1 2017-07-26 Richard Biener <rguenther@suse.de>
3 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4 (decision_tree::gen): Make 'type' const.
5 (write_predicate): Likewise.
7 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11 (rs6000_option_override_internal): Likewise.
12 (rs6000_expand_vector_set): Likewise.
13 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14 (TARGET_UPPER_REGS_SF): Likewise.
15 (TARGET_UPPER_REGS_DI): Likewise.
16 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20 (Splitters for DI constants in Altivec registers): Likewise.
21 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22 (vsx_set_v4sf_p9): Likewise.
23 (vsx_set_v4sf_p9_zero): Likewise.
24 (vsx_insert_extract_v4sf_p9): Likewise.
25 (vsx_insert_extract_v4sf_p9_2): Likewise.
27 2017-07-25 Carl Love <cel@us.ibm.com>
29 * doc/extend.texi: Update the built-in documentation file for the
30 existing built-in functions
31 vector signed char vec_cnttz (vector signed char);
32 vector unsigned char vec_cnttz (vector unsigned char);
33 vector signed short vec_cnttz (vector signed short);
34 vector unsigned short vec_cnttz (vector unsigned short);
35 vector signed int vec_cnttz (vector signed int);
36 vector unsigned int vec_cnttz (vector unsigned int);
37 vector signed long long vec_cnttz (vector signed long long);
38 vector unsigned long long vec_cnttz (vector unsigned long long);
40 2017-07-25 Andrew Pinski <apinski@cavium.com>
42 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
43 accesses where the use is for the first operand of a BIT_INSERT.
45 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
48 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
49 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
51 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
53 * config/i386/gstabs.h: Delete.
54 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
56 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
58 * config/i386/i386.c (ix86_decompose_address): Do not check for
59 register RTX when looking at index_reg or base_reg.
60 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
62 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
64 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
65 to update EH info here.
67 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
69 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
71 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
73 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
75 2017-07-25 Torsten Duwe <duwe@suse.de>
77 * common.opt: Introduce -fpatchable-function-entry
78 command line option, and its variables function_entry_patch_area_size
79 and function_entry_patch_area_start.
80 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
81 including a two-value parser.
82 * target.def (print_patchable_function_entry): New target hook.
83 * targhooks.h (default_print_patchable_function_entry): New function.
84 * targhooks.c (default_print_patchable_function_entry): Likewise.
85 * toplev.c (process_options): Switch off IPA-RA if
86 patchable function entries are being generated.
87 * varasm.c (assemble_start_function): Look at the
88 patchable-function-entry command line switch and current
89 function attributes and maybe generate NOP instructions by
90 calling the print_patchable_function_entry hook.
91 * doc/extend.texi: Document patchable_function_entry attribute.
92 * doc/invoke.texi: Document -fpatchable_function_entry
94 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
96 * doc/tm.texi: Re-generate.
98 2017-07-25 Jakub Jelinek <jakub@redhat.com>
101 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
102 TARGET_AVX512DQ rather than TARGET_AVX512BW.
104 2017-07-25 Tamar Christina <tamar.christina@arm.com>
106 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
108 2017-07-25 Richard Biener <rguenther@suse.de>
110 PR tree-optimization/81455
111 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
112 not walk in cycles when looking for guards.
114 2017-07-25 Richard Biener <rguenther@suse.de>
116 PR tree-optimization/81529
117 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
118 when optimizing backedge uses.
120 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
122 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
124 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
125 to dl_section_ref. On AIX, append an expression to subtract
126 the size of the section length to dl_section_ref.
128 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
130 * configure.ac: If any of the config.* scripts fail, exit 1.
131 * configure: Regenerate.
133 2017-07-25 Richard Biener <rguenther@suse.de>
136 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
137 of immediate uses, be more verbose on errors.
139 2017-07-25 Richard Biener <rguenther@suse.de>
141 PR tree-optimization/81510
142 * tree-vect-loop.c (vect_is_simple_reduction): When the
143 reduction stmt is not inside the loop bail out.
145 2017-07-25 Richard Biener <rguenther@suse.de>
147 PR tree-optimization/81303
148 * tree-vect-loop-manip.c (vect_loop_versioning): Build
149 profitability check against LOOP_VINFO_NITERSM1.
151 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
153 * domwalk.c (cmp_bb_postorder): Simplify.
154 (sort_bbs_postorder): New function. Use it...
155 (dom_walker::walk): ...here to optimize common cases.
157 2017-07-25 Martin Liska <mliska@suse.cz>
160 * ipa-visibility.c (function_and_variable_visibility): Make the
161 redirection just on target that supports aliasing.
162 Fix GNU coding style.
164 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
167 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
169 * config/arm/rtems.h: Likewise.
170 * config/bfin/rtems.h: Likewise.
171 * config/i386/rtemself.h: Likewise.
172 * config/lm32/rtems.h: Likewise.
173 * config/m32c/rtems.h: Likewise.
174 * config/m68k/rtemself.h: Likewise.
175 * config/microblaze/rtems.h: Likewise.
176 * config/mips/rtems.h: Likewise.
177 * config/moxie/rtems.h: Likewise.
178 * config/nios2/rtems.h: Likewise.
179 * config/powerpcspe/rtems.h: Likewise.
180 * config/rs6000/rtems.h: Likewise.
181 * config/rtems.h: Likewise.
182 * config/sh/rtems.h: Likewise.
183 * config/sh/rtemself.h: Likewise.
184 * config/sparc/rtemself.h: Likewise.
186 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
189 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
190 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
191 * tree-ssa-structalias.c (alias_get_name): Same.
193 2017-07-25 Bin Cheng <bin.cheng@arm.com>
196 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
197 instructions if no du chain is found.
199 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
201 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
203 2017-07-25 Richard Biener <rguenther@suse.de>
206 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
209 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
211 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
213 (ISA_2_7_MASKS_SERVER): Likewise.
214 (ISA_3_0_MASKS_IEEE): Likewise.
215 (OTHER_P8_VECTOR_MASKS): Likewise.
216 (OTHER_VSX_VECTOR_MASKS): Likewise.
217 (POWERPC_MASKS): Likewise.
218 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
219 duplicate list of options.
220 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
221 explicit -mupper-regs options.
222 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
223 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
224 alias for -mupper-regs-df.
225 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
226 (rs6000_init_hard_regno_mode_ok): Likewise.
227 (rs6000_option_override_internal): Likewise.
228 (rs6000_opt_masks): Likewise.
229 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
230 options in terms of whether -mvsx or -mpower8-vector was used.
231 (TARGET_UPPER_REGS_DI): Likewise.
232 (TARGET_UPPER_REGS_SF): Likewise.
233 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
234 -mupper-regs-* options.
236 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
238 * passes.c (emergency_dump_function): Print some empty lines and a
239 header before the RTL dump.
241 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
243 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
245 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
248 * config/aarch64/aarch64.c (aarch64_classify_symbol):
249 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
251 2017-07-24 Carl Love <cel@us.ibm.com>
253 * config/rs6000/rs6000-c.c: Add support for built-in functions
254 vector float vec_extract_fp32_from_shorth (vector unsigned short);
255 vector float vec_extract_fp32_from_shortl (vector unsigned short);
256 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
257 vec_extract_fp_from_shortl): Add defines for the two builtins.
258 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
259 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
261 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
262 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
263 * doc/extend.texi: Update the built-in documentation file for the
264 new built-in function.
266 2017-07-24 Jakub Jelinek <jakub@redhat.com>
269 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
271 * doc/generic.texi: Likewise.
272 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
273 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
275 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
277 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
278 (aarch64_mls_elt_merge<mode>): Likewise.
280 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
282 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
285 2017-07-23 Michael Collison <michael.collison@arm.com>
287 * config/arm/arm.c (arm_option_override): Deprecate
288 use of -mstructure-size-boundary.
289 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
290 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
292 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
295 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
296 Reduce cost estimate for direct moves.
298 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
301 * config/i386/i386.c (ix86_option_override_internal): Disable
302 BMI, BMI2 and TBM instructions for -m16.
304 2017-07-21 Carl Love <cel@us.ibm.com>
306 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
307 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
308 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
309 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
310 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
311 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
312 VMULOSW): New enum "unspec" values.
313 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
314 altivec_vmulosw): New patterns.
315 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
316 VMULOSW): Add definitions.
318 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
320 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
322 * config/aarch64/aarch64-options-extensions.def (rdma); New.
323 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
324 (AARCH64_FL_V8_1): Renumber.
325 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
326 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
327 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
328 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
329 rdma to feature modifiers list.
331 2017-07-21 Yury Gribov <tetra2005@gmail.com>
334 * ipa-visibility (function_and_variable_visibility): Convert
335 recursive PLT call to direct call if appropriate.
337 2017-07-21 Andrew Pinski <apinski@cavium.com>
339 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
340 operand 1 to see if the types precision matches.
341 * fold-const.c (operand_equal_p): Likewise.
343 2017-07-21 Richard Biener <rguenther@suse.de>
345 PR tree-optimization/81303
346 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
347 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
348 (vect_peeling_hash_get_lowest_cost): Adjust.
349 (vect_enhance_data_refs_alignment): Likewise. Use
350 vect_get_peeling_costs_all_drs to compute the penalty for no
351 peeling to match up costs.
353 2017-07-21 Richard Biener <rguenther@suse.de>
355 PR tree-optimization/81500
356 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
357 we didn't identify a reduction path.
359 2017-07-21 Tom de Vries <tom@codesourcery.com>
360 Cesar Philippidis <cesar@codesourcery.com>
362 PR gcov-profile/81442
363 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
366 2017-07-21 Tom de Vries <tom@codesourcery.com>
369 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
371 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
372 nvptx_override_options_after_change.
374 2017-07-21 Ulrich Drepper <drepper@redhat.com>
376 * dwarf2out.c (output_file_names): Avoid double testing for
379 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
381 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
383 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
385 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
387 (try_crossjump_to_edge): Do not punt on partitioned functions.
389 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
391 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
392 Put all BBs reachable only via paths crossing cold region to cold
394 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
396 2016-07-21 Richard Biener <rguenther@suse.de>
398 PR tree-optimization/81303
399 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
400 into account prologue and epilogue iterations when raising
401 min_profitable_iters to sth at least covering one vector iteration.
403 2017-07-21 Tamar Christina <tamar.christina@arm.com>
405 * config/arm/arm.c (arm_test_cpu_arch_dat):
408 2017-07-20 Nathan Sidwell <nathan@acm.org>
411 * tree.h (TYPE_METHODS): Delete.
412 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
413 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
414 (dbxout_type_methods): Scan TYPE_FIELDS.
415 (dbxout_type): Don't check TYPE_METHODS here.
416 * function.c (use_register_for_decl): Always ignore register for
417 class types when not optimizing.
418 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
419 * tree.c (free_lang_data_in_type): Stitch out member functions and
420 templates from TYPE_FIELDS.
421 (build_distinct_type_copy, verify_type_variant,
422 verify_type): Member fns are on TYPE_FIELDS.
423 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
424 * tree-pretty-print.c (dump_generic_node): Likewise.
426 2017-07-20 Jakub Jelinek <jakub@redhat.com>
429 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
430 V2TImode and V4TImode.
431 (ix86_expand_vector_extract): Likewise.
432 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
433 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
434 (ssescalarmode): Handle V4TImode and V2TImode.
435 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
436 (*vec_extractv2ti, *vec_extractv4ti): New insns.
437 (VEXTRACTI128_MODE): New mode iterator.
438 (splitter for *vec_extractv?ti first element): New.
439 (VEC_INIT_MODE): New mode iterator.
440 (vec_init<mode>): Consolidate 3 expanders into one using
441 VEC_INIT_MODE mode iterator.
443 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
445 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
446 non_spilled_static_chain_regno_p.
448 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
450 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
452 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
454 * bb-reorder.c (connect_traces): Allow copying of blocks within
457 2017-07-20 Richard Biener <rguenther@suse.de>
459 * gimple.h (gimple_phi_result): Add gphi * overload.
460 (gimple_phi_result_ptr): Likewise.
461 (gimple_phi_arg): Likewise. Adjust index assert to only
462 allow actual argument accesses rather than all slots available
464 (gimple_phi_arg_def): Add gphi * overload.
465 * tree-phinodes.c (make_phi_node): Initialize only actual
467 (resize_phi_node): Clear memory not covered by old node,
468 do not initialize excess argument slots.
469 (reserve_phi_args_for_new_edge): Initialize new argument slot
472 2017-07-20 Bin Cheng <bin.cheng@arm.com>
474 PR tree-optimization/81388
476 2016-07-21 Bin Cheng <bin.cheng@arm.com>
478 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
479 by removing computation of may_be_zero.
481 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
482 Tom de Vries <tom@codesourcery.com>
485 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
486 when gimple level profile disagrees with what RTL expander did.
488 2017-07-20 Richard Biener <rguenther@suse.de>
490 PR tree-optimization/61171
491 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
492 (vect_analyze_stmt): Add slp instance parameter.
493 (vectorizable_reduction): Likewise.
494 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
495 (vect_is_simple_reduction): Deal with chains not detected
496 as SLP reduction chain, specifically not properly associated
497 chains containing a mix of plus/minus.
498 (get_reduction_op): Remove.
499 (get_initial_defs_for_reduction): Simplify, pass in whether
500 this is a reduction chain, pass in the SLP node for the PHIs.
501 (vect_create_epilog_for_reduction): Get the SLP instance as
503 (vectorizable_reduction): Get the SLP instance as arg.
504 During analysis remember the SLP node with the PHIs in the
505 instance. Simplify getting at the vectorized reduction PHIs.
506 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
507 through SLP instance.
508 (vect_slp_analyze_operations): Likewise.
509 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
510 (vect_transform_stmt): Likewise.
512 2017-07-20 Tom de Vries <tom@codesourcery.com>
514 PR tree-optimization/81489
515 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
516 read of phi arg location to before loop that modifies phi.
518 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
520 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
523 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
526 * except.c (execute): Fix ordering issue.
528 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
530 PR rtl-optimization/81423
531 * combine.c (make_compound_operation_int): Don't try to optimize
532 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
534 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
536 PR rtl-optimization/81423
537 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
538 with a constant that is -1 in the truncated to mode.
540 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
542 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
543 (determine_unlikely_bbs): ... here.
544 * predict.h (propagate_unlikely_bbs_forward): Declare.
545 * cfgexpand.c (pass_expand::execute): Use it.
546 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
548 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
549 propagate_unlikely_bbs_forward.
551 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
554 * except.c (maybe_add_nop_after_section_switch): New function.
557 2017-07-19 Tom de Vries <tom@codesourcery.com>
559 * gimple.h (gimple_phi_set_arg): Make assert more strict.
561 2017-07-19 Tom de Vries <tom@codesourcery.com>
563 * gimple.h (gimple_phi_arg): Make assert more strict.
565 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
567 * config.gcc (powerpc*-*-*): Add mmintrin.h.
568 * config/rs6000/mmintrin.h: New file.
569 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
571 2017-07-19 Jakub Jelinek <jakub@redhat.com>
573 PR tree-optimization/81346
574 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
576 2017-07-19 Tom de Vries <tom@codesourcery.com>
578 * config/nvptx/nvptx.md (VECIM): Add V2DI.
580 2017-07-19 Tom de Vries <tom@codesourcery.com>
582 * config/nvptx/nvptx-modes.def: Add V2DImode.
583 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
584 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
585 (nvptx_output_mov_insn): Handle lack of mov.b128.
586 (nvptx_print_operand): Handle 'H' and 'L' codes.
587 (nvptx_vector_mode_supported): Allow V2DImode.
588 (nvptx_preferred_simd_mode): New function.
589 (nvptx_data_alignment): New function.
590 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
591 nvptx_preferred_simd_mode.
592 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
594 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
596 2017-07-19 Tom de Vries <tom@codesourcery.com>
598 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
599 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
600 (nvptx_vector_mode_supported): New function. Allow V2SImode.
601 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
602 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
603 (mov<VECIM>_insn): New define_insn.
604 (define_expand "mov<VECIM>): New define_expand.
606 2017-07-19 Tom de Vries <tom@codesourcery.com>
608 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
610 2017-07-19 Jakub Jelinek <jakub@redhat.com>
612 PR tree-optimization/81346
613 * fold-const.h (fold_div_compare, range_check_type): Declare.
614 * fold-const.c (range_check_type): New function.
615 (build_range_check): Use range_check_type.
616 (fold_div_compare): No longer static, rewritten into
617 a match.pd helper function.
618 (fold_comparison): Don't call fold_div_compare here.
619 * match.pd (X / C1 op C2): New optimization using fold_div_compare
622 2017-07-19 Nathan Sidwell <nathan@acm.org>
624 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
625 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
626 * tree.c (find_decls_types_r, verify_type): Use
627 TYPE_{MIN,MAX}_VALUE_RAW.
628 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
629 (hash_tree): Likewise.
630 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
632 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
635 2017-07-18 Tom de Vries <tom@codesourcery.com>
638 * omp-expand.c (expand_omp_for_static_chunk): Handle
639 equal-argument loop exit phi.
641 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
644 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
645 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
647 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
649 (ror,rol -> rorx splitters): Use const_int_operand as
652 2017-06-18 Richard Biener <rguenther@suse.de>
654 PR tree-optimization/81410
655 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
656 the gap in the ! slp_perm SLP case after each group.
658 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
661 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
664 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
667 * predict.c (set_even_probabilities): Cleanup; do not affect
668 probabilities that are already known.
669 (combine_predictions_for_bb): Call even when count is set.
671 2017-07-18 Nathan Sidwell <nathan@acm.org>
673 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
676 2017-07-18 Bin Cheng <bin.cheng@arm.com>
679 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
680 optimization for loop niter analysis.
682 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
685 * config/avr/avr.c (avr_optimize_casesi): Don't use
686 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
688 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
690 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
691 body_cost_vec from _vect_peel_extended_info.
692 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
693 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
696 2017-07-18 Bin Cheng <bin.cheng@arm.com>
698 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
700 2017-07-18 Richard Biener <rguenther@suse.de>
702 PR tree-optimization/80620
703 PR tree-optimization/81403
704 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
705 info when re-using a VN table entry.
707 2017-07-18 Richard Biener <rguenther@suse.de>
709 PR tree-optimization/81418
710 * tree-vect-loop.c (vectorizable_reduction): Properly compute
711 vectype_in. Verify that with lane-reducing reduction operations
712 we have a single def-use cycle.
714 2017-07-17 Carl Love <cel@us.ibm.com>
716 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
718 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
719 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
720 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
721 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
722 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
723 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
724 VMULOSW): New enum "unspec" values.
725 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
726 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
727 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
728 altivec_vmulosw): New patterns.
729 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
730 VMULOSW): Add definitions.
731 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
733 * config/alpha/alpha.c: Include predict.h.
735 2017-07-17 Yury Gribov <tetra2005@gmail.com>
737 * tree-vrp.c (compare_assert_loc): Fix comparison function
738 to return predictable results.
740 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
742 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
745 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
746 * doc/invoke.texi (mexpand-adddi): Update text.
748 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
750 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
751 that also clobbers the CC register. The old expand code is moved
753 (*arc_clzsi2): ... here.
754 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
755 the CC register. The old expand code is moved to ...
756 (arc_ctzsi2): ... here.
758 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
760 * config/arc/arc.opt (mindexed-loads): Use initial value
761 TARGET_INDEXED_LOADS_DEFAULT.
762 (mauto-modify-reg): Use initial value
763 TARGET_AUTO_MODIFY_REG_DEFAULT.
764 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
765 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
766 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
767 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
769 2017-07-17 Martin Liska <mliska@suse.cz>
772 * opts.c (finish_options): Do not allow -fgnu-tm
773 w/ -fsanitize={kernel-,}address. Say sorry.
775 2017-07-17 Bin Cheng <bin.cheng@arm.com>
778 * tree-loop-distribution.c (classify_partition): Only assert on
780 (merge_dep_scc_partitions): Delete prameter. Update function call.
781 (distribute_loop): Remove code handling loop with unknown niters.
782 (pass_loop_distribution::execute): Skip loop with unknown niters.
784 2017-07-17 Bin Cheng <bin.cheng@arm.com>
787 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
788 function sort_partitions_by_post_order.
790 2017-07-17 Bin Cheng <bin.cheng@arm.com>
792 PR tree-optimization/81374
793 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
794 the max index of basic blocks, rather than number of basic blocks.
796 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
798 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
800 (arc_legitimate_pic_operand_p): Likewise.
801 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
803 (arc_needs_pcl_p): Likewise.
804 (arc_legitimate_pc_offset_p): Likewise.
805 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
806 function is also used in constrains.md.
807 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
808 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
809 PLUS. Only return true/false in known cases, otherwise assert.
810 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
811 is already called in arc_legitimate_constant_p.
812 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
814 (LEGITIMATE_PIC_OPERAND_P): Use
815 arc_raw_symbolic_reference_mentioned_p function.
816 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
821 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
822 Andrew Burgess <andrew.burgess@embecosm.com>
824 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
825 (arc_return_address_register): New function.
826 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
827 (arc_handle_fndecl_attribute): Add naked attribute.
828 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
829 (TARGET_WARN_FUNC_RETURN): Likewise.
830 (arc_allocate_stack_slots_for_args): New function.
831 (arc_warn_func_return): Likewise.
832 (machine_function): Change type fn_type.
833 (arc_compute_function_type): Consider new naked function type,
834 change function return type.
835 (arc_must_save_register): Adapt to handle new
836 arc_compute_function_type's return type.
837 (arc_expand_prologue): Likewise.
838 (arc_expand_epilogue): Likewise.
839 (arc_return_address_regs): Delete.
840 (arc_return_address_register): New function.
841 (arc_epilogue_uses): Use above function.
842 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
843 (arc_function_type): Change encoding, add naked type.
844 (ARC_INTERRUPT_P): Change to handle the new encoding.
845 (ARC_FAST_INTERRUPT_P): Likewise.
846 (ARC_NORMAL_P): Define.
847 (ARC_NAKED_P): Likewise.
848 (arc_compute_function_type): Delete prototype.
849 * config/arc/arc.md (in_ret_delay_slot): Use
850 arc_return_address_register function.
851 (simple_return): Likewise.
852 (p_return_i): Likewise.
854 2017-07-17 Jakub Jelinek <jakub@redhat.com>
856 PR tree-optimization/81428
857 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
858 can't be built for those types.
860 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
862 Remove stuff dead since r239246.
864 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
865 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
866 (avr_inform_devices): Remove dead stuff.
868 2017-07-17 Tamar Christina <tamar.christina@arm.com>
870 * config/arm/arm_neon.h: Fix softp typo.
872 2017-07-17 Jakub Jelinek <jakub@redhat.com>
874 PR tree-optimization/81365
875 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
876 aggregate moves onto bb predecessor edges, make sure there are no
877 loads that could alias the lhs in between the start of bb and the
880 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
883 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
884 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
885 [LSHIFTRT, outer_code = TRUNCATE]: Same.
887 2017-07-17 Jakub Jelinek <jakub@redhat.com>
889 PR tree-optimization/81396
890 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
891 (init_symbolic_number): Initialize it to 1.
892 (perform_symbolic_merge): Add n_ops from both operands into the new
894 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
895 without base_addr as useless if they need more than one operation.
896 (bswap_replace): Handle !bswap case for NULL base_addr.
898 2017-07-17 Tom de Vries <tom@codesourcery.com>
901 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
904 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
906 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
907 conditional builtin define __FIX_LEON3FT_B2BST.
909 2017-07-17 Daniel Cederman <cederman@gaisler.com>
911 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
912 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
915 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
917 PR rtl-optimization/81424
918 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
919 to remove potential trapping from operands if -fnon-call-exceptions.
921 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
923 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
924 profile_proability for scalling.
925 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
927 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
929 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
931 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
933 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
934 fixpoint arithmetics.
936 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
938 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
939 fixpoint arithmetics.
941 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
943 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
944 fixpoint arithmetics.
946 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
948 * profile-count.h (profile_probability::from_reg_br_prob_note,
949 profile_probability::to_reg_br_prob_note): New functions.
950 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
951 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
952 * predict.c (probability_reliable_p): Update.
953 (edge_probability_reliable_p): Update.
954 (br_prob_note_reliable_p): Update.
955 (invert_br_probabilities): Update.
956 (add_reg_br_prob_note): New function.
957 (combine_predictions_for_insn): Update.
958 * asan.c (asan_clear_shadow): Update.
959 * cfgbuild.c (compute_outgoing_frequencies): Update.
960 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
961 (update_br_prob_note): Update.
962 (rtl_verify_edges): Update.
963 (purge_dead_edges): Update.
964 (fixup_reorder_chain): Update.
965 * emit-rtl.c (try_split): Update.
966 * ifcvt.c (cond_exec_process_insns): Update.
967 (cond_exec_process_if_block): Update.
968 (dead_or_predicable): Update.
969 * internal-fn.c (expand_addsub_overflow): Update.
970 (expand_neg_overflow): Update.
971 (expand_mul_overflow): Update.
972 * loop-doloop.c (doloop_modify): Update.
973 * loop-unroll.c (compare_and_jump_seq): Update.
974 * optabs.c (emit_cmp_and_jump_insn_1): Update.
976 * reorg.c (mostly_true_jump): Update.
978 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
979 * config/alpha/alpha.c (emit_unlikely_jump): Update.
980 * config/arc/arc.c: (emit_unlikely_jump): Update.
981 * config/arm/arm.c: (emit_unlikely_jump): Update.
982 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
983 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
984 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
985 (ix86_print_operand): Update.
986 (ix86_split_fp_branch): Update.
987 (predict_jump): Update.
988 * config/ia64/ia64.c (ia64_print_operand): Update.
989 * config/mmix/mmix.c (mmix_print_operand): Update.
990 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
991 (rs6000_expand_split_stack_prologue): Update.
992 * config/rs6000/rs6000.c: Update.
993 * config/s390/s390.c (s390_expand_vec_strlen): Update.
994 (s390_expand_vec_movstr): Update.
995 (s390_expand_cs_tdsi): Update.
996 (s390_expand_split_stack_prologue): Update.
997 * config/sh/sh.c (sh_print_operand): Update.
998 (expand_cbranchsi4): Update.
999 (expand_cbranchdi4): Update.
1000 * config/sparc/sparc.c (output_v9branch): Update.
1001 * config/spu/spu.c (get_branch_target): Update.
1002 (ea_load_store_inline): Update.
1003 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1004 * config/tilepro/tilepro.c: Update.
1006 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1008 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1009 (gimplify_save_expr): Likewise.
1011 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1013 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1015 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1017 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1019 * timevar.def (TV_IPA_FNSUMMARY): Define.
1021 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1023 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1024 to back store errata sensitive sequence from being generated.
1025 (sqrtdf2_fix): Likewise.
1027 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1029 * tree-ssa-threadupdate.c (compute_path_counts,
1030 update_joiner_offpath_counts): Use profile_probability.
1032 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1035 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1037 * config/arm/arm-c.c (arm_cpu_builtins): Define
1038 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1040 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1042 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1043 array entries to represent __ieee128 versions of the
1044 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1045 scalar_extract_sig, and scalar_insert_exp built-in functions.
1046 (altivec_resolve_overloaded_builtin): Add special case handling
1047 for the __builtin_scalar_insert_exp function, as represented by
1048 the P9V_BUILTIN_VEC_VSIEDP constant.
1049 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1050 exponent support for __ieee128 argument.
1051 (VSESQP): Add scalar extract signature support for __ieee128
1053 (VSTDCNQP): Add scalar test negative support for __ieee128
1055 (VSIEQP): Add scalar insert exponent support for __int128 argument
1056 with __ieee128 result.
1057 (VSIEQPF): Add scalar insert exponent support for __ieee128
1058 argument with __ieee128 result.
1059 (VSTDCQP): Add scalar test data class support for __ieee128
1061 (VSTDCNQP): Add overload support for scalar test negative with
1063 (VSTDCQP): Add overload support for scalar test data class
1065 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1067 (UNSPEC_VSX_SIEXPQP): New constant.
1068 (xsxexpqp): New insn for VSX scalar extract exponent quad
1070 (xsxsigqp): New insn for VSX scalar extract significand quad
1072 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1073 precision with floating point argument.
1074 (xststdcqp): New expand for VSX scalar test data class quad
1076 (xststdcnegqp): New expand for VSX scalar test negative quad
1078 (xststdcqp): New insn to match expansions for VSX scalar test data
1079 class quad precision and VSX scalar test negative quad precision.
1080 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1081 special case operand checking to enforce that second operand of
1082 VSX scalar test data class with quad precision argument is a 7-bit
1084 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1085 prototypes and descriptions of __ieee128 versions of
1086 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1087 scalar_test_data_class, and scalar_test_neg built-in functions.
1089 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1091 PR tree-optimization/81162
1092 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1093 replace a negate with an add.
1095 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1097 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1099 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1101 * config/arm/arm-c.c (arm_cpu_builtins): Define
1102 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1104 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1106 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1107 (armv8-r): Set ARM Cortex-R52 as default CPU.
1108 * config/arm/arm-tables.opt: Regenerate.
1109 * config/arm/arm-tune.md: Regenerate.
1110 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1112 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1113 extension for -mcpu=cortex-r52.
1115 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1117 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1118 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1119 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1120 (fp-armv8): Define it as FP_ARMv8 only.
1121 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1122 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1124 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1125 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1126 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1127 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1128 than TARGET_FPU_ARMV8.
1129 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1130 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1131 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1133 * config/arm/neon.md (neon_vrint): Likewise.
1134 (neon_vcvt): Likewise.
1135 (neon_<fmaxmin_op><mode>): Likewise.
1136 (<fmaxmin><mode>3): Likewise.
1137 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1138 * config/arm/predicates.md (arm_cond_move_operator): Check against
1139 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1141 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1143 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1146 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1148 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1149 loop in comment with memset.
1151 2017-07-14 Martin Liska <mliska@suse.cz>
1153 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1154 * dwarf2out.c (is_java): Remove the function.
1155 (output_pubname): Remove usage of the function.
1156 (lower_bound_default): Remove usage of DW_LANG_Java.
1157 (gen_compile_unit_die): Likewise.
1158 * gcc.c: Remove compiler defaults for .java and .zip files.
1159 * gimple-expr.c (remove_suffix): Change as there's no longer
1160 extension than 4-letter one.
1161 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1162 (gimplify_save_expr): Likewise.
1163 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1164 as it's possible even for other languages than Java.
1165 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1166 * lto-opts.c (lto_write_options): Remove reference to Java.
1167 * opts.c (strip_off_ending): Update file extension handling.
1168 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1169 * tree-eh.c (lower_resx): Likewise.
1170 * tree.c (free_lang_data_in_type): Remove dead code.
1171 (find_decls_types_r): Likewise.
1172 (build_common_builtin_nodes): Remove Java from a comment.
1173 (verify_type): Remove dead code.
1174 * varasm.c (assemble_external): Remove Java from a comment.
1176 2017-07-14 Martin Liska <mliska@suse.cz>
1178 * opts.c (finish_options): Add quotes.
1179 (common_handle_option): Likewise.
1181 2017-07-14 Martin Liska <mliska@suse.cz>
1183 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1184 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1186 * dwarf2out.c (lower_bound_default): Do not handle
1188 (gen_compile_unit_die): Likewise.
1189 * gcc.c: Remove default extension binding for GNU Pascal.
1190 * stmt.c: Remove Pascal language from a comment.
1191 * xcoffout.c: Likewise.
1193 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1196 * diagnostic-show-locus.c (fixit_cmp): New function.
1197 (layout::layout): Sort m_fixit_hints.
1198 (column_range::column_range): Assert that the values are valid.
1199 (struct char_span): New struct.
1200 (correction::overwrite): New method.
1201 (struct source_line): New struct.
1202 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1203 calls in terms of classes source_line and char_span, and
1204 correction::overwrite.
1205 (selftest::test_overlapped_fixit_printing_2): New function.
1206 (selftest::diagnostic_show_locus_c_tests): Call it.
1208 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1210 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1211 early if there is no lhs.
1213 2017-07-13 Martin Liska <mliska@suse.cz>
1215 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1216 (gen_reference_type_die): Likewise.
1217 * stor-layout.c: Remove Pascal-related comment.
1219 2017-07-13 Martin Liska <mliska@suse.cz>
1221 * opts.c (finish_options): Add quotes to error messages.
1222 (parse_sanitizer_options): Likewise.
1224 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1226 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1228 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1230 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1232 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1234 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1236 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1238 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1241 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1242 provides the hardware capability bits, define the macro
1243 __BUILTIN_CPU_SUPPORTS__.
1244 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1245 if GLIBC does not provide the hardware capability bits. Add a
1246 gcc_unreachable call if the built-in cpu function is neither
1247 __builtin_cpu_is nor __builtin_cpu_supports.
1248 (rs6000_get_function_versions_dispatcher): Change the warning
1249 that an old GLIBC is used which does not export the capability
1250 bits to be an error.
1251 * doc/extend.texi (target_clones attribute): Document the
1252 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1253 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1254 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1255 the macros defined by GCC if the newer GLIBC is available.
1257 2017-07-12 Jeff Law <law@redhat.com>
1259 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1260 remaining includes slightly.
1261 * config/riscv/riscv-builtins.c: Include profile-count.h.
1263 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1266 * config/avr/avr.c (avr_set_current_function): In diagnostic
1267 messages: Quote keywords and (parts of) identifiers.
1268 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1271 2017-07-12 Carl Love <cel@us.ibm.com>
1273 * config/rs6000/rs6000-c.c: Add support for built-in functions
1274 vector bool char vec_revb (vector bool char);
1275 vector bool short vec_revb (vector short char);
1276 vector bool int vec_revb (vector bool int);
1277 vector bool long long vec_revb (vector bool long long);
1278 * doc/extend.texi: Update the built-in documentation file for the
1279 new built-in functions.
1281 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1283 * config/s390/s390.md: Remove movcc splitter.
1285 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1287 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1288 load/store on condition.
1290 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1293 * config/avr/avr.c (avr_encode_section_info)
1294 [progmem && !TREE_READONLY]: Error if progmem object needs
1297 2017-07-11 Michael Collison <michael.collison@arm.com>
1299 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1302 2017-07-11 Carl Love <cel@us.ibm.com>
1304 * config/rs6000/rs6000-c.c: Add support for builtins
1305 vector unsigned int vec_parity_lsbb (vector signed int);
1306 vector unsigned int vec_parity_lsbb (vector unsigned int);
1307 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1308 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1309 vector unsigned long long vec_parity_lsbb (vector signed long long);
1310 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1311 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1312 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1313 * doc/extend.texi: Update the built-in documentation file for the
1314 new built-in functions.
1316 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1318 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1319 (layout::m_primary_loc): New field.
1320 (layout::layout): Initialize new field. Move location filtering
1321 logic from here to...
1322 (layout::maybe_add_location_range): ...this new method. Add
1323 support for filtering to just the lines already specified by other
1325 (layout::will_show_line_p): New method.
1326 (gcc_rich_location::add_location_if_nearby): New method.
1327 (selftest::test_add_location_if_nearby): New test function.
1328 (selftest::diagnostic_show_locus_c_tests): Call it.
1329 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1332 2017-07-11 Tom de Vries <tom@codesourcery.com>
1334 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1335 (bb_first_real_insn): New function.
1336 (nvptx_single): Add extra initialization of broadcasted condition
1339 2017-07-11 Nathan Sidwell <nathan@acm.org>
1341 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1343 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1345 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1346 Binutils doc as TEXI will mess them up.
1347 * doc/invoke.texi (AVR Options): Same here.
1349 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1351 * config/sparc/sparc.opt (mfix-ut700): New option.
1352 (mfix-gr712rc): Likewise.
1353 (sparc_fix_b2bst): New variable.
1354 * doc/invoke.texi (SPARC options): Document them.
1355 (ARM options): Fix warnings.
1356 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1357 instructions to prevent sequences that can trigger the store-store
1358 errata for certain LEON3FT processors.
1359 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1360 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1361 * config/sparc/sparc.md (fix_b2bst): New attribute.
1362 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1364 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1367 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1369 (*rsqrtsf2_sse): Ditto.
1371 (div<mode>3): Macroize insn from divdf3 and divsf3
1372 using MODEF mode iterator.
1374 2017-07-10 Martin Sebor <msebor@redhat.com>
1376 PR tree-optimization/80397
1377 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1378 instead of testing for equality to INTEGER_TYPE.
1380 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1382 * config.gcc: Remove uclibc from arc target spec.
1384 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1386 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1388 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1391 * lto-wrapper.c (remove_option): New function.
1392 (merge_and_complain): Merge PIC/PIE options more realistically.
1394 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1396 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1400 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1401 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1402 * config.in: Regenerate.
1403 * configure: Regenerate.
1404 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1405 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1406 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1407 (TARGET_GASISR_PROLOGUES): ...target mask.
1408 * common/config/avr/avr-common.c
1409 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1410 Set -mgas-isr-prologues.
1411 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1412 INSERT_PASS_BEFORE for it.
1413 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1414 * config/avr/avr.c (avr_option_override)
1415 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1416 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1417 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1418 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1419 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1421 (make_avr_pass_pre_proep): New function.
1422 (emit_push_sfr) <treg>: Add argument to function and use it
1424 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1425 and set machine->gasisr.yes.
1426 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1427 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1428 __gcc_isr.n_pushed to .L__stack_usage.
1429 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1430 (avr_asm_final_postscan_insn): ...this new static function.
1431 * config/avr/avr.h (machine_function)
1432 <is_no_gccisr, use_L__stack_usage>: New fields.
1433 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1434 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1435 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1436 (gasisr, *gasisr): New expander and insn.
1437 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1438 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1439 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1441 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1443 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1446 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1448 Move jump-tables out of .text again.
1451 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1452 (ASM_OUTPUT_ADDR_VEC): New function.
1453 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1454 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1455 INSN_ADDRESSes as asm comment.
1456 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1457 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1458 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1459 * config/avr/avr.md (*tablejump): Adjust comment.
1460 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1461 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1463 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1464 (avr_output_addr_vec): New proto.
1465 (avr_log_t) <insn_addresses>: New field.
1467 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1470 * config/i386/i386.c (ix86_function_arg_advance): Set
1471 outgoing_args_on_stack to true if there are outgoing arguments
1473 (ix86_function_arg): Likewise.
1474 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1475 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1476 * config/i386/i386.h (machine_function): Add
1477 outgoing_args_on_stack.
1479 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1481 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1483 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1485 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1487 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1488 (REAL_H): Remove $(MACHMODE_H).
1489 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1491 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1492 $(MACHMODE_H) and double-int.h.
1493 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1495 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1498 2017-07-07 Andrew Pinski <apinski@cavium.com>
1500 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1501 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1503 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1505 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1506 Add warning if GCC was not configured to link against a GLIBC that
1507 exports the hardware capability bits.
1508 (make_resolver_func): Make resolver function private and not a
1509 COMDAT function. Create the name with clone_function_name instead
1510 of make_unique_name.
1513 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1514 correct operand in doing the split.
1516 2017-07-07 Carl Love <cel@us.ibm.com>
1518 * config/rs6000/rs6000-c: Add support for built-in function
1519 vector unsigned short vec_pack_to_short_fp32 (vector float,
1521 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1522 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1523 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1524 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1525 (convert_4f32_8i16): Add define_expand.
1526 * doc/extend.texi: Update the built-in documentation file for the
1527 new built-in function.
1529 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1531 * config/sparc/m8.md: New file.
1532 * config/sparc/sparc.md: Include m8.md.
1534 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1536 * config/sparc/sparc.opt: New option -mvis4b.
1537 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
1538 (sparc_option_override): Handle VIS4B.
1539 (enum sparc_builtins): Define
1540 SPARC_BUILTIN_DICTUNPACK{8,16,32},
1541 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
1542 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
1543 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
1544 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
1545 (check_constant_argument): New function.
1546 (sparc_vis_init_builtins): Define builtins
1547 __builtin_vis_dictunpack{8,16,32},
1548 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
1549 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
1550 __builtin_vis_fpcmpde{8,16,32}shl and
1551 __builtin_vis_fpcmpur{8,16,32}shl.
1552 (sparc_expand_builtin): Check that the constant operands to
1553 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
1554 constant and in range.
1555 * config/sparc/sparc-c.c (sparc_target_macros): Handle
1557 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
1558 (SPARC_IMM5_P): Likewise.
1559 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
1560 (enabled): Handle vis4b.
1561 (UNSPEC_DICTUNPACK): New unspec.
1562 (UNSPEC_FPCMPSHL): Likewise.
1563 (UNSPEC_FPUCMPSHL): Likewise.
1564 (UNSPEC_FPCMPDESHL): Likewise.
1565 (UNSPEC_FPCMPURSHL): Likewise.
1566 (cpu_feature): New CPU feature `vis4b'.
1567 (dictunpack{8,16,32}): New insns.
1568 (FPCSMODE): New mode iterator.
1569 (fpcscond): New code iterator.
1570 (fpcsucond): Likewise.
1571 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
1572 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
1573 (fpcmpde{8,16,32}{si,di}shl): Likewise.
1574 (fpcmpur{8,16,32}{si,di}shl): Likewise.
1575 * config/sparc/constraints.md: Define constraints `q' for unsigned
1576 2-bit integer constants and `t' for unsigned 5-bit integer
1578 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
1580 (imm5_operand_dictunpack16): Likewise.
1581 (imm5_operand_dictunpack32): Likewise.
1582 (imm2_operand): Likewise.
1583 * doc/invoke.texi (SPARC Options): Document -mvis4b.
1584 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
1585 ditunpack* and fpcmp*shl builtins.
1587 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1589 * config.gcc: Handle m8 in --with-{cpu,tune} options.
1590 * config.in: Add HAVE_AS_SPARC6 define.
1591 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
1593 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
1595 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
1596 (CPP_CPU_SPEC): Handle m8.
1597 (ASM_CPU_SPEC): Likewise.
1598 * config/sparc/sparc-opts.h (enum processor_type): Add
1600 * config/sparc/sparc.c (m8_costs): New struct.
1601 (sparc_option_override): Handle TARGET_CPU_m8.
1602 (sparc32_initialize_trampoline): Likewise.
1603 (sparc64_initialize_trampoline): Likewise.
1604 (sparc_issue_rate): Likewise.
1605 (sparc_register_move_cost): Likewise.
1606 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
1607 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
1608 (ASM_CPU64_DEFAULT_SPEC): Likewise.
1609 (CPP_CPU_SPEC): Handle M8.
1610 (ASM_CPU_SPEC): Likewise.
1611 (AS_M8_FLAG): Define.
1612 * config/sparc/sparc.md: Add m8 to the cpu attribute.
1613 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
1614 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
1616 * configure: Regenerate.
1617 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
1620 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1622 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
1624 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
1625 ("*movdi_insn_sp32"): Do not set v3pipe.
1626 ("*movsi_insn"): Likewise.
1627 ("*movdi_insn_sp64"): Likewise.
1628 ("*movsf_insn"): Likewise.
1629 ("*movdf_insn_sp32"): Likewise.
1630 ("*movdf_insn_sp64"): Likewise.
1631 ("*zero_extendsidi2_insn_sp64"): Likewise.
1632 ("*sign_extendsidi2_insn"): Likewise.
1633 ("*mov<VM32:mode>_insn"): Likewise.
1634 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1635 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1636 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1637 ("<vlop:code><VL:mode>3"): Likewise.
1638 ("*not_<vlop:code><VL:mode>3"): Likewise.
1639 ("*nand<VL:mode>_vis"): Likewise.
1640 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
1641 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
1642 ("one_cmpl<VL:mode>2"): Likewise.
1643 ("faligndata<VM64:mode>_vis"): Likewise.
1644 ("alignaddrsi_vis"): Likewise.
1645 ("alignaddrdi_vis"): Likweise.
1646 ("alignaddrlsi_vis"): Likewise.
1647 ("alignaddrldi_vis"): Likewise.
1648 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1649 ("bmaskdi_vis"): Likewise.
1650 ("bmasksi_vis"): Likewise.
1651 ("bshuffle<VM64:mode>_vis"): Likewise.
1652 ("cmask8<P:mode>_vis"): Likewise.
1653 ("cmask16<P:mode>_vis"): Likewise.
1654 ("cmask32<P:mode>_vis"): Likewise.
1655 ("pdistn<P:mode>_vis"): Likewise.
1656 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1658 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1660 * config/sparc/sparc.md ("subtype"): New insn attribute.
1661 ("*wrgsr_sp64"): Set insn subtype.
1662 ("*rdgsr_sp64"): Likewise.
1663 ("alignaddrsi_vis"): Likewise.
1664 ("alignaddrdi_vis"): Likewise.
1665 ("alignaddrlsi_vis"): Likewise.
1666 ("alignaddrldi_vis"): Likewise.
1667 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
1668 ("fexpand_vis"): Likewise.
1669 ("fpmerge_vis"): Likewise.
