likewise.
1097 (<fmaxmin><mode>3): Likewise.
1098 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1099 * config/arm/predicates.md (arm_cond_move_operator): Check against
1100 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1102 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1104 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1107 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1109 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1110 loop in comment with memset.
1112 2017-07-14 Martin Liska <mliska@suse.cz>
1114 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1115 * dwarf2out.c (is_java): Remove the function.
1116 (output_pubname): Remove usage of the function.
1117 (lower_bound_default): Remove usage of DW_LANG_Java.
1118 (gen_compile_unit_die): Likewise.
1119 * gcc.c: Remove compiler defaults for .java and .zip files.
1120 * gimple-expr.c (remove_suffix): Change as there's no longer
1121 extension than 4-letter one.
1122 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1123 (gimplify_save_expr): Likewise.
1124 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1125 as it's possible even for other languages than Java.
1126 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1127 * lto-opts.c (lto_write_options): Remove reference to Java.
1128 * opts.c (strip_off_ending): Update file extension handling.
1129 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1130 * tree-eh.c (lower_resx): Likewise.
1131 * tree.c (free_lang_data_in_type): Remove dead code.
1132 (find_decls_types_r): Likewise.
1133 (build_common_builtin_nodes): Remove Java from a comment.
1134 (verify_type): Remove dead code.
1135 * varasm.c (assemble_external): Remove Java from a comment.
1137 2017-07-14 Martin Liska <mliska@suse.cz>
1139 * opts.c (finish_options): Add quotes.
1140 (common_handle_option): Likewise.
1142 2017-07-14 Martin Liska <mliska@suse.cz>
1144 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1145 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1147 * dwarf2out.c (lower_bound_default): Do not handle
1149 (gen_compile_unit_die): Likewise.
1150 * gcc.c: Remove default extension binding for GNU Pascal.
1151 * stmt.c: Remove Pascal language from a comment.
1152 * xcoffout.c: Likewise.
1154 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1157 * diagnostic-show-locus.c (fixit_cmp): New function.
1158 (layout::layout): Sort m_fixit_hints.
1159 (column_range::column_range): Assert that the values are valid.
1160 (struct char_span): New struct.
1161 (correction::overwrite): New method.
1162 (struct source_line): New struct.
1163 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1164 calls in terms of classes source_line and char_span, and
1165 correction::overwrite.
1166 (selftest::test_overlapped_fixit_printing_2): New function.
1167 (selftest::diagnostic_show_locus_c_tests): Call it.
1169 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1171 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1172 early if there is no lhs.
1174 2017-07-13 Martin Liska <mliska@suse.cz>
1176 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1177 (gen_reference_type_die): Likewise.
1178 * stor-layout.c: Remove Pascal-related comment.
1180 2017-07-13 Martin Liska <mliska@suse.cz>
1182 * opts.c (finish_options): Add quotes to error messages.
1183 (parse_sanitizer_options): Likewise.
1185 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1187 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1189 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1191 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1193 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1195 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1197 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1199 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1202 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1203 provides the hardware capability bits, define the macro
1204 __BUILTIN_CPU_SUPPORTS__.
1205 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1206 if GLIBC does not provide the hardware capability bits. Add a
1207 gcc_unreachable call if the built-in cpu function is neither
1208 __builtin_cpu_is nor __builtin_cpu_supports.
1209 (rs6000_get_function_versions_dispatcher): Change the warning
1210 that an old GLIBC is used which does not export the capability
1211 bits to be an error.
1212 * doc/extend.texi (target_clones attribute): Document the
1213 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1214 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1215 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1216 the macros defined by GCC if the newer GLIBC is available.
1218 2017-07-12 Jeff Law <law@redhat.com>
1220 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1221 remaining includes slightly.
1222 * config/riscv/riscv-builtins.c: Include profile-count.h.
1224 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1227 * config/avr/avr.c (avr_set_current_function): In diagnostic
1228 messages: Quote keywords and (parts of) identifiers.
1229 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1232 2017-07-12 Carl Love <cel@us.ibm.com>
1234 * config/rs6000/rs6000-c.c: Add support for built-in functions
1235 vector bool char vec_revb (vector bool char);
1236 vector bool short vec_revb (vector short char);
1237 vector bool int vec_revb (vector bool int);
1238 vector bool long long vec_revb (vector bool long long);
1239 * doc/extend.texi: Update the built-in documentation file for the
1240 new built-in functions.
1242 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1244 * config/s390/s390.md: Remove movcc splitter.
1246 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1248 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1249 load/store on condition.
1251 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1254 * config/avr/avr.c (avr_encode_section_info)
1255 [progmem && !TREE_READONLY]: Error if progmem object needs
1258 2017-07-11 Michael Collison <michael.collison@arm.com>
1260 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1263 2017-07-11 Carl Love <cel@us.ibm.com>
1265 * config/rs6000/rs6000-c.c: Add support for builtins
1266 vector unsigned int vec_parity_lsbb (vector signed int);
1267 vector unsigned int vec_parity_lsbb (vector unsigned int);
1268 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1269 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1270 vector unsigned long long vec_parity_lsbb (vector signed long long);
1271 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1272 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1273 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1274 * doc/extend.texi: Update the built-in documentation file for the
1275 new built-in functions.
1277 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1279 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1280 (layout::m_primary_loc): New field.
1281 (layout::layout): Initialize new field. Move location filtering
1282 logic from here to...
1283 (layout::maybe_add_location_range): ...this new method. Add
1284 support for filtering to just the lines already specified by other
1286 (layout::will_show_line_p): New method.
1287 (gcc_rich_location::add_location_if_nearby): New method.
1288 (selftest::test_add_location_if_nearby): New test function.
1289 (selftest::diagnostic_show_locus_c_tests): Call it.
1290 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1293 2017-07-11 Tom de Vries <tom@codesourcery.com>
1295 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1296 (bb_first_real_insn): New function.
1297 (nvptx_single): Add extra initialization of broadcasted condition
1300 2017-07-11 Nathan Sidwell <nathan@acm.org>
1302 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1304 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1306 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1307 Binutils doc as TEXI will mess them up.
1308 * doc/invoke.texi (AVR Options): Same here.
1310 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1312 * config/sparc/sparc.opt (mfix-ut700): New option.
1313 (mfix-gr712rc): Likewise.
1314 (sparc_fix_b2bst): New variable.
1315 * doc/invoke.texi (SPARC options): Document them.
1316 (ARM options): Fix warnings.
1317 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1318 instructions to prevent sequences that can trigger the store-store
1319 errata for certain LEON3FT processors.
1320 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1321 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1322 * config/sparc/sparc.md (fix_b2bst): New attribute.
1323 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1325 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1328 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1330 (*rsqrtsf2_sse): Ditto.
1332 (div<mode>3): Macroize insn from divdf3 and divsf3
1333 using MODEF mode iterator.
1335 2017-07-10 Martin Sebor <msebor@redhat.com>
1337 PR tree-optimization/80397
1338 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1339 instead of testing for equality to INTEGER_TYPE.
1341 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1343 * config.gcc: Remove uclibc from arc target spec.
1345 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1347 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1349 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1352 * lto-wrapper.c (remove_option): New function.
1353 (merge_and_complain): Merge PIC/PIE options more realistically.
1355 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1357 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1361 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1362 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1363 * config.in: Regenerate.
1364 * configure: Regenerate.
1365 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1366 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1367 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1368 (TARGET_GASISR_PROLOGUES): ...target mask.
1369 * common/config/avr/avr-common.c
1370 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1371 Set -mgas-isr-prologues.
1372 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1373 INSERT_PASS_BEFORE for it.
1374 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1375 * config/avr/avr.c (avr_option_override)
1376 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1377 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1378 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1379 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1380 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1382 (make_avr_pass_pre_proep): New function.
1383 (emit_push_sfr) <treg>: Add argument to function and use it
1385 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1386 and set machine->gasisr.yes.
1387 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1388 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1389 __gcc_isr.n_pushed to .L__stack_usage.
1390 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1391 (avr_asm_final_postscan_insn): ...this new static function.
1392 * config/avr/avr.h (machine_function)
1393 <is_no_gccisr, use_L__stack_usage>: New fields.
1394 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1395 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1396 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1397 (gasisr, *gasisr): New expander and insn.
1398 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1399 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1400 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1402 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1404 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1407 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1409 Move jump-tables out of .text again.
1412 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1413 (ASM_OUTPUT_ADDR_VEC): New function.
1414 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1415 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1416 INSN_ADDRESSes as asm comment.
1417 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1418 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1419 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1420 * config/avr/avr.md (*tablejump): Adjust comment.
1421 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1422 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1424 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1425 (avr_output_addr_vec): New proto.
1426 (avr_log_t) <insn_addresses>: New field.
1428 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1431 * config/i386/i386.c (ix86_function_arg_advance): Set
1432 outgoing_args_on_stack to true if there are outgoing arguments
1434 (ix86_function_arg): Likewise.
1435 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1436 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1437 * config/i386/i386.h (machine_function): Add
1438 outgoing_args_on_stack.
1440 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1442 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1444 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1446 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1448 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1449 (REAL_H): Remove $(MACHMODE_H).
1450 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1452 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1453 $(MACHMODE_H) and double-int.h.
1454 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1456 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1459 2017-07-07 Andrew Pinski <apinski@cavium.com>
1461 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1462 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1464 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1466 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1467 Add warning if GCC was not configured to link against a GLIBC that
1468 exports the hardware capability bits.
1469 (make_resolver_func): Make resolver function private and not a
1470 COMDAT function. Create the name with clone_function_name instead
1471 of make_unique_name.
1474 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1475 correct operand in doing the split.
1477 2017-07-07 Carl Love <cel@us.ibm.com>
1479 * config/rs6000/rs6000-c: Add support for built-in function
1480 vector unsigned short vec_pack_to_short_fp32 (vector float,
1482 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1483 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1484 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1485 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1486 (convert_4f32_8i16): Add define_expand.
1487 * doc/extend.texi: Update the built-in documentation file for the
1488 new built-in function.
1490 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1492 * config/sparc/m8.md: New file.
1493 * config/sparc/sparc.md: Include m8.md.
1495 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1497 * config/sparc/sparc.opt: New option -mvis4b.
1498 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1499 (sparc_option_override): Handle VIS4B.
1500 (enum sparc_builtins): Define
1501 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1502 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1503 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1504 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1505 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1506 (check_constant_argument): New function.
1507 (sparc_vis_init_builtins): Define builtins
1508 __builtin_vis_dictunpack{8,16,32},
1509 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1510 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1511 __builtin_vis_fpcmpde{8,16,32}shl and
1512 __builtin_vis_fpcmpur{8,16,32}shl.
1513 (sparc_expand_builtin): Check that the constant operands to
1514 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1515 constant and in range.
1516 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1518 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1519 (SPARC_IMM5_P): Likewise.
1520 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1521 (enabled): Handle vis4b.
1522 (UNSPEC_DICTUNPACK): New unspec.
1523 (UNSPEC_FPCMPSHL): Likewise.
1524 (UNSPEC_FPUCMPSHL): Likewise.
1525 (UNSPEC_FPCMPDESHL): Likewise.
1526 (UNSPEC_FPCMPURSHL): Likewise.
1527 (cpu_feature): New CPU feature `vis4b'.
1528 (dictunpack{8,16,32}): New insns.
1529 (FPCSMODE): New mode iterator.
1530 (fpcscond): New code iterator.
1531 (fpcsucond): Likewise.
1532 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1533 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1534 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1535 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1536 * config/sparc/constraints.md: Define constraints `q' for unsigned
1537 2-bit integer constants and `t' for unsigned 5-bit integer
1539 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1541 (imm5_operand_dictunpack16): Likewise.
1542 (imm5_operand_dictunpack32): Likewise.
1543 (imm2_operand): Likewise.
1544 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1545 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1546 ditunpack* and fpcmp*shl builtins.
1548 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1550 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1551 * config.in: Add HAVE_AS_SPARC6 define.
1552 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1554 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1556 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1557 (CPP_CPU_SPEC): Handle m8.
1558 (ASM_CPU_SPEC): Likewise.
1559 * config/sparc/sparc-opts.h (enum processor_type): Add
1561 * config/sparc/sparc.c (m8_costs): New struct.
1562 (sparc_option_override): Handle TARGET_CPU_m8.
1563 (sparc32_initialize_trampoline): Likewise.
1564 (sparc64_initialize_trampoline): Likewise.
1565 (sparc_issue_rate): Likewise.
1566 (sparc_register_move_cost): Likewise.
1567 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1568 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1569 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1570 (CPP_CPU_SPEC): Handle M8.
1571 (ASM_CPU_SPEC): Likewise.
1572 (AS_M8_FLAG): Define.
1573 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1574 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1575 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1577 * configure: Regenerate.
1578 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1581 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1583 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1585 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1586 ("*movdi_insn_sp32"): Do not set v3pipe.
1587 ("*movsi_insn"): Likewise.
1588 ("*movdi_insn_sp64"): Likewise.
1589 ("*movsf_insn"): Likewise.
1590 ("*movdf_insn_sp32"): Likewise.
1591 ("*movdf_insn_sp64"): Likewise.
1592 ("*zero_extendsidi2_insn_sp64"): Likewise.
1593 ("*sign_extendsidi2_insn"): Likewise.
1594 ("*mov<VM32:mode>_insn"): Likewise.
1595 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1596 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1597 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1598 ("<vlop:code><VL:mode>3"): Likewise.
1599 ("*not_<vlop:code><VL:mode>3"): Likewise.
1600 ("*nand<VL:mode>_vis"): Likewise.
1601 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1602 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1603 ("one_cmpl<VL:mode>2"): Likewise.
1604 ("faligndata<VM64:mode>_vis"): Likewise.
1605 ("alignaddrsi_vis"): Likewise.
1606 ("alignaddrdi_vis"): Likweise.
1607 ("alignaddrlsi_vis"): Likewise.
1608 ("alignaddrldi_vis"): Likewise.
1609 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1610 ("bmaskdi_vis"): Likewise.
1611 ("bmasksi_vis"): Likewise.
1612 ("bshuffle<VM64:mode>_vis"): Likewise.
1613 ("cmask8<P:mode>_vis"): Likewise.
1614 ("cmask16<P:mode>_vis"): Likewise.
1615 ("cmask32<P:mode>_vis"): Likewise.
1616 ("pdistn<P:mode>_vis"): Likewise.
1617 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1619 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1621 * config/sparc/sparc.md ("subtype"): New insn attribute.
1622 ("*wrgsr_sp64"): Set insn subtype.
1623 ("*rdgsr_sp64"): Likewise.
1624 ("alignaddrsi_vis"): Likewise.
1625 ("alignaddrdi_vis"): Likewise.
1626 ("alignaddrlsi_vis"): Likewise.
1627 ("alignaddrldi_vis"): Likewise.
1628 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1629 ("fexpand_vis"): Likewise.
1630 ("fpmerge_vis"): Likewise.
1631 ("faligndata<VM64:mode>_vis"): Likewise.
1632 ("bshuffle<VM64:mode>_vis"): Likewise.
1633 ("cmask8<P:mode>_vis"): Likewise.
1634 ("cmask16<P:mode>_vis"): Likewise.
1635 ("cmask32<P:mode>_vis"): Likewise.
1636 ("fchksm16_vis"): Likewise.
1637 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1638 ("fmean16_vis"): Likewise.
1639 ("fp<plusminus_insn>64_vis"): Likewise.
1640 ("<plusminus_insn>v8qi3"): Likewise.
1641 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1642 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1643 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1644 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1645 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1646 ("*movqi_insn"): Likewise.
1647 ("*movhi_insn"): Likewise.
1648 ("*movsi_insn"): Likewise.
1649 ("movsi_pic_gotdata_op"): Likewise.
1650 ("*movdi_insn_sp32"): Likewise.
1651 ("*movdi_insn_sp64"): Likewise.
1652 ("movdi_pic_gotdata_op"): Likewise.
1653 ("*movsf_insn"): Likewise.
1654 ("*movdf_insn_sp32"): Likewise.
1655 ("*movdf_insn_sp64"): Likewise.
1656 ("*zero_extendhisi2_insn"): Likewise.
1657 ("*zero_extendqihi2_insn"): Likewise.
1658 ("*zero_extendqisi2_insn"): Likewise.
1659 ("*zero_extendqidi2_insn"): Likewise.
1660 ("*zero_extendhidi2_insn"): Likewise.
1661 ("*zero_extendsidi2_insn_sp64"): Likewise.
1662 ("ldfsr"): Likewise.
1663 ("prefetch_64"): Likewise.
1664 ("prefetch_32"): Likewise.
1665 ("tie_ld32"): Likewise.
1666 ("tie_ld64"): Likewise.
1667 ("*tldo_ldub_sp32"): Likewise.
1668 ("*tldo_ldub1_sp32"): Likewise.
1669 ("*tldo_ldub2_sp32"): Likewise.
1670 ("*tldo_ldub_sp64"): Likewise.
1671 ("*tldo_ldub1_sp64"): Likewise.
1672 ("*tldo_ldub2_sp64"): Likewise.
1673 ("*tldo_ldub3_sp64"): Likewise.
1674 ("*tldo_lduh_sp32"): Likewise.
1675 ("*tldo_lduh1_sp32"): Likewise.
1676 ("*tldo_lduh_sp64"): Likewise.
1677 ("*tldo_lduh1_sp64"): Likewise.
1678 ("*tldo_lduh2_sp64"): Likewise.
1679 ("*tldo_lduw_sp32"): Likewise.
1680 ("*tldo_lduw_sp64"): Likewise.
1681 ("*tldo_lduw1_sp64"): Likewise.
1682 ("*tldo_ldx_sp64"): Likewise.
1683 ("*mov<VM32:mode>_insn"): Likewise.
1684 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1685 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1687 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1689 * config/sparc/sparc.md ("type"): New insn type viscmp.
1690 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1692 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1693 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1694 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1695 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1697 ("n7_vis_logical_11cycle"): Likewise.
1698 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1699 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1700 * config/sparc/niagara.md ("niag_vis"): Likewise.
1701 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1702 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1704 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1706 * config/sparc/sparc.md: New instruction type `bmask'.
1707 (bmaskdi_vis): Use the `bmask' type.
1708 (bmasksi_vis): Likewise.
1709 * config/sparc/ultra3.md (us3_array): Likewise.
1710 * config/sparc/niagara7.md (n7_array): Likewise.
1711 * config/sparc/niagara4.md (n4_array): Likewise.
1712 * config/sparc/niagara2.md (niag2_vis): Likewise.
1713 (niag3_vis): Likewise.
1714 * config/sparc/niagara.md (niag_vis): Likewise.
1716 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1718 * ipa-comdats.c: Remove optimize check from gate.
1719 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1720 for functions not optimized.
1721 (ipa_fn_summary_read): Skip optimize check.
1722 (ipa_fn_summary_write): Likewise.
1723 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1725 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1727 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1728 (check_callers): Give up on caller which is not optimized.
1729 (inline_small_functions): Likewise.
1730 (ipa_inline): Do not give up when not optimizing.
1731 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1732 away unoptimizes cdtors.
1733 (whole_program_function_and_variable_visibility): Do
1734 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1735 * ipa.c (process_references): Do not check optimize.
1736 (symbol_table::remove_unreachable_nodes): Update optimize check.
1737 (set_writeonly_bit): Update optimize check.
1738 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1739 (pass_ipa_single_use::gate): Remove.
1741 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1743 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1744 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1745 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1746 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1747 permute_load, permute_store, adjust_extract, adjust_splat,
1748 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1749 replace_swap_with_copy, dump_swap_insn_table,
1750 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1751 recombine_lvx_pattern, recombine_stvx_pattern,
1752 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1753 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1754 to file rs6000-p8swap.c.
1755 * config/rs6000/rs6000-p8swap.c: New file.
1756 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1757 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1758 and rs6000*-*-* targets.
1760 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1762 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1764 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1766 * lto-wrapper.c (merge_and_complain): Do not merge
1767 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
1768 fsigned_zeros, ftrapping_math, fwrapv.
1769 (append_compiler_options): Do not track these options.
1770 (append_linker_options): Likewie
1772 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1774 * cgraphunit.c (cgraph_node::finalize_function): When
1775 !flag_toplevel_reorde set no_reorder flag.
1776 (varpool_node::finalize_decl): Likewise.
1777 (symbol_table::compile): Drop no toplevel reorder path.
1779 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1781 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
1782 edges; zero probability is not better than uninitialized.
1784 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1786 * asan.h (asan_sanitize_allocas_p): Declare.
1787 * asan.c (asan_sanitize_allocas_p): New function.
1788 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
1789 (handle_builtin_alloca): Likewise.
1790 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
1791 if !asan_sanitize_allocas_p.
1792 * params.def (asan-instrument-allocas): Add new option.
1793 * params.h (ASAN_PROTECT_ALLOCAS): Define.
1794 * opts.c (common_handle_option): Disable allocas sanitization for
1797 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1799 * asan.c: Include gimple-fold.h.
1800 (get_last_alloca_addr): New function.
1801 (handle_builtin_stackrestore): Likewise.
1802 (handle_builtin_alloca): Likewise.
1803 (asan_emit_allocas_unpoison): Likewise.
1804 (get_mem_refs_of_builtin_call): Add new parameter, remove const
1805 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
1806 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
1807 (instrument_builtin_call): Pass gimple iterator to
1808 get_mem_refs_of_builtin_call.
1809 (last_alloca_addr): New global.
1810 * asan.h (asan_emit_allocas_unpoison): Declare.
1811 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
1812 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
1813 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
1814 if function calls alloca.
1815 * gimple-fold.c (replace_call_with_value): Remove static keyword.
1816 * gimple-fold.h (replace_call_with_value): Declare.
1817 * internal-fn.c: Include asan.h.
1818 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
1819 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
1821 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1823 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
1824 (C_SELFTEST_FLAGS): New.
1825 (CPP_SELFTEST_FLAGS): New.
1826 (SELFTEST_DEPS): New, from deps of s-selftest.
1827 (C_SELFTEST_DEPS): New, from deps of s-selftest.
1828 (CPP_SELFTEST_DEPS): New.
1829 (selftest): Add dependency on s-selftest-c++.
1830 (s-selftest): Rename to...
1831 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
1832 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
1833 than SELFTEST_FLAGS.
1834 (selftest-gdb): Rename to...
1835 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
1837 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
1838 (selftest-valgrind): Rename to...
1839 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
1841 (selftest-valgrind): Reintroduce as an alias for
1842 selftest-c-valgrind.
1843 (s-selftest-c++): New.
1844 (selftest-c++-gdb): New.
1845 (selftest-c++-valgrind): New.
1847 2017-07-06 Olivier Hainque <hainque@adacore.com>
1849 * gcc.c (process_command): When deciding if undefined variables
1850 should be ignored when processing specs, accept "gcc -v" as well.
1852 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1854 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
1855 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
1857 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1859 * config/arm/arm-cpus.in (armv8-r): Add new entry.
1860 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
1861 * config/arm/arm-tables.opt: Regenerate.
1862 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
1864 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
1866 2017-07-06 Carl Love <cel@us.ibm.com>
1868 * ChangeLog: Clean up from mid air collision
1870 2017-07-06 Carl Love <cel@us.ibm.com>
1872 * config/rs6000/rs6000-c.c: Add support for built-in functions
1873 vector signed int vec_subc (vector signed int, vector signed int);
1874 vector signed __int128 vec_subc (vector signed __int128,
1875 vector signed __int128);
1876 vector unsigned __int128 vec_subc (vector unsigned __int128,
1877 vector unsigned __int128);
1878 vector signed int vec_sube (vector signed int, vector signed int,
1880 vector unsigned int vec_sube (vector unsigned int,
1881 vector unsigned int,
1882 vector unsigned int);
1883 vector signed __int128 vec_sube (vector signed __int128,
1884 vector signed __int128,
1885 vector signed__int128);
1886 vector unsigned __int128 vec_sube (vector unsigned __int128,
1887 vector unsigned __int128,
1888 vector unsigned __int128);
1889 vector signed int vec_subec (vector signed int, vector signed int,
1891 vector unsigned int vec_subec (vector unsigned int,
1892 vector unsigned int,
1893 vector unsigned int);
1894 vector signed __int128 vec_subec (vector signed __int128,
1895 vector signed __int128,
1896 vector signed__int128);
1897 vector unsigned __int128 vec_subec (vector unsigned __int128,
1898 vector unsigned __int128,
1899 vector unsigned __int128);
1900 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
1901 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
1902 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
1903 BU_ALTIVEC_OVERLOAD_X definitions.
1904 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
1905 * doc/extend.texi: Update the built-in documentation file for the new
1908 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1911 * diagnostic-show-locus.c (layout::layout): Use start and finish
1912 spelling location for the start and finish of each range.
1913 * genmatch.c (linemap_client_expand_location_to_spelling_point):
1914 Add unused aspect param.
1915 * input.c (expand_location_1): Add "aspect" param, and use it
1916 to access the correct part of the location.
1917 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
1919 (expand_location_to_spelling_point): Likewise.
1920 (linemap_client_expand_location_to_spelling_point): Add "aspect"
1921 param, and pass it to expand_location_1.
1923 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
1925 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
1926 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
1927 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
1928 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
1929 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
1930 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
1931 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
1932 _mm_maskz_getmant_ss): New intrinsics.
1933 (__builtin_ia32_getexpss128_mask): Changed to ...
1934 __builtin_ia32_getexpss128_round ... this.
1935 (__builtin_ia32_getexpsd128_mask): Changed to ...
1936 __builtin_ia32_getexpsd128_round ... this.
1937 * config/i386/i386-builtin-types.def
1938 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
1939 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
1940 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
1941 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
1942 __builtin_ia32_getmantss_mask_round): New builtins.
1943 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
1944 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
1945 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
1946 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
1947 * config/i386/sse.md
1948 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
1949 avx512f_sgetexp<mode><mask_scalar_name>
1950 <round_saeonly_scalar_name> ... this.
1951 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
1952 %0, %1, %2<round_saeonly_op3>}): Changed to ...
1953 vgetexp<ssescalarmodesuffix>
1954 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1955 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
1956 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
1957 avx512f_vgetmant<mode><mask_scalar_name>
1958 <round_saeonly_scalar_name> ... this.
1959 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
1960 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
1961 vgetmant<ssescalarmodesuffix>
1962 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
1963 %0<mask_scalar_operand4>, %1, %2
1964 <round_saeonly_scalar_mask_op4>, %3} ... this.
1965 * config/i386/subst.md (mask_scalar_operand4,
1966 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
1967 round_saeonly_scalar_nimm_predicate): New subst attributes.
1969 2017-07-06 Julia Koval <julia.koval@intel.com>
1971 * config/i386/i386.c (ix86_erase_embedded_rounding):
1972 Remove code for old rounding pattern.
1974 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
1976 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
1978 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
1980 * doc/sourcebuild.texi (Test Directives, Variants of
1981 dg-require-support): Add documentation for dg-require-stack-check.
1983 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
1985 * config/i386/subst.md (mask_scalar, round_scalar,
1986 round_saeonly_scalar): New meta-templates.
1987 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
1988 round_scalar_mask_operand3, round_scalar_mask_op3,
1989 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
1990 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
1991 round_saeonly_scalar_constraint,
1992 round_saeonly_scalar_prefix): New subst attribute.
1993 * config/i386/sse.md
1994 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
1995 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
1996 <round_scalar_name> ... this.
1997 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
1998 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
1999 <round_scalar_name> ... this.
2000 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2001 <sse>_vm<code><mode>3<mask_scalar_name>
2002 <round_saeonly_scalar_name> ... this.
2003 (v<plusminus_mnemonic><ssescalarmodesuffix>
2004 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2005 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2006 v<plusminus_mnemonic><ssescalarmodesuffix>
2007 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2008 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2009 (v<multdiv_mnemonic><ssescalarmodesuffix>
2010 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2011 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2012 v<multdiv_mnemonic><ssescalarmodesuffix>
2013 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2014 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2015 (v<maxmin_float><ssescalarmodesuffix>
2016 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2017 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2018 v<maxmin_float><ssescalarmodesuffix>
2019 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2020 %0<mask_scalar_operand3>, %1, %<iptr>2
2021 <round_saeonly_scalar_mask_op3>} ... this.
2023 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2025 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2026 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2028 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2029 Alan Hayward <alan.hayward@arm.com>
2030 David Sherwood <david.sherwood@arm.com>
2032 * combine.c (simplify_if_then_else): Remove "enum" before
2034 * compare-elim.c (can_eliminate_compare): Likewise.
2035 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2037 (aarch64_lookup_simd_builtin_type): Likewise.
2038 (aarch64_simd_builtin_type): Likewise.
2039 (aarch64_init_simd_builtin_types): Likewise.
2040 (aarch64_simd_expand_args): Likewise.
2041 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2043 (aarch64_reverse_mask): Likewise.
2044 (aarch64_simd_emit_reg_reg_move): Likewise.
2045 (aarch64_gen_adjusted_ldpstp): Likewise.
2046 (aarch64_ccmp_mode_to_code): Likewise.
2047 (aarch64_operands_ok_for_ldpstp): Likewise.
2048 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2049 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2051 (aarch64_min_divisions_for_recip_mul): Likewise.
2052 (aarch64_reassociation_width): Likewise.
2053 (aarch64_get_condition_code_1): Likewise.
2054 (aarch64_simd_emit_reg_reg_move): Likewise.
2055 (aarch64_simd_attr_length_rglist): Likewise.
2056 (aarch64_reverse_mask): Likewise.
2057 (aarch64_operands_ok_for_ldpstp): Likewise.
2058 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2059 (aarch64_gen_adjusted_ldpstp): Likewise.
2060 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2062 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2063 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2064 (arm_lookup_simd_builtin_type): Likewise.
2065 (arm_simd_builtin_type): Likewise.
2066 (arm_init_simd_builtin_types): Likewise.
2067 (arm_expand_builtin_args): Likewise.
2068 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2069 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2070 (ft32_setup_incoming_varargs): Likewise.
2071 (ft32_function_arg): Likewise.
2072 (ft32_function_arg_advance): Likewise.
2073 (ft32_pass_by_reference): Likewise.
2074 (ft32_arg_partial_bytes): Likewise.
2075 (ft32_valid_pointer_mode): Likewise.
2076 (ft32_addr_space_pointer_mode): Likewise.
2077 (ft32_addr_space_legitimate_address_p): Likewise.
2078 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2080 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2081 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2082 (iamcu_alignment): Likewise.
2083 (canonicalize_vector_int_perm): Likewise.
2084 (ix86_noce_conversion_profitable_p): Likewise.
2085 (ix86_mpx_bound_mode): Likewise.
2086 (ix86_operands_ok_for_move_multiple): Likewise.
2087 * config/microblaze/microblaze-protos.h
2088 (microblaze_expand_conditional_branch_reg): Likewise.
2089 * config/microblaze/microblaze.c
2090 (microblaze_expand_conditional_branch_reg): Likewise.
2091 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2093 (rs6000_reassociation_width): Likewise.
2094 (rs6000_invalid_binary_op): Likewise.
2095 (fusion_p9_p): Likewise.
2096 (emit_fusion_p9_load): Likewise.
2097 (emit_fusion_p9_store): Likewise.
2098 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2100 (riscv_hard_regno_mode_ok_p): Likewise.
2101 (riscv_address_insns): Likewise.
2102 (riscv_split_symbol): Likewise.
2103 (riscv_legitimize_move): Likewise.
2104 (riscv_function_value): Likewise.
2105 (riscv_hard_regno_nregs): Likewise.
2106 (riscv_expand_builtin): Likewise.
2107 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2108 (riscv_build_integer): Likewise.
2109 (riscv_split_integer): Likewise.
2110 (riscv_legitimate_constant_p): Likewise.
2111 (riscv_cannot_force_const_mem): Likewise.
2112 (riscv_regno_mode_ok_for_base_p): Likewise.
2113 (riscv_valid_base_register_p): Likewise.
2114 (riscv_valid_offset_p): Likewise.
2115 (riscv_valid_lo_sum_p): Likewise.
2116 (riscv_classify_address): Likewise.
2117 (riscv_legitimate_address_p): Likewise.
2118 (riscv_address_insns): Likewise.
2119 (riscv_load_store_insns): Likewise.
2120 (riscv_force_binary): Likewise.
2121 (riscv_split_symbol): Likewise.
2122 (riscv_force_address): Likewise.
2123 (riscv_legitimize_address): Likewise.
2124 (riscv_move_integer): Likewise.
2125 (riscv_legitimize_const_move): Likewise.
2126 (riscv_legitimize_move): Likewise.
2127 (riscv_address_cost): Likewise.
2128 (riscv_subword): Likewise.
2129 (riscv_output_move): Likewise.
2130 (riscv_canonicalize_int_order_test): Likewise.
2131 (riscv_emit_int_order_test): Likewise.
2132 (riscv_function_arg_boundary): Likewise.
2133 (riscv_pass_mode_in_fpr_p): Likewise.
2134 (riscv_pass_fpr_single): Likewise.
2135 (riscv_pass_fpr_pair): Likewise.
2136 (riscv_get_arg_info): Likewise.
2137 (riscv_function_arg): Likewise.
2138 (riscv_function_arg_advance): Likewise.
2139 (riscv_arg_partial_bytes): Likewise.
2140 (riscv_function_value): Likewise.
2141 (riscv_pass_by_reference): Likewise.
2142 (riscv_setup_incoming_varargs): Likewise.
2143 (riscv_print_operand): Likewise.
2144 (riscv_elf_select_rtx_section): Likewise.
2145 (riscv_save_restore_reg): Likewise.
2146 (riscv_for_each_saved_reg): Likewise.
2147 (riscv_register_move_cost): Likewise.
2148 (riscv_hard_regno_mode_ok_p): Likewise.
2149 (riscv_hard_regno_nregs): Likewise.
2150 (riscv_class_max_nregs): Likewise.
2151 (riscv_memory_move_cost): Likewise.
2152 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2153 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2154 (rl78_addr_space_address_mode): Likewise.
2155 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2157 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2158 (rs6000_reassociation_width): Likewise.
2159 (rs6000_invalid_binary_op): Likewise.
2160 (fusion_p9_p): Likewise.
2161 (emit_fusion_p9_load): Likewise.
2162 (emit_fusion_p9_store): Likewise.
2163 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2164 (ok_for_simple_move_operands): Likewise.
2165 (ok_for_simple_move_strict_operands): Likewise.
2166 (ok_for_simple_arith_logic_operands): Likewise.
2167 (visium_legitimize_reload_address): Likewise.
2168 (visium_select_cc_mode): Likewise.
2169 (output_cbranch): Likewise.
2170 (visium_split_double_move): Likewise.
2171 (visium_expand_copysign): Likewise.
2172 (visium_expand_int_cstore): Likewise.
2173 (visium_expand_fp_cstore): Likewise.
2174 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2175 (visium_function_arg): Likewise.
2176 (visium_function_arg_advance): Likewise.
2177 (visium_libcall_value): Likewise.
2178 (visium_setup_incoming_varargs): Likewise.
2179 (visium_legitimate_constant_p): Likewise.
2180 (visium_legitimate_address_p): Likewise.
2181 (visium_legitimize_address): Likewise.
2182 (visium_secondary_reload): Likewise.
2183 (visium_register_move_cost): Likewise.
2184 (visium_memory_move_cost): Likewise.
2185 (prepare_move_operands): Likewise.
2186 (ok_for_simple_move_operands): Likewise.
2187 (ok_for_simple_move_strict_operands): Likewise.
2188 (ok_for_simple_arith_logic_operands): Likewise.
2189 (visium_function_value_1): Likewise.
2190 (rtx_ok_for_offset_p): Likewise.
2191 (visium_legitimize_reload_address): Likewise.
2192 (visium_split_double_move): Likewise.
2193 (visium_expand_copysign): Likewise.
2194 (visium_expand_int_cstore): Likewise.
2195 (visium_expand_fp_cstore): Likewise.
2196 (visium_split_cstore): Likewise.
2197 (visium_select_cc_mode): Likewise.
2198 (visium_split_cbranch): Likewise.
2199 (output_cbranch): Likewise.
2200 (visium_print_operand_address): Likewise.
2201 * expmed.c (flip_storage_order): Likewise.
2202 * expmed.h (emit_cstore): Likewise.
2203 (flip_storage_order): Likewise.
2204 * genrecog.c (validate_pattern): Likewise.
2205 * hsa-gen.c (gen_hsa_addr): Likewise.
2206 * internal-fn.c (expand_arith_overflow): Likewise.
2207 * ira-color.c (allocno_copy_cost_saving): Likewise.
2208 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2209 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2210 (process_invariant_for_inheritance): Likewise.
2211 * lra-eliminations.c (move_plus_up): Likewise.
2212 * omp-low.c (lower_oacc_reductions): Likewise.
2213 * simplify-rtx.c (simplify_subreg): Likewise.
2214 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2215 (TARGET_CHKP_BOUND_MODE): Likewise..
2216 * targhooks.c (default_chkp_bound_mode): Likewise.
2217 (default_setup_incoming_vararg_bounds): Likewise.
2218 * targhooks.h (default_chkp_bound_mode): Likewise.
2219 (default_setup_incoming_vararg_bounds): Likewise.
2220 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2221 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2222 (have_whole_vector_shift): Likewise.
2223 * tree-vect-stmts.c (vectorizable_load): Likewise.
2224 * doc/tm.texi: Regenerate.
2226 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2228 Graceful degrade if Binutils PR21472 is not available.
2231 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2232 .rodata in flash test fails.
2233 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2234 * confgure: Regenerate.
2235 * config.in: Regenerate.
2236 * config/avr/avr.c (avr_asm_named_section)
2237 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2238 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2239 (avr_asm_init_sections): Same.
2241 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2243 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2244 (fma<VH:mode>4_intrinsic): Likewise.
2245 (*fmsub<VCVTF:mode>4): Likewise.
2246 (*fmsub<VH:mode>4_intrinsic): Likewise.
2248 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2251 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2252 Don't depend on "optimize > 0".
2253 (out_movhi_r_mr, out_movqi_mr_r): Same.
2254 (out_movhi_mr_r, out_movqi_r_mr): Same.
2255 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2256 io_address_operand on "optimize > 0".
2258 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2260 * tree-loop-distribution.c: Add general explanantion on the pass.
2261 (generate_loops_for_partition): Mark distributed loop.
2262 (pg_add_dependence_edges): New parameter. Handle alias data
2263 dependence specially and record it in the parameter if asked.
2264 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2265 (init_partition_graph_vertices, add_partition_graph_edge): New.
2266 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2267 (free_partition_graph_vdata, build_partition_graph): New.
2268 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2269 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2270 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2271 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2272 (version_for_distribution_p, finalize_partitions): New.
2273 (distribute_loop): Handle alias data dependence specially. Factor
2274 out loop fusion code as functions and call these functions.
2276 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2278 * tree-loop-distribution.c (classify_partition): New parameter and
2279 better handle reduction statement.
2280 (rdg_build_partitions): Revise comment.
2281 (distribute_loop): Compute statements in all partitions and pass it
2282 to classify_partition.
2284 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2286 * tree-loop-distribution.c (enum partition_type): New.
2287 (struct partition): New field type.
2288 (partition_merge_into): Add parameter. Update partition type.
2289 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2290 (build_rdg_partition_for_vertex): Compute partition type.
2291 (rdg_build_partitions): Dump partition type.
2292 (distribute_loop): Update calls to partition_merge_into.
2294 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2296 * tree-loop-distribution.c (struct ddr_hasher): New.
2297 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2299 (classify_partition): Call get_data_dependence.
2300 (pg_add_dependence_edges): Ditto.
2301 (distribute_loop): Release data dependence hash table.
2303 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2305 * tree-loop-distribution.c (ref_base_address): Delete.
2306 (similar_memory_accesses): Rename ...
2307 (share_memory_accesses): ... to this. Check if partitions access
2308 the same memory reference.
2309 (distribute_loop): Call share_memory_accesses.
2311 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2313 * tree-loop-distribution.c (struct partition): New field recording
2315 (partition_alloc, partition_free): Init and release data refs.
2316 (partition_merge_into): Merge data refs.
2317 (build_rdg_partition_for_vertex): Collect data refs for partition.
2318 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2320 (distribute_loop): Remve data refs from vertice data of partition
2323 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2325 * tree-loop-distribution.c (params.h): Include header file.
2326 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2327 (datarefs_vec): New global var.
2328 (create_rdg_vertices): Use datarefs_vec directly.
2329 (free_rdg): Don't free data references.
2330 (build_rdg): Update use. Don't free data references.
2331 (distribute_loop): Compute global variable for data references.
2332 Bail out if there are too many data references.
2334 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2336 * tree-loop-distribution.c (loop_nest): New global var.
2337 (build_rdg): Use loop directly, rather than loop nest.
2338 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2340 (distribute_loop): Compute global variable loop nest. Update use.
2342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2344 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2345 (partition_merge_into): New parameter. Dump reason for fusion.
2346 (distribute_loop): Update use of partition_merge_into.
2348 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2350 * tree-loop-distribution.c (bb_top_order_index): New.
2351 (bb_top_order_index_size, bb_top_order_cmp): New.
2352 (stmts_from_loop): Use topological order.
2353 (pass_loop_distribution::execute): Compute and release topological
2354 order for basic blocks.
2356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2358 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2363 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2364 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2365 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2366 * internal-fn.def (LOOP_DIST_ALIAS): New.
2367 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2368 (fold_loop_internal_call): ... this.
2369 (vect_loop_dist_alias_call): New function.
2370 (set_uid_loop_bbs): Call fold_loop_internal_call.
2371 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2374 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2377 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2378 Require dead FLAGS_REG at the beginning of a peephole.
2380 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2383 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2384 arguments in the call to __builtin_ia32_sbb_u32.
2385 (_subborrow_u64): Swap _X and _Y arguments in the call to
2386 __builtin_ia32_sbb_u64.
2388 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2391 * tree-vrp.c (compare_assert_loc): Turn into a function template
2392 with stable template parameter. Only test if a->e is NULL,
2393 !a->e == !b->e has been verified already. Use e == NULL or
2394 e != NULL instead of e or ! e tests. If stable is true, don't use
2395 iterative_hash_expr, on the other side allow a or b or both NULL
2396 and sort the NULLs last.
2397 (process_assert_insertions): Sort using compare_assert_loc<false>
2398 instead of compare_assert_loc, later sort using
2399 compare_assert_loc<true> before calling process_assert_insertions_for
2400 in a loop. Use break instead of continue once seen NULL pointer.
2402 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2404 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2405 Cortex-R7 and Cortex-R8 processors.
2407 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2409 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2410 uninitialized while src is not.
2412 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2414 * common/config/arm/arm-common.c: Adjust include path for
2416 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2417 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2418 (arm-cpu-data.h): Likewise.
2419 (arm-cpu-cdata.h): Likewise.
2420 * config/arm/arm-cpu.h: Delete.
2421 * config/arm/arm-cpu-cdata.h: Delete.
2422 * config/arm/arm-cpu-data.h: Delete.
2424 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2426 * config/arm/arm-cpus.in (cortex-a55): New.
2427 (cortex-a75): Likewise.
2428 (cortex-a75.cortex-a55): Likewise.
2429 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2431 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2432 * config/arm/arm-cpu-cdata.h: Regenerate.
2433 * config/arm/arm-cpu-data.h: Regenerate.
2434 * config/arm/arm-cpu.h: Regenerate.
2435 * config/arm/arm-tables.opt: Regenerate.
2436 * config/arm/arm-tune.md: Regenerate.
2438 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2440 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2442 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2444 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2447 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2449 PR tree-optimization/81292
2450 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2451 full_string_p, also call adjust_related_strinfos if the adjustment
2452 is simple, otherwise invalidate related strinfos.
2454 2017-07-04 Martin Liska <mliska@suse.cz>
2457 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2458 newly created variable as DECL_IGNORED_P.
2460 2017-07-04 Martin Liska <mliska@suse.cz>
2463 * ipa-inline.c (inline_small_functions):
2464 Use xstrdup_for_dump.
2466 2017-07-04 Tom de Vries <tom@codesourcery.com>
2468 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2470 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2473 * config/darwin.c (darwin_function_switched_text_sections):
2476 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2478 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2480 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2482 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2484 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2486 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2487 min_profitable_iters, and th as inclusive lower bounds.
2488 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2489 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2490 for min_profitable_iters and min_profitable_estimate.
2491 (vect_transform_loop): Treat th as an inclusive lower bound.
2492 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2494 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2497 * config/darwin.c (darwin_function_switched_text_sections):
2498 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2499 in two pieces, and suppress the use of buf.
2501 2017-07-03 Nathan Sidwell <nathan@acm.org>
2503 * hash-table.h (hash_table_mod1): Fix indentation.
2505 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2508 * predict.c (force_edge_cold): Be more careful about propagation
2510 * profile-count.h (profile_probability::guessed,
2511 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2513 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2515 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2517 * doc/invoke.texi (rcpc architecture extension): Document it.
2519 2017-07-03 Richard Biener <rguenther@suse.de>
2521 PR tree-optimization/60510
2522 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2523 the scalar reduction PHI and use it.
2524 (vectorizable_reduction): Properly guard the single_defuse_cycle
2525 path for non-SLP reduction chains where we cannot use it.
2526 Rework reduc_def/index and vector type deduction. Rework
2527 vector operand gathering during reduction op code-gen.
2528 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2529 chains dissolve the chain and leave it to non-SLP reduction
2532 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2534 * tree-data-ref.h (dr_alignment): Declare.
2535 * tree-data-ref.c (dr_alignment): New function.
2536 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2537 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2539 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2541 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2543 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2544 and base_misalignment fields.
2545 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2546 * tree-data-ref.c: Include builtins.h.
2547 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2548 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2549 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2550 * tree-vect-data-refs.c: Include tree-cfg.h.
2551 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2552 fields instead of calculating an alignment here.
2553 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2554 innermost_loop_behavior fields.
2556 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2558 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2560 (DR_STEP_ALIGNMENT): New macro.
2561 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2562 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2563 (create_data_ref): Print it.
2564 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2565 to tell whether the step preserves vector (mis)alignment.
2566 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2567 Move the check for an integer step and generalise to all INTEGER_CST.
2568 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2569 Print the outer step alignment.
2571 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2573 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2574 with offset_alignment.
2575 (DR_ALIGNED_TO): Delete.
2576 (DR_OFFSET_ALIGNMENT): New macro.
2577 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2578 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2579 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2580 (create_data_ref): Likewise.
2581 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2582 (vect_analyze_data_refs): Likewise.
2583 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2584 creating dummy innermost behavior.
2586 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2588 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2589 with a "innermost_loop_behavior *" and refeence tree.
2590 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2591 (create_data_ref): Update call accordingly.
2592 * tree-predcom.c (find_looparound_phi): Likewise.
2594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2596 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2597 fields with dr_wrt_vec_loop.
2598 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2599 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2600 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2601 (vect_dr_behavior): New function.
2602 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2603 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2604 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2605 track whether the step preserves the misalignment.
2606 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2607 Use vect_dr_behavior.
2608 (vect_setup_realignment): Update call accordingly.
2609 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2610 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2611 call to vect_create_addr_base_for_vector_ref.
2612 (vect_create_cond_for_align_checks): Likewise.
2613 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2614 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2615 (vect_recog_mask_conversion_pattern): Likewise.
2616 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2617 (new_stmt_vec_info): Remove redundant zeroing.
2619 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2621 * common/config/arm/arm-common.c (arm_be8_option): New function.
2622 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2623 (ISA_ARMv6): Add isa_bit_be8.
2624 * config/arm/arm.h (arm_be8_option): Add prototype.
2625 (BE8_SPEC_FUNCTION): New define.
2626 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2627 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2628 (mlittle-endian): Similarly.
2629 (mbe8, mbe32): New options.
2630 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2631 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2633 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2635 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2637 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2639 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2640 (cleanup_tree_cfg_bb): Use it.
2641 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2643 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2645 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2648 * loop-doloop.c (add_test): Update profile.
2650 2017-07-03 Martin Liska <mliska@suse.cz>
2653 * sanopt.c (rewrite_usage_of_param): New function.
2654 (sanitize_rewrite_addressable_params): Likewise.
2655 (pass_sanopt::execute): Call rewrite_usage_of_param.
2657 2017-07-03 Richard Biener <rguenther@suse.de>
2659 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2660 back to using VIEW_CONVERT_EXPR.
2662 2017-07-03 Martin Liska <mliska@suse.cz>
2665 * doc/extend.texi: Document when a resolver function is
2666 generated for target_clones.
2668 2017-07-03 Martin Liska <mliska@suse.cz>
2670 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2671 and shadow memory of auto variables which are subject of
2672 use-after-scope sanitization.
2673 (asan_expand_mark_ifn): Add do set only when is_poison.
2675 2016-07-03 Richard Biener <rguenther@suse.de>
2677 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2679 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2680 (vectorizable_reduction): Perform reduction PHI creation when
2681 visiting a reduction PHI and adjust and simplify code generation
2682 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2683 (vect_transform_loop): Visit reduction PHIs.
2684 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2685 defs into the SLP tree.
2686 (vect_build_slp_tree): Reduction defs terminate the recursion.
2687 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2689 (vect_get_vec_defs_for_stmt_copy): Export.
2690 (vect_get_vec_defs): Likewise.
2691 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2693 (vect_get_vec_defs_for_stmt_copy): Declare.
2694 (vect_get_vec_defs): Likewise.
2696 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2698 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2699 parameter with a "loop" parameter and use it instead of the
2700 loop containing DR_STMT. Don't check simple_iv when doing
2701 BB analysis. Describe the two analysis modes in the comment.
2703 2017-07-03 Tom de Vries <tom@codesourcery.com>
2705 PR tree-optimization/69468
2706 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2707 (find_same_succ_bb): Handle ignore_edge_flags.
2709 2017-07-03 Tom de Vries <tom@codesourcery.com>
2711 PR tree-optimization/81192
2712 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2714 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2716 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2718 2017-07-03 Tom de Vries <tom@codesourcery.com>
2720 PR tree-optimization/81192
2721 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2722 BB_SAME_SUCC (bb) == NULL.
2724 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2726 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2729 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2731 * dumpfile.c: Include profile-count.h
2732 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2734 (insert_cond_bb): Update profile.
2735 * tree-cfg.h (insert_cond_bb): Update prototype.
2736 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2737 * tree-dump.c: Do not include tree-cfg.
2739 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2741 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2743 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2745 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2748 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2750 * tree-complex.c (expand_complex_div_wide): update profile.
2752 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2753 Alan Hayward <alan.hayward@arm.com>
2754 David Sherwood <david.sherwood@arm.com>
2756 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2757 (CORETYPES_H): New define.
2758 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2759 (insn-modes-inline.h, s-modes-inline-h): New rules.
2760 (generated_files): Add insn-modes-inline.h.
2761 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2762 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2763 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2764 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2765 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2766 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2767 (build/gencodes.o, build/genconditions.o): Likewise.
2768 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
2769 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
2770 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
2771 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
2772 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
2773 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
2774 * coretypes.h: Include everything up to real.h for generators.
2775 Include insn-modes.h first. Include wide-int-print.h after
2776 wide-int.h. Include insn-modes-inline.h and then machmode.h.
2777 * machmode.h: Don't include insn-modes.h here.
2778 * function-tests.c: Remove includes of signop.h, machmode.h,
2779 double-int.h and wide-int.h.
2781 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
2783 * optc-save-gen.awk: Likewise.
2784 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
2785 * godump.c: Remove include of wide-int-print.h.
2786 * pretty-print.h: Likewise.
2787 * wide-int-print.cc: Likewise.
2788 * wide-int.cc: Likewise.
2789 * hash-map-tests.c: Remove include of signop.h.
2790 * hash-set-tests.c: Likewise.
2791 * rtl-tests.c: Likewise.
2792 * mkconfig.sh: Remove include of machmode.h.
2793 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
2795 (emit_insn_modes_inline_h): ...this new function. Emit the code
2796 into an insn-modes-inline.h header file, adding appropriate
2797 include guards and end comments.
2798 (emit_insn_modes_c_header): Remove include of machmode.h.
2799 (emit_min_insn_modes_c_header): Include coretypes.h rather than
2801 (main): Handle -i flag and call emit_insn_modes_inline_h when
2804 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2806 * tree-ssa-strlen.c (strinfo): Rename the length field to
2807 nonzero_chars. Add a full_string_p field.
2808 (compare_nonzero_chars, zero_length_string_p): New functions.
2809 (get_addr_stridx): Add an offset_out parameter.
2810 Use compare_nonzero_chars.
2811 (get_stridx): Update accordingly. Use compare_nonzero_chars.
2812 (new_strinfo): Update after above changes to strinfo.
2813 (set_endptr_and_length): Set full_string_p.
2814 (get_string_length): Update after above changes to strinfo.
2815 (unshare_strinfo): Update call to new_strinfo.
2816 (maybe_invalidate): Likewise.
2817 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
2818 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
2819 as a uhwi instead of an shwi. Update after above changes to
2820 strinfo and new_strinfo.
2821 (zero_length_string): Assert that chainsi contains full strings.
2822 Use zero_length_string_p. Update call to new_strinfo.
2823 (adjust_related_strinfos): Update after above changes to strinfo.
2824 Copy full_string_p from origsi.
2825 (adjust_last_stmt): Use zero_length_string_p.
2826 (handle_builtin_strlen): Update after above changes to strinfo and
2827 new_strinfo. Install the lhs as the string length if the previous
2828 entry didn't describe a full string.
2829 (handle_builtin_strchr): Update after above changes to strinfo
2831 (handle_builtin_strcpy): Likewise.
2832 (handle_builtin_strcat): Likewise.
2833 (handle_builtin_malloc): Likewise.
2834 (handle_pointer_plus): Likewise.
2835 (handle_builtin_memcpy): Likewise. Track nonzero characters
2836 that aren't necessarily followed by a nul terminator.
2837 (handle_char_store): Likewise.
2839 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2841 PR tree-optimization/80769
2842 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
2843 for malloc and calloc. Document the new invariant that all related
2844 strinfos have delayed lengths or none do.
2845 (verify_related_strinfos): Move earlier in file.
2846 (set_endptr_and_length): New function, split out from...
2847 (get_string_length): ...here. Also set the lengths of related
2849 (zero_length_string): Assert that chainsi has known (rather than
2851 (adjust_related_strinfos): Likewise.
2853 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2855 PR tree-optimization/81136
2856 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
2857 assert that two references with the same misalignment have the same
2858 compile-time misalignment if those compile-time misalignments
2861 2017-07-01 Andi Kleen <ak@linux.intel.com>
2863 * print-tree.c (print_node): Print all attributes.
2865 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2867 * cfg.c (scale_bbs_frequencies): New function.
2868 * cfg.h (scale_bbs_frequencies): Declare it.
2869 * cfgloopanal.c (single_likely_exit): Cleanup.
2870 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
2872 (scale_loop_profile): Likewise.
2873 (loop_version): Likewise.
2874 (create_empty_loop_on_edge): Update.
2875 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
2876 scale_loop_frequencies, scale_loop_profile, loopify,
2877 loop_version): Update prototypes.
2878 * modulo-sched.c (sms_schedule): Update.
2879 * predict.c (unlikely_executed_edge_p): Also check probability.
2880 (probably_never_executed_edge_p): Fix typo.
2881 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2882 * tree-parloops.c (gen_parallel_loop): Update.
2883 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
2884 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2885 * tree-ssa-loop-split.c (split_loop): Update.
2886 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2887 * tree-vect-loop-manip.c (vect_do_peeling): Update.
2888 (vect_loop_versioning): Update.
2889 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2891 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2893 * trans-mem.c (split_bb_make_tm_edge): Update profile.
2895 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2897 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
2898 to keep profile consistent.
2900 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2902 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
2903 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
2904 * profile-count.h (max_safe_multiplier): Make unsigned.
2905 (profile_count::guessed_zero): New.
2907 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2909 * bb-reorder.c (fix_up_crossing_landing_pad,
2910 fix_crossing_conditional_branches): Use make_single_succ_edge
2911 to keep profile consistent.
2913 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2915 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
2918 2017-07-01 Jakub Jelinek <jakub@redhat.com>
2921 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
2922 the right scopes, make sure cond_jump isn't preserved between multiple
2923 iterations. Search for fallthru edge whenever there are 3+ edges and
2924 use find_fallthru_edge for it.
2926 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2928 Patch by Alexander Monakov <amonakov@ispras.ru>
2929 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
2930 probabilities consistently.
2932 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2934 * pa.c (pa_expand_compare_and_swap_loop): Update call of
2935 emit_cmp_and_jump_insns.
2937 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2940 * tree-inline.c (expand_call_inline): Combine profile statuses.
2942 2017-06-30 Andrew Pinski <apinski@cavium.com>
2944 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
2945 fold_stmt returned true.
2947 2017-06-30 Nathan Sidwell <nathan@acm.org>
2949 * ggc.h (empty_string): Delete.
2950 * cfgexpand.c (expand_asm_stmt): Use plain "".
2951 * optabs.c (expand_asm_memory_barrier): Likewise.
2952 * stringpool.c (empty_string): Delete.
2953 (digit_vector, digit_string): Delete.
2954 (ggc_alloc_string): Use plain "", don't optimize single digit
2955 strings. Use ggc_alloc_atomic.
2957 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
2959 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
2960 comparison set and one other set, use the cost of the non-comparison
2963 2017-06-30 Nathan Sidwell <nathan@acm.org>
2965 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
2968 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
2970 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
2971 loops. Remove now unneeded calls to gimple_switch_set_label() that
2972 just set removed labels to NULL_TREE.
2974 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
2976 * tree-ssanames.c (set_range_info_raw): Abstract from ...
2977 (set_range_info): ...here. Only call set_range_info_raw if domain
2979 (set_nonzero_bits): Call set_range_info_raw.
2980 * tree-ssanames.h (set_range_info_raw): New.
2982 2017-06-30 Jakub Jelinek <jakub@redhat.com>
2985 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
2986 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
2987 of nonimmediate_operand and <store_mask_constraint> instead of m
2988 for the input operand. For V8FI iterator, always split if input
2989 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
2990 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
2991 <store_mask_predicate> instead of register_operand and
2992 <store_mask_constraint> instead of v for the input operand. Make
2993 sure both operands aren't MEMs for if not <mask_applied>.
2995 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
2997 * lto-wrapper.c (copy_file) Close both file descriptors before
3000 2017-06-30 Martin Liska <mliska@suse.cz>
3003 * multiple_target.c (create_dispatcher_calls): Make ifunc
3004 also for function that don't have calls or are not referenced.
3006 2017-06-30 Richard Biener <rguenther@suse.de>
3008 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3009 analyze the first scalar stmt. Move vector type computation
3010 for the BB case here from ...
3011 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3012 live operation processing in the SLP case properly.
3014 2017-06-30 Richard Biener <rguenther@suse.de>
3016 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3018 2017-06-30 Martin Liska <mliska@suse.cz>
3021 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3022 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3024 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3026 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3029 2017-06-30 Richard Biener <rguenther@suse.de>
3031 PR tree-optimization/81249
3032 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3033 condition reduction result to original scalar type.
3035 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3037 * profile-count.h (enum profile_quality): Fix typos and whitespace
3040 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3042 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3043 type for branch probabilities.
3045 2017-06-29 Julian Brown <julian@codesourcery.com>
3046 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3048 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3049 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3050 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3051 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3053 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3055 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3056 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3057 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3058 CC usage from generic code to here.
3059 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3060 CC usage into the target macros.
3062 2017-06-29 Maya Rashish <coypu@sdf.org>
3064 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3067 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3069 * arm/arm-builtins.c: Include profile-count.h
3070 * except.c (sjlj_emit_function_enter): Use
3071 profile_probability::unlikely.
3073 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3075 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3076 and tocrel_offset be pointer args rather than implicitly using
3078 (legitimate_constant_pool_address_p, rs6000_emit_move,
3079 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3080 tocrel_offset and use in toc_relative_expr_p call.
3081 (print_operand, print_operand_address): Use static tocrel_base_oac
3082 and tocrel_offset_oac.
3083 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3086 2017-06-29 Maya Rashish <coypu@sdf.org>
3088 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3090 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3092 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3093 objects, take into account only the alignment of 'op0' and 'mode1' if
3096 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3098 * ccmp.c (ccmp_tree_comparison_p): New function.
3099 (ccmp_candidate_p): Update to use above function.
3100 (get_compare_parts): New function.
3101 (expand_ccmp_next): Update to use new functions.
3102 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3104 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3105 take mode as argument.
3106 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3107 * expr.c (expand_expr_real_1): Pass mode as argument.
3109 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3111 * combine.c (combine_instructions): Print insns to dump_file, together
3114 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3116 * asan.c (asan_emit_stack_protection): Update.
3117 (create_cond_insert_point): Update.
3118 * auto-profile.c (afdo_propagate_circuit): Update.
3119 * basic-block.h (struct edge_def): Turn probability to
3120 profile_probability.
3121 (EDGE_FREQUENCY): Update.
3122 * bb-reorder.c (find_traces_1_round): Update.
3123 (better_edge_p): Update.
3124 (sanitize_hot_paths): Update.
3125 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3126 (make_single_succ_edge): Update.
3127 (check_bb_profile): Update.
3128 (dump_edge_info): Update.
3129 (update_bb_profile_for_threading): Update.
3130 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3131 probabilitycount to 0.
3132 * cfgbuild.c (compute_outgoing_frequencies): Update.
3133 * cfgcleanup.c (try_forward_edges): Update.
3134 (outgoing_edges_match): Update.
3135 (try_crossjump_to_edge): Update.
3136 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3137 (expand_gimple_tailcall): Update.
3138 (construct_init_block): Use make_single_succ_edge.
3139 (construct_exit_block): Use make_single_succ_edge.
3140 * cfghooks.c (verify_flow_info): Update.
3141 (redirect_edge_succ_nodup): Update.
3142 (split_edge): Update.
3143 (account_profile_record): Update.
3144 * cfgloopanal.c (single_likely_exit): Update.
3145 * cfgloopmanip.c (scale_loop_profile): Update.
3146 (set_zero_probability): Remove.
3147 (duplicate_loop_to_header_edge): Update.
3148 * cfgloopmanip.h (loop_version): Update prototype.
3149 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3150 (force_nonfallthru_and_redirect): Update.
3151 (update_br_prob_note): Update.
3152 (rtl_verify_edges): Update.
3153 (purge_dead_edges): Update.
3154 (rtl_lv_add_condition_to_bb): Update.
3155 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3156 * cgraphunit.c (init_lowered_empty_function): Update.
3157 (cgraph_node::expand_thunk): Update.
3158 * cilk-common.c: Include profile-count.h
3159 * dojump.c (inv): Remove.
3160 (jumpifnot): Update.
3161 (jumpifnot_1): Update.
3162 (do_jump_1): Update.
3164 (do_jump_by_parts_greater_rtx): Update.
3165 (do_compare_rtx_and_jump): Update.
3166 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3167 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3168 * dwarf2cfi.c: Include profile-count.h
3169 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3170 (sjlj_emit_dispatch_table): Likewise.
3171 * explow.c: Include profile-count.h
3172 * expmed.c (emit_store_flag_force): Update.
3173 (do_cmp_and_jump): Update.
3174 * expr.c (compare_by_pieces_d::generate): Update.
3175 (compare_by_pieces_d::finish_mode): Update.
3176 (emit_block_move_via_loop): Update.
3177 (store_expr_with_bounds): Update.
3178 (store_constructor): Update.
3179 (expand_expr_real_2): Update.
3180 (expand_expr_real_1): Update.
3181 * expr.h (try_casesi, try_tablejump): Update prototypes.
3182 * gimple-pretty-print.c (dump_probability): Update.
3183 (dump_profile): New.
3184 (dump_gimple_label): Update.
3185 (dump_gimple_bb_header): Update.
3186 * graph.c (draw_cfg_node_succ_edges): Update.
3187 * hsa-gen.c (convert_switch_statements): Update.
3188 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3189 (find_if_case_1): Update.
3190 (find_if_case_2): Update.
3191 * internal-fn.c (expand_arith_overflow_result_store): Update.
3192 (expand_addsub_overflow): Update.
3193 (expand_neg_overflow): Update.
3194 (expand_mul_overflow): Update.
3195 (expand_vector_ubsan_overflow): Update.
3196 * ipa-cp.c (good_cloning_opportunity_p): Update.
3197 * ipa-split.c (split_function): Use make_single_succ_edge.
3198 * ipa-utils.c (ipa_merge_profiles): Update.
3199 * loop-doloop.c (add_test): Update.
3200 (doloop_modify): Update.
3201 * loop-unroll.c (compare_and_jump_seq): Update.
3202 (unroll_loop_runtime_iterations): Update.
3203 * lra-constraints.c (lra_inheritance): Update.
3204 * lto-streamer-in.c (input_cfg): Update.
3205 * lto-streamer-out.c (output_cfg): Update.
3206 * mcf.c (adjust_cfg_counts): Update.
3207 * modulo-sched.c (sms_schedule): Update.
3208 * omp-expand.c (expand_omp_for_init_counts): Update.
3209 (extract_omp_for_update_vars): Update.
3210 (expand_omp_ordered_sink): Update.
3211 (expand_omp_for_ordered_loops): Update.
3212 (expand_omp_for_generic): Update.
3213 (expand_omp_for_static_nochunk): Update.
3214 (expand_omp_for_static_chunk): Update.
3215 (expand_cilk_for): Update.
3216 (expand_omp_simd): Update.
3217 (expand_omp_taskloop_for_outer): Update.
3218 (expand_omp_taskloop_for_inner): Update.
3219 * omp-simd-clone.c (simd_clone_adjust): Update.
3220 * optabs.c (expand_doubleword_shift): Update.
3221 (expand_abs): Update.
3222 (emit_cmp_and_jump_insn_1): Update.
3223 (expand_compare_and_swap_loop): Update.
3224 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3225 * predict.c (predictable_edge_p): Update.
3226 (edge_probability_reliable_p): Update.
3227 (set_even_probabilities): Update.
3228 (combine_predictions_for_insn): Update.
3229 (combine_predictions_for_bb): Update.
3230 (propagate_freq): Update.
3231 (estimate_bb_frequencies): Update.
3232 (force_edge_cold): Update.
3233 * profile-count.c (profile_count::dump): Add missing space into dump.
3234 (profile_count::debug): Add newline.
3235 (profile_count::differs_from_p): Explicitly convert to unsigned.
3236 (profile_count::stream_in): Update.
3237 (profile_probability::dump): New member function.
3238 (profile_probability::debug): New member function.
3239 (profile_probability::differs_from_p): New member function.
3240 (profile_probability::differs_lot_from_p): New member function.
3241 (profile_probability::stream_in): New member function.
3242 (profile_probability::stream_out): New member function.
3243 * profile-count.h (profile_count_quality): Rename to ...
3244 (profile_quality): ... this one.
3245 (profile_probability): New.
3246 (profile_count): Update.
3247 * profile.c (compute_branch_probabilities): Update.
3248 * recog.c (peep2_attempt): Update.
3249 * sched-ebb.c (schedule_ebbs): Update.
3250 * sched-rgn.c (find_single_block_region): Update.
3251 (compute_dom_prob_ps): Update.
3252 (schedule_region): Update.
3253 * sel-sched-ir.c (compute_succs_info): Update.
3254 * stmt.c (struct case_node): Update.
3255 (do_jump_if_equal): Update.
3256 (get_outgoing_edge_probs): Update.
3257 (conditional_probability): Update.
3258 (emit_case_dispatch_table): Update.
3259 (expand_case): Update.
3260 (expand_sjlj_dispatch_table): Update.
3261 (emit_case_nodes): Update.
3262 * targhooks.c: Update.
3263 * tracer.c (better_p): Update.
3264 (find_best_successor): Update.
3265 * trans-mem.c (expand_transaction): Update.
3266 * tree-call-cdce.c: Update.
3267 * tree-cfg.c (gimple_split_edge): Upate.
3268 (move_sese_region_to_fn): Upate.
3269 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3270 * tree-eh.c (lower_resx): Upate.
3271 (cleanup_empty_eh_move_lp): Upate.
3272 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3273 * tree-inline.c (copy_edges_for_bb): Update.
3274 (copy_cfg_body): Update.
3275 * tree-parloops.c (gen_parallel_loop): Update.
3276 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3277 (gimple_gen_time_profiler): Update.
3278 * tree-ssa-dce.c (remove_dead_stmt): Update.
3279 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3280 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3281 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3282 (unloop_loops): Update.
3283 (try_peel_loop): Update.
3284 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3285 * tree-ssa-loop-split.c (connect_loops): Update.
3286 (split_loop): Update.
3287 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3288 (hoist_guard): Update.
3289 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3290 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3291 (value_replacement): Update.
3292 * tree-ssa-reassoc.c (branch_fixup): Update.
3293 * tree-ssa-tail-merge.c (replace_block_by): Update.
3294 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3295 (create_edge_and_update_destination_phis): Update.
3296 (compute_path_counts): Update.
3297 (recompute_probabilities): Update.
3298 (update_joiner_offpath_counts): Update.
3299 (freqs_to_counts_path): Update.
3300 (duplicate_thread_path): Update.
3301 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3302 (struct switch_conv_info): Update.
3303 (gen_inbound_check): Update.
3304 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3305 (vect_do_peeling): Update.
3306 (vect_loop_versioning): Update.
3307 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3308 (optimize_mask_stores): Update.
3309 * ubsan.c (ubsan_expand_null_ifn): Update.
3310 * value-prof.c (gimple_divmod_fixed_value): Update.
3311 (gimple_divmod_fixed_value_transform): Update.
3312 (gimple_mod_pow2): Update.
3313 (gimple_mod_pow2_value_transform): Update.
3314 (gimple_mod_subtract): Update.
3315 (gimple_mod_subtract_transform): Update.
3316 (gimple_ic): Update.
3317 (gimple_stringop_fixed_value): Update.
3318 (gimple_stringops_transform): Update.
3319 * value-prof.h: Update.
3321 2017-06-29 Carl Love <cel@us.ibm.com>
3323 * config/rs6000/rs6000-c.c: Add support for built-in functions
3324 vector signed int vec_signed (vector float);
3325 vector signed long long vec_signed (vector double);
3326 vector signed int vec_signed2 (vector double, vector double);
3327 vector signed int vec_signede (vector double);
3328 vector signed int vec_signedo (vector double);
3329 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3330 instruction generator.
3331 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3332 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3333 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3335 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3336 vunsignede_v2df): Add define_expands.
3337 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3338 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3339 VEC_UNSIGNEDO): Add definitions.
3340 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3341 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3342 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3343 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3344 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3345 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3346 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3347 vec_unsignede, vec_unsignedo): Add builtin defines.
3348 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3350 * doc/extend.texi: Update the built-in documentation file for the
3351 new built-in functions.
3353 2017-06-29 Richard Biener <rguenther@suse.de>
3355 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3356 reduction chains to LOOP_VINFO_REDUCTIONS.
3357 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3358 SLP reductions after processing reduction chains.
3360 2017-06-29 Nathan Sidwell <nathan@acm.org>
3362 * builtins.c (fold_builtin_FUNCTION): Use
3363 lang_hooks.decl_printable_name.
3365 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3368 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3369 with only one label.
3370 * stmt.c (expand_case): Assert NCASES is greater than one.
3372 2017-06-29 Richard Biener <rguenther@suse.de>
3374 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3376 (group_case_labels): Likewise.
3377 (find_taken_edge): Push sanity checking on val to workers...
3378 (find_taken_edge_cond_expr): ... here
3379 (find_taken_edge_switch_expr): ... and here, handle cases
3380 with just a default label.
3381 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3382 (group_case_labels): Likewise.
3383 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3384 group_case_labels does anything cleanup the CFG again.
3386 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3388 PR tree-optimization/81196
3389 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3390 exit condition comparing two IVs.
3392 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3394 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3395 profile to the dummy entry at the end of the list of architectures.
3396 * config/arm/arm-cpu-cdata.h: Regenerated.
3398 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3399 Michael Collison <michael.collison@arm.com>
3402 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3404 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3405 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3406 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3407 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3408 of shift when the shift amount is masked with constant equal to
3409 the size of the mode.
3410 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3413 2017-06-29 Martin Liska <mliska@suse.cz>
3415 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3417 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3420 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3423 (edge_growth_cache_entry::edge_growth_cache_entry): New
3425 (reset_edge_growth_cache): Update to use constructor.
3427 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3429 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3430 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3431 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3433 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3435 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3436 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3438 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3440 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3441 (*-linux-uclibc*): Add t-uclibc tmake_file.
3442 * config/t-musl: New.
3443 * config/t-uclibc: New.
3445 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3447 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3449 (gen_comm_data): Emit architectural setting of arch_prof.
3450 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3452 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3453 (armv8-m.base, armv8-m.main): Likewise.
3454 * arm-protos.h (arm_build_target): Add profile field.
3455 (arch_option): Likewise.
3456 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3458 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3459 arm_active_target.profile.
3461 2017-06-28 Richard Biener <rguenther@suse.de>
3464 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3465 TYPE_OVERFLOW_WRAPS.
3466 * match.pd (negate_expr_p): Likewise.
3467 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3468 fold_build2, not fold_binary.
3470 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3472 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3473 Convert memory address to Pmode.
3474 (aarch64_print_operand): Assert MEM operands are always Pmode.
3476 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3479 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3480 Remove redundant if.
3481 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3482 * config/arm/aarch-common-protos.h
3483 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3484 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3486 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3489 * multiple_target.c (create_dispatcher_calls): Set the default
3490 clone to be static, not public.
3492 2017-06-28 Richard Biener <rguenther@suse.de>
3494 * tree-vect-loop.c (vectorizable_reduction): Move special
3495 cond reduction IV var creation ...
3496 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3497 parameter. Use STMT_VINFO_VECTYPE.
3498 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3501 2017-06-28 Martin Liska <mliska@suse.cz>
3504 * ipa-visibility.c (non_local_p): Handle visibility.
3506 2017-06-28 Martin Liska <mliska@suse.cz>
3509 * common.opt: Add IntegerRange to various options.
3510 * opt-functions.awk (integer_range_info): New function.
3511 * optc-gen.awk: Add integer_range_info to cl_options struct.
3512 * opts-common.c (decode_cmdline_option): Handle
3513 CL_ERR_INT_RANGE_ARG.
3514 (cmdline_handle_error): Likewise.
3515 * opts.c (print_filtered_help): Show valid interval in
3516 when --help is provided.
3517 * opts.h (struct cl_option): Add range_min and range_max fields.
3518 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3520 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3522 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3523 (x * C EQ/NE y * C): New transformation.
3525 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3527 * genmultilib (combination_space): Accept '+' in option names.
3529 2017-06-28 Martin Liska <mliska@suse.cz>
3532 * asan.c (instrument_derefs): Bail out inner references
3533 that are hard register variables.
3535 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3538 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3539 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3541 2017-06-28 Richard Biener <rguenther@suse.de>
3543 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3544 (vect_get_slp_defs): Adjust.
3545 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3547 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3549 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3550 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3551 (vectorizable_reduction): Adjust.
3552 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3554 (vect_get_slp_defs): Likewise.
3555 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3556 (vectorizable_bswap): Adjust.
3557 (vectorizable_call): Likewise.
3558 (vectorizable_conversion): Likewise.
3559 (vectorizable_assignment): Likewise.
3560 (vectorizable_shift): Likewise.
3561 (vectorizable_operation): Likewise.
3562 (vectorizable_store): Likewise.
3563 (vectorizable_condition): Likewise.
3564 (vectorizable_comparison): Likewise.
3566 2017-06-28 Michael Collison <michael.collison@arm.com>
3569 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3571 (arm_gen_movmemqi): Removed unused variable 'i'.
3572 Convert 'for' loop into 'while' loop.
3573 (arm_expand_prologue): Remove last unnecessary set of insn.
3574 (thumb_pop): Remove unused variable 'pushed_words'.
3575 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3577 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3579 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3580 * config/s390/s390-protos.h: Add prototype of
3581 s390_rel_address_ok_p.
3582 * config/s390/s390.c (s390_got_symbol): New function.
3583 (s390_rel_address_ok_p): New function.
3584 (legitimize_pic_address): Use s390_rel_address_ok_p.
3585 (s390_load_got): Use s390_got_symbol.
3586 (s390_option_override): Issue error if
3587 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3588 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3590 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3592 2017-06-27 Andrew Pinski <apinski@cavium.com>
3594 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3595 (X * copysign (1.0, X)): New pattern.
3596 (X * copysign (1.0, -X)): New pattern.
3597 (copysign (-1.0, CST)): New pattern.
3599 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3601 * genmultilib (combination_space): Remove variable.
3602 Validate reuse rules against regular expression for any sequence
3603 of multilib options in any order.
3605 2017-06-27 Michael Collison <michael.collison@arm.com>
3607 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3608 call aarch64_split_simd_combine.
3609 * (aarch64_combine_internal<mode>): Delete pattern.
3610 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3611 Allow register and subreg operands.
3613 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3615 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3616 specific need, just fallback on defaults.
3617 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3619 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3620 Olivier Hainque <hainque@adacore.com>
3622 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3624 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3625 targets. Pick a default if no particular attempt applied.
3626 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3629 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3631 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3632 (x86_64-wrs-vxworks7): Likewise.
3634 2017-06-27 Marek Polacek <polacek@redhat.com>
3637 * ubsan.c (instrument_null): Check get_base_address's result for null.
3639 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3641 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3643 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3645 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3646 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3647 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3648 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3650 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3651 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3652 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3653 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3654 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3655 BUILT_IN_FEUPDATEENV): New builtins.
3656 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3657 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3658 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3659 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3661 (builtin_structptr_types): Adjust size.
3662 * tree.c (builtin_structptr_types): Add four entries.
3664 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3665 Olivier Hainque <hainque@adacore.com>
3667 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3668 (TLS_SYM): New local macro, forcing reference to __tls__ on
3669 link command lines for VxWorks 7 RTPs, triggering initialization
3671 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3672 OS features TLS support, true for RTPs on VxWorks 7.
3673 * config/vxworks.c (vxworks_override_options): Setup emutls
3676 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3678 * predict.c (test_prediction_value_range): Use -1U instead of -1
3679 to avoid narrowing conversion warning.
3680 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3681 to avoid narrowing conversion warning.
3682 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3684 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3686 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3688 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3689 64bit configurations.
3690 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3691 (SIZE_TYPE): Likewise.
3692 * config/vxworks.c (vxworks_emutls_var_fields): Use
3693 long_unsigned_type_node instead of unsigned_type_node as the offset
3694 field type, which is "pointer" mode in emutls.c.
3696 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3699 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3702 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3703 gimple_vuse copying separately from gimple_vdef copying.
3705 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3707 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3710 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3711 Olivier Hainque <hainque@adacore.com>
3713 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3714 for all vxworks7 targets.
3715 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3716 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3717 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3718 variations for VX6/VX7 and 32/64bits later on in ...
3719 (VXWORKS_LIB_SPEC): Leverage new macros.
3720 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3721 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3723 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3725 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3726 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3728 2017-06-26 Carl Love <cel@us.ibm.com>
3730 * config/rs6000/rs6000-c.c: Add support for built-in functions
3731 vector bool char vec_reve (vector bool char);
3732 vector signed char vec_reve (vector signed char);
3733 vector unsigned char vec_reve (vector unsigned char);
3734 vector bool int vec_reve (vector bool int);
3735 vector signed int vec_reve (vector signed int);
3736 vector unsigned int vec_reve (vector unsigned int);
3737 vector bool long long vec_reve (vector bool long long);
3738 vector signed long long vec_reve (vector signed long long);
3739 vector unsigned long long vec_reve (vector unsigned long long);
3740 vector bool short vec_reve (vector bool short);
3741 vector signed short vec_reve (vector signed short);
3742 vector double vec_reve (vector double);
3743 vector float vec_reve (vector float);
3744 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3745 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3746 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3747 (altivec_vreve): New pattern.
3748 * config/rs6000/altivec.h (vec_reve): New define.
3749 * doc/extend.texi (vec_rev): Update the built-in documentation file
3750 for the new built-in functions.
3752 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3754 PR tree-optimization/71815
3755 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3757 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3759 (slsr_process_phi): Likewise.
3760 (replace_uncond_cands_and_profitable_phis): Don't replace a
3761 multiply candidate with a stride of 1 (copy or cast).
3762 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3764 (lowest_cost_path): Likewise.
3765 (total_savings): Likewise.
3767 2017-06-26 Richard Biener <rguenther@suse.de>
3770 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
3771 Use def_builtin_pure for all gather builtins.
3773 2017-06-26 Richard Biener <rguenther@suse.de>
3775 PR tree-optimization/81203
3776 * tree-tailcall.c (find_tail_calls): Do not move stmts into
3779 2017-06-26 Marek Polacek <polacek@redhat.com>
3782 * doc/invoke.texi: Document -Wmultistatement-macros.
3784 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
3786 * doc/sourcebuild.texi (ARM-specific attributes): Document new
3787 arm_neon_ok_no_float_abi effective target.
3789 2017-06-26 Richard Biener <rguenther@suse.de>
3791 PR tree-optimization/80928
3792 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
3793 (copy_bbs): Set BB_DUPLICATED flag early.
3794 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
3796 (execute_on_shrinking_pred): Likewise.
3797 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
3798 BB_DUPLICATED blocks.
3799 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
3800 iterate over all PHIs considering removal of *gsi.
3802 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
3804 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
3807 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3809 * config/rs6000/rs6000-string.c: (expand_block_clear,
3810 do_load_for_compare, select_block_compare_mode,
3811 compute_current_alignment, expand_block_compare,
3812 expand_strncmp_align_check, expand_strn_compare,
3813 expand_block_move, rs6000_output_load_multiple)
3814 Move functions related to string/block move/compare
3816 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
3817 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
3818 for this function which is now used in two files.
3819 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
3820 * config.gcc: Add rs6000-string.o to extra_objs for
3821 targets powerpc*-*-* and rs6000*-*-*.
3823 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3826 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
3827 32-bit, since indexed is not valid for DImode.
3828 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
3829 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
3830 (define_peephole2 for Altivec d-form load): Add 32-bit support.
3831 (define_peephole2 for Altivec d-form store): Likewise.
3834 * multiple_target.c (create_dispatcher_calls): Only create the
3835 dispatcher call if the function is the default clone of a
3838 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
3841 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
3842 a call, force the call to not be a tail call.
3844 2017-06-23 Jeff Law <law@redhat.com>
3846 * doc/contrib.texi: Add entry for Steven Pemberton's work on
3849 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
3851 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
3852 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
3853 handling for early expansion of vector shifts (sl,sr,sra,rl).
3854 (builtin_function_type): Add vector shift right instructions
3855 to the unsigned argument list.
3857 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3859 rtl-optimizatoin/79286
3860 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
3861 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
3862 trap. PIC register plus a const unspec without offset can never trap.
3864 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
3866 * tree.h (builtin_structptr_type): New type.
3867 (builtin_structptr_types): Declare new array.
3868 * tree.c (builtin_structptr_types): New array.
3869 (free_lang_data, build_common_tree_nodes): Use it.
3871 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
3874 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
3877 2017-06-22 Matt Turner <mattst88@gmail.com>
3879 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
3880 Lake models to skylake case. Assume skylake for unknown
3881 models with clflushopt.
3883 2017-06-22 Jeff Law <law@redhat.com>
3885 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
3886 frame sizes that do not satisfy aarch64_uimm12_shift.
3888 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
3890 * profile-count.h (apply_probability,
3891 apply_scale, probability_in): Fix checks for zero.
3893 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3895 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
3896 * doc/cppdiropts.texi (-I @var{dir}): Document it.
3898 2016-06-22 Richard Biener <rguenther@suse.de>
3900 * tree-vect-loop.c (vect_model_reduction_cost): Handle
3901 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
3902 REDUC_MAX_EXPR support.
3903 (vectorizable_reduction): Likewise.
3904 (vect_create_epilog_for_reduction): Likewise.
3906 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
3908 * match.pd (A / (1 << B) -> A >> B): New.
3909 * generic-match-head.c: Include optabs-tree.h.
3910 * gimple-match-head.c: Likewise.
3911 * optabs-tree.h (target_supports_op_p): New.
3912 * optabs-tree.c (target_supports_op_p): New.
3914 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3916 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
3917 $gcc_cv_ld --help output.
3918 (gcc_cv_ld_demangle): Likewise.
3919 (gcc_cv_ld_eh_frame_hdr): Likewise.
3920 (gcc_cv_ld_pie): Likewise.
3921 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
3922 (gcc_cv_ld_buildid): Likewise.
3923 (gcc_cv_ld_sysroot): Likewise.
3924 (ld_bndplt_support): Likewise.
3925 (ld_pushpopstate_support): Likewise.
3926 * configure: Regenerate.
3927 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
3929 2017-06-21 Jakub Jelinek <jakub@redhat.com>
3932 * config/i386/sse.md (round<mode>2): Renumber match_dup and
3933 operands indexes to avoid gap between operands and match_dups.
3935 2017-06-21 Andrew Pinski <apinski@cavium.com>
3937 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
3938 Increment Arith_shift and Arith_shift_reg by 1.
3939 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
3941 * config/aarch64/aarch64.c (thunderx_tunings): Enable
3942 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
3943 (aarch64_strip_extend): Add new argument and test for it.
3944 (aarch64_cheap_mult_shift_p): New function.
3945 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
3946 add a cost if it is true.
3947 Update calls to aarch64_strip_extend.
3948 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
3950 2017-06-21 Andrew Pinski <apinski@cavium.com>
3952 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
3954 (thunderxt88): Likewise.
3955 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
3956 (thunderx_prefetch_tune): New variable.
3957 (thunderx2t99_prefetch_tune): Update for the correct values.
3958 (thunderxt88_tunings): New variable.
3959 (thunderx_tunings): Use thunderx_prefetch_tune instead of
3960 generic_prefetch_tune.
3961 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
3963 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3965 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
3966 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
3967 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
3968 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
3969 (aarch64_atomic_cas<mode>, GPI): Likewise.
3971 2017-06-21 Martin Liska <mliska@suse.cz>
3973 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
3974 statements on cold and hot labels.
3975 * predict.c (tree_estimate_probability_bb): Remove the
3976 prediction from this place.
3978 2017-06-21 Martin Liska <mliska@suse.cz>
3980 PR tree-optimization/79489
3981 * gimplify.c (maybe_add_early_return_predict_stmt): New
3983 (gimplify_return_expr): Call the function.
3984 * predict.c (tree_estimate_probability_bb): Remove handling
3986 * predict.def: Update comment about early return predictor.
3987 * gimple-predict.h (is_gimple_predict): New function.
3988 * predict.def: Change default value of early return to 66.
3989 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
3991 * passes.def: Put pass_strip_predict_hints to the beginning of
3994 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
3996 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
3997 FUNCTION_DECL declarations.
3998 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4000 (dwaf2out_decl): Likewise.
4001 * godump.c (go_early_global_decl): Skip call to the real debug hook
4002 for FUNCTION_DECL declarations.
4003 * passes.c (rest_of_decl_compilation): Skip call to the
4004 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4005 -fdump-go-spec is passed.
4007 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4009 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4010 Reorder for compactness.
4011 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4012 (def_builtin_pure, def_builtin_pure2): New functions.
4013 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4015 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4017 * match.pd (nop_convert): New predicate.
4018 ((A +- CST1) +- CST2): Allow some NOP conversions.
4020 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4023 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4024 with ctors/dtors if GOVD_SHARED is set.
4026 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4028 * config/aarch64/aarch64.md (movti_aarch64):
4029 Emit mov rather than orr.
4030 (movtf_aarch64): Likewise.
4031 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4032 Emit mov rather than orr.
4034 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4036 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4037 Swap alternatives, make integer dup more expensive.
4039 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4041 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4042 Return true for non-tls symbols.
4044 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4046 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4047 (cortex-a75): Likewise.
4048 (cortex-a75.cortex-a55): Likewise.
4049 * config/aarch64/aarch64-tune.md: Regenerate.
4050 * doc/invoke.texi (-mtune): Document new values for -mtune.
4052 2017-06-21 Tom de Vries <tom@codesourcery.com>
4054 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4056 (Effective-Target Keywords, Other attributes): Suggest using
4057 dg-add-options stack_size feature to get stack limit in stack_size
4058 effective target documentation.
4060 2017-06-21 Julian Brown <julian@codesourcery.com>
4061 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4063 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4064 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4065 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4067 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4068 attribute type list for neon_multiply.
4069 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4070 attribute type list for neon_multiply.
4071 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4072 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4073 attribute type list for neon_multiply.
4074 * config/arm/types.md (crypto_pmull): Add.
4075 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4076 attribute type list.
4078 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4080 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4083 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4085 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4086 to make sure not to dereference a NULL cost_classes_ptr pointer.
4088 2017-06-20 Carl Love <cel@us.ibm.com>
4090 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4091 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4092 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4093 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4094 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4095 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4096 VMULOSW): New enum "unspec" values.
4097 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4098 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4099 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4100 altivec_vmulosw): New patterns.
4101 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4102 VMULOSW): Add definitions.
4104 2017-06-20 Julia Koval <julia.koval@intel.com>
4106 * config/i386/i386.c: Fix rounding expand for new pattern.
4107 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4109 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4111 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4112 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4114 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4116 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4119 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4121 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4122 architecture, then by implementer ID.
4123 * config/aarch64/aarch64-tune.md: Regenerate.
4125 2017-06-20 Richard Biener <rguenther@suse.de>
4128 * fold-const.c (split_tree): Fold to type before negating.
4130 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4132 * diagnostic-show-locus.c
4133 (selftest::test_fixit_deletion_affecting_newline): New function.
4134 (selftest::diagnostic_show_locus_c_tests): Call it.
4136 2017-06-20 Andreas Schwab <schwab@suse.de>
4139 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4142 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4144 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4145 __ARM_FEATURE_COPROC according to support.
4147 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4149 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4150 Rewritten to avoid overflow for > 32-bit pointers.
4153 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4154 by removing enum keyword.
4155 (ubsan_type_descriptor): Likewise. Formatting fix.
4158 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4159 splitter): Require TARGET_SSE2 in the condition.
4161 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4164 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4165 for doing vector set of SFmode on ISA 3.0.
4166 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4167 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4169 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4170 SFmode value into a V4SF variable that was extracted from another
4171 V4SF variable without converting the element to double precision
4172 and back to single precision vector format.
4173 (vsx_insert_extract_v4sf_p9_2): Likewise.
4175 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4177 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4178 in UWHI to avoid undefined overflow.
4181 * ubsan.h (enum ubsan_encode_value_phase): New.
4182 (ubsan_encode_value): Change second argument to
4183 enum ubsan_encode_value_phase with default value of
4184 UBSAN_ENCODE_VALUE_GENERIC.
4185 * ubsan.c (ubsan_encode_value): Change second argument to
4186 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4187 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4188 create_tmp_var_raw instead of create_tmp_var and use a
4190 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4191 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4192 ubsan_encode_value callers.
4195 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4196 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4197 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4199 2017-06-19 Richard Biener <rguenther@suse.de>
4202 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4203 estimates if we changed anything.
4205 2017-06-19 Richard Biener <rguenther@suse.de>
4207 PR tree-optimization/80887
4208 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4209 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4210 simplified lookups, then reset mprts_hook.
4211 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4213 (try_to_simplify): Likewise.
4215 2017-06-19 Martin Liska <mliska@suse.cz>
4218 * gimplify.c (gimplify_switch_expr):
4219 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4221 2017-06-19 Martin Liska <mliska@suse.cz>
4223 * doc/install.texi: Document that PGO runs in 4 stages.
4225 2017-06-19 Martin Liska <mliska@suse.cz>
4228 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4229 to dispatcher function name.
4230 * multiple_target.c (replace_function_decl): New function.
4231 (create_dispatcher_calls): Redirect both edges and references.
4233 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4235 * profile-count.c (profile_count::dump): Dump quality.
4236 (profile_count::differs_from_p): Update for unsigned val.
4237 * profile-count.h (profile_count_quality): New enum.
4238 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4240 2017-06-19 Richard Biener <rguenther@suse.de>
4242 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4243 struct function as arg.
4244 (estimate_numbers_of_iterations): Export overload with loop arg.
4245 (free_numbers_of_iterations_estimates_loop): Use an overload of
4246 free_numbers_of_iterations_estimates instead.
4247 * tree-cfg.c (remove_bb): Adjust.
4248 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4249 * tree-parloops.c (gen_parallel_loop): Likewise.
4250 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4252 (tree_unroll_loops_completely): Likewise.
4253 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4254 Use an overload instead and export.
4255 (estimated_loop_iterations): Adjust.
4256 (max_loop_iterations): Likewise.
4257 (likely_max_loop_iterations): Likewise.
4258 (estimate_numbers_of_iterations): Take struct function as arg
4260 (loop_exits_before_overflow): Adjust.
4261 (free_numbers_of_iterations_estimates_loop): Use an overload.
4262 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4263 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4265 2017-06-19 Richard Biener <rguenther@suse.de>
4268 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4269 RANGE_EXPR conservatively.
4271 2017-06-16 Carl Love <cel@us.ibm.com>
4273 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4274 definitions for vec_float, vec_float2, vec_floato,
4275 vec_floate built-ins.
4276 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4277 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4279 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4280 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4281 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4282 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4283 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4284 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4285 vec_floato): Add builtin defines.
4286 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4287 Update the built-in documentation file for the new built-in
4290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4292 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4293 (mthumb): Mark as the negative of -marm.
4295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4297 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4299 (ARM Options, -mtune): Document that this accepts the same
4300 extension options as -mcpu.
4301 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4303 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4305 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4306 permitted extensions.
4308 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4310 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4311 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4312 (armv8-m.main): Add option +nodsp.
4313 * config/arm/arm-cpu-cdata.h: Regenerated.
4315 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4317 * config/arm/t-fuchsia: New file.
4318 * config.gcc (arm*-*-fuchsia*): Use it.
4320 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4322 * config/arm/t-symbian: Rewrite for new option infrastructure.
4324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4326 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4327 (MULTILIB_REQUIRED): Likewise.
4329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4331 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4332 (MULTILIB_RESUE): Likewise.
4333 (MULTILIB_MATCHES): Likewise.
4334 (MULTLIB_REQUIRED): Likewise.
4336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4338 * config/arm/t-rtems: Rewrite for new option framework.
4340 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4342 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4343 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4344 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4345 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4346 * config/arm/t-multilib: ... here.
4347 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4348 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4349 armv7-a and armv8*-a when A-profile libraries have not been built.
4350 * config/arm/t-rmprofile: Rewrite.
4352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4354 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4355 with a backslash. Remove the backslash after substituting unescaped
4357 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4361 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4362 the driver to pass through the default setting of -mfloat-abi.
4363 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4365 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4366 (all_feat_combs): New rule.
4367 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4369 * config/arm/t-aprofile: Rewrite.
4371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4373 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4374 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4375 fpu is not specified by the user/command-line.
4376 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4377 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4378 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4379 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4380 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4381 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4385 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4386 * config/arm/t-arm-elf: Rewritten.
4388 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4390 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4391 have some floating-point instructions.
4392 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4393 (TARGET_MAYBE_HARD_FLOAT): New macro.
4394 * config/arm/arm-builtins.c (arm_init_builtins): Use
4395 TARGET_MAYBE_HARD_FLOAT.
4396 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4400 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4401 (configargs.h): Include it.
4402 (arm_print_hint_for_fpu_option): New function.
4403 (arm_parse_fpu_option): New function.
4404 (candidate_extension): New class.
4405 (arm_canon_for_multilib): New function.
4406 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4407 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4408 (ARCH_CANONICAL_SPECS): New macro.
4409 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4413 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4414 are set after handling multilib fragments. Set target_cpu_default2
4417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4419 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4421 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4422 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4424 (check_arch): Likewise.
4425 * config/arm/arm.c (arm_configure_build_target): Handle
4426 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4427 options in the default.
4429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4431 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4432 when an option is an alias of another.
4433 * config/arm/parsecpu.awk (optalias): New parser token.
4434 (gen_comm_data): Mark non-alias options as such. Emit entries
4435 for extension aliases.
4436 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4437 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4438 (armv6kz, armv6zk, armv6t2): Likewise.
4439 (armv7): Make vfpv3-d16 an alias.
4440 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4442 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4443 Sort in canonical order.
4444 (armv8-a): Sort in canonical order.
4445 (armv8.1-a, armv8.2-a): Likewise.
4446 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4448 (cortex-a9): Sort in canonical order.
4449 * config/arm/arm.c (selftests.h): Include it.
4450 (arm_test_cpu_arch_data): New function.
4451 (arm_run_self_tests): New function.
4452 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4453 (targetm): Move declaration to the end of the file.
4454 * arm-cpu-cdata.h: Regenerated.
4456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4458 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4459 call to target_mode_check describing the type of option passed.
4460 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4461 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4462 arm_parse_cpu_option_name to match parameters against list of
4464 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4465 arm_arch_core_flags data structure.
4466 * config/arm/arm-cpu_cdata.h: Regenerated.
4468 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4470 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4472 (arm_print_hint_for_cpu_option): Likewise.
4473 (arm_print_hint_for_arch_option): Likewise.
4474 (arm_parse_cpu_option_name): Likewise.
4475 (arm_parse_arch_option_name): Likewise.
4476 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4477 of entries in the all_fpus list.
4478 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4479 (arm_parse_cpu_option_name): Declare.
4480 (arm_parse_arch_option_name): Declare.
4481 (arm_parse_option_features): Declare.
4482 (arm_intialize_isa): Declare.
4483 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4485 (gen_comm_data): ... here. Make definitions non-static.
4486 * config/arm/arm-cpu-data.h: Regenerated.
4487 * config/arm/arm-cpu-cdata.h: Regenerated.
4489 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4491 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4492 (cpu_arch_extension): New structure.
4493 (cpu_arch_option, arch_option, cpu_option): New structures.
4494 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4496 (gen_data): Generate new format data tables.
4497 * config/arm/arm.c (cpu_tune): New structure.
4498 (cpu_option, processors): Delete.
4499 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4500 (arm_print_hint_for_cpu_option): ... this and ...
4501 (arm_print_hint_for_arch_option): ... this.
4502 (arm_parse_arch_cpu_name): Delete. Replace with ...
4503 (arm_parse_cpu_option_name): ... this and ...
4504 (arm_parse_arch_option_name): ... this.
4505 (arm_unrecognized_feature): Change type of target parameter to
4507 (arm_parse_arch_cpu_features): Delete. Replace with ...
4508 (arm_parse_option_features): ... this.
4509 (arm_configure_build_target): Rework to use new configuration data
4511 (arm_print_tune_info): Rework for new configuration data tables.
4512 * config/arm/arm-cpu-data.h: Regenerated.
4513 * config/arm/arm-cpu.h: Regenerated.
4515 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4517 * Makefile.in (OBJS): Move sbitmap.o from here ...
4518 (OBJS-libcommon): ... to here.
4520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4522 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4523 (ISA_ALL_CRYPTO): New macro.
4524 (ISA_ALL_SIMD): New macro
4525 (ISA_ALL_FP): New macro.
4526 * config/arm/arm.c (fpu_bitlist): Update initializer.
4527 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4529 (arm9e): Add fpu. Add option for nofp
4530 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4531 (arm926ej-s, arm1026ej-s): Likewise.
4532 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4533 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4534 neon-fp16, neon-vfpv4, nofp and nosimd.
4535 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4536 (cortex-a8): Add fpu. Add option for nofp.
4537 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4538 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4539 (cortex-r4f): Add fpu.
4540 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4541 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4543 (cortex-r8): Likewise.
4544 (cortex-m4): Add fpu. Add option for nofp.
4545 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4546 (cortex-a17.cortex-a7): Likewise.
4547 (cortex-a32): Add fpu. Add options for crypto and nofp.
4548 (cortex-a35, cortex-a53): Likewise.
4549 (cortex-a57): Add fpu. Add option for crypto.
4550 (cortex-a72, cortex-a73): Likewise.
4551 (exynos-m1): Likewise.
4552 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4553 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4554 (cortex-m33): Add fpu. Add option for nofp.
4555 * config/arm/arm-cpu-cdata.h: Regenerated
4556 * config/arm/arm-cpu-data.h: Regenerated.
4558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4560 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4561 (armv5te, armv5tej): Likewise.
4562 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4563 (armv7): Add options fp and vfpv3-d16.
4564 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4565 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4567 (armv7ve): Likewise.
4568 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4569 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4570 (armv8-a): Add nocrypto option.
4571 (armv8.1-a, armv8.2-a): Likewise.
4572 (armv8-m.main): add options fp, fp.dp and nofp.
4574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4576 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4578 (armv8-a+crc): Delete.
4579 (armv8.1-a): Add options simd, crypto and nofp.
4580 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4581 (armv8.2-a+fp16): Delete.
4582 (armv8-m.main): Add option dsp.
4583 (armv8-m.main+dsp): Delete.
4584 (cortex-a8): Add fpu. Add nofp option.
4585 (cortex-a9): Add fpu. Add nofp and nosimd options.
4586 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4587 link to main cpu and architecture data structures.
4588 (gen_comm_data): Only put isa attributes from the main architecture
4590 (option): New statement for architecture and CPU entries.
4591 * arm.c (struct cpu_option): New structure.
4592 (struct processors): Add entry for options.
4593 (arm_unrecognized_feature): New function.
4594 (arm_parse_arch_cpu_name): Ignore any characters after the first
4596 (arm_parse_arch_cpu_feature): New function.
4597 (arm_configure_build_target): Separate out any CPU and architecture
4598 features and parse separately. Don't error out if -mfpu=auto is
4599 used with only an architecture string.
4600 (arm_print_asm_arch_directives): New function.
4601 (arm_file_start): Call it.
4602 * config/arm/arm-cpu-cdata.h: Regenerated.
4603 * config/arm/arm-cpu-data.h: Likewise.
4604 * config/arm/arm-tables.opt: Likewise.
4606 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4608 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4609 assembler when it is not -mfpu=auto.
4611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4613 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4614 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4615 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4616 (ASM_CPU_SPEC): Rewrite.
4617 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4618 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4619 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4620 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4621 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4622 copied string is NUL-terminated. Also strip any characters prefixed
4624 (arm_rewrite_selected_arch): New function.
4625 (arm_rewrite_march): New function.
4627 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4629 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4630 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4631 (march, mcpu, mtune): Convert to string-based options.
4632 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4633 (arm_parse_arch_cpu_name): New function.
4634 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4635 identify selected architecture or CPU.
4636 (arm_option_save): New function.
4637 (TARGET_OPTION_SAVE): Redefine.
4638 (arm_option_restore): Restore string options.
4639 (arm_option_print): Print string options.
4641 2017-06-16 Martin Sebor <msebor@redhat.com>
4643 PR tree-optimization/80933
4644 PR tree-optimization/80934
4645 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4646 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4647 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4648 (gimple_fold_builtin): Call them.
4650 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4652 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4653 as unlikely; update profile.
4655 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4657 * predict.c (force_edge_cold): Handle declaring edges impossible
4660 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4662 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4664 (try_unroll_loop_completely): Fix reporting.
4666 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4668 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4670 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4673 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4674 if given a non-constant argument for an intrinsic which requires a
4677 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4679 * profile.c (compare_freqs): New function.
4680 (branch_prob): Sort edge list.
4681 (find_spanning_tree): Assume that the list is priority sorted.
4683 2017-06-16 Richard Biener <rguenther@suse.de>
4685 PR tree-optimization/81090
4686 * passes.def (pass_record_bounds): Remove.
4687 * tree-pass.h (make_pass_record_bounds): Likewise.
4688 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4689 make_pass_record_bounds): Likewise.
4690 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4691 not free niter estimates at the beginning but at the end.
4692 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4694 2017-06-16 Richard Biener <rguenther@suse.de>
4696 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4697 initializer to workaround ICE in host GCC 4.8.
4699 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4701 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4703 (clone_inlined_nodes): Update.
4705 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4707 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4708 prefetch settings, and enable prefetching by default at -O3.
4710 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4712 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4713 Set flag_prefetch_loop_arrays according to tuning data.
4715 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4717 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4719 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4720 [Unrelated to main purpose of the patch] Place the pointer field last
4721 to enable type checking errors when tune structure are wrongly merged.
4722 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4723 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4724 (thunderx2t99_prefetch_tune): New tune constants.
4725 (tune_params *_tunings): Update all tunings (no functional change).
4726 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4727 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4728 from tunings structures.
4730 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4733 * ubsan.c (instrument_null): Add T argument, use it instead
4734 of computing it based on IS_LHS.
4735 (instrument_object_size): Likewise.
4736 (pass_ubsan::execute): Adjust instrument_null and
4737 instrument_object_size callers to pass gimple_get_lhs or
4738 gimple_assign_rhs1 result to it. Use instrument_null instead of
4739 calling get_base_address and instrument_mem_ref. Handle
4740 aggregate call arguments for object-size sanitization.
4742 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4744 PR tree-optimization/81089
4745 * tree-vrp.c (is_masked_range_test): Validate operands of
4748 2017-06-15 Martin Sebor <msebor@redhat.com>
4751 * dumpfile.c (dump_register): Avoid calling memset to initialize
4752 a class with a default ctor.
4753 * gcc.c (struct compiler): Remove const qualification.
4754 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4755 * hash-table.h: Ditto.
4756 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4758 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4759 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4761 * params.h (struct param_info): Make struct members non-const.
4762 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4763 with default initialization.
4764 * vec.h (vec_copy_construct, vec_default_construct): New helper
4766 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4767 with vec_copy_construct.
4768 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
4769 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
4770 * doc/invoke.texi (-Wclass-memaccess): Document.
4772 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4774 * emit-rtl.h (is_leaf): Update comment about local
4777 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4780 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
4781 for a variable to have a section before checking if the section has a
4783 Set section to.persistent if persistent attribute is set.
4784 Warn if .persistent attribute is used on an automatic variable.
4786 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
4788 PR rtl-optimization/80474
4789 * reorg.c (update_block): Do not ignore instructions in a delay slot.
4791 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
4793 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
4796 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
4798 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
4799 (casesi): Emit bounds checking as RTL.
4800 (casesi_internal_mips16_<mode>): Remove bounds checking.
4802 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
4804 * config/xtensa/xtensa.c (xtensa_option_override): Append
4805 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
4806 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
4807 xtensa_doloop_hooks): Define unconditionally.
4808 (xtensa_reorg_loops): Only call reorg_loops in the presence of
4810 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
4811 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
4812 for it in xtensa_option_override.
4813 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
4814 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
4816 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
4818 * doc/cppopts.texi: Document '-' special value to -MF.
4820 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
4822 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
4823 (cortex_a53_fconst): Likewise.
4824 (cortex_a53_fpmul): Likewise.
4825 (cortex_a53_f_load_64): Likewise.
4826 (cortex_a53_f_load_many): Likewise.
4827 (cortex_a53_advsimd_alu): Likewise.
4828 (cortex_a53_advsimd_alu_q): Likewise.
4829 (cortex_a53_advsimd_mul): Likewise.
4830 (cortex_a53_advsimd_mul_q): Likewise.
4831 (fpmac bypass): Add new bypass for fpmac-fpmac case.
4832 Add missing fmul, r2f_cvt and fconst cases.
4834 2017-06-14 Richard Biener <rguenther@suse.de>
4837 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
4839 (fold_binary_loc): When associating do not treat pre-existing
4840 TREE_OVERFLOW on literal constants as a reason to allow
4841 TREE_OVERFLOW on associated literal constants.
4843 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4845 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
4846 (MASK_FEATURES): New macro.
4847 * config/sparc/sparc.c (sparc_option_override): Remove the special
4848 handling of -mfpu and generalize it to all MASK_FEATURES switches.
4850 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4852 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
4853 a division of 0 if non-call exceptions are enabled.
4855 2017-06-14 Andrew Pinski <apinski@cavium.com>
4856 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4859 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
4860 Improve vector initialization code gen for only variable case.
4862 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4864 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
4866 2017-06-14 Richard Biener <rguenther@suse.de>
4868 PR tree-optimization/81083
4869 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
4872 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4874 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
4875 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
4876 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
4877 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
4878 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
4879 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
4881 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4883 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
4884 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
4886 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4888 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
4890 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4892 * config/rs6000/t-rtems: Don't handle SPE.
4894 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4896 * config/rs6000/t-linux: Don't handle SPE.
4898 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4900 * config/rs6000/eabispe.h: Delete file.
4902 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4904 * config/rs6000/t-spe: Delete file.
4906 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4908 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
4909 (rs6000_legitimate_offset_address_p): Return false for anything in
4910 V2SImode or V2SFmode.
4912 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4914 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
4915 except V2SF and V2SI. Rearrange the vector modes, and add comments.
4916 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
4918 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
4919 (rs6000_legitimate_offset_address_p): Ditto.
4920 (rs6000_emit_move): Ditto.
4921 (rs6000_init_builtins): Remove V4HI_type_node.
4923 2017-06-13 Martin Liska <mliska@suse.cz>
4926 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
4927 (gate_asan): Likewise.
4928 * asan.h (asan_no_sanitize_address_p): Remove the function.
4929 (sanitize_flags_p): New function.
4930 * builtins.def: Fix coding style.
4931 * common.opt: Use renamed enum value.
4932 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
4933 * doc/extend.texi: Document no_sanitize attribute.
4934 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
4935 to SANITIZE_UNDEFINED_NONDEFAULT.
4936 * gcc.c (sanitize_spec_function): Use the renamed enum value.
4937 * gimple-fold.c (optimize_atomic_compare_exchange_p):
4938 Use sanitize_flags_p.
4939 * gimplify.c (gimplify_function_tree): Likewise.
4940 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
4941 * opts.c (parse_no_sanitize_attribute): New function.
4942 (common_handle_option): Use renamed enum value.
4943 * opts.h (parse_no_sanitize_attribute): Declare.
4944 * tree.c (sanitize_flags_p): New function.
4945 * tree.h: Declared here.
4946 * tsan.c: Use sanitize_flags_p.
4947 * ubsan.c (ubsan_expand_null_ifn): Likewise.
4948 (instrument_mem_ref): Likewise.
4949 (instrument_bool_enum_load): Likewise.
4950 (do_ubsan_in_current_function): Remove the function.
4951 (pass_ubsan::execute): Use sanitize_flags_p.
4952 * ubsan.h: Remove do_ubsan_in_current_function
4953 * tree-cfg.c (print_no_sanitize_attr_value): New function.
4954 (dump_function_to_file): Use it here.
4956 2017-06-13 Martin Jambor <mjambor@suse.cz>
4958 PR tree-optimization/80803
4959 PR tree-optimization/81063
4960 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
4961 (propagate_subaccesses_across_link): Enqueue subtree whenever
4962 necessary instead of relying on the caller.
4964 2017-06-13 Martin Jambor <mjambor@suse.cz>
4966 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
4967 that have a first_link.
4968 (sort_and_splice_var_accesses): Do not check first_link before
4970 (subtree_mark_written_and_enqueue): Likewise.
4971 (propagate_all_subaccesses): Likewise and do not stop at first
4972 parent with a first_link.
4974 2017-06-13 Martin Jambor <mjambor@suse.cz>
4976 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
4979 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4981 * match.pd: New pattern.
4983 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4985 * tree-vrp.c (is_masked_range_test): New function.
4986 (register_edge_assert_for): Determine ranges for
4989 2017-06-13 Yury Gribov <tetra2005@gmail.com>
4991 PR tree-optimization/67328
4992 * fold-const.c (maskable_range_p): New function.
4993 (build_range_check): Generate bittests if possible.
4995 2017-06-13 Martin Liska <mliska@suse.cz>
4997 * gimple-pretty-print.c (dump_probability): Add new argument.
4998 (dump_edge_probability): Dump both probability and count.
4999 (dump_gimple_label): Likewise.
5000 (dump_gimple_bb_header): Likewise.
5002 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5005 * config/avr/avr-devices.c: Fix indentation.
5006 * config/avr/gen-avr-mmcu-specs.c: Dito.
5008 2017-06-13 Richard Biener <rguenther@suse.de>
5010 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5011 instead get vector type from stmt_info.
5012 (vectorizable_reduction): Adjust. Remove dead code.
5014 2017-06-13 Richard Biener <rguenther@suse.de>
5017 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5018 case of C * (x * C2 + C3).
5019 (fold_addr_of_array_ref_difference): Properly fold index difference.
5021 2017-06-12 David S. Miller <davem@davemloft.net>
5024 * config/sparc/sparc.md (return expander): Emit frame blockage if
5025 function uses alloca.
5027 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5029 * combine.c (make_field_assignment): Check len rather than the mode
5030 precision when calling force_to_mode.
5032 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5034 Support multilibs and devices that see flash in RAM address range.
5037 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5038 (avr_mcu_t) <flash_pm_offset>: New field.
5039 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5040 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5041 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5042 (AVR_TINY_PM_OFFSET): Remove macro.
5043 * config/avr/avr.opt (-mshort-calls): New option.
5044 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5045 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5046 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5047 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5048 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5049 instead of avr_arch->have_jmp_call.
5050 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5051 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5052 avr_arch->flash_pm_offset to define.
5053 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5054 new field flash_pm_offset. Add entry for avrxmega3.
5055 (avr_texinfo): Add entry for avrxmega3.
5056 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5057 attiny212, attiny214,
5058 attiny412, attiny414, attiny416, attiny417,
5059 attiny814, attiny816, attiny817,
5060 attiny1614, attiny1616, attiny1617,
5061 attiny3214, attiny3216, attiny3217.
5062 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5063 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5064 (avr_print_operand_address) [AVR_TINY]: Same.
5065 (avr_asm_init_sections) <readonly_data_section>: Only patch
5066 callback if avr_arch->flash_pm_offset = 0.
5067 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5068 for rodata if avr_arch->flash_pm_offset != 0.
5069 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5070 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5071 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5072 (m_options): Append opt_rcall.
5073 (m_dirnames): Append dir_rcall.
5074 * config/avr/t-multilib: Regenerate.
5076 * configure.ac [target=avr]: Check whether avrxmega3 default
5077 linker description file works as needed.
5078 * configure: Regenerate.
5079 * doc/avr-mmcu.texi: Regenerate.
5080 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5081 <__AVR_ARCH__>: Document avrxmega3 and 103.
5082 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5083 <__AVR_SHORT_CALLS__>: Document it.
5084 <__AVR_PM_BASE_ADDRESS__>: Document it.
5085 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5086 (AVR Variable Attributes) <progmem>: Document this is
5087 not needed for avrxmega3.
5088 (AVR Named Address Spaces) <__flash>: Dito.
5090 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5092 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5094 2017-06-12 Doug Rupp <rupp@adacore.com>
5096 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5097 Append vxworks-stdint.h to the tm_file list.
5098 * config/vxworks-stdint.h: New file.
5100 2017-06-12 Martin Liska <mliska@suse.cz>
5102 PR tree-optimization/81041
5103 * tree-profile.c (gimple_gen_ic_func_profiler):
5104 Create an extra BB in profile-generate
5105 (gimple_gen_time_profiler): Likewise.
5107 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5109 PR tree-optimization/81003
5110 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5111 (update_range_test): Use it instead of force_gimple_operand_gsi.
5113 2017-06-12 Richard Biener <rguenther@suse.de>
5115 PR tree-optimization/81053
5116 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5117 with backedge value not defined in loop. Simplify def stmt
5120 2017-06-11 Tom de Vries <tom@codesourcery.com>
5123 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5125 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5126 nvptx_cannot_force_const_mem.
5128 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5130 * opts.c (finish_options): Move test for flag_split_stack after
5131 it has been initialized.
5133 2017-06-11 Jason Merrill <jason@redhat.com>
5135 * tree.h (id_equal): New.
5136 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5137 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5138 instead of strcmp of IDENTIFIER_POINTER.
5140 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5142 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5143 (mark_all_inlined_calls_cdtor): Fix formating.
5144 (inline_transform): Rescale profile before inlining.
5146 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5148 * cgraph.h (cgraph_edge::clone): Update prototype.
5149 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5150 (cgraph_node::create_clone): Update.
5151 (cgraph_node::create_version_clone): Update.
5152 * tree-inline.c (copy_bb): Update.
5153 (expand_call_inline): Update.
5155 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5157 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5158 factored out from ...
5159 (rs6000_emit_prologue): ... here.
5161 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5163 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5164 factored out from ...
5165 (rs6000_emit_prologue): ... here.
5167 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5169 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5171 (handle_missing_profiles): Fix computation of tp_first_run.
5172 (counts_to_freqs): Do not touch freqs when count is 0.
5174 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5176 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5179 2017-06-10 Tom de Vries <tom@codesourcery.com>
5181 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5182 attributes): Document signal effective target.
5184 2017-06-10 Tom de Vries <tom@codesourcery.com>
5186 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5187 Document effective target stack_size.
5189 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5191 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5192 to the edit_context if they can be auto-applied.
5194 2017-06-9 Ian Lance Taylor <iant@golang.org>
5196 * opts.c (finish_options): If -fsplit-stack, disable implicit
5197 -forder-blocks-and-partition.
5198 * doc/invoke.texi (Optimize Options): Document that when using
5199 -fsplit-stack -forder-blocks-and-partition is not implicitly
5202 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5204 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5205 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5206 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5207 * builtins.def (abort, trap, unreachable): Declare cold.
5208 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5209 * tree-core.h (ECF_COLD): New.
5210 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5211 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5213 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5215 * predict.c (unlikely_executed_stmt_p): Cleanup.
5217 2017-06-09 Richard Biener <rguenther@suse.de>
5219 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5220 model if the ref is always written to.
5222 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5224 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5226 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5228 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5231 2017-06-09 Tom de Vries <tom@codesourcery.com>
5234 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5235 "target cannot support label values" when encountering LABEL_REF.
5237 2017-06-09 Martin Liska <mliska@suse.cz>
5239 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5240 (gimple_gen_ic_func_profiler): Emit direct comparison
5241 of __gcov_indirect_call_callee with NULL.
5242 (gimple_gen_time_profiler): Change probability from
5243 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5245 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5247 * profile.c (edge_gcov_counts): Turn to pointer.
5248 (compute_branch_probabilities, compute_branch_probabilities): Update.
5249 (branch_prob): Do not clear edge_gcov_count.
5250 * profile.h (edge_gcov_counts): Turn to pointer.
5251 (edge_gcov_count): Update.
5253 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5255 * gimple.h (gimple_check_failed): Mark cold.
5257 2017-06-09 Richard Biener <rguenther@suse.de>
5259 PR tree-optimization/66623
5260 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5261 refactor check_reduction into two parts, properly computing
5262 whether we have to check reduction validity for outer loop
5265 2017-06-09 Richard Biener <rguenther@suse.de>
5267 PR tree-optimization/79483
5268 * graphite-scop-detection.c (order): New global.
5269 (get_order): Compute bb to order mapping that satisfies code
5270 generation constraints.
5271 (cmp_pbbs): New helper.
5272 (build_scops): Start domwalk at entry block, sort generated
5275 2017-06-09 Richard Biener <rguenther@suse.de>
5278 * ipa-polymorphic-call.c
5279 (ipa_polymorphic_call_context::restrict_to_inner_class):
5280 Skip FIELD_DECLs with error_mark_node type.
5281 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5284 2017-06-09 Martin Liska <mliska@suse.cz>
5286 * predict.c (struct branch_predictor): New struct.
5287 (test_prediction_value_range): New test.
5288 (predict_c_tests): New function.
5289 * selftest-run-tests.c (selftest::run_tests): Run the function.
5290 * selftest.h: Declare new tests.
5292 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5295 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5296 gen_add3_insn did not fail.
5297 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5298 r0, construct that number in a temporary reg and add that reg to r0.
5299 If asked to put the result in r0 as well, fail.
5301 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5303 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5304 for early expansion of vec_eqv.
5306 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5309 * ubsan.c (instrument_null): Avoid pointless code temporary.
5310 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5312 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5316 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5319 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5322 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5324 * predict.c (unlikely_executed_edge_p): Move ahead.
5325 (probably_never_executed_edge_p): Use it.
5327 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5330 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5331 gimple_call_builtin_p call.
5333 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5335 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5336 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5337 rtl_check_failed_type2, rtl_check_failed_code1,
5338 rtl_check_failed_code2, rtl_check_failed_code_mode,
5339 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5340 rtvec_check_failed_bounds, rtl_check_failed_flag,
5341 _fatal_insn_not_found, _fatal_insn): Likewise.
5342 * tree.h (tree_contains_struct_check_failed,
5343 tree_check_failed, tree_not_check_failed,
5344 tree_class_check_failed, tree_range_check_failed,
5345 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5346 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5347 tree_operand_check_failed, omp_clause_check_failed,
5348 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5351 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5353 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5354 flag_branch_probabilities.
5355 * ipa-inline.c (edge_badness): Likewise.
5356 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5357 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5358 * predict.c (maybe_hot_frequency_p): Likewise.
5359 (probably_never_executed): Likewise.
5360 * sched-ebb.c (schedule_ebbs): Likewise.
5361 * sched-rgn.c (find_single_block_region): Likewise.
5362 * tracer.c (tail_duplicate): Likewise.
5364 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5366 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5367 longer requires x_flag_profile_use.
5369 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5371 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5372 instead of flag_reorder_blocks_and_partition.
5373 * dbxout.c (dbxout_function_end): Likewise.
5374 * dwarf2out.c (gen_subprogram_die): Likewise.
5375 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5376 * hw-doloop.c (reorg_loops): Likewise.
5377 * varasm.c (assemble_start_function,
5378 assemble_end_function): Likewise.
5379 (decide_function_section): Do not check for
5380 flag_reorder_blocks_and_partition.
5382 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5384 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5386 (chkp_get_hard_register_fake_addr_expr): Ditto.
5387 (chkp_build_addr_expr): Add check for hard reg case.
5388 (chkp_parse_array_and_component_ref): Ditto.
5389 (chkp_find_bounds_1): Ditto.
5390 (chkp_process_stmt): Don't generate bounds store for
5393 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5395 * predict.c (maybe_hot_bb_p): Do not check profile status.
5396 (maybe_hot_edge_p): Likewise.