1670 ("faligndata<VM64:mode>_vis"): Likewise.
1671 ("bshuffle<VM64:mode>_vis"): Likewise.
1672 ("cmask8<P:mode>_vis"): Likewise.
1673 ("cmask16<P:mode>_vis"): Likewise.
1674 ("cmask32<P:mode>_vis"): Likewise.
1675 ("fchksm16_vis"): Likewise.
1676 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
1677 ("fmean16_vis"): Likewise.
1678 ("fp<plusminus_insn>64_vis"): Likewise.
1679 ("<plusminus_insn>v8qi3"): Likewise.
1680 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
1681 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
1682 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
1683 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
1684 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
1685 ("*movqi_insn"): Likewise.
1686 ("*movhi_insn"): Likewise.
1687 ("*movsi_insn"): Likewise.
1688 ("movsi_pic_gotdata_op"): Likewise.
1689 ("*movdi_insn_sp32"): Likewise.
1690 ("*movdi_insn_sp64"): Likewise.
1691 ("movdi_pic_gotdata_op"): Likewise.
1692 ("*movsf_insn"): Likewise.
1693 ("*movdf_insn_sp32"): Likewise.
1694 ("*movdf_insn_sp64"): Likewise.
1695 ("*zero_extendhisi2_insn"): Likewise.
1696 ("*zero_extendqihi2_insn"): Likewise.
1697 ("*zero_extendqisi2_insn"): Likewise.
1698 ("*zero_extendqidi2_insn"): Likewise.
1699 ("*zero_extendhidi2_insn"): Likewise.
1700 ("*zero_extendsidi2_insn_sp64"): Likewise.
1701 ("ldfsr"): Likewise.
1702 ("prefetch_64"): Likewise.
1703 ("prefetch_32"): Likewise.
1704 ("tie_ld32"): Likewise.
1705 ("tie_ld64"): Likewise.
1706 ("*tldo_ldub_sp32"): Likewise.
1707 ("*tldo_ldub1_sp32"): Likewise.
1708 ("*tldo_ldub2_sp32"): Likewise.
1709 ("*tldo_ldub_sp64"): Likewise.
1710 ("*tldo_ldub1_sp64"): Likewise.
1711 ("*tldo_ldub2_sp64"): Likewise.
1712 ("*tldo_ldub3_sp64"): Likewise.
1713 ("*tldo_lduh_sp32"): Likewise.
1714 ("*tldo_lduh1_sp32"): Likewise.
1715 ("*tldo_lduh_sp64"): Likewise.
1716 ("*tldo_lduh1_sp64"): Likewise.
1717 ("*tldo_lduh2_sp64"): Likewise.
1718 ("*tldo_lduw_sp32"): Likewise.
1719 ("*tldo_lduw_sp64"): Likewise.
1720 ("*tldo_lduw1_sp64"): Likewise.
1721 ("*tldo_ldx_sp64"): Likewise.
1722 ("*mov<VM32:mode>_insn"): Likewise.
1723 ("*mov<VM64:mode>_insn_sp64"): Likewise.
1724 ("*mov<VM64:mode>_insn_sp32"): Likewise.
1726 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1728 * config/sparc/sparc.md ("type"): New insn type viscmp.
1729 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
1731 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
1732 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
1733 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
1734 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
1736 ("n7_vis_logical_11cycle"): Likewise.
1737 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
1738 * config/sparc/niagara2.md ("niag3_vis": Likewise.
1739 * config/sparc/niagara.md ("niag_vis"): Likewise.
1740 * config/sparc/ultra3.md ("us3_fga"): Likewise.
1741 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
1743 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
1745 * config/sparc/sparc.md: New instruction type `bmask'.
1746 (bmaskdi_vis): Use the `bmask' type.
1747 (bmasksi_vis): Likewise.
1748 * config/sparc/ultra3.md (us3_array): Likewise.
1749 * config/sparc/niagara7.md (n7_array): Likewise.
1750 * config/sparc/niagara4.md (n4_array): Likewise.
1751 * config/sparc/niagara2.md (niag2_vis): Likewise.
1752 (niag3_vis): Likewise.
1753 * config/sparc/niagara.md (niag_vis): Likewise.
1755 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1757 * ipa-comdats.c: Remove optimize check from gate.
1758 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
1759 for functions not optimized.
1760 (ipa_fn_summary_read): Skip optimize check.
1761 (ipa_fn_summary_write): Likewise.
1762 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
1764 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
1766 (can_inline_edge_p): Check flag_pcc_struct_return for match.
1767 (check_callers): Give up on caller which is not optimized.
1768 (inline_small_functions): Likewise.
1769 (ipa_inline): Do not give up when not optimizing.
1770 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
1771 away unoptimizes cdtors.
1772 (whole_program_function_and_variable_visibility): Do
1773 ipa_discover_readonly_nonaddressable_vars in LTO mode.
1774 * ipa.c (process_references): Do not check optimize.
1775 (symbol_table::remove_unreachable_nodes): Update optimize check.
1776 (set_writeonly_bit): Update optimize check.
1777 (pass_ipa_cdtor_merge::gate): Do not check optimize.
1778 (pass_ipa_single_use::gate): Remove.
1780 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1782 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
1783 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
1784 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
1785 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
1786 permute_load, permute_store, adjust_extract, adjust_splat,
1787 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
1788 replace_swap_with_copy, dump_swap_insn_table,
1789 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
1790 recombine_lvx_pattern, recombine_stvx_pattern,
1791 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
1792 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
1793 to file rs6000-p8swap.c.
1794 * config/rs6000/rs6000-p8swap.c: New file.
1795 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
1796 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
1797 and rs6000*-*-* targets.
1799 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1801 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
1803 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1805 * lto-wrapper.c (merge_and_complain): Do not merge
1806 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
1807 fsigned_zeros, ftrapping_math, fwrapv.
1808 (append_compiler_options): Do not track these options.
1809 (append_linker_options): Likewie
1811 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1813 * cgraphunit.c (cgraph_node::finalize_function): When
1814 !flag_toplevel_reorde set no_reorder flag.
1815 (varpool_node::finalize_decl): Likewise.
1816 (symbol_table::compile): Drop no toplevel reorder path.
1818 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1820 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
1821 edges; zero probability is not better than uninitialized.
1823 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1825 * asan.h (asan_sanitize_allocas_p): Declare.
1826 * asan.c (asan_sanitize_allocas_p): New function.
1827 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
1828 (handle_builtin_alloca): Likewise.
1829 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
1830 if !asan_sanitize_allocas_p.
1831 * params.def (asan-instrument-allocas): Add new option.
1832 * params.h (ASAN_PROTECT_ALLOCAS): Define.
1833 * opts.c (common_handle_option): Disable allocas sanitization for
1836 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
1838 * asan.c: Include gimple-fold.h.
1839 (get_last_alloca_addr): New function.
1840 (handle_builtin_stackrestore): Likewise.
1841 (handle_builtin_alloca): Likewise.
1842 (asan_emit_allocas_unpoison): Likewise.
1843 (get_mem_refs_of_builtin_call): Add new parameter, remove const
1844 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
1845 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
1846 (instrument_builtin_call): Pass gimple iterator to
1847 get_mem_refs_of_builtin_call.
1848 (last_alloca_addr): New global.
1849 * asan.h (asan_emit_allocas_unpoison): Declare.
1850 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
1851 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
1852 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
1853 if function calls alloca.
1854 * gimple-fold.c (replace_call_with_value): Remove static keyword.
1855 * gimple-fold.h (replace_call_with_value): Declare.
1856 * internal-fn.c: Include asan.h.
1857 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
1858 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
1860 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1862 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
1863 (C_SELFTEST_FLAGS): New.
1864 (CPP_SELFTEST_FLAGS): New.
1865 (SELFTEST_DEPS): New, from deps of s-selftest.
1866 (C_SELFTEST_DEPS): New, from deps of s-selftest.
1867 (CPP_SELFTEST_DEPS): New.
1868 (selftest): Add dependency on s-selftest-c++.
1869 (s-selftest): Rename to...
1870 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
1871 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
1872 than SELFTEST_FLAGS.
1873 (selftest-gdb): Rename to...
1874 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
1876 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
1877 (selftest-valgrind): Rename to...
1878 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
1880 (selftest-valgrind): Reintroduce as an alias for
1881 selftest-c-valgrind.
1882 (s-selftest-c++): New.
1883 (selftest-c++-gdb): New.
1884 (selftest-c++-valgrind): New.
1886 2017-07-06 Olivier Hainque <hainque@adacore.com>
1888 * gcc.c (process_command): When deciding if undefined variables
1889 should be ignored when processing specs, accept "gcc -v" as well.
1891 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
1893 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
1894 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
1896 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1898 * config/arm/arm-cpus.in (armv8-r): Add new entry.
1899 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
1900 * config/arm/arm-tables.opt: Regenerate.
1901 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
1903 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
1905 2017-07-06 Carl Love <cel@us.ibm.com>
1907 * ChangeLog: Clean up from mid air collision
1909 2017-07-06 Carl Love <cel@us.ibm.com>
1911 * config/rs6000/rs6000-c.c: Add support for built-in functions
1912 vector signed int vec_subc (vector signed int, vector signed int);
1913 vector signed __int128 vec_subc (vector signed __int128,
1914 vector signed __int128);
1915 vector unsigned __int128 vec_subc (vector unsigned __int128,
1916 vector unsigned __int128);
1917 vector signed int vec_sube (vector signed int, vector signed int,
1919 vector unsigned int vec_sube (vector unsigned int,
1920 vector unsigned int,
1921 vector unsigned int);
1922 vector signed __int128 vec_sube (vector signed __int128,
1923 vector signed __int128,
1924 vector signed__int128);
1925 vector unsigned __int128 vec_sube (vector unsigned __int128,
1926 vector unsigned __int128,
1927 vector unsigned __int128);
1928 vector signed int vec_subec (vector signed int, vector signed int,
1930 vector unsigned int vec_subec (vector unsigned int,
1931 vector unsigned int,
1932 vector unsigned int);
1933 vector signed __int128 vec_subec (vector signed __int128,
1934 vector signed __int128,
1935 vector signed__int128);
1936 vector unsigned __int128 vec_subec (vector unsigned __int128,
1937 vector unsigned __int128,
1938 vector unsigned __int128);
1939 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
1940 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
1941 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
1942 BU_ALTIVEC_OVERLOAD_X definitions.
1943 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
1944 * doc/extend.texi: Update the built-in documentation file for the new
1947 2017-07-06 David Malcolm <dmalcolm@redhat.com>
1950 * diagnostic-show-locus.c (layout::layout): Use start and finish
1951 spelling location for the start and finish of each range.
1952 * genmatch.c (linemap_client_expand_location_to_spelling_point):
1953 Add unused aspect param.
1954 * input.c (expand_location_1): Add "aspect" param, and use it
1955 to access the correct part of the location.
1956 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
1958 (expand_location_to_spelling_point): Likewise.
1959 (linemap_client_expand_location_to_spelling_point): Add "aspect"
1960 param, and pass it to expand_location_1.
1962 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
1964 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
1965 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
1966 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
1967 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
1968 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
1969 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
1970 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
1971 _mm_maskz_getmant_ss): New intrinsics.
1972 (__builtin_ia32_getexpss128_mask): Changed to ...
1973 __builtin_ia32_getexpss128_round ... this.
1974 (__builtin_ia32_getexpsd128_mask): Changed to ...
1975 __builtin_ia32_getexpsd128_round ... this.
1976 * config/i386/i386-builtin-types.def
1977 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
1978 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
1979 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
1980 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
1981 __builtin_ia32_getmantss_mask_round): New builtins.
1982 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
1983 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
1984 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
1985 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
1986 * config/i386/sse.md
1987 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
1988 avx512f_sgetexp<mode><mask_scalar_name>
1989 <round_saeonly_scalar_name> ... this.
1990 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
1991 %0, %1, %2<round_saeonly_op3>}): Changed to ...
1992 vgetexp<ssescalarmodesuffix>
1993 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
1994 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
1995 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
1996 avx512f_vgetmant<mode><mask_scalar_name>
1997 <round_saeonly_scalar_name> ... this.
1998 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
1999 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2000 vgetmant<ssescalarmodesuffix>
2001 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2002 %0<mask_scalar_operand4>, %1, %2
2003 <round_saeonly_scalar_mask_op4>, %3} ... this.
2004 * config/i386/subst.md (mask_scalar_operand4,
2005 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2006 round_saeonly_scalar_nimm_predicate): New subst attributes.
2008 2017-07-06 Julia Koval <julia.koval@intel.com>
2010 * config/i386/i386.c (ix86_erase_embedded_rounding):
2011 Remove code for old rounding pattern.
2013 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2015 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2017 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2019 * doc/sourcebuild.texi (Test Directives, Variants of
2020 dg-require-support): Add documentation for dg-require-stack-check.
2022 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2024 * config/i386/subst.md (mask_scalar, round_scalar,
2025 round_saeonly_scalar): New meta-templates.
2026 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2027 round_scalar_mask_operand3, round_scalar_mask_op3,
2028 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2029 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2030 round_saeonly_scalar_constraint,
2031 round_saeonly_scalar_prefix): New subst attribute.
2032 * config/i386/sse.md
2033 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2034 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2035 <round_scalar_name> ... this.
2036 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2037 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2038 <round_scalar_name> ... this.
2039 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2040 <sse>_vm<code><mode>3<mask_scalar_name>
2041 <round_saeonly_scalar_name> ... this.
2042 (v<plusminus_mnemonic><ssescalarmodesuffix>
2043 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2044 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2045 v<plusminus_mnemonic><ssescalarmodesuffix>
2046 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2047 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2048 (v<multdiv_mnemonic><ssescalarmodesuffix>
2049 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2050 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2051 v<multdiv_mnemonic><ssescalarmodesuffix>
2052 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2053 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2054 (v<maxmin_float><ssescalarmodesuffix>
2055 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2056 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2057 v<maxmin_float><ssescalarmodesuffix>
2058 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2059 %0<mask_scalar_operand3>, %1, %<iptr>2
2060 <round_saeonly_scalar_mask_op3>} ... this.
2062 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2064 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2065 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2067 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2068 Alan Hayward <alan.hayward@arm.com>
2069 David Sherwood <david.sherwood@arm.com>
2071 * combine.c (simplify_if_then_else): Remove "enum" before
2073 * compare-elim.c (can_eliminate_compare): Likewise.
2074 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2076 (aarch64_lookup_simd_builtin_type): Likewise.
2077 (aarch64_simd_builtin_type): Likewise.
2078 (aarch64_init_simd_builtin_types): Likewise.
2079 (aarch64_simd_expand_args): Likewise.
2080 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2082 (aarch64_reverse_mask): Likewise.
2083 (aarch64_simd_emit_reg_reg_move): Likewise.
2084 (aarch64_gen_adjusted_ldpstp): Likewise.
2085 (aarch64_ccmp_mode_to_code): Likewise.
2086 (aarch64_operands_ok_for_ldpstp): Likewise.
2087 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2088 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2090 (aarch64_min_divisions_for_recip_mul): Likewise.
2091 (aarch64_reassociation_width): Likewise.
2092 (aarch64_get_condition_code_1): Likewise.
2093 (aarch64_simd_emit_reg_reg_move): Likewise.
2094 (aarch64_simd_attr_length_rglist): Likewise.
2095 (aarch64_reverse_mask): Likewise.
2096 (aarch64_operands_ok_for_ldpstp): Likewise.
2097 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2098 (aarch64_gen_adjusted_ldpstp): Likewise.
2099 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2101 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2102 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2103 (arm_lookup_simd_builtin_type): Likewise.
2104 (arm_simd_builtin_type): Likewise.
2105 (arm_init_simd_builtin_types): Likewise.
2106 (arm_expand_builtin_args): Likewise.
2107 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2108 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2109 (ft32_setup_incoming_varargs): Likewise.
2110 (ft32_function_arg): Likewise.
2111 (ft32_function_arg_advance): Likewise.
2112 (ft32_pass_by_reference): Likewise.
2113 (ft32_arg_partial_bytes): Likewise.
2114 (ft32_valid_pointer_mode): Likewise.
2115 (ft32_addr_space_pointer_mode): Likewise.
2116 (ft32_addr_space_legitimate_address_p): Likewise.
2117 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2119 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2120 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2121 (iamcu_alignment): Likewise.
2122 (canonicalize_vector_int_perm): Likewise.
2123 (ix86_noce_conversion_profitable_p): Likewise.
2124 (ix86_mpx_bound_mode): Likewise.
2125 (ix86_operands_ok_for_move_multiple): Likewise.
2126 * config/microblaze/microblaze-protos.h
2127 (microblaze_expand_conditional_branch_reg): Likewise.
2128 * config/microblaze/microblaze.c
2129 (microblaze_expand_conditional_branch_reg): Likewise.
2130 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2132 (rs6000_reassociation_width): Likewise.
2133 (rs6000_invalid_binary_op): Likewise.
2134 (fusion_p9_p): Likewise.
2135 (emit_fusion_p9_load): Likewise.
2136 (emit_fusion_p9_store): Likewise.
2137 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2139 (riscv_hard_regno_mode_ok_p): Likewise.
2140 (riscv_address_insns): Likewise.
2141 (riscv_split_symbol): Likewise.
2142 (riscv_legitimize_move): Likewise.
2143 (riscv_function_value): Likewise.
2144 (riscv_hard_regno_nregs): Likewise.
2145 (riscv_expand_builtin): Likewise.
2146 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2147 (riscv_build_integer): Likewise.
2148 (riscv_split_integer): Likewise.
2149 (riscv_legitimate_constant_p): Likewise.
2150 (riscv_cannot_force_const_mem): Likewise.
2151 (riscv_regno_mode_ok_for_base_p): Likewise.
2152 (riscv_valid_base_register_p): Likewise.
2153 (riscv_valid_offset_p): Likewise.
2154 (riscv_valid_lo_sum_p): Likewise.
2155 (riscv_classify_address): Likewise.
2156 (riscv_legitimate_address_p): Likewise.
2157 (riscv_address_insns): Likewise.
2158 (riscv_load_store_insns): Likewise.
2159 (riscv_force_binary): Likewise.
2160 (riscv_split_symbol): Likewise.
2161 (riscv_force_address): Likewise.
2162 (riscv_legitimize_address): Likewise.
2163 (riscv_move_integer): Likewise.
2164 (riscv_legitimize_const_move): Likewise.
2165 (riscv_legitimize_move): Likewise.
2166 (riscv_address_cost): Likewise.
2167 (riscv_subword): Likewise.
2168 (riscv_output_move): Likewise.
2169 (riscv_canonicalize_int_order_test): Likewise.
2170 (riscv_emit_int_order_test): Likewise.
2171 (riscv_function_arg_boundary): Likewise.
2172 (riscv_pass_mode_in_fpr_p): Likewise.
2173 (riscv_pass_fpr_single): Likewise.
2174 (riscv_pass_fpr_pair): Likewise.
2175 (riscv_get_arg_info): Likewise.
2176 (riscv_function_arg): Likewise.
2177 (riscv_function_arg_advance): Likewise.
2178 (riscv_arg_partial_bytes): Likewise.
2179 (riscv_function_value): Likewise.
2180 (riscv_pass_by_reference): Likewise.
2181 (riscv_setup_incoming_varargs): Likewise.
2182 (riscv_print_operand): Likewise.
2183 (riscv_elf_select_rtx_section): Likewise.
2184 (riscv_save_restore_reg): Likewise.
2185 (riscv_for_each_saved_reg): Likewise.
2186 (riscv_register_move_cost): Likewise.
2187 (riscv_hard_regno_mode_ok_p): Likewise.
2188 (riscv_hard_regno_nregs): Likewise.
2189 (riscv_class_max_nregs): Likewise.
2190 (riscv_memory_move_cost): Likewise.
2191 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2192 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2193 (rl78_addr_space_address_mode): Likewise.
2194 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2196 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2197 (rs6000_reassociation_width): Likewise.
2198 (rs6000_invalid_binary_op): Likewise.
2199 (fusion_p9_p): Likewise.
2200 (emit_fusion_p9_load): Likewise.
2201 (emit_fusion_p9_store): Likewise.
2202 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2203 (ok_for_simple_move_operands): Likewise.
2204 (ok_for_simple_move_strict_operands): Likewise.
2205 (ok_for_simple_arith_logic_operands): Likewise.
2206 (visium_legitimize_reload_address): Likewise.
2207 (visium_select_cc_mode): Likewise.
2208 (output_cbranch): Likewise.
2209 (visium_split_double_move): Likewise.
2210 (visium_expand_copysign): Likewise.
2211 (visium_expand_int_cstore): Likewise.
2212 (visium_expand_fp_cstore): Likewise.
2213 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2214 (visium_function_arg): Likewise.
2215 (visium_function_arg_advance): Likewise.
2216 (visium_libcall_value): Likewise.
2217 (visium_setup_incoming_varargs): Likewise.
2218 (visium_legitimate_constant_p): Likewise.
2219 (visium_legitimate_address_p): Likewise.
2220 (visium_legitimize_address): Likewise.
2221 (visium_secondary_reload): Likewise.
2222 (visium_register_move_cost): Likewise.
2223 (visium_memory_move_cost): Likewise.
2224 (prepare_move_operands): Likewise.
2225 (ok_for_simple_move_operands): Likewise.
2226 (ok_for_simple_move_strict_operands): Likewise.
2227 (ok_for_simple_arith_logic_operands): Likewise.
2228 (visium_function_value_1): Likewise.
2229 (rtx_ok_for_offset_p): Likewise.
2230 (visium_legitimize_reload_address): Likewise.
2231 (visium_split_double_move): Likewise.
2232 (visium_expand_copysign): Likewise.
2233 (visium_expand_int_cstore): Likewise.
2234 (visium_expand_fp_cstore): Likewise.
2235 (visium_split_cstore): Likewise.
2236 (visium_select_cc_mode): Likewise.
2237 (visium_split_cbranch): Likewise.
2238 (output_cbranch): Likewise.
2239 (visium_print_operand_address): Likewise.
2240 * expmed.c (flip_storage_order): Likewise.
2241 * expmed.h (emit_cstore): Likewise.
2242 (flip_storage_order): Likewise.
2243 * genrecog.c (validate_pattern): Likewise.
2244 * hsa-gen.c (gen_hsa_addr): Likewise.
2245 * internal-fn.c (expand_arith_overflow): Likewise.
2246 * ira-color.c (allocno_copy_cost_saving): Likewise.
2247 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2248 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2249 (process_invariant_for_inheritance): Likewise.
2250 * lra-eliminations.c (move_plus_up): Likewise.
2251 * omp-low.c (lower_oacc_reductions): Likewise.
2252 * simplify-rtx.c (simplify_subreg): Likewise.
2253 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2254 (TARGET_CHKP_BOUND_MODE): Likewise..
2255 * targhooks.c (default_chkp_bound_mode): Likewise.
2256 (default_setup_incoming_vararg_bounds): Likewise.
2257 * targhooks.h (default_chkp_bound_mode): Likewise.
2258 (default_setup_incoming_vararg_bounds): Likewise.
2259 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2260 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2261 (have_whole_vector_shift): Likewise.
2262 * tree-vect-stmts.c (vectorizable_load): Likewise.
2263 * doc/tm.texi: Regenerate.
2265 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2267 Graceful degrade if Binutils PR21472 is not available.
2270 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2271 .rodata in flash test fails.
2272 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2273 * confgure: Regenerate.
2274 * config.in: Regenerate.
2275 * config/avr/avr.c (avr_asm_named_section)
2276 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2277 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2278 (avr_asm_init_sections): Same.
2280 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2282 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2283 (fma<VH:mode>4_intrinsic): Likewise.
2284 (*fmsub<VCVTF:mode>4): Likewise.
2285 (*fmsub<VH:mode>4_intrinsic): Likewise.
2287 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2290 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2291 Don't depend on "optimize > 0".
2292 (out_movhi_r_mr, out_movqi_mr_r): Same.
2293 (out_movhi_mr_r, out_movqi_r_mr): Same.
2294 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2295 io_address_operand on "optimize > 0".
2297 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2299 * tree-loop-distribution.c: Add general explanantion on the pass.
2300 (generate_loops_for_partition): Mark distributed loop.
2301 (pg_add_dependence_edges): New parameter. Handle alias data
2302 dependence specially and record it in the parameter if asked.
2303 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2304 (init_partition_graph_vertices, add_partition_graph_edge): New.
2305 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2306 (free_partition_graph_vdata, build_partition_graph): New.
2307 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2308 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2309 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2310 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2311 (version_for_distribution_p, finalize_partitions): New.
2312 (distribute_loop): Handle alias data dependence specially. Factor
2313 out loop fusion code as functions and call these functions.
2315 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2317 * tree-loop-distribution.c (classify_partition): New parameter and
2318 better handle reduction statement.
2319 (rdg_build_partitions): Revise comment.
2320 (distribute_loop): Compute statements in all partitions and pass it
2321 to classify_partition.
2323 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2325 * tree-loop-distribution.c (enum partition_type): New.
2326 (struct partition): New field type.
2327 (partition_merge_into): Add parameter. Update partition type.
2328 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2329 (build_rdg_partition_for_vertex): Compute partition type.
2330 (rdg_build_partitions): Dump partition type.
2331 (distribute_loop): Update calls to partition_merge_into.
2333 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2335 * tree-loop-distribution.c (struct ddr_hasher): New.
2336 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2338 (classify_partition): Call get_data_dependence.
2339 (pg_add_dependence_edges): Ditto.
2340 (distribute_loop): Release data dependence hash table.
2342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2344 * tree-loop-distribution.c (ref_base_address): Delete.
2345 (similar_memory_accesses): Rename ...
2346 (share_memory_accesses): ... to this. Check if partitions access
2347 the same memory reference.
2348 (distribute_loop): Call share_memory_accesses.
2350 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2352 * tree-loop-distribution.c (struct partition): New field recording
2354 (partition_alloc, partition_free): Init and release data refs.
2355 (partition_merge_into): Merge data refs.
2356 (build_rdg_partition_for_vertex): Collect data refs for partition.
2357 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2359 (distribute_loop): Remve data refs from vertice data of partition
2362 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2364 * tree-loop-distribution.c (params.h): Include header file.
2365 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2366 (datarefs_vec): New global var.
2367 (create_rdg_vertices): Use datarefs_vec directly.
2368 (free_rdg): Don't free data references.
2369 (build_rdg): Update use. Don't free data references.
2370 (distribute_loop): Compute global variable for data references.
2371 Bail out if there are too many data references.
2373 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2375 * tree-loop-distribution.c (loop_nest): New global var.
2376 (build_rdg): Use loop directly, rather than loop nest.
2377 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2379 (distribute_loop): Compute global variable loop nest. Update use.
2381 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2383 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2384 (partition_merge_into): New parameter. Dump reason for fusion.
2385 (distribute_loop): Update use of partition_merge_into.
2387 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2389 * tree-loop-distribution.c (bb_top_order_index): New.
2390 (bb_top_order_index_size, bb_top_order_cmp): New.
2391 (stmts_from_loop): Use topological order.
2392 (pass_loop_distribution::execute): Compute and release topological
2393 order for basic blocks.
2395 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2397 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2400 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2402 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2403 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2404 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2405 * internal-fn.def (LOOP_DIST_ALIAS): New.
2406 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2407 (fold_loop_internal_call): ... this.
2408 (vect_loop_dist_alias_call): New function.
2409 (set_uid_loop_bbs): Call fold_loop_internal_call.
2410 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2413 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2416 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2417 Require dead FLAGS_REG at the beginning of a peephole.
2419 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2422 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2423 arguments in the call to __builtin_ia32_sbb_u32.
2424 (_subborrow_u64): Swap _X and _Y arguments in the call to
2425 __builtin_ia32_sbb_u64.
2427 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2430 * tree-vrp.c (compare_assert_loc): Turn into a function template
2431 with stable template parameter. Only test if a->e is NULL,
2432 !a->e == !b->e has been verified already. Use e == NULL or
2433 e != NULL instead of e or ! e tests. If stable is true, don't use
2434 iterative_hash_expr, on the other side allow a or b or both NULL
2435 and sort the NULLs last.
2436 (process_assert_insertions): Sort using compare_assert_loc<false>
2437 instead of compare_assert_loc, later sort using
2438 compare_assert_loc<true> before calling process_assert_insertions_for
2439 in a loop. Use break instead of continue once seen NULL pointer.
2441 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2443 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2444 Cortex-R7 and Cortex-R8 processors.
2446 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2448 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2449 uninitialized while src is not.
2451 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2453 * common/config/arm/arm-common.c: Adjust include path for
2455 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2456 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2457 (arm-cpu-data.h): Likewise.
2458 (arm-cpu-cdata.h): Likewise.
2459 * config/arm/arm-cpu.h: Delete.
2460 * config/arm/arm-cpu-cdata.h: Delete.
2461 * config/arm/arm-cpu-data.h: Delete.
2463 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2465 * config/arm/arm-cpus.in (cortex-a55): New.
2466 (cortex-a75): Likewise.
2467 (cortex-a75.cortex-a55): Likewise.
2468 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2470 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2471 * config/arm/arm-cpu-cdata.h: Regenerate.
2472 * config/arm/arm-cpu-data.h: Regenerate.
2473 * config/arm/arm-cpu.h: Regenerate.
2474 * config/arm/arm-tables.opt: Regenerate.
2475 * config/arm/arm-tune.md: Regenerate.
2477 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2479 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2481 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2483 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2486 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2488 PR tree-optimization/81292
2489 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2490 full_string_p, also call adjust_related_strinfos if the adjustment
2491 is simple, otherwise invalidate related strinfos.
2493 2017-07-04 Martin Liska <mliska@suse.cz>
2496 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2497 newly created variable as DECL_IGNORED_P.
2499 2017-07-04 Martin Liska <mliska@suse.cz>
2502 * ipa-inline.c (inline_small_functions):
2503 Use xstrdup_for_dump.
2505 2017-07-04 Tom de Vries <tom@codesourcery.com>
2507 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2509 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2512 * config/darwin.c (darwin_function_switched_text_sections):
2515 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2517 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2519 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2521 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2523 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2525 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2526 min_profitable_iters, and th as inclusive lower bounds.
2527 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
2528 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
2529 for min_profitable_iters and min_profitable_estimate.
2530 (vect_transform_loop): Treat th as an inclusive lower bound.
2531 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2533 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2536 * config/darwin.c (darwin_function_switched_text_sections):
2537 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
2538 in two pieces, and suppress the use of buf.
2540 2017-07-03 Nathan Sidwell <nathan@acm.org>
2542 * hash-table.h (hash_table_mod1): Fix indentation.
2544 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2547 * predict.c (force_edge_cold): Be more careful about propagation
2549 * profile-count.h (profile_probability::guessed,
2550 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
2552 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
2554 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
2556 * doc/invoke.texi (rcpc architecture extension): Document it.
2558 2017-07-03 Richard Biener <rguenther@suse.de>
2560 PR tree-optimization/60510
2561 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
2562 the scalar reduction PHI and use it.
2563 (vectorizable_reduction): Properly guard the single_defuse_cycle
2564 path for non-SLP reduction chains where we cannot use it.
2565 Rework reduc_def/index and vector type deduction. Rework
2566 vector operand gathering during reduction op code-gen.
2567 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
2568 chains dissolve the chain and leave it to non-SLP reduction
2571 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2573 * tree-data-ref.h (dr_alignment): Declare.
2574 * tree-data-ref.c (dr_alignment): New function.
2575 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
2576 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
2578 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
2580 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2582 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
2583 and base_misalignment fields.
2584 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
2585 * tree-data-ref.c: Include builtins.h.
2586 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
2587 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
2588 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
2589 * tree-vect-data-refs.c: Include tree-cfg.h.
2590 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
2591 fields instead of calculating an alignment here.
2592 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
2593 innermost_loop_behavior fields.
2595 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2597 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
2599 (DR_STEP_ALIGNMENT): New macro.
2600 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
2601 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
2602 (create_data_ref): Print it.
2603 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
2604 to tell whether the step preserves vector (mis)alignment.
2605 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2606 Move the check for an integer step and generalise to all INTEGER_CST.
2607 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
2608 Print the outer step alignment.
2610 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2612 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
2613 with offset_alignment.
2614 (DR_ALIGNED_TO): Delete.
2615 (DR_OFFSET_ALIGNMENT): New macro.
2616 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
2617 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
2618 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
2619 (create_data_ref): Likewise.
2620 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
2621 (vect_analyze_data_refs): Likewise.
2622 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
2623 creating dummy innermost behavior.
2625 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2627 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
2628 with a "innermost_loop_behavior *" and refeence tree.
2629 * tree-data-ref.c (dr_analyze_innermost): Likewise.
2630 (create_data_ref): Update call accordingly.
2631 * tree-predcom.c (find_looparound_phi): Likewise.
2633 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2635 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
2636 fields with dr_wrt_vec_loop.
2637 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
2638 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
2639 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
2640 (vect_dr_behavior): New function.
2641 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2642 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
2643 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
2644 track whether the step preserves the misalignment.
2645 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
2646 Use vect_dr_behavior.
2647 (vect_setup_realignment): Update call accordingly.
2648 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
2649 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
2650 call to vect_create_addr_base_for_vector_ref.
2651 (vect_create_cond_for_align_checks): Likewise.
2652 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
2653 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
2654 (vect_recog_mask_conversion_pattern): Likewise.
2655 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
2656 (new_stmt_vec_info): Remove redundant zeroing.
2658 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2660 * common/config/arm/arm-common.c (arm_be8_option): New function.
2661 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
2662 (ISA_ARMv6): Add isa_bit_be8.
2663 * config/arm/arm.h (arm_be8_option): Add prototype.
2664 (BE8_SPEC_FUNCTION): New define.
2665 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
2666 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
2667 (mlittle-endian): Similarly.
2668 (mbe8, mbe32): New options.
2669 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
2670 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
2672 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2674 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
2676 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2678 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
2679 (cleanup_tree_cfg_bb): Use it.
2680 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
2682 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
2684 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2687 * loop-doloop.c (add_test): Update profile.
2689 2017-07-03 Martin Liska <mliska@suse.cz>
2692 * sanopt.c (rewrite_usage_of_param): New function.
2693 (sanitize_rewrite_addressable_params): Likewise.
2694 (pass_sanopt::execute): Call rewrite_usage_of_param.
2696 2017-07-03 Richard Biener <rguenther@suse.de>
2698 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
2699 back to using VIEW_CONVERT_EXPR.
2701 2017-07-03 Martin Liska <mliska@suse.cz>
2704 * doc/extend.texi: Document when a resolver function is
2705 generated for target_clones.
2707 2017-07-03 Martin Liska <mliska@suse.cz>
2709 * asan.c (asan_emit_stack_protection): Unpoison just red zones
2710 and shadow memory of auto variables which are subject of
2711 use-after-scope sanitization.
2712 (asan_expand_mark_ifn): Add do set only when is_poison.
2714 2016-07-03 Richard Biener <rguenther@suse.de>
2716 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
2718 (vect_force_simple_reduction): Record reduction def -> phi mapping.
2719 (vectorizable_reduction): Perform reduction PHI creation when
2720 visiting a reduction PHI and adjust and simplify code generation
2721 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
2722 (vect_transform_loop): Visit reduction PHIs.
2723 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
2724 defs into the SLP tree.
2725 (vect_build_slp_tree): Reduction defs terminate the recursion.
2726 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
2728 (vect_get_vec_defs_for_stmt_copy): Export.
2729 (vect_get_vec_defs): Likewise.
2730 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
2732 (vect_get_vec_defs_for_stmt_copy): Declare.
2733 (vect_get_vec_defs): Likewise.
2735 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2737 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
2738 parameter with a "loop" parameter and use it instead of the
2739 loop containing DR_STMT. Don't check simple_iv when doing
2740 BB analysis. Describe the two analysis modes in the comment.
2742 2017-07-03 Tom de Vries <tom@codesourcery.com>
2744 PR tree-optimization/69468
2745 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
2746 (find_same_succ_bb): Handle ignore_edge_flags.
2748 2017-07-03 Tom de Vries <tom@codesourcery.com>
2750 PR tree-optimization/81192
2751 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
2753 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
2755 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
2757 2017-07-03 Tom de Vries <tom@codesourcery.com>
2759 PR tree-optimization/81192
2760 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
2761 BB_SAME_SUCC (bb) == NULL.
2763 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2765 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
2768 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2770 * dumpfile.c: Include profile-count.h
2771 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
2773 (insert_cond_bb): Update profile.
2774 * tree-cfg.h (insert_cond_bb): Update prototype.
2775 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
2776 * tree-dump.c: Do not include tree-cfg.
2778 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2780 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
2782 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2784 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
2787 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
2789 * tree-complex.c (expand_complex_div_wide): update profile.
2791 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2792 Alan Hayward <alan.hayward@arm.com>
2793 David Sherwood <david.sherwood@arm.com>
2795 * Makefile.in (MACHMODE_H): Remove insn-modes.h
2796 (CORETYPES_H): New define.
2797 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
2798 (insn-modes-inline.h, s-modes-inline-h): New rules.
2799 (generated_files): Add insn-modes-inline.h.
2800 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
2801 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
2802 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
2803 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
2804 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
2805 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
2806 (build/gencodes.o, build/genconditions.o): Likewise.
2807 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
2808 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
2809 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
2810 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
2811 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
2812 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
2813 * coretypes.h: Include everything up to real.h for generators.
2814 Include insn-modes.h first. Include wide-int-print.h after
2815 wide-int.h. Include insn-modes-inline.h and then machmode.h.
2816 * machmode.h: Don't include insn-modes.h here.
2817 * function-tests.c: Remove includes of signop.h, machmode.h,
2818 double-int.h and wide-int.h.
2820 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
2822 * optc-save-gen.awk: Likewise.
2823 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
2824 * godump.c: Remove include of wide-int-print.h.
2825 * pretty-print.h: Likewise.
2826 * wide-int-print.cc: Likewise.
2827 * wide-int.cc: Likewise.
2828 * hash-map-tests.c: Remove include of signop.h.
2829 * hash-set-tests.c: Likewise.
2830 * rtl-tests.c: Likewise.
2831 * mkconfig.sh: Remove include of machmode.h.
2832 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
2834 (emit_insn_modes_inline_h): ...this new function. Emit the code
2835 into an insn-modes-inline.h header file, adding appropriate
2836 include guards and end comments.
2837 (emit_insn_modes_c_header): Remove include of machmode.h.
2838 (emit_min_insn_modes_c_header): Include coretypes.h rather than
2840 (main): Handle -i flag and call emit_insn_modes_inline_h when
2843 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2845 * tree-ssa-strlen.c (strinfo): Rename the length field to
2846 nonzero_chars. Add a full_string_p field.
2847 (compare_nonzero_chars, zero_length_string_p): New functions.
2848 (get_addr_stridx): Add an offset_out parameter.
2849 Use compare_nonzero_chars.
2850 (get_stridx): Update accordingly. Use compare_nonzero_chars.
2851 (new_strinfo): Update after above changes to strinfo.
2852 (set_endptr_and_length): Set full_string_p.
2853 (get_string_length): Update after above changes to strinfo.
2854 (unshare_strinfo): Update call to new_strinfo.
2855 (maybe_invalidate): Likewise.
2856 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
2857 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
2858 as a uhwi instead of an shwi. Update after above changes to
2859 strinfo and new_strinfo.
2860 (zero_length_string): Assert that chainsi contains full strings.
2861 Use zero_length_string_p. Update call to new_strinfo.
2862 (adjust_related_strinfos): Update after above changes to strinfo.
2863 Copy full_string_p from origsi.
2864 (adjust_last_stmt): Use zero_length_string_p.
2865 (handle_builtin_strlen): Update after above changes to strinfo and
2866 new_strinfo. Install the lhs as the string length if the previous
2867 entry didn't describe a full string.
2868 (handle_builtin_strchr): Update after above changes to strinfo
2870 (handle_builtin_strcpy): Likewise.
2871 (handle_builtin_strcat): Likewise.
2872 (handle_builtin_malloc): Likewise.
2873 (handle_pointer_plus): Likewise.
2874 (handle_builtin_memcpy): Likewise. Track nonzero characters
2875 that aren't necessarily followed by a nul terminator.
2876 (handle_char_store): Likewise.
2878 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2880 PR tree-optimization/80769
2881 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
2882 for malloc and calloc. Document the new invariant that all related
2883 strinfos have delayed lengths or none do.
2884 (verify_related_strinfos): Move earlier in file.
2885 (set_endptr_and_length): New function, split out from...
2886 (get_string_length): ...here. Also set the lengths of related
2888 (zero_length_string): Assert that chainsi has known (rather than
2890 (adjust_related_strinfos): Likewise.
2892 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
2894 PR tree-optimization/81136
2895 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
2896 assert that two references with the same misalignment have the same
2897 compile-time misalignment if those compile-time misalignments
2900 2017-07-01 Andi Kleen <ak@linux.intel.com>
2902 * print-tree.c (print_node): Print all attributes.
2904 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2906 * cfg.c (scale_bbs_frequencies): New function.
2907 * cfg.h (scale_bbs_frequencies): Declare it.
2908 * cfgloopanal.c (single_likely_exit): Cleanup.
2909 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
2911 (scale_loop_profile): Likewise.
2912 (loop_version): Likewise.
2913 (create_empty_loop_on_edge): Update.
2914 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
2915 scale_loop_frequencies, scale_loop_profile, loopify,
2916 loop_version): Update prototypes.
2917 * modulo-sched.c (sms_schedule): Update.
2918 * predict.c (unlikely_executed_edge_p): Also check probability.
2919 (probably_never_executed_edge_p): Fix typo.
2920 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2921 * tree-parloops.c (gen_parallel_loop): Update.
2922 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
2923 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2924 * tree-ssa-loop-split.c (split_loop): Update.
2925 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2926 * tree-vect-loop-manip.c (vect_do_peeling): Update.
2927 (vect_loop_versioning): Update.
2928 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2930 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2932 * trans-mem.c (split_bb_make_tm_edge): Update profile.
2934 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2936 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
2937 to keep profile consistent.
2939 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2941 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
2942 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
2943 * profile-count.h (max_safe_multiplier): Make unsigned.
2944 (profile_count::guessed_zero): New.
2946 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2948 * bb-reorder.c (fix_up_crossing_landing_pad,
2949 fix_crossing_conditional_branches): Use make_single_succ_edge
2950 to keep profile consistent.
2952 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
2954 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
2957 2017-07-01 Jakub Jelinek <jakub@redhat.com>
2960 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
2961 the right scopes, make sure cond_jump isn't preserved between multiple
2962 iterations. Search for fallthru edge whenever there are 3+ edges and
2963 use find_fallthru_edge for it.
2965 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2967 Patch by Alexander Monakov <amonakov@ispras.ru>
2968 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
2969 probabilities consistently.
2971 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2973 * pa.c (pa_expand_compare_and_swap_loop): Update call of
2974 emit_cmp_and_jump_insns.
2976 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
2979 * tree-inline.c (expand_call_inline): Combine profile statuses.
2981 2017-06-30 Andrew Pinski <apinski@cavium.com>
2983 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
2984 fold_stmt returned true.
2986 2017-06-30 Nathan Sidwell <nathan@acm.org>
2988 * ggc.h (empty_string): Delete.
2989 * cfgexpand.c (expand_asm_stmt): Use plain "".
2990 * optabs.c (expand_asm_memory_barrier): Likewise.
2991 * stringpool.c (empty_string): Delete.
2992 (digit_vector, digit_string): Delete.
2993 (ggc_alloc_string): Use plain "", don't optimize single digit
2994 strings. Use ggc_alloc_atomic.
2996 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
2998 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
2999 comparison set and one other set, use the cost of the non-comparison
3002 2017-06-30 Nathan Sidwell <nathan@acm.org>
3004 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3007 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3009 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3010 loops. Remove now unneeded calls to gimple_switch_set_label() that
3011 just set removed labels to NULL_TREE.
3013 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3015 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3016 (set_range_info): ...here. Only call set_range_info_raw if domain
3018 (set_nonzero_bits): Call set_range_info_raw.
3019 * tree-ssanames.h (set_range_info_raw): New.
3021 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3024 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3025 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3026 of nonimmediate_operand and <store_mask_constraint> instead of m
3027 for the input operand. For V8FI iterator, always split if input
3028 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3029 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3030 <store_mask_predicate> instead of register_operand and
3031 <store_mask_constraint> instead of v for the input operand. Make
3032 sure both operands aren't MEMs for if not <mask_applied>.
3034 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3036 * lto-wrapper.c (copy_file) Close both file descriptors before
3039 2017-06-30 Martin Liska <mliska@suse.cz>
3042 * multiple_target.c (create_dispatcher_calls): Make ifunc
3043 also for function that don't have calls or are not referenced.
3045 2017-06-30 Richard Biener <rguenther@suse.de>
3047 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3048 analyze the first scalar stmt. Move vector type computation
3049 for the BB case here from ...