5397 (probably_never_executed): Check for zero counts even if profile
5399 (unlikely_executed_edge_p): New function.
5400 (unlikely_executed_stmt_p): New function.
5401 (unlikely_executed_bb_p): New function.
5402 (set_even_probabilities): Use unlikely predicates.
5403 (combine_predictions_for_bb): Likewise.
5404 (predict_paths_for_bb): Likewise.
5405 (predict_paths_leading_to_edge): Likewise.
5406 (determine_unlikely_bbs): New function.
5407 (estimate_bb_frequencies): Use it.
5408 (compute_function_frequency): Use zero counts even if profile is
5410 * profile-count.h: Fix typo.
5412 2017-08-08 Julia Koval <julia.koval@intel.com>
5414 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5415 _mm512_mask_cvtsepi16_storeu_epi8,
5416 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5417 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5418 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5419 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5420 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5421 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5422 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5423 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5424 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5425 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5426 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5427 __builtin_ia32_pmovuswb256mem_mask,
5428 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5429 __builtin_ia32_pmovwb512mem_mask): New builtins.
5431 2017-08-08 Julia Koval <julia.koval@intel.com>
5433 PR target/73350,80862
5434 * config/i386/subst.md (round): Fix round pattern.
5435 * config/i386/i386.c (ix86_erase_embedded_rounding):
5436 Fix erasing rounding for the fixed pattern.
5438 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5440 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5442 2017-06-08 Martin Liska <mliska@suse.cz>
5444 PR gcov-profile/80911
5445 * gcov.c (block_info::block_info): New constructor.
5447 2017-06-07 Carl Love <cel@us.ibm.com>
5449 * config/rs6000/rs6000-c: The return type of the following
5450 built-in functions was implemented as int not long long. Fix sign
5451 of return value for the unsigned version of vec_mulo and vec_mule.
5452 vector unsigned long long vec_bperm (vector unsigned long long,
5453 vector unsigned char)
5454 vector signed long long vec_mule (vector signed int,
5456 vector unsigned long long vec_mule (vector unsigned int,
5457 vector unsigned int)
5458 vector signed long long vec_mulo (vector signed int,
5460 vector unsigned long long vec_mulo (vector unsigned int,
5461 vector unsigned int)
5462 * doc/extend.texi: Fix the documentation for the built-in
5465 2017-06-07 Carl Love <cel@us.ibm.com>
5468 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5471 2017-06-07 Carl Love <cel@us.ibm.com>
5473 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5474 support, Generate doublehv for signed int/float for BE case only.
5476 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5478 * doc/invoke.texi (mcx16): Rewrite.
5480 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5482 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5483 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5484 *mov<mode>_softfloat, and an anonymous splitter): Use
5485 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5487 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5489 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5491 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5492 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5493 (rs6000_debug_reg_global): Adjust.
5494 (rs6000_init_hard_regno_mode_ok): Adjust.
5495 (rs6000_dbx_register_number): Adjust.
5496 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5497 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5498 Remove SPE_ACC and SPEFSCR.
5499 (REG_ALLOC_ORDER): Ditto.
5500 (FRAME_POINTER_REGNUM): Change to 111.
5501 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5502 (REG_CLASS_NAMES): Ditto.
5503 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5504 (REGISTER_NAMES): Ditto.
5505 (ADDITIONAL_REG_NAMES): Ditto.
5506 (rs6000_reg_names): Ditto.
5507 * config/rs6000/rs6000.md: Renumber some register number
5510 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5512 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5514 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5515 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5517 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5518 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5519 Delete the SPE high registers.
5520 (REG_ALLOC_ORDER): Ditto.
5521 (enum reg_class): Remove SPE_HIGH_REGS.
5522 (REG_CLASS_NAMES): Ditto.
5523 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5524 (REGISTER_NAMES): Ditto.
5525 (rs6000_reg_names): Ditto.
5526 * doc/tm.texi.in: Remove SPE as example.
5527 * doc/tm.texi: Regenerate.
5529 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5531 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5532 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5533 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5534 * config/rs6000/rs6000.md (type): Remove "brinc".
5536 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5538 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5539 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5540 * config/rs6000/linuxspe.h: Delete file.
5541 * config/rs6000/rs6000.md: Don't include spe.md.
5542 * config/rs6000/spe.h: Delete file.
5543 * config/rs6000/spe.md: Delete file.
5544 * config/rs6000/t-rs6000: Remove spe.md.
5546 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5548 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5549 (reg_or_none500mem_operand): Delete.
5550 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5551 instead of reg_or_none500mem_operand.
5553 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5555 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5556 handling of SPE flags.
5557 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5559 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5561 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5563 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5564 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5565 paired_divv2sf3): Similar.
5566 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5567 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5568 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5570 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5571 Rename the paired_* instruction patterns.
5572 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5574 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5575 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5576 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5577 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5578 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5580 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5581 (spe_func_has_64bit_regs_p): Delete.
5582 (spe_expand_predicate_builtin): Delete.
5583 (spe_expand_evsel_builtin): Delete.
5584 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5585 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5586 (invalid_e500_subreg): Delete.
5587 (rs6000_legitimize_address): Always force_reg op2 as well, for
5588 paired single memory accesses.
5589 (rs6000_member_type_forces_blk): Delete.
5590 (rs6000_spe_function_arg): Delete.
5591 (rs6000_expand_unop_builtin): Delete SPE handling.
5592 (rs6000_expand_binop_builtin): Ditto.
5593 (spe_expand_stv_builtin): Delete.
5594 (bdesc_2arg_spe): Delete.
5595 (spe_expand_builtin): Delete.
5596 (spe_expand_predicate_builtin): Delete.
5597 (spe_expand_evsel_builtin): Delete.
5598 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5599 (spe_init_builtins): Delete.
5600 (spe_func_has_64bit_regs_p): Delete.
5601 (savres_routine_name): Delete "info" parameter. Adjust callers.
5602 (rs6000_emit_stack_reset): Ditto.
5603 (rs6000_dwarf_register_span): Delete.
5604 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5605 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5606 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5608 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5610 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5611 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5612 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5613 mulv2sf3, divv2sf3): Delete expanders.
5615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5617 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5619 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5621 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5622 * config/rs6000/rs6000.c: Ditto.
5624 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5626 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5627 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5628 comparison_operator.
5630 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5632 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5633 * config/rs6000/rs6000.opt: Ditto.
5634 * config/rs6000/t-rtems: Ditto.
5636 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5638 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5639 TARGET_E500_SINGLE by 0, simplify.
5640 * config/rs6000/rs6000.c: Ditto.
5641 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5642 (spe_build_register_parallel): Delete.
5643 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5644 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5645 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5646 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5647 (E500_CONVERT): Delete.
5648 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5650 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5652 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5653 * config/rs6000/dfp.md: Ditto.
5654 (negdd2, *negdd2_fpr): Merge.
5655 (absdd2, *absdd2_fpr): Merge.
5656 (negtd2, *negtd2_fpr): Merge.
5657 (abstd2, *abstd2_fpr): Merge.
5658 * config/rs6000/e500.h: Delete file.
5659 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5660 TARGET_FPRS by 1 and simplify.
5661 * config/rs6000/rs6000-c.c: Ditto.
5662 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5664 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5666 * config/rs6000/rs6000.md: Ditto.
5667 (floatdidf2, *floatdidf2_fpr): Merge.
5668 (move_from_CR_gt_bit): Delete.
5669 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5670 (E500_CR_IOR_COMPARE): Delete.
5671 (All patterns that require !TARGET_FPRS): Delete.
5672 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5674 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5676 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5678 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5680 * graphds.c (add_edge): Intitialize edge's attached data.
5681 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5682 pointer parameter. Call pointed function on each edge during
5683 graph traversing. Skip traversing the edge when the function
5685 (graphds_dfs, graphds_scc): Ditto.
5686 (for_each_edge): New parameter. Pass the new parameter to callback
5688 * graphds.h (skip_edge_callback): New function pointer type.
5689 (graphds_dfs, graphds_scc): New function pointer parameter.
5690 (graphds_edge_callback, for_each_edge): New parameter.
5692 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5694 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5695 out code checking if runtime alias check is possible to below ...
5696 Call the new function.
5697 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5698 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5700 2017-06-07 Marek Polacek <polacek@redhat.com>
5703 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5704 TYPE_OVERFLOW_WRAPS check.
5706 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5708 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5709 if versioning is required.
5710 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5711 peeling with the check for versioning.
5713 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5715 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5716 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5717 Set true to new parameter if new ssa variable is defined.
5718 (vect_gen_vector_loop_niters): Refactor. Set range information
5719 for the new vector loop bound variable.
5720 (vect_do_peeling): Ditto.
5722 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5724 * tree-affine.c (ssa.h): Include header file.
5725 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5726 has wrapping overflow behavior.
5728 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5730 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5732 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5734 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5735 (tree_to_aff_combination): ... here.
5737 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5739 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5740 reg_pressure model function.
5741 (ivopts_global_cost_for_size): Delete.
5742 (determine_set_costs, iv_ca_recount_cost): Call new model function
5743 ivopts_estimate_reg_pressure.
5745 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5747 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5748 expensive than udiv. Remove floating point cases from mod.
5750 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5752 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5755 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5757 * config/aarch64/aarch64.md
5758 (copysignsf3): Fix mask generation.
5760 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5762 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5764 (class dump_manager): Add register_dumps method.
5765 * dumpfile.c: Include langhooks.h.
5766 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5767 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
5768 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
5769 (dump_manager::dump_register): Start with 512 entries instead of 32.
5770 (dump_manager::register_dumps): New method.
5771 * toplev.c (general_init): Instead of invoking register_dumps
5772 langhook, invoke register_dumps method on the dump manager.
5773 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
5776 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
5778 * doc/md.texi: Clarify the restrictions on a define_insn condition.
5779 Say that # requires an associated define_split to exist, and that
5780 the define_split must be suitable for use after register allocation.
5782 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5784 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
5785 (compute_outgoing_frequencies): Also initialize zero counts.
5786 (find_many_sub_basic_blocks): Do not produce uninitialized profile
5787 around loops; preserve more of profile when nothing changes.
5789 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
5791 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
5793 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
5794 * config/arm/arm-cpu-cdata.h: Regenerate.
5795 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
5796 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
5797 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
5798 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
5799 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
5801 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
5803 * config/arm/t-rmprofile: Likewise.
5804 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
5806 2017-06-06 David S. Miller <davem@davemloft.net>
5809 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
5810 blockage if function uses alloca.
5812 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5814 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
5815 New "uid" fields to hold pretty-print IDs of group and ref.
5816 Memory references are now identified as <group_id>:<ref_id>
5817 instead of using [random] addresses.
5818 (dump_mem_details): Simplify, no functional change.
5819 (dump_mem_ref): Simplify and make output more concise.
5820 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
5821 (find_or_create_group): Initialize group uid.
5822 (record_ref): Initialize ref uid. Improve debug output.
5823 (prune_group_by_reuse, should_issue_prefetch_p,)
5824 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
5825 (mark_nontemporal_store, determine_loop_nest_reuse):
5826 Improve debug output.
5828 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5830 * dbgcnt.def (prefetch): New debug counter.
5831 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
5832 (schedule_prefetches): Stop issueing prefetches if debug counter
5835 2017-06-06 Tom de Vries <tom@codesourcery.com>
5837 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
5838 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
5840 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5842 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
5843 Use aarch64_reg_or_zero predicate for operand 4.
5844 (aarch64_compare_and_swap<mode> define_insn_and_split):
5845 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
5846 (aarch64_store_exclusive<mode>): Likewise for operand 2.
5848 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5850 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
5851 (arm_compute_save_core_reg_mask): This.
5852 (thumb1_compute_save_reg_mask): Rename into ...
5853 (thumb1_compute_save_core_reg_mask): This.
5854 (arm_compute_save_reg0_reg12_mask): Adapt comment.
5855 (arm_compute_frame_layout): Likewise.
5857 2017-06-06 Richard Biener <rguenther@suse.de>
5859 PR tree-optimization/80974
5860 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
5861 keep or clear leaders SSA info.
5863 2017-06-06 Tom de Vries <tom@codesourcery.com>
5865 * config/nvptx/nvptx.c (split_mode_p): New function.
5866 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
5868 2017-06-06 Tom de Vries <tom@codesourcery.com>
5870 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
5872 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5875 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
5878 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5880 * shrink-wrap.c (handle_simple_exit): Update profile.
5881 (try_shrink_wrapping): Upate profile.
5883 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5885 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
5886 (tree_guess_outgoing_edge_probabilities): New.
5887 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
5888 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
5890 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5892 * ipa-split.c (split_function): Initialize return bb profile.
5894 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5896 * profile.c (compute_branch_probabilities): Also initialize
5899 2017-06-06 Richard Biener <rguenther@suse.de>
5901 PR tree-optimization/80928
5902 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
5903 (vect_analyze_loop_operations): Properly guard analysis for
5905 (vect_transform_loop): Likewise.
5906 (vect_analyze_loop_2): Also reset SLP type on PHIs.
5907 (vect_model_induction_cost): Do not cost for pure SLP.
5908 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
5909 of induction in inner loop vectorization.
5910 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
5911 (vect_get_and_check_slp_defs): Handle vect_induction_def.
5912 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
5914 (vect_analyze_slp_cost_1): Cost induction.
5915 (vect_detect_hybrid_slp_stmts): Handle PHIs.
5916 (vect_get_slp_vect_defs): Likewise.
5917 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
5918 (vect_transform_stmt): Handle SLP reductions.
5919 * tree-vectorizer.h (vectorizable_induction): Adjust.
5921 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5923 * config/rs6000/rs6000.c (make_resolver_func): Update
5924 init_lowered_empty_function call.
5926 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
5928 * doc/invoke.texi: Document the -fprofile-abs-path option.
5929 * common.opt (fprofile-abs-path): New option.
5930 * gcov-io.h (gcov_write_filename): Declare.
5931 * gcov-io.c (gcov_write_filename): New function.
5932 * coverage.c (coverage_begin_function): Use gcov_write_filename.
5933 * profile.c (output_location): Likewise.
5935 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5937 * shring-wrap.c: Revert accidental commit.
5939 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
5941 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
5943 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5945 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
5947 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
5948 profile in callgraph edge.
5949 * profile-count.h (apply_probability): If THIS is 0, then result is 0
5950 (apply_scale): Likewise.
5951 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
5952 Also scale profile when inlining function with zero profile.
5953 (initialize_cfun): Update exit block profile even when it is zero.
5954 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
5955 when profile is read.
5957 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5959 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
5960 (CLONE_*): New constants to define the processors we can generate
5961 code for with the target_clone attribute.
5962 (rs6000_clone_map): New array to identify which clone processors
5963 the current program is running on.
5964 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
5965 target_clone attribute.
5966 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
5967 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
5968 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
5969 (cpu_expand_builtin): Add support for target_clone attribute.
5970 (rs6000_valid_attribute_p): Allow "default" attribute.
5971 (get_decl_name): New debug function to simplify printing the
5972 current function name in debugging statements.
5973 (rs6000_clone_priority): New functions to support the target_clone
5974 attribute, and be able to generate code to switch between ISA 2.05
5975 through ISA 3.0 (power6 through power9).
5976 (rs6000_compare_version_priority): Likewise.
5977 (rs6000_get_function_versions_dispatcher): Likewise.
5978 (make_resolver_func): Likewise.
5979 (add_condition_to_bb): Likewise.
5980 (dispatch_function_versions): Likewise.
5981 (rs6000_generate_version_dispatcher_body): Likewise.
5982 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
5983 (fusion_gpr_load_p): Fix a spacing issue.
5984 * doc/extend.texi (Common Function Attributes): Document that the
5985 PowerPC supports the target_clone attribute.
5987 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
5989 * config/arm/arm.h: explain F symbol found in description of ARM
5990 register allocation in its legend.
5992 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5994 * config/mips/frame-header-opt.c: Include profile-count.h.
5995 * config/riscv/riscv.c: Include profile-count.h
5997 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5999 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6001 (sm_set_flag_if_changed): Add bbs field.
6002 (execute_sm_if_changed_flag_set): Pass BBS.
6003 (execute_sm): Update.
6005 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6007 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6010 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6012 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6013 (peephole2): New peephole2 to emit the above.
6014 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6016 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6018 * config/aarch64/aarch64.c (define_peephole2 above
6019 *sub_<shift>_<mode>): New peephole.
6021 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6023 * config/i386/i386.c (make_resolver_func): Update.
6024 * Makefile.in: Add profile-count.h and profile-count.o
6025 * auto-profile.c (afdo_indirect_call): Update to new API.
6026 (afdo_set_bb_count): Update.
6027 (afdo_propagate_edge): Update.
6028 (afdo_propagate_circuit): Update.
6029 (afdo_calculate_branch_prob): Update.
6030 (afdo_annotate_cfg): Update.
6031 * basic-block.h: Include profile-count.h
6032 (struct edge_def): Turn count to profile_count.
6033 (struct basic_block_def): Likewie.
6034 (REG_BR_PROB_BASE): Move to profile-count.h
6035 (RDIV): Move to profile-count.h
6036 * bb-reorder.c (max_entry_count): Turn to profile_count.
6037 (find_traces): Update.
6038 (rotate_loop):Update.
6039 (connect_traces):Update.
6040 (sanitize_hot_paths):Update.
6041 * bt-load.c (migrate_btr_defs): Update.
6042 * cfg.c (RDIV): Remove.
6043 (init_flow): Use alloc_block.
6044 (alloc_block): Uninitialize count.
6045 (unchecked_make_edge): Uninitialize count.
6046 (check_bb_profile): Update.
6047 (dump_edge_info): Update.
6048 (dump_bb_info): Update.
6049 (update_bb_profile_for_threading): Update.
6050 (scale_bbs_frequencies_int): Update.
6051 (scale_bbs_frequencies_gcov_type): Update.
6052 (scale_bbs_frequencies_profile_count): New.
6053 * cfg.h (update_bb_profile_for_threading): Update.
6054 (scale_bbs_frequencies_profile_count): Declare.
6055 * cfgbuild.c (compute_outgoing_frequencies): Update.
6056 (find_many_sub_basic_blocks): Update.
6057 * cfgcleanup.c (try_forward_edges): Update.
6058 (try_crossjump_to_edge): Update.
6059 * cfgexpand.c (expand_gimple_tailcall): Update.
6060 (construct_exit_block): Update.
6061 * cfghooks.c (verify_flow_info): Update.
6062 (dump_bb_for_graph): Update.
6063 (split_edge): Update.
6064 (make_forwarder_block): Update.
6065 (duplicate_block): Update.
6066 (account_profile_record): Update.
6067 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6068 (get_estimated_loop_iterations): Update.
6069 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6070 (single_likely_exit): Update.
6071 * cfgloopmanip.c (scale_loop_profile): Update.
6073 (set_zero_probability): Update.
6074 (lv_adjust_loop_entry_edge): Update.
6075 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6076 (purge_dead_edges): Update.
6077 (rtl_account_profile_record): Update.
6078 * cgraph.c (cgraph_node::create): Uninitialize count.
6079 (symbol_table::create_edge): Uninitialize count.
6080 (cgraph_update_edges_for_call_stmt_node): Update.
6081 (cgraph_edge::dump_edge_flags): Update.
6082 (cgraph_node::dump): Update.
6083 (cgraph_edge::maybe_hot_p): Update.
6084 * cgraph.h: Include profile-count.h
6085 (create_clone), create_edge, create_indirect_edge): Update.
6086 (cgraph_node): Turn count to profile_count.
6087 (cgraph_edge0: Likewise.
6088 (make_speculative, clone): Update.
6089 (create_edge): Update.
6090 (init_lowered_empty_function): Update.
6091 * cgraphclones.c (cgraph_edge::clone): Update.
6092 (duplicate_thunk_for_node): Update.
6093 (cgraph_node::create_clone): Update.
6094 * cgraphunit.c (cgraph_node::analyze): Update.
6095 (cgraph_node::expand_thunk): Update.
6096 * final.c (dump_basic_block_info): Update.
6097 * gimple-streamer-in.c (input_bb): Update.
6098 * gimple-streamer-out.c (output_bb): Update.
6099 * graphite.c (print_global_statistics): Update.
6100 (print_graphite_scop_statistics): Update.
6101 * hsa-brig.c: Include basic-block.h.
6102 * hsa-dump.c: Include basic-block.h.
6103 * hsa-gen.c (T sum_slice): Update.
6104 (convert_switch_statements):Update.
6105 * hsa-regalloc.c: Include basic-block.h.
6106 * ipa-chkp.c (chkp_produce_thunks): Update.
6107 * ipa-cp.c (struct caller_statistics): Update.
6108 (init_caller_stats): Update.
6109 (gather_caller_stats): Update.
6110 (ipcp_cloning_candidate_p): Update.
6111 (good_cloning_opportunity_p): Update.
6112 (get_info_about_necessary_edges): Update.
6113 (dump_profile_updates): Update.
6114 (update_profiling_info): Update.
6115 (update_specialized_profile): Update.
6116 (perhaps_add_new_callers): Update.
6117 (decide_about_value): Update.
6118 (ipa_cp_c_finalize): Update.
6119 * ipa-devirt.c (struct odr_type_warn_count): Update.
6120 (struct decl_warn_count): Update.
6121 (struct final_warning_record): Update.
6122 (possible_polymorphic_call_targets): Update.
6123 (ipa_devirt): Update.
6124 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6125 * ipa-icf.c (sem_function::merge): Update.
6126 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6127 * ipa-inline.c (compute_uninlined_call_time): Update.
6128 (compute_inlined_call_time): Update.
6129 (want_inline_small_function_p): Update.
6130 (want_inline_self_recursive_call_p): Update.
6131 (edge_badness): Update.
6132 (lookup_recursive_calls): Update.
6133 (recursive_inlining): Update.
6134 (inline_small_functions): Update.
6135 (dump_overall_stats): Update.
6136 (dump_inline_stats): Update.
6137 * ipa-profile.c (ipa_profile_generate_summary): Update.
6138 (ipa_propagate_frequency): Update.
6139 (ipa_profile): Update.
6140 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6141 * ipa-utils.c (ipa_merge_profiles): Update.
6142 * loop-doloop.c (doloop_modify): Update.
6143 * loop-unroll.c (report_unroll): Update.
6144 (unroll_loop_runtime_iterations): Update.
6145 * lto-cgraph.c (lto_output_edge): Update.
6146 (lto_output_node): Update.
6147 (input_node): Update.
6148 (input_edge): Update.
6149 (merge_profile_summaries): Update.
6150 * lto-streamer-in.c (input_cfg): Update.
6151 * lto-streamer-out.c (output_cfg): Update.
6152 * mcf.c (create_fixup_graph): Update.
6153 (adjust_cfg_counts): Update.
6154 (sum_edge_counts): Update.
6155 * modulo-sched.c (sms_schedule): Update.
6156 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6157 * predict.c (maybe_hot_count_p): Update.
6158 (probably_never_executed): Update.
6159 (dump_prediction): Update.
6160 (combine_predictions_for_bb): Update.
6161 (propagate_freq): Update.
6162 (handle_missing_profiles): Update.
6163 (counts_to_freqs): Update.
6164 (rebuild_frequencies): Update.
6165 (force_edge_cold): Update.
6166 * predict.h: Include profile-count.h
6167 (maybe_hot_count_p, counts_to_freqs): UPdate.
6168 * print-rtl-function.c: Do not include cfg.h
6169 * print-rtl.c: Include basic-block.h
6170 * profile-count.c: New file.
6171 * profile-count.h: New file.
6172 * profile.c (is_edge_inconsistent): Update.
6173 (correct_negative_edge_counts): Update.
6174 (is_inconsistent): Update.
6175 (set_bb_counts): Update.
6176 (read_profile_edge_counts): Update.
6177 (compute_frequency_overlap): Update.
6178 (compute_branch_probabilities): Update; Initialize and deinitialize
6180 (branch_prob): Update.
6181 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6182 (edge_gcov_count): New.
6183 (bb_gcov_count): New.
6184 * shrink-wrap.c (try_shrink_wrapping): Update.
6185 * tracer.c (better_p): Update.
6186 * trans-mem.c (expand_transaction): Update.
6187 (ipa_tm_insert_irr_call): Update.
6188 (ipa_tm_insert_gettmclone_call): Update.
6189 * tree-call-cdce.c: Update.
6190 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6191 (gimple_duplicate_sese_tail): Update.
6192 (gimple_account_profile_record): Update.
6193 (execute_fixup_cfg): Update.
6194 * tree-inline.c (copy_bb): Update.
6195 (copy_edges_for_bb): Update.
6196 (initialize_cfun): Update.
6197 (freqs_to_counts): Update.
6198 (copy_cfg_body): Update.
6199 (expand_call_inline): Update.
6200 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6201 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6202 (try_unroll_loop_completely): Update.
6203 (try_peel_loop): Update.
6204 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6205 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6206 * tree-ssa-loop-split.c (connect_loops): Update.
6207 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6208 * tree-ssa-reassoc.c (branch_fixup): Update.
6209 * tree-ssa-tail-merge.c (replace_block_by): Update.
6210 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6211 (compute_path_counts): Update.
6212 (update_profile): Update.
6213 (recompute_probabilities): Update.
6214 (update_joiner_offpath_counts): Update.
6215 (estimated_freqs_path): Update.
6216 (freqs_to_counts_path): Update.
6217 (clear_counts_path): Update.
6218 (ssa_fix_duplicate_block_edges): Update.
6219 (duplicate_thread_path): Update.
6220 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6221 (struct switch_conv_info): Update.
6222 * tree-tailcall.c (decrease_profile): Update.
6223 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6224 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6225 * value-prof.c (check_counter): Update.
6226 (gimple_divmod_fixed_value): Update.
6227 (gimple_mod_pow2): Update.
6228 (gimple_mod_subtract): Update.
6229 (gimple_ic_transform): Update.
6230 (gimple_stringop_fixed_value): Update.
6231 * value-prof.h (gimple_ic): Update.
6233 2017-06-02 Carl Love <cel@us.ibm.com>
6235 * config/rs6000/rs6000-c: Add support for built-in functions
6236 vector double vec_doublee (vector signed int);
6237 vector double vec_doublee (vector unsigned int);
6238 vector double vec_doublee (vector float);
6239 vector double vec_doubleh (vector signed int);
6240 vector double vec_doubleh (vector unsigned int);
6241 vector double vec_doubleh (vector float);
6242 vector double vec_doublel (vector signed int);
6243 vector double vec_doublel (vector unsigned int);
6244 vector double vec_doublel (vector float);
6245 vector double vec_doubleo (vector signed int);
6246 vector double vec_doubleo (vector unsigned int);
6247 vector double vec_doubleo (vector float);.
6248 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6249 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6251 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6252 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6253 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6255 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6256 vec_doublel, vec_doubleh.
6257 * doc/extend.texi: Update the built-in documentation file for the
6258 new built-in functions.
6260 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6263 * ipa-inline-analysis.c (free_growth_caches): Set
6264 edge_removal_hook_holder to NULL after removing it.
6266 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6268 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6269 comparision with zero.
6271 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6272 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6273 for early expansion of vec_min and vec_max builtins.
6274 (builtin_function_type): Add min/max unsigned variants to those
6275 identified as having unsigned arguments.
6277 2017-06-02 Olivier Hainque <hainque@adacore.com>
6279 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6281 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6283 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6284 Use VALL_F16 iterator rather than VALL.
6286 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6288 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6289 Emit CBNZ inside loop when doing a strong exchange and comparing
6290 against zero. Generate the CC flags after the loop.
6292 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6294 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6295 (dl_section_ref): New.
6296 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6297 On AIX, append an expression to subtract the size of the
6298 section length to dl_section_ref.
6300 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6302 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6303 for early expansion of vector absolute builtins.
6305 2017-06-02 Richard Biener <rguenther@suse.de>
6307 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6308 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6310 2017-06-02 Richard Biener <rguenther@suse.de>
6312 PR tree-optimization/80948
6313 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6316 2017-06-02 Richard Biener <rguenther@suse.de>
6318 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6320 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6321 for inductions as relevant.
6323 2017-06-02 Richard Biener <rguenther@suse.de>
6325 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6326 (vectorizable_induction): ... this. Remove dead code.
6328 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6330 * builtins. (expand_builtin_alloca): Remove second parameter and
6331 infer its value from the first parameter instead.
6332 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6334 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6336 PR rtl-optimization/80903
6337 * loop-doloop.c (add_test): Unshare sequence.
6339 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6341 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6343 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6345 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6347 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6348 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6349 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6350 (xlogue_layout::get_stub_rtx): Make static.
6351 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6352 (xlogue_layout::compute_stub_managed_regs): Rename to...
6353 (xlogue_layout::count_stub_managed_regs): ...this.
6354 (xlogue_layout::is_stub_managed_reg): New function.
6355 (xlogue_layout::m_stub_names): Rename to...
6356 (xlogue_layout::s_stub_names): ...this, make static.
6357 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6358 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6359 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6360 xlogue_layout::s_stub_names): Instantiate statics.
6361 (stub_managed_regs): Remove.
6362 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6363 (disable_call_ms2sysv_xlogues): Rename to...
6364 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6365 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6367 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6368 change after reload_completed.
6369 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6371 (ix86_expand_prologue): Likewise.
6372 (ix86_expand_epilogue): Likewise.
6373 (ix86_expand_split_stack_prologue): Likewise.
6374 (ix86_compute_frame_layout): Remove frame parameter ...
6375 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6376 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6378 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6379 (ix86_frame): Move from here ...
6380 * config/i386/i386.h (ix86_frame): ... to here.
6381 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6382 complete ix86_frame data structure instead. Remove some_ld_name.
6384 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6386 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6387 symbols that hold a DECL_VALUE_EXPR.
6389 2017-06-01 Martin Jambor <mjambor@suse.cz>
6391 PR tree-optimization/80898
6392 * tree-sra.c (process_subtree_disqualification): Removed.
6393 (disqualify_candidate): Do not acll
6394 process_subtree_disqualification.
6395 (subtree_mark_written_and_enqueue): New function.
6396 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6397 RHS has been disqualified and re-queue LHS if necessary. Apart
6398 from that, ignore disqualified RHS.
6400 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6402 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6403 address fetch for z10 or later.
6405 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6407 * config/arc/arc.md (tst_movb): Add guard when splitting.
6409 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6411 * config/arc/arc.c (arc_can_eliminate): Test against
6412 arc_frame_pointer_needed.
6414 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6416 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6417 to prevent store reordering.
6418 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6419 (type): Add block type.
6420 (stack_tie): Define special instruction to be used in
6423 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6425 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6426 constraint. It is not valid for the pattern.
6427 (noncommutative_binary_comparison): Likewise.
6429 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6431 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6434 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6436 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6437 be used by the reg-alloc.
6439 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6441 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6442 reg-alloc when having mul64 or mul32x16 instructions.
6443 (mulsidi3): Likewise.
6444 (umulsidi3): Likewise.
6445 (mulsi32x16): New pattern.
6446 (mulsi64): Likewise.
6447 (mulsidi64): Likewise.
6448 (umulsidi64): Likewise.
6449 (MUL32x16_REG): Define.
6450 (mul64_600): Use MUL32x16_REG.
6451 (mac64_600): Likewise.
6452 (umul64_600): Likewise.
6453 (umac64_600): Likewise.
6455 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6457 * config/arc/arc.md (mulsi3_700): Make it commutative.
6459 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6461 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6463 (*sign_extendsidi2_insn): Likewise for movstosw.
6465 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6467 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6468 the type of the input discriminant value. Convert the
6469 discriminant value of signedness vary.
6471 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6473 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6474 Add to -Wall section.
6476 2017-06-01 Richard Biener <rguenther@suse.de>
6479 * fold-const.c (fold_plusminus_mult_expr): If the factored
6480 factor may be zero use a wrapping type for the inner operation.
6481 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6482 and handle moved defs.
6483 (process_assignment): Properly guard the unary op case. Return a
6484 tri-state indicating that moving the stmt before the call may allow
6485 to continue. Pass through to_move.
6486 (find_tail_calls): Handle moving unrelated defs before
6489 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6492 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6493 splitter result in the canonical way.
6495 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6497 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6498 also for 32bit target. Update insn attributes.
6499 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6501 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6503 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6504 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6505 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6506 (_mm_maskz_min_ss): New intrinsics.
6508 2017-05-31 Martin Liska <mliska@suse.cz>
6510 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6511 Change comment style to one we normally use.
6512 (vectorizable_reduction): Likewise.
6513 (vectorizable_induction): Likewise.
6514 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6515 (vectorizable_call): Likewise.
6516 (vectorizable_simd_clone_call): Likewise.
6517 (vectorizable_conversion): Likewise.
6518 (vectorizable_assignment): Likewise.
6519 (vectorizable_shift): Likewise.
6520 (vectorizable_operation): Likewise.
6521 (vectorizable_store): Likewise.
6522 (vectorizable_load): Likewise.
6523 * tree-vectorizer.h: Likewise.
6525 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6527 * passes.c (emergency_dump_function): New.
6528 * tree-pass.h (emergency_dump_function): Declare.
6529 * plugin.c (plugins_internal_error_function): Remove.
6530 * plugin.h (plugins_internal_error_function): Remove declaration.
6531 * toplev.c (internal_error_function): New static function. Use it...
6532 (general_init): ...here.
6534 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6536 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6537 (arc_rtx_costs): Add costs for new patterns.
6538 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6539 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6541 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6543 * tree-ssa-strlen.c (get_next_strinfo): New function.
6544 (get_stridx_plus_constant): Use it.
6545 (zero_length_string): Likewise.
6546 (adjust_related_strinfos): Likewise.
6547 (adjust_last_stmt): Likewise.
6549 2017-05-31 Richard Biener <rguenther@suse.de>
6552 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6553 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6555 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6557 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6558 loop_vinfo argument and use of dependence distance vectors.
6559 Check instead whether the two references differ only in their
6560 initial value and assume that they have the same alignment if the
6561 difference is a multiple of the vector alignment.
6562 (vect_analyze_data_refs_alignment): Update call accordingly.
6564 2017-05-31 Martin Liska <mliska@suse.cz>
6567 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6569 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6571 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6572 (create_intersect_range_checks): Move from ...
6573 * tree-data-ref.c (create_intersect_range_checks_index)
6574 (create_intersect_range_checks): ... to here.
6575 (create_runtime_alias_checks): New function factored from ...
6576 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6577 here. Call above function.
6578 * tree-data-ref.h (create_runtime_alias_checks): New function.
6580 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6582 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6583 segment length for dr_b and compute it in wide_int.
6585 2017-05-31 Richard Biener <rguenther@suse.de>
6587 PR tree-optimization/80906
6588 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6589 and pass through iv_map.
6590 (copy_bb_and_scalar_dependences): Adjust.
6591 (translate_pending_phi_nodes): Likewise.
6592 (copy_loop_close_phi_args): Handle code-generating IVs instead
6595 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6597 * diagnostic-color.c (color_dict): Add "type-diff".
6598 (parse_gcc_colors): Update comment.
6599 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6600 -fdiagnostics-show-template-tree and -fno-elide-type.
6601 (GCC_COLORS): Add type-diff to example.
6603 (-fdiagnostics-show-template-tree): New.
6604 (-fno-elide-type): New.
6605 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6606 the pp_format_decoder callback. Call any m_format_postprocessor's
6608 (pretty_printer::pretty_printer): Initialize
6609 m_format_postprocessor.
6610 (pretty_printer::~pretty_printer): Delete any
6611 m_format_postprocessor.
6612 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6613 (class format_postprocessor): New class.
6614 (struct pretty_printer::format_decoder): Document the new parameters.
6615 (struct pretty_printer::m_format_postprocessor): New field.
6616 * tree-diagnostic.c (default_tree_printer): Update for new
6617 bool and const char ** params.
6618 * tree-diagnostic.h (default_tree_printer): Likewise.
6620 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6622 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6623 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6625 rs6000_gen_cell_microcode code.
6626 (rs6000_final_prescan_insn): Delete.
6627 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6628 "warn-cell-microcode" entries.
6629 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6630 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6631 throughout. Change cc_reg_not_micro_cr0_operand to
6632 cc_reg_not_cr0_operand throughout.
6633 (*extendhi<mode>2_noload): Delete.
6634 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6635 (mwarn-cell-microcode): Delete.
6636 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6637 -mgen-cell-microcode and -mwarn-cell-microcode.
6639 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6642 * config/i386/constraints.md (Yd): New constraint.
6644 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6645 and (?Yd, r) alternatives. Update insn attributes.
6646 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6647 and (?*Yd, r) alternatives. Update insn attributes.
6648 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6650 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6652 * gimplify.c (gimplify_modify_expr): Don't create a
6653 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6656 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6658 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6660 2017-05-30 Richard Biener <rguenther@suse.de>
6662 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6663 and reduc_def fields.
6664 (STMT_VINFO_REDUC_TYPE): New define.
6665 (STMT_VINFO_REDUC_DEF): Likewise.
6666 (vect_force_simple_reduction): Adjust prototype.
6667 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6668 (vect_is_simple_reduction): Remove check_reduction argument.
6669 (vect_force_simple_reduction): Adjust and set
6670 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6671 (vectorizable_reduction): Do not re-do reduction analysis
6672 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6673 * tree-parloops.c (gather_scalar_reductions): Adjust.
6675 2017-05-30 Richard Biener <rguenther@suse.de>
6678 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6681 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6683 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6684 Introduce unknown_misalignment parameter and remove vf.
6685 (vect_peeling_hash_get_lowest_cost):
6686 Pass unknown_misalignment parameter.
6687 (vect_enhance_data_refs_alignment):
6688 Fix unsupportable data ref treatment.
6690 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6692 * tree-vect-data-refs.c (vect_get_data_access_cost):
6693 Workaround for SLP handling.
6694 (vect_enhance_data_refs_alignment):
6695 Compute costs for doing no peeling at all, compare to the best
6696 peeling costs so far and avoid peeling if cheaper.
6698 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6700 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6701 Return peeling info and set costs to zero for unlimited cost
6703 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6704 with unknown misalignment. Compute and costs for unknown
6705 misalignment, compare them to the costs for known misalignment
6706 and choose the cheapest for peeling.
6708 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6710 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6711 (vect_get_peeling_costs_all_drs): Create function.
6712 (vect_peeling_hash_get_lowest_cost):
6713 Use vect_get_peeling_costs_all_drs.
6714 (vect_peeling_supportable): Create function.
6715 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6717 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6719 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6720 DR_HAS_NEGATIVE_STEP.
6721 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6722 (vect_enhance_data_refs_alignment): Use.
6723 (vect_duplicate_ssa_name_ptr_info): Use.
6724 * tree-vectorizer.h (dr_misalignment): Use.
6725 (known_alignment_for_access_p): Use.
6727 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6730 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6732 (has_section_name): New function.
6734 2017-05-30 Martin Liska <mliska@suse.cz>
6737 * auto-profile.c (get_function_decl_from_block): Fix
6740 2017-05-30 Richard Biener <rguenther@suse.de>
6743 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6745 2017-05-30 Martin Liska <mliska@suse.cz>
6747 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6748 * dumpfile.h (struct dump_file_info): Remove ctors.
6750 2017-05-30 Martin Liska <mliska@suse.cz>
6752 * predict.def: Fix GNU coding style.
6754 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6756 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6757 Mark 'to' argument with ATTRIBUTE_UNUSED.
6759 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6761 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6762 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6763 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6766 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
6768 * doc/install.texi (Options specification): Restore entry of
6769 --enable-sjlj-exceptions.
6771 2017-05-27 Michael Eager <eager@eagercon.com>
6774 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
6776 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
6778 * config/microblaze/microblaze.h
6779 (FIXED_REGISTERS): Update in macro.
6780 (CALL_USED_REGISTERS): Update in macro.
6782 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
6784 * doc/install.texi: Add links to macOS binary distributions.
6786 2017-05-27 Jakub Jelinek <jakub@redhat.com>
6790 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6792 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6794 2017-05-26 Martin Liska <mliska@suse.cz>
6796 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
6798 2017-05-26 Martin Liska <mliska@suse.cz>
6800 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
6801 always leading ';; '.
6802 (dump_bb_info): Likewise.
6803 (brief_dump_cfg): Likewise.
6804 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
6805 * dumpfile.c: Remove usage of TDF_VERBOSE.
6806 * dumpfile.h (enum dump_kind): Likewise.
6807 (dump_gimple_bb_header): Do not use TDF_COMMENT.
6808 * print-tree.c (debug_verbose): Remove.
6809 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
6810 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
6811 * tree-diagnostic.c (default_tree_printer): Replace
6812 TDF_DIAGNOSTIC with TDF_SLIM.
6814 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6816 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
6817 in parameter loop, rather than loop_vinfo.
6818 (create_intersect_range_checks): Ditto.
6819 (vect_create_cond_for_alias_checks): Update call to above functions.
6821 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6823 PR tree-optimization/80815
6824 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
6825 for merging runtime alias checks. Handle negative DR_STEPs.
6827 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6829 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
6831 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
6832 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
6833 out code pruning runtime alias checks.
6834 * tree-data-ref.c (prune_runtime_alias_test_list): New function
6835 factored out from above.
6836 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6838 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6840 (prune_runtime_alias_test_list): New decalaration.
6842 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6844 * tree-vect-data-refs.c (compare_tree): Rename and move ...
6845 * tree-data-ref.c (data_ref_compare_tree): ... to here.
6846 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
6847 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
6848 (operator==, comp_dr_with_seg_len_pair): Ditto.
6849 (vect_prune_runtime_alias_test_list): Ditto.
6851 2017-05-26 Martin Liska <mliska@suse.cz>
6854 * params.def: Bound partial-inlining-entry-probability param.
6856 2017-05-26 Marek Polacek <polacek@redhat.com>
6859 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
6862 2017-05-26 Richard Biener <rguenther@suse.de>
6864 PR tree-optimization/80842
6865 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
6868 2017-05-26 Richard Biener <rguenther@suse.de>
6870 PR tree-optimization/80844
6871 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
6873 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
6875 * doc/md.texi (Machine Constraints): Update x86 family
6876 machine constraints section to match 'config/i386/constraints.md'.
6878 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
6880 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
6882 2017-05-25 Nathan Sidwell <nathan@acm.org>
6884 * doc/invoke.texi (--enable-languages): Update documentation.
6886 2017-05-25 Martin Liska <mliska@suse.cz>
6888 * dumpfile.c: Add TDF_FOLDING.
6889 * dumpfile.h (enum dump_kind): Likewise.
6890 * genmatch.c (dt_simplify::gen_1): Use it.
6892 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6894 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
6896 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6898 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6899 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
6901 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6903 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
6904 * match.pd (X == C): Rewrite it here.
6905 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
6906 with_certain_nonzero_bits2): New predicates.
6907 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
6909 2017-05-24 Nathan Sidwell <nathan@acm.org>
6911 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
6914 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
6917 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6919 * config/powerpcspe: New port. Files are copied from the rs6000
6920 port, with "rs6000" in filenames replaced by "powerpcspe".
6922 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
6924 PR rtl-optimization/80754
6925 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
6927 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
6929 * config/sparc/sparc.md (length): Return the correct value for -mflat
6930 sibcalls to match output_sibcall.
6932 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6936 * config/rs6000/rs6000.c (struct machine_function): Add new field
6938 (rs6000_get_separate_components): Init that field, use it.
6939 (rs6000_components_for_bb): Use the field.
6941 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
6943 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
6945 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
6948 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
6950 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6953 * config/s390/s390.c (s390_check_qrst_address): Check incoming
6954 address against address_operand predicate.
6955 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
6957 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
6959 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
6960 parameters passed indirectly.
6962 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
6964 * config/i386/i386.md (*movdi_internal): Remove SSE4
6965 alternative 18 (?r, *v). Update insn attributes.
6966 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
6967 Update insn attributes.
6968 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
6969 Update insn attributes.
6970 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
6971 alternative 1 (r, v). Remove isa attribute.
6972 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
6973 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
6974 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
6976 2017-05-23 Tom de Vries <tom@codesourcery.com>
6978 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
6981 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6983 * cgraphunit.c (symbol_table::process_new_functions): Update.
6984 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
6985 (inline_generate_summary): Rename to ...
6986 (ipa_fn_summary_generate): ... this one.
6987 (inline_read_summary): Rename to ...
6988 (ipa_fn_summary_read): ... this one.
6989 (inline_write_summary): Rename to ...
6990 (ipa_fn_summary_write): ... this one.
6991 (inline_free_summary): Rename to ...
6992 (ipa_free_fn_summary): ... this one.
6993 (pass_data_local_fn_summary, pass_local_fn_summary,
6994 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
6995 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
6996 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
6997 make_pass_ipa_fn_summary): New.
6998 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
6999 inline_write_summary, inline_free_summary): Remove.
7000 (ipa_free_fn_summary) : New.
7001 * ipa-inline.c (ipa_inline): Update.
7002 (pass_ipa_inline): Do not generate summaries.
7003 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7005 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7006 and add pass_ipa_fn_summary.
7007 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7009 (make_pass_inline_parameters): Remove.
7011 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7013 * omp-low.c (struct omp_context): Remove "default_kind" member.
7016 * omp-offload.c (execute_oacc_device_lower): Remove the
7017 parallelism dimensions function attributes for unparallelized
7018 OpenACC kernels constructs.
7020 2017-05-23 Martin Liska <mliska@suse.cz>
7022 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7024 (cgraph_edge::make_speculative): Likewise.
7025 (cgraph_edge::resolve_speculation): Likewise.
7026 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7027 (cgraph_node::dump): Likewise.
7028 * cgraph.h: Likewise.
7029 * cgraphunit.c (analyze_functions): Likewise.
7030 (symbol_table::compile): Likewise.
7031 * ipa-cp.c (print_all_lattices): Likewise.
7032 (determine_versionability): Likewise.
7033 (initialize_node_lattices): Likewise.
7034 (ipcp_verify_propagated_values): Likewise.
7035 (estimate_local_effects): Likewise.
7036 (update_profiling_info): Likewise.
7037 (create_specialized_node): Likewise.
7038 (perhaps_add_new_callers): Likewise.
7039 (decide_about_value): Likewise.
7040 (decide_whether_version_node): Likewise.
7041 (identify_dead_nodes): Likewise.
7042 (ipcp_store_bits_results): Likewise.
7043 * ipa-devirt.c (dump_targets): Likewise.
7044 (ipa_devirt): Likewise.
7045 * ipa-icf.c (sem_item::dump): Likewise.
7046 (sem_function::equals): Likewise.
7047 (sem_variable::equals): Likewise.
7048 (sem_item_optimizer::read_section): Likewise.
7049 (sem_item_optimizer::execute): Likewise.
7050 (congruence_class::dump): Likewise.
7051 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7052 (dump_inline_summary): Likewise.
7053 (estimate_node_size_and_time): Likewise.
7054 (inline_analyze_function): Likewise.
7055 * ipa-inline-transform.c (inline_call): Likewise.
7056 * ipa-inline.c (report_inline_failed_reason): Likewise.
7057 (want_early_inline_function_p): Likewise.
7058 (edge_badness): Likewise.
7059 (update_edge_key): Likewise.
7060 (inline_small_functions): Likewise.
7061 * ipa-profile.c (ipa_profile): Likewise.
7062 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7063 (ipa_make_edge_direct_to_target): Likewise.
7064 (remove_described_reference): Likewise.
7065 (ipa_impossible_devirt_target): Likewise.
7066 (propagate_controlled_uses): Likewise.
7067 (ipa_print_node_params): Likewise.
7068 (ipcp_transform_function): Likewise.
7069 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7070 (propagate_pure_const): Likewise.
7071 * ipa-reference.c (generate_summary): Likewise.
7072 (read_write_all_from_decl): Likewise.
7073 (propagate): Likewise.
7074 (ipa_reference_read_optimization_summary): Likewise.
7075 * ipa-utils.c (ipa_merge_profiles): Likewise.
7076 * ipa.c (walk_polymorphic_call_targets): Likewise.
7077 (symbol_table::remove_unreachable_nodes): Likewise.
7078 (ipa_single_use): Likewise.
7079 * passes.c (execute_todo): Likewise.
7080 * predict.c (drop_profile): Likewise.
7081 * symtab.c (symtab_node::get_dump_name): New function.
7082 (symtab_node::dump_name): Likewise.
7083 (symtab_node::dump_asm_name): Likewise.
7084 (symtab_node::dump_references): Likewise.
7085 (symtab_node::dump_referring): Likewise.
7086 (symtab_node::dump_base): Likewise.
7087 (symtab_node::debug_symtab): Likewise.
7088 * tree-sra.c (convert_callers_for_node): Likewise.
7089 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7090 * value-prof.c (init_node_map): Likewise.
7092 2017-05-23 Martin Liska <mliska@suse.cz>
7094 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7095 and symtab_node::debug_symtab to symbol_table::debug.
7096 * cgraphunit.c (analyze_functions): Use the renamed function.
7097 (symbol_table::compile): Likewise.
7098 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7099 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7100 * passes.c (execute_todo): Likewise.
7101 * symtab.c (symbol_table::dump): New function.
7102 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7104 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7106 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7107 that nonconst implies exec.
7109 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7111 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7112 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7113 (inline_edge_summary_vec): Turn into ...
7114 (ipa_call_summaries): ... this one.
7115 (redirect_to_unreachable, edge_set_predicate,
7116 evaluate_properties_for_edge, inline_summary_alloc,
7117 reset_ipa_call_summary, reset_inline_summary,
7118 inline_summary_t::duplicate): Update.
7119 (inline_edge_duplication_hook): Turn to ...
7120 (ipa_call_summary_t::duplicate): ... this one.
7121 (inline_edge_removal_hook): Turn to ...
7122 (ipa_call_summary_t::remove): ... this one.
7123 (dump_inline_edge_summary): Turn to ...
7124 (dump_ipa_call_summary): ... this one.
7125 (estimate_function_body_sizes): Update.
7126 (inline_update_callee_summaries): Update.
7127 (remap_edge_change_prob): Update.
7128 (remap_edge_summaries): Update.
7129 (inline_merge_summary): Update.
7130 (do_estimate_edge_time): Update.
7131 (inline_generate_summary): Update.
7132 (inline_read_section): Update.
7133 (inline_read_summary): Update.
7134 (inline_free_summary): Update.
7135 * ipa-inline.c (can_inline_edge_p): Update.
7136 (compute_inlined_call_time): Update.
7137 (want_inline_small_function_p): Update.
7138 (edge_badness): Update.
7139 (early_inliner): Update.
7140 * ipa-inline.h (inline_edge_summary): Turn to ...
7141 (ipa_call_summary): ... this one.
7142 (ipa_call_summary_t): New class.
7143 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7144 (ipa_call_summaries): New.
7145 (inline_edge_summary): Remove.
7146 (estimate_edge_growth): Update.
7147 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7148 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7149 * ipa-split.c (execute_split_functions): Update.
7150 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7152 2017-05-23 Tom de Vries <tom@codesourcery.com>
7154 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7155 attributes): Document rdrand effective target.
7157 2017-05-23 Tom de Vries <tom@codesourcery.com>
7159 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7160 attributes): Sort alphabetically.
7162 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7164 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7166 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7169 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7170 V2DF/V2DI splat into two separate patterns, one that handles
7171 registers, and the other that only handles memory. Drop support
7172 for splatting from a GPR on ISA 2.07 and then splitting the
7173 splat into direct move and splat.
7174 (vsx_splat_<mode>_reg): Likewise.
7175 (vsx_splat_<mode>_mem): Likewise.
7177 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7179 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7181 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7184 * omp-low.c (finish_taskreg_remap): New function.
7185 (finish_taskreg_scan): If unit size of ctx->record_type
7186 is non-constant, unshare the size expression and replace
7187 decls in it with possible outer var refs.
7190 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7191 GOVD_SHARED rather than GOVD_PRIVATE with it.
7192 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7193 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7196 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7197 as last argument to build_outer_var_ref for pointer bases of array
7200 2017-05-19 Martin Sebor <msebor@redhat.com>
7202 * print-tree.c (print_node): Print DECL_READ_P flag.
7204 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7206 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7207 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7208 * cgraph.c: Likewise.
7209 * cgraphunit.c: Likewise.
7210 * gengtype.c: Likewise.
7211 * ipa-cp.c: Likewise.
7212 * ipa-devirt.c: Likewise.
7213 * ipa-icf.c: Likewise.
7214 * ipa-predicate.c: Likewise.
7215 * ipa-profile.c: Likewise.
7216 * ipa-prop.c: Likewise.
7217 * ipa-split.c: Likewise.
7219 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7220 edge_predicate_pool, dump_inline_hints,
7221 inline_summary::account_size_time, redirect_to_unreachable,
7222 edge_set_predicate, set_hint_predicate,
7223 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7224 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7225 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7226 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7227 ipa_call_summary_t::remove, initialize_growth_caches,
7228 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7229 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7230 mark_modified, unmodified_parm_1, unmodified_parm,
7231 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7232 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7233 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7234 will_be_nonconstant_predicate, record_modified_bb_info,
7235 get_minimal_bb, record_modified, param_change_prob,
7236 phi_result_unknown_predicate, predicate_for_phi_result,
7237 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7238 estimate_function_body_sizes, compute_inline_parameters,
7239 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7240 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7241 inline_update_callee_summaries, remap_edge_change_prob,
7242 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7243 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7244 inline_analyze_function, inline_summary_t::insert,
7245 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7246 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7247 inline_free_summary): Move to ipa-fnsummary.h
7248 (predicate_t): Remove.
7249 * ipa-fnsummary.c: New file.
7250 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7251 (enum inline_hints_vals, inline_hints, agg_position_info,
7252 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7253 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7254 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7255 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7256 inline_read_summary, inline_write_summary, inline_free_summary,
7257 inline_analyze_function, initialize_inline_failed,
7258 inline_merge_summary, inline_update_overall_summary,
7259 compute_inline_parameters): Move to ipa-fnsummary.h
7260 * ipa-fnsummary.h: New file.
7261 * ipa-inline-transform.h: Include ipa-inline.h.
7262 * ipa-inline.c: LIkewise.
7264 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7266 * ipa-inline.c (edge_badness): Use inlined_time instead of
7267 inline_summaries->get.
7269 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7271 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7273 2017-05-22 Nathan Sidwell <nathan@acm.org>
7275 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7276 (fdump-lang): Document 'raw' option.
7277 * dumpfile.h (TDI_tu): Delete.
7278 * dumpfile.c (dump_files): Remove translation-unit.
7279 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7281 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7283 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7284 command option from $(AWK) call.
7285 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7286 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7287 [FORMAT]: Remove handling of variable.
7288 * config/avr/t-multilib: Regenerate.
7290 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7292 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7294 (dump_inline_summary): Do not print self_time.
7295 (estimate_function_body_sizes): Do not set self_time.
7296 (compute_inline_parameters): Likewise.
7297 (inline_read_section, inline_write_summary): Do not stream self_time.
7298 * ipa-inline.h (inline_summary): Drop self_time.
7300 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7302 * ipa-inline-analysis.c (account_size_time): Rename to ...
7303 (inline_summary::account_size_time): ... this one.
7304 (reset_ipa_call_summary): Turn to ...
7305 (ipa_call_summary::reset): ... this one.
7306 (reset_inline_summary): Turn to ...
7307 (inline_summary::reset): ... this one.
7308 (inline_summary_t::remove): Update.
7309 (inline_summary_t::duplicate): Update.
7310 (ipa_call_summary_t::remove): Update.
7311 (dump_inline_summary): Update.
7312 (estimate_function_body_sizes): Update.
7313 (compute_inline_parameters): Update.
7314 (estimate_node_size_and_time): Update.
7315 (inline_merge_summary): Update.
7316 (inline_update_overall_summary): Update.
7317 (inline_read_section): Update.
7318 (inline_write_summary): Update.
7319 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7320 add account_size_time and reset member functions.
7321 (ipa_call_summary): Add reset function.
7322 * ipa-predicate.h (predicate::operator &): Constify.
7324 2017-05-22 Richard Biener <rguenther@suse.de>
7326 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7328 2017-05-19 Jason Merrill <jason@redhat.com>
7330 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7332 2017-05-19 Marek Polacek <polacek@redhat.com>
7335 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7336 TYPE_OVERFLOW_WRAPS checks.
7338 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7340 * tree-core.h (enum omp_clause_default_kind): Add
7341 "OMP_CLAUSE_DEFAULT_PRESENT".
7342 * tree-pretty-print.c (dump_omp_clause): Handle it.
7343 * gimplify.c (enum gimplify_omp_var_data): Add
7344 "GOVD_MAP_FORCE_PRESENT".
7345 (gimplify_adjust_omp_clauses_1): Map it to
7346 "GOMP_MAP_FORCE_PRESENT".
7347 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7349 * gimplify.c (oacc_default_clause): Clarify.
7351 2017-05-19 Nathan Sidwell <nathan@acm.org>
7353 LANG_HOOK_REGISTER_DUMPS
7354 * toplev.c (general_init): Call register dump lang hook.
7355 * doc/invoke.texi: Document -fdump-lang option family.
7356 * dumpfile.c (dump_files): Remove class dump here.
7357 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7358 * dumpfile.h (tree_dump_index): Remove TDI_class.
7359 * langhooks-def.h (lhd_register_dumps): Declare.
7360 (LANG_HOOKS_REGISTER_DUMPS): Define.
7361 (LANG_HOOKS_INITIALIZER): Add it.
7362 * langhooks.c (lhd_register_dumps): Define.
7363 * langhooks.h (struct lang_hooks): Add register_dumps.
7365 2017-05-19 Nathan Sidwell <nathan@acm.org>
7367 * context.h (context::set_passes): New.
7368 * context.c (context::context): Do not create pass manager.
7369 * toplev.c (general_init): Create pass manager here.
7371 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7373 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7374 use this splitter if two add or or instructions would also work for
7375 the constant we want to generate.
7377 2017-05-19 Richard Biener <rguenther@suse.de>
7380 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7381 predicate evaluation.
7383 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7385 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7387 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7390 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7392 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7393 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7394 (inline_edge_summary_vec): Turn into ...
7395 (ipa_call_summaries): ... this one.
7396 (redirect_to_unreachable, edge_set_predicate,
7397 evaluate_properties_for_edge, inline_summary_alloc,
7398 reset_ipa_call_summary, reset_inline_summary,
7399 inline_summary_t::duplicate): Update.
7400 (inline_edge_duplication_hook): Turn to ...
7401 (ipa_call_summary_t::duplicate): ... this one.
7402 (inline_edge_removal_hook): Turn to ...
7403 (ipa_call_summary_t::remove): ... this one.
7404 (dump_inline_edge_summary): Turn to ...
7405 (dump_ipa_call_summary): ... this one.
7406 (estimate_function_body_sizes): Update.
7407 (inline_update_callee_summaries): Update.
7408 (remap_edge_change_prob): Update.
7409 (remap_edge_summaries): Update.
7410 (inline_merge_summary): Update.
7411 (do_estimate_edge_time): Update.
7412 (inline_generate_summary): Update.
7413 (inline_read_section): Update.
7414 (inline_read_summary): Update.
7415 (inline_free_summary): Update.
7416 * ipa-inline.c (can_inline_edge_p): Update.
7417 (compute_inlined_call_time): Update.
7418 (want_inline_small_function_p): Update.
7419 (edge_badness): Update.
7420 (early_inliner): Update.
7421 * ipa-inline.h (inline_edge_summary): Turn to ...
7422 (ipa_call_summary): ... this one.
7423 (ipa_call_summary_t): New class.
7424 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7425 (ipa_call_summaries): New.
7426 (inline_edge_summary): Remove.
7427 (estimate_edge_growth): Update.
7428 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7429 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7430 * ipa-split.c (execute_split_functions): Update.
7431 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7433 2017-05-19 Richard Biener <rguenther@suse.de>
7436 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7438 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7440 * config/rs6000/rs6000.c (struct machine_function): Add field
7441 fpr_is_wrapped_separately.
7442 (rs6000_get_separate_components): Use 64 components. Handle the
7444 (rs6000_components_for_bb): Handle the FPR components.
7445 (rs6000_emit_prologue_components): Handle the FPR components.
7446 (rs6000_emit_epilogue_components): Handle the FPR components.
7447 (rs6000_set_handled_components): Handle the FPR components.
7448 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7449 that are already separately shrink-wrapped.
7450 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7451 that are already separately shrink-wrapped.
7453 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7456 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7459 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7460 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7461 cases where the register allocator uses a move and an offsettable
7462 memory operation to/from a FPR register on ISA 2.06/2.07.
7463 (define_peephole2 for Altivec d-form store): Likewise.
7465 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7468 * config/i386/mmx.md (*mov<mode>_internal): Enable
7469 alternatives 11, 12, 13 and 14 also for 32bit targets.
7470 Remove alternatives 15, 16, 17 and 18.
7471 * config/i386/sse.md (vec_concatv2di): Change
7472 alternative (!x, *y) to (x, ?!*Yn).
7474 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7476 * dumpfile.h (enum dump_kind): Remove stray comma.
7478 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7480 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7481 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7482 predicate::num_conditions
7483 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7484 (CHANGED): turn into predicate::changed.
7485 (agg_position_info): Move to ipa-predicate.h
7486 (add_condition, predicate::add_clause, predicate::operator &=,
7487 predicate::or_with, predicate::evaluate, predicate::probability,
7488 dump_condition, dump_clause, predicate::dump,
7489 predicate::remap_after_duplication, predicate::remap_after_inlining,
7490 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7491 (evaluate_conditions_for_known_args): Update.
7492 (set_cond_stmt_execution_predicate): Update.
7493 * ipa-inline.h: Include ipa-predicate.h
7494 (condition, inline_param_summary, conditions, agg_position_info,
7495 predicate): Move to ipa-predicate.h
7496 * ipa-predicate.c: New file.
7497 * ipa-predicate.h: New file.
7499 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7501 * final.c (leaf_function_p): Check we are not in a sequence.
7503 2017-05-18 Martin Liska <mliska@suse.cz>
7505 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7506 * dumpfile.c (dump_register): Use new enum dump_kind.
7507 (get_dump_file_name): Likewise.
7508 (dump_enable_all): Likewise.
7509 (dump_switch_p_1): Likewise.
7510 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7511 * dumpfile.h (enum dump_kind): New enum type.
7512 (struct dump_file_info): Create constructor and
7513 format fields and comments.
7514 * passes.c (pass_manager::register_one_dump_file):
7516 * statistics.c (statistics_early_init): Likewise.
7517 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7518 TDF_TREE with TDF_SLIM.
7519 (gather_memory_references_ref): Likewise.
7521 2017-05-18 Martin Liska <mliska@suse.cz>
7523 * vec.h (struct vnull): Use it.
7525 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7527 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7528 (true_predicate, false_predicate, true_predicate_p,
7529 false_predicate_p): Remove.
7530 (single_cond_predicate, not_inlined_predicate): Turn to member function
7532 (add_condition): Update.
7533 (add_clause): Turn to...
7534 (predicate::add_clause): ... this one; update; allow passing NULL
7536 (and_predicates): Turn to ...
7537 (predicate::operator &=): ... this one.
7538 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7539 (or_predicates): Turn to ...
7540 (predicate::or_with): ... this one.
7541 (evaluate_predicate): Turn to ...
7542 (predicate::evaluate): ... this one.
7543 (predicate_probability): Turn to ...
7544 (predicate::probability): ... this one.
7545 (dump_condition): Update.
7546 (dump_predicate): Turn to ...
7547 (predicate::dump): ... this one.
7548 (account_size_time): Update.
7549 (edge_set_predicate): Update.
7550 (set_hint_predicate): UPdate.
7551 (evaluate_conditions_for_known_args): Update.
7552 (evaluate_properties_for_edge): Update.
7553 (remap_predicate_after_duplication): Turn to...
7554 (predicate::remap_after_duplication): ... this one.
7555 (remap_hint_predicate_after_duplication): Update.
7556 (inline_summary_t::duplicate): UPdate.
7557 (dump_inline_edge_summary): Update.
7558 (dump_inline_summary): Update.
7559 (set_cond_stmt_execution_predicate): Update.
7560 (set_switch_stmt_execution_predicate): Update.
7561 (compute_bb_predicates): Update.
7562 (will_be_nonconstant_expr_predicate): Update.
7563 (will_be_nonconstant_predicate): Update.
7564 (phi_result_unknown_predicate): Update.
7565 (predicate_for_phi_result): Update.
7566 (array_index_predicate): Update.
7567 (estimate_function_body_sizes): Update.
7568 (estimate_node_size_and_time): Update.
7569 (estimate_ipcp_clone_size_and_time): Update.
7570 (remap_predicate): Rename to ...
7571 (predicate::remap_after_inlining): ... this one.
7572 (remap_hint_predicate): Update.
7573 (inline_merge_summary): Update.
7574 (inline_update_overall_summary): Update.
7575 (estimate_size_after_inlining): Update.
7576 (read_predicate): Rename to ...
7577 (predicate::stream_in): ... this one.
7578 (read_inline_edge_summary): Update.
7579 (write_predicate): Rename to ...
7580 (predicate::stream_out): ... this one.
7581 (write_inline_edge_summary): Update.
7582 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7583 (clause_t): Turn to uint32_t
7584 (predicate): Turn to class; implement constructor and operators
7586 (size_time_entry): Update.
7587 (inline_summary): Update.
7588 (inline_edge_summary): Update.
7590 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7592 * fold-const.c (fold_binary_loc): Move transformation...
7593 * match.pd (C - X CMP X): ... here.
7595 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7597 * config/sparc/sparc.c (sparc_option_override): Set function
7598 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7599 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7601 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7603 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7605 2017-05-18 Marek Polacek <polacek@redhat.com>
7608 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7609 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7610 gimple_assign_load_p.
7612 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7615 * real.c (do_compare): Give decimal_do_compare preference over
7616 comparing just the signs.
7618 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7620 * doc/md.texi (Canonicalization of Instructions): Describe the
7621 canonical form of instructions that inherently set a condition
7624 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7627 * tree-cfg.c: Move deletion of unreachable case statements to after
7628 the merging of consecutive case labels.
7630 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7632 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7633 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7634 restoring of callee-saved registers.
7636 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7638 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7639 * config/visium/visium.c (single_set_and_flags): Likewise.
7640 * config/visium/visium.md (Substitutions): Likewise.
7642 2017-05-17 Martin Liska <mliska@suse.cz>
7644 * cfg.c: Introduce dump_flags_t type and
7645 use it instead of int type.
7647 * cfghooks.c: Likewise.
7648 * cfghooks.h (struct cfg_hooks): Likewise.
7649 * cfgrtl.c: Likewise.
7650 * cfgrtl.h: Likewise.
7651 * cgraph.c (cgraph_node::get_body): Likewise.
7652 * coretypes.h: Likewise.
7653 * domwalk.c: Likewise.
7654 * domwalk.h: Likewise.
7655 * dumpfile.c (struct dump_option_value_info): Likewise.
7656 (dump_enable_all): Likewise.
7657 (dump_switch_p_1): Likewise.
7658 (opt_info_switch_p): Likewise.
7659 * dumpfile.h (enum tree_dump_index): Likewise.
7660 (struct dump_file_info): Likewise.
7661 * genemit.c: Likewise.
7662 * generic-match-head.c: Likewise.
7663 * gengtype.c (open_base_files): Likewise.
7664 * gimple-pretty-print.c: Likewise.
7665 * gimple-pretty-print.h: Likewise.
7666 * graph.c (print_graph_cfg): Likewise.
7667 * graphite-scop-detection.c (dot_all_sese): Likewise.
7668 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7669 * loop-unroll.c (report_unroll): Likewise.
7670 * passes.c (pass_manager::register_one_dump_file): Likewise.
7671 * print-tree.c: Likewise.
7672 * statistics.c: Likewise.
7673 * tree-cfg.c: Likewise.
7674 * tree-cfg.h: Likewise.
7675 * tree-dfa.c: Likewise.
7676 * tree-dfa.h: Likewise.
7677 * tree-dump.c (dump_function): Likewise.
7678 * tree-dump.h (struct dump_info): Likewise.
7679 * tree-pretty-print.c: Likewise.
7680 * tree-pretty-print.h: Likewise.
7681 * tree-ssa-live.c: Likewise.
7682 * tree-ssa-live.h: Likewise.
7683 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7684 * tree-vect-loop.c: Likewise.
7685 * tree-vect-slp.c: Likewise.
7687 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7688 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7690 PR tree-optimization/80457
7691 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7692 of all arguments to a statement as scalar_to_vec operations.
7693 (vectorizable_call): Adjust call to vect_model_simple_cost for
7695 (vectorizable_conversion): Likewise.
7696 (vectorizable_assignment): Likewise.
7697 (vectorizable_shift): Likewise.
7698 (vectorizable_operation): Likewise.
7699 (vectorizable_comparison): Likewise.
7700 (vect_is_simple_cond): Record the def types for operands.
7701 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7702 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7703 for statement argument count.
7705 2017-05-16 Carl Love <cel@us.ibm.com>
7707 * config/rs6000/rs6000-c: Add support for built-in functions
7708 vector unsigned long long vec_bperm (vector unsigned long long,
7709 vector unsigned char)
7710 vector signed long long vec_mule (vector signed int,
7712 vector unsigned long long vec_mule (vector unsigned int,
7713 vector unsigned int)
7714 vector signed long long vec_mulo (vector signed int,
7716 vector unsigned long long vec_mulo (vector unsigned int,
7717 vector unsigned int)
7718 vector signed char vec_sldw (vector signed char,
7721 vector unsigned char vec_sldw (vector unsigned char,
7722 vector unsigned char,
7724 vector signed short vec_sldw (vector signed short,
7725 vector signed short,
7727 vector unsigned short vec_sldw (vector unsigned short,
7728 vector unsigned short,
7730 vector signed int vec_sldw (vector signed int,
7733 vector unsigned int vec_sldw (vector unsigned int,
7734 vector unsigned int,
7736 vector signed long long vec_sldw (vector signed long long,
7737 vector signed long long,
7739 vector unsigned long long vec_sldw (vector unsigned long long,
7740 vector unsigned long long,
7742 * config/rs6000/rs6000-c: Add support for built-in functions
7743 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7744 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7745 * doc/extend.texi: Update the built-in documentation for the
7746 new built-in functions.
7748 2017-05-16 Marek Polacek <polacek@redhat.com>
7752 * tree.c (save_expr): Don't fold the expression.
7754 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7756 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7757 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7758 and (?*y,m). Update insn attributes.
7760 2017-05-16 Martin Liska <mliska@suse.cz>
7762 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7763 flags argument of print_gimple_stmt, print_gimple_expr,
7764 print_generic_stmt and print_generic_expr.
7765 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7766 * coretypes.h: Likewise.
7767 * except.c (dump_eh_tree): Likewise.
7768 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7769 * gimple-pretty-print.h: Likewise.
7770 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
7771 (backprop::push_to_worklist): Likewise.
7772 (backprop::pop_from_worklist): Likewise.
7773 (backprop::process_use): Likewise.
7774 (backprop::intersect_uses): Likewise.
7775 (note_replacement): Likewise.
7776 * gimple-ssa-store-merging.c
7777 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
7778 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7779 (pass_store_merging::execute): Likewise.
7780 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
7781 (ssa_base_cand_dump_callback): Likewise.
7782 (dump_incr_vec): Likewise.
7783 (replace_refs): Likewise.
7784 (replace_mult_candidate): Likewise.
7785 (create_add_on_incoming_edge): Likewise.
7786 (create_phi_basis): Likewise.
7787 (insert_initializers): Likewise.
7788 (all_phi_incrs_profitable): Likewise.
7789 (introduce_cast_before_cand): Likewise.
7790 (replace_one_candidate): Likewise.
7791 * gimplify.c (gimplify_expr): Likewise.
7792 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
7793 (set_rename): Likewise.
7794 (rename_uses): Likewise.
7795 (copy_loop_phi_nodes): Likewise.
7796 (add_close_phis_to_merge_points): Likewise.
7797 (copy_loop_close_phi_args): Likewise.
7798 (copy_cond_phi_args): Likewise.
7799 (graphite_copy_stmts_from_block): Likewise.
7800 (translate_pending_phi_nodes): Likewise.
7801 * graphite-poly.c (print_pdr): Likewise.
7802 (dump_gbb_cases): Likewise.
7803 (dump_gbb_conditions): Likewise.
7804 (print_scop_params): Likewise.
7805 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
7806 (build_cross_bb_scalars_use): Likewise.
7807 (gather_bbs::before_dom_children): Likewise.
7808 * hsa-dump.c (dump_hsa_immed): Likewise.
7809 * ipa-cp.c (print_ipcp_constant_value): Likewise.
7810 (get_replacement_map): Likewise.
7811 * ipa-inline-analysis.c (dump_condition): Likewise.
7812 (estimate_function_body_sizes): Likewise.
7813 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
7814 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
7815 * ipa-prop.c (ipa_dump_param): Likewise.
7816 (ipa_print_node_jump_functions_for_edge): Likewise.
7817 (ipa_modify_call_arguments): Likewise.
7818 (ipa_modify_expr): Likewise.
7819 (ipa_dump_param_adjustments): Likewise.
7820 (ipa_dump_agg_replacement_values): Likewise.
7821 (ipcp_modif_dom_walker::before_dom_children): Likewise.
7822 * ipa-pure-const.c (check_stmt): Likewise.
7823 (pass_nothrow::execute): Likewise.
7824 * ipa-split.c (execute_split_functions): Likewise.
7825 * omp-offload.c (dump_oacc_loop_part): Likewise.
7826 (dump_oacc_loop): Likewise.
7827 * trans-mem.c (tm_log_emit): Likewise.
7828 (tm_memopt_accumulate_memops): Likewise.
7829 (dump_tm_memopt_set): Likewise.
7830 (dump_tm_memopt_transform): Likewise.
7831 * tree-cfg.c (gimple_verify_flow_info): Likewise.
7832 (print_loop): Likewise.
7833 * tree-chkp-opt.c (chkp_print_addr): Likewise.
7834 (chkp_gather_checks_info): Likewise.
7835 (chkp_get_check_result): Likewise.
7836 (chkp_remove_check_if_pass): Likewise.
7837 (chkp_use_outer_bounds_if_possible): Likewise.
7838 (chkp_reduce_bounds_lifetime): Likewise.
7839 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
7840 (chkp_mark_completed_bounds): Likewise.
7841 (chkp_register_incomplete_bounds): Likewise.
7842 (chkp_mark_invalid_bounds): Likewise.
7843 (chkp_maybe_copy_and_register_bounds): Likewise.
7844 (chkp_build_returned_bound): Likewise.
7845 (chkp_get_bound_for_parm): Likewise.
7846 (chkp_build_bndldx): Likewise.
7847 (chkp_get_bounds_by_definition): Likewise.
7848 (chkp_generate_extern_var_bounds): Likewise.
7849 (chkp_get_bounds_for_decl_addr): Likewise.
7850 * tree-chrec.c (chrec_apply): Likewise.
7851 * tree-data-ref.c (dump_data_reference): Likewise.
7852 (dump_subscript): Likewise.
7853 (dump_data_dependence_relation): Likewise.
7854 (analyze_overlapping_iterations): Likewise.
7855 * tree-inline.c (expand_call_inline): Likewise.
7856 (tree_function_versioning): Likewise.
7857 * tree-into-ssa.c (dump_defs_stack): Likewise.
7858 (dump_currdefs): Likewise.
7859 (dump_names_replaced_by): Likewise.
7860 (dump_update_ssa): Likewise.
7861 (update_ssa): Likewise.
7862 * tree-object-size.c (pass_object_sizes::execute): Likewise.
7863 * tree-parloops.c (build_new_reduction): Likewise.
7864 (try_create_reduction_list): Likewise.
7865 (ref_conflicts_with_region): Likewise.
7866 (oacc_entry_exit_ok_1): Likewise.
7867 (oacc_entry_exit_single_gang): Likewise.
7868 * tree-pretty-print.h: Likewise.
7869 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
7870 (get_scalar_evolution): Likewise.
7871 (add_to_evolution): Likewise.
7872 (get_loop_exit_condition): Likewise.
7873 (analyze_evolution_in_loop): Likewise.
7874 (analyze_initial_condition): Likewise.
7875 (analyze_scalar_evolution): Likewise.
7876 (instantiate_scev): Likewise.
7877 (number_of_latch_executions): Likewise.
7878 (gather_chrec_stats): Likewise.
7879 (final_value_replacement_loop): Likewise.
7880 (scev_const_prop): Likewise.
7881 * tree-sra.c (dump_access): Likewise.
7882 (disqualify_candidate): Likewise.
7883 (create_access): Likewise.
7885 (maybe_add_sra_candidate): Likewise.
7886 (create_access_replacement): Likewise.
7887 (analyze_access_subtree): Likewise.
7888 (analyze_all_variable_accesses): Likewise.
7889 (sra_modify_assign): Likewise.
7890 (initialize_constant_pool_replacements): Likewise.
7891 (find_param_candidates): Likewise.
7892 (decide_one_param_reduction): Likewise.
7893 (replace_removed_params_ssa_names): Likewise.
7894 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
7895 * tree-ssa-copy.c (dump_copy_of): Likewise.
7896 (copy_prop_visit_cond_stmt): Likewise.
7897 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
7898 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
7899 (record_equivalences_from_stmt): Likewise.
7900 * tree-ssa-dse.c (compute_trims): Likewise.
7901 (delete_dead_call): Likewise.
7902 (delete_dead_assignment): Likewise.
7903 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
7904 (forward_propagate_into_cond): Likewise.
7905 (pass_forwprop::execute): Likewise.
7906 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7907 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
7909 (move_computations_worker): Likewise.
7910 (execute_sm): Likewise.
7911 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
7912 (remove_exits_and_undefined_stmts): Likewise.
7913 (remove_redundant_iv_tests): Likewise.
7914 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
7915 (adjust_iv_update_pos): Likewise.
7916 * tree-ssa-math-opts.c (bswap_replace): Likewise.
7917 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
7918 (value_replacement): Likewise.
7919 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
7920 * tree-ssa-pre.c (print_pre_expr): Likewise.
7921 (get_representative_for): Likewise.
7922 (create_expression_by_pieces): Likewise.
7923 (insert_into_preds_of_block): Likewise.
7924 (eliminate_insert): Likewise.
7925 (eliminate_dom_walker::before_dom_children): Likewise.
7926 (eliminate): Likewise.
7927 (remove_dead_inserted_code): Likewise.
7928 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
7929 * tree-ssa-reassoc.c (get_rank): Likewise.
7930 (eliminate_duplicate_pair): Likewise.
7931 (eliminate_plus_minus_pair): Likewise.
7932 (eliminate_not_pairs): Likewise.
7933 (undistribute_ops_list): Likewise.
7934 (eliminate_redundant_comparison): Likewise.
7935 (update_range_test): Likewise.
7936 (optimize_range_tests_var_bound): Likewise.
7937 (optimize_vec_cond_expr): Likewise.
7938 (rewrite_expr_tree): Likewise.
7939 (rewrite_expr_tree_parallel): Likewise.
7940 (linearize_expr): Likewise.
7941 (break_up_subtract): Likewise.
7942 (linearize_expr_tree): Likewise.
7943 (attempt_builtin_powi): Likewise.
7944 (attempt_builtin_copysign): Likewise.
7945 (transform_stmt_to_copy): Likewise.
7946 (transform_stmt_to_multiply): Likewise.
7947 (dump_ops_vector): Likewise.
7948 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
7949 (print_scc): Likewise.
7950 (set_ssa_val_to): Likewise.
7951 (visit_reference_op_store): Likewise.
7952 (visit_use): Likewise.
7953 (sccvn_dom_walker::before_dom_children): Likewise.
7954 (run_scc_vn): Likewise.
7955 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7957 (expr_hash_elt::print): Likewise.
7958 (const_and_copies::pop_to_marker): Likewise.
7959 (const_and_copies::record_const_or_copy_raw): Likewise.
7960 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
7961 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
7962 (dump_predicates): Likewise.
7963 (find_uninit_use): Likewise.
7964 (warn_uninitialized_phi): Likewise.
7965 (pass_late_warn_uninitialized::execute): Likewise.
7966 * tree-ssa.c (verify_vssa): Likewise.
7967 (verify_ssa): Likewise.
7968 (maybe_optimize_var): Likewise.
7969 * tree-vrp.c (dump_value_range): Likewise.
7970 (dump_all_value_ranges): Likewise.
7971 (dump_asserts_for): Likewise.
7972 (register_edge_assert_for_2): Likewise.
7973 (vrp_visit_cond_stmt): Likewise.
7974 (vrp_visit_switch_stmt): Likewise.
7975 (vrp_visit_stmt): Likewise.
7976 (vrp_visit_phi_node): Likewise.
7977 (simplify_cond_using_ranges_1): Likewise.
7978 (fold_predicate_in): Likewise.
7979 (evrp_dom_walker::before_dom_children): Likewise.
7980 (evrp_dom_walker::push_value_range): Likewise.
7981 (evrp_dom_walker::pop_value_range): Likewise.
7982 (execute_early_vrp): Likewise.
7984 2017-05-16 Richard Biener <rguenther@suse.de>
7986 * dwarf2out.c (loc_list_from_tree_1): Do not create
7987 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
7989 2017-05-16 Richard Biener <rguenther@suse.de>
7991 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
7993 (note_variable_value_in_expr): If we resolved the decl ref
7994 do not push to the stack.
7996 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
7998 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
7999 operations in fast-math mode.
8000 (vaddq_f16): Likewise.
8001 (vmul_f16): Likewise.
8002 (vmulq_f16): Likewise.
8003 (vsub_f16): Likewise.
8004 (vsubq_f16): Likewise.
8005 * config/arm/neon.md (add<mode>3): New.
8007 (fma:<VH:mode>3): New. Also remove outdated comment.
8010 2017-05-16 Martin Liska <mliska@suse.cz>
8014 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8015 (odr_types_equivalent_p): Likewise.
8017 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8019 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8021 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8024 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8025 non-interunit SSE move alternatives with '?'.
8026 (zero-extendsidi peephole2): New peephole to skip intermediate
8027 general register in SSE zero-extend sequence.
8029 2017-05-15 Jeff Law <law@redhat.com>
8031 * reorg.c (relax_delay_slots): Create a new variable to hold
8032 the temporary target rather than clobbering TARGET_LABEL.
8034 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8035 missing argument to extract_bit_field call.
8036 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8038 2017-05-15 Martin Liska <mliska@suse.cz>
8041 * gcc.c (process_command): Do not allow empty argument of -o option.
8043 2017-05-15 Renlin Li <renlin.li@arm.com>
8045 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8046 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8047 * config/aarch64/constraints.md (Usf): Add long call check.
8048 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8049 (call_value): Likewise.
8050 (sibcall): Likewise.
8051 (sibcall_value): Likewise.
8053 (call_value_insn): New.
8054 (sibcall_insn): Update rtx pattern.
8055 (sibcall_value_insn): Likewise.
8056 (call_internal): Remove.
8057 (call_value_internal): Likewise.
8058 (sibcall_internal): Likewise.
8059 (sibcall_value_internal): Likewise.
8060 (call_reg): Likewise.
8061 (call_symbol): Likewise.
8062 (call_value_reg): Likewise.
8063 (call_value_symbol): Likewise.
8065 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8068 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8070 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8072 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8073 compatible with CCGOCmode and with CCZmode.
8075 2017-05-14 Martin Sebor <msebor@redhat.com>
8078 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8079 (gimple_fold_builtin_snprintf): Same.
8080 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8081 (gimple_fold_builtin_snprintf): Same.
8082 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8084 (is_call_safe): New function.
8085 (try_substitute_return_value): Call it.
8086 (try_simplify_call): New function.
8087 (pass_sprintf_length::handle_gimple_call): Call it.
8089 2017-05-14 Martin Sebor <msebor@redhat.com>
8092 * builtins.c (expand_builtin_stpncpy): Simplify.
8094 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8096 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8097 * config/i386/i386.h
8098 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8099 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8100 (struct machine_function): Add new members call_ms2sysv,
8101 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8102 (struct machine_frame_state): New fields sp_realigned and
8103 sp_realigned_offset.
8104 * config/i386/i386.c
8105 (enum xlogue_stub): New enum.
8106 (enum xlogue_stub_sets): New enum.
8107 (class xlogue_layout): New class.
8108 (struct ix86_frame): New fields stack_realign_allocate_offset,
8109 stack_realign_offset and outlined_save_offset. Modify comments to
8110 detail stack layout when using out-of-line stubs.
8111 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8112 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8113 -mcall-ms2sysv-xlogues.
8114 (stub_managed_regs): New static variable.
8115 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8116 registers managed by out-of-line stub.
8117 (disable_call_ms2sysv_xlogues): New function.
8118 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8119 m->call_ms2sysv when appropriate and compute frame layout for
8121 (sp_valid_at, fp_valid_at): New inline functions.
8122 (choose_basereg): New function.
8123 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8125 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8126 Use align parameter of choose_baseaddr to generated aligned SSE movs
8128 (pro_epilogue_adjust_stack): Modify to track
8129 machine_frame_state::sp_realigned.
8130 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8131 (ix86_nsaved_sseregs): Likewise.
8132 (ix86_emit_save_regs): Likewise.
8133 (ix86_emit_save_regs_using_mov): Likewise.
8134 (ix86_emit_save_sse_regs_using_mov): Likewise.
8135 (get_scratch_register_on_entry): Likewise.
8136 (gen_frame_set): New function.
8137 (gen_frame_load): Likewise.
8138 (gen_frame_store): Likewise.
8139 (emit_outlined_ms2sysv_save): Likewise.
8140 (emit_outlined_ms2sysv_restore): Likewise.
8141 (ix86_expand_prologue): Modify stack re-alignment code and call
8142 emit_outlined_ms2sysv_save when appropriate.
8143 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8144 parameter rtx_insn *insn, which allows the function to be used to only
8146 (ix86_expand_epilogue): Modify validity checks of frame and stack
8147 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8148 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8149 * config/i386/predicates.md
8150 (save_multiple): New predicate.
8151 (restore_multiple): Likewise.
8152 * config/i386/sse.md
8153 (save_multiple<mode>): New pattern.
8154 (save_multiple_realign<mode>): Likewise.
8155 (restore_multiple<mode>): Likewise.
8156 (restore_multiple_and_return<mode>): Likewise.
8157 (restore_multiple_leave_return<mode>): Likewise.
8158 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8160 2017-05-14 Julia Koval <julia.koval@intel.com>
8162 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8163 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8164 (__builtin_ia32_xsetbv): New builtins.
8165 * config/i386/i386.c (ix86_expand_special_args_builtin):
8167 (ix86_expand_builtin): Special expand for new intrinsics.
8168 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8169 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8170 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8172 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8174 * cfganal.c (inverted_post_order_compute): Change argument type
8176 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8177 * df-core.c (rest_of_handle_df_initialize): Adjust.
8178 (rest_of_handle_df_finish): Likewise.
8179 (df_analyze_1): Likewise.
8180 (df_analyze): Likewise.
8181 (loop_inverted_post_order_compute): Change argument to be a vec *.
8182 (df_analyze_loop): Adjust.
8183 (df_get_n_blocks): Likewise.
8184 (df_get_postorder): Likewise.
8185 * df.h (struct df_d): Change field to be a vec.
8186 * lcm.c (compute_laterin): Adjust.
8187 (compute_available): Likewise.
8188 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8189 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8190 * tree-ssa-pre.c (compute_antic): Likewise.
8192 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8194 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8195 (depth_first_search::depth_first_search): Change structure init
8196 function to this constructor.
8197 (depth_first_search::add_bb): Rename function to this member.
8198 (depth_first_search::execute): Likewise.
8199 (flow_dfs_compute_reverse_finish): Adjust.
8201 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8203 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8204 (longest_simple_path): Likewise.
8205 * shrink-wrap.c (spread_components): Likewise.
8206 (disqualify_problematic_components): Likewise.
8207 (emit_common_heads_for_components): Likewise.
8208 (emit_common_tails_for_components): Likewise.
8209 (insert_prologue_epilogue_for_components): Likewise.
8211 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8213 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8216 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8218 * df-core.c (df_set_blocks): Start using auto_bitmap.
8219 (df_compact_blocks): Likewise.
8220 * df-problems.c (df_rd_confluence_n): Likewise.
8221 * df-scan.c (df_insn_rescan_all): Likewise.
8222 (df_process_deferred_rescans): Likewise.
8223 (df_update_entry_block_defs): Likewise.
8224 (df_update_exit_block_uses): Likewise.
8225 (df_entry_block_bitmap_verify): Likewise.
8226 (df_exit_block_bitmap_verify): Likewise.
8227 (df_scan_verify): Likewise.
8228 * lra-constraints.c (lra_constraints): Likewise.
8229 (undo_optional_reloads): Likewise.
8230 (lra_undo_inheritance): Likewise.
8231 * lra-remat.c (calculate_gen_cands): Likewise.
8232 (do_remat): Likewise.
8233 * lra-spills.c (assign_spill_hard_regs): Likewise.
8234 (spill_pseudos): Likewise.
8235 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8236 (bitmap_set_subtract_values): Likewise.
8238 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8240 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8241 management with auto_bitmap.
8242 (fix_inter_tick): Likewise.
8243 (fix_recovery_deps): Likewise.
8244 * ira.c (add_store_equivs): Likewise.
8245 (find_moveable_pseudos): Likewise.
8246 (split_live_ranges_for_shrink_wrap): Likewise.
8247 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8248 (rtx_reuse_manager::seen_def_p): Likewise.
8249 (rtx_reuse_manager::set_seen_def): Likewise.
8250 * print-rtl.h (class rtx_reuse_manager): Likewise.
8252 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8254 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8256 (migrate_btr_def): Likewise.
8257 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8258 * df-core.c (loop_post_order_compute): Likewise.
8259 (loop_inverted_post_order_compute): Likewise.
8260 * hsa-common.h: Likewise.
8261 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8262 * init-regs.c (initialize_uninitialized_regs): Likewise.
8263 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8264 (inline_small_functions): Likewise.
8265 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8266 * ira.c (combine_and_move_insns): Likewise.
8267 (build_insn_chain): Likewise.
8268 * loop-invariant.c (find_invariants): Likewise.
8269 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8270 * predict.c (tree_predict_by_opcode): Likewise.
8271 (predict_paths_leading_to): Likewise.
8272 (predict_paths_leading_to_edge): Likewise.
8273 (estimate_loops_at_level): Likewise.
8274 (estimate_loops): Likewise.
8275 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8276 (spread_components): Likewise.
8277 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8278 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8279 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8280 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8281 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8282 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8283 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8284 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8285 (mark_threaded_blocks): Likewise.
8286 (thread_through_all_blocks): Likewise.
8287 * tree-ssa.c (verify_ssa): Likewise.
8288 (execute_update_addresses_taken): Likewise.
8289 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8291 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8293 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8295 (post_order_compute): Likewise.
8296 (inverted_post_order_compute): Likewise.
8297 (pre_and_rev_post_order_compute_fn): Likewise.
8299 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8301 * genrecog.c (int_set::int_set): Explicitly construct our
8302 auto_vec base class.
8303 * vec.h (auto_vec::auto_vec): New constructor.
8305 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8307 * bitmap.h (class auto_bitmap): New constructor taking
8308 bitmap_obstack * argument.
8310 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8312 * bitmap.h (class auto_bitmap): Change type of m_bits to
8313 bitmap_head, and adjust ctor / dtor and member operators.
8315 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8317 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8318 when returned register mode doesn't match original mode.
8320 2017-05-12 Jeff Law <law@redhat.com>
8321 Jakub Jelinek <jakub@redhat.com>
8323 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8324 we look for cc setter after the compare-elim changes.
8325 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8326 within the vector to match what compare-elim now expects.
8327 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8328 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8330 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8331 after the compare-elim changes.
8332 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8333 the vector to match what compare-elim now expects.
8334 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8335 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8336 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8337 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8338 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8340 * config/visium/visium.c (single_set_and_flags): Fix where
8341 we look for cc setter after the compare-elim changes.
8342 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8343 with the vector to match what compare-elim now expects.
8344 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8345 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8346 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8347 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8348 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8349 (neg<mode>2_insn_set_overflow): Likewise.
8351 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8354 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8355 maybe_expand_insn call, set ops[0].target. If still set after call,
8356 set alt_rtl. Add extra arg to recursive calls.
8357 (extract_bit_field): Add alt_rtl argument. Pass to
8359 * expmed.h (extract_bit_field): Fix prototype.
8360 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8361 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8362 to extract_bit_field_calls.
8363 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8364 Pass alt_rtl to extract_bit_field calls.
8365 * calls.c (store_unaligned_arguments_into_psuedos)
8366 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8367 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8369 * optabs.h (struct expand_operand): Add target bitfield.
8371 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8373 * compare-elim.c (try_eliminate_compare): Canonicalize
8374 operation with embedded compare to
8375 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8376 (set (reg) (operation)].
8378 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8380 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8383 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8384 cost of adding a carry flag for ADC instruction.
8385 [case MINUS]: Ignore the cost of subtracting a carry flag
8386 for SBB instruction.
8388 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8390 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8392 * config/rs6000/bmiintrin.h: New file.
8393 * config/rs6000/bmi2intrin.h: New file.
8394 * config/rs6000/x86intrin.h: New file.
8396 2017-05-12 Jeff Law <law@redhat.com>
8398 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8401 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8404 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8406 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8409 early expansion of vector divide builtins.
8410 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8411 builtins identified as having unsigned arguments.
8413 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8415 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8416 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8417 expansion of vector logical operations (and, andc, or, xor,
8420 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8422 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8423 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8425 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8427 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8428 early GIMPLE expansion of vector multiplies.
8430 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8432 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8433 TARGET_HAVE_MOVT conditional.
8434 (movt splitter): Likewise.
8436 2017-05-12 Richard Biener <rguenther@suse.de>
8438 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8439 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8440 Fold all stmts not inplace.
8442 2017-05-12 Richard Biener <rguenther@suse.de>
8444 PR tree-optimization/80713
8445 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8446 inserted_exprs bit for not removed stmts.
8448 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8451 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8452 parallelized" attribute for parallelized OpenACC kernels.
8453 * omp-offload.c (execute_oacc_device_lower): Use it.
8455 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8456 Set "oacc kernels" attribute.
8457 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8458 parameter. Adjust all users.
8459 (oacc_fn_attrib_kernels_p): Remove function.
8460 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8461 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8462 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8463 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8464 assert "oacc kernels" attribute is set.
8466 2017-05-11 Carl Love <cel@us.ibm.com>
8468 * config/rs6000/rs6000-c: Add support for built-in functions
8469 vector unsigned char vec_popcnt (vector signed char)
8470 vector unsigned char vec_popcnt (vector unsigned char)
8471 vector unsigned short vec_popcnt (vector signed short)
8472 vector unsigned short vec_popcnt (vector unsigned short)
8473 vector unsigned int vec_popcnt (vector signed int)
8474 vector unsigned int vec_popcnt (vector unsigned int)
8475 vector unsigned long long vec_popcnt (vector signed long long)
8476 vector unsigned long long vec_popcnt (vector unsigned long long)
8477 vector signed long long vec_slo (vector signed long long,
8479 vector signed long long vec_slo (vector signed long long,
8480 vector unsigned char)
8481 vector unsigned long long vec_slo (vector unsigned long long,
8483 vector unsigned long long vec_slo (vector unsigned long long,
8484 vector unsigned char)
8485 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8486 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8487 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8488 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8489 * doc/extend.texi: Update the built-in documentation file for the
8490 new built-in functions.
8492 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8494 * attribs.h (sorted_attr_string): Move machine independent
8495 functions for target clone support from the i386 port to common
8496 code. Rename ix86_function_versions to common_function_versions.
8497 Rename make_name to make_unique_name.
8498 (common_function_versions): Likewise.
8499 (make_unique_name): Likewise.
8500 (make_dispatcher_decl): Likewise.
8501 (is_function_default_version): Likewise.
8502 * attribs.c (attr_strcmp): Likewise.
8503 (sorted_attr_string): Likewise.
8504 (common_function_versions): Likewise.
8505 (make_unique_name): Likewise.
8506 (make_dispatcher_decl): Likewise.
8507 (is_function_default_version): Likewise.
8508 * config/i386/i386.c (attr_strcmp): Likewise.
8509 (sorted_attr_string): Likewise.
8510 (ix86_function_versions): Likewise.
8511 (make_name): Likewise.
8512 (make_dispatcher_decl): Likewise.
8513 (is_function_default_version): Likewise.
8514 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8516 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8519 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8520 Account for direct move costs for vec_construct of integer
8523 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8526 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8527 (UNSPEC_STX_ATOMIC): Ditto.
8528 (loaddi_via_sse): New insn.
8529 (storedi_via_sse): Ditto.
8530 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8531 Update corresponding peephole2 patterns.
8532 (atomic_storedi_fpu): Ditto.
8534 2017-05-11 Julia Koval <julia.koval@intel.com>
8536 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8537 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8539 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8540 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8541 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8543 2017-05-11 Nathan Sidwell <nathan@acm.org>
8545 * graphite-poly.c: Include dumpfile.h.
8547 * dumpfle.h (dump_function): Declare here ...
8548 * tree-dump.h (dump_function): ... not here.
8549 * dumpfile.c: #include tree-cfg.h.
8550 (dump_function): Move here from ...
8551 * tree-dump.c (dump_function): ... here.
8552 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8553 * graphite-poly.c: Don't include tree-dump.h.
8554 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8555 * print-tree.c: Likewise.
8556 * stor-layout.c: Likewise.
8557 * tree-nested.c: Likewise.
8559 * dumpfile.c (dump_start): Use TDF_FLAGS.
8560 (dump_enable_all): Fix TDF_KIND check thinko.
8562 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8564 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8565 array entries to represent two legal parameterizations of the
8566 overloaded __builtin_cmpb function, as represented by the
8567 P6_OV_BUILTIN_CMPB constant.
8568 (altivec_resolve_overloaded_builtin): Add special case handling
8569 for the __builtin_cmpb function, as represented by the
8570 P6_OV_BUILTIN_CMPB constant.
8571 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8572 (BU_P6_64BIT_2): New macro.
8573 (BU_P6_OVERLOAD_2): New macro
8574 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8575 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8576 (CMPB): Add overload support to represent both 32-bit and 64-bit
8577 compare-bytes function.
8578 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8579 support for TARGET_CMPB.
8580 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8581 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8582 documentation of the __builtin_cmpb overloaded built-in function.
8584 2017-05-11 Richard Biener <rguenther@suse.de>
8586 PR tree-optimization/80705
8587 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8588 bases are not vectorizable.
8590 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8592 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8593 when counting register pressure.
8595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8597 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8598 (iv_ca_more_deps): Renamed to ...
8599 (iv_ca_compare_deps): ... this.
8600 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8602 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8604 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8606 (determine_group_iv_costs): ... here.
8607 (find_inv_vars_cb): Record inv var if it's not recorded before.
8609 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8611 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8612 (get_shiftadd_cost): Ditto.
8614 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8616 * tree-ssa-address.c: Include header file.
8617 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8619 (add_to_parts): Refactor.
8620 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8621 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8624 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8626 PR tree-optimization/53090
8627 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8629 (extract_cond_operands): Detect condition with IV on both sides
8630 and return COMP_IV_EXPR_2.
8631 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8632 (rewrite_use_compare): Simplify by removing call to function
8633 extract_cond_operands.
8635 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8637 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8638 (extract_cond_operands): Detect condition comparing against non-
8639 invariant bound and return appropriate enum value.
8640 (find_interesting_uses_cond): Update use of extract_cond_operands.
8641 Handle its return value accordingly.
8642 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8644 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8646 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8647 nonlinear iv_use computation in loop invariant sensitive way.
8649 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8651 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8652 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8654 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8656 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8657 (dump_cand): Support iv_cand.inv_exprs.
8658 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8660 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8663 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8665 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8667 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8668 as local function. Include necessary header files.
8669 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8671 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8673 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8675 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8677 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8678 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8679 RSHIFT_EXPR and BIT_NOT_EXPR.
8681 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8683 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8684 (adjust_setup_cost): New parameter supporting round up adjustment.
8685 (struct address_cost_data): Delete.
8686 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8687 (split_address_cost, ptr_difference_cost): Delete.
8688 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8689 (struct ainc_cost_data): New struct.
8690 (get_address_cost_ainc): New function.
8691 (get_address_cost, get_computation_cost): Reimplement.
8692 (determine_group_iv_cost_address): Record inv_expr for all uses of
8694 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8695 (iv_ca_has_deps): Reimplemented to ...
8696 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8698 (iv_ca_extend): Call iv_ca_more_deps.
8700 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8702 * tree-ssa-address.c (struct mem_address): Move to header file.
8703 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8704 * tree-ssa-address.h (struct mem_address): Move from C file.
8705 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8707 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8709 * tree-affine.h (aff_combination_type): New interface.
8710 (aff_combination_zero_p): Remove static.
8711 (aff_combination_const_p): New interface.
8712 (aff_combination_singleton_var_p): New interfaces.
8714 2017-05-11 Richard Biener <rguenther@suse.de>
8716 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8717 Skip unreachable blocks and destinations.
8718 (eliminate): Move stmt removal and fixup ...
8719 (fini_eliminate): ... here. Skip inserted exprs.
8720 (pass_pre::execute): Move fini_pre after fini_eliminate.
8721 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8722 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8723 PRE to get rid of dead code that has invalid SSA form and
8724 split critical edges again.
8726 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8728 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8730 2017-05-11 Richard Biener <rguenther@suse.de>
8732 * passes.c (execute_function_todo): Verify loops if they are
8733 said to be up-to-date.
8734 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8735 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8737 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8740 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8741 handle calling assemble_external ourself.
8744 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8745 modes with zero size. Enhance comment.
8747 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8749 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8750 built-ins for vec_xl and vec_xst with short and char pointer
8753 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8755 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8756 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8757 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8758 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8759 (_mm_maskz_min_round_ss): New intrinsics.
8760 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8761 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8762 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8763 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8764 (__builtin_ia32_minss_mask_round): New builtins.
8765 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8766 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8767 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
8769 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
8770 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
8772 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
8775 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8777 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
8778 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
8779 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
8780 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
8781 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
8782 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
8783 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
8784 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8785 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8786 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
8787 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
8788 (__builtin_ia32_mulss_mask_round): New builtins.
8789 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8790 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8791 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
8793 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
8794 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8796 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8799 2017-05-10 Julia Koval <julia.koval@intel.com>
8801 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
8802 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
8803 (_mm256_setr_m128i): New intrinsics.
8805 2017-05-10 Julia Koval <julia.koval@intel.com>
8807 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
8808 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
8809 (_mm_maskz_rcp14_ss): New intrinsics.
8810 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
8811 (__builtin_ia32_rcp14ss_mask): New builtins.
8812 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
8814 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
8816 PR tree-optimization/51513
8817 * tree-cfg.c (gimple_seq_unreachable_p): New function.
8818 (assert_unreachable_fallthru_edge_p): Use it.
8819 (group_case_labels_stmt): Likewise.
8820 * tree-cfg.h: Prototype it.
8821 * stmt.c: Include cfghooks.h and tree-cfg.h.
8822 (emit_case_dispatch_table) <gap_label>: New local variable.
8823 Use it to fill dispatch table gaps.
8824 Test for default_label before updating probabilities.
8825 (expand_case) <default_label>: Remove unneeded initialization.
8826 Test for unreachable default case statement and remove its edge.
8827 Set default_label accordingly.
8828 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
8830 2017-05-10 Carl Love <cel@us.ibm.com>
8832 * config/rs6000/rs6000-c: Add support for built-in functions
8833 vector signed char vec_neg (vector signed char)
8834 vector signed short int vec_neg (vector short int)
8835 vector signed int vec_neg (vector signed int)
8836 vector signed long long vec_neg (vector signed long long)
8837 vector float vec_neg (vector float)
8838 vector double vec_neg (vector double)
8839 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
8841 * config/rs6000/altivec.h: Add define for vec_neg
8842 * doc/extend.texi: Update the built-in documentation for the
8843 new built-in functions.
8845 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8847 PR tree-optimization/77644
8848 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
8850 2017-05-10 Nathan Sidwell <nathan@acm.org>
8852 * dumpfile.h (TDI_lang_all): New.
8853 (TDF_KIND): New. Renumber others
8854 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
8856 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
8858 (get_dump_file_name): Adjust suffix generation.
8859 (dump_enable_all): Use TDF_KIND.
8860 * doc/invoke.texi (-fdump-lang-all): Document.
8862 * dumpfile.h: Tabify.
8864 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
8867 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
8868 Move member access before delete.
8870 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
8872 * tree-inline.c (expand_call_inline): Split block at stmt
8875 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8878 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
8879 are now unused after splitting mov{sf,sd}_hardfloat.
8880 (f32_lr2): Likewise.
8882 (f32_lm2): Likewise.
8884 (f32_li2): Likewise.
8887 (f32_sr2): Likewise.
8889 (f32_sm2): Likewise.
8891 (f32_si2): Likewise.
8894 (f32_vsx): Likewise.
8896 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
8897 For movsf, order stores so the VSX stores occur before the GPR
8898 store which encourages the register allocator to use a traditional
8899 FPR instead of a GPR. For movsd, order the stores so that the GPR
8900 store comes before the VSX stores to allow the power6 to work.
8901 This is due to the power6 not having a 32-bit integer store
8902 instruction from a FPR.
8903 (movsf_hardfloat): Likewise.
8904 (movsd_hardfloat): Likewise.
8906 2017-05-09 Martin Sebor <msebor@redhat.com>
8908 PR translation/80280
8909 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
8912 PR translation/80280
8913 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
8914 data member added in r247778.
8915 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
8917 2017-05-09 Nathan Sidwell <nathan@acm.org>
8919 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
8921 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
8924 2017-05-09 Marek Polacek <polacek@redhat.com>
8926 * doc/invoke.texi: Fix typo.
8928 2017-05-09 Richard Biener <rguenther@suse.de>
8930 * tree-vrp.c (vrp_val_is_max): Adjust comment.
8931 (vrp_val_is_min): Likewise.
8932 (set_value_range_to_value): Likewise.
8933 (set_value_range_to_nonnegative): Likewise.
8934 (gimple_assign_nonzero_p): Likewise.
8935 (gimple_stmt_nonzero_p): Likewise.
8936 (vrp_int_const_binop): Likewise. Remove unreachable case.
8937 (adjust_range_with_scev): Adjust comments.
8938 (compare_range_with_value): Likewise.
8939 (extract_range_from_phi_node): Likewise.
8940 (test_for_singularity): Likewise.
8942 2017-05-09 Richard Biener <rguenther@suse.de>
8944 * tree-vrp.c (get_single_symbol): Add assert that we don't
8945 get overflowed constants as invariant part.
8946 (compare_values_warnv): Add comment before the TREE_NO_WARNING
8947 checks. Use wi::cmp instead of recursing for integer constants.
8948 (compare_values): Just ignore whether we assumed undefined
8949 overflow instead of failing the compare.
8950 (extract_range_for_var_from_comparison_expr): Add comment before the
8951 TREE_NO_WARNING sets.
8952 (test_for_singularity): Likewise.
8953 (extract_range_from_comparison): Do not disable optimization
8954 when we assumed undefined overflow.
8955 (extract_range_basic): Remove init of unused var.
8957 2017-05-09 Richard Biener <rguenther@suse.de>
8959 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
8960 (extract_range_from_multiplicative_op_1): Adjust.
8961 (extract_range_from_binary_expr_1): Use int_const_binop.
8963 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
8966 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
8967 rs6000_store_data_bypass_p in seven define_bypass directives and
8968 in several comments.
8969 * config/rs6000/rs6000-protos.h: Add prototype for
8970 rs6000_store_data_bypass_p function.
8971 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
8972 function implements slightly different (rs6000-specific) semantics
8973 than store_data_bypass_p, returning false rather than aborting
8974 with assertion error when arguments do not satisfy the
8975 requirements of store data bypass.
8976 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
8977 rs6000_store_data_bypass_p.
8979 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
8981 * config/xtensa/xtensa-protos.h
8982 (xtensa_initial_elimination_offset): New declaration.
8983 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
8984 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
8985 macro definition, add case for FRAME_POINTER_REGNUM when
8986 FRAME_GROWS_DOWNWARD.
8987 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
8988 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
8989 xtensa_initial_elimination_offset.
8991 2017-05-08 Nathan Sidwell <nathan@acm.org>
8993 * doc/invoke.texi: Alphabetize -fdump options.
8995 2017-05-08 Martin Sebor <msebor@redhat.com>
8997 PR translation/80280
8998 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9000 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9002 * target.def (compute_frame_layout): New optional target hook.
9003 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9004 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9005 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9007 * reload1.c (verify_initial_elim_offsets): Likewise.
9008 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9009 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9010 (arm_get_frame_offsets): Split up into this ...
9011 (arm_compute_frame_layout): ... and this function.
9013 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9015 * config/aarch64/constraints.md (Usa): New constraint.
9016 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9018 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9020 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9021 with_multilib_list after it has been checked.
9023 2017-05-08 Richard Biener <rguenther@suse.de>
9025 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9026 (bitmap_set_subtract_values): Likewise.
9028 2017-05-08 Richard Biener <rguenther@suse.de>
9030 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9031 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9033 (gimple_stmt_nonzero_warnv_p): Rename to ...
9034 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9036 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9037 (extract_range_basic): Adjust, do not disable propagation on
9038 strict overflow sensitive simplification.
9039 (vrp_visit_cond_stmt): Likewise.
9041 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9043 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9044 body size unconditionally.
9046 2017-05-07 Jeff Law <law@redhat.com>
9049 2017-05-06 Jeff Law <law@redhat.com>
9050 PR tree-optimization/78496
9051 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9054 PR tree-optimization/78496
9055 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9056 (simplify_stmt_using_ranges): Call it.
9057 (vrp_dom_walker::before_dom_children): Extract equivalences
9058 from an ASSERT_EXPR with an equality comparison against a
9061 2017-05-06 Jeff Law <law@redhat.com>
9063 PR tree-optimization/78496
9064 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9067 PR tree-optimization/78496
9068 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9069 (simplify_stmt_using_ranges): Call it.
9070 (vrp_dom_walker::before_dom_children): Extract equivalences
9071 from an ASSERT_EXPR with an equality comparison against a
9074 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9076 * lra-constraints.c (lra_copy_reg_equiv): New function.
9077 (split_reg): Use it to copy equivalence information from the
9078 original register to the spill register.
9080 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9082 PR rtl-optimization/75964
9083 * simplify-rtx.c (simplify_const_relational_operation): Remove
9084 invalid handling of comparisons of integer ABS.
9086 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9088 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9090 (init_regs): Remove declaration.
9091 (function_arg_advance_32): Initialize error_p as boolean variable.
9093 2017-05-05 Nathan Sidwell <nathan@acm.org>
9095 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9096 lines. Use for (;;).
9098 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9100 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9101 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9102 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9103 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9104 VF=2 that require versioning.
9106 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9108 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9111 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9113 * diagnostic.h (diagnostic_override_option_index): Convert from
9114 macro to inline function.
9116 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9118 * diagnostic.c (last_module_changed_p): New function.
9119 (set_last_module): New function.
9120 (diagnostic_report_current_module): Convert macro usage to
9121 the above functions.
9122 * diagnostic.h (diagnostic_context::last_module): Strengthen
9123 from const line_map * to const line_map_ordinary *.
9124 (diagnostic_last_module_changed): Delete macro.
9125 (diagnostic_set_last_module): Delete macro.
9127 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9129 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9130 with diagnostic_report_diagnostic.
9131 (diagnostic_n_impl_richloc): Likewise.
9132 * diagnostic.h (report_diagnostic): Delete macro.
9133 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9134 with diagnostic_report_diagnostic.
9135 * substring-locations.c (format_warning_va): Likewise.
9137 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9139 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9140 save/restor of format_spec. Move option-printing code to...
9141 (print_option_information): ...this new function, and
9142 reimplement by simply printing to the pretty_printer,
9143 rather than appending to the format string.
9145 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9147 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9148 handling logic into...
9149 (update_effective_level_from_pragmas): ...this new function.
9151 2017-05-04 Andrew Waterman <andrew@sifive.com>
9153 * config/riscv/riscv.opt (mstrict-align): New option.
9154 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9155 (SLOW_UNALIGNED_ACCESS): Define.
9156 (riscv_slow_unaligned_access): Declare.
9157 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9159 (riscv_slow_unaligned_access): New variable.
9160 (rocket_tune_info): Set slow_unaligned_access to true.
9161 (optimize_size_tune_info): Set slow_unaligned_access to false.
9162 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9163 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9164 (riscv_option_override): Set riscv_slow_unaligned_access.
9165 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9167 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9169 * config/riscv/riscv.md: Unify indentation.
9171 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9176 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9178 (extendsi<mode>2): Add support for doing sign extension via
9179 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9180 don't have ISA 3.0 instructions.
9181 (extendsi<mode>2 splitter): Likewise.
9182 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9183 generate the normal insns since SImode can now go in vector
9184 registers. Disallow the special UNSPECs needed for previous
9185 machines to hide SImode being used. Add new insns
9186 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9187 (fix_trunc<mode>si2_stfiwx): Likewise.
9188 (fix_trunc<mode>si2_internal): Likewise.
9189 (fixuns_trunc<mode>si2): Likewise.
9190 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9191 (fctiw<u>z_<mode>_smallint): Likewise.
9192 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9193 of floating point to 32-bit integer from doing a direct move to
9194 the GPR registers to do a store.
9195 (fctiwz_<mode>): Break long line.
9197 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9199 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9200 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9201 (addr_list, addr_offset_valid_p): New.
9202 (split_address_groups): Check offset validity with above function.
9203 (gt-tree-ssa-loop-ivopts.h): Include header file.
9205 2017-05-05 Nathan Sidwell <nathan@acm.org>
9207 * config.gcc (arm*-*-*): Add missing 'fi'.
9209 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9211 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9212 included in -fopt-info does not matter.
9213 * doc/optinfo.texi (-fopt-info): Fix description of default
9214 behavour. Explicitly say order of options included in -fopt-info
9217 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9219 * config.gcc: Allow combinations of aprofile and rmprofile values for
9220 --with-multilib-list.
9221 * config/arm/t-multilib: New file.
9222 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9223 variables. Remove setting of ISA and floating-point ABI in
9224 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9225 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9226 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9227 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9229 * config/arm/t-rmprofile: Likewise except for the matches changes.
9230 * doc/install.texi (--with-multilib-list): Document the combination of
9231 aprofile and rmprofile values and warn about pitfalls in doing that.
9233 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9235 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9236 (movdi_aarch64): Likewise.
9238 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9240 PR tree-optimization/80632
9241 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9243 (build_arrays): Initialize it for virtual phis.
9244 (fix_phi_nodes): Use it for virtual phis.
9246 PR tree-optimization/80558
9247 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9248 [x, y] op z into [x op, y op z] for op & or | if conditions
9251 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9252 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9255 * config/arm/arm.md (use_literal_pool): Remove.
9256 (64-bit immediate split): No longer takes cost into consideration
9257 if arm_disable_literal_pool is enabled.
9258 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9259 used when arm_disable_literal_pool is enabled.
9260 (arm_max_const_double_inline_cost): Remove use of
9261 arm_disable_literal_pool.
9262 (push_minipool_fix): Add assert.
9263 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9264 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9265 (no_literal_pool_sf_immediate): New.
9267 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9269 PR tree-optimization/80613
9270 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9271 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9273 2017-05-05 Richard Biener <rguenther@suse.de>
9275 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9277 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9279 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9280 of this flag from insn conditions due to removal from r247495.
9282 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9284 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9286 (arm_early_store_addr_dep_ptr): Likewise.
9287 * config/arm/aarch-common-protos.h
9288 (arm_early_load_addr_dep_ptr): Add prototype.
9289 (arm_early_store_addr_dep_ptr): Likewise.
9290 * config/arm/cortex-a53.md: Add new bypasses.
9292 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9294 * tree.c (next_type_uid): Change type to unsigned.
9295 (type_hash_canon): Decrement back next_type_uid if
9296 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9297 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9300 2017-05-04 Martin Sebor <msebor@redhat.com>
9302 * builtins.c: Fix a trivial typo in a comment.
9305 * builtins.c (check_sizes): Adjust to handle reading past the end.
9306 Avoid printing excessive upper bound of ranges. Use %E to print
9307 tree nodes instead of converting them to %wu.
9308 (expand_builtin_memchr): New function.
9309 (compute_dest_size): Rename...
9310 (compute_objsize): ...to this.
9311 (expand_builtin_memcpy): Adjust.
9312 (expand_builtin_mempcpy): Adjust.
9313 (expand_builtin_strcat): Adjust.
9314 (expand_builtin_strcpy): Adjust.
9315 (check_strncat_sizes): Adjust.
9316 (expand_builtin_strncat): Adjust.
9317 (expand_builtin_strncpy): Adjust and simplify.
9318 (expand_builtin_memset): Adjust.
9319 (expand_builtin_bzero): Adjust.
9320 (expand_builtin_memcmp): Adjust.
9321 (expand_builtin): Handle memcmp.
9322 (maybe_emit_chk_warning): Check strncat just once.
9324 2017-05-04 Martin Sebor <msebor@redhat.com>
9326 PR preprocessor/79214
9329 * builtins.c (check_sizes): Add inlining context and issue
9330 warnings even when -Wno-system-headers is set.
9331 (check_strncat_sizes): Same.
9332 (expand_builtin_strncat): Same.
9333 (expand_builtin_memmove): New function.
9334 (expand_builtin_stpncpy): Same.
9335 (expand_builtin): Handle memmove and stpncpy.
9337 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9339 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9340 which is not used any more.
9342 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9344 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9346 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9348 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9349 (cortexa53_tunings): Likewise.
9350 (cortexa57_tunings): Likewise.
9351 (cortexa72_tunings): Likewise.
9352 (cortexa73_tunings): Likewise.
9354 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9356 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9357 Set loop alignment to 8.
9359 2017-05-04 Martin Sebor <msebor@redhat.com>
9361 PR translation/80280
9362 * builtins.c (expand_builtin_object_size): Add missing quoting to
9363 %D and like directives.
9364 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9365 (hsa_type_for_tree_type): Same.
9366 (verify_function_arguments): Same.
9367 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9368 * varasm.c (get_section): Same.
9371 2017-05-04 Martin Sebor <msebor@redhat.com>
9373 PR translation/80280
9374 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9376 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9378 * config/aarch64/aarch64.c (generic_addrcost_table):
9379 Change HI/TI mode setting.
9381 2017-05-04 Martin Jambor <mjambor@suse.cz>
9383 PR tree-optimization/80622
9384 * tree-sra.c (comes_initialized_p): New function.
9385 (build_accesses_from_assign): Only set write lazily when
9386 comes_initialized_p is false.
9387 (analyze_access_subtree): Use comes_initialized_p.
9388 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9389 instead of testing for PARM_DECL.
9391 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9393 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9394 constraint on operand 0 to allow more general addressing modes.
9395 Adjust output template.
9396 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9398 * config/aarch64/aarch64-protos.h
9399 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9400 * config/aarch64/constraints.md (Dp): New address constraint.
9401 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9404 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9406 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9407 update use of estimate_ipcp_clone_size_and_time.
9408 (estimate_local_effects): Update use of
9409 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9410 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9411 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9412 Return nonspecialized time.
9414 2017-05-04 Richard Biener <rguenther@suse.de>
9416 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9417 for the last VUSE which def dominates the PHI. Directly call
9419 (get_continuation_for_phi_1): Remove.
9421 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9423 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9424 to explain the use of truncating division. Cap the number of
9425 iterations to the maximum given by nb_iterations_upper_bound,
9428 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9430 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9431 * configure: Regenerate.
9432 * config.in: Regenerate.
9433 * config/i386/driver-mingw32.c: new file.
9434 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9435 * config.host: Link driver-mingw32.o on MinGW host.
9436 * doc/install.texi: Document new --enable-mingw-wildcard configure
9439 2017-05-04 Marek Polacek <polacek@redhat.com>
9441 PR tree-optimization/80612
9442 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9444 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9445 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9447 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9448 (movt splitter): Likewise.
9449 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9450 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9451 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9452 block for Thumb-1 with MOVT.
9453 (thumb2_legitimate_address_p): Move code block ...
9454 (can_avoid_literal_pool_for_label_p): ... into this new function.
9455 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9457 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9458 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9459 "M-profile targets with the MOVT instruction".
9461 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9463 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9464 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9465 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9467 2017-05-04 Martin Liska <mliska@suse.cz>
9469 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9472 2017-05-04 Richard Biener <rguenther@suse.de>
9474 * tree.c (array_at_struct_end_p): Handle arrays at struct
9475 end with flexarrays more conservatively. Refactor and treat
9476 arrays of arrays or aggregates more strict. Fix
9477 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9478 * tree.c (array_at_struct_end_p): Adjust prototype.
9479 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9480 * gimple-fold.c (get_range_strlen): Likewise.
9481 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9483 2017-05-04 Richard Biener <rguenther@suse.de>
9485 PR tree-optimization/31130
9486 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9488 (supports_overflow_infinity): Likewise.
9489 (is_negative_overflow_infinity): Likewise.
9490 (is_positive_overflow_infinity): Likewise.
9491 (is_overflow_infinity): Likewise.
9492 (stmt_overflow_infinity): Likewise.
9493 (overflow_infinity_range_p): Likewise.
9494 (usable_range_p): Remove as always returning true.
9495 (make_overflow_infinity): Remove.
9496 (negative_overflow_infinity): Likewise.
9497 (positive_overflow_infinity): Likewise.
9498 (avoid_overflow_infinity): Likewise.
9499 (set_value_range): Adjust accordingly.
9500 (set_value_range_to_nonnegative): Likewise, remove now unused
9501 overflow_infinity arg.
9502 (vrp_operand_equal_p): Adjust.
9503 (update_value_range): Likewise.
9504 (range_int_cst_singleton_p): Likewise.
9505 (operand_less_p): Likewise.
9506 (compare_values_warnv): Likewise.