3050 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3051 live operation processing in the SLP case properly.
3053 2017-06-30 Richard Biener <rguenther@suse.de>
3055 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3057 2017-06-30 Martin Liska <mliska@suse.cz>
3060 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3061 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3063 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3065 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3068 2017-06-30 Richard Biener <rguenther@suse.de>
3070 PR tree-optimization/81249
3071 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3072 condition reduction result to original scalar type.
3074 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3076 * profile-count.h (enum profile_quality): Fix typos and whitespace
3079 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3081 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3082 type for branch probabilities.
3084 2017-06-29 Julian Brown <julian@codesourcery.com>
3085 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3087 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3088 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3089 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3090 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3092 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3094 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3095 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3096 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3097 CC usage from generic code to here.
3098 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3099 CC usage into the target macros.
3101 2017-06-29 Maya Rashish <coypu@sdf.org>
3103 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3106 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3108 * arm/arm-builtins.c: Include profile-count.h
3109 * except.c (sjlj_emit_function_enter): Use
3110 profile_probability::unlikely.
3112 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3114 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3115 and tocrel_offset be pointer args rather than implicitly using
3117 (legitimate_constant_pool_address_p, rs6000_emit_move,
3118 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3119 tocrel_offset and use in toc_relative_expr_p call.
3120 (print_operand, print_operand_address): Use static tocrel_base_oac
3121 and tocrel_offset_oac.
3122 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3125 2017-06-29 Maya Rashish <coypu@sdf.org>
3127 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3129 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3131 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3132 objects, take into account only the alignment of 'op0' and 'mode1' if
3135 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3137 * ccmp.c (ccmp_tree_comparison_p): New function.
3138 (ccmp_candidate_p): Update to use above function.
3139 (get_compare_parts): New function.
3140 (expand_ccmp_next): Update to use new functions.
3141 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3143 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3144 take mode as argument.
3145 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3146 * expr.c (expand_expr_real_1): Pass mode as argument.
3148 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3150 * combine.c (combine_instructions): Print insns to dump_file, together
3153 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3155 * asan.c (asan_emit_stack_protection): Update.
3156 (create_cond_insert_point): Update.
3157 * auto-profile.c (afdo_propagate_circuit): Update.
3158 * basic-block.h (struct edge_def): Turn probability to
3159 profile_probability.
3160 (EDGE_FREQUENCY): Update.
3161 * bb-reorder.c (find_traces_1_round): Update.
3162 (better_edge_p): Update.
3163 (sanitize_hot_paths): Update.
3164 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3165 (make_single_succ_edge): Update.
3166 (check_bb_profile): Update.
3167 (dump_edge_info): Update.
3168 (update_bb_profile_for_threading): Update.
3169 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3170 probabilitycount to 0.
3171 * cfgbuild.c (compute_outgoing_frequencies): Update.
3172 * cfgcleanup.c (try_forward_edges): Update.
3173 (outgoing_edges_match): Update.
3174 (try_crossjump_to_edge): Update.
3175 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3176 (expand_gimple_tailcall): Update.
3177 (construct_init_block): Use make_single_succ_edge.
3178 (construct_exit_block): Use make_single_succ_edge.
3179 * cfghooks.c (verify_flow_info): Update.
3180 (redirect_edge_succ_nodup): Update.
3181 (split_edge): Update.
3182 (account_profile_record): Update.
3183 * cfgloopanal.c (single_likely_exit): Update.
3184 * cfgloopmanip.c (scale_loop_profile): Update.
3185 (set_zero_probability): Remove.
3186 (duplicate_loop_to_header_edge): Update.
3187 * cfgloopmanip.h (loop_version): Update prototype.
3188 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3189 (force_nonfallthru_and_redirect): Update.
3190 (update_br_prob_note): Update.
3191 (rtl_verify_edges): Update.
3192 (purge_dead_edges): Update.
3193 (rtl_lv_add_condition_to_bb): Update.
3194 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3195 * cgraphunit.c (init_lowered_empty_function): Update.
3196 (cgraph_node::expand_thunk): Update.
3197 * cilk-common.c: Include profile-count.h
3198 * dojump.c (inv): Remove.
3199 (jumpifnot): Update.
3200 (jumpifnot_1): Update.
3201 (do_jump_1): Update.
3203 (do_jump_by_parts_greater_rtx): Update.
3204 (do_compare_rtx_and_jump): Update.
3205 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3206 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3207 * dwarf2cfi.c: Include profile-count.h
3208 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3209 (sjlj_emit_dispatch_table): Likewise.
3210 * explow.c: Include profile-count.h
3211 * expmed.c (emit_store_flag_force): Update.
3212 (do_cmp_and_jump): Update.
3213 * expr.c (compare_by_pieces_d::generate): Update.
3214 (compare_by_pieces_d::finish_mode): Update.
3215 (emit_block_move_via_loop): Update.
3216 (store_expr_with_bounds): Update.
3217 (store_constructor): Update.
3218 (expand_expr_real_2): Update.
3219 (expand_expr_real_1): Update.
3220 * expr.h (try_casesi, try_tablejump): Update prototypes.
3221 * gimple-pretty-print.c (dump_probability): Update.
3222 (dump_profile): New.
3223 (dump_gimple_label): Update.
3224 (dump_gimple_bb_header): Update.
3225 * graph.c (draw_cfg_node_succ_edges): Update.
3226 * hsa-gen.c (convert_switch_statements): Update.
3227 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3228 (find_if_case_1): Update.
3229 (find_if_case_2): Update.
3230 * internal-fn.c (expand_arith_overflow_result_store): Update.
3231 (expand_addsub_overflow): Update.
3232 (expand_neg_overflow): Update.
3233 (expand_mul_overflow): Update.
3234 (expand_vector_ubsan_overflow): Update.
3235 * ipa-cp.c (good_cloning_opportunity_p): Update.
3236 * ipa-split.c (split_function): Use make_single_succ_edge.
3237 * ipa-utils.c (ipa_merge_profiles): Update.
3238 * loop-doloop.c (add_test): Update.
3239 (doloop_modify): Update.
3240 * loop-unroll.c (compare_and_jump_seq): Update.
3241 (unroll_loop_runtime_iterations): Update.
3242 * lra-constraints.c (lra_inheritance): Update.
3243 * lto-streamer-in.c (input_cfg): Update.
3244 * lto-streamer-out.c (output_cfg): Update.
3245 * mcf.c (adjust_cfg_counts): Update.
3246 * modulo-sched.c (sms_schedule): Update.
3247 * omp-expand.c (expand_omp_for_init_counts): Update.
3248 (extract_omp_for_update_vars): Update.
3249 (expand_omp_ordered_sink): Update.
3250 (expand_omp_for_ordered_loops): Update.
3251 (expand_omp_for_generic): Update.
3252 (expand_omp_for_static_nochunk): Update.
3253 (expand_omp_for_static_chunk): Update.
3254 (expand_cilk_for): Update.
3255 (expand_omp_simd): Update.
3256 (expand_omp_taskloop_for_outer): Update.
3257 (expand_omp_taskloop_for_inner): Update.
3258 * omp-simd-clone.c (simd_clone_adjust): Update.
3259 * optabs.c (expand_doubleword_shift): Update.
3260 (expand_abs): Update.
3261 (emit_cmp_and_jump_insn_1): Update.
3262 (expand_compare_and_swap_loop): Update.
3263 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3264 * predict.c (predictable_edge_p): Update.
3265 (edge_probability_reliable_p): Update.
3266 (set_even_probabilities): Update.
3267 (combine_predictions_for_insn): Update.
3268 (combine_predictions_for_bb): Update.
3269 (propagate_freq): Update.
3270 (estimate_bb_frequencies): Update.
3271 (force_edge_cold): Update.
3272 * profile-count.c (profile_count::dump): Add missing space into dump.
3273 (profile_count::debug): Add newline.
3274 (profile_count::differs_from_p): Explicitly convert to unsigned.
3275 (profile_count::stream_in): Update.
3276 (profile_probability::dump): New member function.
3277 (profile_probability::debug): New member function.
3278 (profile_probability::differs_from_p): New member function.
3279 (profile_probability::differs_lot_from_p): New member function.
3280 (profile_probability::stream_in): New member function.
3281 (profile_probability::stream_out): New member function.
3282 * profile-count.h (profile_count_quality): Rename to ...
3283 (profile_quality): ... this one.
3284 (profile_probability): New.
3285 (profile_count): Update.
3286 * profile.c (compute_branch_probabilities): Update.
3287 * recog.c (peep2_attempt): Update.
3288 * sched-ebb.c (schedule_ebbs): Update.
3289 * sched-rgn.c (find_single_block_region): Update.
3290 (compute_dom_prob_ps): Update.
3291 (schedule_region): Update.
3292 * sel-sched-ir.c (compute_succs_info): Update.
3293 * stmt.c (struct case_node): Update.
3294 (do_jump_if_equal): Update.
3295 (get_outgoing_edge_probs): Update.
3296 (conditional_probability): Update.
3297 (emit_case_dispatch_table): Update.
3298 (expand_case): Update.
3299 (expand_sjlj_dispatch_table): Update.
3300 (emit_case_nodes): Update.
3301 * targhooks.c: Update.
3302 * tracer.c (better_p): Update.
3303 (find_best_successor): Update.
3304 * trans-mem.c (expand_transaction): Update.
3305 * tree-call-cdce.c: Update.
3306 * tree-cfg.c (gimple_split_edge): Upate.
3307 (move_sese_region_to_fn): Upate.
3308 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3309 * tree-eh.c (lower_resx): Upate.
3310 (cleanup_empty_eh_move_lp): Upate.
3311 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3312 * tree-inline.c (copy_edges_for_bb): Update.
3313 (copy_cfg_body): Update.
3314 * tree-parloops.c (gen_parallel_loop): Update.
3315 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3316 (gimple_gen_time_profiler): Update.
3317 * tree-ssa-dce.c (remove_dead_stmt): Update.
3318 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3319 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3320 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3321 (unloop_loops): Update.
3322 (try_peel_loop): Update.
3323 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3324 * tree-ssa-loop-split.c (connect_loops): Update.
3325 (split_loop): Update.
3326 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3327 (hoist_guard): Update.
3328 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3329 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3330 (value_replacement): Update.
3331 * tree-ssa-reassoc.c (branch_fixup): Update.
3332 * tree-ssa-tail-merge.c (replace_block_by): Update.
3333 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3334 (create_edge_and_update_destination_phis): Update.
3335 (compute_path_counts): Update.
3336 (recompute_probabilities): Update.
3337 (update_joiner_offpath_counts): Update.
3338 (freqs_to_counts_path): Update.
3339 (duplicate_thread_path): Update.
3340 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3341 (struct switch_conv_info): Update.
3342 (gen_inbound_check): Update.
3343 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3344 (vect_do_peeling): Update.
3345 (vect_loop_versioning): Update.
3346 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3347 (optimize_mask_stores): Update.
3348 * ubsan.c (ubsan_expand_null_ifn): Update.
3349 * value-prof.c (gimple_divmod_fixed_value): Update.
3350 (gimple_divmod_fixed_value_transform): Update.
3351 (gimple_mod_pow2): Update.
3352 (gimple_mod_pow2_value_transform): Update.
3353 (gimple_mod_subtract): Update.
3354 (gimple_mod_subtract_transform): Update.
3355 (gimple_ic): Update.
3356 (gimple_stringop_fixed_value): Update.
3357 (gimple_stringops_transform): Update.
3358 * value-prof.h: Update.
3360 2017-06-29 Carl Love <cel@us.ibm.com>
3362 * config/rs6000/rs6000-c.c: Add support for built-in functions
3363 vector signed int vec_signed (vector float);
3364 vector signed long long vec_signed (vector double);
3365 vector signed int vec_signed2 (vector double, vector double);
3366 vector signed int vec_signede (vector double);
3367 vector signed int vec_signedo (vector double);
3368 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3369 instruction generator.
3370 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3371 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3372 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3374 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3375 vunsignede_v2df): Add define_expands.
3376 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3377 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3378 VEC_UNSIGNEDO): Add definitions.
3379 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3380 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3381 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3382 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3383 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3384 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3385 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3386 vec_unsignede, vec_unsignedo): Add builtin defines.
3387 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3389 * doc/extend.texi: Update the built-in documentation file for the
3390 new built-in functions.
3392 2017-06-29 Richard Biener <rguenther@suse.de>
3394 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3395 reduction chains to LOOP_VINFO_REDUCTIONS.
3396 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3397 SLP reductions after processing reduction chains.
3399 2017-06-29 Nathan Sidwell <nathan@acm.org>
3401 * builtins.c (fold_builtin_FUNCTION): Use
3402 lang_hooks.decl_printable_name.
3404 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3407 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3408 with only one label.
3409 * stmt.c (expand_case): Assert NCASES is greater than one.
3411 2017-06-29 Richard Biener <rguenther@suse.de>
3413 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3415 (group_case_labels): Likewise.
3416 (find_taken_edge): Push sanity checking on val to workers...
3417 (find_taken_edge_cond_expr): ... here
3418 (find_taken_edge_switch_expr): ... and here, handle cases
3419 with just a default label.
3420 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3421 (group_case_labels): Likewise.
3422 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3423 group_case_labels does anything cleanup the CFG again.
3425 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3427 PR tree-optimization/81196
3428 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3429 exit condition comparing two IVs.
3431 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3433 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3434 profile to the dummy entry at the end of the list of architectures.
3435 * config/arm/arm-cpu-cdata.h: Regenerated.
3437 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3438 Michael Collison <michael.collison@arm.com>
3441 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3443 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3444 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3445 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3446 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3447 of shift when the shift amount is masked with constant equal to
3448 the size of the mode.
3449 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3452 2017-06-29 Martin Liska <mliska@suse.cz>
3454 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3456 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3459 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3462 (edge_growth_cache_entry::edge_growth_cache_entry): New
3464 (reset_edge_growth_cache): Update to use constructor.
3466 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3468 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3469 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3470 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3472 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3474 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3475 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3477 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3479 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3480 (*-linux-uclibc*): Add t-uclibc tmake_file.
3481 * config/t-musl: New.
3482 * config/t-uclibc: New.
3484 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3486 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3488 (gen_comm_data): Emit architectural setting of arch_prof.
3489 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3491 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3492 (armv8-m.base, armv8-m.main): Likewise.
3493 * arm-protos.h (arm_build_target): Add profile field.
3494 (arch_option): Likewise.
3495 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3497 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3498 arm_active_target.profile.
3500 2017-06-28 Richard Biener <rguenther@suse.de>
3503 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3504 TYPE_OVERFLOW_WRAPS.
3505 * match.pd (negate_expr_p): Likewise.
3506 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3507 fold_build2, not fold_binary.
3509 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3511 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3512 Convert memory address to Pmode.
3513 (aarch64_print_operand): Assert MEM operands are always Pmode.
3515 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3518 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3519 Remove redundant if.
3520 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3521 * config/arm/aarch-common-protos.h
3522 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3523 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3525 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3528 * multiple_target.c (create_dispatcher_calls): Set the default
3529 clone to be static, not public.
3531 2017-06-28 Richard Biener <rguenther@suse.de>
3533 * tree-vect-loop.c (vectorizable_reduction): Move special
3534 cond reduction IV var creation ...
3535 (vect_create_epilog_for_reduction): ... here. Remove induction_index
3536 parameter. Use STMT_VINFO_VECTYPE.
3537 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
3540 2017-06-28 Martin Liska <mliska@suse.cz>
3543 * ipa-visibility.c (non_local_p): Handle visibility.
3545 2017-06-28 Martin Liska <mliska@suse.cz>
3548 * common.opt: Add IntegerRange to various options.
3549 * opt-functions.awk (integer_range_info): New function.
3550 * optc-gen.awk: Add integer_range_info to cl_options struct.
3551 * opts-common.c (decode_cmdline_option): Handle
3552 CL_ERR_INT_RANGE_ARG.
3553 (cmdline_handle_error): Likewise.
3554 * opts.c (print_filtered_help): Show valid interval in
3555 when --help is provided.
3556 * opts.h (struct cl_option): Add range_min and range_max fields.
3557 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
3559 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
3561 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
3562 (x * C EQ/NE y * C): New transformation.
3564 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
3566 * genmultilib (combination_space): Accept '+' in option names.
3568 2017-06-28 Martin Liska <mliska@suse.cz>
3571 * asan.c (instrument_derefs): Bail out inner references
3572 that are hard register variables.
3574 2017-06-28 Jakub Jelinek <jakub@redhat.com>
3577 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
3578 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
3580 2017-06-28 Richard Biener <rguenther@suse.de>
3582 * tree-vectorizer.h (vect_get_vec_defs): Remove.
3583 (vect_get_slp_defs): Adjust.
3584 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
3586 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
3588 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3589 get_initial_defs_for_reduction instead of vect_get_vec_defs.
3590 (vectorizable_reduction): Adjust.
3591 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
3593 (vect_get_slp_defs): Likewise.
3594 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
3595 (vectorizable_bswap): Adjust.
3596 (vectorizable_call): Likewise.
3597 (vectorizable_conversion): Likewise.
3598 (vectorizable_assignment): Likewise.
3599 (vectorizable_shift): Likewise.
3600 (vectorizable_operation): Likewise.
3601 (vectorizable_store): Likewise.
3602 (vectorizable_condition): Likewise.
3603 (vectorizable_comparison): Likewise.
3605 2017-06-28 Michael Collison <michael.collison@arm.com>
3608 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
3610 (arm_gen_movmemqi): Removed unused variable 'i'.
3611 Convert 'for' loop into 'while' loop.
3612 (arm_expand_prologue): Remove last unnecessary set of insn.
3613 (thumb_pop): Remove unused variable 'pushed_words'.
3614 (thumb_exit): Remove last unnecessary set of regs_to_pop.
3616 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3618 * config/s390/predicates.md: Use s390_rel_address_ok_p.
3619 * config/s390/s390-protos.h: Add prototype of
3620 s390_rel_address_ok_p.
3621 * config/s390/s390.c (s390_got_symbol): New function.
3622 (s390_rel_address_ok_p): New function.
3623 (legitimize_pic_address): Use s390_rel_address_ok_p.
3624 (s390_load_got): Use s390_got_symbol.
3625 (s390_option_override): Issue error if
3626 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
3627 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
3629 * config/s390/s390.opt: New option mpic-data-is-text-relative.
3631 2017-06-27 Andrew Pinski <apinski@cavium.com>
3633 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
3634 (X * copysign (1.0, X)): New pattern.
3635 (X * copysign (1.0, -X)): New pattern.
3636 (copysign (-1.0, CST)): New pattern.
3638 2017-06-27 Joseph Myers <joseph@codesourcery.com>
3640 * genmultilib (combination_space): Remove variable.
3641 Validate reuse rules against regular expression for any sequence
3642 of multilib options in any order.
3644 2017-06-27 Michael Collison <michael.collison@arm.com>
3646 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
3647 call aarch64_split_simd_combine.
3648 * (aarch64_combine_internal<mode>): Delete pattern.
3649 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
3650 Allow register and subreg operands.
3652 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3654 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
3655 specific need, just fallback on defaults.
3656 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
3658 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3659 Olivier Hainque <hainque@adacore.com>
3661 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
3663 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
3664 targets. Pick a default if no particular attempt applied.
3665 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
3668 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3670 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
3671 (x86_64-wrs-vxworks7): Likewise.
3673 2017-06-27 Marek Polacek <polacek@redhat.com>
3676 * ubsan.c (instrument_null): Check get_base_address's result for null.
3678 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3680 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
3682 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
3684 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
3685 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
3686 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
3687 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
3689 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
3690 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
3691 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
3692 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
3693 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
3694 BUILT_IN_FEUPDATEENV): New builtins.
3695 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
3696 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
3697 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
3698 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
3700 (builtin_structptr_types): Adjust size.
3701 * tree.c (builtin_structptr_types): Add four entries.
3703 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3704 Olivier Hainque <hainque@adacore.com>
3706 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
3707 (TLS_SYM): New local macro, forcing reference to __tls__ on
3708 link command lines for VxWorks 7 RTPs, triggering initialization
3710 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
3711 OS features TLS support, true for RTPs on VxWorks 7.
3712 * config/vxworks.c (vxworks_override_options): Setup emutls
3715 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3717 * predict.c (test_prediction_value_range): Use -1U instead of -1
3718 to avoid narrowing conversion warning.
3719 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
3720 to avoid narrowing conversion warning.
3721 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
3723 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
3725 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
3727 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
3728 64bit configurations.
3729 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
3730 (SIZE_TYPE): Likewise.
3731 * config/vxworks.c (vxworks_emutls_var_fields): Use
3732 long_unsigned_type_node instead of unsigned_type_node as the offset
3733 field type, which is "pointer" mode in emutls.c.
3735 2017-06-27 Jakub Jelinek <jakub@redhat.com>
3738 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
3741 * gimple-fold.c (replace_call_with_call_and_fold): Handle
3742 gimple_vuse copying separately from gimple_vdef copying.
3744 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3746 * value-prof.c (free_hist): Remove call to memset and the enclosing if
3749 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3750 Olivier Hainque <hainque@adacore.com>
3752 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
3753 for all vxworks7 targets.
3754 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
3755 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
3756 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
3757 variations for VX6/VX7 and 32/64bits later on in ...
3758 (VXWORKS_LIB_SPEC): Leverage new macros.
3759 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
3760 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
3762 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
3764 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
3765 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
3767 2017-06-26 Carl Love <cel@us.ibm.com>
3769 * config/rs6000/rs6000-c.c: Add support for built-in functions
3770 vector bool char vec_reve (vector bool char);
3771 vector signed char vec_reve (vector signed char);
3772 vector unsigned char vec_reve (vector unsigned char);
3773 vector bool int vec_reve (vector bool int);
3774 vector signed int vec_reve (vector signed int);
3775 vector unsigned int vec_reve (vector unsigned int);
3776 vector bool long long vec_reve (vector bool long long);
3777 vector signed long long vec_reve (vector signed long long);
3778 vector unsigned long long vec_reve (vector unsigned long long);
3779 vector bool short vec_reve (vector bool short);
3780 vector signed short vec_reve (vector signed short);
3781 vector double vec_reve (vector double);
3782 vector float vec_reve (vector float);
3783 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
3784 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
3785 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
3786 (altivec_vreve): New pattern.
3787 * config/rs6000/altivec.h (vec_reve): New define.
3788 * doc/extend.texi (vec_rev): Update the built-in documentation file
3789 for the new built-in functions.
3791 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3793 PR tree-optimization/71815
3794 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
3796 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
3798 (slsr_process_phi): Likewise.
3799 (replace_uncond_cands_and_profitable_phis): Don't replace a
3800 multiply candidate with a stride of 1 (copy or cast).
3801 (phi_incr_cost): Call uses_consumed_by_stmt rather than
3803 (lowest_cost_path): Likewise.
3804 (total_savings): Likewise.
3806 2017-06-26 Richard Biener <rguenther@suse.de>
3809 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
3810 Use def_builtin_pure for all gather builtins.
3812 2017-06-26 Richard Biener <rguenther@suse.de>
3814 PR tree-optimization/81203
3815 * tree-tailcall.c (find_tail_calls): Do not move stmts into
3818 2017-06-26 Marek Polacek <polacek@redhat.com>
3821 * doc/invoke.texi: Document -Wmultistatement-macros.
3823 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
3825 * doc/sourcebuild.texi (ARM-specific attributes): Document new
3826 arm_neon_ok_no_float_abi effective target.
3828 2017-06-26 Richard Biener <rguenther@suse.de>
3830 PR tree-optimization/80928
3831 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
3832 (copy_bbs): Set BB_DUPLICATED flag early.
3833 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
3835 (execute_on_shrinking_pred): Likewise.
3836 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
3837 BB_DUPLICATED blocks.
3838 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
3839 iterate over all PHIs considering removal of *gsi.
3841 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
3843 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
3846 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3848 * config/rs6000/rs6000-string.c: (expand_block_clear,
3849 do_load_for_compare, select_block_compare_mode,
3850 compute_current_alignment, expand_block_compare,
3851 expand_strncmp_align_check, expand_strn_compare,
3852 expand_block_move, rs6000_output_load_multiple)
3853 Move functions related to string/block move/compare
3855 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
3856 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
3857 for this function which is now used in two files.
3858 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
3859 * config.gcc: Add rs6000-string.o to extra_objs for
3860 targets powerpc*-*-* and rs6000*-*-*.
3862 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3865 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
3866 32-bit, since indexed is not valid for DImode.
3867 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
3868 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
3869 (define_peephole2 for Altivec d-form load): Add 32-bit support.
3870 (define_peephole2 for Altivec d-form store): Likewise.
3873 * multiple_target.c (create_dispatcher_calls): Only create the
3874 dispatcher call if the function is the default clone of a
3877 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
3880 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
3881 a call, force the call to not be a tail call.
3883 2017-06-23 Jeff Law <law@redhat.com>
3885 * doc/contrib.texi: Add entry for Steven Pemberton's work on
3888 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
3890 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
3891 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
3892 handling for early expansion of vector shifts (sl,sr,sra,rl).
3893 (builtin_function_type): Add vector shift right instructions
3894 to the unsigned argument list.
3896 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3898 rtl-optimizatoin/79286
3899 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
3900 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
3901 trap. PIC register plus a const unspec without offset can never trap.
3903 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
3905 * tree.h (builtin_structptr_type): New type.
3906 (builtin_structptr_types): Declare new array.
3907 * tree.c (builtin_structptr_types): New array.
3908 (free_lang_data, build_common_tree_nodes): Use it.
3910 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
3913 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
3916 2017-06-22 Matt Turner <mattst88@gmail.com>
3918 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
3919 Lake models to skylake case. Assume skylake for unknown
3920 models with clflushopt.
3922 2017-06-22 Jeff Law <law@redhat.com>
3924 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
3925 frame sizes that do not satisfy aarch64_uimm12_shift.
3927 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
3929 * profile-count.h (apply_probability,
3930 apply_scale, probability_in): Fix checks for zero.
3932 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3934 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
3935 * doc/cppdiropts.texi (-I @var{dir}): Document it.
3937 2016-06-22 Richard Biener <rguenther@suse.de>
3939 * tree-vect-loop.c (vect_model_reduction_cost): Handle
3940 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
3941 REDUC_MAX_EXPR support.
3942 (vectorizable_reduction): Likewise.
3943 (vect_create_epilog_for_reduction): Likewise.
3945 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
3947 * match.pd (A / (1 << B) -> A >> B): New.
3948 * generic-match-head.c: Include optabs-tree.h.
3949 * gimple-match-head.c: Likewise.
3950 * optabs-tree.h (target_supports_op_p): New.
3951 * optabs-tree.c (target_supports_op_p): New.
3953 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3955 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
3956 $gcc_cv_ld --help output.
3957 (gcc_cv_ld_demangle): Likewise.
3958 (gcc_cv_ld_eh_frame_hdr): Likewise.
3959 (gcc_cv_ld_pie): Likewise.
3960 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
3961 (gcc_cv_ld_buildid): Likewise.
3962 (gcc_cv_ld_sysroot): Likewise.
3963 (ld_bndplt_support): Likewise.
3964 (ld_pushpopstate_support): Likewise.
3965 * configure: Regenerate.
3966 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
3968 2017-06-21 Jakub Jelinek <jakub@redhat.com>
3971 * config/i386/sse.md (round<mode>2): Renumber match_dup and
3972 operands indexes to avoid gap between operands and match_dups.
3974 2017-06-21 Andrew Pinski <apinski@cavium.com>
3976 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
3977 Increment Arith_shift and Arith_shift_reg by 1.
3978 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
3980 * config/aarch64/aarch64.c (thunderx_tunings): Enable
3981 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
3982 (aarch64_strip_extend): Add new argument and test for it.
3983 (aarch64_cheap_mult_shift_p): New function.
3984 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
3985 add a cost if it is true.
3986 Update calls to aarch64_strip_extend.
3987 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
3989 2017-06-21 Andrew Pinski <apinski@cavium.com>
3991 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
3993 (thunderxt88): Likewise.
3994 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
3995 (thunderx_prefetch_tune): New variable.
3996 (thunderx2t99_prefetch_tune): Update for the correct values.
3997 (thunderxt88_tunings): New variable.
3998 (thunderx_tunings): Use thunderx_prefetch_tune instead of
3999 generic_prefetch_tune.
4000 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4002 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4004 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4005 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4006 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4007 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4008 (aarch64_atomic_cas<mode>, GPI): Likewise.
4010 2017-06-21 Martin Liska <mliska@suse.cz>
4012 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4013 statements on cold and hot labels.
4014 * predict.c (tree_estimate_probability_bb): Remove the
4015 prediction from this place.
4017 2017-06-21 Martin Liska <mliska@suse.cz>
4019 PR tree-optimization/79489
4020 * gimplify.c (maybe_add_early_return_predict_stmt): New
4022 (gimplify_return_expr): Call the function.
4023 * predict.c (tree_estimate_probability_bb): Remove handling
4025 * predict.def: Update comment about early return predictor.
4026 * gimple-predict.h (is_gimple_predict): New function.
4027 * predict.def: Change default value of early return to 66.
4028 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4030 * passes.def: Put pass_strip_predict_hints to the beginning of
4033 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4035 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4036 FUNCTION_DECL declarations.
4037 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4039 (dwaf2out_decl): Likewise.
4040 * godump.c (go_early_global_decl): Skip call to the real debug hook
4041 for FUNCTION_DECL declarations.
4042 * passes.c (rest_of_decl_compilation): Skip call to the
4043 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4044 -fdump-go-spec is passed.
4046 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4048 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4049 Reorder for compactness.
4050 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4051 (def_builtin_pure, def_builtin_pure2): New functions.
4052 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4054 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4056 * match.pd (nop_convert): New predicate.
4057 ((A +- CST1) +- CST2): Allow some NOP conversions.
4059 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4062 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4063 with ctors/dtors if GOVD_SHARED is set.
4065 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4067 * config/aarch64/aarch64.md (movti_aarch64):
4068 Emit mov rather than orr.
4069 (movtf_aarch64): Likewise.
4070 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4071 Emit mov rather than orr.
4073 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4075 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4076 Swap alternatives, make integer dup more expensive.
4078 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4080 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4081 Return true for non-tls symbols.
4083 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4085 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4086 (cortex-a75): Likewise.
4087 (cortex-a75.cortex-a55): Likewise.
4088 * config/aarch64/aarch64-tune.md: Regenerate.
4089 * doc/invoke.texi (-mtune): Document new values for -mtune.
4091 2017-06-21 Tom de Vries <tom@codesourcery.com>
4093 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4095 (Effective-Target Keywords, Other attributes): Suggest using
4096 dg-add-options stack_size feature to get stack limit in stack_size
4097 effective target documentation.
4099 2017-06-21 Julian Brown <julian@codesourcery.com>
4100 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4102 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4103 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4104 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4106 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4107 attribute type list for neon_multiply.
4108 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4109 attribute type list for neon_multiply.
4110 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4111 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4112 attribute type list for neon_multiply.
4113 * config/arm/types.md (crypto_pmull): Add.
4114 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4115 attribute type list.
4117 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4119 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4122 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4124 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4125 to make sure not to dereference a NULL cost_classes_ptr pointer.
4127 2017-06-20 Carl Love <cel@us.ibm.com>
4129 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4130 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4131 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4132 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4133 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4134 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4135 VMULOSW): New enum "unspec" values.
4136 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4137 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4138 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4139 altivec_vmulosw): New patterns.
4140 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4141 VMULOSW): Add definitions.
4143 2017-06-20 Julia Koval <julia.koval@intel.com>
4145 * config/i386/i386.c: Fix rounding expand for new pattern.
4146 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4148 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4150 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4151 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4153 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4155 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4158 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4160 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4161 architecture, then by implementer ID.
4162 * config/aarch64/aarch64-tune.md: Regenerate.
4164 2017-06-20 Richard Biener <rguenther@suse.de>
4167 * fold-const.c (split_tree): Fold to type before negating.
4169 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4171 * diagnostic-show-locus.c
4172 (selftest::test_fixit_deletion_affecting_newline): New function.
4173 (selftest::diagnostic_show_locus_c_tests): Call it.
4175 2017-06-20 Andreas Schwab <schwab@suse.de>
4178 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4181 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4183 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4184 __ARM_FEATURE_COPROC according to support.
4186 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4188 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4189 Rewritten to avoid overflow for > 32-bit pointers.
4192 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4193 by removing enum keyword.
4194 (ubsan_type_descriptor): Likewise. Formatting fix.
4197 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4198 splitter): Require TARGET_SSE2 in the condition.
4200 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4203 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4204 for doing vector set of SFmode on ISA 3.0.
4205 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4206 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4208 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4209 SFmode value into a V4SF variable that was extracted from another
4210 V4SF variable without converting the element to double precision
4211 and back to single precision vector format.
4212 (vsx_insert_extract_v4sf_p9_2): Likewise.
4214 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4216 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4217 in UWHI to avoid undefined overflow.
4220 * ubsan.h (enum ubsan_encode_value_phase): New.
4221 (ubsan_encode_value): Change second argument to
4222 enum ubsan_encode_value_phase with default value of
4223 UBSAN_ENCODE_VALUE_GENERIC.
4224 * ubsan.c (ubsan_encode_value): Change second argument to
4225 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4226 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4227 create_tmp_var_raw instead of create_tmp_var and use a
4229 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4230 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4231 ubsan_encode_value callers.
4234 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4235 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4236 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4238 2017-06-19 Richard Biener <rguenther@suse.de>
4241 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4242 estimates if we changed anything.
4244 2017-06-19 Richard Biener <rguenther@suse.de>
4246 PR tree-optimization/80887
4247 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4248 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4249 simplified lookups, then reset mprts_hook.
4250 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4252 (try_to_simplify): Likewise.
4254 2017-06-19 Martin Liska <mliska@suse.cz>
4257 * gimplify.c (gimplify_switch_expr):
4258 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4260 2017-06-19 Martin Liska <mliska@suse.cz>
4262 * doc/install.texi: Document that PGO runs in 4 stages.
4264 2017-06-19 Martin Liska <mliska@suse.cz>
4267 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4268 to dispatcher function name.
4269 * multiple_target.c (replace_function_decl): New function.
4270 (create_dispatcher_calls): Redirect both edges and references.
4272 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4274 * profile-count.c (profile_count::dump): Dump quality.
4275 (profile_count::differs_from_p): Update for unsigned val.
4276 * profile-count.h (profile_count_quality): New enum.
4277 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4279 2017-06-19 Richard Biener <rguenther@suse.de>
4281 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4282 struct function as arg.
4283 (estimate_numbers_of_iterations): Export overload with loop arg.
4284 (free_numbers_of_iterations_estimates_loop): Use an overload of
4285 free_numbers_of_iterations_estimates instead.
4286 * tree-cfg.c (remove_bb): Adjust.
4287 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4288 * tree-parloops.c (gen_parallel_loop): Likewise.
4289 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4291 (tree_unroll_loops_completely): Likewise.
4292 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4293 Use an overload instead and export.
4294 (estimated_loop_iterations): Adjust.
4295 (max_loop_iterations): Likewise.
4296 (likely_max_loop_iterations): Likewise.
4297 (estimate_numbers_of_iterations): Take struct function as arg
4299 (loop_exits_before_overflow): Adjust.
4300 (free_numbers_of_iterations_estimates_loop): Use an overload.
4301 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4302 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4304 2017-06-19 Richard Biener <rguenther@suse.de>
4307 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4308 RANGE_EXPR conservatively.
4310 2017-06-16 Carl Love <cel@us.ibm.com>
4312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4313 definitions for vec_float, vec_float2, vec_floato,
4314 vec_floate built-ins.
4315 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4316 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4318 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4319 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4320 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4321 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4322 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4323 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4324 vec_floato): Add builtin defines.
4325 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4326 Update the built-in documentation file for the new built-in
4329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4331 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4332 (mthumb): Mark as the negative of -marm.
4334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4336 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4338 (ARM Options, -mtune): Document that this accepts the same
4339 extension options as -mcpu.
4340 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4342 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4344 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4345 permitted extensions.
4347 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4349 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4350 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4351 (armv8-m.main): Add option +nodsp.
4352 * config/arm/arm-cpu-cdata.h: Regenerated.
4354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4356 * config/arm/t-fuchsia: New file.
4357 * config.gcc (arm*-*-fuchsia*): Use it.
4359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4361 * config/arm/t-symbian: Rewrite for new option infrastructure.
4363 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4365 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4366 (MULTILIB_REQUIRED): Likewise.
4368 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4370 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4371 (MULTILIB_RESUE): Likewise.
4372 (MULTILIB_MATCHES): Likewise.
4373 (MULTLIB_REQUIRED): Likewise.
4375 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4377 * config/arm/t-rtems: Rewrite for new option framework.
4379 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4381 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4382 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4383 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4384 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4385 * config/arm/t-multilib: ... here.
4386 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4387 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4388 armv7-a and armv8*-a when A-profile libraries have not been built.
4389 * config/arm/t-rmprofile: Rewrite.
4391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4393 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4394 with a backslash. Remove the backslash after substituting unescaped
4396 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4400 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4401 the driver to pass through the default setting of -mfloat-abi.
4402 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4404 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4405 (all_feat_combs): New rule.
4406 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4408 * config/arm/t-aprofile: Rewrite.
4410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4412 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4413 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4414 fpu is not specified by the user/command-line.
4415 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4416 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4417 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4418 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4419 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4420 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4422 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4424 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4425 * config/arm/t-arm-elf: Rewritten.
4427 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4429 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4430 have some floating-point instructions.
4431 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4432 (TARGET_MAYBE_HARD_FLOAT): New macro.
4433 * config/arm/arm-builtins.c (arm_init_builtins): Use
4434 TARGET_MAYBE_HARD_FLOAT.
4435 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4437 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4439 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4440 (configargs.h): Include it.
4441 (arm_print_hint_for_fpu_option): New function.
4442 (arm_parse_fpu_option): New function.
4443 (candidate_extension): New class.
4444 (arm_canon_for_multilib): New function.
4445 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4446 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4447 (ARCH_CANONICAL_SPECS): New macro.
4448 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4452 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4453 are set after handling multilib fragments. Set target_cpu_default2
4456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4458 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4460 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4461 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4463 (check_arch): Likewise.
4464 * config/arm/arm.c (arm_configure_build_target): Handle
4465 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4466 options in the default.
4468 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4470 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4471 when an option is an alias of another.
4472 * config/arm/parsecpu.awk (optalias): New parser token.
4473 (gen_comm_data): Mark non-alias options as such. Emit entries
4474 for extension aliases.
4475 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4476 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4477 (armv6kz, armv6zk, armv6t2): Likewise.
4478 (armv7): Make vfpv3-d16 an alias.
4479 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4481 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4482 Sort in canonical order.
4483 (armv8-a): Sort in canonical order.
4484 (armv8.1-a, armv8.2-a): Likewise.
4485 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4487 (cortex-a9): Sort in canonical order.
4488 * config/arm/arm.c (selftests.h): Include it.
4489 (arm_test_cpu_arch_data): New function.
4490 (arm_run_self_tests): New function.
4491 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4492 (targetm): Move declaration to the end of the file.
4493 * arm-cpu-cdata.h: Regenerated.
4495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4497 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4498 call to target_mode_check describing the type of option passed.
4499 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4500 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4501 arm_parse_cpu_option_name to match parameters against list of
4503 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4504 arm_arch_core_flags data structure.
4505 * config/arm/arm-cpu_cdata.h: Regenerated.
4507 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4509 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4511 (arm_print_hint_for_cpu_option): Likewise.
4512 (arm_print_hint_for_arch_option): Likewise.
4513 (arm_parse_cpu_option_name): Likewise.
4514 (arm_parse_arch_option_name): Likewise.
4515 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4516 of entries in the all_fpus list.
4517 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4518 (arm_parse_cpu_option_name): Declare.
4519 (arm_parse_arch_option_name): Declare.
4520 (arm_parse_option_features): Declare.
4521 (arm_intialize_isa): Declare.
4522 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4524 (gen_comm_data): ... here. Make definitions non-static.
4525 * config/arm/arm-cpu-data.h: Regenerated.
4526 * config/arm/arm-cpu-cdata.h: Regenerated.
4528 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4530 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
4531 (cpu_arch_extension): New structure.
4532 (cpu_arch_option, arch_option, cpu_option): New structures.
4533 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
4535 (gen_data): Generate new format data tables.
4536 * config/arm/arm.c (cpu_tune): New structure.
4537 (cpu_option, processors): Delete.
4538 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
4539 (arm_print_hint_for_cpu_option): ... this and ...
4540 (arm_print_hint_for_arch_option): ... this.
4541 (arm_parse_arch_cpu_name): Delete. Replace with ...
4542 (arm_parse_cpu_option_name): ... this and ...
4543 (arm_parse_arch_option_name): ... this.
4544 (arm_unrecognized_feature): Change type of target parameter to
4546 (arm_parse_arch_cpu_features): Delete. Replace with ...
4547 (arm_parse_option_features): ... this.
4548 (arm_configure_build_target): Rework to use new configuration data
4550 (arm_print_tune_info): Rework for new configuration data tables.
4551 * config/arm/arm-cpu-data.h: Regenerated.
4552 * config/arm/arm-cpu.h: Regenerated.
4554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4556 * Makefile.in (OBJS): Move sbitmap.o from here ...
4557 (OBJS-libcommon): ... to here.
4559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4561 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
4562 (ISA_ALL_CRYPTO): New macro.
4563 (ISA_ALL_SIMD): New macro
4564 (ISA_ALL_FP): New macro.
4565 * config/arm/arm.c (fpu_bitlist): Update initializer.
4566 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
4568 (arm9e): Add fpu. Add option for nofp
4569 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
4570 (arm926ej-s, arm1026ej-s): Likewise.
4571 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
4572 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
4573 neon-fp16, neon-vfpv4, nofp and nosimd.
4574 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
4575 (cortex-a8): Add fpu. Add option for nofp.
4576 (cortex-a9): Add fpu. Add options for nosimd and nofp.
4577 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
4578 (cortex-r4f): Add fpu.
4579 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
4580 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
4582 (cortex-r8): Likewise.
4583 (cortex-m4): Add fpu. Add option for nofp.
4584 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
4585 (cortex-a17.cortex-a7): Likewise.
4586 (cortex-a32): Add fpu. Add options for crypto and nofp.
4587 (cortex-a35, cortex-a53): Likewise.
4588 (cortex-a57): Add fpu. Add option for crypto.
4589 (cortex-a72, cortex-a73): Likewise.
4590 (exynos-m1): Likewise.
4591 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
4592 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
4593 (cortex-m33): Add fpu. Add option for nofp.
4594 * config/arm/arm-cpu-cdata.h: Regenerated
4595 * config/arm/arm-cpu-data.h: Regenerated.
4597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4599 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
4600 (armv5te, armv5tej): Likewise.
4601 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
4602 (armv7): Add options fp and vfpv3-d16.
4603 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
4604 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
4606 (armv7ve): Likewise.
4607 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
4608 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
4609 (armv8-a): Add nocrypto option.
4610 (armv8.1-a, armv8.2-a): Likewise.
4611 (armv8-m.main): add options fp, fp.dp and nofp.
4613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4615 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
4617 (armv8-a+crc): Delete.
4618 (armv8.1-a): Add options simd, crypto and nofp.
4619 (armv8.2-a): Add options fp16, simd, crypto and nofp.
4620 (armv8.2-a+fp16): Delete.
4621 (armv8-m.main): Add option dsp.
4622 (armv8-m.main+dsp): Delete.
4623 (cortex-a8): Add fpu. Add nofp option.
4624 (cortex-a9): Add fpu. Add nofp and nosimd options.
4625 * config/arm/parsecpu.awk (gen_data): Generate option tables and
4626 link to main cpu and architecture data structures.
4627 (gen_comm_data): Only put isa attributes from the main architecture
4629 (option): New statement for architecture and CPU entries.
4630 * arm.c (struct cpu_option): New structure.
4631 (struct processors): Add entry for options.
4632 (arm_unrecognized_feature): New function.
4633 (arm_parse_arch_cpu_name): Ignore any characters after the first
4635 (arm_parse_arch_cpu_feature): New function.
4636 (arm_configure_build_target): Separate out any CPU and architecture
4637 features and parse separately. Don't error out if -mfpu=auto is
4638 used with only an architecture string.
4639 (arm_print_asm_arch_directives): New function.
4640 (arm_file_start): Call it.
4641 * config/arm/arm-cpu-cdata.h: Regenerated.
4642 * config/arm/arm-cpu-data.h: Likewise.
4643 * config/arm/arm-tables.opt: Likewise.
4645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4647 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
4648 assembler when it is not -mfpu=auto.
4650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4652 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
4653 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
4654 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
4655 (ASM_CPU_SPEC): Rewrite.
4656 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
4657 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
4658 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
4659 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
4660 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
4661 copied string is NUL-terminated. Also strip any characters prefixed
4663 (arm_rewrite_selected_arch): New function.
4664 (arm_rewrite_march): New function.
4666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4668 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
4669 (x_arm_cpu_string, x_arm_tune_string): Likewise.
4670 (march, mcpu, mtune): Convert to string-based options.
4671 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
4672 (arm_parse_arch_cpu_name): New function.
4673 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
4674 identify selected architecture or CPU.
4675 (arm_option_save): New function.
4676 (TARGET_OPTION_SAVE): Redefine.
4677 (arm_option_restore): Restore string options.
4678 (arm_option_print): Print string options.
4680 2017-06-16 Martin Sebor <msebor@redhat.com>
4682 PR tree-optimization/80933
4683 PR tree-optimization/80934
4684 * builtins.c (fold_builtin_3): Do not handle bcmp here.
4685 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
4686 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
4687 (gimple_fold_builtin): Call them.
4689 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4691 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
4692 as unlikely; update profile.
4694 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4696 * predict.c (force_edge_cold): Handle declaring edges impossible
4699 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4701 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
4703 (try_unroll_loop_completely): Fix reporting.
4705 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4707 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
4709 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
4712 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
4713 if given a non-constant argument for an intrinsic which requires a
4716 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4718 * profile.c (compare_freqs): New function.
4719 (branch_prob): Sort edge list.
4720 (find_spanning_tree): Assume that the list is priority sorted.
4722 2017-06-16 Richard Biener <rguenther@suse.de>
4724 PR tree-optimization/81090
4725 * passes.def (pass_record_bounds): Remove.
4726 * tree-pass.h (make_pass_record_bounds): Likewise.
4727 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
4728 make_pass_record_bounds): Likewise.
4729 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
4730 not free niter estimates at the beginning but at the end.
4731 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
4733 2017-06-16 Richard Biener <rguenther@suse.de>
4735 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
4736 initializer to workaround ICE in host GCC 4.8.
4738 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
4740 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
4742 (clone_inlined_nodes): Update.
4744 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4746 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
4747 prefetch settings, and enable prefetching by default at -O3.
4749 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4751 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4752 Set flag_prefetch_loop_arrays according to tuning data.
4754 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4756 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
4758 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
4759 [Unrelated to main purpose of the patch] Place the pointer field last
4760 to enable type checking errors when tune structure are wrongly merged.
4761 * config/aarch64/aarch64.c (generic_prefetch_tune,)
4762 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
4763 (thunderx2t99_prefetch_tune): New tune constants.
4764 (tune_params *_tunings): Update all tunings (no functional change).
4765 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
4766 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
4767 from tunings structures.
4769 2017-06-16 Jakub Jelinek <jakub@redhat.com>
4772 * ubsan.c (instrument_null): Add T argument, use it instead
4773 of computing it based on IS_LHS.
4774 (instrument_object_size): Likewise.
4775 (pass_ubsan::execute): Adjust instrument_null and
4776 instrument_object_size callers to pass gimple_get_lhs or
4777 gimple_assign_rhs1 result to it. Use instrument_null instead of
4778 calling get_base_address and instrument_mem_ref. Handle
4779 aggregate call arguments for object-size sanitization.
4781 2017-06-16 Yury Gribov <tetra2005@gmail.com>
4783 PR tree-optimization/81089
4784 * tree-vrp.c (is_masked_range_test): Validate operands of
4787 2017-06-15 Martin Sebor <msebor@redhat.com>
4790 * dumpfile.c (dump_register): Avoid calling memset to initialize
4791 a class with a default ctor.
4792 * gcc.c (struct compiler): Remove const qualification.
4793 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
4794 * hash-table.h: Ditto.
4795 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
4797 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
4798 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
4800 * params.h (struct param_info): Make struct members non-const.
4801 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
4802 with default initialization.
4803 * vec.h (vec_copy_construct, vec_default_construct): New helper
4805 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
4806 with vec_copy_construct.
4807 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
4808 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
4809 * doc/invoke.texi (-Wclass-memaccess): Document.
4811 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4813 * emit-rtl.h (is_leaf): Update comment about local
4816 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4819 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
4820 for a variable to have a section before checking if the section has a
4822 Set section to.persistent if persistent attribute is set.
4823 Warn if .persistent attribute is used on an automatic variable.
4825 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
4827 PR rtl-optimization/80474
4828 * reorg.c (update_block): Do not ignore instructions in a delay slot.
4830 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
4832 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
4835 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
4837 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
4838 (casesi): Emit bounds checking as RTL.
4839 (casesi_internal_mips16_<mode>): Remove bounds checking.
4841 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
4843 * config/xtensa/xtensa.c (xtensa_option_override): Append
4844 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
4845 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
4846 xtensa_doloop_hooks): Define unconditionally.
4847 (xtensa_reorg_loops): Only call reorg_loops in the presence of
4849 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
4850 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
4851 for it in xtensa_option_override.
4852 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
4853 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
4855 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
4857 * doc/cppopts.texi: Document '-' special value to -MF.
4859 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
4861 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
4862 (cortex_a53_fconst): Likewise.
4863 (cortex_a53_fpmul): Likewise.
4864 (cortex_a53_f_load_64): Likewise.
4865 (cortex_a53_f_load_many): Likewise.
4866 (cortex_a53_advsimd_alu): Likewise.
4867 (cortex_a53_advsimd_alu_q): Likewise.
4868 (cortex_a53_advsimd_mul): Likewise.
4869 (cortex_a53_advsimd_mul_q): Likewise.
4870 (fpmac bypass): Add new bypass for fpmac-fpmac case.
4871 Add missing fmul, r2f_cvt and fconst cases.
4873 2017-06-14 Richard Biener <rguenther@suse.de>
4876 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
4878 (fold_binary_loc): When associating do not treat pre-existing
4879 TREE_OVERFLOW on literal constants as a reason to allow
4880 TREE_OVERFLOW on associated literal constants.
4882 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4884 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
4885 (MASK_FEATURES): New macro.
4886 * config/sparc/sparc.c (sparc_option_override): Remove the special
4887 handling of -mfpu and generalize it to all MASK_FEATURES switches.
4889 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4891 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
4892 a division of 0 if non-call exceptions are enabled.
4894 2017-06-14 Andrew Pinski <apinski@cavium.com>
4895 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4898 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
4899 Improve vector initialization code gen for only variable case.
4901 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
4903 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
4905 2017-06-14 Richard Biener <rguenther@suse.de>
4907 PR tree-optimization/81083
4908 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
4911 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4913 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
4914 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
4915 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
4916 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
4917 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
4918 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
4920 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4922 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
4923 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
4925 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4927 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
4929 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4931 * config/rs6000/t-rtems: Don't handle SPE.
4933 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4935 * config/rs6000/t-linux: Don't handle SPE.
4937 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4939 * config/rs6000/eabispe.h: Delete file.
4941 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4943 * config/rs6000/t-spe: Delete file.
4945 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4947 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
4948 (rs6000_legitimate_offset_address_p): Return false for anything in
4949 V2SImode or V2SFmode.
4951 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
4953 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
4954 except V2SF and V2SI. Rearrange the vector modes, and add comments.
4955 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
4957 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
4958 (rs6000_legitimate_offset_address_p): Ditto.
4959 (rs6000_emit_move): Ditto.
4960 (rs6000_init_builtins): Remove V4HI_type_node.
4962 2017-06-13 Martin Liska <mliska@suse.cz>
4965 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
4966 (gate_asan): Likewise.
4967 * asan.h (asan_no_sanitize_address_p): Remove the function.
4968 (sanitize_flags_p): New function.
4969 * builtins.def: Fix coding style.
4970 * common.opt: Use renamed enum value.
4971 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
4972 * doc/extend.texi: Document no_sanitize attribute.
4973 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
4974 to SANITIZE_UNDEFINED_NONDEFAULT.
4975 * gcc.c (sanitize_spec_function): Use the renamed enum value.
4976 * gimple-fold.c (optimize_atomic_compare_exchange_p):
4977 Use sanitize_flags_p.
4978 * gimplify.c (gimplify_function_tree): Likewise.
4979 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
4980 * opts.c (parse_no_sanitize_attribute): New function.
4981 (common_handle_option): Use renamed enum value.
4982 * opts.h (parse_no_sanitize_attribute): Declare.
4983 * tree.c (sanitize_flags_p): New function.
4984 * tree.h: Declared here.
4985 * tsan.c: Use sanitize_flags_p.
4986 * ubsan.c (ubsan_expand_null_ifn): Likewise.
4987 (instrument_mem_ref): Likewise.
4988 (instrument_bool_enum_load): Likewise.
4989 (do_ubsan_in_current_function): Remove the function.
4990 (pass_ubsan::execute): Use sanitize_flags_p.
4991 * ubsan.h: Remove do_ubsan_in_current_function
4992 * tree-cfg.c (print_no_sanitize_attr_value): New function.
4993 (dump_function_to_file): Use it here.
4995 2017-06-13 Martin Jambor <mjambor@suse.cz>
4997 PR tree-optimization/80803
4998 PR tree-optimization/81063
4999 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5000 (propagate_subaccesses_across_link): Enqueue subtree whenever
5001 necessary instead of relying on the caller.
5003 2017-06-13 Martin Jambor <mjambor@suse.cz>
5005 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5006 that have a first_link.
5007 (sort_and_splice_var_accesses): Do not check first_link before
5009 (subtree_mark_written_and_enqueue): Likewise.
5010 (propagate_all_subaccesses): Likewise and do not stop at first
5011 parent with a first_link.
5013 2017-06-13 Martin Jambor <mjambor@suse.cz>
5015 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5018 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5020 * match.pd: New pattern.
5022 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5024 * tree-vrp.c (is_masked_range_test): New function.
5025 (register_edge_assert_for): Determine ranges for
5028 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5030 PR tree-optimization/67328
5031 * fold-const.c (maskable_range_p): New function.
5032 (build_range_check): Generate bittests if possible.
5034 2017-06-13 Martin Liska <mliska@suse.cz>
5036 * gimple-pretty-print.c (dump_probability): Add new argument.
5037 (dump_edge_probability): Dump both probability and count.
5038 (dump_gimple_label): Likewise.
5039 (dump_gimple_bb_header): Likewise.
5041 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5044 * config/avr/avr-devices.c: Fix indentation.
5045 * config/avr/gen-avr-mmcu-specs.c: Dito.
5047 2017-06-13 Richard Biener <rguenther@suse.de>
5049 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5050 instead get vector type from stmt_info.
5051 (vectorizable_reduction): Adjust. Remove dead code.
5053 2017-06-13 Richard Biener <rguenther@suse.de>
5056 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5057 case of C * (x * C2 + C3).
5058 (fold_addr_of_array_ref_difference): Properly fold index difference.
5060 2017-06-12 David S. Miller <davem@davemloft.net>
5063 * config/sparc/sparc.md (return expander): Emit frame blockage if
5064 function uses alloca.
5066 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5068 * combine.c (make_field_assignment): Check len rather than the mode
5069 precision when calling force_to_mode.
5071 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5073 Support multilibs and devices that see flash in RAM address range.
5076 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5077 (avr_mcu_t) <flash_pm_offset>: New field.
5078 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5079 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5080 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5081 (AVR_TINY_PM_OFFSET): Remove macro.
5082 * config/avr/avr.opt (-mshort-calls): New option.
5083 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5084 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5085 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5086 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5087 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5088 instead of avr_arch->have_jmp_call.
5089 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5090 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5091 avr_arch->flash_pm_offset to define.
5092 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5093 new field flash_pm_offset. Add entry for avrxmega3.
5094 (avr_texinfo): Add entry for avrxmega3.
5095 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5096 attiny212, attiny214,
5097 attiny412, attiny414, attiny416, attiny417,
5098 attiny814, attiny816, attiny817,
5099 attiny1614, attiny1616, attiny1617,
5100 attiny3214, attiny3216, attiny3217.
5101 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5102 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5103 (avr_print_operand_address) [AVR_TINY]: Same.
5104 (avr_asm_init_sections) <readonly_data_section>: Only patch
5105 callback if avr_arch->flash_pm_offset = 0.
5106 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5107 for rodata if avr_arch->flash_pm_offset != 0.
5108 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5109 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5110 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5111 (m_options): Append opt_rcall.
5112 (m_dirnames): Append dir_rcall.
5113 * config/avr/t-multilib: Regenerate.
5115 * configure.ac [target=avr]: Check whether avrxmega3 default
5116 linker description file works as needed.
5117 * configure: Regenerate.
5118 * doc/avr-mmcu.texi: Regenerate.
5119 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5120 <__AVR_ARCH__>: Document avrxmega3 and 103.
5121 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5122 <__AVR_SHORT_CALLS__>: Document it.
5123 <__AVR_PM_BASE_ADDRESS__>: Document it.
5124 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5125 (AVR Variable Attributes) <progmem>: Document this is
5126 not needed for avrxmega3.
5127 (AVR Named Address Spaces) <__flash>: Dito.
5129 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5131 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5133 2017-06-12 Doug Rupp <rupp@adacore.com>
5135 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5136 Append vxworks-stdint.h to the tm_file list.
5137 * config/vxworks-stdint.h: New file.
5139 2017-06-12 Martin Liska <mliska@suse.cz>
5141 PR tree-optimization/81041
5142 * tree-profile.c (gimple_gen_ic_func_profiler):
5143 Create an extra BB in profile-generate
5144 (gimple_gen_time_profiler): Likewise.
5146 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5148 PR tree-optimization/81003
5149 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5150 (update_range_test): Use it instead of force_gimple_operand_gsi.
5152 2017-06-12 Richard Biener <rguenther@suse.de>
5154 PR tree-optimization/81053
5155 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5156 with backedge value not defined in loop. Simplify def stmt
5159 2017-06-11 Tom de Vries <tom@codesourcery.com>
5162 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5164 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5165 nvptx_cannot_force_const_mem.
5167 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5169 * opts.c (finish_options): Move test for flag_split_stack after
5170 it has been initialized.
5172 2017-06-11 Jason Merrill <jason@redhat.com>
5174 * tree.h (id_equal): New.
5175 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5176 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5177 instead of strcmp of IDENTIFIER_POINTER.
5179 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5181 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5182 (mark_all_inlined_calls_cdtor): Fix formating.
5183 (inline_transform): Rescale profile before inlining.
5185 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5187 * cgraph.h (cgraph_edge::clone): Update prototype.
5188 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5189 (cgraph_node::create_clone): Update.
5190 (cgraph_node::create_version_clone): Update.
5191 * tree-inline.c (copy_bb): Update.
5192 (expand_call_inline): Update.
5194 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5196 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5197 factored out from ...
5198 (rs6000_emit_prologue): ... here.
5200 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5202 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5203 factored out from ...
5204 (rs6000_emit_prologue): ... here.
5206 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5208 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5210 (handle_missing_profiles): Fix computation of tp_first_run.
5211 (counts_to_freqs): Do not touch freqs when count is 0.
5213 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5215 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5218 2017-06-10 Tom de Vries <tom@codesourcery.com>
5220 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5221 attributes): Document signal effective target.
5223 2017-06-10 Tom de Vries <tom@codesourcery.com>
5225 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5226 Document effective target stack_size.
5228 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5230 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5231 to the edit_context if they can be auto-applied.
5233 2017-06-9 Ian Lance Taylor <iant@golang.org>
5235 * opts.c (finish_options): If -fsplit-stack, disable implicit
5236 -forder-blocks-and-partition.
5237 * doc/invoke.texi (Optimize Options): Document that when using
5238 -fsplit-stack -forder-blocks-and-partition is not implicitly
5241 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5243 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5244 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5245 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5246 * builtins.def (abort, trap, unreachable): Declare cold.
5247 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5248 * tree-core.h (ECF_COLD): New.
5249 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5250 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5252 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5254 * predict.c (unlikely_executed_stmt_p): Cleanup.
5256 2017-06-09 Richard Biener <rguenther@suse.de>
5258 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5259 model if the ref is always written to.
5261 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5263 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5265 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5267 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5270 2017-06-09 Tom de Vries <tom@codesourcery.com>
5273 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5274 "target cannot support label values" when encountering LABEL_REF.
5276 2017-06-09 Martin Liska <mliska@suse.cz>
5278 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5279 (gimple_gen_ic_func_profiler): Emit direct comparison
5280 of __gcov_indirect_call_callee with NULL.
5281 (gimple_gen_time_profiler): Change probability from
5282 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5284 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5286 * profile.c (edge_gcov_counts): Turn to pointer.
5287 (compute_branch_probabilities, compute_branch_probabilities): Update.
5288 (branch_prob): Do not clear edge_gcov_count.
5289 * profile.h (edge_gcov_counts): Turn to pointer.
5290 (edge_gcov_count): Update.
5292 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5294 * gimple.h (gimple_check_failed): Mark cold.
5296 2017-06-09 Richard Biener <rguenther@suse.de>
5298 PR tree-optimization/66623
5299 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5300 refactor check_reduction into two parts, properly computing
5301 whether we have to check reduction validity for outer loop
5304 2017-06-09 Richard Biener <rguenther@suse.de>
5306 PR tree-optimization/79483
5307 * graphite-scop-detection.c (order): New global.
5308 (get_order): Compute bb to order mapping that satisfies code
5309 generation constraints.
5310 (cmp_pbbs): New helper.
5311 (build_scops): Start domwalk at entry block, sort generated
5314 2017-06-09 Richard Biener <rguenther@suse.de>
5317 * ipa-polymorphic-call.c
5318 (ipa_polymorphic_call_context::restrict_to_inner_class):
5319 Skip FIELD_DECLs with error_mark_node type.
5320 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5323 2017-06-09 Martin Liska <mliska@suse.cz>
5325 * predict.c (struct branch_predictor): New struct.
5326 (test_prediction_value_range): New test.
5327 (predict_c_tests): New function.
5328 * selftest-run-tests.c (selftest::run_tests): Run the function.
5329 * selftest.h: Declare new tests.
5331 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5334 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5335 gen_add3_insn did not fail.
5336 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5337 r0, construct that number in a temporary reg and add that reg to r0.
5338 If asked to put the result in r0 as well, fail.
5340 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5342 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5343 for early expansion of vec_eqv.
5345 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5348 * ubsan.c (instrument_null): Avoid pointless code temporary.
5349 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5351 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5355 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5358 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5361 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5363 * predict.c (unlikely_executed_edge_p): Move ahead.
5364 (probably_never_executed_edge_p): Use it.
5366 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5369 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5370 gimple_call_builtin_p call.
5372 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5374 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5375 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5376 rtl_check_failed_type2, rtl_check_failed_code1,
5377 rtl_check_failed_code2, rtl_check_failed_code_mode,
5378 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5379 rtvec_check_failed_bounds, rtl_check_failed_flag,
5380 _fatal_insn_not_found, _fatal_insn): Likewise.
5381 * tree.h (tree_contains_struct_check_failed,
5382 tree_check_failed, tree_not_check_failed,
5383 tree_class_check_failed, tree_range_check_failed,
5384 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5385 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5386 tree_operand_check_failed, omp_clause_check_failed,
5387 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5390 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5392 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5393 flag_branch_probabilities.
5394 * ipa-inline.c (edge_badness): Likewise.
5395 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5396 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5397 * predict.c (maybe_hot_frequency_p): Likewise.
5398 (probably_never_executed): Likewise.
5399 * sched-ebb.c (schedule_ebbs): Likewise.
5400 * sched-rgn.c (find_single_block_region): Likewise.
5401 * tracer.c (tail_duplicate): Likewise.
5403 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5405 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5406 longer requires x_flag_profile_use.
5408 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5410 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5411 instead of flag_reorder_blocks_and_partition.
5412 * dbxout.c (dbxout_function_end): Likewise.
5413 * dwarf2out.c (gen_subprogram_die): Likewise.
5414 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5415 * hw-doloop.c (reorg_loops): Likewise.
5416 * varasm.c (assemble_start_function,
5417 assemble_end_function): Likewise.
5418 (decide_function_section): Do not check for
5419 flag_reorder_blocks_and_partition.
5421 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5423 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5425 (chkp_get_hard_register_fake_addr_expr): Ditto.
5426 (chkp_build_addr_expr): Add check for hard reg case.
5427 (chkp_parse_array_and_component_ref): Ditto.
5428 (chkp_find_bounds_1): Ditto.
5429 (chkp_process_stmt): Don't generate bounds store for
5432 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5434 * predict.c (maybe_hot_bb_p): Do not check profile status.
5435 (maybe_hot_edge_p): Likewise.
5436 (probably_never_executed): Check for zero counts even if profile
5438 (unlikely_executed_edge_p): New function.
5439 (unlikely_executed_stmt_p): New function.
5440 (unlikely_executed_bb_p): New function.
5441 (set_even_probabilities): Use unlikely predicates.
5442 (combine_predictions_for_bb): Likewise.
5443 (predict_paths_for_bb): Likewise.
5444 (predict_paths_leading_to_edge): Likewise.
5445 (determine_unlikely_bbs): New function.
5446 (estimate_bb_frequencies): Use it.
5447 (compute_function_frequency): Use zero counts even if profile is
5449 * profile-count.h: Fix typo.
5451 2017-08-08 Julia Koval <julia.koval@intel.com>
5453 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5454 _mm512_mask_cvtsepi16_storeu_epi8,
5455 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5456 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5457 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5458 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5459 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5460 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5461 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5462 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5463 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5464 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5465 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5466 __builtin_ia32_pmovuswb256mem_mask,
5467 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5468 __builtin_ia32_pmovwb512mem_mask): New builtins.
5470 2017-08-08 Julia Koval <julia.koval@intel.com>
5472 PR target/73350,80862
5473 * config/i386/subst.md (round): Fix round pattern.
5474 * config/i386/i386.c (ix86_erase_embedded_rounding):
5475 Fix erasing rounding for the fixed pattern.
5477 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5479 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5481 2017-06-08 Martin Liska <mliska@suse.cz>
5483 PR gcov-profile/80911
5484 * gcov.c (block_info::block_info): New constructor.
5486 2017-06-07 Carl Love <cel@us.ibm.com>
5488 * config/rs6000/rs6000-c: The return type of the following
5489 built-in functions was implemented as int not long long. Fix sign
5490 of return value for the unsigned version of vec_mulo and vec_mule.
5491 vector unsigned long long vec_bperm (vector unsigned long long,
5492 vector unsigned char)
5493 vector signed long long vec_mule (vector signed int,
5495 vector unsigned long long vec_mule (vector unsigned int,
5496 vector unsigned int)
5497 vector signed long long vec_mulo (vector signed int,
5499 vector unsigned long long vec_mulo (vector unsigned int,
5500 vector unsigned int)
5501 * doc/extend.texi: Fix the documentation for the built-in
5504 2017-06-07 Carl Love <cel@us.ibm.com>
5507 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5510 2017-06-07 Carl Love <cel@us.ibm.com>
5512 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5513 support, Generate doublehv for signed int/float for BE case only.
5515 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5517 * doc/invoke.texi (mcx16): Rewrite.
5519 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5521 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5522 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5523 *mov<mode>_softfloat, and an anonymous splitter): Use
5524 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5526 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5528 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
5530 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5531 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
5532 (rs6000_debug_reg_global): Adjust.
5533 (rs6000_init_hard_regno_mode_ok): Adjust.
5534 (rs6000_dbx_register_number): Adjust.
5535 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
5536 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5537 Remove SPE_ACC and SPEFSCR.
5538 (REG_ALLOC_ORDER): Ditto.
5539 (FRAME_POINTER_REGNUM): Change to 111.
5540 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
5541 (REG_CLASS_NAMES): Ditto.
5542 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
5543 (REGISTER_NAMES): Ditto.
5544 (ADDITIONAL_REG_NAMES): Ditto.
5545 (rs6000_reg_names): Ditto.
5546 * config/rs6000/rs6000.md: Renumber some register number
5549 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5551 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
5553 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
5554 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
5556 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
5557 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
5558 Delete the SPE high registers.
5559 (REG_ALLOC_ORDER): Ditto.
5560 (enum reg_class): Remove SPE_HIGH_REGS.
5561 (REG_CLASS_NAMES): Ditto.
5562 (REG_CLASS_CONTENTS): Delete the SPE high registers.
5563 (REGISTER_NAMES): Ditto.
5564 (rs6000_reg_names): Ditto.
5565 * doc/tm.texi.in: Remove SPE as example.
5566 * doc/tm.texi: Regenerate.
5568 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5570 * config/rs6000/8540.md (ppc8540_brinc): Delete.
5571 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
5572 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
5573 * config/rs6000/rs6000.md (type): Remove "brinc".
5575 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5577 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
5578 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
5579 * config/rs6000/linuxspe.h: Delete file.
5580 * config/rs6000/rs6000.md: Don't include spe.md.
5581 * config/rs6000/spe.h: Delete file.
5582 * config/rs6000/spe.md: Delete file.
5583 * config/rs6000/t-rs6000: Remove spe.md.
5585 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5587 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
5588 (reg_or_none500mem_operand): Delete.
5589 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
5590 instead of reg_or_none500mem_operand.
5592 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5595 handling of SPE flags.
5596 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
5598 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5600 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
5602 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
5603 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
5604 paired_divv2sf3): Similar.
5605 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
5606 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
5607 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
5609 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
5610 Rename the paired_* instruction patterns.
5611 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
5613 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
5614 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
5615 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
5616 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
5617 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
5619 (struct machine_function): Delete field spe_insn_chain_scanned_p.
5620 (spe_func_has_64bit_regs_p): Delete.
5621 (spe_expand_predicate_builtin): Delete.
5622 (spe_expand_evsel_builtin): Delete.
5623 (TARGET_DWARF_REGISTER_SPAN): Do not define.
5624 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
5625 (invalid_e500_subreg): Delete.
5626 (rs6000_legitimize_address): Always force_reg op2 as well, for
5627 paired single memory accesses.
5628 (rs6000_member_type_forces_blk): Delete.
5629 (rs6000_spe_function_arg): Delete.
5630 (rs6000_expand_unop_builtin): Delete SPE handling.
5631 (rs6000_expand_binop_builtin): Ditto.
5632 (spe_expand_stv_builtin): Delete.
5633 (bdesc_2arg_spe): Delete.
5634 (spe_expand_builtin): Delete.
5635 (spe_expand_predicate_builtin): Delete.
5636 (spe_expand_evsel_builtin): Delete.
5637 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
5638 (spe_init_builtins): Delete.
5639 (spe_func_has_64bit_regs_p): Delete.
5640 (savres_routine_name): Delete "info" parameter. Adjust callers.
5641 (rs6000_emit_stack_reset): Ditto.
5642 (rs6000_dwarf_register_span): Delete.
5643 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
5644 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
5645 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
5647 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
5649 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
5650 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
5651 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
5652 mulv2sf3, divv2sf3): Delete expanders.
5654 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5656 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
5658 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5660 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
5661 * config/rs6000/rs6000.c: Ditto.
5663 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5665 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
5666 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
5667 comparison_operator.
5669 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5671 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
5672 * config/rs6000/rs6000.opt: Ditto.
5673 * config/rs6000/t-rtems: Ditto.
5675 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5677 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
5678 TARGET_E500_SINGLE by 0, simplify.
5679 * config/rs6000/rs6000.c: Ditto.
5680 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
5681 (spe_build_register_parallel): Delete.
5682 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
5683 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
5684 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
5685 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
5686 (E500_CONVERT): Delete.
5687 * config/rs6000/spe.md: Remove many patterns and all define_constants.
5689 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5691 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
5692 * config/rs6000/dfp.md: Ditto.
5693 (negdd2, *negdd2_fpr): Merge.
5694 (absdd2, *absdd2_fpr): Merge.
5695 (negtd2, *negtd2_fpr): Merge.
5696 (abstd2, *abstd2_fpr): Merge.
5697 * config/rs6000/e500.h: Delete file.
5698 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
5699 TARGET_FPRS by 1 and simplify.
5700 * config/rs6000/rs6000-c.c: Ditto.
5701 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
5703 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
5705 * config/rs6000/rs6000.md: Ditto.
5706 (floatdidf2, *floatdidf2_fpr): Merge.
5707 (move_from_CR_gt_bit): Delete.
5708 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
5709 (E500_CR_IOR_COMPARE): Delete.
5710 (All patterns that require !TARGET_FPRS): Delete.
5711 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
5713 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5715 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
5717 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5719 * graphds.c (add_edge): Intitialize edge's attached data.
5720 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
5721 pointer parameter. Call pointed function on each edge during
5722 graph traversing. Skip traversing the edge when the function
5724 (graphds_dfs, graphds_scc): Ditto.
5725 (for_each_edge): New parameter. Pass the new parameter to callback
5727 * graphds.h (skip_edge_callback): New function pointer type.
5728 (graphds_dfs, graphds_scc): New function pointer parameter.
5729 (graphds_edge_callback, for_each_edge): New parameter.
5731 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5733 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
5734 out code checking if runtime alias check is possible to below ...
5735 Call the new function.
5736 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
5737 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
5739 2017-06-07 Marek Polacek <polacek@redhat.com>
5742 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
5743 TYPE_OVERFLOW_WRAPS check.
5745 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5747 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
5748 if versioning is required.
5749 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
5750 peeling with the check for versioning.
5752 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5754 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
5755 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
5756 Set true to new parameter if new ssa variable is defined.
5757 (vect_gen_vector_loop_niters): Refactor. Set range information
5758 for the new vector loop bound variable.
5759 (vect_do_peeling): Ditto.
5761 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5763 * tree-affine.c (ssa.h): Include header file.
5764 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
5765 has wrapping overflow behavior.
5767 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5769 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
5771 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5773 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
5774 (tree_to_aff_combination): ... here.
5776 2017-06-07 Bin Cheng <bin.cheng@arm.com>
5778 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
5779 reg_pressure model function.
5780 (ivopts_global_cost_for_size): Delete.
5781 (determine_set_costs, iv_ca_recount_cost): Call new model function
5782 ivopts_estimate_reg_pressure.
5784 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5786 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
5787 expensive than udiv. Remove floating point cases from mod.
5789 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5791 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
5794 2017-06-07 Tamar Christina <tamar.christina@arm.com>
5796 * config/aarch64/aarch64.md
5797 (copysignsf3): Fix mask generation.
5799 2017-06-07 Jakub Jelinek <jakub@redhat.com>
5801 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
5803 (class dump_manager): Add register_dumps method.
5804 * dumpfile.c: Include langhooks.h.
5805 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
5806 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
5807 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
5808 (dump_manager::dump_register): Start with 512 entries instead of 32.
5809 (dump_manager::register_dumps): New method.
5810 * toplev.c (general_init): Instead of invoking register_dumps
5811 langhook, invoke register_dumps method on the dump manager.
5812 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
5815 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
5817 * doc/md.texi: Clarify the restrictions on a define_insn condition.
5818 Say that # requires an associated define_split to exist, and that
5819 the define_split must be suitable for use after register allocation.
5821 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5823 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
5824 (compute_outgoing_frequencies): Also initialize zero counts.
5825 (find_many_sub_basic_blocks): Do not produce uninitialized profile
5826 around loops; preserve more of profile when nothing changes.
5828 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
5830 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
5832 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
5833 * config/arm/arm-cpu-cdata.h: Regenerate.
5834 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
5835 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
5836 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
5837 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
5838 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
5840 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
5842 * config/arm/t-rmprofile: Likewise.
5843 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
5845 2017-06-06 David S. Miller <davem@davemloft.net>
5848 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
5849 blockage if function uses alloca.
5851 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5853 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
5854 New "uid" fields to hold pretty-print IDs of group and ref.
5855 Memory references are now identified as <group_id>:<ref_id>
5856 instead of using [random] addresses.
5857 (dump_mem_details): Simplify, no functional change.
5858 (dump_mem_ref): Simplify and make output more concise.
5859 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
5860 (find_or_create_group): Initialize group uid.
5861 (record_ref): Initialize ref uid. Improve debug output.
5862 (prune_group_by_reuse, should_issue_prefetch_p,)
5863 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
5864 (mark_nontemporal_store, determine_loop_nest_reuse):
5865 Improve debug output.
5867 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5869 * dbgcnt.def (prefetch): New debug counter.
5870 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
5871 (schedule_prefetches): Stop issueing prefetches if debug counter
5874 2017-06-06 Tom de Vries <tom@codesourcery.com>
5876 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
5877 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
5879 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5881 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
5882 Use aarch64_reg_or_zero predicate for operand 4.
5883 (aarch64_compare_and_swap<mode> define_insn_and_split):
5884 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
5885 (aarch64_store_exclusive<mode>): Likewise for operand 2.
5887 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5889 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
5890 (arm_compute_save_core_reg_mask): This.
5891 (thumb1_compute_save_reg_mask): Rename into ...
5892 (thumb1_compute_save_core_reg_mask): This.
5893 (arm_compute_save_reg0_reg12_mask): Adapt comment.
5894 (arm_compute_frame_layout): Likewise.
5896 2017-06-06 Richard Biener <rguenther@suse.de>
5898 PR tree-optimization/80974
5899 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
5900 keep or clear leaders SSA info.
5902 2017-06-06 Tom de Vries <tom@codesourcery.com>
5904 * config/nvptx/nvptx.c (split_mode_p): New function.
5905 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
5907 2017-06-06 Tom de Vries <tom@codesourcery.com>
5909 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
5911 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5914 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
5917 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5919 * shrink-wrap.c (handle_simple_exit): Update profile.
5920 (try_shrink_wrapping): Upate profile.
5922 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5924 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
5925 (tree_guess_outgoing_edge_probabilities): New.
5926 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
5927 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
5929 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5931 * ipa-split.c (split_function): Initialize return bb profile.
5933 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
5935 * profile.c (compute_branch_probabilities): Also initialize
5938 2017-06-06 Richard Biener <rguenther@suse.de>
5940 PR tree-optimization/80928
5941 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
5942 (vect_analyze_loop_operations): Properly guard analysis for
5944 (vect_transform_loop): Likewise.
5945 (vect_analyze_loop_2): Also reset SLP type on PHIs.
5946 (vect_model_induction_cost): Do not cost for pure SLP.
5947 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
5948 of induction in inner loop vectorization.
5949 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
5950 (vect_get_and_check_slp_defs): Handle vect_induction_def.
5951 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
5953 (vect_analyze_slp_cost_1): Cost induction.
5954 (vect_detect_hybrid_slp_stmts): Handle PHIs.
5955 (vect_get_slp_vect_defs): Likewise.
5956 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
5957 (vect_transform_stmt): Handle SLP reductions.
5958 * tree-vectorizer.h (vectorizable_induction): Adjust.
5960 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5962 * config/rs6000/rs6000.c (make_resolver_func): Update
5963 init_lowered_empty_function call.
5965 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
5967 * doc/invoke.texi: Document the -fprofile-abs-path option.
5968 * common.opt (fprofile-abs-path): New option.
5969 * gcov-io.h (gcov_write_filename): Declare.
5970 * gcov-io.c (gcov_write_filename): New function.
5971 * coverage.c (coverage_begin_function): Use gcov_write_filename.
5972 * profile.c (output_location): Likewise.
5974 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5976 * shring-wrap.c: Revert accidental commit.
5978 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
5980 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
5982 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
5984 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
5986 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
5987 profile in callgraph edge.
5988 * profile-count.h (apply_probability): If THIS is 0, then result is 0
5989 (apply_scale): Likewise.
5990 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
5991 Also scale profile when inlining function with zero profile.
5992 (initialize_cfun): Update exit block profile even when it is zero.
5993 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
5994 when profile is read.
5996 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
5998 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
5999 (CLONE_*): New constants to define the processors we can generate
6000 code for with the target_clone attribute.
6001 (rs6000_clone_map): New array to identify which clone processors
6002 the current program is running on.
6003 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6004 target_clone attribute.
6005 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6006 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6007 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6008 (cpu_expand_builtin): Add support for target_clone attribute.
6009 (rs6000_valid_attribute_p): Allow "default" attribute.
6010 (get_decl_name): New debug function to simplify printing the
6011 current function name in debugging statements.
6012 (rs6000_clone_priority): New functions to support the target_clone
6013 attribute, and be able to generate code to switch between ISA 2.05
6014 through ISA 3.0 (power6 through power9).
6015 (rs6000_compare_version_priority): Likewise.
6016 (rs6000_get_function_versions_dispatcher): Likewise.
6017 (make_resolver_func): Likewise.
6018 (add_condition_to_bb): Likewise.
6019 (dispatch_function_versions): Likewise.
6020 (rs6000_generate_version_dispatcher_body): Likewise.
6021 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6022 (fusion_gpr_load_p): Fix a spacing issue.
6023 * doc/extend.texi (Common Function Attributes): Document that the
6024 PowerPC supports the target_clone attribute.
6026 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6028 * config/arm/arm.h: explain F symbol found in description of ARM
6029 register allocation in its legend.
6031 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6033 * config/mips/frame-header-opt.c: Include profile-count.h.
6034 * config/riscv/riscv.c: Include profile-count.h
6036 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6038 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6040 (sm_set_flag_if_changed): Add bbs field.
6041 (execute_sm_if_changed_flag_set): Pass BBS.
6042 (execute_sm): Update.
6044 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6046 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6049 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6051 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6052 (peephole2): New peephole2 to emit the above.
6053 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6055 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6057 * config/aarch64/aarch64.c (define_peephole2 above
6058 *sub_<shift>_<mode>): New peephole.
6060 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6062 * config/i386/i386.c (make_resolver_func): Update.
6063 * Makefile.in: Add profile-count.h and profile-count.o
6064 * auto-profile.c (afdo_indirect_call): Update to new API.
6065 (afdo_set_bb_count): Update.
6066 (afdo_propagate_edge): Update.
6067 (afdo_propagate_circuit): Update.
6068 (afdo_calculate_branch_prob): Update.
6069 (afdo_annotate_cfg): Update.
6070 * basic-block.h: Include profile-count.h
6071 (struct edge_def): Turn count to profile_count.
6072 (struct basic_block_def): Likewie.
6073 (REG_BR_PROB_BASE): Move to profile-count.h
6074 (RDIV): Move to profile-count.h
6075 * bb-reorder.c (max_entry_count): Turn to profile_count.
6076 (find_traces): Update.
6077 (rotate_loop):Update.
6078 (connect_traces):Update.
6079 (sanitize_hot_paths):Update.
6080 * bt-load.c (migrate_btr_defs): Update.
6081 * cfg.c (RDIV): Remove.
6082 (init_flow): Use alloc_block.
6083 (alloc_block): Uninitialize count.
6084 (unchecked_make_edge): Uninitialize count.
6085 (check_bb_profile): Update.
6086 (dump_edge_info): Update.
6087 (dump_bb_info): Update.
6088 (update_bb_profile_for_threading): Update.
6089 (scale_bbs_frequencies_int): Update.
6090 (scale_bbs_frequencies_gcov_type): Update.
6091 (scale_bbs_frequencies_profile_count): New.
6092 * cfg.h (update_bb_profile_for_threading): Update.
6093 (scale_bbs_frequencies_profile_count): Declare.
6094 * cfgbuild.c (compute_outgoing_frequencies): Update.
6095 (find_many_sub_basic_blocks): Update.
6096 * cfgcleanup.c (try_forward_edges): Update.
6097 (try_crossjump_to_edge): Update.
6098 * cfgexpand.c (expand_gimple_tailcall): Update.
6099 (construct_exit_block): Update.
6100 * cfghooks.c (verify_flow_info): Update.
6101 (dump_bb_for_graph): Update.
6102 (split_edge): Update.
6103 (make_forwarder_block): Update.
6104 (duplicate_block): Update.
6105 (account_profile_record): Update.
6106 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6107 (get_estimated_loop_iterations): Update.
6108 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6109 (single_likely_exit): Update.
6110 * cfgloopmanip.c (scale_loop_profile): Update.
6112 (set_zero_probability): Update.
6113 (lv_adjust_loop_entry_edge): Update.
6114 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6115 (purge_dead_edges): Update.
6116 (rtl_account_profile_record): Update.
6117 * cgraph.c (cgraph_node::create): Uninitialize count.
6118 (symbol_table::create_edge): Uninitialize count.
6119 (cgraph_update_edges_for_call_stmt_node): Update.
6120 (cgraph_edge::dump_edge_flags): Update.
6121 (cgraph_node::dump): Update.
6122 (cgraph_edge::maybe_hot_p): Update.
6123 * cgraph.h: Include profile-count.h
6124 (create_clone), create_edge, create_indirect_edge): Update.
6125 (cgraph_node): Turn count to profile_count.
6126 (cgraph_edge0: Likewise.
6127 (make_speculative, clone): Update.
6128 (create_edge): Update.
6129 (init_lowered_empty_function): Update.
6130 * cgraphclones.c (cgraph_edge::clone): Update.
6131 (duplicate_thunk_for_node): Update.
6132 (cgraph_node::create_clone): Update.
6133 * cgraphunit.c (cgraph_node::analyze): Update.
6134 (cgraph_node::expand_thunk): Update.
6135 * final.c (dump_basic_block_info): Update.
6136 * gimple-streamer-in.c (input_bb): Update.
6137 * gimple-streamer-out.c (output_bb): Update.
6138 * graphite.c (print_global_statistics): Update.
6139 (print_graphite_scop_statistics): Update.
6140 * hsa-brig.c: Include basic-block.h.
6141 * hsa-dump.c: Include basic-block.h.
6142 * hsa-gen.c (T sum_slice): Update.
6143 (convert_switch_statements):Update.
6144 * hsa-regalloc.c: Include basic-block.h.
6145 * ipa-chkp.c (chkp_produce_thunks): Update.
6146 * ipa-cp.c (struct caller_statistics): Update.
6147 (init_caller_stats): Update.
6148 (gather_caller_stats): Update.
6149 (ipcp_cloning_candidate_p): Update.
6150 (good_cloning_opportunity_p): Update.
6151 (get_info_about_necessary_edges): Update.
6152 (dump_profile_updates): Update.
6153 (update_profiling_info): Update.
6154 (update_specialized_profile): Update.
6155 (perhaps_add_new_callers): Update.
6156 (decide_about_value): Update.
6157 (ipa_cp_c_finalize): Update.
6158 * ipa-devirt.c (struct odr_type_warn_count): Update.
6159 (struct decl_warn_count): Update.
6160 (struct final_warning_record): Update.
6161 (possible_polymorphic_call_targets): Update.
6162 (ipa_devirt): Update.
6163 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6164 * ipa-icf.c (sem_function::merge): Update.
6165 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6166 * ipa-inline.c (compute_uninlined_call_time): Update.
6167 (compute_inlined_call_time): Update.
6168 (want_inline_small_function_p): Update.
6169 (want_inline_self_recursive_call_p): Update.
6170 (edge_badness): Update.
6171 (lookup_recursive_calls): Update.
6172 (recursive_inlining): Update.
6173 (inline_small_functions): Update.
6174 (dump_overall_stats): Update.
6175 (dump_inline_stats): Update.
6176 * ipa-profile.c (ipa_profile_generate_summary): Update.
6177 (ipa_propagate_frequency): Update.
6178 (ipa_profile): Update.
6179 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6180 * ipa-utils.c (ipa_merge_profiles): Update.
6181 * loop-doloop.c (doloop_modify): Update.
6182 * loop-unroll.c (report_unroll): Update.
6183 (unroll_loop_runtime_iterations): Update.
6184 * lto-cgraph.c (lto_output_edge): Update.
6185 (lto_output_node): Update.
6186 (input_node): Update.
6187 (input_edge): Update.
6188 (merge_profile_summaries): Update.
6189 * lto-streamer-in.c (input_cfg): Update.
6190 * lto-streamer-out.c (output_cfg): Update.
6191 * mcf.c (create_fixup_graph): Update.
6192 (adjust_cfg_counts): Update.
6193 (sum_edge_counts): Update.
6194 * modulo-sched.c (sms_schedule): Update.
6195 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6196 * predict.c (maybe_hot_count_p): Update.
6197 (probably_never_executed): Update.
6198 (dump_prediction): Update.
6199 (combine_predictions_for_bb): Update.
6200 (propagate_freq): Update.
6201 (handle_missing_profiles): Update.
6202 (counts_to_freqs): Update.
6203 (rebuild_frequencies): Update.
6204 (force_edge_cold): Update.
6205 * predict.h: Include profile-count.h
6206 (maybe_hot_count_p, counts_to_freqs): UPdate.
6207 * print-rtl-function.c: Do not include cfg.h
6208 * print-rtl.c: Include basic-block.h
6209 * profile-count.c: New file.
6210 * profile-count.h: New file.
6211 * profile.c (is_edge_inconsistent): Update.
6212 (correct_negative_edge_counts): Update.
6213 (is_inconsistent): Update.
6214 (set_bb_counts): Update.
6215 (read_profile_edge_counts): Update.
6216 (compute_frequency_overlap): Update.
6217 (compute_branch_probabilities): Update; Initialize and deinitialize
6219 (branch_prob): Update.
6220 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6221 (edge_gcov_count): New.