9507 (extract_range_for_var_from_comparison_expr): Likewise.
9508 (vrp_int_const_binop): Likewise.
9509 (zero_nonzero_bits_from_vr): Likewise.
9510 (extract_range_from_multiplicative_op_1): Likewise.
9511 (extract_range_from_binary_expr_1): Likewise.
9512 (extract_range_from_unary_expr): Likewise.
9513 (extract_range_from_comparison): Likewise.
9514 (extract_range_basic): Likewise.
9515 (adjust_range_with_scev): Likewise.
9516 (compare_ranges): Likewise.
9517 (compare_range_with_value): Likewise.
9518 (dump_value_range): Likewise.
9519 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9521 (simplify_cond_using_ranges): Adjust.
9523 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9525 * brig-builtins.def: Added a builtin for class_f64.
9526 * builtin-types.def: Added a builtin type needed by class_f64.
9528 2017-05-03 Jason Merrill <jason@redhat.com>
9530 * timevar.def: Add TV_CONSTEXPR.
9532 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9534 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9536 2017-05-03 Martin Jambor <mjambor@suse.cz>
9538 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9539 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9540 * ipa-cp.c (ipcp_propagate_stage): Do not call
9541 ipa_update_after_lto_read.
9542 * ipa-inline.c (ipa_inline): Likewise.
9544 2017-05-03 Martin Jambor <mjambor@suse.cz>
9546 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9547 tag. Added a default constructor and a destructor.
9548 (ipa_edge_args_sum_t): New class;
9549 (ipa_edge_args_sum): Declare.
9550 (ipa_edge_args_vector): Remove declaration.
9551 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9552 (ipa_free_edge_args_substructures): Remove declaration.
9553 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9554 (ipa_edge_args_info_available_for_edge_p): Likewise.
9555 * ipa-prop.c (ipa_edge_args_vector): Removed.
9556 (edge_removal_hook_holder): Likewise.
9557 (edge_duplication_hook_holder): Likewise.
9558 (ipa_edge_args_sum): New variable.
9559 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9560 ipa_edge_args_vector.
9561 (ipa_free_edge_args_substructures): Likewise.
9562 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9563 ipa_edge_args_vector.
9564 (ipa_edge_removal_hook): Turned into method
9565 ipa_edge_args_sum_t::remove.
9566 (ipa_edge_duplication_hook): Turned into method
9567 ipa_edge_args_sum_t::duplicate.
9568 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9569 registering edge hooks.
9570 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9571 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9572 ipa_edge_args_sum instead of ipa_edge_args_vector.
9573 * ipa-profile.c (ipa_profile): Likewise.
9575 2017-05-03 Martin Jambor <mjambor@suse.cz>
9577 * symbol-summary.h (function_summary): New method exists.
9578 (function_summary::symtab_removal): Deallocate through release.
9579 (call_summary): New class.
9580 (gt_ggc_mx): New overload.
9581 (gt_pch_nx): Likewise.
9582 (gt_pch_nx): Likewise.
9584 2017-05-03 Jeff Law <law@redhat.com>
9586 PR tree-optimization/78496
9587 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9588 from simplify_cond_using_ranges. Split off code to walk
9589 backwards through casts into ...
9590 (simplify_cond_using_ranges_2): New function.
9591 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9592 (execute_vrp): After identifying jump threads, call
9593 simplify_cond_using_ranges_2.
9595 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9598 * ipa-inline.h (inline_summary): Add ctor.
9599 (create_ggc): Do not use ggc_cleared_alloc.
9601 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9602 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9604 * gcc.c (handle_braces): Support escaping in switch matching
9606 * doc/invoke.texi (Spec Files): Document it.
9607 Remove superfluous @code markup in items.
9609 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9611 * diagnostic-show-locus.c (struct column_range): New struct.
9612 (get_affected_columns): New function.
9613 (get_printed_columns): New function.
9614 (struct correction): New struct.
9615 (correction::ensure_capacity): New function.
9616 (correction::ensure_terminated): New function.
9617 (struct line_corrections): New struct.
9618 (line_corrections::~line_corrections): New dtor.
9619 (line_corrections::add_hint): New function.
9620 (layout::print_trailing_fixits): Reimplement in terms of the new
9622 (selftest::test_overlapped_fixit_printing): New function.
9623 (selftest::diagnostic_show_locus_c_tests): Call it.
9625 2017-05-03 Nathan Sidwell <nathan@acm.org>
9627 Canonicalize canonical type hashing
9628 * tree.h (type_hash_canon_hash): Declare.
9629 * tree.c (type_hash_list, attribute_hash_list): Move into
9630 type_hash_canon_hash.
9631 (build_type_attribute_qual_variant): Break out hash code calc into
9632 type_hash_canon_hash.
9633 (type_hash_canon_hash): New. Generic type hash computation.
9634 (build_range_type_1, build_array_type_1, build_function_type,
9635 build_method_type_directly, build_offset_type, build_complex_type,
9636 make_vector_type): Call it.
9638 2017-05-03 Richard Biener <rguenther@suse.de>
9640 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9641 When all DRs have unknown misaligned do not always peel
9642 when there is a store but apply the same costing model as if
9643 there were only loads.
9645 2017-05-03 Richard Biener <rguenther@suse.de>
9648 PR tree-optimization/80492
9649 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9650 compare_base_decls returning dont-know properly.
9652 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9654 * config/arm/iterators.md (CCSI): New mode iterator.
9655 (arch): New mode attribute.
9656 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9657 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9658 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9659 code iterator for success result mode.
9660 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9661 the corresponding new insn generators.
9663 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9666 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9667 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9669 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9671 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9672 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9673 (DDR_A): Wrap DDR argument in brackets.
9674 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9675 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9676 (DDR_REVERSED_P): Likewise.
9678 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9680 PR tree-optimization/79472
9681 * tree-switch-conversion.c (struct switch_conv_info): Add
9682 contiguous_range and default_case_nonstandard fields.
9683 (collect_switch_conv_info): Compute contiguous_range and
9684 default_case_nonstandard fields, don't clear final_bb if
9685 contiguous_range and only the default case doesn't have the required
9687 (check_all_empty_except_final): Set default_case_nonstandard instead
9688 of failing if contiguous_range and the default case doesn't have empty
9690 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9691 and only the default case doesn't have the required constants. Skip
9693 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9694 if default_case_nonstandard.
9695 (build_constructors): Build constant 1 just once. Assert that default
9696 values aren't inserted in between cases if contiguous_range. Skip
9698 (build_arrays): Skip virtual phis.
9699 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9700 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9701 Handle virtual phis.
9702 (gen_inbound_check): Handle default_case_nonstandard case.
9703 (process_switch): Adjust check_final_bb caller. Call
9704 gather_default_values with the first non-default case instead of
9705 default case if default_case_nonstandard.
9707 2017-05-02 Nathan Sidwell <nathan@acm.org>
9709 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9710 check. Fix formatting.
9712 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9714 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9715 errors when comparing specialized and unspecialized times.
9717 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9719 * diagnostic-show-locus.c
9720 (layout::should_print_annotation_line_p): Make private.
9721 (layout::print_annotation_line): Make private.
9722 (layout::annotation_line_showed_range_p): Make private.
9723 (layout::show_ruler): Make private.
9724 (layout::print_source_line): Make private. Pass in line and
9725 line_width, rather than calling location_get_source_line. Drop
9727 (layout::print_leading_fixits): New method.
9728 (layout::print_any_fixits): Rename to...
9729 (layout::print_trailing_fixits): ...this, and make private.
9730 Don't print newline fixits.
9731 (diagnostic_show_locus): Move logic for printing one row into...
9732 (layout::print_line): ...this new function. Move the
9733 location_get_source_line call and error-handling from
9734 print_source_line to here. Call print_leading_fixits, and rename
9735 print_any_fixits to print_trailing_fixits.
9736 (selftest::test_fixit_insert_containing_newline): Update now that
9737 newlines are partially supported.
9738 (selftest::test_fixit_insert_containing_newline_2): New test.
9739 (selftest::test_fixit_replace_containing_newline): Update comments.
9740 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9741 * edit-context.c (class added_line): New class.
9742 (class edited_line): Describe newline handling in comment.
9743 (edited_line::actually_edited_p): New method.
9744 (edited_line::print_content): Delete redundant decl.
9745 (edited_line::m_predecessors): New field.
9746 (edited_file::print_content): Call edited_line::print_content.
9747 (edited_file::print_diff): Update to support newlines.
9748 (edited_file::print_diff_hunk): Likewise.
9749 (edited_file::print_run_of_changed_lines): New function.
9750 (edited_file::print_diff_line): Convert to...
9751 (print_diff_line): ...this.
9752 (edited_file::get_effective_line_count): New function.
9753 (edited_line::edited_line): Initialize new field m_predecessors.
9754 (edited_line::~edited_line): Clean up m_predecessors.
9755 (edited_line::apply_fixit): Handle newlines.
9756 (edited_line::get_effective_line_count): New function.
9757 (edited_line::print_content): New function.
9758 (edited_line::print_diff_lines): New function.
9759 (selftest::test_applying_fixits_insert_containing_newline): New
9761 (selftest::test_applying_fixits_replace_containing_newline): New
9763 (selftest::insert_line): New function.
9764 (selftest::test_applying_fixits_multiple_lines): Add example of
9766 (selftest::edit_context_c_tests): Call the new tests.
9768 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9770 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
9771 parameter cand. Update dump information.
9772 (get_computation_cost): Update uses.
9774 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9776 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
9777 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
9778 (get_computation_at, rewrite_use_address): Update use of
9779 get_computation_aff.
9781 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9783 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
9784 (get_computation): Delete.
9785 (get_computation_cost): Implement like get_computation_cost_at.
9786 Use get_computation_at.
9787 (get_computation_cost_at): Delete.
9788 (rewrite_use_nonlinear_expr): Use get_computation_at.
9789 (rewrite_use_compare, remove_unused_ivs): Ditto.
9791 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9793 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
9795 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9797 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
9798 (ivopts_global_cost_for_size): Rename parameter and update uses.
9799 (iv_ca_recount_cost): Update uses.
9800 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
9801 candidates seperately in n_invs and n_cands.
9802 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
9804 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9806 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
9807 (find_inv_vars_cb): New.
9808 (find_depends): Renamed to ...
9809 (find_inv_vars): ... this.
9810 (add_candidate_1, force_var_cost): Call find_inv_vars.
9811 (split_address_cost, determine_group_iv_cost_cond): Ditto.
9813 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9815 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
9816 inv_vars. Add inv_exprs.
9817 (struct iv_cand): Rename depends_on to inv_vars.
9818 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
9819 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
9820 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
9821 (dump_cand): Dump inv_vars.
9822 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
9823 (record_invariant, find_depends, add_candidate_1): Ditto.
9824 (set_group_iv_cost, force_var_cost): Ditto.
9825 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
9826 (get_computation_cost_at, get_computation_cost): Ditto.
9827 (determine_group_iv_cost_generic): Ditto.
9828 (determine_group_iv_cost_address): Ditto.
9829 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
9830 (determine_group_iv_costs): Ditto.
9831 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
9832 (iv_ca_set_remove_invariants): Renamed to ...
9833 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
9834 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
9835 (iv_ca_set_add_invariants): Renamed to ...
9836 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
9837 (iv_ca_set_cp): Use iv_ca_set_add_invs.
9838 (iv_ca_has_deps): Support inv_vars and inv_exprs.
9839 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
9840 (create_new_ivs): Remove useless dump.
9842 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9844 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
9846 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
9847 (iv_ca_set_no_cp, create_new_iv): Ditto.
9849 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9851 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9853 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9855 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
9856 function tree_check2.
9858 2017-05-02 Martin Liska <mliska@suse.cz>
9860 * doc/gcov.texi: Add missing preposition.
9861 * gcov.c (function_info::function_info): Properly fill up
9862 all member variables.
9864 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9866 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
9868 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9870 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
9872 2017-05-02 Martin Liska <mliska@suse.cz>
9875 * lto-streamer-in.c (lto_read_tree_1): Remove
9877 * lto-streamer.c (struct tree_hash_entry): Likewise.
9878 (struct tree_entry_hasher): Likewise.
9879 (tree_entry_hasher::hash): Likewise.
9880 (tree_entry_hasher::equal): Likewise.
9881 (lto_streamer_init): Likewise.
9882 (lto_orig_address_map): Likewise.
9883 (lto_orig_address_get): Likewise.
9884 (lto_orig_address_remove): Likewise.
9885 * lto-streamer.h: Likewise.
9886 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
9887 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
9889 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
9891 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
9892 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
9893 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
9894 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
9895 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
9896 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
9897 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
9898 (mm_maskz_sub_ss): New intrinsics.
9899 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9900 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9901 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
9902 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
9903 (__builtin_ia32_subss_mask_round): New builtins.
9904 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9905 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9906 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
9908 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
9909 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9911 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9914 2017-05-02 Martin Jambor <mjambor@suse.cz>
9916 PR tree-optimization/78687
9917 * tree-sra.c (access): New field parent.
9918 (process_subtree_disqualification): New function.
9919 (disqualify_candidate): Call it.
9920 (build_accesses_from_assign): Reset write flag if creating an
9922 (build_access_subtree): Fill in parent field and also prpagate
9923 down grp_write flag.
9924 (create_artificial_child_access): New parameter set_grp_write, set
9925 grp_write to its value.
9926 (propagate_subaccesses_across_link): Also propagate grp_write flag
9928 (propagate_all_subaccesses): Push the closest parent back to work
9929 queue if add_access_to_work_queue returned true.
9931 2017-05-02 Richard Biener <rguenther@suse.de>
9933 * common.opt (fstrict-overflow): Alias negative to fwrapv.
9934 * doc/invoke.texi (fstrict-overflow): Remove all traces of
9935 -fstrict-overflow documentation.
9936 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
9937 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
9938 flag_strict_overflow.
9939 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
9940 * lto-opts.c (lto_write_options): Do not stream it.
9941 * lto-wrapper.c (merge_and_complain): Do not handle it.
9942 * opts.c (default_options_table): Do not set -fstrict-overflow.
9943 (finish_options): Likewise do not clear it when sanitizing.
9944 * simplify-rtx.c (simplify_const_relational_operation): Do not
9945 test flag_strict_overflow.
9947 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9949 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
9950 using enabled attribute.
9951 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
9952 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
9953 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
9954 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
9955 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
9956 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
9957 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
9958 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
9959 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
9960 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
9962 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9964 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
9966 2017-05-02 Richard Biener <rguenther@suse.de>
9968 PR tree-optimization/80591
9970 2017-04-10 Richard Biener <rguenther@suse.de>
9972 * tree-ssa-structalias.c (find_func_aliases): Properly handle
9975 2017-05-02 Richard Biener <rguenther@suse.de>
9977 PR tree-optimization/80549
9978 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
9979 (cleanup_tree_cfg_noloop): Create forwarders to known loop
9980 headers if they do not have a preheader.
9982 2017-05-02 Martin Liska <mliska@suse.cz>
9985 * common.opt: Fix typo.
9986 * doc/invoke.texi: Likewise.
9988 2017-05-01 Jan Beulich <jbeulich@suse.com>
9990 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
9991 swapping, add (x,x,m,x,n) alternative.
9993 2017-05-01 Nathan Sidwell <nathan@acm.org>
9995 * calls.c (combine_pending_stack_adjustment_and_call): Remove
9996 unnecessary unadjusted_alignment check.
9998 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10001 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10003 * gimplify.c (gimplify_cilk_detach): New function.
10004 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10005 * tree-core.h: Document EXPR_CILK_SPAWN.
10006 * tree.h (EXPR_CILK_SPAWN): Define.
10008 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10010 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10011 to use new fixit_hint representation, using the "replace" logic.
10012 (get_line_span_for_fixit_hint): Likewise.
10013 (layout::print_any_fixits): Likewise.
10014 (selftest::test_one_liner_many_fixits): Rename to...
10015 (selftest::test_one_liner_many_fixits_1): ...this, and update
10016 comment and expected output to reflect that the multiple fix-it
10017 hints are now consolidated into one insertion.
10018 (selftest::test_one_liner_many_fixits_2): New test.
10019 (selftest::test_diagnostic_show_locus_one_liner): Update for
10021 (selftest::test_fixit_consolidation): Update for fix-it API
10023 * diagnostic.c (print_parseable_fixits): Likewise.
10024 * edit-context.c (edited_line::m_line_events): Convert from
10025 auto_vec <line_event *> to auto_vec <line_event>.
10026 (class line_event): Convert from abstract base class to a concrete
10027 class, taking over the role of replace_event.
10028 (class insert_event): Delete.
10029 (class replace_event): Rename to class line_event. Convert to
10031 (edit_context::add_fixits): Reimplement.
10032 (edit_context::apply_insert): Delete.
10033 (edit_context::apply_replace): Rename to...
10034 (edit_context::apply_fixit): ...this. Convert to half-open range.
10035 (edited_file::apply_insert): Delete.
10036 (edited_file::apply_replace): Rename to...
10037 (edited_file::apply_fixit): ...this.
10038 (edited_line::~edited_line): Drop deletion of events.
10039 (edited_line::apply_insert): Delete.
10040 (edited_line::apply_replace): Rename to...
10041 (edited_line::apply_fixit): ...this. Convert to half-open range.
10042 Update for change to type of m_line_events.
10043 * edit-context.h (edit_context::apply_insert): Delete.
10044 (edit_context::apply_replace): Rename to...
10045 (edit_context::apply_fixit): ...this.
10047 2017-05-01 Martin Sebor <msebor@redhat.com>
10049 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10052 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10055 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10056 __get_cpuid_max returns 0.
10057 (__get_cpuid_count): Ditto.
10059 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10061 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10062 replacement expression is another instance of one of its arguments.
10064 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10067 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10068 check for stack push/pop autoinc.
10069 * config/i386/i386.c (ix86_agi_dependent): Return false
10070 if the only reason why modified_in_p returned true is that
10071 addr is SP based and set_insn is a push or pop.
10073 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10075 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10078 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10081 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10082 (account_size_time): Use two predicates - exec_pred and
10084 (evaluate_conditions_for_known_args): Compute both clause and
10086 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10087 (inline_summary_t::duplicate): Update.
10088 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10090 (compute_inline_parameters): Likewise.
10091 (estimate_edge_size_and_time): Update caluclation of time.
10092 (estimate_node_size_and_time): Compute both time and nonspecialized
10094 (estimate_ipcp_clone_size_and_time): Update.
10095 (inline_merge_summary): Update.
10096 (do_estimate_edge_time): Update.
10097 (do_estimate_edge_size): Update.
10098 (do_estimate_edge_hints): Update.
10099 (inline_read_section, inline_write_summary): Stream both new predicates.
10100 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10102 (compute_inlined_call_time): Cleanup.
10103 (big_speedup_p): Update.
10104 (edge_badness): Update.
10105 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10106 (size_time_entry): Replace predicate by exec_predicate and
10107 nonconst_predicate.
10108 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10109 (estimate_edge_time): Return also nonspec_time.
10110 (reset_edge_growth_cache): Update.
10112 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10114 PR rtl-optimization/80491
10115 * ifcvt.c (noce_process_if_block): When looking for x setter
10116 with missing else_bb, don't check only the insn right before
10117 cond_earliest, but look for the last insn that x is modified in
10118 within the same bb.
10120 PR rtl-optimization/80491
10121 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10123 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10125 PR tree-optimization/80487
10126 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10128 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10130 PR tree-optimization/79697
10131 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10132 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10133 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10135 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10136 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10138 2017-04-28 Martin Sebor <msebor@redhat.com>
10140 PR tree-optimization/80523
10141 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10142 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10144 (maybe_warn, format_directive, parse_directive): Use new functions.
10145 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10147 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10149 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10151 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10153 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10154 target_header_dir): Set correctly.
10155 * configure: Regenerated.
10156 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10157 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10158 instead of SYSTEM_HEADER_DIR.
10160 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10162 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10163 (estimate_local_effects): Likewise.
10164 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10165 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10166 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10167 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10168 do_estimate_edge_time, estimate_edge_time): Likewise.
10169 * ipa-inline-analysis.c (estimate_node_size_and_time,
10170 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10171 (estimate_time_after_inlining): Remove.
10173 2017-04-28 Martin Liska <mliska@suse.cz>
10175 * doc/gcov.texi: Enhance documentation of gcov.
10177 2017-04-28 Martin Liska <mliska@suse.cz>
10179 * doc/gcov.texi: Sort options in alphabetic order.
10180 * doc/gcov-dump.texi: Likewise.
10181 * doc/gcov-tool.texi: Likewise.
10182 * gcov.c (print_usage): Likewise.
10183 * gcov-dump.c (print_usage): Likewise.
10184 * gcov-tool.c (print_merge_usage_message): Likewise.
10185 (print_rewrite_usage_message): Likewise.
10186 (print_overlap_usage_message): Likewise.
10188 2017-04-28 Martin Liska <mliska@suse.cz>
10190 PR gcov-profile/53915
10191 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10193 2017-04-28 Martin Liska <mliska@suse.cz>
10195 PR gcov-profile/79891
10196 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10197 is marked by compiler as living on a line.
10198 (get_cycles_count): Remove usage of the union.
10199 (output_intermediate_file): Likewise.
10200 (find_source): Fix GNU coding style.
10201 (accumulate_line_counts): Remove old non-all block mode.
10202 (output_lines): Remove usage of the union.
10203 * profile.c (output_location): Include all BBs, even if
10204 belonging to a same line (and file) as a previous BB.
10206 2017-04-28 Martin Liska <mliska@suse.cz>
10208 * gcov.c (process_args): Handle new argument 'w'.
10209 (read_graph_file): Assign ID to BBs.
10210 (output_branch_count): Display BB # if verbose flag is set.
10211 (output_lines): Likewise for arcs.
10212 (print_usage): Add '--verbose' option help.
10213 * doc/gcov.texi: Document --verbose (-w) option.
10215 2017-04-28 Martin Liska <mliska@suse.cz>
10217 * gcov.c (struct block_location_info): New struct.
10218 (process_file): Fill up the new structure.
10219 (read_graph_file): Replace usage of encoding by the newly added
10221 (add_line_counts): Likewise.
10222 (accumulate_line_counts): Remove usage of the union.
10223 (function_info::function_info): New function.
10224 (function_info::~function_info): Likewise.
10225 (process_file): Call delete instead of release_function.
10226 (release_function): Release the function.
10227 (release_structures): Call delete instead of release_function.
10228 (solve_flow_graph): Replace usage of num_blocks.
10229 (find_exception_blocks): Likewise.
10230 (output_lines): Fix GNU coding style.
10232 2017-04-28 Martin Liska <mliska@suse.cz>
10235 * coverage.c (coverage_remove_note_file): New function.
10236 * coverage.h: Declare the function.
10237 * toplev.c (finalize): Clean if an error has been seen.
10239 2017-04-28 Martin Liska <mliska@suse.cz>
10241 PR gcov-profile/80031
10242 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10243 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10244 * gcov.c (read_graph_file): Read just number of blocks.
10245 * profile.c (branch_prob): Do not stream 0 flags per a basic
10248 2017-04-28 Martin Liska <mliska@suse.cz>
10250 * gcov-dump.c (tag_*): Add new argument to declarations.
10251 (dump_gcov_file): Likewise.
10252 (tag_blocks): Add and use new argument depth.
10253 (tag_arcs): Likewise.
10254 (tag_lines): Likewise.
10255 (tag_counters): Likewise.
10256 (tag_summary): Likewise.
10257 (dump_working_sets): Use depth to do a proper indentation.
10259 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10262 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10263 * symtab.c (symtab_node::debug_symtab): Move definition here.
10265 2017-04-28 Richard Biener <rguenther@suse.de>
10267 * lto-streamer.h (LTO_major_version): Bump to 7.
10269 2017-04-28 Richard Biener <rguenther@suse.de>
10271 * tree-vrp.c (assert_info): New struct.
10272 (add_assert_info): New helper.
10273 (register_edge_assert_for_2): Refactor to add asserts to a vector
10275 (register_edge_assert_for_1): Likewise.
10276 (register_edge_assert_for): Likewise.
10277 (finish_register_edge_assert_for): New helper actually registering
10278 asserts where live on edge.
10279 (find_conditional_asserts): Adjust.
10280 (find_switch_asserts): Likewise.
10281 (evrp_dom_walker::try_find_new_range): Generalize.
10282 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10284 2017-04-27 Marek Polacek <polacek@redhat.com>
10287 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10288 arg10 and arg11 to itype.
10290 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10292 * doc/extend.texi (Object Size Checking): Improve grammar.
10294 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10297 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10298 that the logic for permitting reciprocal estimates matches that
10301 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10304 * tree.c (type_cache_hasher::equal): Only compare
10305 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10306 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10307 non-aggregate element types.
10308 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10309 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10311 2017-04-27 Richard Biener <rguenther@suse.de>
10313 PR middle-end/80533
10314 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10315 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10316 keeping a reference to a trailing array.
10318 2017-04-27 Richard Biener <rguenther@suse.de>
10320 PR middle-end/80539
10321 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10322 being in loop-closed SSA form conservatively.
10323 (chrec_fold_multiply_poly_poly): Likewise.
10325 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10327 PR middle-end/79665
10328 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10329 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10331 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10334 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10335 (aarch64_function_arg_alignment): Return unsigned int again, but still
10336 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10337 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10338 Don't emit -Wpsabi note.
10339 (aarch64_function_arg_boundary): Likewise.
10340 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10343 2017-04-26 Nathan Sidwell <nathan@acm.org>
10345 * tree.h (crc32_unsigned_n): Declare.
10346 (crc32_unsigned, crc32_unsigned): Make inline.
10347 * tree.c (crc32_unsigned_bits): Replace with ...
10348 (crc32_unsigned_n): ... this.
10349 (crc32_unsigned, crc32_byte): Remove.
10350 (crc32_string): Remove unnecessary braces.
10352 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10354 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10355 * ipa-inline-analysis.c (MAX_TIME): Remove.
10356 (account_size_time): Use sreal for time.
10357 (dump_inline_summary): Update.
10358 (estimate_function_body_sizes): Update.
10359 (estimate_edge_size_and_time): Update.
10360 (estimate_calls_size_and_time): Update.
10361 (estimate_node_size_and_time): Update.
10362 (inline_merge_summary): Update.
10363 (inline_update_overall_summary): Update.
10364 (estimate_time_after_inlining): Update.
10365 (inline_read_section): Update.
10366 (inline_write_summary): Update.
10367 * ipa-inline.c (compute_uninlined_call_time): Update.
10368 (compute_inlined_call_time): Update.
10369 (recursive_inlining): Update.
10370 (inline_small_functions): Update.
10371 (dump_overall_stats): Update.
10372 * ipa-inline.h: Include sreal.h.
10373 (size_time_entry): Turn time to sreal.
10374 (inline_summary): Turn self_time nad time to sreal.
10376 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10378 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10380 (sreal::stream_out, sreal::stream_in): New.
10381 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10383 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10385 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10388 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10391 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10393 (dimode_scalar_chain::compute_convert_gain): Ditto.
10394 (dimode_scalar_chain::make_vector_copies): Ditto.
10395 (dimode_scalar_chain::convert_reg): Ditto.
10396 (dimode_scalar_chain::convert_insn): Ditto.
10397 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10398 (VI248_AVX512BW_1): New mode iterator.
10399 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10400 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10403 2017-04-25 Martin Sebor <msebor@redhat.com>
10405 PR tree-optimization/80497
10406 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10407 constants are representable in HOST_WIDE_INT.
10408 (parse_directive): Ditto.
10410 2017-04-25 Martin Sebor <msebor@redhat.com>
10413 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10414 (new_zero_array): Adjust signature.
10415 (dom_info::dom_init): Used unsigned rather that size_t.
10416 (dom_info::dom_info): Same.
10418 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10419 Jakub Jelinek <jakub@redhat.com>
10422 * config/arm/arm.c: Include gimple.h.
10423 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10424 returns negative, increment ncrn only if it returned positive.
10425 (arm_needs_doubleword_align): Return int instead of bool,
10426 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10427 members, but if there is any such non-FIELD_DECL
10428 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10429 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10430 returns negative, increment nregs only if it returned positive.
10431 (arm_setup_incoming_varargs): Likewise.
10432 (arm_function_arg_boundary): Emit -Wpsabi note if
10433 arm_needs_doubleword_align returns negative, return
10434 DOUBLEWORD_ALIGNMENT only if it returned positive.
10436 2017-04-25 Marek Polacek <polacek@redhat.com>
10439 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10440 first argument to type.
10442 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10445 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10446 type checks to test for compatibility instead of equality.
10448 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10449 Jakub Jelinek <jakub@redhat.com>
10452 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10454 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10455 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10456 the alignment computation, but return their maximum in warn_alignment.
10457 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10458 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10460 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10461 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10464 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10466 * config/arc/simdext.md (dmpyh): Fix typo.
10468 2017-04-25 Richard Biener <rguenther@suse.de>
10470 PR tree-optimization/80492
10471 * alias.c (compare_base_decls): Handle registers with asm
10472 specification conservatively.
10473 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10474 compare_base_decls returning dont-know properly.
10476 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10478 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10479 (legitimate_offset_address_p): New function.
10480 (arc_legitimate_address_p): Use above function.
10482 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10484 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10486 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10488 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10489 ACCH registers whenever they are available.
10491 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10493 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10494 double regs fix when not used.
10496 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10498 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10500 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10501 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10503 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10505 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10506 when emitting short ADD instructions.
10508 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10510 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10512 (cmpsi_cc_c_insn): Likewise.
10513 (cbranchsi4_scratch): Compute proper instruction length using
10514 compact_hreg_operand.
10515 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10517 2017-04-25 Richard Biener <rguenther@suse.de>
10519 PR middle-end/80509
10520 * passes.c (pass_manager::pass_manager): Initialize
10521 m_name_to_pass_map.
10523 2017-04-25 Richard Biener <rguenther@suse.de>
10525 PR tree-optimization/79201
10526 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10528 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10531 * config/s390/vector.md: Split MEM->GPR vector moves for
10532 non-s_operand addresses.
10534 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10537 * config/s390/predicates.md (reload_const_wide_int_operand): New
10539 * config/s390/s390.md ("movti"): Remove d/P alternative.
10540 ("movti_bigconst"): New pattern definition.
10542 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10545 * s390-protos.h (s390_expand_cs_hqi): Removed.
10546 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10547 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10548 modes as well as CCZ1mode and CCZmode.
10549 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10550 signature of s390_emit_compare_and_swap.
10551 (s390_expand_cs_hqi): Likewise, make static.
10552 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10553 compare-and-swap, in some cases.
10554 (s390_expand_cs): Wrapper function.
10555 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10557 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10558 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10559 patterns for small and large integers. Forbid symref memory operands.
10560 Move expander to s390.c. Require cc register.
10561 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10562 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10563 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10564 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10565 symref memory operands. Remove CC mode and call s390_match_ccmode
10567 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10569 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10571 * config/s390/s390.md (define_peephole2): New peephole to help
10572 combining the load-and-test pattern with volatile memory.
10574 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10576 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10577 with CCZmode for TARGET_Z196.
10579 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10581 PR rtl-optimization/80501
10582 * combine.c (make_compound_operation_int): Set subreg_code to SET
10583 even for AND with mask of the sign bit of mode.
10585 PR rtl-optimization/80500
10586 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10587 sum's initial value.
10589 2017-04-25 Julian Brown <julian@codesourcery.com>
10590 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10592 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10594 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10596 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10598 2017-04-25 Julian Brown <julian@codesourcery.com>
10599 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10601 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10602 (thunderx2t99_sha): New Reservation.
10604 2017-04-25 Julian Brown <julian@codesourcery.com>
10605 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10607 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10608 type for 1-element load.
10610 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10612 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10614 2017-04-24 Martin Jambor <mjambor@suse.cz>
10616 PR tree-optimization/80293
10617 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10618 char arrays not totally scalarizable if it is false.
10619 (analyze_all_variable_accesses): Pass correct value in the new
10620 parameter. Add a statistics counter.
10622 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10624 PR middle-end/79931
10625 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10627 2017-04-24 Richard Biener <rguenther@suse.de>
10629 PR tree-optimization/80494
10630 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10631 out for complex types.
10633 2017-04-24 Richard Biener <rguenther@suse.de>
10635 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10636 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10637 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10638 (DFS): Adjust and never fail.
10639 (sccvn_dom_walker::fail): Remove.
10640 (sccvn_dom_walker::before_dom_children): Adjust.
10641 (run_scc_vn): Likewise and never fail.
10642 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10643 (pass_fre::execute): Likewise.
10645 2017-04-24 Richard Biener <rguenther@suse.de>
10647 PR tree-optimization/79725
10648 * tree-ssa-sink.c (statement_sink_location): Return whether
10649 failure reason was zero uses. Move that check later.
10650 (sink_code_in_bb): Deal with zero uses by removing the stmt
10653 2017-04-24 Richard Biener <rguenther@suse.de>
10656 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10657 pointer-based references.
10659 2017-04-24 Richard Biener <rguenther@suse.de>
10662 * pass_manager.h (pass_manager::operator new): Remove.
10663 (pass_manager::operator delete): Likewise.
10664 * passes.c (pass_manager::operator new): Remove.
10665 (pass_manager::operator delete): Likewise.
10666 (pass_manager::pass_manager): Zero individual pass members.
10668 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10671 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10672 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10673 Check "XEXP (src, 1)" operand here.
10674 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10675 Check "XEXP (src, 1)" operand here.
10676 (dimode_scalar_chain::make_vector_copies): Detect count register
10677 of a shift instruction. Zero extend count register from QImode
10678 to DImode to satisfy vector shift pattern count operand predicate.
10679 Substitute vector shift count operand with a DImode copy.
10680 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10683 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10685 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10686 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10687 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10688 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10689 (UNSPEC_NOREX_MEM): Remove definition.
10691 2017-04-21 Richard Biener <rguenther@suse.de>
10693 PR tree-optimization/79547
10694 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10695 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10696 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10697 without any constraints.
10699 2017-04-21 Richard Biener <rguenther@suse.de>
10701 PR tree-optimization/78847
10702 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10704 2017-04-21 Richard Biener <rguenther@suse.de>
10706 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10707 (build_distinct_type_copy): Likewise.
10708 (build_variant_type_copy): Likewise.
10709 * tree.c (build_qualified_type): Pass down mem-stat info.
10710 (build_distinct_type_copy): Likewise.
10711 (build_variant_type_copy): Likewise.
10713 2017-04-21 Richard Biener <rguenther@suse.de>
10715 PR tree-optimization/80237
10716 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10718 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10719 for a simplified result.
10721 2016-04-21 Richard Biener <rguenther@suse.de>
10723 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10724 sth as strict as a simple_iv but a chrec without symbols and an
10725 operand defined in the loop we are peeling (and not some subloop).
10726 (propagate_constants_for_unrolling): Propagate all constants.
10728 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10731 * config/i386/i386.c (print_reg): Remove assert for disalowed
10732 regno values, call output_operand_lossage instead.
10734 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10737 * config/i386/constraints.md (Yc): New register constraint.
10738 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10739 Use Yc constraint for alternative 2 of operand 0. Remove
10740 preferred_for_speed attribute.
10742 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10744 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10745 lastprivate clauses in SIMT case.
10747 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10749 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10751 2017-04-20 Richard Biener <rguenther@suse.de>
10753 PR tree-optimization/57796
10754 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10755 as N scalar stores.
10756 (vect_model_load_cost): Cost gathers as N scalar loads.
10758 2017-04-20 Richard Biener <rguenther@suse.de>
10760 * ggc-page.c (ggc_allocated_p): Rename to ...
10761 (safe_lookup_page_table_entry): ... this and return the lookup
10763 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10765 2017-04-20 Richard Biener <rguenther@suse.de>
10767 PR tree-optimization/80453
10768 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
10769 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
10770 from the conditions.
10771 (vn_phi_eq): Pass them down.
10772 (vn_phi_lookup): Record them.
10773 (vn_phi_insert): Likewise.
10775 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
10777 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
10778 uninitialized variable warning to avoid buffer overrun.
10780 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10783 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
10784 is suppressed for '{ 0 }' in C.
10786 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10788 * BASE-VER: Set to 8.0.0.
10790 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
10792 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
10793 priority .init_array and .fini_array section with SECTION_NOTYPE
10796 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10798 PR middle-end/80423
10799 * tree.h (build_array_type): Add typeless_storage default argument.
10800 * tree.c (type_cache_hasher::equal): Also compare
10801 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
10802 (build_array_type): Add typeless_storage argument, set
10803 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
10805 (build_nonshared_array_type): Adjust build_array_type_1 caller.
10806 (build_array_type): Likewise. Add typeless_storage argument.
10808 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10809 Jakub Jelinek <jakub@redhat.com>
10811 PR tree-optimization/80426
10812 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
10813 operation on symbolic operands, also compute the overflow for the
10814 invariant part when the operation degenerates into a negation.
10816 2017-04-19 Jakub Jelinek <jakub@redhat.com>
10819 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
10820 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
10823 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
10825 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
10828 * config/avr/avr.c (tree.h): Include it.
10829 (cgraph.h): Include it.
10830 (avr_encode_section_info): Don't warn for uninitialized progmem
10831 variable if it's just an alias.
10833 2017-04-19 Richard Biener <rguenther@suse.de>
10836 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
10837 when needed by AutoPGO.
10839 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
10842 * doc/lto.texi: Remove an extra 'that'.
10844 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
10846 PR rtl-optimization/80429
10847 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
10848 are only used in debug insns.
10850 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10851 Vladimir Makarov <vmakarov@redhat.com>
10853 * config/sparc/predicates.md (input_operand): Add comment. Return
10854 true for any memory operand when LRA is in progress.
10855 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
10857 2017-04-18 Jeff Law <law@redhat.com>
10860 * mips.md ({return,simple_return}_internal): Do not overwrite
10863 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10865 PR tree-optimization/80443
10866 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
10867 instead of adding 1, subtract -1 and similarly instead of subtracting
10870 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
10872 PR rtl-optimization/80357
10873 * haifa-sched.c (tmp_bitmap): New variable.
10874 (model_recompute): Handle duplicate use records.
10875 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
10876 (free_global_sched_pressure_data): Free it.
10878 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
10881 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10882 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10883 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10884 instead of SYSTEM_HEADER_DIR.
10886 2017-04-18 Jeff Law <law@redhat.com>
10888 PR middle-end/80422
10889 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
10890 predecessors after walking up the insn chain.
10892 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10895 * dwarf2out.c (modified_type_die): Try harder not to emit internal
10896 sizetype type into debug info.
10898 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
10901 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
10902 unneeded test for TARGET_UPPER_REGS_SF.
10903 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
10905 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10908 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
10909 instead of gsi_after_labels.
10911 2017-04-18 Jeff Law <law@redhat.com>
10913 * regcprop.c (maybe_mode_change): Avoid creating copies of the
10917 2017-04-13 Jeff Law <law@redhat.com>
10918 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10919 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10921 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
10924 * config/avr/avr.c (intl.h): Include it.
10925 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
10927 2017-04-18 Martin Liska <mliska@suse.cz>
10929 PR gcov-profile/78783
10930 * gcov-tool.c (gcov_output_files): Validate that destination
10931 file is either removed by the tool or by a user.
10933 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10934 Guy Benyei <guybe@mellanox.com>
10936 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
10937 block, and do not negate it, the stored id is already negative.
10939 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10941 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
10943 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10946 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
10947 masks of options that should be turned off if the VSX vector
10948 options are turned off.
10949 (OTHER_P8_VECTOR_MASKS): Likewise.
10950 (OTHER_VSX_VECTOR_MASKS): Likewise.
10951 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
10952 rs6000_disable_incompatible_switches to validate no type switches
10954 (rs6000_incompatible_switch): New function to disallow turning on
10955 other vector options if -mno-vsx, -mno-power8-vector, or
10956 -mno-power9-vector are specified.
10958 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10960 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
10962 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10964 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
10965 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
10966 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
10967 (ARG_POINTER_CFA_OFFSET): Likewise.
10969 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10971 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
10972 conditions to take advantage of various optimizations.
10974 2017-04-13 Jeff Law <law@redhat.com>
10976 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10977 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10978 (zero_extendsidi2_dext): Likewise.
10980 2017-04-13 Jakub Jelinek <jakub@redhat.com>
10983 * fold-const.c (fold_ternary_loc): Revert
10984 use op0 instead of fold_convert_loc (loc, type, arg0) part of
10987 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
10989 PR rtl-optimization/80343
10990 * lra-remat.c (update_scratch_ops): Assign original hard reg to
10991 new scratch pseudo.
10993 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
10996 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
10997 to ubsan_encode_value.
10999 2017-04-13 Jeff Law <law@redhat.com>
11001 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11002 appearing in DEBUG_INSNs.
11004 2017-04-13 Martin Liska <mliska@suse.cz>
11006 PR gcov-profile/80413
11007 * gcov-io.c (gcov_write_string): Copy to buffer just when
11008 allocated size is greater than zero.
11010 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11013 * dwarf2out.c (decls_for_scope): Ignore declarations of
11014 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11016 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11019 * ipa-visibility.c (non_local_p): Fix typos.
11020 (localize_node): When localizing symbol in same comdat group,
11021 dissolve the group only when we know external symbols are going
11023 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11025 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11027 PR tree-optimization/79390
11028 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11029 order does not result in usable sequence, retry with reversed operand
11035 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11036 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11037 op0 instead of fold_convert_loc (loc, type, arg0).
11039 2017-04-12 Jeff Law <law@redhat.com>
11041 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11042 has a delay slot in the generated code.
11044 * config/cris/cris.md (cris_preferred_reload_class): Return
11045 GENNONACR_REGS rather than GENERAL_REGS.
11047 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11050 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11051 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11052 signedness of the result type.
11054 2017-04-12 Richard Biener <rguenther@suse.de>
11055 Jeff Law <law@redhat.com>
11057 PR tree-optimization/80359
11058 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11059 trim stores to TARGET_MEM_REFs.
11061 2017-04-12 Richard Biener <rguenther@suse.de>
11063 PR tree-optimization/79390
11064 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11065 threading case even more.
11067 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11070 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11071 for quad_address_p for TImode, instead of just not indexed_address.
11073 2017-04-12 Richard Biener <rguenther@suse.de>
11074 Bernd Edlinger <bernd.edlinger@hotmail.de>
11076 PR middle-end/79671
11077 * alias.c (component_uses_parent_alias_set_from): Handle
11078 TYPE_TYPELESS_STORAGE.
11079 (get_alias_set): Likewise.
11080 * tree-core.h (tree_type_common): Add typeless_storage flag.
11081 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11082 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11083 for types containing members with TYPE_TYPELESS_STORAGE.
11084 (place_field): Likewise.
11085 (layout_type): Likewise for ARRAY_TYPE.
11086 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11087 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11088 TYPE_TYPELESS_STORAGE.
11089 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11091 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11094 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11095 first argument to type.
11097 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11101 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11102 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11103 (rs6000_expand_binop_builtin): Likewise.
11104 (rs6000_expand_ternop_builtin): Likewise; also add missing
11105 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11106 vshasigma built-ins.
11107 * doc/extend.texi: Document that vec_xxpermdi's third argument
11108 must be a constant.
11110 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11112 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11113 Use shift_const cost parameter when calculating gain of STV shifts.
11115 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11117 PR rtl-optimization/70478
11118 * lra-constraints.c (process_alt_operands): Check memory for
11119 disfavoring memory insn operand.
11121 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11123 PR middle-end/80100
11124 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11125 left shift in unsigned HOST_WIDE_INT type.
11127 PR rtl-optimization/80385
11128 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11129 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11132 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11133 if they have any depend clauses.
11135 2017-04-11 Martin Liska <mliska@suse.cz>
11138 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11139 * ipa-split.c (split_function): Create a local comdat symbol
11140 if caller is in a comdat group.
11142 2017-04-11 Martin Liska <mliska@suse.cz>
11145 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11148 2017-04-11 Martin Sebor <msebor@redhat.com>
11150 PR middle-end/80364
11151 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11152 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11154 (directive::set_width, directive::set_precision, format_character):
11156 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11159 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11162 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11163 conflict, set target->arch_name instead of target->cpu_name.
11165 2017-04-11 Richard Biener <rguenther@suse.de>
11167 PR tree-optimization/80374
11168 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11169 build_zero_cst, remove fold_convertible_p check again.
11171 2017-04-11 Martin Liska <mliska@suse.cz>
11174 * ubsan.c (instrument_object_size): Do not instrument register
11177 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11180 * config/i386/i386-builtin-types.def
11181 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11182 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11183 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11184 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11185 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11186 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11187 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11188 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11189 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11190 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11191 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11192 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11193 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11194 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11195 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11196 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11197 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11198 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11199 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11200 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11201 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11202 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11203 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11204 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11205 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11206 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11207 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11208 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11209 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11210 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11211 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11212 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11213 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11214 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11215 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11216 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11217 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11218 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11219 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11220 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11221 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11222 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11223 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11224 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11225 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11226 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11228 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11229 flag to second_arg_count, handle 4 argument function type _COUNT
11230 aliases, handle second_arg_count on second argument rather than last.
11232 2017-04-10 Jeff Law <law@redhat.com>
11234 PR tree-optimization/80374
11235 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11236 record anything if we can not convert integer_zero_node to the
11239 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11242 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11243 Enhance special handling given to the TARGET_P9_MINMAX option in
11244 relation to certain other options.
11246 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11248 PR tree-optimization/80153
11249 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11250 remove POINTER_PLUS_EXPR's base part directly, rather than through
11253 2017-04-10 Richard Biener <rguenther@suse.de>
11254 Bin Cheng <bin.cheng@arm.com>
11256 PR tree-optimization/80153
11257 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11258 the first element of pointer type aff_tree. Build result expr in
11260 (add_elt_to_tree): Convert to type unconditionally. Remove other
11261 fold_convert calls.
11262 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11263 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11265 2017-04-10 Richard Biener <rguenther@suse.de>
11267 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11270 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11272 PR rtl-optimization/70478
11273 * lra-constraints.c (curr_small_class_check): New.
11274 (update_and_check_small_class_inputs): New.
11275 (process_alt_operands): Update curr_small_class_check. Disfavor
11276 alternative insn memory operands. Check available regs for small
11279 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11282 * config/mips/mips.opt (-mvirt): Update description.
11283 * doc/invoke.texi (-mvirt): Likewise.
11285 2017-04-10 Richard Biener <rguenther@suse.de>
11287 PR middle-end/80362
11288 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11289 looking for NEGATE_EXPR in -A / -B to A / B folding.
11291 2017-04-10 Martin Liska <mliska@suse.cz>
11293 PR gcov-profile/80224
11294 * gcov.c (print_usage): Fix usage string.
11295 (get_gcov_intermediate_filename): Remove.
11296 (output_gcov_file): Use both for normal and intermediate format.
11297 (generate_results): Do not initialize special file for
11298 intermediate format.
11300 2017-04-10 Richard Biener <rguenther@suse.de>
11302 PR tree-optimization/80304
11303 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11306 2017-04-10 Nathan Sidwell <nathan@acm.org>
11309 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11310 (rs6000_init_builtins): Use it.
11312 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11314 * config/arm/arm.md (<mrc>): Add mode to SET source.
11315 (<mrrc>): Likewise.
11317 2017-04-10 Richard Biener <rguenther@suse.de>
11319 PR middle-end/80344
11320 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11322 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11325 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11326 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11327 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11328 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11329 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11330 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11331 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11332 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11333 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11334 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11335 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11336 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11337 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11338 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11339 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11340 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11341 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11342 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11343 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11344 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11345 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11346 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11347 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11349 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11351 PR rtl-optimization/70478
11352 * lra-constraints.c: Reverse the last patch.
11354 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11356 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11357 Add comment for WCHAR_T.
11359 2017-04-08 Martin Liska <mliska@suse.cz>
11362 2017-04-07 Martin Liska <mliska@suse.cz>
11365 * ipa-split.c (split_function): Add function part to a same comdat
11368 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11371 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11373 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11375 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11376 (vec_pairing): ...to this.
11377 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11378 (rs6000_sched_init): Adjust for name change.
11379 (struct rs6000_sched_context): Likewise.
11380 (rs6000_init_sched_context): Likewise.
11381 (rs6000_set_sched_context): Likewise.
11383 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11389 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11391 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11392 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11393 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11395 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11397 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11399 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11401 PR rtl-optimization/70703
11402 * ira-color.c (update_conflict_hard_regno_costs): Use
11403 int64_t instead of HOST_WIDE_INT.
11405 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11407 PR rtl-optimization/70478
11408 * lra-constraints.c (process_alt_operands): Disfavor alternative
11409 insn memory operands.
11411 2017-04-07 Jeff Law <law@redhat.com>
11413 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11414 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11416 2017-04-07 Martin Liska <mliska@suse.cz>
11419 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11420 Show error message instead of an ICE.
11422 2017-04-07 Martin Liska <mliska@suse.cz>
11425 * ipa-split.c (split_function): Add function part to a same comdat
11428 2017-04-07 Richard Biener <rguenther@suse.de>
11430 PR middle-end/80341
11431 * tree.c (get_unwidened): Also handle ! for_type case for
11433 * convert.c (do_narrow): Split out from ...
11434 (convert_to_integer_1): ... here. Do not pass final truncation
11435 type to get_unwidened for TRUNC_DIV_EXPR.
11437 2017-04-07 Richard Biener <rguenther@suse.de>
11439 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11441 (aff_combination_const): Adjust.
11442 (aff_combination_scale): Likewise.
11443 (aff_combination_add_elt): Likewise.
11444 (aff_combination_add_cst): Likewise.
11445 (aff_combination_convert): Likewise.
11446 (add_elt_to_tree): Likewise. Remove unused argument.
11447 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11449 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11451 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11453 * config/arm/arm.c (arm_default_short_enums): Use
11454 ARM_DEFAULT_SHORT_ENUMS.
11455 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11457 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11460 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11461 members with redundant out-of-class redeclaration.
11463 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11466 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11467 * config/i386/i386.md (*zero_extendsidi2):
11468 Add (?*x,*x) and (?*v,*v) alternatives.
11470 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11473 * config/i386/i386.c (ix86_expand_builtin)
11474 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11475 mode from insn data. Convert operands to insn operand mode.
11476 Copy operands that don't satisfy insn predicate to a register.
11478 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11480 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11483 2017-04-06 Richard Biener <rguenther@suse.de>
11485 PR tree-optimization/80334
11486 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11487 preserve alignment of accesses.
11489 2017-04-06 Richard Biener <rguenther@suse.de>
11491 PR tree-optimization/80262
11492 * tree-sra.c (build_ref_for_offset): Preserve address-space
11494 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11495 Drop useless address-space information on MEM_REF offsets.
11497 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11499 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11501 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11503 PR rtl-optimization/70703
11504 * ira-color.c (update_conflict_hard_regno_costs): Use
11505 HOST_WIDE_INT instead of long.
11507 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11510 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11511 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11513 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11514 for x86_64 target. Handle -m3dnowa option.
11516 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11518 PR rtl-optimization/70703
11519 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11520 (update_conflict_hard_regno_costs): Use long instead of unsigned
11521 arithmetic for cost calculation.
11523 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11524 Bernd Edlinger <bernd.edlinger@hotmail.de>
11527 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11530 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11533 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11534 ptr_mode with Pmode throughout.
11535 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11536 into probe_stack_range and use DImode.
11538 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11541 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11542 call_eh_return is true.
11544 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11546 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11547 Initialize last_match_fntype_index.
11549 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11552 * tree-nvr.c: Include internal-fn.h.
11553 (pass_return_slot::execute): Ignore internal calls without
11556 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11557 Richard Biener <rguenther@suse.de>
11560 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11561 captures used multiple times, except for the last use.
11562 * generic-match-head.c: Include gimplify.h.
11564 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11566 PR tree-optimization/79390
11567 * target.h (struct noce_if_info): Declare.
11568 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11569 * target.def (noce_conversion_profitable_p): New target hook.
11570 * ifcvt.h (struct noce_if_info): New type, moved from ...
11571 * ifcvt.c (struct noce_if_info): ... here.
11572 (noce_conversion_profitable_p): Renamed to ...
11573 (default_noce_conversion_profitable_p): ... this. No longer
11575 (noce_try_store_flag_constants, noce_try_addcc,
11576 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11577 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11578 instead of noce_conversion_profitable_p.
11579 * config/i386/i386.c: Include ifcvt.h.
11580 (ix86_option_override_internal): Don't override
11581 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11582 (ix86_noce_conversion_profitable_p): New function.
11583 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11584 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11585 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11586 * doc/tm.texi: Regenerated.
11588 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11590 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11593 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11596 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11597 instructions for small multiply cores.
11599 2017-04-04 Jeff Law <law@redhat.com>
11601 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11603 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11604 that are not set by the loop over the elements.
11606 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11609 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11610 int mode, convert_modes it to mode as unsigned, otherwise use
11611 lowpart_subreg to mode rather than SImode.
11612 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11613 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11614 Use DImode instead of SImode for the shift count operand.
11615 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11618 2017-04-04 Richard Biener <rguenther@suse.de>
11620 PR middle-end/80281
11621 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11622 arithmetic done for the negate or the plus. Simplify.
11623 (A - (-B) -> A + B): Likewise.
11624 * fold-const.c (split_tree): Make sure to not negate pointers.
11626 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11628 PR rtl-optimization/60818
11629 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11630 a compare of comparisons with the thing compared if this results
11631 in a different machine mode.
11633 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11635 * alias.c (base_alias_check): Fix typo in comment.
11636 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11637 * cgraphunit.c (symbol_table::compile): Likewise.
11638 * collect2.c (maybe_run_lto_and_relink): Likewise.
11639 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11640 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11641 * config/avr/avr.c (avr_map_op_t): Likewise.
11642 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11643 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11644 * config/epiphany/epiphany.md (movcc): Likewise.
11645 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11646 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11648 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11649 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11650 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11651 * config/sh/sh.c (sh_rtx_costs): Likewise.
11652 * fold-const.c (fold_truth_andor): Likewise.
11653 * genautomata.c (collapse_flag): Likewise.
11654 * gengtype.h (struct type::u::s): Likewise.
11655 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11656 * input.c (FORMAT_AMOUNT): Likewise.
11657 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11658 (known_aggs_to_agg_replacement_list): Likewise.
11659 * ipa-inline-analysis.c: Likewise.
11660 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11661 * ipa-polymorphic-call.c
11662 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11663 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11664 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11666 * modulo-sched.c (apply_reg_moves): Likewise.
11667 * omp-expand.c (build_omp_regions_1): Likewise.
11668 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11669 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11670 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11671 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11672 * value-prof.c: Likewise.
11673 * var-tracking.c (val_reset): Likewise.
11675 2017-04-03 Richard Biener <rguenther@suse.de>
11677 PR tree-optimization/80275
11678 * fold-const.c (split_address_to_core_and_offset): Handle
11681 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11683 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11684 descriptors is at least equal to that of functions.
11686 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11688 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11690 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11693 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11694 (mov<IMOD4:mode>): New expander.
11695 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11697 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11699 PR rtl-optimization/79405
11700 * fwprop.c (propagations_left): New variable.
11701 (forward_propagate_into): Decrement it.
11702 (fwprop_init): Initialize it.
11703 (fw_prop): If the variable has reached zero, stop propagating.
11704 (fwprop_addr): Ditto.
11706 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11709 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11710 a FUNCTION_DECL, pass it as decl instead of origin to
11713 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11715 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11718 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11721 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11722 TARGET_VSX_SMALL_INTEGER.
11724 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11726 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11727 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11729 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11731 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11732 extraction from odd-numbered MSA register.
11734 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11736 PR middle-end/80173
11737 * expmed.c (store_bit_field_1): Don't attempt to create
11738 a word subreg out of hard registers wider than word if they
11739 have HARD_REGNO_NREGS of 1 for their mode.
11741 PR middle-end/80163
11742 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11743 conversions to integer types wider than word and pointer.
11746 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11747 (rtx_equal_for_cselib_p): Pass 0 to it.
11748 * cselib.c (cselib_hasher::equal): Likewise.
11749 (rtx_equal_for_cselib_1): Add depth argument. If depth
11750 is 128, don't look up VALUE locs and punt. Increment
11751 depth in recursive calls when walking VALUE locs.
11753 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11755 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11756 (make_gcov_file_name): Use the canonical path name for generating
11758 (read_line): Fix handling of files with ascii null bytes.
11760 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11762 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11763 to initialise a vector register instead
11764 of using a const_int.
11766 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11768 PR translation/80189
11769 * gimplify.c (omp_default_clause): Use %qs instead of %s in
11770 diagnostic messages.
11772 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
11775 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
11776 (dfp_diex_<mode>): Update mode of operand 1.
11777 * doc/extend.texi (dxex, dxexq): Document change to return type.
11778 (diex, diexq): Document change to argument type.
11780 2017-03-30 Martin Jambor <mjambor@suse.cz>
11783 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
11784 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
11785 it reflects the signature changes performed at the callee side.
11786 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
11787 to cgraph_build_function_type_skip_args.
11788 (build_function_decl_skip_args): Adjust call to the above function.
11790 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11793 * config/i386/sse.md
11794 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
11795 register as dest whenever it is a MEM not rtx_equal_p to the
11796 corresponding dup operand, and when forcing into reg move the
11797 reg into the memory afterwards.
11798 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
11799 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
11800 for the force_reg mode.
11801 (avx512vl_vextractf128<mode>): Use register as dest either
11802 always when a MEM, or when it is a MEM not rtx_equal_p to the
11803 corresponding dup operand, or even not when it is a CONST_VECTOR
11804 depending on the mode and lo vs. hi.
11805 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
11807 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
11808 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
11809 Likewise. Require that operands[2] is even.
11810 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
11811 Remove extraneous parens. Require that operands[2] is a multiple
11813 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
11814 operands[0] is a MEM if <mask_applied>, the predicates/constraints
11815 disallow memory then.
11817 2017-03-30 Richard Biener <rguenther@suse.de>
11819 PR tree-optimization/77498
11820 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
11821 to non-constants over backedges.
11823 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
11825 PR rtl-optimization/80233
11826 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
11827 as last_combined_insn. Do not test for BARRIER_P separately.
11829 2017-03-29 Andreas Schwab <schwab@suse.de>
11832 * calls.c (prepare_call_address): Convert funexp to Pmode before
11833 copying to temp reg.
11835 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11837 PR tree-optimization/80158
11838 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
11839 Handle possible future case of more than one alternate
11841 (replace_rhs_if_not_dup): Likewise.
11842 (replace_one_candidate): Likewise.
11844 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
11846 PR rtl-optimization/80193
11847 * ira.c (ira): Do not check allocation for LRA.
11849 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
11851 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
11852 (nvptx_output_simt_exit): Declare.
11853 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
11854 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
11855 (init_softstack_frame): Move initialization of crtl->is_leaf to...
11856 (nvptx_declare_function_name): ...here. Emit declaration of local
11857 memory space buffer for omp_simt_enter insn.
11858 (nvptx_output_unisimt_switch): New.
11859 (nvptx_output_softstack_switch): New.
11860 (nvptx_output_simt_enter): New.
11861 (nvptx_output_simt_exit): New.
11862 * config/nvptx/nvptx.h (struct machine_function): New fields
11863 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
11864 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
11865 (UNSPECV_SIMT_EXIT): Ditto.
11866 (omp_simt_enter_insn): New insn.
11867 (omp_simt_enter): New expansion.
11868 (omp_simt_exit): New insn.
11869 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
11871 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
11872 (expand_GOMP_SIMT_ENTER_ALLOC): New.
11873 (expand_GOMP_SIMT_EXIT): New.
11874 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
11875 (GOMP_SIMT_ENTER_ALLOC): Ditto.
11876 (GOMP_SIMT_EXIT): Ditto.
11877 * target-insns.def (omp_simt_enter): New insn.
11878 (omp_simt_exit): Ditto.
11879 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
11881 (lower_rec_simd_input_clauses): Implement SIMT privatization.
11882 (lower_rec_input_clauses): Likewise.
11883 (lower_lastprivate_clauses): Handle SIMT privatization.
11885 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
11886 (ompdevlow_adjust_simt_enter): New.
11887 (find_simtpriv_var_op): New.
11888 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
11889 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
11891 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
11892 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
11893 (copy_decl_for_dup_finish): Ditto.
11895 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
11897 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
11900 * config/i386/i386.c (ix86_option_override_internal): Always
11901 allow -mpreferred-stack-boundary=3 for 64-bit targets.
11903 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11905 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
11907 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11909 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
11910 mark new edge's irreducible flag accordign to it.
11911 (vect_do_peeling): Check loop preheader edge's irreducible flag
11912 and pass it to function slpeel_add_loop_guard.
11914 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
11916 PR tree-optimization/80218
11917 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
11918 Update block frequencies and counts.
11920 2017-03-28 Richard Biener <rguenther@suse.de>
11922 PR tree-optimization/78644
11923 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
11924 of a simplification result we may not use it at all.
11926 2017-03-28 Richard Biener <rguenther@suse.de>
11929 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
11930 without arguments, generate default definition of a SSA name.