6222 (bb_gcov_count): New.
6223 * shrink-wrap.c (try_shrink_wrapping): Update.
6224 * tracer.c (better_p): Update.
6225 * trans-mem.c (expand_transaction): Update.
6226 (ipa_tm_insert_irr_call): Update.
6227 (ipa_tm_insert_gettmclone_call): Update.
6228 * tree-call-cdce.c: Update.
6229 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6230 (gimple_duplicate_sese_tail): Update.
6231 (gimple_account_profile_record): Update.
6232 (execute_fixup_cfg): Update.
6233 * tree-inline.c (copy_bb): Update.
6234 (copy_edges_for_bb): Update.
6235 (initialize_cfun): Update.
6236 (freqs_to_counts): Update.
6237 (copy_cfg_body): Update.
6238 (expand_call_inline): Update.
6239 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6240 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6241 (try_unroll_loop_completely): Update.
6242 (try_peel_loop): Update.
6243 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6244 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6245 * tree-ssa-loop-split.c (connect_loops): Update.
6246 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6247 * tree-ssa-reassoc.c (branch_fixup): Update.
6248 * tree-ssa-tail-merge.c (replace_block_by): Update.
6249 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6250 (compute_path_counts): Update.
6251 (update_profile): Update.
6252 (recompute_probabilities): Update.
6253 (update_joiner_offpath_counts): Update.
6254 (estimated_freqs_path): Update.
6255 (freqs_to_counts_path): Update.
6256 (clear_counts_path): Update.
6257 (ssa_fix_duplicate_block_edges): Update.
6258 (duplicate_thread_path): Update.
6259 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6260 (struct switch_conv_info): Update.
6261 * tree-tailcall.c (decrease_profile): Update.
6262 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6263 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6264 * value-prof.c (check_counter): Update.
6265 (gimple_divmod_fixed_value): Update.
6266 (gimple_mod_pow2): Update.
6267 (gimple_mod_subtract): Update.
6268 (gimple_ic_transform): Update.
6269 (gimple_stringop_fixed_value): Update.
6270 * value-prof.h (gimple_ic): Update.
6272 2017-06-02 Carl Love <cel@us.ibm.com>
6274 * config/rs6000/rs6000-c: Add support for built-in functions
6275 vector double vec_doublee (vector signed int);
6276 vector double vec_doublee (vector unsigned int);
6277 vector double vec_doublee (vector float);
6278 vector double vec_doubleh (vector signed int);
6279 vector double vec_doubleh (vector unsigned int);
6280 vector double vec_doubleh (vector float);
6281 vector double vec_doublel (vector signed int);
6282 vector double vec_doublel (vector unsigned int);
6283 vector double vec_doublel (vector float);
6284 vector double vec_doubleo (vector signed int);
6285 vector double vec_doubleo (vector unsigned int);
6286 vector double vec_doubleo (vector float);.
6287 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6288 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6290 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6291 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6292 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6294 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6295 vec_doublel, vec_doubleh.
6296 * doc/extend.texi: Update the built-in documentation file for the
6297 new built-in functions.
6299 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6302 * ipa-inline-analysis.c (free_growth_caches): Set
6303 edge_removal_hook_holder to NULL after removing it.
6305 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6307 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6308 comparision with zero.
6310 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6311 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6312 for early expansion of vec_min and vec_max builtins.
6313 (builtin_function_type): Add min/max unsigned variants to those
6314 identified as having unsigned arguments.
6316 2017-06-02 Olivier Hainque <hainque@adacore.com>
6318 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6320 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6322 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6323 Use VALL_F16 iterator rather than VALL.
6325 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6327 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6328 Emit CBNZ inside loop when doing a strong exchange and comparing
6329 against zero. Generate the CC flags after the loop.
6331 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6333 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6334 (dl_section_ref): New.
6335 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6336 On AIX, append an expression to subtract the size of the
6337 section length to dl_section_ref.
6339 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6341 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6342 for early expansion of vector absolute builtins.
6344 2017-06-02 Richard Biener <rguenther@suse.de>
6346 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6347 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6349 2017-06-02 Richard Biener <rguenther@suse.de>
6351 PR tree-optimization/80948
6352 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6355 2017-06-02 Richard Biener <rguenther@suse.de>
6357 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6359 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6360 for inductions as relevant.
6362 2017-06-02 Richard Biener <rguenther@suse.de>
6364 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6365 (vectorizable_induction): ... this. Remove dead code.
6367 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6369 * builtins. (expand_builtin_alloca): Remove second parameter and
6370 infer its value from the first parameter instead.
6371 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6373 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6375 PR rtl-optimization/80903
6376 * loop-doloop.c (add_test): Unshare sequence.
6378 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6380 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6382 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6384 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6386 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6387 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6388 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6389 (xlogue_layout::get_stub_rtx): Make static.
6390 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6391 (xlogue_layout::compute_stub_managed_regs): Rename to...
6392 (xlogue_layout::count_stub_managed_regs): ...this.
6393 (xlogue_layout::is_stub_managed_reg): New function.
6394 (xlogue_layout::m_stub_names): Rename to...
6395 (xlogue_layout::s_stub_names): ...this, make static.
6396 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6397 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6398 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6399 xlogue_layout::s_stub_names): Instantiate statics.
6400 (stub_managed_regs): Remove.
6401 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6402 (disable_call_ms2sysv_xlogues): Rename to...
6403 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6404 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6406 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6407 change after reload_completed.
6408 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6410 (ix86_expand_prologue): Likewise.
6411 (ix86_expand_epilogue): Likewise.
6412 (ix86_expand_split_stack_prologue): Likewise.
6413 (ix86_compute_frame_layout): Remove frame parameter ...
6414 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6415 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6417 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6418 (ix86_frame): Move from here ...
6419 * config/i386/i386.h (ix86_frame): ... to here.
6420 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6421 complete ix86_frame data structure instead. Remove some_ld_name.
6423 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6425 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6426 symbols that hold a DECL_VALUE_EXPR.
6428 2017-06-01 Martin Jambor <mjambor@suse.cz>
6430 PR tree-optimization/80898
6431 * tree-sra.c (process_subtree_disqualification): Removed.
6432 (disqualify_candidate): Do not acll
6433 process_subtree_disqualification.
6434 (subtree_mark_written_and_enqueue): New function.
6435 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6436 RHS has been disqualified and re-queue LHS if necessary. Apart
6437 from that, ignore disqualified RHS.
6439 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6441 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6442 address fetch for z10 or later.
6444 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6446 * config/arc/arc.md (tst_movb): Add guard when splitting.
6448 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6450 * config/arc/arc.c (arc_can_eliminate): Test against
6451 arc_frame_pointer_needed.
6453 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6455 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6456 to prevent store reordering.
6457 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6458 (type): Add block type.
6459 (stack_tie): Define special instruction to be used in
6462 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6464 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6465 constraint. It is not valid for the pattern.
6466 (noncommutative_binary_comparison): Likewise.
6468 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6470 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6473 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6475 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6476 be used by the reg-alloc.
6478 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6480 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6481 reg-alloc when having mul64 or mul32x16 instructions.
6482 (mulsidi3): Likewise.
6483 (umulsidi3): Likewise.
6484 (mulsi32x16): New pattern.
6485 (mulsi64): Likewise.
6486 (mulsidi64): Likewise.
6487 (umulsidi64): Likewise.
6488 (MUL32x16_REG): Define.
6489 (mul64_600): Use MUL32x16_REG.
6490 (mac64_600): Likewise.
6491 (umul64_600): Likewise.
6492 (umac64_600): Likewise.
6494 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6496 * config/arc/arc.md (mulsi3_700): Make it commutative.
6498 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6500 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6502 (*sign_extendsidi2_insn): Likewise for movstosw.
6504 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6506 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6507 the type of the input discriminant value. Convert the
6508 discriminant value of signedness vary.
6510 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6512 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6513 Add to -Wall section.
6515 2017-06-01 Richard Biener <rguenther@suse.de>
6518 * fold-const.c (fold_plusminus_mult_expr): If the factored
6519 factor may be zero use a wrapping type for the inner operation.
6520 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6521 and handle moved defs.
6522 (process_assignment): Properly guard the unary op case. Return a
6523 tri-state indicating that moving the stmt before the call may allow
6524 to continue. Pass through to_move.
6525 (find_tail_calls): Handle moving unrelated defs before
6528 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6531 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
6532 splitter result in the canonical way.
6534 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
6536 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
6537 also for 32bit target. Update insn attributes.
6538 (zero-extendsidi2 splitter): Allow all registers for operand 1.
6540 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
6542 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
6543 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
6544 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
6545 (_mm_maskz_min_ss): New intrinsics.
6547 2017-05-31 Martin Liska <mliska@suse.cz>
6549 * tree-vect-loop.c (vect_create_epilog_for_reduction):
6550 Change comment style to one we normally use.
6551 (vectorizable_reduction): Likewise.
6552 (vectorizable_induction): Likewise.
6553 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6554 (vectorizable_call): Likewise.
6555 (vectorizable_simd_clone_call): Likewise.
6556 (vectorizable_conversion): Likewise.
6557 (vectorizable_assignment): Likewise.
6558 (vectorizable_shift): Likewise.
6559 (vectorizable_operation): Likewise.
6560 (vectorizable_store): Likewise.
6561 (vectorizable_load): Likewise.
6562 * tree-vectorizer.h: Likewise.
6564 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
6566 * passes.c (emergency_dump_function): New.
6567 * tree-pass.h (emergency_dump_function): Declare.
6568 * plugin.c (plugins_internal_error_function): Remove.
6569 * plugin.h (plugins_internal_error_function): Remove declaration.
6570 * toplev.c (internal_error_function): New static function. Use it...
6571 (general_init): ...here.
6573 2017-05-31 Graham Markall <graham.markall@embecosm.com>
6575 * config/arc/arc.c (arc_print_operand): Handle constant operands.
6576 (arc_rtx_costs): Add costs for new patterns.
6577 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
6578 * config/arc/predicates.md: Add _1_2_3_operand predicate.
6580 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6582 * tree-ssa-strlen.c (get_next_strinfo): New function.
6583 (get_stridx_plus_constant): Use it.
6584 (zero_length_string): Likewise.
6585 (adjust_related_strinfos): Likewise.
6586 (adjust_last_stmt): Likewise.
6588 2017-05-31 Richard Biener <rguenther@suse.de>
6591 * config/i386/i386.c (ix86_expand_builtin): Remove assert
6592 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
6594 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
6596 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
6597 loop_vinfo argument and use of dependence distance vectors.
6598 Check instead whether the two references differ only in their
6599 initial value and assume that they have the same alignment if the
6600 difference is a multiple of the vector alignment.
6601 (vect_analyze_data_refs_alignment): Update call accordingly.
6603 2017-05-31 Martin Liska <mliska@suse.cz>
6606 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
6608 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6610 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
6611 (create_intersect_range_checks): Move from ...
6612 * tree-data-ref.c (create_intersect_range_checks_index)
6613 (create_intersect_range_checks): ... to here.
6614 (create_runtime_alias_checks): New function factored from ...
6615 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
6616 here. Call above function.
6617 * tree-data-ref.h (create_runtime_alias_checks): New function.
6619 2017-05-31 Bin Cheng <bin.cheng@arm.com>
6621 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
6622 segment length for dr_b and compute it in wide_int.
6624 2017-05-31 Richard Biener <rguenther@suse.de>
6626 PR tree-optimization/80906
6627 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
6628 and pass through iv_map.
6629 (copy_bb_and_scalar_dependences): Adjust.
6630 (translate_pending_phi_nodes): Likewise.
6631 (copy_loop_close_phi_args): Handle code-generating IVs instead
6634 2017-05-30 David Malcolm <dmalcolm@redhat.com>
6636 * diagnostic-color.c (color_dict): Add "type-diff".
6637 (parse_gcc_colors): Update comment.
6638 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6639 -fdiagnostics-show-template-tree and -fno-elide-type.
6640 (GCC_COLORS): Add type-diff to example.
6642 (-fdiagnostics-show-template-tree): New.
6643 (-fno-elide-type): New.
6644 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
6645 the pp_format_decoder callback. Call any m_format_postprocessor's
6647 (pretty_printer::pretty_printer): Initialize
6648 m_format_postprocessor.
6649 (pretty_printer::~pretty_printer): Delete any
6650 m_format_postprocessor.
6651 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
6652 (class format_postprocessor): New class.
6653 (struct pretty_printer::format_decoder): Document the new parameters.
6654 (struct pretty_printer::m_format_postprocessor): New field.
6655 * tree-diagnostic.c (default_tree_printer): Update for new
6656 bool and const char ** params.
6657 * tree-diagnostic.h (default_tree_printer): Likewise.
6659 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
6661 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
6662 (lwa_operand): Delete rs6000_gen_cell_microcode test.
6663 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6664 rs6000_gen_cell_microcode code.
6665 (rs6000_final_prescan_insn): Delete.
6666 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
6667 "warn-cell-microcode" entries.
6668 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
6669 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
6670 throughout. Change cc_reg_not_micro_cr0_operand to
6671 cc_reg_not_cr0_operand throughout.
6672 (*extendhi<mode>2_noload): Delete.
6673 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
6674 (mwarn-cell-microcode): Delete.
6675 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
6676 -mgen-cell-microcode and -mwarn-cell-microcode.
6678 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
6681 * config/i386/constraints.md (Yd): New constraint.
6683 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
6684 and (?Yd, r) alternatives. Update insn attributes.
6685 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
6686 and (?*Yd, r) alternatives. Update insn attributes.
6687 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
6689 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
6691 * gimplify.c (gimplify_modify_expr): Don't create a
6692 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
6695 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6697 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
6699 2017-05-30 Richard Biener <rguenther@suse.de>
6701 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
6702 and reduc_def fields.
6703 (STMT_VINFO_REDUC_TYPE): New define.
6704 (STMT_VINFO_REDUC_DEF): Likewise.
6705 (vect_force_simple_reduction): Adjust prototype.
6706 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
6707 (vect_is_simple_reduction): Remove check_reduction argument.
6708 (vect_force_simple_reduction): Adjust and set
6709 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6710 (vectorizable_reduction): Do not re-do reduction analysis
6711 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
6712 * tree-parloops.c (gather_scalar_reductions): Adjust.
6714 2017-05-30 Richard Biener <rguenther@suse.de>
6717 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
6720 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
6722 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
6723 Introduce unknown_misalignment parameter and remove vf.
6724 (vect_peeling_hash_get_lowest_cost):
6725 Pass unknown_misalignment parameter.
6726 (vect_enhance_data_refs_alignment):
6727 Fix unsupportable data ref treatment.
6729 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6731 * tree-vect-data-refs.c (vect_get_data_access_cost):
6732 Workaround for SLP handling.
6733 (vect_enhance_data_refs_alignment):
6734 Compute costs for doing no peeling at all, compare to the best
6735 peeling costs so far and avoid peeling if cheaper.
6737 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6739 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
6740 Return peeling info and set costs to zero for unlimited cost
6742 (vect_enhance_data_refs_alignment): Also inspect all datarefs
6743 with unknown misalignment. Compute and costs for unknown
6744 misalignment, compare them to the costs for known misalignment
6745 and choose the cheapest for peeling.
6747 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6749 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
6750 (vect_get_peeling_costs_all_drs): Create function.
6751 (vect_peeling_hash_get_lowest_cost):
6752 Use vect_get_peeling_costs_all_drs.
6753 (vect_peeling_supportable): Create function.
6754 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
6756 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
6758 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
6759 DR_HAS_NEGATIVE_STEP.
6760 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
6761 (vect_enhance_data_refs_alignment): Use.
6762 (vect_duplicate_ssa_name_ptr_info): Use.
6763 * tree-vectorizer.h (dr_misalignment): Use.
6764 (known_alignment_for_access_p): Use.
6766 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6769 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
6771 (has_section_name): New function.
6773 2017-05-30 Martin Liska <mliska@suse.cz>
6776 * auto-profile.c (get_function_decl_from_block): Fix
6779 2017-05-30 Richard Biener <rguenther@suse.de>
6782 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
6784 2017-05-30 Martin Liska <mliska@suse.cz>
6786 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
6787 * dumpfile.h (struct dump_file_info): Remove ctors.
6789 2017-05-30 Martin Liska <mliska@suse.cz>
6791 * predict.def: Fix GNU coding style.
6793 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6795 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6796 Mark 'to' argument with ATTRIBUTE_UNUSED.
6798 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
6800 * config/xtensa/xtensa.c (xtensa_emit_call): Use
6801 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
6802 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
6805 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
6807 * doc/install.texi (Options specification): Restore entry of
6808 --enable-sjlj-exceptions.
6810 2017-05-27 Michael Eager <eager@eagercon.com>
6813 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
6815 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
6817 * config/microblaze/microblaze.h
6818 (FIXED_REGISTERS): Update in macro.
6819 (CALL_USED_REGISTERS): Update in macro.
6821 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
6823 * doc/install.texi: Add links to macOS binary distributions.
6825 2017-05-27 Jakub Jelinek <jakub@redhat.com>
6829 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6831 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6833 2017-05-26 Martin Liska <mliska@suse.cz>
6835 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
6837 2017-05-26 Martin Liska <mliska@suse.cz>
6839 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
6840 always leading ';; '.
6841 (dump_bb_info): Likewise.
6842 (brief_dump_cfg): Likewise.
6843 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
6844 * dumpfile.c: Remove usage of TDF_VERBOSE.
6845 * dumpfile.h (enum dump_kind): Likewise.
6846 (dump_gimple_bb_header): Do not use TDF_COMMENT.
6847 * print-tree.c (debug_verbose): Remove.
6848 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
6849 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
6850 * tree-diagnostic.c (default_tree_printer): Replace
6851 TDF_DIAGNOSTIC with TDF_SLIM.
6853 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6855 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
6856 in parameter loop, rather than loop_vinfo.
6857 (create_intersect_range_checks): Ditto.
6858 (vect_create_cond_for_alias_checks): Update call to above functions.
6860 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6862 PR tree-optimization/80815
6863 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
6864 for merging runtime alias checks. Handle negative DR_STEPs.
6866 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6868 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
6870 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
6871 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
6872 out code pruning runtime alias checks.
6873 * tree-data-ref.c (prune_runtime_alias_test_list): New function
6874 factored out from above.
6875 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6877 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
6879 (prune_runtime_alias_test_list): New decalaration.
6881 2017-05-26 Bin Cheng <bin.cheng@arm.com>
6883 * tree-vect-data-refs.c (compare_tree): Rename and move ...
6884 * tree-data-ref.c (data_ref_compare_tree): ... to here.
6885 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
6886 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
6887 (operator==, comp_dr_with_seg_len_pair): Ditto.
6888 (vect_prune_runtime_alias_test_list): Ditto.
6890 2017-05-26 Martin Liska <mliska@suse.cz>
6893 * params.def: Bound partial-inlining-entry-probability param.
6895 2017-05-26 Marek Polacek <polacek@redhat.com>
6898 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
6901 2017-05-26 Richard Biener <rguenther@suse.de>
6903 PR tree-optimization/80842
6904 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
6907 2017-05-26 Richard Biener <rguenther@suse.de>
6909 PR tree-optimization/80844
6910 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
6912 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
6914 * doc/md.texi (Machine Constraints): Update x86 family
6915 machine constraints section to match 'config/i386/constraints.md'.
6917 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
6919 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
6921 2017-05-25 Nathan Sidwell <nathan@acm.org>
6923 * doc/invoke.texi (--enable-languages): Update documentation.
6925 2017-05-25 Martin Liska <mliska@suse.cz>
6927 * dumpfile.c: Add TDF_FOLDING.
6928 * dumpfile.h (enum dump_kind): Likewise.
6929 * genmatch.c (dt_simplify::gen_1): Use it.
6931 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6933 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
6935 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6937 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
6938 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
6940 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
6942 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
6943 * match.pd (X == C): Rewrite it here.
6944 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
6945 with_certain_nonzero_bits2): New predicates.
6946 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
6948 2017-05-24 Nathan Sidwell <nathan@acm.org>
6950 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
6953 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
6956 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6958 * config/powerpcspe: New port. Files are copied from the rs6000
6959 port, with "rs6000" in filenames replaced by "powerpcspe".
6961 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
6963 PR rtl-optimization/80754
6964 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
6966 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
6968 * config/sparc/sparc.md (length): Return the correct value for -mflat
6969 sibcalls to match output_sibcall.
6971 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
6975 * config/rs6000/rs6000.c (struct machine_function): Add new field
6977 (rs6000_get_separate_components): Init that field, use it.
6978 (rs6000_components_for_bb): Use the field.
6980 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
6982 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
6984 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
6987 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
6989 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6992 * config/s390/s390.c (s390_check_qrst_address): Check incoming
6993 address against address_operand predicate.
6994 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
6996 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
6998 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
6999 parameters passed indirectly.
7001 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7003 * config/i386/i386.md (*movdi_internal): Remove SSE4
7004 alternative 18 (?r, *v). Update insn attributes.
7005 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7006 Update insn attributes.
7007 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7008 Update insn attributes.
7009 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7010 alternative 1 (r, v). Remove isa attribute.
7011 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7012 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7013 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7015 2017-05-23 Tom de Vries <tom@codesourcery.com>
7017 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7020 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7022 * cgraphunit.c (symbol_table::process_new_functions): Update.
7023 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7024 (inline_generate_summary): Rename to ...
7025 (ipa_fn_summary_generate): ... this one.
7026 (inline_read_summary): Rename to ...
7027 (ipa_fn_summary_read): ... this one.
7028 (inline_write_summary): Rename to ...
7029 (ipa_fn_summary_write): ... this one.
7030 (inline_free_summary): Rename to ...
7031 (ipa_free_fn_summary): ... this one.
7032 (pass_data_local_fn_summary, pass_local_fn_summary,
7033 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7034 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7035 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7036 make_pass_ipa_fn_summary): New.
7037 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7038 inline_write_summary, inline_free_summary): Remove.
7039 (ipa_free_fn_summary) : New.
7040 * ipa-inline.c (ipa_inline): Update.
7041 (pass_ipa_inline): Do not generate summaries.
7042 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7044 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7045 and add pass_ipa_fn_summary.
7046 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7048 (make_pass_inline_parameters): Remove.
7050 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7052 * omp-low.c (struct omp_context): Remove "default_kind" member.
7055 * omp-offload.c (execute_oacc_device_lower): Remove the
7056 parallelism dimensions function attributes for unparallelized
7057 OpenACC kernels constructs.
7059 2017-05-23 Martin Liska <mliska@suse.cz>
7061 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7063 (cgraph_edge::make_speculative): Likewise.
7064 (cgraph_edge::resolve_speculation): Likewise.
7065 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7066 (cgraph_node::dump): Likewise.
7067 * cgraph.h: Likewise.
7068 * cgraphunit.c (analyze_functions): Likewise.
7069 (symbol_table::compile): Likewise.
7070 * ipa-cp.c (print_all_lattices): Likewise.
7071 (determine_versionability): Likewise.
7072 (initialize_node_lattices): Likewise.
7073 (ipcp_verify_propagated_values): Likewise.
7074 (estimate_local_effects): Likewise.
7075 (update_profiling_info): Likewise.
7076 (create_specialized_node): Likewise.
7077 (perhaps_add_new_callers): Likewise.
7078 (decide_about_value): Likewise.
7079 (decide_whether_version_node): Likewise.
7080 (identify_dead_nodes): Likewise.
7081 (ipcp_store_bits_results): Likewise.
7082 * ipa-devirt.c (dump_targets): Likewise.
7083 (ipa_devirt): Likewise.
7084 * ipa-icf.c (sem_item::dump): Likewise.
7085 (sem_function::equals): Likewise.
7086 (sem_variable::equals): Likewise.
7087 (sem_item_optimizer::read_section): Likewise.
7088 (sem_item_optimizer::execute): Likewise.
7089 (congruence_class::dump): Likewise.
7090 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7091 (dump_inline_summary): Likewise.
7092 (estimate_node_size_and_time): Likewise.
7093 (inline_analyze_function): Likewise.
7094 * ipa-inline-transform.c (inline_call): Likewise.
7095 * ipa-inline.c (report_inline_failed_reason): Likewise.
7096 (want_early_inline_function_p): Likewise.
7097 (edge_badness): Likewise.
7098 (update_edge_key): Likewise.
7099 (inline_small_functions): Likewise.
7100 * ipa-profile.c (ipa_profile): Likewise.
7101 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7102 (ipa_make_edge_direct_to_target): Likewise.
7103 (remove_described_reference): Likewise.
7104 (ipa_impossible_devirt_target): Likewise.
7105 (propagate_controlled_uses): Likewise.
7106 (ipa_print_node_params): Likewise.
7107 (ipcp_transform_function): Likewise.
7108 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7109 (propagate_pure_const): Likewise.
7110 * ipa-reference.c (generate_summary): Likewise.
7111 (read_write_all_from_decl): Likewise.
7112 (propagate): Likewise.
7113 (ipa_reference_read_optimization_summary): Likewise.
7114 * ipa-utils.c (ipa_merge_profiles): Likewise.
7115 * ipa.c (walk_polymorphic_call_targets): Likewise.
7116 (symbol_table::remove_unreachable_nodes): Likewise.
7117 (ipa_single_use): Likewise.
7118 * passes.c (execute_todo): Likewise.
7119 * predict.c (drop_profile): Likewise.
7120 * symtab.c (symtab_node::get_dump_name): New function.
7121 (symtab_node::dump_name): Likewise.
7122 (symtab_node::dump_asm_name): Likewise.
7123 (symtab_node::dump_references): Likewise.
7124 (symtab_node::dump_referring): Likewise.
7125 (symtab_node::dump_base): Likewise.
7126 (symtab_node::debug_symtab): Likewise.
7127 * tree-sra.c (convert_callers_for_node): Likewise.
7128 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7129 * value-prof.c (init_node_map): Likewise.
7131 2017-05-23 Martin Liska <mliska@suse.cz>
7133 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7134 and symtab_node::debug_symtab to symbol_table::debug.
7135 * cgraphunit.c (analyze_functions): Use the renamed function.
7136 (symbol_table::compile): Likewise.
7137 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7138 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7139 * passes.c (execute_todo): Likewise.
7140 * symtab.c (symbol_table::dump): New function.
7141 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7143 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7145 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7146 that nonconst implies exec.
7148 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7150 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7151 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7152 (inline_edge_summary_vec): Turn into ...
7153 (ipa_call_summaries): ... this one.
7154 (redirect_to_unreachable, edge_set_predicate,
7155 evaluate_properties_for_edge, inline_summary_alloc,
7156 reset_ipa_call_summary, reset_inline_summary,
7157 inline_summary_t::duplicate): Update.
7158 (inline_edge_duplication_hook): Turn to ...
7159 (ipa_call_summary_t::duplicate): ... this one.
7160 (inline_edge_removal_hook): Turn to ...
7161 (ipa_call_summary_t::remove): ... this one.
7162 (dump_inline_edge_summary): Turn to ...
7163 (dump_ipa_call_summary): ... this one.
7164 (estimate_function_body_sizes): Update.
7165 (inline_update_callee_summaries): Update.
7166 (remap_edge_change_prob): Update.
7167 (remap_edge_summaries): Update.
7168 (inline_merge_summary): Update.
7169 (do_estimate_edge_time): Update.
7170 (inline_generate_summary): Update.
7171 (inline_read_section): Update.
7172 (inline_read_summary): Update.
7173 (inline_free_summary): Update.
7174 * ipa-inline.c (can_inline_edge_p): Update.
7175 (compute_inlined_call_time): Update.
7176 (want_inline_small_function_p): Update.
7177 (edge_badness): Update.
7178 (early_inliner): Update.
7179 * ipa-inline.h (inline_edge_summary): Turn to ...
7180 (ipa_call_summary): ... this one.
7181 (ipa_call_summary_t): New class.
7182 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7183 (ipa_call_summaries): New.
7184 (inline_edge_summary): Remove.
7185 (estimate_edge_growth): Update.
7186 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7187 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7188 * ipa-split.c (execute_split_functions): Update.
7189 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7191 2017-05-23 Tom de Vries <tom@codesourcery.com>
7193 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7194 attributes): Document rdrand effective target.
7196 2017-05-23 Tom de Vries <tom@codesourcery.com>
7198 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7199 attributes): Sort alphabetically.
7201 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7203 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7205 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7208 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7209 V2DF/V2DI splat into two separate patterns, one that handles
7210 registers, and the other that only handles memory. Drop support
7211 for splatting from a GPR on ISA 2.07 and then splitting the
7212 splat into direct move and splat.
7213 (vsx_splat_<mode>_reg): Likewise.
7214 (vsx_splat_<mode>_mem): Likewise.
7216 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7218 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7220 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7223 * omp-low.c (finish_taskreg_remap): New function.
7224 (finish_taskreg_scan): If unit size of ctx->record_type
7225 is non-constant, unshare the size expression and replace
7226 decls in it with possible outer var refs.
7229 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7230 GOVD_SHARED rather than GOVD_PRIVATE with it.
7231 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7232 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7235 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7236 as last argument to build_outer_var_ref for pointer bases of array
7239 2017-05-19 Martin Sebor <msebor@redhat.com>
7241 * print-tree.c (print_node): Print DECL_READ_P flag.
7243 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7245 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7246 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7247 * cgraph.c: Likewise.
7248 * cgraphunit.c: Likewise.
7249 * gengtype.c: Likewise.
7250 * ipa-cp.c: Likewise.
7251 * ipa-devirt.c: Likewise.
7252 * ipa-icf.c: Likewise.
7253 * ipa-predicate.c: Likewise.
7254 * ipa-profile.c: Likewise.
7255 * ipa-prop.c: Likewise.
7256 * ipa-split.c: Likewise.
7258 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7259 edge_predicate_pool, dump_inline_hints,
7260 inline_summary::account_size_time, redirect_to_unreachable,
7261 edge_set_predicate, set_hint_predicate,
7262 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7263 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7264 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7265 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7266 ipa_call_summary_t::remove, initialize_growth_caches,
7267 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7268 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7269 mark_modified, unmodified_parm_1, unmodified_parm,
7270 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7271 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7272 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7273 will_be_nonconstant_predicate, record_modified_bb_info,
7274 get_minimal_bb, record_modified, param_change_prob,
7275 phi_result_unknown_predicate, predicate_for_phi_result,
7276 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7277 estimate_function_body_sizes, compute_inline_parameters,
7278 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7279 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7280 inline_update_callee_summaries, remap_edge_change_prob,
7281 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7282 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7283 inline_analyze_function, inline_summary_t::insert,
7284 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7285 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7286 inline_free_summary): Move to ipa-fnsummary.h
7287 (predicate_t): Remove.
7288 * ipa-fnsummary.c: New file.
7289 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7290 (enum inline_hints_vals, inline_hints, agg_position_info,
7291 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7292 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7293 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7294 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7295 inline_read_summary, inline_write_summary, inline_free_summary,
7296 inline_analyze_function, initialize_inline_failed,
7297 inline_merge_summary, inline_update_overall_summary,
7298 compute_inline_parameters): Move to ipa-fnsummary.h
7299 * ipa-fnsummary.h: New file.
7300 * ipa-inline-transform.h: Include ipa-inline.h.
7301 * ipa-inline.c: LIkewise.
7303 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7305 * ipa-inline.c (edge_badness): Use inlined_time instead of
7306 inline_summaries->get.
7308 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7310 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7312 2017-05-22 Nathan Sidwell <nathan@acm.org>
7314 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7315 (fdump-lang): Document 'raw' option.
7316 * dumpfile.h (TDI_tu): Delete.
7317 * dumpfile.c (dump_files): Remove translation-unit.
7318 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7320 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7322 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7323 command option from $(AWK) call.
7324 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7325 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7326 [FORMAT]: Remove handling of variable.
7327 * config/avr/t-multilib: Regenerate.
7329 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7331 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7333 (dump_inline_summary): Do not print self_time.
7334 (estimate_function_body_sizes): Do not set self_time.
7335 (compute_inline_parameters): Likewise.
7336 (inline_read_section, inline_write_summary): Do not stream self_time.
7337 * ipa-inline.h (inline_summary): Drop self_time.
7339 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7341 * ipa-inline-analysis.c (account_size_time): Rename to ...
7342 (inline_summary::account_size_time): ... this one.
7343 (reset_ipa_call_summary): Turn to ...
7344 (ipa_call_summary::reset): ... this one.
7345 (reset_inline_summary): Turn to ...
7346 (inline_summary::reset): ... this one.
7347 (inline_summary_t::remove): Update.
7348 (inline_summary_t::duplicate): Update.
7349 (ipa_call_summary_t::remove): Update.
7350 (dump_inline_summary): Update.
7351 (estimate_function_body_sizes): Update.
7352 (compute_inline_parameters): Update.
7353 (estimate_node_size_and_time): Update.
7354 (inline_merge_summary): Update.
7355 (inline_update_overall_summary): Update.
7356 (inline_read_section): Update.
7357 (inline_write_summary): Update.
7358 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7359 add account_size_time and reset member functions.
7360 (ipa_call_summary): Add reset function.
7361 * ipa-predicate.h (predicate::operator &): Constify.
7363 2017-05-22 Richard Biener <rguenther@suse.de>
7365 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7367 2017-05-19 Jason Merrill <jason@redhat.com>
7369 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7371 2017-05-19 Marek Polacek <polacek@redhat.com>
7374 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7375 TYPE_OVERFLOW_WRAPS checks.
7377 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7379 * tree-core.h (enum omp_clause_default_kind): Add
7380 "OMP_CLAUSE_DEFAULT_PRESENT".
7381 * tree-pretty-print.c (dump_omp_clause): Handle it.
7382 * gimplify.c (enum gimplify_omp_var_data): Add
7383 "GOVD_MAP_FORCE_PRESENT".
7384 (gimplify_adjust_omp_clauses_1): Map it to
7385 "GOMP_MAP_FORCE_PRESENT".
7386 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7388 * gimplify.c (oacc_default_clause): Clarify.
7390 2017-05-19 Nathan Sidwell <nathan@acm.org>
7392 LANG_HOOK_REGISTER_DUMPS
7393 * toplev.c (general_init): Call register dump lang hook.
7394 * doc/invoke.texi: Document -fdump-lang option family.
7395 * dumpfile.c (dump_files): Remove class dump here.
7396 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7397 * dumpfile.h (tree_dump_index): Remove TDI_class.
7398 * langhooks-def.h (lhd_register_dumps): Declare.
7399 (LANG_HOOKS_REGISTER_DUMPS): Define.
7400 (LANG_HOOKS_INITIALIZER): Add it.
7401 * langhooks.c (lhd_register_dumps): Define.
7402 * langhooks.h (struct lang_hooks): Add register_dumps.
7404 2017-05-19 Nathan Sidwell <nathan@acm.org>
7406 * context.h (context::set_passes): New.
7407 * context.c (context::context): Do not create pass manager.
7408 * toplev.c (general_init): Create pass manager here.
7410 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7412 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7413 use this splitter if two add or or instructions would also work for
7414 the constant we want to generate.
7416 2017-05-19 Richard Biener <rguenther@suse.de>
7419 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7420 predicate evaluation.
7422 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7424 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7426 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7429 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7431 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7432 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7433 (inline_edge_summary_vec): Turn into ...
7434 (ipa_call_summaries): ... this one.
7435 (redirect_to_unreachable, edge_set_predicate,
7436 evaluate_properties_for_edge, inline_summary_alloc,
7437 reset_ipa_call_summary, reset_inline_summary,
7438 inline_summary_t::duplicate): Update.
7439 (inline_edge_duplication_hook): Turn to ...
7440 (ipa_call_summary_t::duplicate): ... this one.
7441 (inline_edge_removal_hook): Turn to ...
7442 (ipa_call_summary_t::remove): ... this one.
7443 (dump_inline_edge_summary): Turn to ...
7444 (dump_ipa_call_summary): ... this one.
7445 (estimate_function_body_sizes): Update.
7446 (inline_update_callee_summaries): Update.
7447 (remap_edge_change_prob): Update.
7448 (remap_edge_summaries): Update.
7449 (inline_merge_summary): Update.
7450 (do_estimate_edge_time): Update.
7451 (inline_generate_summary): Update.
7452 (inline_read_section): Update.
7453 (inline_read_summary): Update.
7454 (inline_free_summary): Update.
7455 * ipa-inline.c (can_inline_edge_p): Update.
7456 (compute_inlined_call_time): Update.
7457 (want_inline_small_function_p): Update.
7458 (edge_badness): Update.
7459 (early_inliner): Update.
7460 * ipa-inline.h (inline_edge_summary): Turn to ...
7461 (ipa_call_summary): ... this one.
7462 (ipa_call_summary_t): New class.
7463 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7464 (ipa_call_summaries): New.
7465 (inline_edge_summary): Remove.
7466 (estimate_edge_growth): Update.
7467 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7468 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7469 * ipa-split.c (execute_split_functions): Update.
7470 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7472 2017-05-19 Richard Biener <rguenther@suse.de>
7475 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7477 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7479 * config/rs6000/rs6000.c (struct machine_function): Add field
7480 fpr_is_wrapped_separately.
7481 (rs6000_get_separate_components): Use 64 components. Handle the
7483 (rs6000_components_for_bb): Handle the FPR components.
7484 (rs6000_emit_prologue_components): Handle the FPR components.
7485 (rs6000_emit_epilogue_components): Handle the FPR components.
7486 (rs6000_set_handled_components): Handle the FPR components.
7487 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7488 that are already separately shrink-wrapped.
7489 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7490 that are already separately shrink-wrapped.
7492 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7495 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7498 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7499 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7500 cases where the register allocator uses a move and an offsettable
7501 memory operation to/from a FPR register on ISA 2.06/2.07.
7502 (define_peephole2 for Altivec d-form store): Likewise.
7504 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7507 * config/i386/mmx.md (*mov<mode>_internal): Enable
7508 alternatives 11, 12, 13 and 14 also for 32bit targets.
7509 Remove alternatives 15, 16, 17 and 18.
7510 * config/i386/sse.md (vec_concatv2di): Change
7511 alternative (!x, *y) to (x, ?!*Yn).
7513 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7515 * dumpfile.h (enum dump_kind): Remove stray comma.
7517 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7519 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7520 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7521 predicate::num_conditions
7522 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7523 (CHANGED): turn into predicate::changed.
7524 (agg_position_info): Move to ipa-predicate.h
7525 (add_condition, predicate::add_clause, predicate::operator &=,
7526 predicate::or_with, predicate::evaluate, predicate::probability,
7527 dump_condition, dump_clause, predicate::dump,
7528 predicate::remap_after_duplication, predicate::remap_after_inlining,
7529 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
7530 (evaluate_conditions_for_known_args): Update.
7531 (set_cond_stmt_execution_predicate): Update.
7532 * ipa-inline.h: Include ipa-predicate.h
7533 (condition, inline_param_summary, conditions, agg_position_info,
7534 predicate): Move to ipa-predicate.h
7535 * ipa-predicate.c: New file.
7536 * ipa-predicate.h: New file.
7538 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
7540 * final.c (leaf_function_p): Check we are not in a sequence.
7542 2017-05-18 Martin Liska <mliska@suse.cz>
7544 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
7545 * dumpfile.c (dump_register): Use new enum dump_kind.
7546 (get_dump_file_name): Likewise.
7547 (dump_enable_all): Likewise.
7548 (dump_switch_p_1): Likewise.
7549 (enable_rtl_dump_file): Remove usage of TDF_RTL.
7550 * dumpfile.h (enum dump_kind): New enum type.
7551 (struct dump_file_info): Create constructor and
7552 format fields and comments.
7553 * passes.c (pass_manager::register_one_dump_file):
7555 * statistics.c (statistics_early_init): Likewise.
7556 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
7557 TDF_TREE with TDF_SLIM.
7558 (gather_memory_references_ref): Likewise.
7560 2017-05-18 Martin Liska <mliska@suse.cz>
7562 * vec.h (struct vnull): Use it.
7564 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7566 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
7567 (true_predicate, false_predicate, true_predicate_p,
7568 false_predicate_p): Remove.
7569 (single_cond_predicate, not_inlined_predicate): Turn to member function
7571 (add_condition): Update.
7572 (add_clause): Turn to...
7573 (predicate::add_clause): ... this one; update; allow passing NULL
7575 (and_predicates): Turn to ...
7576 (predicate::operator &=): ... this one.
7577 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
7578 (or_predicates): Turn to ...
7579 (predicate::or_with): ... this one.
7580 (evaluate_predicate): Turn to ...
7581 (predicate::evaluate): ... this one.
7582 (predicate_probability): Turn to ...
7583 (predicate::probability): ... this one.
7584 (dump_condition): Update.
7585 (dump_predicate): Turn to ...
7586 (predicate::dump): ... this one.
7587 (account_size_time): Update.
7588 (edge_set_predicate): Update.
7589 (set_hint_predicate): UPdate.
7590 (evaluate_conditions_for_known_args): Update.
7591 (evaluate_properties_for_edge): Update.
7592 (remap_predicate_after_duplication): Turn to...
7593 (predicate::remap_after_duplication): ... this one.
7594 (remap_hint_predicate_after_duplication): Update.
7595 (inline_summary_t::duplicate): UPdate.
7596 (dump_inline_edge_summary): Update.
7597 (dump_inline_summary): Update.
7598 (set_cond_stmt_execution_predicate): Update.
7599 (set_switch_stmt_execution_predicate): Update.
7600 (compute_bb_predicates): Update.
7601 (will_be_nonconstant_expr_predicate): Update.
7602 (will_be_nonconstant_predicate): Update.
7603 (phi_result_unknown_predicate): Update.
7604 (predicate_for_phi_result): Update.
7605 (array_index_predicate): Update.
7606 (estimate_function_body_sizes): Update.
7607 (estimate_node_size_and_time): Update.
7608 (estimate_ipcp_clone_size_and_time): Update.
7609 (remap_predicate): Rename to ...
7610 (predicate::remap_after_inlining): ... this one.
7611 (remap_hint_predicate): Update.
7612 (inline_merge_summary): Update.
7613 (inline_update_overall_summary): Update.
7614 (estimate_size_after_inlining): Update.
7615 (read_predicate): Rename to ...
7616 (predicate::stream_in): ... this one.
7617 (read_inline_edge_summary): Update.
7618 (write_predicate): Rename to ...
7619 (predicate::stream_out): ... this one.
7620 (write_inline_edge_summary): Update.
7621 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
7622 (clause_t): Turn to uint32_t
7623 (predicate): Turn to class; implement constructor and operators
7625 (size_time_entry): Update.
7626 (inline_summary): Update.
7627 (inline_edge_summary): Update.
7629 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
7631 * fold-const.c (fold_binary_loc): Move transformation...
7632 * match.pd (C - X CMP X): ... here.
7634 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
7636 * config/sparc/sparc.c (sparc_option_override): Set function
7637 alignment for -mcpu=niagara7 to 64 to match the I$ line.
7638 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
7640 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
7642 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
7644 2017-05-18 Marek Polacek <polacek@redhat.com>
7647 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
7648 (pass_ubsan::execute): Call gimple_assign_single_p instead of
7649 gimple_assign_load_p.
7651 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
7654 * real.c (do_compare): Give decimal_do_compare preference over
7655 comparing just the signs.
7657 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
7659 * doc/md.texi (Canonicalization of Instructions): Describe the
7660 canonical form of instructions that inherently set a condition
7663 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
7666 * tree-cfg.c: Move deletion of unreachable case statements to after
7667 the merging of consecutive case labels.
7669 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7671 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
7672 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
7673 restoring of callee-saved registers.
7675 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
7677 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
7678 * config/visium/visium.c (single_set_and_flags): Likewise.
7679 * config/visium/visium.md (Substitutions): Likewise.
7681 2017-05-17 Martin Liska <mliska@suse.cz>
7683 * cfg.c: Introduce dump_flags_t type and
7684 use it instead of int type.
7686 * cfghooks.c: Likewise.
7687 * cfghooks.h (struct cfg_hooks): Likewise.
7688 * cfgrtl.c: Likewise.
7689 * cfgrtl.h: Likewise.
7690 * cgraph.c (cgraph_node::get_body): Likewise.
7691 * coretypes.h: Likewise.
7692 * domwalk.c: Likewise.
7693 * domwalk.h: Likewise.
7694 * dumpfile.c (struct dump_option_value_info): Likewise.
7695 (dump_enable_all): Likewise.
7696 (dump_switch_p_1): Likewise.
7697 (opt_info_switch_p): Likewise.
7698 * dumpfile.h (enum tree_dump_index): Likewise.
7699 (struct dump_file_info): Likewise.
7700 * genemit.c: Likewise.
7701 * generic-match-head.c: Likewise.
7702 * gengtype.c (open_base_files): Likewise.
7703 * gimple-pretty-print.c: Likewise.
7704 * gimple-pretty-print.h: Likewise.
7705 * graph.c (print_graph_cfg): Likewise.