11932 2017-03-28 Richard Biener <rguenther@suse.de>
11934 PR middle-end/80222
11935 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
11936 TYPE_REF_CAN_ALIAS_ALL references.
11937 * fold-const.c (fold_indirect_ref_1): Likewise.
11939 2017-03-28 Martin Liska <mliska@suse.cz>
11942 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
11943 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
11945 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11946 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
11948 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
11949 (EXTRA_SPECS): Define.
11950 (SUBTARGET_EXTRA_SPECS): Likewise.
11951 (SUBTARGET_CPP_SPEC): Likewise.
11952 * config/arc/elf.h (EXTRA_SPECS): Renamed to
11953 SUBTARGET_EXTRA_SPECS.
11954 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
11956 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11958 * config/arc/simdext.md (vst64_insn): Update pattern.
11959 (vld32wh_insn): Likewise.
11960 (vld32wl_insn): Likewise.
11961 (vld64_insn): Likewise.
11962 (vld32_insn): Likewise.
11964 2017-03-28 Marek Polacek <polacek@redhat.com>
11967 * fold-const.c (fold_comparison): Use protected_set_expr_location
11968 instead of SET_EXPR_LOCATION.
11970 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
11972 * tree.c (add_expr): Avoid name lookup warning.
11974 2017-03-27 Jeff Law <law@redhat.com>
11976 PR tree-optimization/80216
11977 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
11978 function name. Limit recursion depth.
11979 (record_temporary_equivalences): Corresponding changes.
11981 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
11983 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
11986 2017-03-27 Jakub Jelinek <jakub@redhat.com>
11989 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
11991 * cfgcleanup.c (reg_note_cfa_p): New array.
11992 (insns_have_identical_cfa_notes): New function.
11993 (old_insns_match_p): Don't cross-jump in between /f
11994 and non-/f instructions. If both i1 and i2 are frame related,
11995 verify all CFA notes, their order and content.
11997 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12000 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12001 HImode and SImode with zero extend to DImode to one insn.
12002 (bswap<mode>2_extenddi): Likewise.
12003 (bswapsi2_extenddi): Likewise.
12004 (bswaphi2_extendsi): Likewise.
12005 (bswaphi2): Combine bswap HImode and SImode into one insn.
12006 Separate memory insns from swapping register.
12007 (bswapsi2): Likewise.
12008 (bswap<mode>2): Likewise.
12009 (bswaphi2_internal): Delete, no longer used.
12010 (bswapsi2_internal): Likewise.
12011 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12012 store, and gpr<-gpr swap insns.
12013 (bswap<mode>2_store): Likewise.
12014 (bswaphi2_reg): Register only splitter, combine with the splitter.
12015 (bswaphi2 splitter): Likewise.
12016 (bswapsi2_reg): Likewise.
12017 (bswapsi2 splitter): Likewise.
12018 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12019 the insns into load, store, and register/register insns.
12020 (bswapdi2_ldbrx): Likewise.
12021 (bswapdi2_load): Likewise.
12022 (bswapdi2_store): Likewise.
12023 (bswapdi2_reg): Likewise.
12025 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12027 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12028 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12030 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12033 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12035 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12036 special handling for target option conflicts between dform
12037 options (-mpower9-dform, -mpower9-dform-vector,
12038 -mpower9-dform-scalar) and -mno-direct-move.
12040 2017-03-27 Richard Biener <rguenther@suse.de>
12042 PR tree-optimization/80181
12043 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12045 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12047 * config/arc/predicates.md (move_double_src_operand): Replace the
12048 call to move_double_src_operand with a call to address_operand.
12050 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12052 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12053 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12054 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12056 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12058 * config/arc/predicates.md (long_immediate_loadstore_operand):
12059 Consider scaled addresses cases.
12061 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12063 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12064 restored when in interrupt.
12065 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12066 doesn't have delay slot.
12068 2017-03-27 Richard Biener <rguenther@suse.de>
12071 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12072 inlined thunk clones.
12074 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12077 * asan.c (instrument_derefs): Copy over last operand from
12078 original COMPONENT_REF to the new COMPONENT_REF with
12079 DECL_BIT_FIELD_REPRESENTATIVE.
12080 * ubsan.c (instrument_object_size): Likewise.
12082 2017-03-27 Richard Biener <rguenther@suse.de>
12084 PR tree-optimization/80170
12085 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12086 sure DR/SCEV didnt fold in constants we do not see when looking
12087 at the reference base alignment.
12089 2017-03-27 Richard Biener <rguenther@suse.de>
12091 PR middle-end/80171
12092 * gimple-fold.c (fold_ctor_reference): Properly guard against
12093 NULL return value from canonicalize_constructor_val.
12095 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12098 * config/i386/i386.c (ix86_expand_builtin)
12099 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12100 flags reg setting and flags reg using instructions.
12101 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12102 clobbering instructions to zero extend op2.
12104 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12106 * doc/install.texi (Configuration) <--with-aix-soname>:
12107 Update link to AIX ld.
12109 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12111 PR rtl-optimization/80160
12112 PR rtl-optimization/80159
12113 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12114 reg_alternate_class into account.
12116 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12119 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12120 to consider in curr_insn_transform.
12122 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12124 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12125 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12126 and emit_mode_inner.
12128 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12130 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12131 argument to the overloaded builtin variants. Use the new flag to
12132 deprecate certain builtin variants.
12133 * config/s390/s390-builtin-types.def: Add new builtin types.
12134 * config/s390/s390-builtins.h: Support new flags field for
12135 overloaded builtins.
12136 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12137 (s390_macro_to_expand): Enable vector float data type.
12138 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12139 builtins by incrementing the __VEC__ version number.
12140 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12142 (s390_resolve_overloaded_builtin): Emit error messages depending
12143 on the builtin flags.
12144 * config/s390/s390.c (s390_expand_builtin): Support additional
12145 flags argument. Change error message to match the messages
12146 emitted in s390-c.c.
12147 * config/s390/s390.md: New UNSPEC_* constants.
12148 (op_type): Add new instruction types.
12149 * config/s390/vecintrin.h: Add new builtins and test data class
12151 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12152 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12153 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12154 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12155 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12156 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12158 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12159 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12160 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12161 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12163 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12164 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12165 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12166 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12168 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12169 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12170 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12171 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12172 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12173 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12174 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12176 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12177 ("vec_scatter_element<V_HW_4:mode>_DI")
12178 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12179 ("vec_fpint<mode>", "vflls")
12180 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12181 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12182 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12183 ("*vec_cmphe<mode>_cc"): ... these.
12185 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12186 mode constant instead of magic value.
12188 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12190 * config/s390/s390.c (s390_expand_vec_compare): Support other
12191 vector floating point modes than just V2DF.
12192 (s390_expand_vcond): Likewise.
12193 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12194 (s390_cannot_change_mode_class): Prevent mode changes between TF
12195 and V1TF in vector registers.
12196 * config/s390/s390.md (DF, SF): New mode attributes.
12197 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12198 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12199 SFmode support for VRs.
12200 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12202 (VFT, VF_HW): New mode iterators.
12203 (vw, sdx): New mode attributes.
12204 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12205 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12206 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12207 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12208 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12209 also the new vector floating point modes. Renaming to ...
12211 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12212 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12213 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12214 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12215 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12216 ("vec_unordered<mode>"): ... these.
12218 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12219 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12220 ("*vec_extendv2df"): New insn definitions.
12222 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12224 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12225 ("mulditi3_2", "*muldi3_sign"): New patterns.
12226 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12227 rename the pattern definition.
12229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12231 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12233 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12235 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12237 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12238 instruction if possible.
12239 * config/s390/vector.md (vec_halfnumelts): New mode
12241 ("*vec_vllezlf<mode>"): New pattern.
12243 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12245 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12246 ("popcountv4si2", "popcountv2di2"): Rename to ...
12247 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12248 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12250 ("popcount<mode>2_vxe"): New pattern.
12252 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12254 * common/config/s390/s390-common.c (processor_flags_table): Add
12256 * config.gcc: Add arch12.
12257 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12258 Default to arch12 for unknown CPU model numbers.
12259 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12260 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12261 PROCESSOR_max sanity check.
12262 * config/s390/s390-opts.h (enum processor_type): Add
12264 * config/s390/s390.c (processor_table): Add arch12.
12265 (s390_expand_builtin): Add check for B_VXE flag.
12266 (s390_issue_rate): Add PROCESSOR_ARCH12.
12267 (s390_get_sched_attrmask): Likewise.
12268 (s390_get_unit_mask): Likewise.
12269 (s390_sched_score): Enable z13 scheduling for arch12.
12270 (s390_sched_reorder): Likewise.
12271 (s390_sched_variable_issue): Likewise.
12272 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12274 (s390_tune_attr): Use z13 scheduling also for arch12.
12275 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12276 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12277 (TARGET_VXE_P): New macros.
12278 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12279 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12280 * config/s390/s390.opt: Add arch12 as processor_type.
12282 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12284 * config/s390/s390.md
12285 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12286 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12287 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12289 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12290 Rename expanders to ...
12292 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12293 ("fixuns_truncdddi2_emu"): ... these.
12295 ("fixuns_trunc<mode>si2_emu"): New expander.
12297 ("*fixuns_truncdfdi2_z13"): Rename to ...
12298 ("*fixuns_truncdfdi2_vx"): ... this.
12300 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12302 * config/s390/2964.md: Remove the single element vector compare
12303 instructions which are no longer used.
12304 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12306 (s390_canonicalize_comparison): Remove handling of DFmode
12308 (s390_expand_vec_compare_scalar): Remove function.
12309 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12310 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12312 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12314 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12316 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12318 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12319 will anyway by matched by mov<mode>_64dfp.
12321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12323 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12324 vlef/vstef. Add missing operand to vleif.
12326 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12328 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12329 pair for all vector types with 64 bit elements.
12330 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12331 * config/s390/vector.md (V_HW_64): ... here.
12332 (V_128_NOSINGLE): New mode iterator.
12333 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12334 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12335 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12336 ("*vec_load_pairv2di"): Change to ...
12337 ("*vec_load_pair<mode>"): ... this one.
12339 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12341 * config/s390/constraints.md: Add comments.
12342 (jKK): Reject element sizes > 8 bytes.
12343 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12345 * config/s390/s390.md: Add the s_operand checks formerly in
12346 s390_split_ok_p to various splitters where they are still
12348 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12349 for 128 bit vectors. Plus two splitters.
12351 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12353 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12356 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12359 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12360 error if the boundary argument is not constant.
12362 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12364 PR rtl-optimization/80112
12365 * loop-doloop.c (doloop_condition_get): Don't check condition
12366 if cmp isn't SET with IF_THEN_ELSE src.
12368 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12370 PR tree-optimization/80158
12371 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12372 replacing a candidate statement, also replace it for the
12373 candidate's alternate interpretation.
12374 (replace_rhs_if_not_dup): Likewise.
12375 (replace_one_candidate): Likewise.
12377 2017-03-24 Richard Biener <rguenther@suse.de>
12379 PR tree-optimization/80167
12380 * graphite-isl-ast-to-gimple.c
12381 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12383 (translate_isl_ast_to_gimple::get_rename): Likewise.
12385 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12387 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12388 handling of certain combinations of target options, including the
12389 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12390 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12392 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12395 * config/arm/arm.md (*load_multiple): Add reload_completed to
12396 matching condition.
12398 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12399 Richard Biener <rguenth@suse.de>
12401 PR tree-optimization/79908
12402 PR tree-optimization/80136
12403 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12404 been cast away, gimplify_and_add suffices.
12406 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12408 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12410 2017-03-23 Richard Biener <rguenther@suse.de>
12412 PR tree-optimization/80032
12413 * gimplify.c (gimple_push_cleanup): Forced unconditional
12414 cleanups still have to go to the conditional_cleanups
12417 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12419 PR tree-optimization/80072
12420 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12422 (next_operand_entry_id): Change type to unsigned int.
12423 (sort_by_operand_rank): Make sure to return the right return value
12424 even if unsigned fields are bigger than INT_MAX.
12425 (struct oecount): Change cnt and id type to unsigned int.
12426 (oecount_hasher::equal): Formatting fix.
12427 (oecount_cmp): Make sure to return the right return value
12428 even if unsigned fields are bigger than INT_MAX.
12429 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12432 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12433 TREE_READONLY on result if writing it more than once.
12436 * doc/invoke.texi (-fsanitize=thread): Document that with
12437 -fnon-call-exceptions atomics are not able to throw
12441 * tsan.c: Include tree-eh.h.
12442 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12443 maybe_clean_or_replace_eh_stmt where needed.
12444 (instrument_memory_accesses): Add cfg_changed argument.
12445 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12446 if it returned true.
12447 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12449 PR rtl-optimization/63191
12450 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12451 wrapper function, moved the whole old content into ...
12452 (ix86_delegitimize_address_1): ... this. New inline function.
12453 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12454 true as last argument instead of ix86_delegitimize_address.
12456 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12458 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12459 cortexa57_branch_cost.
12461 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12463 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12465 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12468 * doc/md.texi (Constraints): Document wA constraint.
12469 * config/rs6000/constraints.md (wA): New.
12470 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12471 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12472 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12473 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12475 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12478 * gimplify.c (is_oacc_declared): New function.
12479 (oacc_default_clause): Use it to set default flags for acc declared
12480 variables inside parallel regions.
12481 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12482 declared variables.
12483 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12484 declare attribute to any decl as necessary.
12486 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12489 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12490 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12491 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12492 (arm_arch_lpae): This.
12493 * config/arm/arm.c (arm_arch7ve): Rename into ...
12494 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12495 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12498 2017-03-22 Martin Liska <mliska@suse.cz>
12501 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12502 error message instead of an ICE.
12504 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12506 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12508 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12510 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12512 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12515 2017-03-21 Martin Sebor <msebor@redhat.com>
12517 * doc/extend.texi: Use "cannot" instead of "can't."
12518 * doc/hostconfig.texi: Same.
12519 * doc/install.texi: Same.
12520 * doc/invoke.texi: Same.
12521 * doc/loop.texi: Same.
12522 * doc/md.texi: Same.
12523 * doc/objc.texi: Same.
12524 * doc/rtl.texi: Same.
12525 * doc/tm.texi: Same.
12526 * doc/tm.texi.in: Same.
12527 * doc/trouble.texi: Same.
12529 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12532 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12533 (collect_checksum_attributes): Set it.
12534 (die_checksum_ordered): Use it.
12536 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12538 PR tree-optimization/79908
12539 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12540 change: For a VA_ARG whose LHS has been cast away, use
12541 force_gimple_operand to construct the side effects.
12543 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12545 PR translation/80001
12546 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12547 more amenable to translation.
12548 (oacc_loop_auto_partitions): Likewise.
12550 2017-03-21 Marek Polacek <polacek@redhat.com>
12551 Martin Sebor <msebor@redhat.com>
12553 PR tree-optimization/80109
12554 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12555 on INTEGRAL_TYPE_P.
12557 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12558 Segher Boessenkool <segher@kernel.crashing.org>
12561 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12562 check reg_used_between_p between insn and one of succ or succ2
12563 depending on if succ is artificial insn not inserted into insn
12566 2017-03-21 Martin Liska <mliska@suse.cz>
12568 PR gcov-profile/80081
12569 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12570 * doc/gcc.texi: Include gcov-dump stuff.
12571 * doc/gcov-dump.texi: New file.
12573 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12575 PR rtl-optimization/79150
12576 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12577 conditional jump, if the jump is the last insn of the loop.
12579 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12580 Richard Biener <rguenth@suse.de>
12582 PR tree-optimization/79908
12583 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12584 been cast away, use force_gimple_operand to construct the side
12587 2017-03-21 Martin Liska <mliska@suse.cz>
12589 PR libfortran/79956
12590 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12593 2017-03-21 Brad Spengler <spender@grsecurity.net>
12596 * plugin.c (htab_hash_plugin): New function.
12597 (add_new_plugin): Use it and adjust.
12598 (parse_plugin_arg_opt): Adjust.
12599 (init_one_plugin): Likewise.
12601 2017-03-21 Richard Biener <rguenther@suse.de>
12603 PR tree-optimization/80032
12604 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12605 if set force the cleanup to happen unconditionally.
12606 (gimplify_target_expr): Push inserted clobbers with force_uncond
12607 to avoid them being removed by control-dependent DCE.
12609 2017-03-21 Richard Biener <rguenther@suse.de>
12611 PR tree-optimization/80122
12612 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12613 va_arg_pack_len when the inlined call stmt requires pack
12615 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12617 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12620 * tsan.c (instrument_builtin_call): If the memory model argument
12621 is not a constant, assume it is valid.
12624 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12627 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12629 PR rtl-optimization/79910
12630 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12631 if its dest is used by an insn before I2 (other than the combined
12632 insns themselves, which are properly handled already).
12634 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12637 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12639 * combine.c (record_used_regs): New static function.
12640 (try_combine): Handle situations where there is an additional
12641 instruction between I2 and I3 which needs to have a LOG_LINK
12645 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12647 * combine.c (try_combine): Delete redundant i1 test. Call
12648 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12650 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12653 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12654 alternatives 13/14.
12656 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12658 PR tree-optimization/80054
12659 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12660 the optimization if a PHI or any of its arguments is not dominated
12661 by the candidate's basis. Use gphi* rather than gimple* as
12663 (replace_profitable_candidates): Clean up a gimple* variable that
12664 should be a gphi* variable.
12666 2017-03-20 Martin Sebor <msebor@redhat.com>
12669 * doc/extend.texi (attribute constructor): Document present limitation.
12671 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12674 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12675 __POWER9_VECTOR__ #ifdef control, change template definition to
12676 use Power9-specific built-in function.
12677 (vec_any_eq): Likewise.
12678 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12679 to control outcomes from this test.
12680 (vector_ae_<mode>p): For VEC_F modes, likewise.
12682 2017-03-20 Ian Lance Taylor <iant@google.com>
12684 * config/i386/i386.c (ix86_function_regparm): Save an extra
12685 register for -fsplit-stack with DECL_STATIC_CHAIN.
12687 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12690 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12691 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12693 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12695 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12697 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12700 2017-03-20 Marek Polacek <polacek@redhat.com>
12703 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12705 2017-03-20 Richard Biener <rguenther@suse.de>
12707 PR tree-optimization/80113
12708 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12709 allocate extra SSA name for PHI def.
12710 (add_close_phis_to_outer_loops): Likewise.
12711 (add_close_phis_to_merge_points): Likewise.
12712 (copy_loop_close_phi_args): Likewise.
12713 (copy_cond_phi_nodes): Likewise.
12715 2017-03-20 Martin Liska <mliska@suse.cz>
12717 PR middle-end/79753
12718 * tree-chkp.c (chkp_build_returned_bound): Do not build
12719 returned bounds for a LHS that's not a BOUNDED_P type.
12721 2017-03-20 Martin Liska <mliska@suse.cz>
12725 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12726 COMPLEX_CST and VECTOR_CST.
12728 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12731 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12732 target operand. A new splitter adds the clobber statement in case
12733 the target operand is dead anyway.
12735 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12737 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12738 to age-old versions of binutils and glibc.
12740 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12742 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12744 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12746 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12748 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12750 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12751 requirement for binutils 2.13.
12753 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12755 * combine.c (try_combine): Delete redundant i1 test. Call
12756 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12758 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12760 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12761 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12763 <riscv64-*-elf>: Re-arrange section
12764 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12765 <riscv32-*-linux>: Likewise.
12766 <riscv64-*-elf>: Likewise
12767 <riscv64-*-linux>: Likewise.
12769 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
12772 * aarch64.opt(verbose-cost-dump): Fix typo.
12774 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
12777 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
12778 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
12780 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12782 * reload.c (find_reloads): When reloading a nonoffsettable address,
12783 use RELOAD_OTHER for it and its address reloads.
12785 PR rtl-optimization/79910
12786 * combine.c (record_used_regs): New static function.
12787 (try_combine): Handle situations where there is an additional
12788 instruction between I2 and I3 which needs to have a LOG_LINK
12791 2017-03-17 Jeff Law <law@redhat.com>
12793 PR tree-optimization/71437
12794 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
12795 conditional in the hash table first.
12796 (vrp_dom_walker::before_dom_children): Extract condition from
12797 ASSERT_EXPR. Record condition, its inverion and any implied
12798 conditions as well.
12800 2017-03-17 Marek Polacek <polacek@redhat.com>
12801 Markus Trippelsdorf <markus@trippelsdorf.de>
12803 PR tree-optimization/80079
12804 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
12807 2017-03-17 Richard Biener <rguenther@suse.de>
12809 PR middle-end/80075
12810 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
12811 Properly verify the LHS before the RHS possibly claims to be
12813 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
12816 2017-03-17 Martin Jambor <mjambor@suse.cz>
12818 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
12819 (List of -O2 options): Likewise.
12820 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
12823 2017-03-17 Tom de Vries <tom@codesourcery.com>
12825 * gcov-dump.c (print_usage): Print bug_report_url.
12827 2017-03-17 Richard Biener <rguenther@suse.de>
12829 PR middle-end/80050
12830 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
12831 (parser::peek): Likewise.
12833 2017-03-17 Richard Biener <rguenther@suse.de>
12835 PR tree-optimization/80048
12836 * sese.c (free_sese_info): Properly release rename_map and
12837 copied_bb_map elements.
12839 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
12841 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
12842 Add linked-list forward and backlinks. Insert on
12843 construction, remove on destruction.
12844 (class pass_store_merging): Add m_stores_head field.
12845 (pass_store_merging::terminate_and_process_all_chains):
12846 Iterate over m_stores_head list.
12847 (pass_store_merging::terminate_all_aliasing_chains):
12849 (pass_store_merging::execute): Check for debug stmts first.
12850 Push new chains onto the m_stores_head stack.
12852 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
12855 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
12856 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
12857 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
12859 2017-03-16 Jeff Law <law@redhat.com>
12861 PR tree-optimization/71437
12862 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
12863 member function. Implementation moved into after_dom_children
12864 member function and into the threader's thread_outgoing_edges
12866 (dom_opt_dom_walker::after_dom_children): Simplify by moving
12867 some code into new thread_outgoing_edges.
12868 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
12869 definition. Simplify marker handling (do it here). Assume we always
12870 have the available expression and the const/copies tables.
12871 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
12873 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
12874 * tree-vrp.c (equiv_stack): No longer file scoped.
12875 (vrp_dom_walker): New class.
12876 (vrp_dom_walker::before_dom_children): New member function.
12877 (vrp_dom_walker::after_dom_children): Likewise.
12878 (identify_jump_threads): Setup domwalker. Use it rather than
12879 walking edges in a random order by hand. Simplify setup/finalization.
12880 (finalize_jump_threads): Remove.
12881 (vrp_finalize): Do not call identify_jump_threads here.
12882 (execute_vrp): Do it here instead and call thread_through_all_blocks
12885 PR tree-optimization/71437
12886 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
12888 (simplify_stmt_for_jump_threading): Add basic_block argument. All
12890 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
12891 (dom_opt_dom_walker::thread_across_edge): Remove
12892 handle_dominating_asserts argument. All callers changed.
12893 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
12894 changes. Remove calls to lhs_of_dominating_assert. Other
12895 uses of handle_dominating_asserts turn into unconditional code
12896 (simplify_control_stmt_condition_1): Likewise.
12897 (simplify_control_stmt_condition): Likewise.
12898 (thread_through_normal_block, thread_across_edge): Likewise.
12899 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
12900 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
12901 object if it is not an SSA_NAME.
12902 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
12903 before calling into the VRP specific simplifiers.
12904 (identify_jump_threads): Remove handle_dominating_asserts
12907 2017-03-16 Jakub Jelinek <jakub@redhat.com>
12910 * tree-diagnostic.c (default_tree_printer): No longer static.
12911 * tree-diagnostic.h (default_tree_printer): New prototype.
12913 2017-03-16 Tamar Christina <tamar.christina@arm.com>
12915 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
12916 Change ins into fmov.
12918 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12920 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
12921 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
12922 Use h_con constraint for operand 1.
12923 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
12924 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
12926 2017-03-15 Jeff Law <law@redhat.com>
12928 PR tree-optimization/71437
12929 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
12930 (record_temporary_equivalences): Use it.
12932 PR tree-optimization/71437
12933 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
12934 tree-ssa-scopedtables.
12935 (lookup_avail_expr, build_and_record_new_cond): Likewise.
12936 (record_conditions, record_cond, vuse_eq): Likewise.
12937 (record_edge_info): Adjust to API tweak of record_conditions.
12938 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
12939 (record_temporary_equivalences, optimize_stmt): Likewise.
12940 (eliminate_redundant_computations): Likewise.
12941 (record_equivalences_from_stmt): Likewise.
12942 * tree-ssa-scopedtables.c: Include options.h and params.h.
12943 (vuse_eq): New function, moved from tree-ssa-dom.c
12944 (build_and_record_new_cond): Likewise.
12945 (record_conditions): Likewise. Accept vector of conditions rather
12946 than edge_equivalence structure for first argument.
12947 for the first argument.
12948 (avail_exprs_stack::lookup_avail_expr): New member function, moved
12949 from tree-ssa-dom.c.
12950 (avail_exprs_stack::record_cond): Likewise.
12951 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
12952 from tree-ssa-dom.c.
12953 (avail_exprs_stack): Add new member functions lookup_avail_expr
12955 (record_conditions): Declare.
12957 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
12960 * lra-constraints.c (process_alt_operands): Increase reject for
12961 reloading an input/output operand.
12963 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
12966 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
12967 insns to convert from signed/unsigned char/short to IEEE 128-bit
12969 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
12971 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
12974 * config/i386/i386.c (ix86_vector_duplicate_value): Create
12975 subreg of inner mode for values already in registers.
12977 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
12979 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
12980 iteration reg is used after the loop.
12982 2017-03-14 Martin Sebor <msebor@redhat.com>
12984 PR tree-optimization/79800
12985 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
12986 precision in negative-positive range.
12987 (format_floating): Call non-const overload with adjusted precision.
12989 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12992 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
12995 2017-03-14 Martin Sebor <msebor@redhat.com>
12997 PR middle-end/80020
12998 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
12999 * builtins.def (aligned_alloc): Use it.
13002 * Makefile.in (GTFILES): Add calls.c.
13003 * calls.c: Include "gt-calls.h".
13005 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13007 PR rtl-optimization/79728
13008 * regs.h (struct target_regs): New field
13009 x_contains_allocatable_regs_of_mode.
13010 (contains_allocatable_regs_of_mode): New macro.
13011 * reginfo.c (init_reg_sets_1): Initialize it, and change
13012 contains_reg_of_mode so it includes global regs as well.
13013 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13014 rather than contains_regs_of_mode.
13016 2017-03-14 Martin Liska <mliska@suse.cz>
13018 * doc/invoke.texi: Document options that can't be combined with
13019 -fcheck-pointer-bounds.
13021 2017-03-14 Martin Liska <mliska@suse.cz>
13023 PR middle-end/79831
13024 * doc/invoke.texi (-Wchkp): Document the option.
13026 2017-03-14 Martin Liska <mliska@suse.cz>
13028 * Makefile.in: Install gcov-dump.
13030 2017-03-14 Martin Liska <mliska@suse.cz>
13032 * multiple_target.c (expand_target_clones): Bail out for
13033 an invalid attribute.
13035 2017-03-14 Richard Biener <rguenther@suse.de>
13037 * alias.c (struct alias_set_entry): Pack properly.
13038 * cfgloop.h (struct loop): Likewise.
13039 * cse.c (struct set): Likewise.
13040 * ipa-utils.c (struct searchc_env): Likewise.
13041 * loop-invariant.c (struct invariant): Likewise.
13042 * lra-remat.c (struct cand): Likewise.
13043 * recog.c (struct change_t): Likewise.
13044 * rtl.h (struct address_info): Likewise.
13045 * symbol-summary.h (function_summary): Likewise.
13046 * tree-loop-distribution.c (struct partition): Likewise.
13047 * tree-object-size.c (struct object_size_info): Likewise.
13048 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13049 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13050 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13051 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13052 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13053 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13054 (struct _stmt_vec_info): Likewise.
13056 2017-03-14 Martin Liska <mliska@suse.cz>
13059 * multiple_target.c (create_dispatcher_calls): Check that
13060 a target can create a function dispatcher.
13062 2017-03-14 Martin Liska <mliska@suse.cz>
13065 * multiple_target.c (expand_target_clones): Drop local.local
13066 flag for default implementation.
13068 2017-03-14 Richard Biener <rguenther@suse.de>
13070 PR tree-optimization/80030
13071 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13073 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13075 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13076 gcc_fallthrough() instead of __attribute__((fallthrough));
13078 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13080 * doc/gcc.texi: Remove "up" link to (DIR).
13081 * doc/gccint.texi: Ditto.
13083 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13085 * doc/install.texi (Specific) <avr>: Remove reference to
13088 2017-03-13 Jeff Law <law@redhat.com>
13090 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13091 attribute rather than comments.
13093 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13094 match_scratch operand is highest.
13096 2017-03-13 Martin Liska <mliska@suse.cz>
13098 PR middle-end/78339
13099 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13100 is a CHKP clone, use original declaration.
13102 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13104 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13105 (arc_conditional_register_usage): Use a different allocation order
13106 when optimizing for size.
13107 * common/config/arc/arc-common.c (arc_option_optimization_table):
13108 Section anchors default on when optimizing for size.
13110 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13112 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13114 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13116 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13117 * config/arc/arc.md (cpu_facility): Add cd variant.
13118 (*movqi_insn): Add code density variant.
13119 (*movhi_insn): Likewise.
13120 (*movqi_insn): Likewise.
13121 (*addsi3_mixed): Likewise.
13122 (subsi3_insn): Likewise.
13124 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13126 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13128 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13130 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13131 expressions with MINUS and UNARY ops.
13133 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13136 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13138 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13139 between vec_select and vector argument.
13140 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13141 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13142 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13143 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13144 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13145 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13147 2017-03-13 Richard Biener <rguenther@suse.de>
13150 * params.def (vect-max-peeling-for-alignment): Fix typo.
13152 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13154 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13155 issue that only occurred with binutils below 2.18.
13157 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13159 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13160 refer to binutils 2.11/2.12 minimum.
13162 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13164 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13165 ftp.kernel.org and simplify binutils requirement.
13167 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13169 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13171 (Optimize Options): Ditto. Also remove redundancy.
13173 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13175 PR translation/79848
13176 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13178 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13179 to G_ to avoid double translation.
13181 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13183 PR translation/79923
13184 * auto-profile.c (get_combined_location): Convert leading
13185 character of diagnostics to lower case and remove trailing period.
13186 (read_profile): Likewise for various diagnostics.
13187 * config/arm/arm.c (arm_option_override): Remove trailing period
13188 from various diagnostics.
13189 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13190 (msp430_expand_delay_cycles): Likewise.
13192 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13195 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13196 full command-line argument, rather than just "str".
13197 (aarch64_validate_march): Likewise.
13198 (aarch64_validate_mtune): Likewise.
13200 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13202 PR rtl-optimization/78911
13203 * lra-assigns.c (must_not_spill_p): New function.
13204 (spill_for): Use it.
13206 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13208 PR tree-optimization/79981
13209 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13210 ATOMIC_COMPARE_EXCHANGE ifn result.
13211 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13212 IFN_ATOMIC_COMPARE_EXCHANGE.
13214 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13217 * opts.c (parse_sanitizer_options): Add missing question mark to
13218 "did you mean" message.
13220 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13222 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13224 (VMULEUH_UNS): Likewise.
13225 (VMULOUB_UNS): Likewise.
13226 (VMULOUH_UNS): Likewise.
13227 * config/rs6000/rs6000.c (builtin_function_type): Remove
13228 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13230 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13233 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13234 x with result of extra_parsing_for_operand_code_0.
13235 (function_reader::extra_parsing_for_operand_code_0): Convert
13236 return type from void to rtx, returning x. When reading
13237 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13238 larger size containing struct block_symbol.
13240 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13242 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13243 -mfloat128-hardware without -m64.
13245 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13248 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13249 entries to the case statement that marks unsigned arguments to
13250 overloaded functions.
13252 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13254 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13255 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13257 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13260 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13261 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13263 2017-03-10 Martin Liska <mliska@suse.cz>
13267 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13268 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13271 2017-03-10 Olivier Hainque <hainque@adacore.com>
13273 * tree-switch-conversion (array_value_type): Start by resetting
13274 candidate type to it's main variant.
13276 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13278 PR rtl-optimization/79909
13279 * combine.c (try_combine): Use simplify_replace_rtx on individual
13280 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13281 of the whole CALL_INSN_FUNCTION_USAGE.
13283 PR tree-optimization/79972
13284 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13285 get_range_info on SSA_NAMEs. Formatting fixes.
13287 2017-03-10 Richard Biener <rguenther@suse.de>
13288 Jakub Jelinek <jakub@redhat.com>
13290 PR tree-optimization/77975
13291 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13292 edge to be constant.
13293 (get_val_for): For constant x return it. Formatting fix.
13294 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13295 would use the same bases as the current one.
13297 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13299 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13300 instead of vec_select for V1TImode.
13301 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13303 (VSX_LE_128): Add V1TI to this mode iterator.
13304 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13305 (*vsx_le_perm_store_<mode>): Likewise.
13306 (pre-reload splitter for VSX stores): Likewise.
13307 (post-reload splitter for VSX stores): Likewise.
13308 (*vsx_xxpermdi2_le_<mode>): Likewise.
13309 (*vsx_lxvd2x2_le_<mode>): Likewise.
13310 (*vsx_stxvd2x2_le_<mode>): Likewise.
13312 2017-03-09 Michael Eager <eager@eagercon.com>
13314 Correct failures with --enable-checking=yes,rtl.
13316 * config/microblaze/microblaze.c (microblaze_expand_shift):
13317 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13318 test for const0_rtx.
13319 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13320 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13322 2017-03-09 Richard Biener <rguenther@suse.de>
13324 PR tree-optimization/79977
13325 * graphite-scop-detection.c (scop_detection::merge_sese):
13326 Handle the case of extra exits to blocks dominating the entry.
13328 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13330 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13333 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13335 PR rtl-optimization/79949
13336 * lra-constraints.c (process_alt_operands): Check memory when
13337 trying to predict a cycle. Print about the overall increase.
13339 2017-03-09 Richard Biener <rguenther@suse.de>
13341 PR middle-end/79971
13342 * gimple-expr.c (useless_type_conversion_p): Preserve
13343 TYPE_SATURATING for fixed-point types.
13345 2017-03-09 Richard Biener <rguenther@suse.de>
13348 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13349 alignment of BLKmode params.
13351 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13354 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13355 (VALL_NO_V2Q): Likewise.
13357 * config/aarch64/aarch64-simd.md
13358 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13360 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13361 VALL_NO_V2Q mode iterator.
13362 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13364 2017-03-09 Martin Liska <mliska@suse.cz>
13366 PR tree-optimization/79631
13367 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13368 tree_int_cst_sign_bit just for INTEGER constants.
13370 2017-03-09 Martin Liska <mliska@suse.cz>
13374 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13377 2017-03-09 Marek Polacek <polacek@redhat.com>
13380 * tree.c (inchash::add_expr): Handle TREE_VEC.
13382 2017-03-09 Martin Liska <mliska@suse.cz>
13385 (chkp_narrow_size_and_offset): New function.
13386 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13387 (void chkp_parse_bit_field_ref): New function.
13388 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13389 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13391 2017-03-09 Martin Liska <mliska@suse.cz>
13394 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13395 (chkp_find_bounds_1): Remove gcc_unreachable.
13397 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13400 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13401 BUILT_IN_SYNC*, determine the access type from the size suffix and
13402 always build a MEM_REF with that type. Handle forgotten
13403 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13406 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13407 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13408 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13409 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13410 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13411 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13412 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13413 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13414 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13415 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13416 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13417 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13418 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13419 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13420 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13421 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13422 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13423 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13424 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13425 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13426 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13427 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13428 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13429 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13430 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13431 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13432 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13433 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13434 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13435 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13436 definitions outside of __OPTIMIZE__ guarded section.
13439 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13440 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13441 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13442 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13445 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13447 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13448 ("vfenez<mode>"): Add missing constraints.
13450 2017-03-08 Martin Sebor <msebor@redhat.com>
13453 * config/nds32/nds32.c (nds32_option_override):
13454 Fix misspelled diagnostic.
13456 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13459 * gimplify.c (gimplify_omp_for): Replace index var in outer
13460 taskloop statement with an artificial variable and add
13461 OMP_CLAUSE_PRIVATE clause for it.
13463 2017-03-08 Richard Biener <rguenther@suse.de>
13465 PR tree-optimization/79955
13466 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13467 for accesses that are completely outside of the variable.
13469 2017-03-08 Andrew Haley <aph@redhat.com>
13471 PR tree-optimization/79943
13472 * tree-ssa-loop-split.c (compute_new_first_bound): When
13473 calculating the new upper bound, (END-BEG) should be added, not
13476 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13478 * config/avr/avr.md (setmemhi): Make sure match_dup
13479 operand number comes before match_scratch.
13481 2017-03-08 Richard Biener <rguenther@suse.de>
13483 PR tree-optimization/79920
13484 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13485 with ncopies == 1 to ...
13486 (vect_transform_slp_perm_load): ... here. Properly compute
13487 all element loads by iterating VF times over the group. Do
13488 not handle ncopies (computed in a broken way) in
13489 vect_create_mask_and_perm.
13491 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13494 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13495 is a uniform vector, use uniform_vector_p return value instead of
13496 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13498 2017-03-07 Marek Polacek <polacek@redhat.com>
13500 PR middle-end/79809
13501 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13502 (alloca_call_type): Likewise.
13504 2017-03-07 Martin Liska <mliska@suse.cz>
13506 * gcov.c (process_args): Put comment to correct location.
13508 2017-03-07 Martin Liska <mliska@suse.cz>
13510 PR middle-end/68270
13511 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13512 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13513 (chkp_narrow_bounds_for_field): Likewise.
13514 (chkp_parse_array_and_component_ref): Pass one more argument to
13517 2017-03-07 Richard Biener <rguenther@suse.de>
13519 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13522 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13524 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13525 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13527 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13530 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13531 to end of description.
13532 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13534 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13536 PR rtl-optimization/79901
13537 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13539 (*avx512f_<code><mode>3<mask_name>): ... this.
13540 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13541 iterator instead of VI8_AVX2_AVX512BW.
13543 PR rtl-optimization/79901
13544 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13545 min/max expander, expand it using expand_vec_cond_expr.
13548 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13551 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13554 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13555 to void * for PCH reasons.
13556 * dwarf2out.c (output_loc_operands, output_die): Cast
13557 v.val_vec.array to unsigned char *.
13559 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13562 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13565 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13567 PR rtl-optimization/79571
13568 * lra-constraints.c (process_alt_operands): Calculate static
13569 reject and subtract it from overall when only addresses will be
13572 2017-03-06 Julia Koval <julia.koval@intel.com>
13575 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13576 incoming stack boundary to 128 for 64-bit targets.
13578 2017-03-06 Richard Biener <rguenther@suse.de>
13580 PR tree-optimization/79894
13581 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13582 to NULL after folding it.
13584 2017-03-06 Richard Biener <rguenther@suse.de>
13586 PR tree-optimization/79824
13587 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13588 check disabling peeling for gaps.
13590 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13592 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13593 attributes): Document gettimeofday.
13595 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13597 * config/s390/s390.c (s390_option_override_internal): Set
13598 PARAM_MIN_VECT_LOOP_BOUND
13600 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13602 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13603 * config/s390/s390.md: Likewise.
13605 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13608 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13609 (<avx2_avx512>_perm<mode>): Rename to ...
13610 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13612 (<avx512>_perm<mode>_mask): Rename to ...
13613 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13615 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13616 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13617 instead of VI8F_256_512.
13618 (avx512f_perm<mode>): New define_expand.
13619 (avx512f_perm<mode>_mask): Likewise.
13620 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13621 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13623 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13625 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13626 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13628 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13630 2017-03-06 Martin Liska <mliska@suse.cz>
13633 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13634 when having a SSA NAME w/o VAR_DECL assigned to it.
13636 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13638 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13639 msa_dpsub_<su>_d): Fix MODE for vec_select.
13641 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13643 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13645 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13647 2017-03-06 Richard Biener <rguenther@suse.de>
13649 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13650 * plugin.c (register_plugin_info): Likewise.
13651 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13653 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13655 * config/i386/sse.md (sse_storehps, sse_storelps,
13656 avx_<castmode><avxsizesuffix>_<castmode>,
13657 avx512f_<castmode><avxsizesuffix>_<castmode>,
13658 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13659 in condition that at least one operand is not a MEM.
13661 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13663 PR middle-end/79805
13664 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13665 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13667 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13668 gimple_call_nothrow_p flag based on whether original builtin can throw.
13669 If it can, emit following stmts on the fallthrough edge.
13670 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13671 don't create new bb if inserting just debug stmts on the edge, try to
13672 insert them on the fallthru bb or just reset debug stmts.
13674 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13677 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13678 restore recog_data (including the operand rtxes inside it) around
13679 the call to get_insn_template.
13681 2017-03-03 Martin Sebor <msebor@redhat.com>
13683 PR tree-optimization/79699
13684 * context.c (context::~context): Free MPFR caches to avoid
13685 a memory leak on program exit.
13687 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13689 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13690 Use wide_int::ulow () instead of .elt (0).
13692 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13694 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13695 (*pushxf): Limit oF constraint to 32bit targets and add oC
13696 constraint for 64bit targets.
13697 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13698 (*pushdf): Change rmF constraint to rmC.
13700 2017-03-03 Martin Liska <mliska@suse.cz>
13702 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13703 Remove unused variable.
13705 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13708 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13709 is a memory operand, increase num_memory.
13710 (ix86_expand_args_builtin): Likewise.
13712 2017-03-03 Jan Hubicka <jh@suse.cz>
13715 * ipa-devirt.c (maybe_record_node): Properly handle
13716 __cxa_pure_virtual visibility.
13718 2017-03-03 Martin Liska <mliska@suse.cz>
13720 PR tree-optimization/79803
13721 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13723 (pass_loop_prefetch::execute): Disabled optimization if an
13724 assumption about L1 cache size is not met.
13726 2017-03-03 Martin Liska <mliska@suse.cz>
13728 PR rtl-optimization/79574
13729 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13730 (hash_scan_set): Likewise.
13731 (dump_hash_table): Likewise.
13732 (hoist_code): Likewise.
13734 2017-03-03 Richard Biener <rguenther@suse.de>
13736 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13737 in place of uhigh ().
13738 (fixed_convert_from_real): Likewise.
13740 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13743 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13745 2017-03-03 Richard Biener <rguenther@suse.de>
13747 PR middle-end/79818
13748 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13749 TYPE_OVERFLOW_UNDEFINED check.
13751 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13753 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13755 (vector_ae_<mode>_p): Likewise.
13756 (vector_nez_<mode>_p): Likewise.
13757 (vector_ne_v2di_p): Likewise.
13758 (vector_ae_v2di_p): Likewise.
13759 (vector_ne_<mode>_p): Likewise.
13760 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13762 (vsx_tsqrt<mode>2_fe): Likewise.
13764 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13767 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
13769 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13771 PR rtl-optimization/79780
13772 * cprop.c (one_cprop_pass): When second and further conditional trap
13773 in a single basic block is turned into an unconditional trap, turn it
13774 into a deleted note to avoid RTL verification failures.
13776 2017-03-02 Richard Biener <rguenther@suse.de>
13778 * fold-const.c (const_binop): Use ulow () instead of elt (0).
13780 2017-03-02 Richard Biener <rguenther@suse.de>
13782 PR tree-optimization/79345
13784 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
13785 param and abort the walk, returning -1 if it is hit.
13786 (walk_aliased_vdefs): Take a limit param and pass it on.
13787 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
13788 defaulting to 0 and return a signed int.
13789 * tree-ssa-uninit.c (struct check_defs_data): New struct.
13790 (check_defs): New helper.
13791 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
13792 about uninitialized memory.
13793 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
13794 bogus uninitialized warning.
13795 (fixed_convert_from_real): Likewise.
13797 2017-03-02 Bin Cheng <bin.cheng@arm.com>
13799 PR tree-optimization/66768
13800 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
13801 iv_use if base object can't be determined.
13803 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13805 PR tree-optimization/79345
13806 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
13807 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
13808 (get_pattern_stats): Initialize it.
13809 * genemit.c (gen_expand): Verify match_scratch numbers come after
13810 match_operand/match_dup numbers.
13811 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
13812 match_scratch numbers.
13813 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
13815 * config/s390/s390.md (trunctdsd2): Likewise.
13817 2017-03-02 Richard Biener <rguenther@suse.de>
13819 * wide-int.h (wide_int_storage::operator=): Implement in terms
13822 2017-03-02 Richard Biener <rguenther@suse.de>
13824 PR tree-optimization/79777
13825 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
13826 the to insert expression to sth existing.
13828 2017-03-01 Martin Sebor <msebor@redhat.com>
13830 PR middle-end/79692
13831 * gimple-ssa-sprintf.c
13832 (directive::known_width_and_precision): New function.
13833 (format_integer): Use it.
13834 (get_mpfr_format_length): Consider the full range of precision
13835 when computing %g output with the # flag. Set the likely byte
13836 count to 3 rather than 1 when precision is indeterminate.
13837 (format_floating): Correct the lower bound of precision.
13839 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13841 * doc/invoke.texi: Document default code model for 64-bit Linux.
13843 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13846 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
13847 udiv rather than div since input pattern is unsigned.
13849 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
13851 * config/i386/i386.c (print_reg): Warn for values of
13852 unsupported size in integer register.
13854 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
13857 * config/rs6000/predicates.md (current_file_function_operand): Do
13858 not allow self calls to be local if the function is replaceable.
13860 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
13863 * config/rs6000/altivec.h (vec_ctz and others): Change the
13864 preprocessor macro that controls conditional compilation from
13865 _ARCH_PWR9 to __POWER9_VECTOR__.
13866 (vec_all_ne): Change parameterization of __altivec_scalar_pred
13867 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
13868 control (instead of _ARCH_PWR9 control) so that template
13869 definition uses power9-specific function.
13870 (vec_any_eq): Likewise.
13871 (vec_all_ne): Change macro definition to use a power9-specific
13872 expansion under #ifdef __POWER9_VECTOR__ control (instead of
13873 _ARCH_PWR9 control).
13874 (vec_any_eq) Likewise.
13875 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
13876 expansion for CMPNEF to remove support for xvcmpnesp instruction.
13877 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
13878 support for xvcmpnedp instruction.
13879 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
13880 macro expansion so that Power9 implementation of vec_all_ne does
13881 not use the AltiVec predicate framework.
13882 (VCMPNEH_P): Likewise.
13883 (VCMPNEW_P): Likewise.
13884 (VCMPNED_P): Likewise.
13885 (VCMPNEFP_P): Likewise.
13886 (VCMPNEDP_P): Likewise.
13887 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
13888 implementation of vec_any_eq to not use AltiVec predicate
13890 (VCMPAEH_P): Likewise.
13891 (VCMPAEW_P): Likewise.
13892 (VCMPAED_P): Likewise.
13893 (VCMPAEFP_P): Likewise.
13894 (VCMPAEDP_P): Likewise.
13895 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
13896 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
13897 not use the AltiVec predicate framework.
13898 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
13899 of vec_any_eq to not use AltiVec predicate framework.
13900 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13901 support for predefined __POWER9_VECTOR__ macro to indicate that
13902 Power9 instruction selection is enabled.
13903 (altivec_overloaded_builtins): Remove extraneous
13904 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
13905 function argument types RS6000_BTI_bool_V16QI and
13906 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
13907 entry for overloaded function argument types RS6000_BTI_bool_V4SI
13908 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
13909 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
13910 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
13911 Power9 for implementations of vec_cmpne. Change the signature for
13912 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
13913 (representing vec_all_ne) to remove the previously described first
13914 argument of type RS6000_BTI_INTSI, as this was an artifact of
13915 reliance on the AltiVec predicate framework, which is no longer
13916 used in the implementation of these functions. Add
13917 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
13918 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
13919 since, unlike the AltiVec predicate framework implementation, we
13920 do not share function descriptors between vec_alle and vec_anyeq.
13921 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
13922 set of modes that receive special treatment even when
13923 TARGET_P9_VECTOR is true. The special treatment emits code that
13924 does not depend on Power9 instructions.
13925 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
13926 define_expand to not rely on AltiVec predicate framework.
13927 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13929 (vector_ne_v2di_p): Change this define_expand to not rely on
13930 AltiVec predicate framework.
13931 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
13933 (vector_ne_<mode>_p): Change this define_expand to not rely on
13934 AltiVec predicate framework.
13935 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13937 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
13938 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
13939 define_insn pattern.
13940 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
13941 define_insn pattern because the xvcmpne<VSs>. instruction is not
13943 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
13944 instruction is not supported.
13946 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13948 * config/nvptx/nvptx.c: Include intl.h.
13950 2017-03-01 Martin Jambor <mjambor@suse.cz>
13953 * ipa-prop.h (ipa_bits): Removed field known.
13954 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
13955 to pointers. Adjusted their comments to warn about their sharing.
13956 (ipcp_transformation_summary): Change bits to a vector of pointers.
13957 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
13958 (ipa_get_ipa_bits_for_value): Declare.
13959 * tree-vrp.h (value_range): Mark as GTY((for_user)).
13960 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
13961 (ipa_bits_hash_table): Likewise.
13962 (ipa_vr_ggc_hash_traits): Likewise.
13963 (ipa_vr_hash_table): Likewise.
13964 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
13965 being pointers and vr_known being removed.
13966 (ipa_set_jf_unknown): Likewise.
13967 (ipa_get_ipa_bits_for_value): New function.
13968 (ipa_set_jfunc_bits): Likewise.
13969 (ipa_get_value_range): New overloaded functions.
13970 (ipa_set_jfunc_vr): Likewise.
13971 (ipa_compute_jump_functions_for_edge): Use the above functions to
13972 construct bits and vr parts of jump functions.
13973 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
13974 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
13976 (ipcp_grow_transformations_if_necessary): Also allocate
13977 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
13979 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
13980 them. Fix too long lines.
13981 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
13982 vr_known being removed.
13983 (ipa_read_jump_function): Use new setter functions to construct bits
13984 and vr parts of jump functions or set them to NULL.
13985 (write_ipcp_transformation_info): Adjust for bits being pointers.
13986 (read_ipcp_transformation_info): Likewise.
13987 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
13989 Include gt-ipa-prop.h.
13990 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
13992 (ipcp_store_bits_results): Likewise.
13993 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
13994 Do not write to existing jump functions but use a temporary instead.
13996 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13999 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14000 attempt to use its first operand as BIT_FIELD_REF base.
14002 2017-03-01 Richard Biener <rguenther@suse.de>
14004 PR middle-end/79721
14005 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14006 interpolating formula in wrapping arithmetic.
14007 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14009 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14011 PR tree-optimization/79734
14012 * tree-vect-generic.c (expand_vector_condition): Optimize
14013 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14014 Handle VEC_COND_EXPR where comparison has different inner width from
14015 type's inner width.
14017 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14019 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14020 markup, and similar issues. Remove @opindex entries for things
14021 that aren't options. Add missing -mmpy-option entries.
14023 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14025 PR tree-optimization/79737
14026 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14027 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14028 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14029 instead of byte_size. Formatting fix.
14030 (shift_bytes_in_array_right): Formatting fix.
14032 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14035 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14036 condition on optimize for the leaf function test.
14038 2017-02-28 Martin Liska <mliska@suse.cz>
14041 * read-rtl-function.c (function_reader::handle_unknown_directive):
14042 Bail out when one uses -flto.
14044 2017-02-28 Martin Liska <mliska@suse.cz>
14046 * common.opt: Replace space with tabular for options of <number>
14048 * config/i386/i386.opt: Show <number> value for
14049 -mlarge-data-threshold.
14050 * opts.c (print_filtered_help): Do not display number in hexadecimal
14053 2017-02-28 Martin Liska <mliska@suse.cz>
14055 * common.opt: Fix --help=option -Q for options which are of
14058 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14060 * config/i386/i386.c (print_reg): Error out for values
14061 of 8-bit size in invalid integer register.
14063 2017-02-28 Martin Sebor <msebor@redhat.com>
14065 PR tree-optimization/79691
14066 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14068 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14071 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14072 gcc_unreachable with output_operand_lossage.
14074 2017-02-28 Richard Biener <rguenther@suse.de>
14076 PR tree-optimization/79740
14077 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14079 (visit_nary_op): Insert the nary into the hashtable if we
14080 pattern-matched sth.
14081 * tree-ssa-pre.c (eliminate_insert): Robustify.
14083 2017-02-28 Richard Biener <rguenther@suse.de>
14085 PR middle-end/79731
14086 * fold-const.c (decode_field_reference): Reject out-of-bound
14089 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14091 * config/i386/i386.c: Include intl.h.
14092 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14093 instead of just cond ? "..." : "...".
14094 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14095 * coverage.c (read_counts_file): Likewise.
14096 * omp-offload.c: Include intl.h.
14097 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14098 of just cond ? "..." : "...".
14099 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14100 of just cond ? "..." : "...".
14102 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14105 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14107 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14108 'tune for' CPU name.
14109 * config/arm/arm-cpu-data.h: Regenerated.
14111 2017-02-28 Richard Biener <rguenther@suse.de>
14113 PR tree-optimization/79732
14114 * tree-inline.c (expand_call_inline): Do not shadow var.
14116 2017-02-28 Richard Biener <rguenther@suse.de>
14118 PR tree-optimization/79723
14119 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14120 address-space properly.
14122 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14124 * doc/optinfo.texi (Optimization groups): Fix option used for
14126 * doc/invoke.texi (-fopt-info): Document "omp".
14127 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14128 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14129 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14130 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14131 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14133 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14135 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14136 OPTGROUP_OMP with "omp".
14138 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14141 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14142 for arithmetic shift of unsigned V2DI.
14144 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14146 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14147 arc/linux.h headers.
14148 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14149 (LINK_SPEC): Likewise.
14150 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14151 (EXTRA_SPECS): Likewise.
14152 (STARTFILE_SPEC): Likewise.
14153 (ENDFILE_SPEC): Likewise.
14154 (LIB_SPEC): Likewise.
14155 (TARGET_SDATA_DEFAULT): Likewise.
14156 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14157 (MULTILIB_DEFAULTS): Likewise.
14158 (DWARF2_UNWIND_INFO): Likewise.
14159 * config/arc/big.h: New file.
14160 * config/arc/elf.h: Likewise.
14161 * config/arc/linux.h: Likewise.
14162 * config/arc/t-uClibc: Remove.
14164 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14166 PR tree-optimization/77536
14167 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14168 (tree_transform_and_unroll_loop): Use above function to compute the
14169 estimated niter of unrolled loop and use it when scaling profile.
14170 Also use count info rather than frequency if it's non-zero.
14171 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14172 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14173 (vect_transform_loop): Call above function.
14175 2017-02-27 Richard Biener <rguenther@suse.de>
14177 PR tree-optimization/45397
14178 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14179 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14180 (visit_nary_op): Add pattern matching for CSEing sign-changed
14181 or truncated operations with wider ones.
14183 2017-02-27 Richard Biener <rguenther@suse.de>
14185 PR tree-optimization/79690
14186 * tree-vect-stmts.c (vectorizable_store): Use vector type
14187 built from the DR with address-space.
14189 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14191 * doc/invoke.texi (Optimize Options): Refine the description
14192 of asan-use-after-return.
14194 2017-02-25 Alan Modra <amodra@gmail.com>
14196 PR rtl-optimization/79584
14197 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14198 base, not ad->base_term, the reg within base. Remove assertion
14199 that ad->base == ad->base_term. Replace gen_int_mode using
14200 bogus mode with const0_rtx.
14202 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14204 PR middle-end/79396
14205 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14206 FMA_EXPR like tcc_binary or tcc_unary.
14208 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14211 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14213 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14214 (output_loc_operands): Handle DW_OP_call_ref and
14215 DW_OP_GNU_variable_value.
14216 (struct variable_value_struct): New type.
14217 (struct variable_value_hasher): Likewise.
14218 (variable_value_hash): New variable.
14219 (string_types): Remove.
14220 (copy_loc_descr): New function.
14221 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14222 (prepend_loc_descr_to_each): New function.
14223 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14224 instead of add_loc_descr_to_each if the first argument is single
14225 location list and the second has multiple.
14226 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14227 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14228 when looking for variable value which doesn't have other location info.
14229 (loc_list_from_tree): Formatting fix.
14230 (gen_array_type_die): Simplify DW_AT_string_length handling.
14231 (adjust_string_types): Remove.
14232 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14233 string_types. Call resolve_variable_values.
14234 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14235 (resolve_addr_in_expr): Likewise. Add A argument.
14236 (copy_deref_exprloc): Remove deref argument. Adjust for the
14237 original expression being DW_OP_GNU_variable_value with optionally
14238 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14239 optionally after it.
14240 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14241 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14242 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14243 (variable_value_hasher::hash, variable_value_hasher::equal): New
14245 (resolve_variable_value_in_expr, resolve_variable_value,
14246 resolve_variable_values, note_variable_value_in_expr,
14247 note_variable_value): New functions.
14248 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14251 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14254 * opts.h (handle_generated_option): Add GENERATED_P argument.
14255 * opts-common.c (handle_option): Adjust function comment.
14256 (handle_generated_option): Add GENERATED_P argument, pass it to
14258 (control_warning_option): Pass false to handle_generated_option
14260 * opts.c (maybe_default_option): Pass true to handle_generated_option
14262 * optc-gen.awk: Likewise.
14264 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14266 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14267 a REG, look at the REG it is a SUBREG of.
14268 (splitter for cmpeqsi_t): Ditto.
14270 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14272 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14273 the special USEs with the pattern of the insn, not the insn itself.
14275 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14278 * doc/invoke.texi: Document -mload-store-pairs.
14280 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14281 Sandra Loosemore <sandra@codesourcery.com>
14283 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14284 argument isn't a CONST_INT.
14285 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14286 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14287 (nios2_validate_compare): Bypass alternate compare logic if *op2
14288 is not a CONST_INT.
14289 (ldstwm_operation_p): Return false if first_base is not a REG or
14290 if first_offset is not a CONST_INT.
14292 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14294 * config/cris/cris.md: Use correct operand in a define_peephole2.
14296 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14298 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14300 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14302 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14303 this_insn if it is an INSN or JUMP_INSN.
14304 (force_offsettable): Look at base, not at addr.
14305 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14306 on things that aren't necessarily CONST_INTs.
14308 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14310 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14311 -mfpmath=sse is the default also for x86-32 targets with SSE2
14312 instruction set when @option{-ffast-math} is enabled
14314 2017-02-24 Jeff Law <law@redhat.com>
14316 PR rtl-optimizatoin/79286
14317 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14320 2017-02-24 Richard Biener <rguenther@suse.de>
14322 PR tree-optimization/79389
14323 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14326 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14328 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14329 function comment to reflect reality.
14330 (loop_exits_before_overflow): Fix typo in function description.
14332 2017-02-24 Richard Biener <rguenther@suse.de>
14334 PR tree-optimization/79389
14335 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14336 properly that a threading opportunity exists. Detect conditional
14337 copy/constant propagation opportunities.
14339 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14341 * config/visium/visium.md (type): Add trap.
14342 (b): New mode attribute.
14343 (*btst): Rename into...
14344 (*btst<mode>): ...this and adjust.
14345 (*cbranchsi4_btst_insn): Rename into...
14346 (*cbranch<mode>4_btst_insn): ...this and adjust.
14347 (trap): New define_insn.
14349 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14351 PR tree-optimization/79389
14352 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14353 (noce_reversed_cond_code): New function.
14354 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14355 reversed_comparison_code. Formatting fix.
14356 (noce_try_store_flag): Test rev_cond != NULL in addition to
14357 reversed_comparison_code.
14358 (noce_try_store_flag_constants): Likewise.
14359 (noce_try_store_flag_mask): Likewise.
14360 (noce_try_addcc): Use rev_cond if non-NULL instead of
14361 reversed_comparison_code.
14362 (noce_try_cmove_arith): Likewise. Formatting fixes.
14363 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14364 (noce_find_if_block): Initialize rev_cond.
14365 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14366 instead of false as last argument never attempt to reverse it
14369 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14371 PR tree-optimization/79663
14372 * tree-predcom.c (combine_chains): Process refs in reverse order
14373 only for ZERO length chains, and add explaining comment.
14375 2017-02-23 Jeff Law <law@redhat.com>
14377 PR tree-optimization/79578
14378 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14379 in call to operand_equal_p.
14381 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14384 * config/i386/cpuid.h: Fix another undefined behavior.
14386 2017-02-23 Richard Biener <rguenther@suse.de>
14388 PR tree-optimization/79683
14389 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14390 vector types for data-refs.
14392 2017-02-23 Martin Liska <mliska@suse.cz>
14394 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14396 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14398 PR middle-end/79665
14399 * internal-fn.c (get_range_pos_neg): Moved to ...
14400 * tree.c (get_range_pos_neg): ... here. No longer static.
14401 * tree.h (get_range_pos_neg): New prototype.