7706 * graphite-scop-detection.c (dot_all_sese): Likewise.
7707 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7708 * loop-unroll.c (report_unroll): Likewise.
7709 * passes.c (pass_manager::register_one_dump_file): Likewise.
7710 * print-tree.c: Likewise.
7711 * statistics.c: Likewise.
7712 * tree-cfg.c: Likewise.
7713 * tree-cfg.h: Likewise.
7714 * tree-dfa.c: Likewise.
7715 * tree-dfa.h: Likewise.
7716 * tree-dump.c (dump_function): Likewise.
7717 * tree-dump.h (struct dump_info): Likewise.
7718 * tree-pretty-print.c: Likewise.
7719 * tree-pretty-print.h: Likewise.
7720 * tree-ssa-live.c: Likewise.
7721 * tree-ssa-live.h: Likewise.
7722 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
7723 * tree-vect-loop.c: Likewise.
7724 * tree-vect-slp.c: Likewise.
7726 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
7727 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7729 PR tree-optimization/80457
7730 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
7731 of all arguments to a statement as scalar_to_vec operations.
7732 (vectorizable_call): Adjust call to vect_model_simple_cost for
7734 (vectorizable_conversion): Likewise.
7735 (vectorizable_assignment): Likewise.
7736 (vectorizable_shift): Likewise.
7737 (vectorizable_operation): Likewise.
7738 (vectorizable_comparison): Likewise.
7739 (vect_is_simple_cond): Record the def types for operands.
7740 (vectorizable_condition): Likewise, call vect_model_simple_cost.
7741 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
7742 for statement argument count.
7744 2017-05-16 Carl Love <cel@us.ibm.com>
7746 * config/rs6000/rs6000-c: Add support for built-in functions
7747 vector unsigned long long vec_bperm (vector unsigned long long,
7748 vector unsigned char)
7749 vector signed long long vec_mule (vector signed int,
7751 vector unsigned long long vec_mule (vector unsigned int,
7752 vector unsigned int)
7753 vector signed long long vec_mulo (vector signed int,
7755 vector unsigned long long vec_mulo (vector unsigned int,
7756 vector unsigned int)
7757 vector signed char vec_sldw (vector signed char,
7760 vector unsigned char vec_sldw (vector unsigned char,
7761 vector unsigned char,
7763 vector signed short vec_sldw (vector signed short,
7764 vector signed short,
7766 vector unsigned short vec_sldw (vector unsigned short,
7767 vector unsigned short,
7769 vector signed int vec_sldw (vector signed int,
7772 vector unsigned int vec_sldw (vector unsigned int,
7773 vector unsigned int,
7775 vector signed long long vec_sldw (vector signed long long,
7776 vector signed long long,
7778 vector unsigned long long vec_sldw (vector unsigned long long,
7779 vector unsigned long long,
7781 * config/rs6000/rs6000-c: Add support for built-in functions
7782 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
7783 * config/rs6000/altivec.h: Add defintion for vec_sldw.
7784 * doc/extend.texi: Update the built-in documentation for the
7785 new built-in functions.
7787 2017-05-16 Marek Polacek <polacek@redhat.com>
7791 * tree.c (save_expr): Don't fold the expression.
7793 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
7795 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
7796 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
7797 and (?*y,m). Update insn attributes.
7799 2017-05-16 Martin Liska <mliska@suse.cz>
7801 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
7802 flags argument of print_gimple_stmt, print_gimple_expr,
7803 print_generic_stmt and print_generic_expr.
7804 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
7805 * coretypes.h: Likewise.
7806 * except.c (dump_eh_tree): Likewise.
7807 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7808 * gimple-pretty-print.h: Likewise.
7809 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
7810 (backprop::push_to_worklist): Likewise.
7811 (backprop::pop_from_worklist): Likewise.
7812 (backprop::process_use): Likewise.
7813 (backprop::intersect_uses): Likewise.
7814 (note_replacement): Likewise.
7815 * gimple-ssa-store-merging.c
7816 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
7817 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7818 (pass_store_merging::execute): Likewise.
7819 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
7820 (ssa_base_cand_dump_callback): Likewise.
7821 (dump_incr_vec): Likewise.
7822 (replace_refs): Likewise.
7823 (replace_mult_candidate): Likewise.
7824 (create_add_on_incoming_edge): Likewise.
7825 (create_phi_basis): Likewise.
7826 (insert_initializers): Likewise.
7827 (all_phi_incrs_profitable): Likewise.
7828 (introduce_cast_before_cand): Likewise.
7829 (replace_one_candidate): Likewise.
7830 * gimplify.c (gimplify_expr): Likewise.
7831 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
7832 (set_rename): Likewise.
7833 (rename_uses): Likewise.
7834 (copy_loop_phi_nodes): Likewise.
7835 (add_close_phis_to_merge_points): Likewise.
7836 (copy_loop_close_phi_args): Likewise.
7837 (copy_cond_phi_args): Likewise.
7838 (graphite_copy_stmts_from_block): Likewise.
7839 (translate_pending_phi_nodes): Likewise.
7840 * graphite-poly.c (print_pdr): Likewise.
7841 (dump_gbb_cases): Likewise.
7842 (dump_gbb_conditions): Likewise.
7843 (print_scop_params): Likewise.
7844 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
7845 (build_cross_bb_scalars_use): Likewise.
7846 (gather_bbs::before_dom_children): Likewise.
7847 * hsa-dump.c (dump_hsa_immed): Likewise.
7848 * ipa-cp.c (print_ipcp_constant_value): Likewise.
7849 (get_replacement_map): Likewise.
7850 * ipa-inline-analysis.c (dump_condition): Likewise.
7851 (estimate_function_body_sizes): Likewise.
7852 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
7853 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
7854 * ipa-prop.c (ipa_dump_param): Likewise.
7855 (ipa_print_node_jump_functions_for_edge): Likewise.
7856 (ipa_modify_call_arguments): Likewise.
7857 (ipa_modify_expr): Likewise.
7858 (ipa_dump_param_adjustments): Likewise.
7859 (ipa_dump_agg_replacement_values): Likewise.
7860 (ipcp_modif_dom_walker::before_dom_children): Likewise.
7861 * ipa-pure-const.c (check_stmt): Likewise.
7862 (pass_nothrow::execute): Likewise.
7863 * ipa-split.c (execute_split_functions): Likewise.
7864 * omp-offload.c (dump_oacc_loop_part): Likewise.
7865 (dump_oacc_loop): Likewise.
7866 * trans-mem.c (tm_log_emit): Likewise.
7867 (tm_memopt_accumulate_memops): Likewise.
7868 (dump_tm_memopt_set): Likewise.
7869 (dump_tm_memopt_transform): Likewise.
7870 * tree-cfg.c (gimple_verify_flow_info): Likewise.
7871 (print_loop): Likewise.
7872 * tree-chkp-opt.c (chkp_print_addr): Likewise.
7873 (chkp_gather_checks_info): Likewise.
7874 (chkp_get_check_result): Likewise.
7875 (chkp_remove_check_if_pass): Likewise.
7876 (chkp_use_outer_bounds_if_possible): Likewise.
7877 (chkp_reduce_bounds_lifetime): Likewise.
7878 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
7879 (chkp_mark_completed_bounds): Likewise.
7880 (chkp_register_incomplete_bounds): Likewise.
7881 (chkp_mark_invalid_bounds): Likewise.
7882 (chkp_maybe_copy_and_register_bounds): Likewise.
7883 (chkp_build_returned_bound): Likewise.
7884 (chkp_get_bound_for_parm): Likewise.
7885 (chkp_build_bndldx): Likewise.
7886 (chkp_get_bounds_by_definition): Likewise.
7887 (chkp_generate_extern_var_bounds): Likewise.
7888 (chkp_get_bounds_for_decl_addr): Likewise.
7889 * tree-chrec.c (chrec_apply): Likewise.
7890 * tree-data-ref.c (dump_data_reference): Likewise.
7891 (dump_subscript): Likewise.
7892 (dump_data_dependence_relation): Likewise.
7893 (analyze_overlapping_iterations): Likewise.
7894 * tree-inline.c (expand_call_inline): Likewise.
7895 (tree_function_versioning): Likewise.
7896 * tree-into-ssa.c (dump_defs_stack): Likewise.
7897 (dump_currdefs): Likewise.
7898 (dump_names_replaced_by): Likewise.
7899 (dump_update_ssa): Likewise.
7900 (update_ssa): Likewise.
7901 * tree-object-size.c (pass_object_sizes::execute): Likewise.
7902 * tree-parloops.c (build_new_reduction): Likewise.
7903 (try_create_reduction_list): Likewise.
7904 (ref_conflicts_with_region): Likewise.
7905 (oacc_entry_exit_ok_1): Likewise.
7906 (oacc_entry_exit_single_gang): Likewise.
7907 * tree-pretty-print.h: Likewise.
7908 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
7909 (get_scalar_evolution): Likewise.
7910 (add_to_evolution): Likewise.
7911 (get_loop_exit_condition): Likewise.
7912 (analyze_evolution_in_loop): Likewise.
7913 (analyze_initial_condition): Likewise.
7914 (analyze_scalar_evolution): Likewise.
7915 (instantiate_scev): Likewise.
7916 (number_of_latch_executions): Likewise.
7917 (gather_chrec_stats): Likewise.
7918 (final_value_replacement_loop): Likewise.
7919 (scev_const_prop): Likewise.
7920 * tree-sra.c (dump_access): Likewise.
7921 (disqualify_candidate): Likewise.
7922 (create_access): Likewise.
7924 (maybe_add_sra_candidate): Likewise.
7925 (create_access_replacement): Likewise.
7926 (analyze_access_subtree): Likewise.
7927 (analyze_all_variable_accesses): Likewise.
7928 (sra_modify_assign): Likewise.
7929 (initialize_constant_pool_replacements): Likewise.
7930 (find_param_candidates): Likewise.
7931 (decide_one_param_reduction): Likewise.
7932 (replace_removed_params_ssa_names): Likewise.
7933 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
7934 * tree-ssa-copy.c (dump_copy_of): Likewise.
7935 (copy_prop_visit_cond_stmt): Likewise.
7936 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
7937 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
7938 (record_equivalences_from_stmt): Likewise.
7939 * tree-ssa-dse.c (compute_trims): Likewise.
7940 (delete_dead_call): Likewise.
7941 (delete_dead_assignment): Likewise.
7942 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
7943 (forward_propagate_into_cond): Likewise.
7944 (pass_forwprop::execute): Likewise.
7945 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7946 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
7948 (move_computations_worker): Likewise.
7949 (execute_sm): Likewise.
7950 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
7951 (remove_exits_and_undefined_stmts): Likewise.
7952 (remove_redundant_iv_tests): Likewise.
7953 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
7954 (adjust_iv_update_pos): Likewise.
7955 * tree-ssa-math-opts.c (bswap_replace): Likewise.
7956 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
7957 (value_replacement): Likewise.
7958 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
7959 * tree-ssa-pre.c (print_pre_expr): Likewise.
7960 (get_representative_for): Likewise.
7961 (create_expression_by_pieces): Likewise.
7962 (insert_into_preds_of_block): Likewise.
7963 (eliminate_insert): Likewise.
7964 (eliminate_dom_walker::before_dom_children): Likewise.
7965 (eliminate): Likewise.
7966 (remove_dead_inserted_code): Likewise.
7967 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
7968 * tree-ssa-reassoc.c (get_rank): Likewise.
7969 (eliminate_duplicate_pair): Likewise.
7970 (eliminate_plus_minus_pair): Likewise.
7971 (eliminate_not_pairs): Likewise.
7972 (undistribute_ops_list): Likewise.
7973 (eliminate_redundant_comparison): Likewise.
7974 (update_range_test): Likewise.
7975 (optimize_range_tests_var_bound): Likewise.
7976 (optimize_vec_cond_expr): Likewise.
7977 (rewrite_expr_tree): Likewise.
7978 (rewrite_expr_tree_parallel): Likewise.
7979 (linearize_expr): Likewise.
7980 (break_up_subtract): Likewise.
7981 (linearize_expr_tree): Likewise.
7982 (attempt_builtin_powi): Likewise.
7983 (attempt_builtin_copysign): Likewise.
7984 (transform_stmt_to_copy): Likewise.
7985 (transform_stmt_to_multiply): Likewise.
7986 (dump_ops_vector): Likewise.
7987 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
7988 (print_scc): Likewise.
7989 (set_ssa_val_to): Likewise.
7990 (visit_reference_op_store): Likewise.
7991 (visit_use): Likewise.
7992 (sccvn_dom_walker::before_dom_children): Likewise.
7993 (run_scc_vn): Likewise.
7994 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7996 (expr_hash_elt::print): Likewise.
7997 (const_and_copies::pop_to_marker): Likewise.
7998 (const_and_copies::record_const_or_copy_raw): Likewise.
7999 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8000 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8001 (dump_predicates): Likewise.
8002 (find_uninit_use): Likewise.
8003 (warn_uninitialized_phi): Likewise.
8004 (pass_late_warn_uninitialized::execute): Likewise.
8005 * tree-ssa.c (verify_vssa): Likewise.
8006 (verify_ssa): Likewise.
8007 (maybe_optimize_var): Likewise.
8008 * tree-vrp.c (dump_value_range): Likewise.
8009 (dump_all_value_ranges): Likewise.
8010 (dump_asserts_for): Likewise.
8011 (register_edge_assert_for_2): Likewise.
8012 (vrp_visit_cond_stmt): Likewise.
8013 (vrp_visit_switch_stmt): Likewise.
8014 (vrp_visit_stmt): Likewise.
8015 (vrp_visit_phi_node): Likewise.
8016 (simplify_cond_using_ranges_1): Likewise.
8017 (fold_predicate_in): Likewise.
8018 (evrp_dom_walker::before_dom_children): Likewise.
8019 (evrp_dom_walker::push_value_range): Likewise.
8020 (evrp_dom_walker::pop_value_range): Likewise.
8021 (execute_early_vrp): Likewise.
8023 2017-05-16 Richard Biener <rguenther@suse.de>
8025 * dwarf2out.c (loc_list_from_tree_1): Do not create
8026 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8028 2017-05-16 Richard Biener <rguenther@suse.de>
8030 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8032 (note_variable_value_in_expr): If we resolved the decl ref
8033 do not push to the stack.
8035 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8037 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8038 operations in fast-math mode.
8039 (vaddq_f16): Likewise.
8040 (vmul_f16): Likewise.
8041 (vmulq_f16): Likewise.
8042 (vsub_f16): Likewise.
8043 (vsubq_f16): Likewise.
8044 * config/arm/neon.md (add<mode>3): New.
8046 (fma:<VH:mode>3): New. Also remove outdated comment.
8049 2017-05-16 Martin Liska <mliska@suse.cz>
8053 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8054 (odr_types_equivalent_p): Likewise.
8056 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8058 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8060 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8063 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8064 non-interunit SSE move alternatives with '?'.
8065 (zero-extendsidi peephole2): New peephole to skip intermediate
8066 general register in SSE zero-extend sequence.
8068 2017-05-15 Jeff Law <law@redhat.com>
8070 * reorg.c (relax_delay_slots): Create a new variable to hold
8071 the temporary target rather than clobbering TARGET_LABEL.
8073 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8074 missing argument to extract_bit_field call.
8075 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8077 2017-05-15 Martin Liska <mliska@suse.cz>
8080 * gcc.c (process_command): Do not allow empty argument of -o option.
8082 2017-05-15 Renlin Li <renlin.li@arm.com>
8084 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8085 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8086 * config/aarch64/constraints.md (Usf): Add long call check.
8087 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8088 (call_value): Likewise.
8089 (sibcall): Likewise.
8090 (sibcall_value): Likewise.
8092 (call_value_insn): New.
8093 (sibcall_insn): Update rtx pattern.
8094 (sibcall_value_insn): Likewise.
8095 (call_internal): Remove.
8096 (call_value_internal): Likewise.
8097 (sibcall_internal): Likewise.
8098 (sibcall_value_internal): Likewise.
8099 (call_reg): Likewise.
8100 (call_symbol): Likewise.
8101 (call_value_reg): Likewise.
8102 (call_value_symbol): Likewise.
8104 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8107 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8109 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8111 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8112 compatible with CCGOCmode and with CCZmode.
8114 2017-05-14 Martin Sebor <msebor@redhat.com>
8117 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8118 (gimple_fold_builtin_snprintf): Same.
8119 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8120 (gimple_fold_builtin_snprintf): Same.
8121 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8123 (is_call_safe): New function.
8124 (try_substitute_return_value): Call it.
8125 (try_simplify_call): New function.
8126 (pass_sprintf_length::handle_gimple_call): Call it.
8128 2017-05-14 Martin Sebor <msebor@redhat.com>
8131 * builtins.c (expand_builtin_stpncpy): Simplify.
8133 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8135 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8136 * config/i386/i386.h
8137 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8138 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8139 (struct machine_function): Add new members call_ms2sysv,
8140 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8141 (struct machine_frame_state): New fields sp_realigned and
8142 sp_realigned_offset.
8143 * config/i386/i386.c
8144 (enum xlogue_stub): New enum.
8145 (enum xlogue_stub_sets): New enum.
8146 (class xlogue_layout): New class.
8147 (struct ix86_frame): New fields stack_realign_allocate_offset,
8148 stack_realign_offset and outlined_save_offset. Modify comments to
8149 detail stack layout when using out-of-line stubs.
8150 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8151 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8152 -mcall-ms2sysv-xlogues.
8153 (stub_managed_regs): New static variable.
8154 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8155 registers managed by out-of-line stub.
8156 (disable_call_ms2sysv_xlogues): New function.
8157 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8158 m->call_ms2sysv when appropriate and compute frame layout for
8160 (sp_valid_at, fp_valid_at): New inline functions.
8161 (choose_basereg): New function.
8162 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8164 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8165 Use align parameter of choose_baseaddr to generated aligned SSE movs
8167 (pro_epilogue_adjust_stack): Modify to track
8168 machine_frame_state::sp_realigned.
8169 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8170 (ix86_nsaved_sseregs): Likewise.
8171 (ix86_emit_save_regs): Likewise.
8172 (ix86_emit_save_regs_using_mov): Likewise.
8173 (ix86_emit_save_sse_regs_using_mov): Likewise.
8174 (get_scratch_register_on_entry): Likewise.
8175 (gen_frame_set): New function.
8176 (gen_frame_load): Likewise.
8177 (gen_frame_store): Likewise.
8178 (emit_outlined_ms2sysv_save): Likewise.
8179 (emit_outlined_ms2sysv_restore): Likewise.
8180 (ix86_expand_prologue): Modify stack re-alignment code and call
8181 emit_outlined_ms2sysv_save when appropriate.
8182 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8183 parameter rtx_insn *insn, which allows the function to be used to only
8185 (ix86_expand_epilogue): Modify validity checks of frame and stack
8186 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8187 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8188 * config/i386/predicates.md
8189 (save_multiple): New predicate.
8190 (restore_multiple): Likewise.
8191 * config/i386/sse.md
8192 (save_multiple<mode>): New pattern.
8193 (save_multiple_realign<mode>): Likewise.
8194 (restore_multiple<mode>): Likewise.
8195 (restore_multiple_and_return<mode>): Likewise.
8196 (restore_multiple_leave_return<mode>): Likewise.
8197 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8199 2017-05-14 Julia Koval <julia.koval@intel.com>
8201 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8202 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8203 (__builtin_ia32_xsetbv): New builtins.
8204 * config/i386/i386.c (ix86_expand_special_args_builtin):
8206 (ix86_expand_builtin): Special expand for new intrinsics.
8207 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8208 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8209 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8211 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8213 * cfganal.c (inverted_post_order_compute): Change argument type
8215 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8216 * df-core.c (rest_of_handle_df_initialize): Adjust.
8217 (rest_of_handle_df_finish): Likewise.
8218 (df_analyze_1): Likewise.
8219 (df_analyze): Likewise.
8220 (loop_inverted_post_order_compute): Change argument to be a vec *.
8221 (df_analyze_loop): Adjust.
8222 (df_get_n_blocks): Likewise.
8223 (df_get_postorder): Likewise.
8224 * df.h (struct df_d): Change field to be a vec.
8225 * lcm.c (compute_laterin): Adjust.
8226 (compute_available): Likewise.
8227 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8228 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8229 * tree-ssa-pre.c (compute_antic): Likewise.
8231 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8233 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8234 (depth_first_search::depth_first_search): Change structure init
8235 function to this constructor.
8236 (depth_first_search::add_bb): Rename function to this member.
8237 (depth_first_search::execute): Likewise.
8238 (flow_dfs_compute_reverse_finish): Adjust.
8240 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8242 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8243 (longest_simple_path): Likewise.
8244 * shrink-wrap.c (spread_components): Likewise.
8245 (disqualify_problematic_components): Likewise.
8246 (emit_common_heads_for_components): Likewise.
8247 (emit_common_tails_for_components): Likewise.
8248 (insert_prologue_epilogue_for_components): Likewise.
8250 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8252 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8255 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8257 * df-core.c (df_set_blocks): Start using auto_bitmap.
8258 (df_compact_blocks): Likewise.
8259 * df-problems.c (df_rd_confluence_n): Likewise.
8260 * df-scan.c (df_insn_rescan_all): Likewise.
8261 (df_process_deferred_rescans): Likewise.
8262 (df_update_entry_block_defs): Likewise.
8263 (df_update_exit_block_uses): Likewise.
8264 (df_entry_block_bitmap_verify): Likewise.
8265 (df_exit_block_bitmap_verify): Likewise.
8266 (df_scan_verify): Likewise.
8267 * lra-constraints.c (lra_constraints): Likewise.
8268 (undo_optional_reloads): Likewise.
8269 (lra_undo_inheritance): Likewise.
8270 * lra-remat.c (calculate_gen_cands): Likewise.
8271 (do_remat): Likewise.
8272 * lra-spills.c (assign_spill_hard_regs): Likewise.
8273 (spill_pseudos): Likewise.
8274 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8275 (bitmap_set_subtract_values): Likewise.
8277 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8279 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8280 management with auto_bitmap.
8281 (fix_inter_tick): Likewise.
8282 (fix_recovery_deps): Likewise.
8283 * ira.c (add_store_equivs): Likewise.
8284 (find_moveable_pseudos): Likewise.
8285 (split_live_ranges_for_shrink_wrap): Likewise.
8286 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8287 (rtx_reuse_manager::seen_def_p): Likewise.
8288 (rtx_reuse_manager::set_seen_def): Likewise.
8289 * print-rtl.h (class rtx_reuse_manager): Likewise.
8291 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8293 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8295 (migrate_btr_def): Likewise.
8296 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8297 * df-core.c (loop_post_order_compute): Likewise.
8298 (loop_inverted_post_order_compute): Likewise.
8299 * hsa-common.h: Likewise.
8300 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8301 * init-regs.c (initialize_uninitialized_regs): Likewise.
8302 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8303 (inline_small_functions): Likewise.
8304 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8305 * ira.c (combine_and_move_insns): Likewise.
8306 (build_insn_chain): Likewise.
8307 * loop-invariant.c (find_invariants): Likewise.
8308 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8309 * predict.c (tree_predict_by_opcode): Likewise.
8310 (predict_paths_leading_to): Likewise.
8311 (predict_paths_leading_to_edge): Likewise.
8312 (estimate_loops_at_level): Likewise.
8313 (estimate_loops): Likewise.
8314 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8315 (spread_components): Likewise.
8316 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8317 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8318 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8319 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8320 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8321 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8322 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8323 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8324 (mark_threaded_blocks): Likewise.
8325 (thread_through_all_blocks): Likewise.
8326 * tree-ssa.c (verify_ssa): Likewise.
8327 (execute_update_addresses_taken): Likewise.
8328 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8330 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8332 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8334 (post_order_compute): Likewise.
8335 (inverted_post_order_compute): Likewise.
8336 (pre_and_rev_post_order_compute_fn): Likewise.
8338 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8340 * genrecog.c (int_set::int_set): Explicitly construct our
8341 auto_vec base class.
8342 * vec.h (auto_vec::auto_vec): New constructor.
8344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8346 * bitmap.h (class auto_bitmap): New constructor taking
8347 bitmap_obstack * argument.
8349 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8351 * bitmap.h (class auto_bitmap): Change type of m_bits to
8352 bitmap_head, and adjust ctor / dtor and member operators.
8354 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8356 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8357 when returned register mode doesn't match original mode.
8359 2017-05-12 Jeff Law <law@redhat.com>
8360 Jakub Jelinek <jakub@redhat.com>
8362 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8363 we look for cc setter after the compare-elim changes.
8364 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8365 within the vector to match what compare-elim now expects.
8366 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8367 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8369 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8370 after the compare-elim changes.
8371 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8372 the vector to match what compare-elim now expects.
8373 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8374 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8375 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8376 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8377 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8379 * config/visium/visium.c (single_set_and_flags): Fix where
8380 we look for cc setter after the compare-elim changes.
8381 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8382 with the vector to match what compare-elim now expects.
8383 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8384 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8385 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8386 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8387 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8388 (neg<mode>2_insn_set_overflow): Likewise.
8390 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8393 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8394 maybe_expand_insn call, set ops[0].target. If still set after call,
8395 set alt_rtl. Add extra arg to recursive calls.
8396 (extract_bit_field): Add alt_rtl argument. Pass to
8398 * expmed.h (extract_bit_field): Fix prototype.
8399 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8400 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8401 to extract_bit_field_calls.
8402 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8403 Pass alt_rtl to extract_bit_field calls.
8404 * calls.c (store_unaligned_arguments_into_psuedos)
8405 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8406 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8408 * optabs.h (struct expand_operand): Add target bitfield.
8410 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8412 * compare-elim.c (try_eliminate_compare): Canonicalize
8413 operation with embedded compare to
8414 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8415 (set (reg) (operation)].
8417 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8419 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8422 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8423 cost of adding a carry flag for ADC instruction.
8424 [case MINUS]: Ignore the cost of subtracting a carry flag
8425 for SBB instruction.
8427 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8429 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8431 * config/rs6000/bmiintrin.h: New file.
8432 * config/rs6000/bmi2intrin.h: New file.
8433 * config/rs6000/x86intrin.h: New file.
8435 2017-05-12 Jeff Law <law@redhat.com>
8437 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8440 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8443 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8445 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8447 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8448 early expansion of vector divide builtins.
8449 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8450 builtins identified as having unsigned arguments.
8452 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8454 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8455 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8456 expansion of vector logical operations (and, andc, or, xor,
8459 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8461 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8462 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8464 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8466 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8467 early GIMPLE expansion of vector multiplies.
8469 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8471 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8472 TARGET_HAVE_MOVT conditional.
8473 (movt splitter): Likewise.
8475 2017-05-12 Richard Biener <rguenther@suse.de>
8477 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8478 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8479 Fold all stmts not inplace.
8481 2017-05-12 Richard Biener <rguenther@suse.de>
8483 PR tree-optimization/80713
8484 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8485 inserted_exprs bit for not removed stmts.
8487 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8490 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8491 parallelized" attribute for parallelized OpenACC kernels.
8492 * omp-offload.c (execute_oacc_device_lower): Use it.
8494 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8495 Set "oacc kernels" attribute.
8496 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8497 parameter. Adjust all users.
8498 (oacc_fn_attrib_kernels_p): Remove function.
8499 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8500 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8501 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8502 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8503 assert "oacc kernels" attribute is set.
8505 2017-05-11 Carl Love <cel@us.ibm.com>
8507 * config/rs6000/rs6000-c: Add support for built-in functions
8508 vector unsigned char vec_popcnt (vector signed char)
8509 vector unsigned char vec_popcnt (vector unsigned char)
8510 vector unsigned short vec_popcnt (vector signed short)
8511 vector unsigned short vec_popcnt (vector unsigned short)
8512 vector unsigned int vec_popcnt (vector signed int)
8513 vector unsigned int vec_popcnt (vector unsigned int)
8514 vector unsigned long long vec_popcnt (vector signed long long)
8515 vector unsigned long long vec_popcnt (vector unsigned long long)
8516 vector signed long long vec_slo (vector signed long long,
8518 vector signed long long vec_slo (vector signed long long,
8519 vector unsigned char)
8520 vector unsigned long long vec_slo (vector unsigned long long,
8522 vector unsigned long long vec_slo (vector unsigned long long,
8523 vector unsigned char)
8524 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8525 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8526 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8527 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
8528 * doc/extend.texi: Update the built-in documentation file for the
8529 new built-in functions.
8531 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8533 * attribs.h (sorted_attr_string): Move machine independent
8534 functions for target clone support from the i386 port to common
8535 code. Rename ix86_function_versions to common_function_versions.
8536 Rename make_name to make_unique_name.
8537 (common_function_versions): Likewise.
8538 (make_unique_name): Likewise.
8539 (make_dispatcher_decl): Likewise.
8540 (is_function_default_version): Likewise.
8541 * attribs.c (attr_strcmp): Likewise.
8542 (sorted_attr_string): Likewise.
8543 (common_function_versions): Likewise.
8544 (make_unique_name): Likewise.
8545 (make_dispatcher_decl): Likewise.
8546 (is_function_default_version): Likewise.
8547 * config/i386/i386.c (attr_strcmp): Likewise.
8548 (sorted_attr_string): Likewise.
8549 (ix86_function_versions): Likewise.
8550 (make_name): Likewise.
8551 (make_dispatcher_decl): Likewise.
8552 (is_function_default_version): Likewise.
8553 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
8555 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8558 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8559 Account for direct move costs for vec_construct of integer
8562 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
8565 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
8566 (UNSPEC_STX_ATOMIC): Ditto.
8567 (loaddi_via_sse): New insn.
8568 (storedi_via_sse): Ditto.
8569 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
8570 Update corresponding peephole2 patterns.
8571 (atomic_storedi_fpu): Ditto.
8573 2017-05-11 Julia Koval <julia.koval@intel.com>
8575 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
8576 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
8578 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
8579 (__builtin_ia32_rsqrt14ss_mask): New builtins.
8580 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
8582 2017-05-11 Nathan Sidwell <nathan@acm.org>
8584 * graphite-poly.c: Include dumpfile.h.
8586 * dumpfle.h (dump_function): Declare here ...
8587 * tree-dump.h (dump_function): ... not here.
8588 * dumpfile.c: #include tree-cfg.h.
8589 (dump_function): Move here from ...
8590 * tree-dump.c (dump_function): ... here.
8591 * gimplify.c: #include splay-tree.h, not tree-dump.h.
8592 * graphite-poly.c: Don't include tree-dump.h.
8593 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
8594 * print-tree.c: Likewise.
8595 * stor-layout.c: Likewise.
8596 * tree-nested.c: Likewise.
8598 * dumpfile.c (dump_start): Use TDF_FLAGS.
8599 (dump_enable_all): Fix TDF_KIND check thinko.
8601 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8603 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8604 array entries to represent two legal parameterizations of the
8605 overloaded __builtin_cmpb function, as represented by the
8606 P6_OV_BUILTIN_CMPB constant.
8607 (altivec_resolve_overloaded_builtin): Add special case handling
8608 for the __builtin_cmpb function, as represented by the
8609 P6_OV_BUILTIN_CMPB constant.
8610 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
8611 (BU_P6_64BIT_2): New macro.
8612 (BU_P6_OVERLOAD_2): New macro
8613 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
8614 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
8615 (CMPB): Add overload support to represent both 32-bit and 64-bit
8616 compare-bytes function.
8617 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8618 support for TARGET_CMPB.
8619 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
8620 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8621 documentation of the __builtin_cmpb overloaded built-in function.
8623 2017-05-11 Richard Biener <rguenther@suse.de>
8625 PR tree-optimization/80705
8626 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
8627 bases are not vectorizable.
8629 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8631 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
8632 when counting register pressure.
8634 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8636 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
8637 (iv_ca_more_deps): Renamed to ...
8638 (iv_ca_compare_deps): ... this.
8639 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
8641 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8643 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
8645 (determine_group_iv_costs): ... here.
8646 (find_inv_vars_cb): Record inv var if it's not recorded before.
8648 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8650 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
8651 (get_shiftadd_cost): Ditto.
8653 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8655 * tree-ssa-address.c: Include header file.
8656 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
8658 (add_to_parts): Refactor.
8659 (addr_to_parts): New parameter. Update use of move_hint_to_base.
8660 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
8663 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8665 PR tree-optimization/53090
8666 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
8668 (extract_cond_operands): Detect condition with IV on both sides
8669 and return COMP_IV_EXPR_2.
8670 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
8671 (rewrite_use_compare): Simplify by removing call to function
8672 extract_cond_operands.
8674 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8676 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
8677 (extract_cond_operands): Detect condition comparing against non-
8678 invariant bound and return appropriate enum value.
8679 (find_interesting_uses_cond): Update use of extract_cond_operands.
8680 Handle its return value accordingly.
8681 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
8683 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8685 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
8686 nonlinear iv_use computation in loop invariant sensitive way.
8688 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8690 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
8691 (find_iv_candidates): Call relate_compare_use_with_all_cands.
8693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8695 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
8696 (dump_cand): Support iv_cand.inv_exprs.
8697 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
8699 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
8702 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8704 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
8706 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
8707 as local function. Include necessary header files.
8708 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
8710 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8712 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
8714 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8716 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
8717 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
8718 RSHIFT_EXPR and BIT_NOT_EXPR.
8720 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8722 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
8723 (adjust_setup_cost): New parameter supporting round up adjustment.
8724 (struct address_cost_data): Delete.
8725 (force_expr_to_var_cost): Don't bound cost with spill_cost.
8726 (split_address_cost, ptr_difference_cost): Delete.
8727 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
8728 (struct ainc_cost_data): New struct.
8729 (get_address_cost_ainc): New function.
8730 (get_address_cost, get_computation_cost): Reimplement.
8731 (determine_group_iv_cost_address): Record inv_expr for all uses of
8733 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
8734 (iv_ca_has_deps): Reimplemented to ...
8735 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
8737 (iv_ca_extend): Call iv_ca_more_deps.
8739 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8741 * tree-ssa-address.c (struct mem_address): Move to header file.
8742 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
8743 * tree-ssa-address.h (struct mem_address): Move from C file.
8744 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
8746 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8748 * tree-affine.h (aff_combination_type): New interface.
8749 (aff_combination_zero_p): Remove static.
8750 (aff_combination_const_p): New interface.
8751 (aff_combination_singleton_var_p): New interfaces.
8753 2017-05-11 Richard Biener <rguenther@suse.de>
8755 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8756 Skip unreachable blocks and destinations.
8757 (eliminate): Move stmt removal and fixup ...
8758 (fini_eliminate): ... here. Skip inserted exprs.
8759 (pass_pre::execute): Move fini_pre after fini_eliminate.
8760 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
8761 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
8762 PRE to get rid of dead code that has invalid SSA form and
8763 split critical edges again.
8765 2017-05-11 Bin Cheng <bin.cheng@arm.com>
8767 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8769 2017-05-11 Richard Biener <rguenther@suse.de>
8771 * passes.c (execute_function_todo): Verify loops if they are
8772 said to be up-to-date.
8773 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
8774 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
8776 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
8779 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
8780 handle calling assemble_external ourself.
8783 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
8784 modes with zero size. Enhance comment.
8786 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8788 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
8789 built-ins for vec_xl and vec_xst with short and char pointer
8792 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8794 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
8795 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
8796 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
8797 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
8798 (_mm_maskz_min_round_ss): New intrinsics.
8799 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
8800 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
8801 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
8802 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
8803 (__builtin_ia32_minss_mask_round): New builtins.
8804 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8805 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8806 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
8808 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
8809 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
8811 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
8814 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
8816 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
8817 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
8818 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
8819 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
8820 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
8821 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
8822 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
8823 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8824 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8825 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
8826 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
8827 (__builtin_ia32_mulss_mask_round): New builtins.
8828 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8829 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8830 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
8832 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
8833 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8835 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8838 2017-05-10 Julia Koval <julia.koval@intel.com>
8840 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
8841 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
8842 (_mm256_setr_m128i): New intrinsics.
8844 2017-05-10 Julia Koval <julia.koval@intel.com>
8846 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
8847 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
8848 (_mm_maskz_rcp14_ss): New intrinsics.
8849 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
8850 (__builtin_ia32_rcp14ss_mask): New builtins.
8851 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
8853 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
8855 PR tree-optimization/51513
8856 * tree-cfg.c (gimple_seq_unreachable_p): New function.
8857 (assert_unreachable_fallthru_edge_p): Use it.
8858 (group_case_labels_stmt): Likewise.
8859 * tree-cfg.h: Prototype it.
8860 * stmt.c: Include cfghooks.h and tree-cfg.h.
8861 (emit_case_dispatch_table) <gap_label>: New local variable.
8862 Use it to fill dispatch table gaps.
8863 Test for default_label before updating probabilities.
8864 (expand_case) <default_label>: Remove unneeded initialization.
8865 Test for unreachable default case statement and remove its edge.
8866 Set default_label accordingly.
8867 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
8869 2017-05-10 Carl Love <cel@us.ibm.com>
8871 * config/rs6000/rs6000-c: Add support for built-in functions
8872 vector signed char vec_neg (vector signed char)
8873 vector signed short int vec_neg (vector short int)
8874 vector signed int vec_neg (vector signed int)
8875 vector signed long long vec_neg (vector signed long long)
8876 vector float vec_neg (vector float)
8877 vector double vec_neg (vector double)
8878 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
8880 * config/rs6000/altivec.h: Add define for vec_neg
8881 * doc/extend.texi: Update the built-in documentation for the
8882 new built-in functions.
8884 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8886 PR tree-optimization/77644
8887 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
8889 2017-05-10 Nathan Sidwell <nathan@acm.org>
8891 * dumpfile.h (TDI_lang_all): New.
8892 (TDF_KIND): New. Renumber others
8893 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
8895 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
8897 (get_dump_file_name): Adjust suffix generation.
8898 (dump_enable_all): Use TDF_KIND.
8899 * doc/invoke.texi (-fdump-lang-all): Document.
8901 * dumpfile.h: Tabify.
8903 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
8906 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
8907 Move member access before delete.
8909 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
8911 * tree-inline.c (expand_call_inline): Split block at stmt
8914 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8917 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
8918 are now unused after splitting mov{sf,sd}_hardfloat.
8919 (f32_lr2): Likewise.
8921 (f32_lm2): Likewise.
8923 (f32_li2): Likewise.
8926 (f32_sr2): Likewise.
8928 (f32_sm2): Likewise.
8930 (f32_si2): Likewise.
8933 (f32_vsx): Likewise.
8935 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
8936 For movsf, order stores so the VSX stores occur before the GPR
8937 store which encourages the register allocator to use a traditional
8938 FPR instead of a GPR. For movsd, order the stores so that the GPR
8939 store comes before the VSX stores to allow the power6 to work.
8940 This is due to the power6 not having a 32-bit integer store
8941 instruction from a FPR.
8942 (movsf_hardfloat): Likewise.
8943 (movsd_hardfloat): Likewise.
8945 2017-05-09 Martin Sebor <msebor@redhat.com>
8947 PR translation/80280
8948 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
8951 PR translation/80280
8952 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
8953 data member added in r247778.
8954 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
8956 2017-05-09 Nathan Sidwell <nathan@acm.org>
8958 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
8960 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
8963 2017-05-09 Marek Polacek <polacek@redhat.com>
8965 * doc/invoke.texi: Fix typo.
8967 2017-05-09 Richard Biener <rguenther@suse.de>
8969 * tree-vrp.c (vrp_val_is_max): Adjust comment.
8970 (vrp_val_is_min): Likewise.
8971 (set_value_range_to_value): Likewise.
8972 (set_value_range_to_nonnegative): Likewise.
8973 (gimple_assign_nonzero_p): Likewise.
8974 (gimple_stmt_nonzero_p): Likewise.
8975 (vrp_int_const_binop): Likewise. Remove unreachable case.
8976 (adjust_range_with_scev): Adjust comments.
8977 (compare_range_with_value): Likewise.
8978 (extract_range_from_phi_node): Likewise.
8979 (test_for_singularity): Likewise.
8981 2017-05-09 Richard Biener <rguenther@suse.de>
8983 * tree-vrp.c (get_single_symbol): Add assert that we don't
8984 get overflowed constants as invariant part.
8985 (compare_values_warnv): Add comment before the TREE_NO_WARNING
8986 checks. Use wi::cmp instead of recursing for integer constants.
8987 (compare_values): Just ignore whether we assumed undefined
8988 overflow instead of failing the compare.
8989 (extract_range_for_var_from_comparison_expr): Add comment before the
8990 TREE_NO_WARNING sets.
8991 (test_for_singularity): Likewise.
8992 (extract_range_from_comparison): Do not disable optimization
8993 when we assumed undefined overflow.
8994 (extract_range_basic): Remove init of unused var.
8996 2017-05-09 Richard Biener <rguenther@suse.de>
8998 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
8999 (extract_range_from_multiplicative_op_1): Adjust.
9000 (extract_range_from_binary_expr_1): Use int_const_binop.
9002 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9005 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9006 rs6000_store_data_bypass_p in seven define_bypass directives and
9007 in several comments.
9008 * config/rs6000/rs6000-protos.h: Add prototype for
9009 rs6000_store_data_bypass_p function.
9010 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9011 function implements slightly different (rs6000-specific) semantics
9012 than store_data_bypass_p, returning false rather than aborting
9013 with assertion error when arguments do not satisfy the
9014 requirements of store data bypass.
9015 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9016 rs6000_store_data_bypass_p.
9018 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9020 * config/xtensa/xtensa-protos.h
9021 (xtensa_initial_elimination_offset): New declaration.
9022 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9023 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9024 macro definition, add case for FRAME_POINTER_REGNUM when
9025 FRAME_GROWS_DOWNWARD.
9026 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9027 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9028 xtensa_initial_elimination_offset.
9030 2017-05-08 Nathan Sidwell <nathan@acm.org>
9032 * doc/invoke.texi: Alphabetize -fdump options.
9034 2017-05-08 Martin Sebor <msebor@redhat.com>
9036 PR translation/80280
9037 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9039 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9041 * target.def (compute_frame_layout): New optional target hook.
9042 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9043 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9044 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9046 * reload1.c (verify_initial_elim_offsets): Likewise.
9047 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9048 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9049 (arm_get_frame_offsets): Split up into this ...
9050 (arm_compute_frame_layout): ... and this function.
9052 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9054 * config/aarch64/constraints.md (Usa): New constraint.
9055 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9057 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9059 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9060 with_multilib_list after it has been checked.
9062 2017-05-08 Richard Biener <rguenther@suse.de>
9064 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9065 (bitmap_set_subtract_values): Likewise.
9067 2017-05-08 Richard Biener <rguenther@suse.de>
9069 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9070 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9072 (gimple_stmt_nonzero_warnv_p): Rename to ...
9073 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9075 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9076 (extract_range_basic): Adjust, do not disable propagation on
9077 strict overflow sensitive simplification.
9078 (vrp_visit_cond_stmt): Likewise.
9080 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9082 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9083 body size unconditionally.
9085 2017-05-07 Jeff Law <law@redhat.com>
9088 2017-05-06 Jeff Law <law@redhat.com>
9089 PR tree-optimization/78496
9090 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9093 PR tree-optimization/78496
9094 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9095 (simplify_stmt_using_ranges): Call it.
9096 (vrp_dom_walker::before_dom_children): Extract equivalences
9097 from an ASSERT_EXPR with an equality comparison against a
9100 2017-05-06 Jeff Law <law@redhat.com>
9102 PR tree-optimization/78496
9103 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9106 PR tree-optimization/78496
9107 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9108 (simplify_stmt_using_ranges): Call it.
9109 (vrp_dom_walker::before_dom_children): Extract equivalences
9110 from an ASSERT_EXPR with an equality comparison against a
9113 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9115 * lra-constraints.c (lra_copy_reg_equiv): New function.
9116 (split_reg): Use it to copy equivalence information from the
9117 original register to the spill register.
9119 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9121 PR rtl-optimization/75964
9122 * simplify-rtx.c (simplify_const_relational_operation): Remove
9123 invalid handling of comparisons of integer ABS.
9125 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9127 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9129 (init_regs): Remove declaration.
9130 (function_arg_advance_32): Initialize error_p as boolean variable.
9132 2017-05-05 Nathan Sidwell <nathan@acm.org>
9134 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9135 lines. Use for (;;).
9137 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9139 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9140 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9141 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9142 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9143 VF=2 that require versioning.
9145 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9147 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9150 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9152 * diagnostic.h (diagnostic_override_option_index): Convert from
9153 macro to inline function.
9155 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9157 * diagnostic.c (last_module_changed_p): New function.
9158 (set_last_module): New function.
9159 (diagnostic_report_current_module): Convert macro usage to
9160 the above functions.
9161 * diagnostic.h (diagnostic_context::last_module): Strengthen
9162 from const line_map * to const line_map_ordinary *.
9163 (diagnostic_last_module_changed): Delete macro.
9164 (diagnostic_set_last_module): Delete macro.
9166 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9168 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9169 with diagnostic_report_diagnostic.
9170 (diagnostic_n_impl_richloc): Likewise.
9171 * diagnostic.h (report_diagnostic): Delete macro.
9172 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9173 with diagnostic_report_diagnostic.
9174 * substring-locations.c (format_warning_va): Likewise.
9176 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9178 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9179 save/restor of format_spec. Move option-printing code to...
9180 (print_option_information): ...this new function, and
9181 reimplement by simply printing to the pretty_printer,
9182 rather than appending to the format string.
9184 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9186 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9187 handling logic into...
9188 (update_effective_level_from_pragmas): ...this new function.
9190 2017-05-04 Andrew Waterman <andrew@sifive.com>
9192 * config/riscv/riscv.opt (mstrict-align): New option.
9193 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9194 (SLOW_UNALIGNED_ACCESS): Define.
9195 (riscv_slow_unaligned_access): Declare.
9196 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9198 (riscv_slow_unaligned_access): New variable.
9199 (rocket_tune_info): Set slow_unaligned_access to true.
9200 (optimize_size_tune_info): Set slow_unaligned_access to false.
9201 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9202 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9203 (riscv_option_override): Set riscv_slow_unaligned_access.
9204 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9206 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9208 * config/riscv/riscv.md: Unify indentation.
9210 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9215 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9217 (extendsi<mode>2): Add support for doing sign extension via
9218 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9219 don't have ISA 3.0 instructions.
9220 (extendsi<mode>2 splitter): Likewise.
9221 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9222 generate the normal insns since SImode can now go in vector
9223 registers. Disallow the special UNSPECs needed for previous
9224 machines to hide SImode being used. Add new insns
9225 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9226 (fix_trunc<mode>si2_stfiwx): Likewise.
9227 (fix_trunc<mode>si2_internal): Likewise.
9228 (fixuns_trunc<mode>si2): Likewise.
9229 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9230 (fctiw<u>z_<mode>_smallint): Likewise.
9231 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9232 of floating point to 32-bit integer from doing a direct move to
9233 the GPR registers to do a store.
9234 (fctiwz_<mode>): Break long line.
9236 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9238 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9239 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9240 (addr_list, addr_offset_valid_p): New.
9241 (split_address_groups): Check offset validity with above function.
9242 (gt-tree-ssa-loop-ivopts.h): Include header file.
9244 2017-05-05 Nathan Sidwell <nathan@acm.org>
9246 * config.gcc (arm*-*-*): Add missing 'fi'.
9248 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9250 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9251 included in -fopt-info does not matter.
9252 * doc/optinfo.texi (-fopt-info): Fix description of default
9253 behavour. Explicitly say order of options included in -fopt-info
9256 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9258 * config.gcc: Allow combinations of aprofile and rmprofile values for
9259 --with-multilib-list.
9260 * config/arm/t-multilib: New file.
9261 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9262 variables. Remove setting of ISA and floating-point ABI in
9263 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9264 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9265 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9266 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9268 * config/arm/t-rmprofile: Likewise except for the matches changes.
9269 * doc/install.texi (--with-multilib-list): Document the combination of
9270 aprofile and rmprofile values and warn about pitfalls in doing that.
9272 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9274 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9275 (movdi_aarch64): Likewise.
9277 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9279 PR tree-optimization/80632
9280 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9282 (build_arrays): Initialize it for virtual phis.
9283 (fix_phi_nodes): Use it for virtual phis.
9285 PR tree-optimization/80558
9286 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9287 [x, y] op z into [x op, y op z] for op & or | if conditions
9290 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9291 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9294 * config/arm/arm.md (use_literal_pool): Remove.
9295 (64-bit immediate split): No longer takes cost into consideration
9296 if arm_disable_literal_pool is enabled.
9297 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9298 used when arm_disable_literal_pool is enabled.
9299 (arm_max_const_double_inline_cost): Remove use of
9300 arm_disable_literal_pool.
9301 (push_minipool_fix): Add assert.
9302 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9303 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9304 (no_literal_pool_sf_immediate): New.
9306 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9308 PR tree-optimization/80613
9309 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9310 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9312 2017-05-05 Richard Biener <rguenther@suse.de>
9314 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9316 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9318 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9319 of this flag from insn conditions due to removal from r247495.
9321 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9323 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9325 (arm_early_store_addr_dep_ptr): Likewise.
9326 * config/arm/aarch-common-protos.h
9327 (arm_early_load_addr_dep_ptr): Add prototype.
9328 (arm_early_store_addr_dep_ptr): Likewise.
9329 * config/arm/cortex-a53.md: Add new bypasses.
9331 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9333 * tree.c (next_type_uid): Change type to unsigned.
9334 (type_hash_canon): Decrement back next_type_uid if
9335 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9336 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9339 2017-05-04 Martin Sebor <msebor@redhat.com>
9341 * builtins.c: Fix a trivial typo in a comment.
9344 * builtins.c (check_sizes): Adjust to handle reading past the end.
9345 Avoid printing excessive upper bound of ranges. Use %E to print
9346 tree nodes instead of converting them to %wu.
9347 (expand_builtin_memchr): New function.
9348 (compute_dest_size): Rename...
9349 (compute_objsize): ...to this.
9350 (expand_builtin_memcpy): Adjust.
9351 (expand_builtin_mempcpy): Adjust.
9352 (expand_builtin_strcat): Adjust.
9353 (expand_builtin_strcpy): Adjust.
9354 (check_strncat_sizes): Adjust.
9355 (expand_builtin_strncat): Adjust.
9356 (expand_builtin_strncpy): Adjust and simplify.
9357 (expand_builtin_memset): Adjust.
9358 (expand_builtin_bzero): Adjust.
9359 (expand_builtin_memcmp): Adjust.
9360 (expand_builtin): Handle memcmp.
9361 (maybe_emit_chk_warning): Check strncat just once.
9363 2017-05-04 Martin Sebor <msebor@redhat.com>
9365 PR preprocessor/79214
9368 * builtins.c (check_sizes): Add inlining context and issue
9369 warnings even when -Wno-system-headers is set.
9370 (check_strncat_sizes): Same.
9371 (expand_builtin_strncat): Same.
9372 (expand_builtin_memmove): New function.
9373 (expand_builtin_stpncpy): Same.
9374 (expand_builtin): Handle memmove and stpncpy.
9376 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9378 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9379 which is not used any more.
9381 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9383 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9385 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9387 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9388 (cortexa53_tunings): Likewise.
9389 (cortexa57_tunings): Likewise.
9390 (cortexa72_tunings): Likewise.
9391 (cortexa73_tunings): Likewise.
9393 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9395 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9396 Set loop alignment to 8.
9398 2017-05-04 Martin Sebor <msebor@redhat.com>
9400 PR translation/80280
9401 * builtins.c (expand_builtin_object_size): Add missing quoting to
9402 %D and like directives.
9403 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9404 (hsa_type_for_tree_type): Same.
9405 (verify_function_arguments): Same.
9406 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9407 * varasm.c (get_section): Same.
9410 2017-05-04 Martin Sebor <msebor@redhat.com>
9412 PR translation/80280
9413 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9415 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9417 * config/aarch64/aarch64.c (generic_addrcost_table):
9418 Change HI/TI mode setting.
9420 2017-05-04 Martin Jambor <mjambor@suse.cz>
9422 PR tree-optimization/80622
9423 * tree-sra.c (comes_initialized_p): New function.
9424 (build_accesses_from_assign): Only set write lazily when
9425 comes_initialized_p is false.
9426 (analyze_access_subtree): Use comes_initialized_p.
9427 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9428 instead of testing for PARM_DECL.
9430 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9432 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9433 constraint on operand 0 to allow more general addressing modes.
9434 Adjust output template.
9435 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9437 * config/aarch64/aarch64-protos.h
9438 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9439 * config/aarch64/constraints.md (Dp): New address constraint.
9440 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9443 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9445 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9446 update use of estimate_ipcp_clone_size_and_time.
9447 (estimate_local_effects): Update use of
9448 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9449 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9450 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9451 Return nonspecialized time.
9453 2017-05-04 Richard Biener <rguenther@suse.de>
9455 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9456 for the last VUSE which def dominates the PHI. Directly call
9458 (get_continuation_for_phi_1): Remove.
9460 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9462 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9463 to explain the use of truncating division. Cap the number of
9464 iterations to the maximum given by nb_iterations_upper_bound,
9467 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9469 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9470 * configure: Regenerate.
9471 * config.in: Regenerate.
9472 * config/i386/driver-mingw32.c: new file.
9473 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9474 * config.host: Link driver-mingw32.o on MinGW host.
9475 * doc/install.texi: Document new --enable-mingw-wildcard configure
9478 2017-05-04 Marek Polacek <polacek@redhat.com>
9480 PR tree-optimization/80612
9481 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9483 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9484 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9486 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9487 (movt splitter): Likewise.
9488 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9489 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9490 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9491 block for Thumb-1 with MOVT.
9492 (thumb2_legitimate_address_p): Move code block ...
9493 (can_avoid_literal_pool_for_label_p): ... into this new function.
9494 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9496 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9497 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9498 "M-profile targets with the MOVT instruction".
9500 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9502 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9503 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9504 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9506 2017-05-04 Martin Liska <mliska@suse.cz>
9508 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9511 2017-05-04 Richard Biener <rguenther@suse.de>
9513 * tree.c (array_at_struct_end_p): Handle arrays at struct
9514 end with flexarrays more conservatively. Refactor and treat
9515 arrays of arrays or aggregates more strict. Fix
9516 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9517 * tree.c (array_at_struct_end_p): Adjust prototype.
9518 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9519 * gimple-fold.c (get_range_strlen): Likewise.
9520 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9522 2017-05-04 Richard Biener <rguenther@suse.de>
9524 PR tree-optimization/31130
9525 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9527 (supports_overflow_infinity): Likewise.
9528 (is_negative_overflow_infinity): Likewise.
9529 (is_positive_overflow_infinity): Likewise.
9530 (is_overflow_infinity): Likewise.
9531 (stmt_overflow_infinity): Likewise.
9532 (overflow_infinity_range_p): Likewise.
9533 (usable_range_p): Remove as always returning true.
9534 (make_overflow_infinity): Remove.
9535 (negative_overflow_infinity): Likewise.
9536 (positive_overflow_infinity): Likewise.
9537 (avoid_overflow_infinity): Likewise.
9538 (set_value_range): Adjust accordingly.
9539 (set_value_range_to_nonnegative): Likewise, remove now unused
9540 overflow_infinity arg.
9541 (vrp_operand_equal_p): Adjust.
9542 (update_value_range): Likewise.
9543 (range_int_cst_singleton_p): Likewise.
9544 (operand_less_p): Likewise.
9545 (compare_values_warnv): Likewise.
9546 (extract_range_for_var_from_comparison_expr): Likewise.
9547 (vrp_int_const_binop): Likewise.
9548 (zero_nonzero_bits_from_vr): Likewise.
9549 (extract_range_from_multiplicative_op_1): Likewise.
9550 (extract_range_from_binary_expr_1): Likewise.
9551 (extract_range_from_unary_expr): Likewise.
9552 (extract_range_from_comparison): Likewise.
9553 (extract_range_basic): Likewise.
9554 (adjust_range_with_scev): Likewise.
9555 (compare_ranges): Likewise.
9556 (compare_range_with_value): Likewise.
9557 (dump_value_range): Likewise.
9558 (test_for_singularity): Likewise, remove strict_overflow_p parameter
9560 (simplify_cond_using_ranges): Adjust.
9562 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
9564 * brig-builtins.def: Added a builtin for class_f64.
9565 * builtin-types.def: Added a builtin type needed by class_f64.
9567 2017-05-03 Jason Merrill <jason@redhat.com>
9569 * timevar.def: Add TV_CONSTEXPR.
9571 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9573 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
9575 2017-05-03 Martin Jambor <mjambor@suse.cz>
9577 * ipa-prop.c (ipa_update_after_lto_read): Removed.
9578 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
9579 * ipa-cp.c (ipcp_propagate_stage): Do not call
9580 ipa_update_after_lto_read.
9581 * ipa-inline.c (ipa_inline): Likewise.
9583 2017-05-03 Martin Jambor <mjambor@suse.cz>
9585 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
9586 tag. Added a default constructor and a destructor.
9587 (ipa_edge_args_sum_t): New class;
9588 (ipa_edge_args_sum): Declare.
9589 (ipa_edge_args_vector): Remove declaration.
9590 (IPA_EDGE_REF): Use ipa_edge_args_sum.
9591 (ipa_free_edge_args_substructures): Remove declaration.
9592 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
9593 (ipa_edge_args_info_available_for_edge_p): Likewise.
9594 * ipa-prop.c (ipa_edge_args_vector): Removed.
9595 (edge_removal_hook_holder): Likewise.
9596 (edge_duplication_hook_holder): Likewise.
9597 (ipa_edge_args_sum): New variable.
9598 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
9599 ipa_edge_args_vector.
9600 (ipa_free_edge_args_substructures): Likewise.
9601 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
9602 ipa_edge_args_vector.
9603 (ipa_edge_removal_hook): Turned into method
9604 ipa_edge_args_sum_t::remove.
9605 (ipa_edge_duplication_hook): Turned into method
9606 ipa_edge_args_sum_t::duplicate.
9607 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
9608 registering edge hooks.
9609 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
9610 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
9611 ipa_edge_args_sum instead of ipa_edge_args_vector.
9612 * ipa-profile.c (ipa_profile): Likewise.
9614 2017-05-03 Martin Jambor <mjambor@suse.cz>
9616 * symbol-summary.h (function_summary): New method exists.
9617 (function_summary::symtab_removal): Deallocate through release.
9618 (call_summary): New class.
9619 (gt_ggc_mx): New overload.
9620 (gt_pch_nx): Likewise.
9621 (gt_pch_nx): Likewise.
9623 2017-05-03 Jeff Law <law@redhat.com>
9625 PR tree-optimization/78496
9626 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
9627 from simplify_cond_using_ranges. Split off code to walk
9628 backwards through casts into ...
9629 (simplify_cond_using_ranges_2): New function.
9630 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
9631 (execute_vrp): After identifying jump threads, call
9632 simplify_cond_using_ranges_2.
9634 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
9637 * ipa-inline.h (inline_summary): Add ctor.
9638 (create_ggc): Do not use ggc_cleared_alloc.
9640 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
9641 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9643 * gcc.c (handle_braces): Support escaping in switch matching
9645 * doc/invoke.texi (Spec Files): Document it.
9646 Remove superfluous @code markup in items.
9648 2017-05-03 David Malcolm <dmalcolm@redhat.com>
9650 * diagnostic-show-locus.c (struct column_range): New struct.
9651 (get_affected_columns): New function.
9652 (get_printed_columns): New function.
9653 (struct correction): New struct.
9654 (correction::ensure_capacity): New function.
9655 (correction::ensure_terminated): New function.
9656 (struct line_corrections): New struct.
9657 (line_corrections::~line_corrections): New dtor.
9658 (line_corrections::add_hint): New function.
9659 (layout::print_trailing_fixits): Reimplement in terms of the new
9661 (selftest::test_overlapped_fixit_printing): New function.
9662 (selftest::diagnostic_show_locus_c_tests): Call it.
9664 2017-05-03 Nathan Sidwell <nathan@acm.org>
9666 Canonicalize canonical type hashing
9667 * tree.h (type_hash_canon_hash): Declare.
9668 * tree.c (type_hash_list, attribute_hash_list): Move into
9669 type_hash_canon_hash.
9670 (build_type_attribute_qual_variant): Break out hash code calc into
9671 type_hash_canon_hash.
9672 (type_hash_canon_hash): New. Generic type hash computation.
9673 (build_range_type_1, build_array_type_1, build_function_type,
9674 build_method_type_directly, build_offset_type, build_complex_type,
9675 make_vector_type): Call it.
9677 2017-05-03 Richard Biener <rguenther@suse.de>
9679 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
9680 When all DRs have unknown misaligned do not always peel
9681 when there is a store but apply the same costing model as if
9682 there were only loads.
9684 2017-05-03 Richard Biener <rguenther@suse.de>
9687 PR tree-optimization/80492
9688 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9689 compare_base_decls returning dont-know properly.
9691 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
9693 * config/arm/iterators.md (CCSI): New mode iterator.
9694 (arch): New mode attribute.
9695 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
9696 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
9697 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
9698 code iterator for success result mode.
9699 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
9700 the corresponding new insn generators.
9702 2017-05-03 Bin Cheng <bin.cheng@arm.com>
9705 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9706 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9708 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
9710 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
9711 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
9712 (DDR_A): Wrap DDR argument in brackets.
9713 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
9714 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
9715 (DDR_REVERSED_P): Likewise.
9717 2017-05-03 Jakub Jelinek <jakub@redhat.com>
9719 PR tree-optimization/79472
9720 * tree-switch-conversion.c (struct switch_conv_info): Add
9721 contiguous_range and default_case_nonstandard fields.
9722 (collect_switch_conv_info): Compute contiguous_range and
9723 default_case_nonstandard fields, don't clear final_bb if
9724 contiguous_range and only the default case doesn't have the required
9726 (check_all_empty_except_final): Set default_case_nonstandard instead
9727 of failing if contiguous_range and the default case doesn't have empty
9729 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
9730 and only the default case doesn't have the required constants. Skip
9732 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
9733 if default_case_nonstandard.
9734 (build_constructors): Build constant 1 just once. Assert that default
9735 values aren't inserted in between cases if contiguous_range. Skip
9737 (build_arrays): Skip virtual phis.
9738 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
9739 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
9740 Handle virtual phis.
9741 (gen_inbound_check): Handle default_case_nonstandard case.
9742 (process_switch): Adjust check_final_bb caller. Call
9743 gather_default_values with the first non-default case instead of
9744 default case if default_case_nonstandard.
9746 2017-05-02 Nathan Sidwell <nathan@acm.org>
9748 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
9749 check. Fix formatting.
9751 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
9753 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
9754 errors when comparing specialized and unspecialized times.
9756 2017-05-02 David Malcolm <dmalcolm@redhat.com>
9758 * diagnostic-show-locus.c
9759 (layout::should_print_annotation_line_p): Make private.
9760 (layout::print_annotation_line): Make private.
9761 (layout::annotation_line_showed_range_p): Make private.
9762 (layout::show_ruler): Make private.
9763 (layout::print_source_line): Make private. Pass in line and
9764 line_width, rather than calling location_get_source_line. Drop
9766 (layout::print_leading_fixits): New method.
9767 (layout::print_any_fixits): Rename to...
9768 (layout::print_trailing_fixits): ...this, and make private.
9769 Don't print newline fixits.
9770 (diagnostic_show_locus): Move logic for printing one row into...
9771 (layout::print_line): ...this new function. Move the
9772 location_get_source_line call and error-handling from
9773 print_source_line to here. Call print_leading_fixits, and rename
9774 print_any_fixits to print_trailing_fixits.
9775 (selftest::test_fixit_insert_containing_newline): Update now that
9776 newlines are partially supported.
9777 (selftest::test_fixit_insert_containing_newline_2): New test.
9778 (selftest::test_fixit_replace_containing_newline): Update comments.
9779 (selftest::diagnostic_show_locus_c_tests): Call the new test.
9780 * edit-context.c (class added_line): New class.
9781 (class edited_line): Describe newline handling in comment.
9782 (edited_line::actually_edited_p): New method.
9783 (edited_line::print_content): Delete redundant decl.
9784 (edited_line::m_predecessors): New field.
9785 (edited_file::print_content): Call edited_line::print_content.
9786 (edited_file::print_diff): Update to support newlines.
9787 (edited_file::print_diff_hunk): Likewise.
9788 (edited_file::print_run_of_changed_lines): New function.
9789 (edited_file::print_diff_line): Convert to...
9790 (print_diff_line): ...this.
9791 (edited_file::get_effective_line_count): New function.
9792 (edited_line::edited_line): Initialize new field m_predecessors.
9793 (edited_line::~edited_line): Clean up m_predecessors.
9794 (edited_line::apply_fixit): Handle newlines.
9795 (edited_line::get_effective_line_count): New function.
9796 (edited_line::print_content): New function.
9797 (edited_line::print_diff_lines): New function.
9798 (selftest::test_applying_fixits_insert_containing_newline): New
9800 (selftest::test_applying_fixits_replace_containing_newline): New
9802 (selftest::insert_line): New function.
9803 (selftest::test_applying_fixits_multiple_lines): Add example of
9805 (selftest::edit_context_c_tests): Call the new tests.
9807 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9809 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
9810 parameter cand. Update dump information.
9811 (get_computation_cost): Update uses.
9813 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9815 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
9816 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
9817 (get_computation_at, rewrite_use_address): Update use of
9818 get_computation_aff.
9820 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9822 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
9823 (get_computation): Delete.
9824 (get_computation_cost): Implement like get_computation_cost_at.
9825 Use get_computation_at.
9826 (get_computation_cost_at): Delete.
9827 (rewrite_use_nonlinear_expr): Use get_computation_at.
9828 (rewrite_use_compare, remove_unused_ivs): Ditto.
9830 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9832 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
9834 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9836 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
9837 (ivopts_global_cost_for_size): Rename parameter and update uses.
9838 (iv_ca_recount_cost): Update uses.
9839 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
9840 candidates seperately in n_invs and n_cands.
9841 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
9843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9845 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
9846 (find_inv_vars_cb): New.
9847 (find_depends): Renamed to ...
9848 (find_inv_vars): ... this.
9849 (add_candidate_1, force_var_cost): Call find_inv_vars.
9850 (split_address_cost, determine_group_iv_cost_cond): Ditto.
9852 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9854 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
9855 inv_vars. Add inv_exprs.
9856 (struct iv_cand): Rename depends_on to inv_vars.
9857 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
9858 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
9859 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
9860 (dump_cand): Dump inv_vars.
9861 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
9862 (record_invariant, find_depends, add_candidate_1): Ditto.
9863 (set_group_iv_cost, force_var_cost): Ditto.
9864 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
9865 (get_computation_cost_at, get_computation_cost): Ditto.
9866 (determine_group_iv_cost_generic): Ditto.
9867 (determine_group_iv_cost_address): Ditto.
9868 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
9869 (determine_group_iv_costs): Ditto.
9870 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
9871 (iv_ca_set_remove_invariants): Renamed to ...
9872 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
9873 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
9874 (iv_ca_set_add_invariants): Renamed to ...
9875 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
9876 (iv_ca_set_cp): Use iv_ca_set_add_invs.
9877 (iv_ca_has_deps): Support inv_vars and inv_exprs.
9878 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
9879 (create_new_ivs): Remove useless dump.
9881 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9883 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
9885 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
9886 (iv_ca_set_no_cp, create_new_iv): Ditto.
9888 2017-05-02 Bin Cheng <bin.cheng@arm.com>
9890 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9892 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
9894 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
9895 function tree_check2.
9897 2017-05-02 Martin Liska <mliska@suse.cz>
9899 * doc/gcov.texi: Add missing preposition.
9900 * gcov.c (function_info::function_info): Properly fill up
9901 all member variables.
9903 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9905 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
9907 2017-05-02 Tamar Christina <tamar.christina@arm.com>
9909 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
9911 2017-05-02 Martin Liska <mliska@suse.cz>
9914 * lto-streamer-in.c (lto_read_tree_1): Remove
9916 * lto-streamer.c (struct tree_hash_entry): Likewise.
9917 (struct tree_entry_hasher): Likewise.
9918 (tree_entry_hasher::hash): Likewise.
9919 (tree_entry_hasher::equal): Likewise.
9920 (lto_streamer_init): Likewise.
9921 (lto_orig_address_map): Likewise.
9922 (lto_orig_address_get): Likewise.
9923 (lto_orig_address_remove): Likewise.
9924 * lto-streamer.h: Likewise.
9925 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
9926 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
9928 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
9930 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
9931 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
9932 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
9933 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
9934 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
9935 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
9936 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
9937 (mm_maskz_sub_ss): New intrinsics.
9938 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9939 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9940 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
9941 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
9942 (__builtin_ia32_subss_mask_round): New builtins.
9943 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9944 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9945 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
9947 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
9948 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9950 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9953 2017-05-02 Martin Jambor <mjambor@suse.cz>
9955 PR tree-optimization/78687
9956 * tree-sra.c (access): New field parent.
9957 (process_subtree_disqualification): New function.
9958 (disqualify_candidate): Call it.
9959 (build_accesses_from_assign): Reset write flag if creating an
9961 (build_access_subtree): Fill in parent field and also prpagate
9962 down grp_write flag.
9963 (create_artificial_child_access): New parameter set_grp_write, set
9964 grp_write to its value.
9965 (propagate_subaccesses_across_link): Also propagate grp_write flag
9967 (propagate_all_subaccesses): Push the closest parent back to work
9968 queue if add_access_to_work_queue returned true.
9970 2017-05-02 Richard Biener <rguenther@suse.de>
9972 * common.opt (fstrict-overflow): Alias negative to fwrapv.
9973 * doc/invoke.texi (fstrict-overflow): Remove all traces of
9974 -fstrict-overflow documentation.
9975 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
9976 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
9977 flag_strict_overflow.
9978 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
9979 * lto-opts.c (lto_write_options): Do not stream it.
9980 * lto-wrapper.c (merge_and_complain): Do not handle it.
9981 * opts.c (default_options_table): Do not set -fstrict-overflow.
9982 (finish_options): Likewise do not clear it when sanitizing.
9983 * simplify-rtx.c (simplify_const_relational_operation): Do not
9984 test flag_strict_overflow.
9986 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
9988 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
9989 using enabled attribute.
9990 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
9991 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
9992 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
9993 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
9994 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
9995 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
9996 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
9997 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
9998 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
9999 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10001 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10003 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10005 2017-05-02 Richard Biener <rguenther@suse.de>
10007 PR tree-optimization/80591
10009 2017-04-10 Richard Biener <rguenther@suse.de>
10011 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10014 2017-05-02 Richard Biener <rguenther@suse.de>
10016 PR tree-optimization/80549
10017 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10018 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10019 headers if they do not have a preheader.
10021 2017-05-02 Martin Liska <mliska@suse.cz>
10024 * common.opt: Fix typo.
10025 * doc/invoke.texi: Likewise.
10027 2017-05-01 Jan Beulich <jbeulich@suse.com>
10029 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10030 swapping, add (x,x,m,x,n) alternative.
10032 2017-05-01 Nathan Sidwell <nathan@acm.org>
10034 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10035 unnecessary unadjusted_alignment check.
10037 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10040 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10042 * gimplify.c (gimplify_cilk_detach): New function.
10043 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10044 * tree-core.h: Document EXPR_CILK_SPAWN.
10045 * tree.h (EXPR_CILK_SPAWN): Define.
10047 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10049 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10050 to use new fixit_hint representation, using the "replace" logic.
10051 (get_line_span_for_fixit_hint): Likewise.
10052 (layout::print_any_fixits): Likewise.
10053 (selftest::test_one_liner_many_fixits): Rename to...
10054 (selftest::test_one_liner_many_fixits_1): ...this, and update
10055 comment and expected output to reflect that the multiple fix-it
10056 hints are now consolidated into one insertion.
10057 (selftest::test_one_liner_many_fixits_2): New test.
10058 (selftest::test_diagnostic_show_locus_one_liner): Update for
10060 (selftest::test_fixit_consolidation): Update for fix-it API
10062 * diagnostic.c (print_parseable_fixits): Likewise.
10063 * edit-context.c (edited_line::m_line_events): Convert from
10064 auto_vec <line_event *> to auto_vec <line_event>.
10065 (class line_event): Convert from abstract base class to a concrete
10066 class, taking over the role of replace_event.
10067 (class insert_event): Delete.
10068 (class replace_event): Rename to class line_event. Convert to
10070 (edit_context::add_fixits): Reimplement.
10071 (edit_context::apply_insert): Delete.
10072 (edit_context::apply_replace): Rename to...
10073 (edit_context::apply_fixit): ...this. Convert to half-open range.
10074 (edited_file::apply_insert): Delete.
10075 (edited_file::apply_replace): Rename to...
10076 (edited_file::apply_fixit): ...this.
10077 (edited_line::~edited_line): Drop deletion of events.
10078 (edited_line::apply_insert): Delete.
10079 (edited_line::apply_replace): Rename to...
10080 (edited_line::apply_fixit): ...this. Convert to half-open range.
10081 Update for change to type of m_line_events.
10082 * edit-context.h (edit_context::apply_insert): Delete.
10083 (edit_context::apply_replace): Rename to...
10084 (edit_context::apply_fixit): ...this.
10086 2017-05-01 Martin Sebor <msebor@redhat.com>
10088 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10091 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10094 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10095 __get_cpuid_max returns 0.
10096 (__get_cpuid_count): Ditto.
10098 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10100 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10101 replacement expression is another instance of one of its arguments.
10103 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10106 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10107 check for stack push/pop autoinc.
10108 * config/i386/i386.c (ix86_agi_dependent): Return false
10109 if the only reason why modified_in_p returned true is that
10110 addr is SP based and set_insn is a push or pop.
10112 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10114 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10117 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10120 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10121 (account_size_time): Use two predicates - exec_pred and
10123 (evaluate_conditions_for_known_args): Compute both clause and
10125 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10126 (inline_summary_t::duplicate): Update.
10127 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10129 (compute_inline_parameters): Likewise.
10130 (estimate_edge_size_and_time): Update caluclation of time.
10131 (estimate_node_size_and_time): Compute both time and nonspecialized
10133 (estimate_ipcp_clone_size_and_time): Update.
10134 (inline_merge_summary): Update.
10135 (do_estimate_edge_time): Update.
10136 (do_estimate_edge_size): Update.
10137 (do_estimate_edge_hints): Update.
10138 (inline_read_section, inline_write_summary): Stream both new predicates.
10139 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10141 (compute_inlined_call_time): Cleanup.
10142 (big_speedup_p): Update.
10143 (edge_badness): Update.
10144 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10145 (size_time_entry): Replace predicate by exec_predicate and
10146 nonconst_predicate.
10147 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10148 (estimate_edge_time): Return also nonspec_time.
10149 (reset_edge_growth_cache): Update.
10151 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10153 PR rtl-optimization/80491
10154 * ifcvt.c (noce_process_if_block): When looking for x setter
10155 with missing else_bb, don't check only the insn right before
10156 cond_earliest, but look for the last insn that x is modified in
10157 within the same bb.
10159 PR rtl-optimization/80491
10160 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10162 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10164 PR tree-optimization/80487
10165 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10167 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10169 PR tree-optimization/79697
10170 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10171 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10172 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10174 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10175 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10177 2017-04-28 Martin Sebor <msebor@redhat.com>
10179 PR tree-optimization/80523
10180 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10181 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10183 (maybe_warn, format_directive, parse_directive): Use new functions.
10184 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10186 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10188 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10190 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10192 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10193 target_header_dir): Set correctly.
10194 * configure: Regenerated.
10195 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10196 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10197 instead of SYSTEM_HEADER_DIR.
10199 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10201 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10202 (estimate_local_effects): Likewise.
10203 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10204 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10205 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10206 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10207 do_estimate_edge_time, estimate_edge_time): Likewise.
10208 * ipa-inline-analysis.c (estimate_node_size_and_time,
10209 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10210 (estimate_time_after_inlining): Remove.
10212 2017-04-28 Martin Liska <mliska@suse.cz>
10214 * doc/gcov.texi: Enhance documentation of gcov.
10216 2017-04-28 Martin Liska <mliska@suse.cz>
10218 * doc/gcov.texi: Sort options in alphabetic order.
10219 * doc/gcov-dump.texi: Likewise.
10220 * doc/gcov-tool.texi: Likewise.
10221 * gcov.c (print_usage): Likewise.
10222 * gcov-dump.c (print_usage): Likewise.
10223 * gcov-tool.c (print_merge_usage_message): Likewise.
10224 (print_rewrite_usage_message): Likewise.
10225 (print_overlap_usage_message): Likewise.
10227 2017-04-28 Martin Liska <mliska@suse.cz>
10229 PR gcov-profile/53915
10230 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10232 2017-04-28 Martin Liska <mliska@suse.cz>
10234 PR gcov-profile/79891
10235 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10236 is marked by compiler as living on a line.
10237 (get_cycles_count): Remove usage of the union.
10238 (output_intermediate_file): Likewise.
10239 (find_source): Fix GNU coding style.
10240 (accumulate_line_counts): Remove old non-all block mode.
10241 (output_lines): Remove usage of the union.
10242 * profile.c (output_location): Include all BBs, even if
10243 belonging to a same line (and file) as a previous BB.
10245 2017-04-28 Martin Liska <mliska@suse.cz>
10247 * gcov.c (process_args): Handle new argument 'w'.
10248 (read_graph_file): Assign ID to BBs.
10249 (output_branch_count): Display BB # if verbose flag is set.
10250 (output_lines): Likewise for arcs.
10251 (print_usage): Add '--verbose' option help.
10252 * doc/gcov.texi: Document --verbose (-w) option.
10254 2017-04-28 Martin Liska <mliska@suse.cz>
10256 * gcov.c (struct block_location_info): New struct.
10257 (process_file): Fill up the new structure.
10258 (read_graph_file): Replace usage of encoding by the newly added
10260 (add_line_counts): Likewise.
10261 (accumulate_line_counts): Remove usage of the union.
10262 (function_info::function_info): New function.
10263 (function_info::~function_info): Likewise.
10264 (process_file): Call delete instead of release_function.
10265 (release_function): Release the function.
10266 (release_structures): Call delete instead of release_function.
10267 (solve_flow_graph): Replace usage of num_blocks.
10268 (find_exception_blocks): Likewise.
10269 (output_lines): Fix GNU coding style.
10271 2017-04-28 Martin Liska <mliska@suse.cz>
10274 * coverage.c (coverage_remove_note_file): New function.
10275 * coverage.h: Declare the function.
10276 * toplev.c (finalize): Clean if an error has been seen.
10278 2017-04-28 Martin Liska <mliska@suse.cz>
10280 PR gcov-profile/80031
10281 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10282 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10283 * gcov.c (read_graph_file): Read just number of blocks.
10284 * profile.c (branch_prob): Do not stream 0 flags per a basic
10287 2017-04-28 Martin Liska <mliska@suse.cz>
10289 * gcov-dump.c (tag_*): Add new argument to declarations.
10290 (dump_gcov_file): Likewise.
10291 (tag_blocks): Add and use new argument depth.
10292 (tag_arcs): Likewise.
10293 (tag_lines): Likewise.
10294 (tag_counters): Likewise.
10295 (tag_summary): Likewise.
10296 (dump_working_sets): Use depth to do a proper indentation.
10298 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10301 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10302 * symtab.c (symtab_node::debug_symtab): Move definition here.
10304 2017-04-28 Richard Biener <rguenther@suse.de>
10306 * lto-streamer.h (LTO_major_version): Bump to 7.
10308 2017-04-28 Richard Biener <rguenther@suse.de>
10310 * tree-vrp.c (assert_info): New struct.
10311 (add_assert_info): New helper.
10312 (register_edge_assert_for_2): Refactor to add asserts to a vector
10314 (register_edge_assert_for_1): Likewise.
10315 (register_edge_assert_for): Likewise.
10316 (finish_register_edge_assert_for): New helper actually registering
10317 asserts where live on edge.
10318 (find_conditional_asserts): Adjust.
10319 (find_switch_asserts): Likewise.
10320 (evrp_dom_walker::try_find_new_range): Generalize.
10321 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10323 2017-04-27 Marek Polacek <polacek@redhat.com>
10326 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10327 arg10 and arg11 to itype.
10329 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10331 * doc/extend.texi (Object Size Checking): Improve grammar.
10333 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10336 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10337 that the logic for permitting reciprocal estimates matches that
10340 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10343 * tree.c (type_cache_hasher::equal): Only compare
10344 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10345 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10346 non-aggregate element types.
10347 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10348 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10350 2017-04-27 Richard Biener <rguenther@suse.de>
10352 PR middle-end/80533
10353 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10354 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10355 keeping a reference to a trailing array.
10357 2017-04-27 Richard Biener <rguenther@suse.de>
10359 PR middle-end/80539
10360 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10361 being in loop-closed SSA form conservatively.
10362 (chrec_fold_multiply_poly_poly): Likewise.
10364 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10366 PR middle-end/79665
10367 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10368 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10370 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10373 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10374 (aarch64_function_arg_alignment): Return unsigned int again, but still
10375 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10376 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10377 Don't emit -Wpsabi note.
10378 (aarch64_function_arg_boundary): Likewise.
10379 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10382 2017-04-26 Nathan Sidwell <nathan@acm.org>
10384 * tree.h (crc32_unsigned_n): Declare.
10385 (crc32_unsigned, crc32_unsigned): Make inline.
10386 * tree.c (crc32_unsigned_bits): Replace with ...
10387 (crc32_unsigned_n): ... this.
10388 (crc32_unsigned, crc32_byte): Remove.
10389 (crc32_string): Remove unnecessary braces.
10391 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10393 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10394 * ipa-inline-analysis.c (MAX_TIME): Remove.
10395 (account_size_time): Use sreal for time.
10396 (dump_inline_summary): Update.
10397 (estimate_function_body_sizes): Update.
10398 (estimate_edge_size_and_time): Update.
10399 (estimate_calls_size_and_time): Update.
10400 (estimate_node_size_and_time): Update.
10401 (inline_merge_summary): Update.
10402 (inline_update_overall_summary): Update.
10403 (estimate_time_after_inlining): Update.
10404 (inline_read_section): Update.
10405 (inline_write_summary): Update.
10406 * ipa-inline.c (compute_uninlined_call_time): Update.
10407 (compute_inlined_call_time): Update.
10408 (recursive_inlining): Update.
10409 (inline_small_functions): Update.
10410 (dump_overall_stats): Update.
10411 * ipa-inline.h: Include sreal.h.
10412 (size_time_entry): Turn time to sreal.
10413 (inline_summary): Turn self_time nad time to sreal.
10415 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10417 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10419 (sreal::stream_out, sreal::stream_in): New.
10420 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10422 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10424 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10427 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10430 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10432 (dimode_scalar_chain::compute_convert_gain): Ditto.
10433 (dimode_scalar_chain::make_vector_copies): Ditto.
10434 (dimode_scalar_chain::convert_reg): Ditto.
10435 (dimode_scalar_chain::convert_insn): Ditto.
10436 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10437 (VI248_AVX512BW_1): New mode iterator.
10438 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10439 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10442 2017-04-25 Martin Sebor <msebor@redhat.com>
10444 PR tree-optimization/80497
10445 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10446 constants are representable in HOST_WIDE_INT.
10447 (parse_directive): Ditto.
10449 2017-04-25 Martin Sebor <msebor@redhat.com>
10452 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10453 (new_zero_array): Adjust signature.
10454 (dom_info::dom_init): Used unsigned rather that size_t.
10455 (dom_info::dom_info): Same.
10457 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10458 Jakub Jelinek <jakub@redhat.com>
10461 * config/arm/arm.c: Include gimple.h.
10462 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10463 returns negative, increment ncrn only if it returned positive.
10464 (arm_needs_doubleword_align): Return int instead of bool,
10465 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10466 members, but if there is any such non-FIELD_DECL
10467 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10468 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10469 returns negative, increment nregs only if it returned positive.
10470 (arm_setup_incoming_varargs): Likewise.
10471 (arm_function_arg_boundary): Emit -Wpsabi note if
10472 arm_needs_doubleword_align returns negative, return
10473 DOUBLEWORD_ALIGNMENT only if it returned positive.
10475 2017-04-25 Marek Polacek <polacek@redhat.com>
10478 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10479 first argument to type.
10481 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10484 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10485 type checks to test for compatibility instead of equality.
10487 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10488 Jakub Jelinek <jakub@redhat.com>
10491 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10493 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10494 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10495 the alignment computation, but return their maximum in warn_alignment.
10496 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10497 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10499 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10500 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10503 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10505 * config/arc/simdext.md (dmpyh): Fix typo.
10507 2017-04-25 Richard Biener <rguenther@suse.de>
10509 PR tree-optimization/80492
10510 * alias.c (compare_base_decls): Handle registers with asm
10511 specification conservatively.
10512 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10513 compare_base_decls returning dont-know properly.
10515 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10517 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10518 (legitimate_offset_address_p): New function.
10519 (arc_legitimate_address_p): Use above function.
10521 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10523 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10525 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10527 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
10528 ACCH registers whenever they are available.
10530 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10532 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
10533 double regs fix when not used.
10535 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10537 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
10539 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
10540 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
10542 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10544 * config/arc/arc.c (arc_output_addsi): Check for h-register class
10545 when emitting short ADD instructions.
10547 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10549 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
10551 (cmpsi_cc_c_insn): Likewise.
10552 (cbranchsi4_scratch): Compute proper instruction length using
10553 compact_hreg_operand.
10554 * config/arc/predicates.md (compact_hreg_operand): New predicate.
10556 2017-04-25 Richard Biener <rguenther@suse.de>
10558 PR middle-end/80509
10559 * passes.c (pass_manager::pass_manager): Initialize
10560 m_name_to_pass_map.
10562 2017-04-25 Richard Biener <rguenther@suse.de>
10564 PR tree-optimization/79201
10565 * tree-ssa-sink.c (statement_sink_location): Handle calls.
10567 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10570 * config/s390/vector.md: Split MEM->GPR vector moves for
10571 non-s_operand addresses.
10573 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10576 * config/s390/predicates.md (reload_const_wide_int_operand): New
10578 * config/s390/s390.md ("movti"): Remove d/P alternative.
10579 ("movti_bigconst"): New pattern definition.
10581 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10584 * s390-protos.h (s390_expand_cs_hqi): Removed.
10585 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
10586 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
10587 modes as well as CCZ1mode and CCZmode.
10588 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
10589 signature of s390_emit_compare_and_swap.
10590 (s390_expand_cs_hqi): Likewise, make static.
10591 (s390_expand_cs_tdsi): Generate an explicit compare before trying
10592 compare-and-swap, in some cases.
10593 (s390_expand_cs): Wrapper function.
10594 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
10596 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
10597 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
10598 patterns for small and large integers. Forbid symref memory operands.
10599 Move expander to s390.c. Require cc register.
10600 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
10601 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
10602 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
10603 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
10604 symref memory operands. Remove CC mode and call s390_match_ccmode
10606 ("atomic_exchange<mode>"): Allow and implement all integer modes.
10608 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10610 * config/s390/s390.md (define_peephole2): New peephole to help
10611 combining the load-and-test pattern with volatile memory.
10613 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10615 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
10616 with CCZmode for TARGET_Z196.
10618 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10620 PR rtl-optimization/80501
10621 * combine.c (make_compound_operation_int): Set subreg_code to SET
10622 even for AND with mask of the sign bit of mode.
10624 PR rtl-optimization/80500
10625 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
10626 sum's initial value.
10628 2017-04-25 Julian Brown <julian@codesourcery.com>
10629 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10631 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
10633 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
10635 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
10637 2017-04-25 Julian Brown <julian@codesourcery.com>
10638 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10640 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
10641 (thunderx2t99_sha): New Reservation.
10643 2017-04-25 Julian Brown <julian@codesourcery.com>
10644 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10646 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
10647 type for 1-element load.
10649 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
10651 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
10653 2017-04-24 Martin Jambor <mjambor@suse.cz>
10655 PR tree-optimization/80293
10656 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
10657 char arrays not totally scalarizable if it is false.
10658 (analyze_all_variable_accesses): Pass correct value in the new
10659 parameter. Add a statistics counter.
10661 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
10663 PR middle-end/79931
10664 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
10666 2017-04-24 Richard Biener <rguenther@suse.de>
10668 PR tree-optimization/80494
10669 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
10670 out for complex types.
10672 2017-04-24 Richard Biener <rguenther@suse.de>
10674 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
10675 * tree-ssa-sccvn.c (print_scc): Print SCC size.
10676 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
10677 (DFS): Adjust and never fail.
10678 (sccvn_dom_walker::fail): Remove.
10679 (sccvn_dom_walker::before_dom_children): Adjust.
10680 (run_scc_vn): Likewise and never fail.
10681 * tree-ssa-pre.c (pass_pre::execute): Adjust.
10682 (pass_fre::execute): Likewise.
10684 2017-04-24 Richard Biener <rguenther@suse.de>
10686 PR tree-optimization/79725
10687 * tree-ssa-sink.c (statement_sink_location): Return whether
10688 failure reason was zero uses. Move that check later.
10689 (sink_code_in_bb): Deal with zero uses by removing the stmt
10692 2017-04-24 Richard Biener <rguenther@suse.de>
10695 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
10696 pointer-based references.
10698 2017-04-24 Richard Biener <rguenther@suse.de>
10701 * pass_manager.h (pass_manager::operator new): Remove.
10702 (pass_manager::operator delete): Likewise.
10703 * passes.c (pass_manager::operator new): Remove.