14402 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14403 are known to be in between 0 and signed maximum inclusive, try to
14404 expand both unsigned and signed divmod and use the cheaper one from
14407 2017-02-22 Jeff Law <law@redhat.com>
14409 PR tree-optimization/79578
14410 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14411 to compare base operands.
14413 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14416 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14417 gpc_reg_operand instead of fpr_reg_operand.
14419 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14421 * config/mips/mips.c (mips_return_in_memory): Force FP
14422 vector types to be returned in memory for o32 ABI.
14424 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14426 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14427 instead of DW_TAG_member for static data member declarations and don't
14428 set no_linkage_name for static inline data members.
14429 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14432 2017-02-22 Martin Liska <mliska@suse.cz>
14434 * doc/invoke.texi: Replace inequality signs with square brackets
14437 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14439 PR tree-optimization/68644
14440 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14442 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14445 * lra-constraints.c (simplify_operand_subreg): Handle
14446 WORD_REGISTER_OPERATIONS targets.
14448 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14451 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14452 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14453 elimination by swapping fld*.
14455 2017-02-22 Richard Biener <rguenther@suse.de>
14457 PR tree-optimization/79673
14458 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14459 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14460 irrelevant address-space qualifiers and avoiding a
14461 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14463 2017-02-22 Richard Biener <rguenther@suse.de>
14465 PR tree-optimization/79666
14466 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14467 to not symbolically negate if that may introduce undefined
14470 2017-02-22 Martin Liska <mliska@suse.cz>
14473 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14474 * data-streamer-out.c (streamer_write_gcov_count_stream):
14476 * value-prof.c (stream_out_histogram_value): Make assert more
14477 precise based on type of counter.
14479 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14482 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14483 Use nonimmediate_operand instead of memory_operand for operand 1.
14484 (float-extend standard_x87sse_constant_load splitter): Ditto.
14486 2017-02-21 Jeff Law <law@redhat.com>
14488 PR tree-optimization/79621
14489 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14490 blocks with edges to themselves.
14492 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14495 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14496 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14497 Use gimple_call_builtin_p.
14500 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14501 on temporarily removed DEBUG_INSNs.
14503 PR tree-optimization/79649
14504 * tree-loop-distribution.c (classify_partition): Give up on
14505 non-generic address space loads/stores.
14507 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14509 * doc/loop.texi (Loop manipulation): Remove nonexistent
14510 tree_ssa_loop_version from the documentation.
14511 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14513 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14516 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14517 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14518 * config/rs6000/rs6000.c: Include except.h.
14519 (rs6000_expand_split_stack_prologue): Call
14520 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14522 2017-02-21 Martin Jambor <mjambor@suse.cz>
14525 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14526 have been analyzed.
14528 2017-02-21 Martin Jambor <mjambor@suse.cz>
14530 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14531 for backward compatibility only.
14532 * doc/invoke.texi (Option Summary): Remove all references to
14533 -fipa-cp-alignment.
14535 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14539 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14541 * lra-constraints.c (curr_insn_transform): Handle
14542 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14544 2017-02-21 Martin Liska <mliska@suse.cz>
14546 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14548 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14550 PR translation/79638
14551 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14553 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14556 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14557 (arm_function_ok_for_sibcall): Return false for an indirect call by
14558 descriptor if all the argument registers are used.
14559 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14560 alignment of the function.
14562 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14564 PR tree-optimization/61441
14565 * simplify-rtx.c (simplify_const_unary_operation): For
14566 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14567 the sNaN unmodified.
14569 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14571 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14572 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14573 instead of SYSTEM_HEADER_DIR.
14575 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14576 Martin Liška <mliska@suse.cz>
14578 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14579 Fix typos and grammar, use active voice, and clarify.
14581 2017-02-20 Marek Polacek <polacek@redhat.com>
14583 PR middle-end/79537
14584 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14587 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14589 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14592 * config/i386/i386.c (ix86_expand_builtin): Handle
14593 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14594 ix86_builtins_isa[fcode].isa as a requirement of those
14595 flags and any other flag in the bitmask.
14596 (ix86_init_mmx_sse_builtins): Use 0 instead of
14597 ~OPTION_MASK_ISA_64BIT as mask.
14598 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14599 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14600 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14601 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14603 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14606 * lra-constraints.c (split_reg): Check requested split mode
14607 is supported by the register.
14609 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14611 * lra-constraints.c (simplify_operand_subreg): Remove early
14614 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14617 * lra-constraints.c (curr_insn_transform): Tighten condition
14618 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14620 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14623 * lra-constraints.c (curr_insn_transform): Handle
14624 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14626 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14629 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14631 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14633 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14636 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14639 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14641 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14644 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14646 * final.c (last_columnnum, override_columnnum): New variables.
14647 (final_start_function): Set last_columnnum, pass it to begin_prologue
14648 hook and pass 0 to dwarf2out_begin_prologue.
14649 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14650 to source_line debug hook.
14651 (notice_source_line): Compute last_columnnum and for debug_column_info
14652 return true on column changes.
14653 * debug.h (struct gcc_debug_hooks): Add column argument to
14654 source_line and begin_prologue hooks.
14655 (debug_nothing_int_charstar_int_bool): Remove prototype.
14656 (debug_nothing_int_int_charstar,
14657 debug_nothing_int_int_charstar_int_bool): New prototypes.
14658 (dwarf2out_begin_prologue): Add column argument.
14659 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14660 begin_prologue hooks.
14661 (debug_nothing_int_charstar_int_bool): Remove.
14662 (debug_nothing_int_int_charstar,
14663 debug_nothing_int_int_charstar_int_bool): New functions.
14664 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14665 through to dwarf2out_source_line.
14666 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14667 (dwarf2out_source_line): Add column argument, emit it if requested.
14668 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14670 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14671 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14672 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14673 through to dwarf2out_begin_prologue.
14674 (vmsdbgout_source_line): Add column argument, pass it through to
14675 dwarf2out_source_line.
14676 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14677 dbxout_source_line caller.
14678 (dbxout_source_line): Add column argument.
14680 * common.opt (gno-column-info, gcolumn-info): New options.
14681 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14682 (check_die): Also test for multiple DW_AT_decl_column attributes.
14683 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14684 DW_AT_decl_column if requested.
14685 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14687 (gen_variable_die): Likewise.
14688 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14689 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14692 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14693 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14694 (ix86_handle_option): Handle OPT_m3dnowa.
14695 * doc/invoke.texi (-m3dnowa): Document.
14696 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14697 -m3dnowa instead of -m3dnow -march=athlon.
14700 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14701 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14703 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14706 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14707 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14708 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14709 generator for vsx_xxpermdi_<mode>_be.
14710 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14711 force big-endian semantics.
14712 (vsx_xxpermdi_<mode>_be): New define_expand with same
14713 implementation as previous version of vsx_xxpermdi_<mode>.
14715 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14717 PR tree-optimization/79327
14718 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14719 variable, its initialization and use.
14721 2017-02-17 Julia Koval <julia.koval@intel.com>
14723 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14724 (OPTION_MASK_ISA_PKU_UNSET): New.
14725 (ix86_handle_option): Handle -mrdpid.
14726 * config/i386/cpuid.h (bit_RDPID): New.
14727 * config/i386/driver-i386.c (host_detect_local_cpu):
14728 Detect RDPID feature.
14729 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14730 * config/i386/i386-c.c (ix86_target_macros_internal):
14732 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14733 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14734 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14735 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14736 * config/i386/i386.md (define_insn "rdpid"): New.
14737 * config/i386/i386.opt Add -mrdpid.
14738 * config/i386/immintrin.h (_rdpid_u32): New.
14740 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14742 PR rtl-optimization/79541
14743 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14744 instead of transforming it into USE.
14746 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14748 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14749 If HONOR_SNANS (SFmode) force the input to a register.
14750 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14751 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14752 an frsp or similar insn.
14754 2017-02-17 Martin Liska <mliska@suse.cz>
14756 PR rtl-optimization/79577
14757 * params.def (selsched-max-sched-times): Increase minimum to 1.
14759 2017-02-17 Martin Liska <mliska@suse.cz>
14761 PR rtl-optimization/79574
14762 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14764 2017-02-17 Martin Liska <mliska@suse.cz>
14766 PR tree-optimization/79529
14767 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
14768 ssa_defined_default_def_p to handle cases which are implicitly
14770 * tree-ssa.c (ssa_defined_default_def_p): New function.
14771 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
14772 which are implicitly defined.
14773 * tree-ssa.h (ssa_defined_default_def_p): Declare.
14775 2017-02-17 Richard Biener <rguenther@suse.de>
14777 PR middle-end/79576
14778 * params.def (max-ssa-name-query-depth): Limit to 10.
14780 2017-02-17 Richard Biener <rguenther@suse.de>
14782 PR tree-optimization/79552
14783 * tree-ssa-structalias.c (visit_loadstore): Properly verify
14786 2017-02-17 Richard Biener <rguenther@suse.de>
14789 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
14791 2017-02-17 Marek Polacek <polacek@redhat.com>
14793 PR middle-end/79536
14794 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
14795 (fold_negate_expr): New wrapper.
14797 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
14799 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
14800 Correct terminology and de-emphasize pre-standard behavior.
14802 2017-02-16 Alan Modra <amodra@gmail.com>
14804 PR rtl-optimization/79286
14805 * ira.c (def_dominates_uses): New function.
14806 (update_equiv_regs): Don't create an equivalence for insns that
14807 may trap where the register def does not dominate the use.
14809 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
14811 PR rtl-optimization/78127
14812 * lra.c (lra): Call lra_eliminate before finish the loop after
14815 2017-02-16 Richard Biener <rguenther@suse.de>
14817 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
14819 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
14820 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
14821 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
14822 (isl_val_int_from_wi): New function.
14823 (extract_affine_gmp): Rename to ...
14824 (extract_affine_wi): ... this, take a widest_int.
14825 (extract_affine_int): Just wrap extract_affine_wi.
14826 (add_param_constraints): Use isl_val_int_from_wi.
14827 (add_loop_constraints): Likewise, and extract_affine_wi.
14829 2017-02-15 Jeff Law <law@redhat.com>
14831 PR middle-end/79521
14832 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
14833 ira_init_register_move_cost_if_necessary.
14835 2017-02-15 Martin Sebor <msebor@redhat.com>
14837 PR middle-end/32003
14838 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
14839 removed in a prior commit.
14841 2017-02-15 Bin Cheng <bin.cheng@arm.com>
14843 PR tree-optimization/79347
14844 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
14845 counters during peeling.
14847 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
14849 * Makefile.in (site.exp): Remove "set ISLVER".
14851 2017-02-15 Jakub Jelinek <jakub@redhat.com>
14854 * real.c (real_from_integer): Call real_convert even for decimal.
14856 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
14859 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
14861 2017-02-14 Andrew Pinski <apinski@cavium.com>
14863 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
14864 cores and change the partno/implementer to be correct.
14865 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
14866 the 'B" as the implementer.
14867 * config/aarch64/aarch64-tune.md: Regenerate.
14869 2017-02-14 Carl Love <cel@us.ibm.com>
14871 * config/rs6000/rs6000.c: Add case statement entry to make the
14872 xvcvuxdsp built-in argument unsigned.
14873 * config/rs6000/vsx.md: Fix the source and return operand types so they
14874 match the instruction definitions from the ISA document. Fix typo
14875 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
14878 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
14881 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
14882 member early_clobber_alts.
14883 * lra-lives.c (reg_early_clobber_p): New.
14884 (process_bb_lives): Use it.
14885 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
14886 (debug_operand_data): Initialize early_clobber_alts.
14887 (setup_operand_alternative): Set up early_clobber_alts.
14888 (collect_non_operand_hard_regs): Ditto. Pass early clobber
14889 alternatives to new_insn_reg.
14890 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
14892 (lra_update_insn_regno_info): Pass the new arg.
14894 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14896 PR middle-end/79505
14897 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
14898 (new_oacc_loop_raw): Don't clear already cleared fields.
14901 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
14902 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
14903 _mm512_prefetch_i64gather_ps): New inline functions and macros.
14905 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
14908 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
14910 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
14913 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
14914 the extra instruction to the right place to store 128-bit constant
14917 2017-02-14 Martin Sebor <msebor@redhat.com>
14919 PR middle-end/79448
14920 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
14921 warning for strings of unknown length.
14923 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
14925 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
14927 2017-02-14 Jeff Law <law@redhat.com>
14930 * ira-costs.c (scan_one_insn): Initialize register move costs
14931 for pseudos seen in USE/CLOBBER insns.
14933 PR tree-optimization/79095
14934 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
14935 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
14936 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
14937 if the operands are known to be not equal, then the resulting range
14939 (intersect_ranges): If the new range is ~[0,0] and the old range is
14940 wide, then prefer ~[0,0].
14941 * tree-vrp.c (overflow_comparison_p_1): New function.
14942 (overflow_comparison_p): New function.
14943 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
14944 if NAME is used in an overflow test.
14945 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
14946 overflow check that can be expressed as an equality test, then adjust
14947 ops to be that equality test.
14949 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14951 * config/s390/s390-builtin-types.def: Remove flags argument.
14952 * config/s390/s390.c (s390_init_builtins): Likewise.
14954 2017-02-14 Martin Liska <mliska@suse.cz>
14956 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
14957 vector. Fix trailing white spaces.
14959 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
14961 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
14964 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14966 PR rtl-optimization/68664
14967 * config/arm/arm.c (arm_sched_can_speculate_insn):
14968 New function. Declare prototype.
14969 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
14971 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14973 PR rtl-optimization/68664
14974 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
14976 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
14978 2017-02-14 Amit Pawar <amit.pawar@amd.com>
14980 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
14981 max skip bytes for function, loop and jump.
14983 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14985 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
14986 ABS_EXPR for gimple dump.
14988 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14991 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
14993 PR tree-optimization/79408
14994 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
14995 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
14996 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
14997 also if rhs1 is INTEGER_CST.
14999 2017-02-14 Richard Biener <rguenther@suse.de>
15001 PR middle-end/79432
15002 * tree-into-ssa.c (insert_phi_nodes): When the function can
15003 have abnormal edges rewrite SSA names with broken use-def
15004 dominance out of SSA and register them for PHI insertion.
15006 2017-02-13 Martin Sebor <msebor@redhat.com>
15008 PR middle-end/79496
15009 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15010 clearing info.nowrite flag when snprintf size argument is a range.
15012 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15014 * cprop.c (cprop_jump): Add missing space in string literal.
15015 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15016 (get_constraint_for_component_ref): Likewise.
15017 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15018 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15019 * lra-constraints.c (process_alt_operands): Likewise.
15020 * ipa-inline.c (inline_small_functions): Likewise.
15021 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15022 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15023 * trans-mem.c (diagnose_tm_1_op): Likewise.
15024 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15025 (grid_parallel_clauses_gridifiable): Likewise.
15027 * config/nvptx/mkoffload.c (process): Add space in between
15030 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15031 "MOD4_SSE_REGS" and "ALL_REGS".
15033 * spellcheck.c (test_data): Add , in between "foo" and "food".
15035 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15038 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15039 boundary crossing check and subsequent code generation agree.
15041 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15043 * config/aarch64/aarch64.c (has_memory_op): Delete.
15044 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15047 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15049 PR rtl-optimization/79388
15050 PR rtl-optimization/79450
15051 * combine.c (distribute_notes): When removing TEM_INSN for which
15052 corresponding dest has last value recorded, invalidate that last
15055 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15057 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15058 of explicit '@'. Add missing assembly comment marker on branch costs
15061 2017-02-13 Nathan Sidwell <nathan@acm.org>
15063 * gengtype-lex.l (<in_struct>): Add '/'.
15065 2017-02-13 Martin Liska <mliska@suse.cz>
15068 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15070 2017-02-13 Richard Biener <rguenther@suse.de>
15072 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15074 * configure: Re-generate.
15075 * config.in: Likewise.
15076 * graphite-dependences.c: Simplify as if
15077 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15078 * graphite-isl-ast-to-gimple.c: Likewise.
15079 * graphite-optimize-isl.c: Likewise.
15080 * graphite-poly.c: Likewise.
15081 * graphite-sese-to-poly.c: Likewise.
15082 * graphite.h: Likewise.
15083 * toplev.c: Include isl/version.h and use isl_version () for
15084 printing the ISL version.
15085 * doc/install.texi: Update ISL requirement.
15087 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15089 * doc/standards.texi (Standards): Update reference to
15092 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15094 * doc/extend.texi (Named Address Spaces): sourceware.org now
15096 * doc/install.texi (Binaries): Ditto.
15099 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15101 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15102 standard terminology "stringize"/"stringizing" throughout.
15103 * doc/cppinternals.texi: Likewise.
15105 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15107 * doc/extend.texi: Fix some spelling mistakes and typos.
15108 * doc/invoke.texi: Likewise.
15110 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15113 * params.def (inline-min-speedup) Change from 10 to 8.
15115 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15117 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15120 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15123 * ipa-inline-analysis.c (get_minimal_bb): New function.
15124 (record_modified): Use it.
15125 (remap_edge_change_prob): Handle also ancestor functions.
15127 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15129 * doc/contrib.texi (Contributors): Remove broken link into
15130 the Mauve CVS repository.
15132 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15134 PR middle-end/79454
15135 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15136 result computation whenever lhs doesn't have vector mode, not
15137 just when it has BLKmode.
15139 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15141 * doc/makefile.texi (profiledbootstrap): Refer to the
15142 installation instructions only in textual form.
15144 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15147 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15149 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15151 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15152 (Specific): Update mingw-w64 reference.
15154 (Specific): Remove broken link to Renesas RX processor.
15156 2017-02-10 Richard Biener <rguenther@suse.de>
15158 * toplev.c (process_options): Do not mention obsolete graphite
15159 options when printing sorry message about missing graphite support.
15160 Mention -floop-nest-optimize.
15162 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15164 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15165 (vtst_p16): Likewise.
15166 (vtstq_p8): Likewise.
15167 (vtstq_p16): Likewise.
15169 (vtstq_p64): Likewise.
15170 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15171 (vset_lane_p64): New.
15172 (vsetq_lane_p64): New.
15174 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15176 PR tree-optimization/79411
15177 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15178 stmt operands are SSA_NAMEs used in abnormal phis.
15179 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15182 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15185 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15188 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15190 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15192 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15194 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15198 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15199 not arbitrary TREE_CONSTANT.
15202 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15203 "omp declare target link" attribute unless is_global_var.
15204 * omp-offload.c (find_link_var_op): Likewise.
15206 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15207 Chung-Lin Tang <cltang@codesourcery.com>
15209 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15211 (gimplify_adjust_omp_clauses): Don't delete TILE.
15212 (gimplify_omp_for): Deal with TILE.
15213 * internal-fn.c (expand_GOACC_TILE): New function.
15214 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15216 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15217 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15219 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15220 avoid DIV for outermost collapse var.
15221 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15222 Remove out of date comments, fix whitespace.
15223 * omp-general.c (omp_extract_for_data): Deal with tiling.
15224 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15225 adjust OLF_DIM_BASE value.
15226 (struct omp_for_data): Add tiling field.
15227 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15228 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15229 for auto loops. Remove default auto determining, moved to
15230 oacc_loop_fixed_partitions.
15231 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15232 stmts, add e_mask field.
15233 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15234 (oacc_thread_numbers): Use oacc_dim_call.
15235 (oacc_xform_tile): New.
15236 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15237 (finish_oacc_loop): Adjust for ifns vector.
15238 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15239 add case for GOACC_TILE fns.
15240 (oacc_loop_xform_loop): Delete.
15241 (oacc_loop_process): Iterate over call list directly, and add
15242 handling for GOACC_TILE fns.
15243 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15245 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15246 vector partitioning to outer loops. Assign 2 partitions to loops
15247 when available. Add TILE handling.
15248 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15249 (execite_oacc_device_lower): Process GOACC_TILE fns,
15250 ignore unknown specs.
15251 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15252 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15253 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15255 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15257 * configure.ac (ACX_BUGURL): Update.
15258 * configure: Regenerate.
15260 2017-02-09 Richard Biener <rguenther@suse.de>
15262 PR tree-optimization/69823
15263 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15264 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15266 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15268 * config/arc/arc-c.def: Add __NPS400__ definition.
15269 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15270 (TARGET_NPS400): Define.
15272 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15274 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15276 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15277 pointer, arch_info.
15278 (arc_cpu_types): Fill the arch_info field with a pointer into the
15279 arc_arch_types table.
15280 (arc_selected_cpu): Declare.
15281 * config/arc/arc.c (arc_selected_cpu): Make global.
15282 (arc_selected_arch): Delete.
15283 (arc_base_cpu): Delete.
15284 (arc_override_options): Remove references to deleted variables,
15285 update access to arch information.
15286 (ARC_OPT): Update access to arch information.
15287 (ARC_OPTX): Likewise.
15288 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15289 (TARGET_ARC600): Update access to arch information.
15290 (TARGET_ARC601): Likewise.
15291 (TARGET_ARC700): Likewise.
15292 (TARGET_EM): Likewise.
15293 (TARGET_HS): Likewise.
15294 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15297 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15300 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15301 condition/operands for integer GE/LE/GEU/LEU operations.
15303 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15305 PR translation/79397
15306 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15309 2017-02-08 Martin Jambor <mjambor@suse.cz>
15312 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15313 whether allocation happened.
15314 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15315 nothing was allocated.
15317 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15319 PR tree-optimization/79408
15320 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15321 constant, but SSA_NAME with a known integer range, use the minimum
15322 of that range instead of op1 to determine if modulo can be replaced
15323 with its first operand.
15325 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15327 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15329 2017-02-08 Richard Biener <rguenther@suse.de>
15331 PR tree-optimization/71824
15332 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15333 Check all loops contained in the merged region.
15335 2017-02-07 Andrew Pinski <apinski@cavium.com>
15337 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15339 2017-02-07 Andrew Pinski <apinski@cavium.com>
15341 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15342 (thunderxt88): Likewise.
15343 (thunderxt81): Disable LSE and change v8.1 to v8.
15344 (thunderxt83): Likewise.
15346 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15347 Richard Biener <rguenther@suse.de>
15349 PR middle-end/79399
15350 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15351 type from int to size_t.
15352 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15354 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15356 PR rtl-optimization/79386
15357 * cprop.c (bypass_conditional_jumps): Initialize
15358 bypass_last_basic_block already before splitting bbs after
15359 unconditional traps...
15360 (bypass_conditional_jumps): ... rather than here.
15363 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15364 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15365 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15366 fix -masm=intel patterns.
15368 2017-02-07 Richard Biener <rguenther@suse.de>
15370 PR tree-optimization/79256
15371 PR middle-end/79278
15372 * builtins.c (get_object_alignment_2): Use min_align_of_type
15373 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15374 and ADJUST_FIELD_ALIGN.
15376 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15378 * doc/tm.texi: Regenerate.
15379 * stor-layout.c (layout_decl): Adjust.
15380 (update_alignment_for_field): Likewise.
15381 (place_field): Likewise.
15382 (min_align_of_type): Likewise.
15383 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15384 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15385 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15386 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15387 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15388 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15389 * config/i386/i386.c (x86_field_alignment): Likewise.
15390 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15391 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15392 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15393 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15394 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15395 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15399 2017-01-30 Richard Biener <rguenther@suse.de>
15401 PR tree-optimization/79256
15402 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15403 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15406 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15408 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15409 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15410 builtins to SImode and emit a zero-extend, if necessary.
15412 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15414 * docs/invoke.texi (RISC-V Options): Alphabetize.
15416 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15418 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15421 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15423 * config/riscv/riscv.c: New file.
15424 * common/config/riscv/riscv-common.c: Likewise.
15425 * config.gcc: Likewise.
15426 * config/riscv/constraints.md: Likewise.
15427 * config/riscv/elf.h: Likewise.
15428 * config/riscv/generic.md: Likewise.
15429 * config/riscv/linux.h: Likewise.
15430 * config/riscv/multilib-generator: Likewise.
15431 * config/riscv/peephole.md: Likewise.
15432 * config/riscv/pic.md: Likewise.
15433 * config/riscv/predicates.md: Likewise.
15434 * config/riscv/riscv-builtins.c: Likewise.
15435 * config/riscv/riscv-c.c: Likewise.
15436 * config/riscv/riscv-ftypes.def: Likewise.
15437 * config/riscv/riscv-modes.def: Likewise.
15438 * config/riscv/riscv-opts.h: Likewise.
15439 * config/riscv/riscv-protos.h: Likewise.
15440 * config/riscv/riscv.h: Likewise.
15441 * config/riscv/riscv.md: Likewise.
15442 * config/riscv/riscv.opt: Likewise.
15443 * config/riscv/sync.md: Likewise.
15444 * config/riscv/t-elf-multilib: Likewise.
15445 * config/riscv/t-linux: Likewise.
15446 * config/riscv/t-linux-multilib: Likewise.
15447 * config/riscv/t-riscv: Likewise.
15448 * configure.ac: Likewise.
15449 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15450 Waterman as RISC-V maintainers.
15451 * doc/install.texi: Add RISC-V entries.
15452 * doc/invoke.texi: Add RISC-V options section.
15453 * doc/md.texi: Add RISC-V constraints section.
15454 * configure: Regenerated.
15456 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15459 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15460 false values to be constant vectors with all 0 or all 1 bits set.
15461 (vcondu<mode><mode>): Likewise.
15462 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15464 (fpmask_comparison_operator): Update comment.
15465 (vecint_comparison_operator): New predicate.
15466 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15467 vector conditionals when the true and false values are constant
15468 vectors with all 0 bits or all 1 bits set.
15470 2017-02-06 Martin Sebor <msebor@redhat.com>
15472 PR tree-optimization/79376
15473 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15475 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15477 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15478 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15479 to simplify split condition.
15481 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15483 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15484 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15487 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15489 PR rtl-optimization/68664
15490 * target.def (can_speculate_insn): New hook.
15491 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15492 * doc/tm.texi: Regenerate.
15493 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15494 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15495 (rs6000_sched_can_speculate_insn): New function.
15497 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15499 PR tree-optimization/79284
15500 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15501 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15502 vectorizable_mask_load_store, vectorizable_operation,
15503 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15504 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15505 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15506 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15507 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15508 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15509 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15510 is_gimple_assign (stmt). Replace another such test with
15511 is_gimple_assign (stmt).
15513 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15516 * config/avr/avr.c (rtl-iter.h): Include it.
15517 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15518 (avr_legitimate_combined_insn): ...and implementation.
15520 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15522 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15523 * config/s390/s390.c (s390_const_operand_ok)
15524 (s390_canonicalize_comparison, s390_extract_part)
15525 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15526 (s390_contiguous_bitmask_p, s390_rtx_costs)
15527 (legitimize_pic_address): Likewise.
15528 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15529 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15530 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15531 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15532 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15534 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15536 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15537 REGNO($0) == REGNO($1).
15539 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15541 * config/s390/linux.h(SIZE_TYPE): Add comment.
15543 2017-02-06 Julian Brown <julian@codesourcery.com>
15544 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15545 Virendra Pathak <virendra.pathak@broadcom.com>
15547 * config/aarch64/aarch64-cores.def: Change the scheduler
15549 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15550 * config/aarch64/thunderx2t99.md: New file.
15552 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15554 * doc/standards.texi (Go Language): Update link to language
15557 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15559 * tree-eh.c (lower_resx): Sanitize profile.
15560 (cleanup_empty_eh_move_lp): Likewise.
15562 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15565 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15567 * cfgloopmanip.h (loop_version): Update prototype.
15568 * modulo-sched.c (sms_schedule): Update call of loop_version.
15569 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15570 * tree-parloops.c (gen_parallel_loop): Likewise.
15571 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15572 * tree-ssa-loop-split.c (split_loop): Likewise.
15573 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15574 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15576 2017-02-05 Martin Liska <mliska@suse.cz>
15579 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15581 (print_operand_address): Initialize a struct to zero.
15583 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15585 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15586 garbage collector only in textual form.
15588 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15590 * doc/extend.texi (x86 specific memory model extensions for
15591 transactional memory): Simplify a phrase.
15593 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15596 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15597 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15598 (atomic_storedi_1): Likewise.
15600 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15602 PR tree-optimization/79338
15603 * tree-parloops.c (gather_scalar_reductions): Don't call
15604 vect_analyze_loop_form for loop->inner before destroying loop's
15607 2017-02-03 Martin Sebor <msebor@redhat.com>
15609 PR tree-optimization/79327
15610 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15611 when precision has resulted in leading zeros.
15612 (format_integer): Adjust the likely counter to assume an unknown
15613 argument that may be zero is non-zero.
15615 2017-02-03 Jason Merrill <jason@redhat.com>
15618 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15619 avoid copying non-taken branch.
15621 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15623 PR tree-optimization/79340
15624 * tree-vect-loop.c (vectorizable_reduction): Release
15625 vec_defs elements after safe_splicing them into other vectors.
15628 PR tree-optimization/79327
15629 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15630 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15632 (format_integer): Use wide_int_to_tree instead of build_int_cst
15633 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15634 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15635 of shortest and longest sequence.
15637 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15639 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15640 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15642 2017-02-03 Walter Lee <walt@tilera.com>
15645 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15646 after initial stackframe link reg save.
15647 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15649 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15652 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15653 wu for stxssp alternative.
15655 2017-02-03 Martin Sebor <msebor@redhat.com>
15657 PR tree-optimization/79352
15658 * gimple-fold.c (get_range_strlen): Add argument.
15659 (get_range_strlen): Change return type to bool.
15660 (get_maxval_strlen): Pass in a dummy argument.
15661 * gimple-fold.h (get_range_strlen): Change return type to bool.
15662 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15663 * tree.h (array_at_struct_end_p): Add argument.
15664 * tree.c (array_at_struct_end_p): Handle it.
15666 2017-02-03 Martin Liska <mliska@suse.cz>
15669 * multiple_target.c (create_dispatcher_calls): Redirect edge
15670 from a caller of a dispatcher.
15671 (expand_target_clones): Make the clones local.
15672 (ipa_target_clone): Do both target clones and resolvers.
15673 (ipa_dispatcher_calls): Remove the pass.
15674 (pass_dispatcher_calls::gate): Likewise.
15675 (make_pass_dispatcher_calls): Likewise.
15676 * passes.def (pass_target_clone): Put as very first IPA early
15679 2017-02-03 Martin Liska <mliska@suse.cz>
15681 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15682 in case of a function with ifunc attribute.
15684 2017-02-03 Martin Liska <mliska@suse.cz>
15686 * cgraph.c (cgraph_node::dump): Dump function version info.
15687 * symtab.c (symtab_node::dump_base): Add missing new line.
15689 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15691 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15692 (ifcombine_ifandif): Use it.
15694 2017-02-03 Martin Liska <mliska@suse.cz>
15696 * doc/invoke.texi: Document default value for
15697 use-after-scope-direct-emission-threshold.
15699 2017-02-03 Martin Liska <mliska@suse.cz>
15701 PR tree-optimization/79339
15702 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15703 (format_floating): Likewise.
15705 2017-02-03 Martin Liska <mliska@suse.cz>
15708 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15710 (ipa_node_params_t::remove): Likewise.
15711 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15712 initialization from removed ipa_node_params_t::insert.
15713 (ipa_node_params::~ipa_node_params): Move from removed
15714 ipa_node_params_t::release.
15715 * symbol-summary.h (symbol_summary::m_released): New member.
15716 Do not release a summary twice. Do not allow to call finalizer
15717 for types of a summary that live in GGC memory.
15719 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15721 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15724 2017-02-02 Martin Sebor <msebor@redhat.com>
15726 PR middle-end/79275
15727 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15728 (format_string): Tighten up the range of output for non-constant
15729 strings and correct the expected range for wide non-constant strings.
15731 2017-02-02 Martin Sebor <msebor@redhat.com>
15733 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15735 PR middle-end/32003
15736 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15738 (-fdump-tree-@var): Add to index and document how to come up
15739 with pass-specific option and dump file names.
15740 (-fdump-passes): Clarify where to look for output.
15742 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15744 PR middle-end/77445
15745 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15746 statistics of the analyzed path; allow threading for speed when
15747 any of BBs along the path are optimized for speed.
15749 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15751 PR middle-end/78468
15752 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15753 settings of the virtual registers.
15756 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15758 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15759 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15762 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15764 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15765 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15767 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15769 * config/s390/s390.md: Add missing comments with the expanded
15771 * config/s390/vector.md: Likewise.
15772 * config/s390/vx-builtins.md: Likewise.
15774 2017-02-02 Jakub Jelinek <jakub@redhat.com>
15777 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
15778 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
15779 conditions on a single line.
15781 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15783 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15784 __S390_VX__ to __VX__.
15786 2017-02-01 Andrew Pinski <apinski@cavium.com>
15788 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
15789 stmt_info to record_stmt_cost.
15790 (vect_get_known_peeling_cost): Pass stmt_info if known to
15792 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
15793 cpu_vector_cost field into
15794 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
15795 field into vec_int_stmt_cost and vec_fp_stmt_cost.
15796 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
15797 splitting of scalar_stmt_cost and vec_stmt_cost.
15798 (thunderx_vector_cost): Likewise.
15799 (cortexa57_vector_cost): LIkewise.
15800 (exynosm1_vector_cost): Likewise.
15801 (xgene1_vector_cost): Likewise.
15802 (thunderx2t99_vector_cost): Improve after the splitting of the two
15804 (aarch64_builtin_vectorization_cost): Update for the splitting of
15805 scalar_stmt_cost and vec_stmt_cost.
15807 2017-02-01 Torvald Riegel <triegel@redhat.com>
15808 Richard Henderson <rth@redhat.com>
15810 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
15811 conditional on existance of a fast atomic load.
15812 * optabs-query.c (can_atomic_load_p): New function.
15813 * optabs-query.h (can_atomic_load_p): Declare it.
15814 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
15815 no fast atomic load is available for the particular size of access.
15816 (expand_atomic_compare_and_swap): Likewise.
15817 (expand_atomic_load): Likewise.
15818 (expand_atomic_store): Likewise.
15819 (expand_atomic_fetch_op): Likewise.
15820 * testsuite/lib/target-supports.exp
15821 (check_effective_target_sync_int_128): Remove x86 because it provides
15822 no fast atomic load.
15823 (check_effective_target_sync_int_128_runtime): Likewise.
15825 2017-02-01 Richard Biener <rguenther@suse.de>
15827 * graphite.c: Include tree-vectorizer.h for find_loop_location.
15828 (graphite_transform_loops): Provide opt-info for optimized nests.
15829 * tree-parloop.c (parallelize_loops): Provide opt-info for
15830 parallelized loops.
15832 2017-02-01 Richard Biener <rguenther@suse.de>
15834 PR middle-end/79315
15835 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
15836 was not set before.
15838 2017-02-01 Richard Biener <rguenther@suse.de>
15840 PR tree-optimization/71824
15841 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15842 Verify the loops are valid in the merged SESE region.
15843 (scop_detection::can_represent_loop_1): Check analyzing the
15844 evolution of the number of iterations in the region succeeds.
15846 2017-01-31 Ian Lance Taylor <iant@golang.org>
15848 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
15849 REG_ARGS_SIZE note to 32-bit push insns and call insn.
15851 2017-01-31 David Malcolm <dmalcolm@redhat.com>
15853 PR preprocessor/79210
15854 * input.c (get_substring_ranges_for_loc): Replace line_width
15855 assertion with error-handling.
15857 2017-01-31 Richard Biener <rguenther@suse.de>
15859 PR tree-optimization/77318
15860 * graphite-sese-to-poly.c (extract_affine): Fix assert.
15861 (create_pw_aff_from_tree): Take loop parameter.
15862 (add_condition_to_pbb): Pass loop of the condition to
15863 create_pw_aff_from_tree.
15865 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15867 * config/s390/s390.c (s390_asan_shadow_offset): New function.
15868 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
15870 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
15874 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
15876 (convert_int_to_float128): Likewise.
15877 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
15878 (convert_int_to_float128): Likewise.
15879 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
15880 (UNSPEC_IEEE128_CONVERT): Likewise.
15881 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
15882 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
15883 Use local variables for IBM extended format.
15884 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
15885 (fix_trunc<mode>si2_fprs): Likewise.
15886 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
15887 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
15888 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
15889 to know that we can now have integers of all sizes in vector
15891 (fix<uns>_<mode>di2_hw): Likewise.
15892 (float<uns>_<mode>si2_hw): Likewise.
15893 (fix_<mode>si2_hw): Likewise.
15894 (fixuns_<mode>si2_hw): Likewise.
15895 (float<uns>_<mode>di2_hw): Likewise.
15896 (float_<mode>di2_hw): Likewise.
15897 (float_<mode>si2_hw): Likewise.
15898 (floatuns_<mode>di2_hw): Likewise.
15899 (floatuns_<mode>si2_hw): Likewise.
15900 (xscvqp<su>wz_<mode>): Delete, no longer used.
15901 (xscvqp<su>dz_<mode>): Likewise.
15902 (xscv<su>dqp_<mode>): Likewise.
15903 (ieee128_mfvsrd_64bit): Likewise.
15904 (ieee128_mfvsrd_32bit): Likewise.
15905 (ieee128_mfvsrwz): Likewise.
15906 (ieee128_mtvsrw): Likewise.
15907 (ieee128_mtvsrd_64bit): Likewise.
15908 (ieee128_mtvsrd_32bit): Likewise.
15910 2017-01-31 Martin Liska <mliska@suse.cz>
15913 * ipa-prop.c (ipa_free_all_node_params): Call release method
15914 instead of ~sumbol_summary to not to trigger double times
15917 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
15919 PR tree-optimization/71691
15920 * bitmap.h (class auto_bitmap): New.
15921 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
15922 is_maybe_undefined instead of ssa_undefined_value_p.
15924 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15926 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15927 __S390_ARCH_LEVEL__ to __ARCH__.
15929 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15931 PR tree-optimization/79267
15932 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
15933 if should_remove_lhs_p is true.
15935 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
15938 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
15939 (add_alignment_attribute): New.
15940 (base_type_die): Add alignment attribute.
15941 (subrange_type_die): Likewise.
15942 (modified_type_die): Likewise.
15943 (gen_array_type_die): Likewise.
15944 (gen_descr_array_type_die: Likewise.
15945 (gen_enumeration_type_die): Likewise.
15946 (gen_subprogram_die): Likewise.
15947 (gen_variable_die): Likewise.
15948 (gen_field_die): Likewise.
15949 (gen_ptr_to_mbr_type_die): Likewise.
15950 (gen_struct_or_union_type_die): Likewise.
15951 (gen_subroutine_type_die): Likewise.
15952 (gen_typedef_die): Likewise.
15953 (base_type_cmp): Compare alignment attribute.
15955 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15958 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
15959 (setb_unsigned) New pattern for setb with CCUNS.
15960 * config/rs6000/rs6000.c (expand_block_compare): Use a different
15961 subfc./subfe sequence to avoid overflow problems. Generate a
15962 shorter sequence with cmpld/setb for power9.
15963 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
15964 for generating subfc. instruction.
15965 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
15966 now uses this instruction.
15968 2017-01-30 Ian Lance Taylor <iant@google.com>
15971 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
15972 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
15974 2017-01-30 Martin Sebor <msebor@redhat.com>
15976 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
15977 Move constant to the right of a relational operator.
15978 (get_mpfr_format_length, format_character, format_string): Ditto.
15979 (should_warn_p, maybe_warn): Same.
15981 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
15983 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
15986 * asan.c (get_translation_unit_decl): Remove function.
15987 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
15989 2017-01-30 Martin Liska <mliska@suse.cz>
15991 PR gcov-profile/79259
15992 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
15993 -fprofile-generate.
15995 2017-01-30 Martin Liska <mliska@suse.cz>
15998 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
15999 Initialize variables with NULL value.
16001 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16004 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16006 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16008 2017-01-30 Richard Biener <rguenther@suse.de>
16010 PR tree-optimization/79276
16011 * tree-vrp.c (process_assert_insertions): Properly adjust common
16012 when removing a duplicate.
16014 * gcc.dg/torture/pr79276.c: New testcase.
16016 2017-01-30 Richard Biener <rguenther@suse.de>
16018 PR tree-optimization/79256
16019 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16020 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16022 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16024 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16027 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16028 ("*r<noxa>sbg_<mode>_sll_bitmask")
16029 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16030 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16031 Use contiguous_bitmask_nowrap_operand.
16033 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16036 * config/rs6000/altivec.h (vec_xl): Revise #define.
16037 (vec_xst): Likewise.
16039 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16041 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16043 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16045 PR rtl-optimization/79194
16046 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16047 traps before call to bypass_conditional_jumps.
16049 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16051 PR tree-optimization/71374
16052 * lra-constraints.c (check_conflict_input_operands): New.
16053 (match_reload): Use it.
16055 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16058 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16059 account to calculate conflict_set.
16061 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16063 PR rtl-optimization/78559
16064 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16065 other_insn in combine.
16067 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16069 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16070 uint16_type_node for BT_UINT16.
16072 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16074 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16075 "RTL Tests" to menu.
16076 (GIMPLE Tests): New node.
16077 (RTL Tests): New node.
16079 2017-01-27 Richard Biener <rguenther@suse.de>
16081 PR tree-optimization/79245
16082 * tree-loop-distribution.c (distribute_loop): Apply cost
16083 modeling also to detected patterns.
16085 2017-01-27 Richard Biener <rguenther@suse.de>
16087 PR tree-optimization/71433
16088 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16089 (compare_assert_loc): New function.
16090 (process_assert_insertions): Sort and optimize assert locations
16091 to remove duplicates and push down identical assertions on
16092 edges to their destination block.
16094 2017-01-27 Richard Biener <rguenther@suse.de>
16096 PR tree-optimization/79244
16097 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16098 out SSA names even if abnormal.
16100 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16102 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16103 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16104 instead of MPFR_RNDN.
16106 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16109 * arm.c (arm_option_override): Don't call build_target_option_node
16110 until after doing all option overrides.
16111 (arm_valid_target_attribute_tree): Likewise.
16113 2017-01-27 Martin Liska <mliska@suse.cz>
16115 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16116 for {cd}tors and C++ {cd}tors.
16118 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16120 * config/s390/s390.md ("*setmem_long_and")
16121 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16123 2017-01-26 Martin Sebor <msebor@redhat.com>
16125 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16128 2017-01-26 Martin Sebor <msebor@redhat.com>
16130 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16131 HAVE_DFmode before using XFmode or DFmode.
16132 (parse_directive): Avoid using the z length modifier to avoid
16133 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16135 PR middle-end/78703
16136 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16137 to accept adjustment as an array.
16138 (get_int_range): New function.
16139 (struct directive): Make width and prec arrays.
16140 (directive::set_width, directive::set_precision): Call get_int_range.
16141 (format_integer, format_floating): Handle width and precision ranges.
16142 (format_string, parse_directive): Same.
16144 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16147 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16148 comdat_type_p set, just clone them, but keep the children in the
16152 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16153 which have direct callers with -fvar-tracking-assignments enabled
16155 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16156 inside of type units.
16158 2017-01-26 Martin Sebor <msebor@redhat.com>
16160 PR middle-end/78703
16161 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16163 (struct format_result): Replace number_chars, number_chars_min,
16164 and number_chars_max with a single member of struct result_range.
16166 (format_result::operator+=): Adjust.
16167 (struct fmtresult): Remove bounded. Handle likely and unlikely
16169 (fmtresult::adjust_for_width_or_precision): New function.
16170 (fmtresult:type_max_digits): New function.
16171 (bytes_remaining): Handle likely and unlikely counters.
16172 (min_bytes_remaining): Remove.
16173 (format_percent): Simplify.
16174 (format_integer, format_floating): Set likely and unlikely counters.
16175 (get_string_length, format_character, format_string): Same.
16176 (format_plain, should_warn_p): New function.
16177 (maybe_warn): Call should_warn_p. Update diagnostic messages
16178 and handle those for all directives, including plain strings.
16179 (format_directive): Handle likely and unlikely counters.
16180 Remove unnecessary quoting from diagnostics. Add an informational
16182 (add_bytes): Remove.
16183 (pass_sprintf_length::compute_format_length): Simplify.
16184 (try_substitute_return_value): Handle likely and unlikely counters.
16186 2017-01-26 Carl Love <cel@us.ibm.com>
16188 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16189 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16191 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16194 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16195 endianess for subregs into account.
16196 * lra-constraints.c (lra_constraints): Do risky transformations
16197 always on the first iteration.
16198 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16199 last_call_used_reg_set.
16200 (process_bb_lives): Define and use last_call_used_reg_set.
16201 * lra.c (lra): Always continue after lra_constraints on the first
16204 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16206 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16208 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16210 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16212 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16213 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16214 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16215 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16216 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16217 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16218 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16219 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16220 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16222 2017-01-26 Marek Polacek <polacek@redhat.com>
16225 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16226 for the third operand.
16228 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16230 PR middle-end/79236
16231 * omp-low.c (struct omp_context): Add simt_stmt field.
16232 (scan_omp_for): Return omp_context *.
16233 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16234 context to the _simt_ SIMD stmt.
16235 (lower_omp_for): For combined SIMD with sibling _simt_
16236 SIMD, make sure to use the same decls in _looptemp_
16237 clauses as in the sibling.
16239 2017-01-26 David Sherwood <david.sherwood@arm.com>
16241 PR middle-end/79212
16242 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16245 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16248 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16249 emit fld b; fld a; if possible.
16251 * brig-builtins.def: Update copyright years.
16252 * config/arm/arm_acle_builtins.def: Update copyright years.
16254 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16257 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16258 constraint instead of o for the stxsd instruction.
16260 2017-01-25 Carl Love <cel@us.ibm.com>
16262 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16263 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16265 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16267 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16269 2017-01-25 Richard Biener <rguenther@suse.de>
16271 PR tree-optimization/69264
16272 * target.def (vector_alignment_reachable): Improve documentation.
16273 * doc/tm.texi: Regenerate.
16274 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16276 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16277 earlier changes with respect to TYPE_USER_ALIGN.
16278 (vector_alignment_reachable_p): Likewise. Improve dumping.
16280 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16283 * config/arm/arm.md (xordi3): Force constant operand into a register
16286 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16288 * doc/invoke.texi (-fstore-merging): Correct default optimization
16289 levels at which it is enabled.
16290 (-O): Move -fstore-merging from list to...
16293 2017-01-25 Richard Biener <rguenther@suse.de>
16296 * omp-expand.c: Include debug.h.
16297 (expand_omp_taskreg): Make sure to generate early debug before
16298 outlining anything from a function.
16299 (expand_omp_target): Likewise.
16300 (grid_expand_target_grid_body): Likewise.
16302 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16305 * asan.c (get_translation_unit_decl): New function.
16306 (asan_add_global): Extract modules file name from globals
16307 TRANSLATION_UNIT_DECL name.
16309 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16312 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16313 for long calls with APCS frame and VFP.
16315 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16317 * cfg.c (original_copy_tables_initialized_p): New function.
16318 * cfg.h (original_copy_tables_initialized_p): New decl.
16319 * cfgrtl.c (relink_block_chain): Guard the call to
16320 free_original_copy_tables with a call to
16321 original_copy_tables_initialized_p.
16322 * cgraph.h (symtab_node::native_rtl_p): New decl.
16323 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16324 (symtab_node::needed_p): Don't assert for early assembly output
16325 for __RTL functions.
16326 (cgraph_node::finalize_function): Set "force_output" for __RTL
16328 (cgraph_node::analyze): Bail out early for __RTL functions.
16329 (analyze_functions): Update assertion to support __RTL functions.
16330 (cgraph_node::expand): Bail out early for __RTL functions.
16331 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16333 * function.h (struct function): Update comment for field
16335 * gimple-expr.c: Include "tree-pass.h".
16336 (gimple_has_body_p): Return false for __RTL functions.
16337 * Makefile.in (OBJS): Add run-rtl-passes.o.
16338 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16340 (gcc::pass_manager::get_clean_slate): New accessor.
16341 * passes.c: Include "insn-addr.h".
16342 (should_skip_pass_p): Add logging. Update logic for running
16343 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16344 property-provider override so it is only done for gimple passes.
16346 (skip_pass): New function.
16347 (execute_one_pass): Call skip_pass when skipping passes.
16348 * read-md.c (md_reader::read_char): Support filtering
16349 the input to a subset of line numbers.
16350 (md_reader::md_reader): Initialize fields
16351 m_first_line and m_last_line.
16352 (md_reader::read_file_fragment): New function.
16353 * read-md.h (md_reader::read_file_fragment): New decl.
16354 (md_reader::m_first_line): New field.
16355 (md_reader::m_last_line): New field.
16356 * read-rtl-function.c (function_reader::create_function): Only
16357 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16358 curr_properties. Set DECL_INITIAL to a dummy block.
16359 (read_rtl_function_body_from_file_range): New function.
16360 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16362 * run-rtl-passes.c: New file.
16363 * run-rtl-passes.h: New file.
16365 2017-01-24 Jeff Law <law@redhat.com>
16367 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16370 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16372 PR tree-optimization/79159
16373 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16374 (record_nonwrapping_iv): Improve boundary using above function if no
16375 value range information.
16377 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16378 Martin Jambor <mjambor@suse.cz>
16380 * brig-builtins.def: New file.
16381 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16382 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16383 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16384 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16385 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16386 * builtin-types.def (BT_INT8): New.
16387 (BT_INT16): Likewise.
16388 (BT_UINT8): Likewise.
16389 (BT_UINT16): Likewise.
16390 (BT_FN_ULONG): Likewise.
16391 (BT_FN_UINT_INT): Likewise.
16392 (BT_FN_UINT_ULONG): Likewise.
16393 (BT_FN_UINT_LONG): Likewise.
16394 (BT_FN_UINT_PTR): Likewise.
16395 (BT_FN_ULONG_PTR): Likewise.
16396 (BT_FN_INT8_FLOAT): Likewise.
16397 (BT_FN_INT16_FLOAT): Likewise.
16398 (BT_FN_UINT32_FLOAT): Likewise.
16399 (BT_FN_UINT16_FLOAT): Likewise.
16400 (BT_FN_UINT8_FLOAT): Likewise.
16401 (BT_FN_UINT64_FLOAT): Likewise.
16402 (BT_FN_UINT16_UINT32): Likewise.
16403 (BT_FN_UINT32_UINT16): Likewise.
16404 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16405 (BT_FN_INT_PTR_INT): Likewise.
16406 (BT_FN_UINT_PTR_UINT): Likewise.
16407 (BT_FN_LONG_PTR_LONG): Likewise.
16408 (BT_FN_ULONG_PTR_ULONG): Likewise.
16409 (BT_FN_VOID_UINT64_UINT64): Likewise.
16410 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16411 (BT_FN_INT8_INT8_INT8): Likewise.
16412 (BT_FN_INT16_INT16_INT16): Likewise.
16413 (BT_FN_INT_INT_INT): Likewise.
16414 (BT_FN_UINT_FLOAT_UINT): Likewise.
16415 (BT_FN_FLOAT_UINT_UINT): Likewise.
16416 (BT_FN_ULONG_UINT_UINT): Likewise.
16417 (BT_FN_ULONG_UINT_PTR): Likewise.
16418 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16419 (BT_FN_UINT_UINT_UINT): Likewise.
16420 (BT_FN_VOID_UINT_PTR): Likewise.
16421 (BT_FN_UINT_UINT_PTR: Likewise.
16422 (BT_FN_UINT32_UINT64_PTR): Likewise.
16423 (BT_FN_INT_INT_UINT_UINT): Likewise.
16424 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16425 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16426 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16427 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16428 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16429 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16430 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16431 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16432 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16433 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16434 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16435 * doc/frontends.texi: List BRIG FE.
16436 * doc/install.texi (Testing): Add BRIG tesring requirements.
16437 * doc/invoke.texi (Overall Options): Mention BRIG.
16438 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16440 2017-01-24 Richard Biener <rguenther@suse.de>
16442 PR translation/79208
16443 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16445 2017-01-24 Martin Jambor <mjambor@suse.cz>
16448 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16449 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16450 and known_contexts.
16452 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16454 PR middle-end/79123
16455 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16456 casts from signed to unsigned really don't have a range.
16458 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16460 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16461 GMP_RNDx for compatiblity.
16463 2017-01-24 Martin Liska <mliska@suse.cz>
16466 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16467 that would prevent us to call alloca with -1 as argument.
16469 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16471 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16472 Avoid -Wformat-security warning.
16474 2017-01-23 Andrew Pinski <apinski@cavium.com>
16476 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16479 2017-01-23 Martin Sebor <msebor@redhat.com>
16481 PR middle-end/78703
16482 * gimple-ssa-sprintf.c (warn_level): New global.
16483 (format_integer): Use it here and throughout the rest of the file.
16484 Use the same switch to compute sign as base.
16485 (maybe_warn): New function.
16486 (format_directive): Factor out warnings into maybe_warn.
16487 Add debugging output. Use warn_level.
16488 (add_bytes): Use warn_level.
16489 (pass_sprintf_length::compute_format_length): Add debugging output.
16490 (try_substitute_return_value): Same.
16491 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16493 PR middle-end/78703
16494 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16495 (struct fmtresult, format_integer, format_floating): Adjust.
16496 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16497 (get_string_length, format_string,format_directive): Same.
16498 (pass_sprintf_length::compute_format_length): Same.
16499 (try_substitute_return_value): Simplify slightly.
16501 PR middle-end/78703
16502 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16503 (fmtresult::operator+=): Outlined.
16504 (struct fmtresult): Add ctors.
16505 (struct conversion_spec): Rename...
16506 (struct directive): ...to this. Add and remove data members.
16507 (directive::set_width, directive::set_precision): New functions.
16508 (format_percent): Use fmtresult ctor.
16509 (get_width_and_precision): Remove.
16510 (format_integer): Make naming changes. Avoid computing width and
16512 (format_floating): Same. Adjust indentation.
16513 (format_character, format_none): New functions.
16514 (format_string): Moved character handling to format_character.
16515 (format_directive): Remove arguments, change return type.
16516 (parse_directive): New function.
16517 (pass_sprintf_length::compute_format_length): Move directive
16518 parsing to parse_directive.
16520 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16522 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16523 (assign_assembler_name_if_needed): ... this.
16524 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16525 (assign_assembler_name_if_needed): ... this.
16526 (free_lang_data_in_cgraph): Adjust callers.
16527 * cgraphunit.c (cgraph_node::analyze): Likewise.
16528 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16531 2017-01-23 Richard Biener <rguenther@suse.de>
16533 PR tree-optimization/79088
16534 PR tree-optimization/79188
16535 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16536 resetting loop bounds after last path deletion. Reset loop
16537 bounds of the target loop, make code match the comments.
16538 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16539 Make sure loops need no fixups.
16541 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16543 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16544 exponent support with double type for first argument.
16545 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16546 type returned by __builtin_vec_extract_sig,
16547 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16548 functions from "vector int" to "vector unsigned int" or from
16549 "vector long long int" to "vector unsigned long long int".
16550 Changed type returned by __builtin_vec_extract_exp,
16551 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16552 functions from "vector int" to "vector unsigned int" or from
16553 "vector long long int" to "vector unsigned long long int".
16554 Changed return type of __builtin_vec_test_data_class,
16555 __builtin_vec_test_data_class_sp, and
16556 __builtin_vec_test_data_class_dp from "vector int" to
16557 "vector bool int" or from "vector long long int" to "vector bool
16558 long long int" and changed second argument type from "unsigned
16559 int" to "int". Added new overloaded function forms "vector float
16560 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16561 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16562 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16563 double, vector unsigned long long int)" and "vector double
16564 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16565 long int)". Changed return type of
16566 __builtin_scalar_test_data_class and
16567 __builtin_scalar_test_data_class_sp and
16568 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16569 int" and changed second argument from "unsigned int" to "int".
16570 Changed type returned by __builtin_scalar_test_neg,
16571 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16572 from "int" to "bool int". Added new overloaded function form
16573 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16574 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16575 exponent double-precision with floating point first argument.
16576 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16577 documentation of scalar_test_data_class, scalar_test_neg,
16578 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16579 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16580 vec_test_data_class built-in functions to reflect refinements in
16581 their type signatures.
16583 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16585 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16587 (aarch64_elf_asm_destructor): Likewise.
16589 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16591 PR rtl-optimization/78634
16592 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16593 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16594 * ifcvt.c (noce_try_cmove): Add missing cost check.
16596 PR rtl-optimization/71724
16597 * combine.c (if_then_else_cond): Look for situations where it is
16598 beneficial to undo the work of one of the recursive calls.
16600 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16602 PR tree-optimization/70754
16603 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16604 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16605 combined stmt before it if not NULL.
16606 (combine_chains): Process refs reversely and compute dominance point
16609 2017-01-23 Martin Liska <mliska@suse.cz>
16611 PR tree-optimization/79196
16612 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16613 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16615 (strlen_optimize_stmt): Call the renamed function.
16617 2017-01-23 Michael Matz <matz@suse.de>
16619 PR tree-optimization/78384
16620 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16622 2017-01-23 Richard Biener <rguenther@suse.de>
16624 PR tree-optimization/79186
16625 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16626 both incoming edges before moving an assert.
16628 2017-01-23 Martin Jambor <mjambor@suse.cz>
16630 * ipa-prop.c (load_from_param_1): Removed.
16631 (load_from_unmodified_param): Bits from load_from_param_1 put back
16633 (load_from_param): Removed.
16634 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16635 with stmt. Reverted back to use of load_from_unmodified_param.
16637 2017-01-23 Martin Jambor <mjambor@suse.cz>
16640 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16641 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16642 field a pointer to garbage collected vector, mark lattices and
16643 ipcp_orig_node with GTY((skip)).
16644 (ipa_get_param_count): Adjust to descriptors being a pointer.
16645 (ipa_get_param): Likewise.
16646 (ipa_get_type): Likewise.
16647 (ipa_get_param_move_cost): Likewise.
16648 (ipa_set_param_used): Likewise.
16649 (ipa_get_controlled_uses): Likewise.
16650 (ipa_set_controlled_uses): Likewise.
16651 (ipa_is_param_used): Likewise.
16652 (ipa_node_params_t): Move into garbage collector. New methods insert
16654 (ipa_node_params_sum): Annotate wth GTY(()).
16655 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16657 (ipa_load_from_parm_agg): Adjust declaration.
16658 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16659 * ipa-profile.c (ipa_profile): Likewise.
16660 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16661 (ipa_populate_param_decls): Make descriptors parameter garbage
16663 (ipa_dump_param): Adjust to descriptors being a pointer.
16664 (ipa_alloc_node_params): Likewise.
16665 (ipa_initialize_node_params): Likewise.
16666 (load_from_param_1): Make descriptors parameter garbage collected.
16667 (load_from_unmodified_param): Likewise.
16668 (load_from_param): Likewise.
16669 (ipa_load_from_parm_agg): Likewise.
16670 (ipa_node_params::~ipa_node_params): Removed.
16671 (ipa_free_all_node_params): Remove call to delete operator.
16672 (ipa_node_params_t::insert): New.
16673 (ipa_node_params_t::remove): Likewise.
16674 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16675 copy known_csts and known_contexts vectors.
16676 (ipa_read_node_info): Adjust to descriptors being a pointer.
16677 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16679 (ipcp_transform_function): Make descriptors variable garbage
16682 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16684 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16685 * config/i386/avx512dqintrin.h: Ditto.
16686 * config/i386/avx512fintrin.h: Ditto.
16687 * config/i386/i386.c: Handle new builtins.
16688 * config/i386/i386-builtin.def: Add new builtins.
16689 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16690 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16692 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16693 Martin Liska <mliska@suse.cz>
16695 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16696 * asan.c (asan_expand_poison_ifn): Support stores and use
16697 appropriate ASAN report function.
16698 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16699 * internal-fn.def (ASAN_POISON_USE): Declare.
16700 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16701 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16702 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16703 ASAN_POISON calls w/o LHS.
16704 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16705 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16706 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16707 * gimplify.c (asan_poison_variables): Add attribute
16708 use_after_scope_memory to variables that really needs to live
16710 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16711 having the attribute.
16713 2017-01-23 Martin Liska <mliska@suse.cz>
16715 * asan.c (create_asan_shadow_var): New function.
16716 (asan_expand_poison_ifn): Likewise.
16717 * asan.h (asan_expand_poison_ifn): New declaration.
16718 * internal-fn.c (expand_ASAN_POISON): Likewise.
16719 * internal-fn.def (ASAN_POISON): New builtin.
16720 * sanopt.c (pass_sanopt::execute): Expand
16721 asan_expand_poison_ifn.
16722 * tree-inline.c (copy_decl_for_dup_finish): Make function
16724 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16725 * tree-ssa.c (is_asan_mark_p): New function.
16726 (execute_update_addresses_taken): Rewrite local variables
16727 (identified just by use-after-scope as addressable) into SSA.
16729 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16731 * doc/install.texi (Specific): opensource.apple.com uses https
16732 now. Remove trailing slash.
16734 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16736 * README.Portability: Remove note on an Irix compatibility issue.
16738 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16740 * gcov.c (INCLUDE_ALGORITHM): Define.
16741 (INCLUDE_VECTOR): Define.
16742 No longer include <vector> and <algorithm> directly.
16744 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16746 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16748 * doc/invoke.texi (Code Gen Options): Ditto.
16750 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16753 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16755 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16757 rtl-optimization/79125
16758 * cprop.c (local_cprop_pass): Handle cases where we make an
16759 unconditional trap.
16761 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16765 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16766 read from, for big endian.
16768 2017-01-20 Jiong Wang <jiong.wang@arm.com>
16770 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
16771 register pauth builtins for LP64 only.
16773 2017-01-20 Marek Polacek <polacek@redhat.com>
16776 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
16779 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16781 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
16783 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
16784 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
16785 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
16787 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16790 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
16791 in big-endian mode when they are not a single duplicated value.
16793 2017-01-20 Richard Biener <rguenther@suse.de>
16795 * BASE-VER: Bump to 7.0.1.
16797 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16799 * omp-low.c (omplow_simd_context): New struct. Use it...
16800 (lower_rec_simd_input_clauses): ...here and...
16801 (lower_rec_input_clauses): ...here to hold common data. Adjust all
16802 references to idx, lane, max_vf, is_simt.
16804 2017-01-20 Graham Markall <graham.markall@embecosm.com>
16806 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
16809 2017-01-20 Martin Jambor <mjambor@suse.cz>
16811 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
16812 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
16814 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
16815 (GTFILES): Rename hsa.c to hsa-common.c.
16816 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
16817 * hsa-dump.c: Likewise.
16818 * hsa-gen.c: Likewise.
16819 * hsa-regalloc.c: Likewise.
16820 * ipa-hsa.c: Likewise.
16821 * omp-expand.c: Likewise.
16822 * omp-low.c: Likewise.
16823 * toplev.c: Likewise.
16825 2017-01-20 Marek Polacek <polacek@redhat.com>
16828 * doc/invoke.texi: Document -Wduplicated-branches.
16829 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
16830 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
16831 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
16832 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
16833 return 0 only when not OEP_LEXICOGRAPHIC.
16834 (fold_build_cleanup_point_expr): Use the expression
16835 location when building CLEANUP_POINT_EXPR.
16836 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
16837 * tree.c (add_expr): Handle error_mark_node.
16839 2017-01-20 Martin Liska <mliska@suse.cz>
16842 * tree-profile.c (init_ic_make_global_vars): Do not call
16844 (gimple_init_gcov_profiler): Likewise.
16846 2017-01-20 Martin Liska <mliska@suse.cz>
16849 * cgraph.h (maybe_create_reference): Remove argument and
16851 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
16853 * ipa-cp.c (create_specialized_node): Likewise.
16854 * symtab.c (symtab_node::maybe_create_reference): Handle
16855 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
16857 2017-01-20 Martin Liska <mliska@suse.cz>
16859 * read-rtl-function.c (function_reader::create_function): Use
16860 build_decl instread of build_decl_stat.
16862 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
16864 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
16865 * config/i386/avx512dqintrin.h: Ditto.
16866 * config/i386/avx512fintrin.h: Ditto.
16867 * config/i386/i386-builtin-types.def: Add new types.
16868 * config/i386/i386.c: Handle new types.
16869 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
16870 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
16871 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
16872 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
16873 (__builtin_ia32_kshiftridi): New.
16874 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
16876 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
16880 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
16881 define to rs6000_init_stack_protect_guard.
16882 (rs6000_init_stack_protect_guard): New function.
16884 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16885 Yunqiang Su <yunqiang.su@imgtec.com>
16887 * config.gcc (supported_defaults): Add madd4.
16888 (with_madd4): Add validation.
16889 (all_defaults): Add madd4.
16890 * config/mips/mips.opt (mmadd4): New option.
16891 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16893 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
16895 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
16896 (ISA_HAS_FUSED_MADD4): Likewise.
16897 * doc/invoke.texi (-mmadd4): Document the new option.
16898 * doc/install.texi (--with-madd4): Document the new option.
16900 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16902 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
16903 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
16904 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
16905 (aarch64_init_pauth_hint_builtins): New.
16906 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
16907 (aarch64_expand_builtin): Expand new builtins.
16909 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16911 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
16912 * combine-stack-adj.c (no_unhandled_cfa): Handle
16913 REG_CFA_TOGGLE_RA_MANGLE.
16914 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
16915 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
16916 info for return address signing.
16917 (aarch64_expand_epilogue): Likewise.
16919 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16921 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
16922 * config/aarch64/aarch64-protos.h
16923 (aarch64_return_address_signing_enabled): New declaration.
16924 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
16926 (aarch64_expand_prologue): Sign return address before it's pushed onto
16928 (aarch64_expand_epilogue): Authenticate return address fetched from
16930 (aarch64_override_options): Sanity check for ILP32 and ISA level.
16931 (aarch64_attributes): New function attributes for "sign-return-address".
16932 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
16933 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
16934 ("*do_return"): Generate combined instructions according to key index.
16935 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
16936 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
16938 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
16939 * config/aarch64/aarch64.opt (msign-return-address=): New.
16940 * doc/extend.texi (AArch64 Function Attributes): Documents
16941 "sign-return-address=".
16942 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
16944 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16946 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
16947 overall option summary.
16949 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16951 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
16952 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
16953 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
16954 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
16956 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
16958 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
16959 -mpower9-minmax by default for -mcpu=power9.
16960 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
16961 128-bit floating point.
16963 2017-01-20 Alan Modra <amodra@gmail.com>
16965 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
16966 optimizing for size.
16968 2017-01-20 Alan Modra <amodra@gmail.com>
16971 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
16972 for strcmp and strncmp from corresponding builtin decl.
16974 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
16976 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
16977 instead of i386/rtems-64.h.
16978 * config/i386/rtems-64.h: Remove.
16980 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
16984 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
16986 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
16988 2017-01-19 Tamar Christina <tamar.christina@arm.com>
16990 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
16991 Change int to HOST_WIDE_INT.
16992 * config/aarch64/aarch64-protos.h
16993 (aarch64_simd_gen_const_vector_dup): Likewise.
16994 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
16996 2017-01-19 David Malcolm <dmalcolm@redhat.com>
16998 * langhooks-def.h (lhd_type_for_size): New decl.
16999 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17000 * langhooks.c (lhd_type_for_size): New function, taken from
17003 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17005 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17006 define_bypass for CR latency.
17007 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17008 (power9-alu2): Add define_bypass for CR latency.
17010 (power9-mul): Update insn latency.
17011 (power9-mul-compare): Update insn latency, bypass latency and remove
17014 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17016 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17018 * config/aarch64/aarch64.md
17019 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17020 aarch64_nopcrelative_literal_loads.
17021 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17023 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17025 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17026 TARGET_LOONGSON_3A.
17027 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17029 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17032 * config.gcc (supported_defaults): Add lxc1-sxc1.
17033 (with_lxc1_sxc1): Add validation.
17034 (all_defaults): Add lxc1-sxc1.
17035 * config/mips/mips.opt (mlxc1-sxc1): New option.
17036 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17038 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17039 __mips_no_lxc1_sxc1.
17040 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17041 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17042 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17044 2017-01-19 Richard Biener <rguenther@suse.de>
17046 PR tree-optimization/72488
17047 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17048 sure to restore SSA info.
17049 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17051 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17053 PR rtl-optimization/79121
17054 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17055 of the inner type when shifting an extended value.
17057 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17060 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17061 interposable aliases.
17063 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17066 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17067 Use the evmergelohi instruction.
17068 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17069 (mov_sitf_e500_subreg8_2_be): Likewise.
17070 (mov_sitf_e500_subreg12_2_le): Likewise.
17071 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17072 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17073 (mov_sitf_e500_subreg8_2_le): Likewise.
17074 (mov_sitf_e500_subreg12_2_be): Likewise.
17076 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17078 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17079 attribute from vecsimple to vecperm.
17080 (altivec_vbpermq2): Likewise.
17082 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17085 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17087 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17088 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17089 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17090 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17091 case where N arg is SIZE_MAX.
17092 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17093 (cmpstrsi): Add pattern.
17095 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17097 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17098 __builtin_vec_revb builtins.
17099 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17100 built-in functions to support generation of the ISA 3.0 XXBR<x>
17101 vector byte reverse instructions.
17102 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17103 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17104 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17105 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17106 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17107 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17108 (P9V_BUILTIN_VEC_REVB): Likewise.
17109 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17110 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17111 (p9_xxbrq_v16qi): Likewise.
17112 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17113 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17114 (p9_xxbrh_v8hi): Likewise.
17115 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17116 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17117 vec_revb built-in functions.
17119 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17121 PR rtl-optimization/78952
17122 * config/i386/i386.md (any_extract): New code iterator.
17123 (*insvqi_2): Use any_extract for source operand.
17124 (*insvqi_3): Use any_shiftrt for source operand.
17126 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17128 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17130 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17132 2017-01-18 Matthias Klose <doko@ubuntu.com>
17134 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17136 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17138 * config/rs6000/altivec.h (vec_bperm): Change #define.
17139 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17140 (altivec_vbpermq2): New define_insn.
17141 (altivec_vbpermd): Likewise.
17142 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17143 function interface.
17144 (VBPERMD): Likewise.
17145 (VBPERM): New polymorphic function interface.
17146 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17147 Add entries for P9V_BUILTIN_VEC_VBPERM.
17148 * doc/extend.texi: Add interfaces for vec_bperm.
17150 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17152 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17153 first letter of error messages.
17154 (s390_resolve_overloaded_builtin): Likewise.
17155 * config/s390/s390.c (s390_expand_builtin): Likewise.
17156 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17157 (s390_valid_target_attribute_inner_p): Likewise.
17158 * config/s390/s390.md ("tabort"): Likewise.
17160 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17162 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17163 (ISA_AVOID_DIV_HILO): New macro.
17164 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17165 (ISA_HAS_DDIV): Likewise.
17167 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17169 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17171 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17173 * doc/invoke.texi (fabi-version): Spelling fix.
17175 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17178 * doc/invoke.texi (fabi-version): Mention mangling fix for
17181 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17184 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17186 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17189 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17190 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17192 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17194 (stack_protect_test): Ditto.
17195 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17196 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17198 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17199 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17200 -mstack-protector-guard-offset=.
17201 (RS/6000 and PowerPC Options): Ditto.
17203 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17205 * config/i386/i386.h (MASK_CLASS_P): New define.
17206 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17207 there are no registers from different register sets also when
17208 mask registers are used. Update function comment.
17209 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17210 to (*k/*r) and (*k/*km) alternatives.
17212 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17214 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17215 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17216 (EH_RETURN_HANDLER_RTX): New define.
17217 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17218 Force frame pointer in EH return functions.
17219 (aarch64_expand_epilogue): Add barrier for eh_return.
17220 (aarch64_final_eh_return_addr): Remove.
17221 (aarch64_eh_return_handler_rtx): New function.
17222 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17224 (aarch64_eh_return_handler_rtx): New prototype.
17226 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17228 * config/rs6000/altivec.h (vec_rlmi): New #define.
17229 (vec_vrlnm): Likewise.
17230 (vec_rlnm): Likewise.
17231 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17232 (UNSPEC_VRLNM): Likewise.
17233 (VIlong): New mode iterator.
17234 (altivec_vrl<VI_char>mi): New define_insn.
17235 (altivec_vrl<VI_char>nm): Likewise.
17236 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17238 (VRLDNM): Likewise.
17239 (RLNM): New polymorphic function entry.
17240 (VRLWMI): New monomorphic function entry.
17241 (VRLDMI): Likewise.
17242 (RLMI): New polymorphic function entry.
17243 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17244 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17245 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17248 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17251 * dwarf2out.c (field_byte_offset): Restore the
17252 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17253 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17254 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17257 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17260 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17262 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17265 * dwarf2out.c (add_data_member_location_attribute): For constant
17266 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17267 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17268 DW_AT_byte_size attributes.
17270 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17272 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17273 after forcing to constant memory when the code model is medium.
17275 2017-01-17 Julia Koval <julia.koval@intel.com>
17278 * config/i386/avx512fintrin.h
17279 (_mm512_i32gather_ps): Change __addr type to void const*.
17280 (_mm512_mask_i32gather_ps): Ditto.
17281 (_mm512_i32gather_pd): Ditto.
17282 (_mm512_mask_i32gather_pd): Ditto.
17283 (_mm512_i64gather_ps): Ditto.
17284 (_mm512_mask_i64gather_ps): Ditto.
17285 (_mm512_i64gather_pd): Ditto.
17286 (_mm512_mask_i64gather_pd): Ditto.
17287 (_mm512_i32gather_epi32): Ditto.
17288 (_mm512_mask_i32gather_epi32): Ditto.
17289 (_mm512_i32gather_epi64): Ditto.
17290 (_mm512_mask_i32gather_epi64): Ditto.
17291 (_mm512_i64gather_epi32): Ditto.
17292 (_mm512_mask_i64gather_epi32): Ditto.
17293 (_mm512_i64gather_epi64): Ditto.
17294 (_mm512_mask_i64gather_epi64): Ditto.
17295 (_mm512_i32scatter_ps): Change __addr type to void*.
17296 (_mm512_mask_i32scatter_ps): Ditto.
17297 (_mm512_i32scatter_pd): Ditto.
17298 (_mm512_mask_i32scatter_pd): Ditto.
17299 (_mm512_i64scatter_ps): Ditto.
17300 (_mm512_mask_i64scatter_ps): Ditto.
17301 (_mm512_i64scatter_pd): Ditto.
17302 (_mm512_mask_i64scatter_pd): Ditto.
17303 (_mm512_i32scatter_epi32): Ditto.
17304 (_mm512_mask_i32scatter_epi32): Ditto.
17305 (_mm512_i32scatter_epi64): Ditto.
17306 (_mm512_mask_i32scatter_epi64): Ditto.
17307 (_mm512_i64scatter_epi32): Ditto.
17308 (_mm512_mask_i64scatter_epi32): Ditto.
17309 (_mm512_i64scatter_epi64): Ditto.
17310 (_mm512_mask_i64scatter_epi64): Ditto.
17311 * config/i386/avx512pfintrin.h
17312 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17313 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17314 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17315 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17316 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17317 (_mm512_prefetch_i32scatter_ps): Ditto.
17318 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17319 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17320 (_mm512_prefetch_i64scatter_pd): Ditto.
17321 (_mm512_prefetch_i64scatter_ps): Ditto.
17322 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17323 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17324 * config/i386/avx512vlintrin.h
17325 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17326 (_mm_mmask_i32gather_ps): Ditto.
17327 (_mm256_mmask_i32gather_pd): Ditto.
17328 (_mm_mmask_i32gather_pd): Ditto.
17329 (_mm256_mmask_i64gather_ps): Ditto.
17330 (_mm_mmask_i64gather_ps): Ditto.
17331 (_mm256_mmask_i64gather_pd): Ditto.
17332 (_mm_mmask_i64gather_pd): Ditto.
17333 (_mm256_mmask_i32gather_epi32): Ditto.
17334 (_mm_mmask_i32gather_epi32): Ditto.
17335 (_mm256_mmask_i32gather_epi64): Ditto.
17336 (_mm_mmask_i32gather_epi64): Ditto.
17337 (_mm256_mmask_i64gather_epi32): Ditto.
17338 (_mm_mmask_i64gather_epi32): Ditto.
17339 (_mm256_mmask_i64gather_epi64): Ditto.
17340 (_mm_mmask_i64gather_epi64): Ditto.
17341 (_mm256_i32scatter_ps): Change __addr type to void*.
17342 (_mm256_mask_i32scatter_ps): Ditto.
17343 (_mm_i32scatter_ps): Ditto.
17344 (_mm_mask_i32scatter_ps): Ditto.
17345 (_mm256_i32scatter_pd): Ditto.
17346 (_mm256_mask_i32scatter_pd): Ditto.
17347 (_mm_i32scatter_pd): Ditto.
17348 (_mm_mask_i32scatter_pd): Ditto.
17349 (_mm256_i64scatter_ps): Ditto.
17350 (_mm256_mask_i64scatter_ps): Ditto.
17351 (_mm_i64scatter_ps): Ditto.
17352 (_mm_mask_i64scatter_ps): Ditto.
17353 (_mm256_i64scatter_pd): Ditto.
17354 (_mm256_mask_i64scatter_pd): Ditto.
17355 (_mm_i64scatter_pd): Ditto.
17356 (_mm_mask_i64scatter_pd): Ditto.
17357 (_mm256_i32scatter_epi32): Ditto.
17358 (_mm256_mask_i32scatter_epi32): Ditto.
17359 (_mm_i32scatter_epi32): Ditto.
17360 (_mm_mask_i32scatter_epi32): Ditto.
17361 (_mm256_i32scatter_epi64): Ditto.
17362 (_mm256_mask_i32scatter_epi64): Ditto.
17363 (_mm_i32scatter_epi64): Ditto.
17364 (_mm_mask_i32scatter_epi64): Ditto.
17365 (_mm256_i64scatter_epi32): Ditto.
17366 (_mm256_mask_i64scatter_epi32): Ditto.
17367 (_mm_i64scatter_epi32): Ditto.
17368 (_mm_mask_i64scatter_epi32): Ditto.
17369 (_mm256_i64scatter_epi64): Ditto.
17370 (_mm256_mask_i64scatter_epi64): Ditto.
17371 (_mm_i64scatter_epi64): Ditto.
17372 (_mm_mask_i64scatter_epi64): Ditto.
17373 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17374 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17375 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17376 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17377 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17378 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17379 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17380 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17381 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17382 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17383 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17384 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17385 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17386 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17387 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17388 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17389 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17390 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17391 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17392 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17393 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17394 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17395 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17396 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17397 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17398 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17399 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17400 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17401 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17402 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17403 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17404 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17405 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17406 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17407 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17408 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17409 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17410 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17411 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17412 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17413 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17414 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17415 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17416 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17417 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17418 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17419 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17420 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17421 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17422 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17423 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17424 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17425 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17426 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17427 definitions accordingly.
17429 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17430 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17433 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17436 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17439 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17440 hard regs for inner regno.
17442 2017-01-17 Martin Liska <mliska@suse.cz>
17445 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17446 assumption and add comment.
17448 2017-01-17 Nathan Sidwell <nathan@acm.org>
17450 * ipa-visibility.c (localize_node): New function, broken out of ...
17451 (function_and_variable_visibility): ... here. Call it.
17453 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17455 PR middle-end/77445
17456 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17457 correctly set frequency of oudgoing edge.
17458 (duplicate_thread_path): Fix profile updating.
17460 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17463 * configure.ac: Add GCC_BASE_VER.
17464 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17465 version from BASE-VER file.
17466 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17467 (gcc.o): Depend on $(BASEVER).
17468 * common.opt (dumpfullversion): New option.
17469 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17470 * doc/invoke.texi: Document -dumpfullversion.
17471 * doc/install.texi: Document --with-gcc-major-version-only.
17472 * configure: Regenerated.
17474 2017-01-17 Richard Biener <rguenther@suse.de>
17476 PR tree-optimization/71433
17477 * tree-vrp.c (register_new_assert_for): Merge same asserts
17478 on all incoming edges.
17479 (process_assert_insertions_for): Handle insertions at the
17482 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17484 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17485 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17487 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17490 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17493 2017-01-17 Alan Modra <amodra@gmail.com>
17496 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17497 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17498 symbolic stack limit when pic.
17500 2017-01-16 Martin Sebor <msebor@redhat.com>
17502 PR tree-optimization/78608
17503 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17505 2017-01-16 Jeff Law <law@redhat.com>
17508 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17509 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17510 for several include directories that may be relative to sysroot.
17511 * config/i386/x-mingw32 (gplus_includedir): Define.
17512 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17513 (native_system_includedir): Likewise.
17514 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17515 override if TARGET_SYSTEM_ROOT is defined.
17516 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17518 PR tree-optimization/79090
17519 PR tree-optimization/33562
17520 PR tree-optimization/61912
17521 PR tree-optimization/77485
17522 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17523 and computed trims into the dump file.
17525 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17527 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17529 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17532 * gimplify.c (gimplify_init_constructor): If want_value and
17533 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17537 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17538 sequence. Formatting fixes.
17539 (doloop_optimize): Formatting fixes.
17542 * gcc.c (debug_level_greater_than_spec_func): New function.
17543 (static_spec_functions): Add debug-level-gt spec function.
17544 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17546 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17547 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17548 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17549 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17550 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17551 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17553 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17555 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17556 QImode fixups to general and mask registers only.
17558 2017-01-16 Carl Love <cel@us.ibm.com>
17560 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17561 for built-in functions
17562 vector signed char vec_nabs (vector signed char)
17563 vector signed short vec_nabs (vector signed short)
17564 vector signed int vec_nabs (vector signed int)
17565 vector signed long long vec_nabs (vector signed long long)
17566 vector float vec_nabs (vector float)
17567 vector double vec_nabs (vector double)
17568 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17570 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17571 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17572 * doc/extend.texi: Update the documentation file for the new built-in
17575 2017-01-16 Martin Sebor <msebor@redhat.com>
17577 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17580 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17582 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17583 UNSPEC_VSX__XXSPLTD to require special splat handling.
17585 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17588 * system.h: Poison strndup.
17590 2017-01-16 Alan Modra <amodra@gmail.com>
17593 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17596 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17598 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17600 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17602 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17603 call recog here. Assert that INSN_CODE (insn) is non-negative.
17605 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17608 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17610 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17611 in the currently scheduled RTL fragment.
17613 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17615 PR rtl-optimization/78751
17616 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17619 2017-01-14 Jeff Law <law@redhat.com>
17621 PR tree-optimization/79090
17622 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17623 variable length stores.
17624 (compute_trims): Delete dead assignment to *trim_tail.
17625 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17628 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17630 PR rtl-optimization/78626
17631 PR rtl-optimization/78727
17632 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17633 of a block, and split such blocks after everything else is finished.
17635 2017-01-14 Alan Modra <amodra@gmail.com>
17638 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17639 target legitimate_combined_insn.
17640 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17641 (rs6000_legitimate_combined_insn): New function.
17642 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17644 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17645 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17646 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17648 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17650 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17652 2017-01-13 Jeff Law <law@redhat.com>
17654 PR tree-optimization/33562
17655 PR tree-optimization/61912
17656 PR tree-optimization/77485
17657 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17659 (delete_dead_assignment): Likewise.
17660 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17661 statement to delete_dead_call and delete_dead_assignment.
17663 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17666 * substring-locations.c (format_warning_va): Strengthen case 1 so
17667 that both endpoints of the substring must be within the format
17668 range for just the substring to be printed.
17670 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17672 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17673 * config/i386/i386.c (ix86_target_string): Add missing options
17674 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17675 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17676 flags_other and ix86_target_other to flags2_other. Display unknown
17678 (ix86_valid_target_attribute_inner_p): Add missing options and
17679 reorder options by implied ISAs, as in ix86_target_string.
17681 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17683 * hash-table.h (hash_table::too_empty_p): New function.
17684 (hash_table::expand): Use it.
17685 (hash_table::traverse): Likewise.
17686 (hash_table::empty_slot): Use sizeof (value_type) instead of
17687 sizeof (PTR) to convert bytes to elements. Shrink the table
17688 if the current size is excessive for the current number of
17691 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17693 * ira-costs.c (record_reg_classes): Break from the inner loop
17694 early once alt_fail is known to be true. Update outer loop
17695 handling accordingly.
17697 2017-01-13 Jeff Law <law@redhat.com>
17699 * tree-ssa-dse.c (decrement_count): New function.
17700 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17701 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17702 when we know the partially dead statement is a mem* function.
17704 PR tree-optimization/61912
17705 PR tree-optimization/77485
17706 * tree-ssa-dse.c: Include expr.h.
17707 (maybe_trim_constructor_store): New function.
17708 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17710 PR tree-optimization/33562
17711 PR tree-optimization/61912
17712 PR tree-optimization/77485
17713 * doc/invoke.texi: Document new dse-max-object-size param.
17714 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17715 * tree-ssa-dse.c: Include params.h.
17716 (dse_store_status): New enum.
17717 (initialize_ao_ref_for_dse): New, partially extracted from
17719 (valid_ao_ref_for_dse, normalize_ref): New.
17720 (setup_live_bytes_from_ref, compute_trims): Likewise.
17721 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17722 (maybe_trim_partially_dead_store): Likewise.
17723 (maybe_trim_complex_store): Likewise.
17724 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17725 Track what bytes live from the original store. Return tri-state
17726 for dead, partially dead or live.
17727 (dse_dom_walker): Add constructor, destructor and new private members.
17728 (delete_dead_call, delete_dead_assignment): New extracted from
17730 (dse_optimize_stmt): Make a member of dse_dom_walker.
17731 Use initialize_ao_ref_for_dse.
17733 PR tree-optimization/33562
17734 PR tree-optimization/61912
17735 PR tree-optimization/77485
17736 * sbitmap.h (bitmap_count_bits): Prototype.
17737 (bitmap_clear_range, bitmap_set_range): Likewise.
17738 * sbitmap.c (bitmap_clear_range): New function.
17739 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17741 2017-01-13 Martin Liska <mliska@suse.cz>
17744 * function.c (set_cfun): Add new argument force.
17745 * function.h (set_cfun): Likewise.
17746 * ipa-inline-transform.c (inline_call): Use the function when
17747 strict alising from is dropped for function we inline to.
17749 2017-01-13 Richard Biener <rguenther@suse.de>
17751 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17752 for dumping GIMPLE INTEGER_CSTs.
17754 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17756 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17757 to 201112L since C++17.
17759 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17762 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17763 if -fsanitize=kernel-address is present.
17765 2017-01-13 Richard Biener <rguenther@suse.de>
17767 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
17768 as _Literal ( type ) number in case usual suffixes do not
17769 preserve all information.
17771 2017-01-13 Richard Biener <rguenther@suse.de>
17773 PR tree-optimization/77283
17774 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
17775 and ssa-iterators.h.
17776 (is_feasible_trace): Implement a cost model based on joiner
17779 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17782 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
17783 char or short to __float128/_Float128 directly.
17785 2017-01-12 Martin Sebor <msebor@redhat.com>
17787 to -Wformat-overflow.
17788 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
17789 (min_bytes_remaining): Same.
17790 (get_string_length): Same.
17791 (format_string): Same.
17792 (format_directive): Same.
17794 (pass_sprintf_length::handle_gimple_call): Same.
17796 2017-01-12 Jakub Jelinek <jakub@redhat.com>
17798 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
17799 info.nowrite calls with no lhs that can't throw. Return bool
17800 whether gsi_remove has been called or not.
17801 (pass_sprintf_length::handle_gimple_call): Return bool whether
17802 try_substitute_return_value called gsi_remove. Formatting fix.
17803 (pass_sprintf_length::execute): Don't use gsi_remove if
17804 handle_gimple_call returned true.
17807 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
17808 be removed due to side-effects, don't remove following barrier nor
17809 turn the successor edge into fallthru edge.
17811 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17814 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
17815 element-reversing loads and stores as not swappable.
17817 2017-01-12 Nathan Sidwell <nathan@acm.org>
17818 Nicolai Stange <nicstange@gmail.com>
17820 * combine.c (try_combine): Don't ignore result of overlap checking
17821 loop. Combine overlap & asm check into single loop.
17823 2017-01-12 Richard Biener <rguenther@suse.de>
17825 * tree-pretty-print.c (dump_generic_node): Provide -gimple
17826 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
17828 2017-01-12 Richard Biener <rguenther@suse.de>
17830 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
17831 and TS_TARGET_OPTION directly derive from TS_BASE.
17832 * tree-core.h (tree_optimization_option): Derive from tree_base.
17833 (tree_target_option): Likewise.
17835 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
17837 * config/i386/i386.c (memory_address_length): Increase len
17838 only when rip_relative_addr_p returns false.
17840 2017-01-11 Julia Koval <julia.koval@intel.com>
17842 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
17843 (OPTION_MASK_ISA_SGX_SET): New.
17844 (ix86_handle_option): Handle OPT_msgx.
17845 * config.gcc: Added sgxintrin.h.
17846 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
17847 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
17848 * config/i386/i386.c (ix86_target_string): Add -msgx.
17850 (ix86_option_override_internal): Handle new options.
17851 (ix86_valid_target_attribute_inner_p): Add sgx.
17852 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
17853 * config/i386/i386.opt: Add msgx.
17854 * config/i386/sgxintrin.h: New file.
17855 * config/i386/x86intrin.h: Add sgxintrin.h.
17857 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17860 * fold-const.c (maybe_nonzero_address): Return 1 for function
17862 (tree_single_nonzero_warnv_p): Don't handle function local objects
17866 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
17869 2017-01-11 David Malcolm <dmalcolm@redhat.com>
17872 * opts.c: Include "spellcheck.h"
17873 (struct string_fragment): New struct.
17874 (struct edit_distance_traits<const string_fragment &>): New
17876 (get_closest_sanitizer_option): New function.
17877 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
17879 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17881 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
17883 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
17884 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
17885 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
17886 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
17887 for initial die_offset if dwarf_split_debug_info.
17888 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
17889 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
17891 (output_skeleton_debug_sections): Formatting fix. Use
17892 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
17893 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
17895 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
17897 * config/arm/cortex-a53.md: Add bypasses for
17898 cortex_a53_r2f_cvt.
17899 (cortex_a53_r2f): Only use for transfers.
17900 (cortex_a53_f2r): Likewise.
17901 (cortex_a53_r2f_cvt): Add reservation for conversions.
17902 (cortex_a53_f2r_cvt): Likewise.
17904 2017-01-11 Tamar Christina <tamar.christina@arm.com>
17906 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
17907 to all inlined functions, change static to extern.
17909 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
17912 * config/arm/arm.c (legitimize_pic_address): Handle reference to
17914 (arm_assemble_integer): Likewise.
17916 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
17918 * config.gcc: Use new awk script to check CPU, FPU and architecture
17919 parameters for --with-... options.
17920 * config/arm/parsecpu.awk: New file
17921 * config/arm/arm-cpus.in: New file.
17922 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
17924 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
17926 * config/arm/t-arm: Update dependency rules.
17927 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
17928 of processing .def files.
17929 * config/arm/genopt.sh: Deleted.
17930 * config/arm/gentune.sh: Deleted.
17931 * config/arm/arm-cores.def: Deleted.
17932 * config/arm/arm-arches.def: Deleted.
17933 * config/arm/arm-fpus.def: Deleted.
17934 * config/arm/arm-tune.md: Regenerated.
17935 * config/arm/arm-tables.opt: Regenerated.
17936 * config/arm/arm-cpu.h: New generated file.
17937 * config/arm/arm-cpu-data.h: New generated file.
17938 * config/arm/arm-cpu-cdata.h: New generated file.
17940 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
17943 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
17945 (input_varpool_node): Unpack dynamically_initialized bit.
17947 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
17949 PR rtl-optimization/79032
17950 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
17951 the alignment of the adjusted memory reference against that of MODE,
17952 instead of the alignment of the original memory reference.
17954 2017-01-11 Martin Jambor <mjambor@suse.cz>
17956 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
17958 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
17959 decorated functions.
17961 2017-01-11 Richard Biener <rguenther@suse.de>
17963 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
17964 set range/nonnull info for PHI results. Do not set it on
17965 stmts marked for removal.
17967 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
17969 * expr.c (store_field): In the bitfield case, fetch the return value
17970 from the registers before applying a single big-endian adjustment.
17971 Always do a final load for a BLKmode value not larger than a word.
17973 2017-01-10 David Malcolm <dmalcolm@redhat.com>
17976 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
17977 that we correctly handle column numbers greater than
17978 LINE_MAP_MAX_COLUMN_NUMBER.
17980 2017-01-10 Martin Sebor <msebor@redhat.com>
17982 PR middle-end/78245
17983 * gimple-ssa-sprintf.c (get_destination_size): Call
17984 {init,fini}object_sizes.
17985 * tree-object-size.c (addr_object_size): Adjust.
17986 (pass_through_call): Adjust.
17987 (pass_object_sizes::execute): Adjust.
17988 * tree-object-size.h (fini_object_sizes): Declare.
17990 2017-01-10 Martin Sebor <msebor@redhat.com>
17992 PR tree-optimization/78775
17993 * builtins.c (get_size_range): Move...
17994 * calls.c: ...to here.
17995 (alloc_max_size): Accept zero argument.
17996 (operand_signed_p): Remove.
17997 (maybe_warn_alloc_args_overflow): Call get_size_range.
17998 * calls.h (get_size_range): Declare.
18000 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18002 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18003 from TI's devices.csv file as of September 2016.
18004 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18006 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18008 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18009 * doc/invoke.texi: Likewise.
18010 * doc/md.texi: Likewise.
18011 * doc/objc.texi: Likewise.
18013 2017-01-10 Joshua Conner <joshconner@google.com>
18015 * config/arm/fuchsia-elf.h: New file.
18016 * config/fuchsia.h: New file.
18017 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18018 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18020 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18022 2016-01-10 Richard Biener <rguenther@suse.de>
18024 PR tree-optimization/79034
18025 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18026 Propagate out degenerate PHIs in the joiner.
18028 2017-01-10 Martin Liska <mliska@suse.cz>
18030 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18031 (sort_congruence_classes_by_decl_uid): Likewise.
18032 (sort_congruence_class_groups_by_decl_uid): Likewise.
18033 (sem_item_optimizer::merge_classes): Sort class, groups in these
18034 classes and members in the groups by DECL_UID of declarations.
18035 This would make merge operations stable.
18037 2017-01-10 Martin Liska <mliska@suse.cz>
18039 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18040 usage of m_classes_vec.
18041 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18042 (sem_item_optimizer::get_group_by_hash): Likewise.
18043 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18044 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18045 (sem_item_optimizer::verify_classes): Likewise.
18046 (sem_item_optimizer::process_cong_reduction): Likewise.
18047 (sem_item_optimizer::dump_cong_classes): Likewise.
18048 (sem_item_optimizer::merge_classes): Likewise.
18049 * ipa-icf.h (congruence_class_hash): Rename from
18050 congruence_class_group_hash. Remove declaration of m_classes_vec.
18052 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18054 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18055 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18056 * config.gcc: Add avx512vpopcntdqintrin.h.
18057 * config/i386/avx512vpopcntdqintrin.h: New.
18058 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18059 * config/i386/i386-builtin-types.def: Add new types.
18060 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18061 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18062 __builtin_ia32_vpopcountq_v8di_mask): New.
18063 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18064 __AVX512VPOPCNTDQ__.
18065 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18066 (PTA_AVX512VPOPCNTDQ): Define.
18067 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18068 TARGET_AVX512VPOPCNTDQ_P): Define.
18069 * config/i386/i386.opt: Add mavx512vpopcntdq.
18070 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18071 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18073 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18075 PR middle-end/77484
18076 * predict.def (PRED_CALL): Set to 67.
18078 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18080 * expr.c (store_field): In the bitfield case, if the value comes from
18081 a function call and is of an aggregate type returned in registers, do
18082 not modify the field mode; extract the value in all cases if the mode
18083 is BLKmode and the size is not larger than a word.
18085 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18088 * config/i386/cpuid.h: Fix undefined behavior.
18090 2017-01-04 Jeff Law <law@redhat.com>
18092 PR tree-optimization/79007
18093 PR tree-optimization/67955
18094 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18095 conservative for pt.null when flag_non_call_exceptions is on.
18097 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18099 PR translation/79019
18100 PR translation/79020
18101 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18102 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18103 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18105 * config/avr/avr.opt (maccumulate-args): Likewise.
18106 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18107 * common.opt (freport-bug): Likewise.
18108 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18109 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18110 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18111 translatable string.
18112 * config/i386/i386.c (function_value_32): Likewise.
18113 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18114 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18116 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18117 * common/config/msp430/msp430-common.c (msp430_handle_option):
18119 * symtab.c (symtab_node::verify_base): Likewise.
18120 * opts.c (set_debug_level): Likewise.
18121 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18122 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18123 missing whitespace to translatable strings.
18124 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18125 * config/sh/superh.h: Likewise.
18126 * config/i386/xopintrin.h: Likewise.
18127 * config/i386/znver1.md: Likewise.
18128 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18129 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18130 * double-int.h (struct double_int): Likewise.
18131 * double-int.c (div_and_round_double): Likewise.
18132 * wide-int.cc: Likewise.
18133 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18134 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18135 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18136 (crossjumps_occurred): ... this.
18137 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18140 PR tree-optimization/78899
18141 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18142 returning bool return struct loop *, NULL for failure and the new
18144 (versionable_outer_loop_p): Don't version outer loop if it has
18145 dont_vectorized bit set.
18146 (tree_if_conversion): When versioning outer loop, ensure
18147 tree_if_conversion is performed also on the inner loop of the
18148 non-vectorizable outer loop copy.
18149 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18150 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18151 prevent vectorization of it.
18152 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18153 the outer loop vectorization of the non-scalar version is attempted
18154 before vectorization of the inner loop in scalar version. If
18155 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18156 vectorization of its inner loop.
18157 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18158 has 2 inner loops, rename also on edges from bb whose single pred
18159 is outer_loop->header. Fix typo in function comment.
18161 2017-01-09 Martin Sebor <msebor@redhat.com>
18164 * asan.c (asan_emit_stack_protection): Increase local buffer size
18165 to avoid snprintf truncation warning.
18167 2017-01-09 Andrew Pinski <apinski@cavium.com>
18169 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18170 to reference thunderx2t99 for the tuning structure
18171 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18173 (thunderx2t99_extra_costs): This.
18174 * config/aarch64/aarch64-tune.md: Regenerate.
18175 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18176 (vulcan_addrcost_table): This.
18177 (vulcan_regmove_cost): Rename to ...
18178 (thunderx2t99_regmove_cost): This.
18179 (vulcan_vector_cost): Rename to ...
18180 (thunderx2t99_vector_cost): this.
18181 (vulcan_branch_cost): Rename to ...
18182 (thunderx2t99_branch_cost): This.
18183 (vulcan_tunings): Rename to ...
18184 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18185 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18187 2017-01-09 Martin Jambor <mjambor@suse.cz>
18191 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18192 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18193 (propagate_vr_accross_jump_function): Use the above function for all
18194 value range computations for pass-through jump functions and type
18195 converasion from explicit value range values.
18196 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18197 parameters from TYPE_ARG_TYPES.
18198 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18199 (ipa_write_node_info): Stream type of the actual argument.
18200 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18202 2017-01-09 Martin Liska <mliska@suse.cz>
18205 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18206 (lookup_compiler): Do not show error message with have_E.
18208 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18210 PR tree-optimization/78938
18211 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18212 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18213 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18214 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18217 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18219 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18222 2017-01-09 Richard Biener <rguenther@suse.de>
18224 PR tree-optimization/78997
18225 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18226 name condition properly.
18228 2017-01-09 Richard Biener <rguenther@suse.de>
18231 * dwarf2out.c (is_cxx): New overload with context.
18232 (is_naming_typedef_decl): Use it.
18234 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18236 * invoke.texi (Option Summary): Correct spacing in option lists
18237 and add line breaks to fix over-long lines.
18239 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18241 PR middle-end/17660
18243 * extend.texi (Common Variable Attributes): Add xref to GCC
18244 Internals manual to explain mode attribute keywords.
18246 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18249 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18250 and Preprocessor Options.
18251 (Options for Linking): Document -pthread here....
18252 (RS/6000 and PowerPC Options): ...not here.
18253 (Solaris 2 Options): ...or here.
18254 * doc/cppopts.texi: Document -pthread.
18256 2017-01-08 Martin Sebor <msebor@redhat.com>
18258 PR middle-end/77708
18259 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18260 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18261 New member functions.
18262 (format_directive): Used them.
18264 (pass_sprintf_length::handle_gimple_call): Same.
18265 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18266 to avoid truncation for any argument.
18267 (extract_affine_mul): Same.
18268 * tree.c (get_file_function_name): Same.
18270 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18272 PR middle-end/77484
18273 * predict.def (PRED_INDIR_CALL): Set to 86.
18275 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18277 PR preprocessor/54124
18278 * doc/cppopts.texi: Reformat -d subtable to list the full name
18279 of the options. Add cross-reference to the docs for the general
18280 compiler -d options.
18281 * doc/invoke.texi (Developer Options): Add cross-reference to the
18282 preprocessor-specific -d option documentation.
18284 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18286 PR preprocessor/13498
18287 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18288 redudant material, and reflect new command-line options.
18289 (System Headers): Likewise.
18291 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18293 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18294 -isystem, and -idirafter. Copy-edit.
18295 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18296 default for -ftrack-macro-expansion. Delete obsolete and
18297 badly-formatted implementation details about -fdebug-cpp output.
18298 * doc/cppwarnopts.texi: Copy-edit.
18300 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18303 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18304 that the transition from a max line width >= 1<<10 to narrower
18305 lines works correctly.
18307 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18309 * doc/options.texi (PerFunction): New.
18310 * opt-functions.awk (switch_flags): Map both Optimization and
18311 PerFunction to CL_OPTIMIZATION.
18312 * opth-gen.awk: Test for PerFunction flag along with
18314 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18315 it only when the latter is present. Skip those that don't in
18316 the hash function generator.
18317 * common.opt (fvar-tracking): Mark as PerFunction instead of
18319 (fvar-tracking-assignments): Likewise.
18320 (fvar-tracking-assignments-toggle): Likewise.
18321 (fvar-tracking-uninit): Likewise.
18323 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18325 PR translation/79018
18326 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18329 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18332 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18333 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18334 TARGET_LONG_JUMP_TABLE_OFFSETS.
18335 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18336 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18337 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18338 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18339 * config/m68k/m68k.md (tablejump expander): Likewise.
18340 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18341 TARGET_LONG_JUMP_TABLE_OFFSETS.
18342 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18343 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18345 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18346 David Holsgrove <david.holsgrove@xilinx.com>
18348 * common/config/microblaze/microblaze-common.c
18349 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18350 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18352 * config/microblaze/microblaze.c (microblaze_must_save_register)
18353 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18355 (microblaze_eh_return): New function.
18356 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18357 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18358 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18359 * config/microblaze/microblaze.md (eh_return): New pattern.
18361 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18363 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18364 GCC_DIAGNOSTIC_STRINGIFY): Define.
18366 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18368 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18370 * config/arm/arm.md (<mcrr>): New.
18372 * config/arm/arm.c (arm_arch5te): New.
18373 (arm_option_override): Set arm_arch5te.
18374 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18376 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18377 (arm_mcrr_qualifiers): ... this. New.
18378 (MRRC_QUALIFIERS): Define to...
18379 (arm_mrrc_qualifiers): ... this. New.
18380 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18382 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18383 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18384 (MRRCI, mrrc, MRRC): New.
18385 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18386 VUNSPEC_MRRC2): New.
18388 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18390 * config/arm/arm.md (<mcr>): New.
18392 * config/arm/arm.c (arm_coproc_builtin_available): Add
18393 support for mcr, mrc, mcr2 and mrc2.
18394 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18395 (arm_mcr_qualifiers): ... this. New.
18396 (MRC_QUALIFIERS): Define to ...
18397 (arm_mrc_qualifiers): ... this. New.
18398 (MCR_QUALIFIERS): Define to ...
18399 (arm_mcr_qualifiers): ... this. New.
18400 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18402 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18403 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18404 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18405 VUNSPEC_MRC2): New.
18407 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18409 * config/arm/arm.md (*ldc): New.
18413 * config/arm/arm.c (arm_coproc_builtin_available): Add
18414 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18415 (arm_coproc_ldc_stc_legitimate_address): New.
18416 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18417 'qualifier_const_pointer'.
18418 (LDC_QUALIFIERS): Define to...
18419 (arm_ldc_qualifiers): ... this. New.
18420 (STC_QUALIFIERS): Define to...
18421 (arm_stc_qualifiers): ... this. New.
18422 * config/arm/arm-protos.h
18423 (arm_coproc_ldc_stc_legitimate_address): New.
18424 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18425 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18426 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18427 stc2, stcl, stc2l): New.
18428 * config/arm/constraints.md (Uz): New.
18429 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18430 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18431 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18432 VUNSPEC_STC2L): New.
18434 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18436 * config/arm/arm.md (<cdp>): New.
18437 * config/arm/arm.c (neon_const_bounds): Rename this ...
18438 (arm_const_bounds): ... this.
18439 (arm_coproc_builtin_available): New.
18440 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18441 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18442 (CDP_QUALIFIERS): Define to...
18443 (arm_cdp_qualifiers): ... this. New.
18445 (arm_expand_builtin_args): Add case for 6 arguments.
18446 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18447 (arm_const_bounds): ... this.
18448 (arm_coproc_builtin_available): New.
18449 * config/arm/arm_acle.h (__arm_cdp): New.
18451 * config/arm/arm_acle_builtins.def (cdp): New.
18453 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18454 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18455 'arm_const_bounds'.
18456 * config/arm/types.md (coproc): New.
18457 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18458 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18459 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18460 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18462 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18464 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18465 (UBINOP_QUALIFIERS): New.
18467 (acle_builtin_data): New. Change comment.
18468 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18469 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18470 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18471 arm_acle_builtins.def.
18472 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18473 (arm_init_acle_builtins): New.
18474 (CRC32_BUILTIN): Remove.
18475 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18476 crc32cb, crc32ch and crc32cw.
18477 (arm_init_crc32_builtins): Remove.
18478 (arm_init_builtins): Use arm_init_acle_builtins rather
18479 than arm_init_crc32_builtins.
18480 (arm_expand_acle_builtin): New.
18481 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18482 * config/arm/arm_acle_builtins.def: New.
18484 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18486 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18487 (arm_builtin_datum): ... this.
18488 (arm_init_neon_builtin): Rename to ...
18489 (arm_init_builtin): ... this. Add a new parameters PREFIX
18490 and USE_SIG_IN_NAME.
18491 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18492 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18493 'arm_builtin_datum'.
18494 (arm_init_vfp_builtins): Likewise.
18495 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18496 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18497 (arm_expand_neon_args): Rename to ...
18498 (arm_expand_builtin_args): ... this. Rename builtin_arg
18499 enum values and differentiate between ARG_BUILTIN_MEMORY
18500 and ARG_BUILTIN_NEON_MEMORY.
18501 (arm_expand_neon_builtin_1): Rename to ...
18502 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18503 values, arm_expand_builtin_args and add bool parameter NEON.
18504 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18505 (arm_expand_vfp_builtin): Likewise.
18506 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18508 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18510 PR middle-end/77484
18511 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18512 * predict.c (tree_estimate_probability_bb): Reverse direction of
18513 polymorphic call predictor.
18515 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18517 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18518 (determine_pass_name_match): ...this new function and...
18519 (should_skip_pass_p): ...this new function.
18521 2017-01-06 Nathan Sidwell <nathan@acm.org>
18523 * ipa-visibility.c (function_and_variable_visibility): Reformat
18524 comments and long lines. Remove extrneous if.
18525 * symtab.c (symtab_node::make_decl_local): Fix code format.
18526 (symtab_node::set_section_for_node): Fix comment typo.
18528 2017-01-06 Martin Liska <mliska@suse.cz>
18531 * lra-constraints.c: Rename invariant to lra_invariant.
18532 * predict.c (set_even_probabilities): Initialize e to NULL.
18534 2017-01-05 Martin Sebor <msebor@redhat.com>
18536 PR tree-optimization/78910
18537 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18538 (format_integer): Correct off-by-one error in the handling
18539 of precision with negative numbers in signed conversions..
18541 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18543 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18545 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18547 PR tree-optimization/71016
18548 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18549 factor_out_conditional_conversion. Formatting fix.
18550 (factor_out_conditional_conversion): Add cond_stmt argument.
18551 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18552 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18555 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18557 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18558 read-rtl-function.o, and selftest-rtl.o.
18559 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18560 (selftest::aarch64_test_loading_full_dump): New function.
18561 (selftest::aarch64_run_selftests): New function.
18562 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18563 selftest::aarch64_run_selftests.
18564 * config/i386/i386.c
18565 (selftest::ix86_test_loading_dump_fragment_1): New function.
18566 (selftest::ix86_test_loading_call_insn): New function.
18567 (selftest::ix86_test_loading_full_dump): New function.
18568 (selftest::ix86_test_loading_unspec): New function.
18569 (selftest::ix86_run_selftests): Call the new functions.
18570 * emit-rtl.c (maybe_set_max_label_num): New function.
18571 * emit-rtl.h (maybe_set_max_label_num): New decl.
18572 * function.c (instantiate_decls): Guard call to
18573 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18574 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18576 * gensupport.c (gen_reader::gen_reader): Pass "false"
18577 for new "compact" param of rtx_reader.
18578 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18579 rather than an empty string for NULL strings.
18580 * read-md.c: Potentially include config.h rather than bconfig.h.
18581 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18582 (have_error): New global, copied from errors.c.
18583 (md_reader::read_name): Rename to...
18584 (md_reader::read_name_1): ...this, adding "out_loc" param,
18585 and converting "missing name or number" to returning false, rather
18587 (md_reader::read_name): Reimplement in terms of read_name_1.
18588 (md_reader::read_name_or_nil): New function.
18589 (md_reader::read_string): Handle "(nil)" by returning NULL.
18590 (md_reader::md_reader): Add new param "compact".
18591 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18592 (md_reader::read_file): New method.
18593 * read-md.h (md_reader::md_reader): Add new param "compact".
18594 (md_reader::read_file): New method.
18595 (md_reader::is_compact): New accessor.
18596 (md_reader::read_name): Convert return type from void to file_location.
18597 (md_reader::read_name_or_nil): New decl.
18598 (md_reader::read_name_1): New decl.
18599 (md_reader::m_compact): New field.
18600 (noop_reader::noop_reader): Pass "false" for new "compact" param
18602 (rtx_reader::rtx_reader): Add new "compact" param.
18603 (rtx_reader::read_rtx_operand): Make virtual and convert return
18604 type from void to rtx.
18605 (rtx_reader::read_until): New decl.
18606 (rtx_reader::handle_any_trailing_information): New virtual function.
18607 (rtx_reader::postprocess): New virtual function.
18608 (rtx_reader::finalize_string): New virtual function.
18609 (rtx_reader::m_in_call_function_usage): New field.
18610 (rtx_reader::m_reuse_rtx_by_id): New field.
18611 * read-rtl-function.c: New file.
18612 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18613 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18614 (selftest::verify_three_block_rtl_cfg): New decl.
18615 * read-rtl-function.h: New file.
18616 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18617 For host, include function.h, memmodel.h, and emit-rtl.h.
18618 (one_time_initialization): New function.
18619 (struct compact_insn_name): New struct.
18620 (compact_insn_names): New array.
18621 (find_code): Handle insn codes in compact dumps.
18622 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18623 (bind_subst_iter_and_attr): Likewise.
18624 (add_condition_to_string): Likewise.
18625 (add_condition_to_rtx): Likewise.
18626 (apply_attribute_uses): Likewise.
18627 (add_current_iterators): Likewise.
18628 (apply_iterators): Likewise.
18629 (initialize_iterators): Guard usage of apply_subst_iterator with
18630 #ifdef GENERATOR_FILE.
18631 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18632 (md_reader::read_mapping): Likewise.
18633 (add_define_attr_for_define_subst): Likewise.
18634 (add_define_subst_attr): Likewise.
18635 (read_subst_mapping): Likewise.
18636 (check_code_iterator): Likewise.
18637 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18639 (one_time_initialization): New function.
18640 (rtx_reader::read_until): New method.
18641 (read_flags): New function.
18642 (parse_reg_note_name): New function.
18643 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18644 Handle reuse_rtx ids.
18645 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18646 Add parsing support for RTL dumps, mirroring the special-cases in
18647 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18648 values, and calling handle_any_trailing_information.
18649 (rtx_reader::read_rtx_operand): Convert return type from void
18650 to rtx, returning return_rtx. Handle case 'e'. Call
18651 finalize_string on XSTR and XTMPL fields.
18652 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18653 "(nil)" values were omitted. Call the postprocess vfunc on the
18655 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18656 class ctor. Initialize m_in_call_function_usage. Call
18657 one_time_initialization.
18658 * rtl-tests.c (selftest::test_uncond_jump): Call
18659 set_new_first_and_last_insn.
18660 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18661 * selftest-rtl.c: New file.
18662 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18663 (selftest::get_insn_by_uid): New decl.
18664 * selftest-run-tests.c (selftest::run_tests): Call
18665 read_rtl_function_c_tests.
18666 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18667 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18670 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18672 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18673 operands in a special way. Assert that pos+len <= mode precision.
18675 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18677 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18678 3 argument Alias with unlimited for the negative form.
18679 (fno-vect-cost-model): Removed.
18681 2017-01-05 Martin Liska <mliska@suse.cz>
18683 * hsa-gen.c (gen_hsa_divmod): New function.
18684 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18686 2017-01-05 Martin Liska <mliska@suse.cz>
18689 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18692 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18694 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18695 small constant length operands.
18697 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18699 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18700 between loop iterations.
18702 2017-01-05 Martin Liska <mliska@suse.cz>
18705 * gimplify.c (gimplify_decl_expr): Compare to
18706 asan_poisoned_variables instread of checking flags.
18707 (gimplify_target_expr): Likewise.
18708 (gimplify_expr): Likewise.
18709 (gimplify_function_tree): Conditionally initialize
18710 asan_poisoned_variables.
18712 2017-01-04 Jeff Law <law@redhat.com>
18714 PR tree-optimizatin/78812
18715 * rtl.h (contains_mem_rtx_p): Prototype.
18716 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18717 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18718 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18719 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18720 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18722 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18724 * input.c (assert_char_at_range): Default-initialize actual_range.
18726 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18728 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18729 to match the caller.
18731 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18733 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18734 insns after final jump in test to emit dummy move.
18736 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18738 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18739 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18741 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18743 * multiple_target.c (create_dispatcher_calls): Init e_next.
18744 * tree-ssa-loop-split.c (split_loop): Init border.
18745 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18748 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18753 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18754 (altivec_register_operand): Do not return true if the operand
18755 contains a SUBREG mixing SImode and SFmode.
18756 (vsx_register_operand): Likewise.
18757 (vsx_reg_sfsubreg_ok): New predicate.
18758 (vfloat_operand): Do not return true if the operand contains a
18759 SUBREG mixing SImode and SFmode.
18760 (vint_operand): Likewise.
18761 (vlogical_operand): Likewise.
18762 (gpc_reg_operand): Likewise.
18763 (int_reg_operand): Likewise.
18764 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18765 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18766 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
18768 (rs6000_emit_move_si_sf_subreg): New helper function.
18769 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
18770 fixup SUBREGs involving SImode and SFmode.
18771 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
18772 numbers for the new peephole2 optimization.
18773 (peephole2 for SFmode unions): New peephole2 to optimize cases in
18774 the GLIBC math library that do AND/IOR/XOR operations on single
18775 precision floating point.
18776 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
18777 target macros to say whether we need to avoid SUBREGs mixing
18779 (TARGET_ALLOW_SF_SUBREG): Likewise.
18780 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
18781 (UNSPEC_SI_FROM_SF): Likewise.
18782 (iorxor): Change spacing.
18783 (and_ior_xor): New iterator for AND, IOR, and XOR.
18784 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
18785 (movdi_from_sf_zero_ext): Likewise.
18786 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
18787 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
18788 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
18789 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
18790 (fms<mode>4): Likewise.
18791 (fnma<mode>4): Likewise.
18792 (fnms<mode>4): Likewise.
18793 (nfma<mode>4): Likewise.
18794 (nfms<mode>4): Likewise.
18796 2017-01-04 Marek Polacek <polacek@redhat.com>
18799 * doc/invoke.texi: Document -Wpointer-compare.
18801 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18803 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
18806 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
18807 descriptions for -gdwarf-5 and emit them as uleb128 instead of
18810 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
18813 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
18814 documentation of the powerpc_popcntb_ok attribute.
18815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18816 code to issue warning messages if a requested CPU configuration is
18817 not supported by the binary (assembler and loader) toolchain.
18818 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
18819 made to define a built-in function that has been disabled.
18820 (paired_init_builtins): Add assertion to prevent ICE if attempt is
18821 made to define a built-in function that has been disabled.
18822 (altivec_init_builtins): Add comment explaining why definition
18823 of the DST built-in functions is not preceded by an assertion
18824 check. Add assertions to prevent ICE if attempts are made to
18825 define an altivec predicate or an abs* built-in function that has
18827 (htm_init_builtins): Add comment explaining why definition of the
18828 htm built-in functions is not preceded by an assertion check.
18830 2017-01-04 Jeff Law <law@redhat.com>
18832 PR tree-optimizatin/67955
18833 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
18834 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
18835 the points-to solution does not include pt_null. Use DECL_PT_UID
18838 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
18840 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
18841 Use gen_int_mode instead of gen_lopwart for const_int operands.
18843 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18845 PR tree-optimization/71563
18846 * match.pd: Simplify X << Y into X if Y is known to be 0 or
18847 out of range value - has low bits known to be zero.
18849 2017-01-04 Alan Modra <amodra@gmail.com>
18851 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
18852 * configure: Regenerate.
18853 * config.in: Regenerate.
18855 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18858 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
18859 a substring of the message, but strcmp with the whole message. Ifdef
18860 ENABLE_NLS, translate the message first using dgettext.
18862 2017-01-03 Jeff Law <law@redhat.com>
18864 PR tree-optimizatin/78856
18865 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
18866 (mark_threaded_blocks): Remove code to truncate thread paths that
18867 cross multiple loop headers. Instead invalidate the cached loop
18868 iteration information and handle case of a thread path walking
18869 into an irreducible region.
18871 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
18874 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
18875 assertions. Add support for doing the signbit if the IEEE 128-bit
18876 floating point value is in a GPR.
18877 * config/rs6000/rs6000.md (Fsignbit): Delete.
18878 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
18879 Update the length attribute if the value is in a GPR.
18880 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
18881 the sign or zero extension instruction, since the value is always 0/1.
18882 (signbit<mode>2_dm2): Delete using <Fsignbit>.
18885 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
18886 extracting SImode to a GPR register so that we can generate a
18887 store, limit the vector to be in a traditional Altivec register
18888 for the vextuwrx instruction.
18890 2017-01-03 Ian Lance Taylor <iant@google.com>
18892 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
18894 2017-01-03 Martin Sebor <msebor@redhat.com>
18896 PR tree-optimization/78696
18897 * gimple-ssa-sprintf.c (format_floating): Correct handling of
18898 precision. Use MPFR for %f for greater fidelity. Correct handling
18900 (pass_sprintf_length::compute_format_length): Set width and precision
18901 specified by asrerisk to void_node for vararg functions.
18902 (try_substitute_return_value): Adjust dump output.
18904 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
18906 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
18908 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
18910 * doc/invoke.texi (SPARC options): Document -mlra as the default.
18911 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
18912 -mlra/-mno-lra was passed to the compiler.
18914 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
18916 PR rtl-optimization/65618
18917 * emit-rtl.c (try_split): Move initialization of "before" and
18918 "after" to just before the call to emit_insn_after_setloc.
18920 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
18922 * doc/md.texi (Standard Names): Remove reference to Java frontend.
18924 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
18926 * dwarf2out.c (gen_enumeration_type_die): When
18927 -gno-strict-dwarf, add a DW_AT_encoding attribute.
18929 2017-01-03 Jakub Jelinek <jakub@redhat.com>
18931 PR tree-optimization/78965
18932 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
18933 Change first argument from const call_info & to call_info &. For %n
18934 set info.nowrite to false.
18936 PR middle-end/78901
18937 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
18938 possibly throwing calls.
18940 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
18941 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
18942 and fns handling, rather than in a separate case SSA_NAME.
18944 2017-01-02 Jeff Law <law@redhat.com>
18946 * config/darwin-driver.c (darwin_driver_init): Const-correctness
18947 fixes for first_period and second_period variables.
18949 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
18952 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
18953 (*insvqi_1): New insn pattern.
18954 (*insvqi_1_mem_rex64): Ditto.
18955 (*insvqi_2): Ditto.
18956 (*insvqi_3): Rename from *insvqi.
18958 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
18960 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
18962 * doc/cfg.texi (Edges): Remove reference to Java.
18963 (Maintaining the CFG): Ditto.
18965 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18967 PR middle-end/77674
18968 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
18969 transparent aliases.
18971 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18973 PR middle-end/77484
18974 * predict.def (PRED_CALL): Update hitrate.
18975 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
18976 * predict.c (tree_estimate_probability_bb): Split CALL predictor
18977 into direct/indirect/polymorphic variants.
18979 2017-01-01 Jakub Jelinek <jakub@redhat.com>
18981 Update copyright years.
18983 * gcc.c (process_command): Update copyright notice dates.
18984 * gcov-dump.c (print_version): Ditto.
18985 * gcov.c (print_version): Ditto.
18986 * gcov-tool.c (print_version): Ditto.
18987 * gengtype.c (create_file): Ditto.
18988 * doc/cpp.texi: Bump @copying's copyright year.
18989 * doc/cppinternals.texi: Ditto.
18990 * doc/gcc.texi: Ditto.
18991 * doc/gccint.texi: Ditto.
18992 * doc/gcov.texi: Ditto.
18993 * doc/install.texi: Ditto.
18994 * doc/invoke.texi: Ditto.
18996 Copyright (C) 2017 Free Software Foundation, Inc.
18998 Copying and distribution of this file, with or without modification,
18999 are permitted in any medium without royalty provided the copyright
19000 notice and this notice are preserved.