10704 (pass_manager::operator delete): Likewise.
10705 (pass_manager::pass_manager): Zero individual pass members.
10707 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
10710 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
10711 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
10712 Check "XEXP (src, 1)" operand here.
10713 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
10714 Check "XEXP (src, 1)" operand here.
10715 (dimode_scalar_chain::make_vector_copies): Detect count register
10716 of a shift instruction. Zero extend count register from QImode
10717 to DImode to satisfy vector shift pattern count operand predicate.
10718 Substitute vector shift count operand with a DImode copy.
10719 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
10722 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
10724 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
10725 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10726 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
10727 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
10728 (UNSPEC_NOREX_MEM): Remove definition.
10730 2017-04-21 Richard Biener <rguenther@suse.de>
10732 PR tree-optimization/79547
10733 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10734 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
10735 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
10736 without any constraints.
10738 2017-04-21 Richard Biener <rguenther@suse.de>
10740 PR tree-optimization/78847
10741 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
10743 2017-04-21 Richard Biener <rguenther@suse.de>
10745 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
10746 (build_distinct_type_copy): Likewise.
10747 (build_variant_type_copy): Likewise.
10748 * tree.c (build_qualified_type): Pass down mem-stat info.
10749 (build_distinct_type_copy): Likewise.
10750 (build_variant_type_copy): Likewise.
10752 2017-04-21 Richard Biener <rguenther@suse.de>
10754 PR tree-optimization/80237
10755 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
10757 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
10758 for a simplified result.
10760 2016-04-21 Richard Biener <rguenther@suse.de>
10762 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
10763 sth as strict as a simple_iv but a chrec without symbols and an
10764 operand defined in the loop we are peeling (and not some subloop).
10765 (propagate_constants_for_unrolling): Propagate all constants.
10767 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10770 * config/i386/i386.c (print_reg): Remove assert for disalowed
10771 regno values, call output_operand_lossage instead.
10773 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
10776 * config/i386/constraints.md (Yc): New register constraint.
10777 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
10778 Use Yc constraint for alternative 2 of operand 0. Remove
10779 preferred_for_speed attribute.
10781 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10783 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
10784 lastprivate clauses in SIMT case.
10786 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
10788 * doc/invoke.texi (-Wextra-semi): Document new warning option.
10790 2017-04-20 Richard Biener <rguenther@suse.de>
10792 PR tree-optimization/57796
10793 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
10794 as N scalar stores.
10795 (vect_model_load_cost): Cost gathers as N scalar loads.
10797 2017-04-20 Richard Biener <rguenther@suse.de>
10799 * ggc-page.c (ggc_allocated_p): Rename to ...
10800 (safe_lookup_page_table_entry): ... this and return the lookup
10802 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
10804 2017-04-20 Richard Biener <rguenther@suse.de>
10806 PR tree-optimization/80453
10807 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
10808 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
10809 from the conditions.
10810 (vn_phi_eq): Pass them down.
10811 (vn_phi_lookup): Record them.
10812 (vn_phi_insert): Likewise.
10814 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
10816 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
10817 uninitialized variable warning to avoid buffer overrun.
10819 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
10822 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
10823 is suppressed for '{ 0 }' in C.
10825 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10827 * BASE-VER: Set to 8.0.0.
10829 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
10831 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
10832 priority .init_array and .fini_array section with SECTION_NOTYPE
10835 2017-04-20 Jakub Jelinek <jakub@redhat.com>
10837 PR middle-end/80423
10838 * tree.h (build_array_type): Add typeless_storage default argument.
10839 * tree.c (type_cache_hasher::equal): Also compare
10840 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
10841 (build_array_type): Add typeless_storage argument, set
10842 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
10844 (build_nonshared_array_type): Adjust build_array_type_1 caller.
10845 (build_array_type): Likewise. Add typeless_storage argument.
10847 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10848 Jakub Jelinek <jakub@redhat.com>
10850 PR tree-optimization/80426
10851 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
10852 operation on symbolic operands, also compute the overflow for the
10853 invariant part when the operation degenerates into a negation.
10855 2017-04-19 Jakub Jelinek <jakub@redhat.com>
10858 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
10859 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
10862 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
10864 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
10867 * config/avr/avr.c (tree.h): Include it.
10868 (cgraph.h): Include it.
10869 (avr_encode_section_info): Don't warn for uninitialized progmem
10870 variable if it's just an alias.
10872 2017-04-19 Richard Biener <rguenther@suse.de>
10875 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
10876 when needed by AutoPGO.
10878 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
10881 * doc/lto.texi: Remove an extra 'that'.
10883 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
10885 PR rtl-optimization/80429
10886 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
10887 are only used in debug insns.
10889 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
10890 Vladimir Makarov <vmakarov@redhat.com>
10892 * config/sparc/predicates.md (input_operand): Add comment. Return
10893 true for any memory operand when LRA is in progress.
10894 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
10896 2017-04-18 Jeff Law <law@redhat.com>
10899 * mips.md ({return,simple_return}_internal): Do not overwrite
10902 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10904 PR tree-optimization/80443
10905 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
10906 instead of adding 1, subtract -1 and similarly instead of subtracting
10909 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
10911 PR rtl-optimization/80357
10912 * haifa-sched.c (tmp_bitmap): New variable.
10913 (model_recompute): Handle duplicate use records.
10914 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
10915 (free_global_sched_pressure_data): Free it.
10917 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
10920 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10921 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10922 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10923 instead of SYSTEM_HEADER_DIR.
10925 2017-04-18 Jeff Law <law@redhat.com>
10927 PR middle-end/80422
10928 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
10929 predecessors after walking up the insn chain.
10931 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10934 * dwarf2out.c (modified_type_die): Try harder not to emit internal
10935 sizetype type into debug info.
10937 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
10940 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
10941 unneeded test for TARGET_UPPER_REGS_SF.
10942 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
10944 2017-04-18 Jakub Jelinek <jakub@redhat.com>
10947 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
10948 instead of gsi_after_labels.
10950 2017-04-18 Jeff Law <law@redhat.com>
10952 * regcprop.c (maybe_mode_change): Avoid creating copies of the
10956 2017-04-13 Jeff Law <law@redhat.com>
10957 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
10958 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
10960 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
10963 * config/avr/avr.c (intl.h): Include it.
10964 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
10966 2017-04-18 Martin Liska <mliska@suse.cz>
10968 PR gcov-profile/78783
10969 * gcov-tool.c (gcov_output_files): Validate that destination
10970 file is either removed by the tool or by a user.
10972 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10973 Guy Benyei <guybe@mellanox.com>
10975 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
10976 block, and do not negate it, the stored id is already negative.
10978 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
10980 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
10982 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10985 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
10986 masks of options that should be turned off if the VSX vector
10987 options are turned off.
10988 (OTHER_P8_VECTOR_MASKS): Likewise.
10989 (OTHER_VSX_VECTOR_MASKS): Likewise.
10990 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
10991 rs6000_disable_incompatible_switches to validate no type switches
10993 (rs6000_incompatible_switch): New function to disallow turning on
10994 other vector options if -mno-vsx, -mno-power8-vector, or
10995 -mno-power9-vector are specified.
10997 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
10999 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11001 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11003 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11004 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11005 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11006 (ARG_POINTER_CFA_OFFSET): Likewise.
11008 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11010 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11011 conditions to take advantage of various optimizations.
11013 2017-04-13 Jeff Law <law@redhat.com>
11015 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11016 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11017 (zero_extendsidi2_dext): Likewise.
11019 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11022 * fold-const.c (fold_ternary_loc): Revert
11023 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11026 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11028 PR rtl-optimization/80343
11029 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11030 new scratch pseudo.
11032 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11035 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11036 to ubsan_encode_value.
11038 2017-04-13 Jeff Law <law@redhat.com>
11040 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11041 appearing in DEBUG_INSNs.
11043 2017-04-13 Martin Liska <mliska@suse.cz>
11045 PR gcov-profile/80413
11046 * gcov-io.c (gcov_write_string): Copy to buffer just when
11047 allocated size is greater than zero.
11049 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11052 * dwarf2out.c (decls_for_scope): Ignore declarations of
11053 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11055 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11058 * ipa-visibility.c (non_local_p): Fix typos.
11059 (localize_node): When localizing symbol in same comdat group,
11060 dissolve the group only when we know external symbols are going
11062 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11064 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11066 PR tree-optimization/79390
11067 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11068 order does not result in usable sequence, retry with reversed operand
11074 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11075 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11076 op0 instead of fold_convert_loc (loc, type, arg0).
11078 2017-04-12 Jeff Law <law@redhat.com>
11080 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11081 has a delay slot in the generated code.
11083 * config/cris/cris.md (cris_preferred_reload_class): Return
11084 GENNONACR_REGS rather than GENERAL_REGS.
11086 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11089 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11090 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11091 signedness of the result type.
11093 2017-04-12 Richard Biener <rguenther@suse.de>
11094 Jeff Law <law@redhat.com>
11096 PR tree-optimization/80359
11097 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11098 trim stores to TARGET_MEM_REFs.
11100 2017-04-12 Richard Biener <rguenther@suse.de>
11102 PR tree-optimization/79390
11103 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11104 threading case even more.
11106 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11109 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11110 for quad_address_p for TImode, instead of just not indexed_address.
11112 2017-04-12 Richard Biener <rguenther@suse.de>
11113 Bernd Edlinger <bernd.edlinger@hotmail.de>
11115 PR middle-end/79671
11116 * alias.c (component_uses_parent_alias_set_from): Handle
11117 TYPE_TYPELESS_STORAGE.
11118 (get_alias_set): Likewise.
11119 * tree-core.h (tree_type_common): Add typeless_storage flag.
11120 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11121 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11122 for types containing members with TYPE_TYPELESS_STORAGE.
11123 (place_field): Likewise.
11124 (layout_type): Likewise for ARRAY_TYPE.
11125 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11126 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11127 TYPE_TYPELESS_STORAGE.
11128 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11130 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11133 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11134 first argument to type.
11136 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11140 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11141 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11142 (rs6000_expand_binop_builtin): Likewise.
11143 (rs6000_expand_ternop_builtin): Likewise; also add missing
11144 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11145 vshasigma built-ins.
11146 * doc/extend.texi: Document that vec_xxpermdi's third argument
11147 must be a constant.
11149 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11151 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11152 Use shift_const cost parameter when calculating gain of STV shifts.
11154 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11156 PR rtl-optimization/70478
11157 * lra-constraints.c (process_alt_operands): Check memory for
11158 disfavoring memory insn operand.
11160 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11162 PR middle-end/80100
11163 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11164 left shift in unsigned HOST_WIDE_INT type.
11166 PR rtl-optimization/80385
11167 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11168 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11171 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11172 if they have any depend clauses.
11174 2017-04-11 Martin Liska <mliska@suse.cz>
11177 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11178 * ipa-split.c (split_function): Create a local comdat symbol
11179 if caller is in a comdat group.
11181 2017-04-11 Martin Liska <mliska@suse.cz>
11184 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11187 2017-04-11 Martin Sebor <msebor@redhat.com>
11189 PR middle-end/80364
11190 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11191 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11193 (directive::set_width, directive::set_precision, format_character):
11195 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11198 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11201 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11202 conflict, set target->arch_name instead of target->cpu_name.
11204 2017-04-11 Richard Biener <rguenther@suse.de>
11206 PR tree-optimization/80374
11207 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11208 build_zero_cst, remove fold_convertible_p check again.
11210 2017-04-11 Martin Liska <mliska@suse.cz>
11213 * ubsan.c (instrument_object_size): Do not instrument register
11216 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11219 * config/i386/i386-builtin-types.def
11220 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11221 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11222 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11223 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11224 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11225 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11226 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11227 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11228 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11229 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11230 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11231 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11232 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11233 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11234 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11235 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11236 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11237 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11238 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11239 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11240 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11241 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11242 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11243 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11244 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11245 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11246 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11247 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11248 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11249 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11250 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11251 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11252 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11253 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11254 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11255 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11256 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11257 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11258 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11259 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11260 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11261 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11262 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11263 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11264 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11265 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11267 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11268 flag to second_arg_count, handle 4 argument function type _COUNT
11269 aliases, handle second_arg_count on second argument rather than last.
11271 2017-04-10 Jeff Law <law@redhat.com>
11273 PR tree-optimization/80374
11274 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11275 record anything if we can not convert integer_zero_node to the
11278 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11281 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11282 Enhance special handling given to the TARGET_P9_MINMAX option in
11283 relation to certain other options.
11285 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11287 PR tree-optimization/80153
11288 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11289 remove POINTER_PLUS_EXPR's base part directly, rather than through
11292 2017-04-10 Richard Biener <rguenther@suse.de>
11293 Bin Cheng <bin.cheng@arm.com>
11295 PR tree-optimization/80153
11296 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11297 the first element of pointer type aff_tree. Build result expr in
11299 (add_elt_to_tree): Convert to type unconditionally. Remove other
11300 fold_convert calls.
11301 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11302 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11304 2017-04-10 Richard Biener <rguenther@suse.de>
11306 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11309 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11311 PR rtl-optimization/70478
11312 * lra-constraints.c (curr_small_class_check): New.
11313 (update_and_check_small_class_inputs): New.
11314 (process_alt_operands): Update curr_small_class_check. Disfavor
11315 alternative insn memory operands. Check available regs for small
11318 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11321 * config/mips/mips.opt (-mvirt): Update description.
11322 * doc/invoke.texi (-mvirt): Likewise.
11324 2017-04-10 Richard Biener <rguenther@suse.de>
11326 PR middle-end/80362
11327 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11328 looking for NEGATE_EXPR in -A / -B to A / B folding.
11330 2017-04-10 Martin Liska <mliska@suse.cz>
11332 PR gcov-profile/80224
11333 * gcov.c (print_usage): Fix usage string.
11334 (get_gcov_intermediate_filename): Remove.
11335 (output_gcov_file): Use both for normal and intermediate format.
11336 (generate_results): Do not initialize special file for
11337 intermediate format.
11339 2017-04-10 Richard Biener <rguenther@suse.de>
11341 PR tree-optimization/80304
11342 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11345 2017-04-10 Nathan Sidwell <nathan@acm.org>
11348 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11349 (rs6000_init_builtins): Use it.
11351 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11353 * config/arm/arm.md (<mrc>): Add mode to SET source.
11354 (<mrrc>): Likewise.
11356 2017-04-10 Richard Biener <rguenther@suse.de>
11358 PR middle-end/80344
11359 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11361 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11364 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11365 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11366 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11367 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11368 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11369 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11370 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11371 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11372 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11373 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11374 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11375 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11376 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11377 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11378 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11379 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11380 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11381 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11382 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11383 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11384 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11385 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11386 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11388 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11390 PR rtl-optimization/70478
11391 * lra-constraints.c: Reverse the last patch.
11393 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11395 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11396 Add comment for WCHAR_T.
11398 2017-04-08 Martin Liska <mliska@suse.cz>
11401 2017-04-07 Martin Liska <mliska@suse.cz>
11404 * ipa-split.c (split_function): Add function part to a same comdat
11407 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11410 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11412 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11414 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11415 (vec_pairing): ...to this.
11416 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11417 (rs6000_sched_init): Adjust for name change.
11418 (struct rs6000_sched_context): Likewise.
11419 (rs6000_init_sched_context): Likewise.
11420 (rs6000_set_sched_context): Likewise.
11422 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11428 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11430 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11431 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11432 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11434 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11436 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11438 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11440 PR rtl-optimization/70703
11441 * ira-color.c (update_conflict_hard_regno_costs): Use
11442 int64_t instead of HOST_WIDE_INT.
11444 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11446 PR rtl-optimization/70478
11447 * lra-constraints.c (process_alt_operands): Disfavor alternative
11448 insn memory operands.
11450 2017-04-07 Jeff Law <law@redhat.com>
11452 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11453 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11455 2017-04-07 Martin Liska <mliska@suse.cz>
11458 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11459 Show error message instead of an ICE.
11461 2017-04-07 Martin Liska <mliska@suse.cz>
11464 * ipa-split.c (split_function): Add function part to a same comdat
11467 2017-04-07 Richard Biener <rguenther@suse.de>
11469 PR middle-end/80341
11470 * tree.c (get_unwidened): Also handle ! for_type case for
11472 * convert.c (do_narrow): Split out from ...
11473 (convert_to_integer_1): ... here. Do not pass final truncation
11474 type to get_unwidened for TRUNC_DIV_EXPR.
11476 2017-04-07 Richard Biener <rguenther@suse.de>
11478 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11480 (aff_combination_const): Adjust.
11481 (aff_combination_scale): Likewise.
11482 (aff_combination_add_elt): Likewise.
11483 (aff_combination_add_cst): Likewise.
11484 (aff_combination_convert): Likewise.
11485 (add_elt_to_tree): Likewise. Remove unused argument.
11486 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11488 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11490 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11492 * config/arm/arm.c (arm_default_short_enums): Use
11493 ARM_DEFAULT_SHORT_ENUMS.
11494 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11496 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11499 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11500 members with redundant out-of-class redeclaration.
11502 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11505 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11506 * config/i386/i386.md (*zero_extendsidi2):
11507 Add (?*x,*x) and (?*v,*v) alternatives.
11509 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11512 * config/i386/i386.c (ix86_expand_builtin)
11513 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11514 mode from insn data. Convert operands to insn operand mode.
11515 Copy operands that don't satisfy insn predicate to a register.
11517 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11519 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11522 2017-04-06 Richard Biener <rguenther@suse.de>
11524 PR tree-optimization/80334
11525 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11526 preserve alignment of accesses.
11528 2017-04-06 Richard Biener <rguenther@suse.de>
11530 PR tree-optimization/80262
11531 * tree-sra.c (build_ref_for_offset): Preserve address-space
11533 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
11534 Drop useless address-space information on MEM_REF offsets.
11536 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
11538 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
11540 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11542 PR rtl-optimization/70703
11543 * ira-color.c (update_conflict_hard_regno_costs): Use
11544 HOST_WIDE_INT instead of long.
11546 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
11549 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
11550 not defined for x86_64 target. Add -mmmx target option when __SSE2__
11552 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
11553 for x86_64 target. Handle -m3dnowa option.
11555 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
11557 PR rtl-optimization/70703
11558 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
11559 (update_conflict_hard_regno_costs): Use long instead of unsigned
11560 arithmetic for cost calculation.
11562 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11563 Bernd Edlinger <bernd.edlinger@hotmail.de>
11566 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
11569 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
11572 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
11573 ptr_mode with Pmode throughout.
11574 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
11575 into probe_stack_range and use DImode.
11577 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
11580 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
11581 call_eh_return is true.
11583 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11585 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
11586 Initialize last_match_fntype_index.
11588 2017-04-05 Jakub Jelinek <jakub@redhat.com>
11591 * tree-nvr.c: Include internal-fn.h.
11592 (pass_return_slot::execute): Ignore internal calls without
11595 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11596 Richard Biener <rguenther@suse.de>
11599 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
11600 captures used multiple times, except for the last use.
11601 * generic-match-head.c: Include gimplify.h.
11603 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11605 PR tree-optimization/79390
11606 * target.h (struct noce_if_info): Declare.
11607 * targhooks.h (default_noce_conversion_profitable_p): Declare.
11608 * target.def (noce_conversion_profitable_p): New target hook.
11609 * ifcvt.h (struct noce_if_info): New type, moved from ...
11610 * ifcvt.c (struct noce_if_info): ... here.
11611 (noce_conversion_profitable_p): Renamed to ...
11612 (default_noce_conversion_profitable_p): ... this. No longer
11614 (noce_try_store_flag_constants, noce_try_addcc,
11615 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
11616 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
11617 instead of noce_conversion_profitable_p.
11618 * config/i386/i386.c: Include ifcvt.h.
11619 (ix86_option_override_internal): Don't override
11620 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
11621 (ix86_noce_conversion_profitable_p): New function.
11622 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
11623 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
11624 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
11625 * doc/tm.texi: Regenerated.
11627 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11629 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
11632 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11635 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
11636 instructions for small multiply cores.
11638 2017-04-04 Jeff Law <law@redhat.com>
11640 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
11642 (mips_expand_vec_perm_const): Initialize elements in orig_perm
11643 that are not set by the loop over the elements.
11645 2017-04-04 Jakub Jelinek <jakub@redhat.com>
11648 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
11649 int mode, convert_modes it to mode as unsigned, otherwise use
11650 lowpart_subreg to mode rather than SImode.
11651 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
11652 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
11653 Use DImode instead of SImode for the shift count operand.
11654 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
11657 2017-04-04 Richard Biener <rguenther@suse.de>
11659 PR middle-end/80281
11660 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
11661 arithmetic done for the negate or the plus. Simplify.
11662 (A - (-B) -> A + B): Likewise.
11663 * fold-const.c (split_tree): Make sure to not negate pointers.
11665 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
11667 PR rtl-optimization/60818
11668 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
11669 a compare of comparisons with the thing compared if this results
11670 in a different machine mode.
11672 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
11674 * alias.c (base_alias_check): Fix typo in comment.
11675 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
11676 * cgraphunit.c (symbol_table::compile): Likewise.
11677 * collect2.c (maybe_run_lto_and_relink): Likewise.
11678 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
11679 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
11680 * config/avr/avr.c (avr_map_op_t): Likewise.
11681 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
11682 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
11683 * config/epiphany/epiphany.md (movcc): Likewise.
11684 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
11685 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
11687 * config/mips/mips.c (mips_save_restore_reg): Likewise.
11688 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
11689 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
11690 * config/sh/sh.c (sh_rtx_costs): Likewise.
11691 * fold-const.c (fold_truth_andor): Likewise.
11692 * genautomata.c (collapse_flag): Likewise.
11693 * gengtype.h (struct type::u::s): Likewise.
11694 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
11695 * input.c (FORMAT_AMOUNT): Likewise.
11696 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
11697 (known_aggs_to_agg_replacement_list): Likewise.
11698 * ipa-inline-analysis.c: Likewise.
11699 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
11700 * ipa-polymorphic-call.c
11701 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
11702 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
11703 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
11705 * modulo-sched.c (apply_reg_moves): Likewise.
11706 * omp-expand.c (build_omp_regions_1): Likewise.
11707 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
11708 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
11709 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
11710 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
11711 * value-prof.c: Likewise.
11712 * var-tracking.c (val_reset): Likewise.
11714 2017-04-03 Richard Biener <rguenther@suse.de>
11716 PR tree-optimization/80275
11717 * fold-const.c (split_address_to_core_and_offset): Handle
11720 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
11722 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
11723 descriptors is at least equal to that of functions.
11725 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11727 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
11729 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
11732 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
11733 (mov<IMOD4:mode>): New expander.
11734 (*mov<IMOD4:mode>_internal): New insn and split pattern.
11736 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11738 PR rtl-optimization/79405
11739 * fwprop.c (propagations_left): New variable.
11740 (forward_propagate_into): Decrement it.
11741 (fwprop_init): Initialize it.
11742 (fw_prop): If the variable has reached zero, stop propagating.
11743 (fwprop_addr): Ditto.
11745 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11748 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
11749 a FUNCTION_DECL, pass it as decl instead of origin to
11752 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
11754 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
11757 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
11760 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
11761 TARGET_VSX_SMALL_INTEGER.
11763 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11765 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11766 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
11768 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11770 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
11771 extraction from odd-numbered MSA register.
11773 2017-03-31 Jakub Jelinek <jakub@redhat.com>
11775 PR middle-end/80173
11776 * expmed.c (store_bit_field_1): Don't attempt to create
11777 a word subreg out of hard registers wider than word if they
11778 have HARD_REGNO_NREGS of 1 for their mode.
11780 PR middle-end/80163
11781 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
11782 conversions to integer types wider than word and pointer.
11785 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
11786 (rtx_equal_for_cselib_p): Pass 0 to it.
11787 * cselib.c (cselib_hasher::equal): Likewise.
11788 (rtx_equal_for_cselib_1): Add depth argument. If depth
11789 is 128, don't look up VALUE locs and punt. Increment
11790 depth in recursive calls when walking VALUE locs.
11792 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
11794 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
11795 (make_gcov_file_name): Use the canonical path name for generating
11797 (read_line): Fix handling of files with ascii null bytes.
11799 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
11801 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
11802 to initialise a vector register instead
11803 of using a const_int.
11805 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11807 PR translation/80189
11808 * gimplify.c (omp_default_clause): Use %qs instead of %s in
11809 diagnostic messages.
11811 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
11814 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
11815 (dfp_diex_<mode>): Update mode of operand 1.
11816 * doc/extend.texi (dxex, dxexq): Document change to return type.
11817 (diex, diexq): Document change to argument type.
11819 2017-03-30 Martin Jambor <mjambor@suse.cz>
11822 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
11823 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
11824 it reflects the signature changes performed at the callee side.
11825 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
11826 to cgraph_build_function_type_skip_args.
11827 (build_function_decl_skip_args): Adjust call to the above function.
11829 2017-03-30 Jakub Jelinek <jakub@redhat.com>
11832 * config/i386/sse.md
11833 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
11834 register as dest whenever it is a MEM not rtx_equal_p to the
11835 corresponding dup operand, and when forcing into reg move the
11836 reg into the memory afterwards.
11837 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
11838 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
11839 for the force_reg mode.
11840 (avx512vl_vextractf128<mode>): Use register as dest either
11841 always when a MEM, or when it is a MEM not rtx_equal_p to the
11842 corresponding dup operand, or even not when it is a CONST_VECTOR
11843 depending on the mode and lo vs. hi.
11844 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
11846 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
11847 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
11848 Likewise. Require that operands[2] is even.
11849 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
11850 Remove extraneous parens. Require that operands[2] is a multiple
11852 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
11853 operands[0] is a MEM if <mask_applied>, the predicates/constraints
11854 disallow memory then.
11856 2017-03-30 Richard Biener <rguenther@suse.de>
11858 PR tree-optimization/77498
11859 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
11860 to non-constants over backedges.
11862 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
11864 PR rtl-optimization/80233
11865 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
11866 as last_combined_insn. Do not test for BARRIER_P separately.
11868 2017-03-29 Andreas Schwab <schwab@suse.de>
11871 * calls.c (prepare_call_address): Convert funexp to Pmode before
11872 copying to temp reg.
11874 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11876 PR tree-optimization/80158
11877 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
11878 Handle possible future case of more than one alternate
11880 (replace_rhs_if_not_dup): Likewise.
11881 (replace_one_candidate): Likewise.
11883 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
11885 PR rtl-optimization/80193
11886 * ira.c (ira): Do not check allocation for LRA.
11888 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
11890 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
11891 (nvptx_output_simt_exit): Declare.
11892 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
11893 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
11894 (init_softstack_frame): Move initialization of crtl->is_leaf to...
11895 (nvptx_declare_function_name): ...here. Emit declaration of local
11896 memory space buffer for omp_simt_enter insn.
11897 (nvptx_output_unisimt_switch): New.
11898 (nvptx_output_softstack_switch): New.
11899 (nvptx_output_simt_enter): New.
11900 (nvptx_output_simt_exit): New.
11901 * config/nvptx/nvptx.h (struct machine_function): New fields
11902 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
11903 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
11904 (UNSPECV_SIMT_EXIT): Ditto.
11905 (omp_simt_enter_insn): New insn.
11906 (omp_simt_enter): New expansion.
11907 (omp_simt_exit): New insn.
11908 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
11910 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
11911 (expand_GOMP_SIMT_ENTER_ALLOC): New.
11912 (expand_GOMP_SIMT_EXIT): New.
11913 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
11914 (GOMP_SIMT_ENTER_ALLOC): Ditto.
11915 (GOMP_SIMT_EXIT): Ditto.
11916 * target-insns.def (omp_simt_enter): New insn.
11917 (omp_simt_exit): Ditto.
11918 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
11920 (lower_rec_simd_input_clauses): Implement SIMT privatization.
11921 (lower_rec_input_clauses): Likewise.
11922 (lower_lastprivate_clauses): Handle SIMT privatization.
11924 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
11925 (ompdevlow_adjust_simt_enter): New.
11926 (find_simtpriv_var_op): New.
11927 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
11928 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
11930 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
11931 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
11932 (copy_decl_for_dup_finish): Ditto.
11934 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
11936 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
11939 * config/i386/i386.c (ix86_option_override_internal): Always
11940 allow -mpreferred-stack-boundary=3 for 64-bit targets.
11942 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11944 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
11946 2017-03-28 Bin Cheng <bin.cheng@arm.com>
11948 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
11949 mark new edge's irreducible flag accordign to it.
11950 (vect_do_peeling): Check loop preheader edge's irreducible flag
11951 and pass it to function slpeel_add_loop_guard.
11953 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
11955 PR tree-optimization/80218
11956 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
11957 Update block frequencies and counts.
11959 2017-03-28 Richard Biener <rguenther@suse.de>
11961 PR tree-optimization/78644
11962 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
11963 of a simplification result we may not use it at all.
11965 2017-03-28 Richard Biener <rguenther@suse.de>
11968 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
11969 without arguments, generate default definition of a SSA name.
11971 2017-03-28 Richard Biener <rguenther@suse.de>
11973 PR middle-end/80222
11974 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
11975 TYPE_REF_CAN_ALIAS_ALL references.
11976 * fold-const.c (fold_indirect_ref_1): Likewise.
11978 2017-03-28 Martin Liska <mliska@suse.cz>
11981 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
11982 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
11984 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11985 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
11987 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
11988 (EXTRA_SPECS): Define.
11989 (SUBTARGET_EXTRA_SPECS): Likewise.
11990 (SUBTARGET_CPP_SPEC): Likewise.
11991 * config/arc/elf.h (EXTRA_SPECS): Renamed to
11992 SUBTARGET_EXTRA_SPECS.
11993 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
11995 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
11997 * config/arc/simdext.md (vst64_insn): Update pattern.
11998 (vld32wh_insn): Likewise.
11999 (vld32wl_insn): Likewise.
12000 (vld64_insn): Likewise.
12001 (vld32_insn): Likewise.
12003 2017-03-28 Marek Polacek <polacek@redhat.com>
12006 * fold-const.c (fold_comparison): Use protected_set_expr_location
12007 instead of SET_EXPR_LOCATION.
12009 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12011 * tree.c (add_expr): Avoid name lookup warning.
12013 2017-03-27 Jeff Law <law@redhat.com>
12015 PR tree-optimization/80216
12016 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12017 function name. Limit recursion depth.
12018 (record_temporary_equivalences): Corresponding changes.
12020 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12022 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12025 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12028 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12030 * cfgcleanup.c (reg_note_cfa_p): New array.
12031 (insns_have_identical_cfa_notes): New function.
12032 (old_insns_match_p): Don't cross-jump in between /f
12033 and non-/f instructions. If both i1 and i2 are frame related,
12034 verify all CFA notes, their order and content.
12036 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12039 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12040 HImode and SImode with zero extend to DImode to one insn.
12041 (bswap<mode>2_extenddi): Likewise.
12042 (bswapsi2_extenddi): Likewise.
12043 (bswaphi2_extendsi): Likewise.
12044 (bswaphi2): Combine bswap HImode and SImode into one insn.
12045 Separate memory insns from swapping register.
12046 (bswapsi2): Likewise.
12047 (bswap<mode>2): Likewise.
12048 (bswaphi2_internal): Delete, no longer used.
12049 (bswapsi2_internal): Likewise.
12050 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12051 store, and gpr<-gpr swap insns.
12052 (bswap<mode>2_store): Likewise.
12053 (bswaphi2_reg): Register only splitter, combine with the splitter.
12054 (bswaphi2 splitter): Likewise.
12055 (bswapsi2_reg): Likewise.
12056 (bswapsi2 splitter): Likewise.
12057 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12058 the insns into load, store, and register/register insns.
12059 (bswapdi2_ldbrx): Likewise.
12060 (bswapdi2_load): Likewise.
12061 (bswapdi2_store): Likewise.
12062 (bswapdi2_reg): Likewise.
12064 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12066 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12067 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12069 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12072 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12074 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12075 special handling for target option conflicts between dform
12076 options (-mpower9-dform, -mpower9-dform-vector,
12077 -mpower9-dform-scalar) and -mno-direct-move.
12079 2017-03-27 Richard Biener <rguenther@suse.de>
12081 PR tree-optimization/80181
12082 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12084 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12086 * config/arc/predicates.md (move_double_src_operand): Replace the
12087 call to move_double_src_operand with a call to address_operand.
12089 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12091 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12092 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12093 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12095 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12097 * config/arc/predicates.md (long_immediate_loadstore_operand):
12098 Consider scaled addresses cases.
12100 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12102 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12103 restored when in interrupt.
12104 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12105 doesn't have delay slot.
12107 2017-03-27 Richard Biener <rguenther@suse.de>
12110 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12111 inlined thunk clones.
12113 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12116 * asan.c (instrument_derefs): Copy over last operand from
12117 original COMPONENT_REF to the new COMPONENT_REF with
12118 DECL_BIT_FIELD_REPRESENTATIVE.
12119 * ubsan.c (instrument_object_size): Likewise.
12121 2017-03-27 Richard Biener <rguenther@suse.de>
12123 PR tree-optimization/80170
12124 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12125 sure DR/SCEV didnt fold in constants we do not see when looking
12126 at the reference base alignment.
12128 2017-03-27 Richard Biener <rguenther@suse.de>
12130 PR middle-end/80171
12131 * gimple-fold.c (fold_ctor_reference): Properly guard against
12132 NULL return value from canonicalize_constructor_val.
12134 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12137 * config/i386/i386.c (ix86_expand_builtin)
12138 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12139 flags reg setting and flags reg using instructions.
12140 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12141 clobbering instructions to zero extend op2.
12143 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12145 * doc/install.texi (Configuration) <--with-aix-soname>:
12146 Update link to AIX ld.
12148 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12150 PR rtl-optimization/80160
12151 PR rtl-optimization/80159
12152 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12153 reg_alternate_class into account.
12155 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12158 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12159 to consider in curr_insn_transform.
12161 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12163 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12164 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12165 and emit_mode_inner.
12167 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12169 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12170 argument to the overloaded builtin variants. Use the new flag to
12171 deprecate certain builtin variants.
12172 * config/s390/s390-builtin-types.def: Add new builtin types.
12173 * config/s390/s390-builtins.h: Support new flags field for
12174 overloaded builtins.
12175 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12176 (s390_macro_to_expand): Enable vector float data type.
12177 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12178 builtins by incrementing the __VEC__ version number.
12179 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12181 (s390_resolve_overloaded_builtin): Emit error messages depending
12182 on the builtin flags.
12183 * config/s390/s390.c (s390_expand_builtin): Support additional
12184 flags argument. Change error message to match the messages
12185 emitted in s390-c.c.
12186 * config/s390/s390.md: New UNSPEC_* constants.
12187 (op_type): Add new instruction types.
12188 * config/s390/vecintrin.h: Add new builtins and test data class
12190 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12191 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12192 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12193 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12194 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12195 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12197 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12198 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12199 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12200 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12202 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12203 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12204 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12205 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12207 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12208 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12209 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12210 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12211 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12212 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12213 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12215 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12216 ("vec_scatter_element<V_HW_4:mode>_DI")
12217 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12218 ("vec_fpint<mode>", "vflls")
12219 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12220 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12221 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12222 ("*vec_cmphe<mode>_cc"): ... these.
12224 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12225 mode constant instead of magic value.
12227 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12229 * config/s390/s390.c (s390_expand_vec_compare): Support other
12230 vector floating point modes than just V2DF.
12231 (s390_expand_vcond): Likewise.
12232 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12233 (s390_cannot_change_mode_class): Prevent mode changes between TF
12234 and V1TF in vector registers.
12235 * config/s390/s390.md (DF, SF): New mode attributes.
12236 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12237 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12238 SFmode support for VRs.
12239 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12241 (VFT, VF_HW): New mode iterators.
12242 (vw, sdx): New mode attributes.
12243 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12244 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12245 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12246 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12247 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12248 also the new vector floating point modes. Renaming to ...
12250 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12251 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12252 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12253 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12254 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12255 ("vec_unordered<mode>"): ... these.
12257 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12258 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12259 ("*vec_extendv2df"): New insn definitions.
12261 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12263 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12264 ("mulditi3_2", "*muldi3_sign"): New patterns.
12265 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12266 rename the pattern definition.
12268 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12270 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12272 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12274 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12276 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12277 instruction if possible.
12278 * config/s390/vector.md (vec_halfnumelts): New mode
12280 ("*vec_vllezlf<mode>"): New pattern.
12282 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12284 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12285 ("popcountv4si2", "popcountv2di2"): Rename to ...
12286 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12287 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12289 ("popcount<mode>2_vxe"): New pattern.
12291 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12293 * common/config/s390/s390-common.c (processor_flags_table): Add
12295 * config.gcc: Add arch12.
12296 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12297 Default to arch12 for unknown CPU model numbers.
12298 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12299 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12300 PROCESSOR_max sanity check.
12301 * config/s390/s390-opts.h (enum processor_type): Add
12303 * config/s390/s390.c (processor_table): Add arch12.
12304 (s390_expand_builtin): Add check for B_VXE flag.
12305 (s390_issue_rate): Add PROCESSOR_ARCH12.
12306 (s390_get_sched_attrmask): Likewise.
12307 (s390_get_unit_mask): Likewise.
12308 (s390_sched_score): Enable z13 scheduling for arch12.
12309 (s390_sched_reorder): Likewise.
12310 (s390_sched_variable_issue): Likewise.
12311 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12313 (s390_tune_attr): Use z13 scheduling also for arch12.
12314 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12315 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12316 (TARGET_VXE_P): New macros.
12317 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12318 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12319 * config/s390/s390.opt: Add arch12 as processor_type.
12321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12323 * config/s390/s390.md
12324 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12325 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12326 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12328 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12329 Rename expanders to ...
12331 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12332 ("fixuns_truncdddi2_emu"): ... these.
12334 ("fixuns_trunc<mode>si2_emu"): New expander.
12336 ("*fixuns_truncdfdi2_z13"): Rename to ...
12337 ("*fixuns_truncdfdi2_vx"): ... this.
12339 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12341 * config/s390/2964.md: Remove the single element vector compare
12342 instructions which are no longer used.
12343 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12345 (s390_canonicalize_comparison): Remove handling of DFmode
12347 (s390_expand_vec_compare_scalar): Remove function.
12348 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12349 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12351 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12353 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12355 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12357 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12358 will anyway by matched by mov<mode>_64dfp.
12360 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12362 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12363 vlef/vstef. Add missing operand to vleif.
12365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12367 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12368 pair for all vector types with 64 bit elements.
12369 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12370 * config/s390/vector.md (V_HW_64): ... here.
12371 (V_128_NOSINGLE): New mode iterator.
12372 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12373 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12374 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12375 ("*vec_load_pairv2di"): Change to ...
12376 ("*vec_load_pair<mode>"): ... this one.
12378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12380 * config/s390/constraints.md: Add comments.
12381 (jKK): Reject element sizes > 8 bytes.
12382 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12384 * config/s390/s390.md: Add the s_operand checks formerly in
12385 s390_split_ok_p to various splitters where they are still
12387 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12388 for 128 bit vectors. Plus two splitters.
12390 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12392 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12398 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12399 error if the boundary argument is not constant.
12401 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12403 PR rtl-optimization/80112
12404 * loop-doloop.c (doloop_condition_get): Don't check condition
12405 if cmp isn't SET with IF_THEN_ELSE src.
12407 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12409 PR tree-optimization/80158
12410 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12411 replacing a candidate statement, also replace it for the
12412 candidate's alternate interpretation.
12413 (replace_rhs_if_not_dup): Likewise.
12414 (replace_one_candidate): Likewise.
12416 2017-03-24 Richard Biener <rguenther@suse.de>
12418 PR tree-optimization/80167
12419 * graphite-isl-ast-to-gimple.c
12420 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12422 (translate_isl_ast_to_gimple::get_rename): Likewise.
12424 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12426 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12427 handling of certain combinations of target options, including the
12428 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12429 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12431 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12434 * config/arm/arm.md (*load_multiple): Add reload_completed to
12435 matching condition.
12437 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12438 Richard Biener <rguenth@suse.de>
12440 PR tree-optimization/79908
12441 PR tree-optimization/80136
12442 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12443 been cast away, gimplify_and_add suffices.
12445 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12447 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12449 2017-03-23 Richard Biener <rguenther@suse.de>
12451 PR tree-optimization/80032
12452 * gimplify.c (gimple_push_cleanup): Forced unconditional
12453 cleanups still have to go to the conditional_cleanups
12456 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12458 PR tree-optimization/80072
12459 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12461 (next_operand_entry_id): Change type to unsigned int.
12462 (sort_by_operand_rank): Make sure to return the right return value
12463 even if unsigned fields are bigger than INT_MAX.
12464 (struct oecount): Change cnt and id type to unsigned int.
12465 (oecount_hasher::equal): Formatting fix.
12466 (oecount_cmp): Make sure to return the right return value
12467 even if unsigned fields are bigger than INT_MAX.
12468 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12471 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12472 TREE_READONLY on result if writing it more than once.
12475 * doc/invoke.texi (-fsanitize=thread): Document that with
12476 -fnon-call-exceptions atomics are not able to throw
12480 * tsan.c: Include tree-eh.h.
12481 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12482 maybe_clean_or_replace_eh_stmt where needed.
12483 (instrument_memory_accesses): Add cfg_changed argument.
12484 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12485 if it returned true.
12486 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12488 PR rtl-optimization/63191
12489 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12490 wrapper function, moved the whole old content into ...
12491 (ix86_delegitimize_address_1): ... this. New inline function.
12492 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12493 true as last argument instead of ix86_delegitimize_address.
12495 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12497 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12498 cortexa57_branch_cost.
12500 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12502 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12504 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12507 * doc/md.texi (Constraints): Document wA constraint.
12508 * config/rs6000/constraints.md (wA): New.
12509 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12510 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12511 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12512 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12514 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12517 * gimplify.c (is_oacc_declared): New function.
12518 (oacc_default_clause): Use it to set default flags for acc declared
12519 variables inside parallel regions.
12520 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12521 declared variables.
12522 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12523 declare attribute to any decl as necessary.
12525 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12528 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
12529 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
12530 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
12531 (arm_arch_lpae): This.
12532 * config/arm/arm.c (arm_arch7ve): Rename into ...
12533 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
12534 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
12537 2017-03-22 Martin Liska <mliska@suse.cz>
12540 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
12541 error message instead of an ICE.
12543 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12545 * doc/extend.texi (6.11 Additional Floating Types): Revise.
12547 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
12549 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12551 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12554 2017-03-21 Martin Sebor <msebor@redhat.com>
12556 * doc/extend.texi: Use "cannot" instead of "can't."
12557 * doc/hostconfig.texi: Same.
12558 * doc/install.texi: Same.
12559 * doc/invoke.texi: Same.
12560 * doc/loop.texi: Same.
12561 * doc/md.texi: Same.
12562 * doc/objc.texi: Same.
12563 * doc/rtl.texi: Same.
12564 * doc/tm.texi: Same.
12565 * doc/tm.texi.in: Same.
12566 * doc/trouble.texi: Same.
12568 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
12571 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
12572 (collect_checksum_attributes): Set it.
12573 (die_checksum_ordered): Use it.
12575 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12577 PR tree-optimization/79908
12578 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
12579 change: For a VA_ARG whose LHS has been cast away, use
12580 force_gimple_operand to construct the side effects.
12582 2017-03-21 David Malcolm <dmalcolm@redhat.com>
12584 PR translation/80001
12585 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
12586 more amenable to translation.
12587 (oacc_loop_auto_partitions): Likewise.
12589 2017-03-21 Marek Polacek <polacek@redhat.com>
12590 Martin Sebor <msebor@redhat.com>
12592 PR tree-optimization/80109
12593 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
12594 on INTEGRAL_TYPE_P.
12596 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12597 Segher Boessenkool <segher@kernel.crashing.org>
12600 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
12601 check reg_used_between_p between insn and one of succ or succ2
12602 depending on if succ is artificial insn not inserted into insn
12605 2017-03-21 Martin Liska <mliska@suse.cz>
12607 PR gcov-profile/80081
12608 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
12609 * doc/gcc.texi: Include gcov-dump stuff.
12610 * doc/gcov-dump.texi: New file.
12612 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
12614 PR rtl-optimization/79150
12615 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
12616 conditional jump, if the jump is the last insn of the loop.
12618 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12619 Richard Biener <rguenth@suse.de>
12621 PR tree-optimization/79908
12622 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12623 been cast away, use force_gimple_operand to construct the side
12626 2017-03-21 Martin Liska <mliska@suse.cz>
12628 PR libfortran/79956
12629 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
12632 2017-03-21 Brad Spengler <spender@grsecurity.net>
12635 * plugin.c (htab_hash_plugin): New function.
12636 (add_new_plugin): Use it and adjust.
12637 (parse_plugin_arg_opt): Adjust.
12638 (init_one_plugin): Likewise.
12640 2017-03-21 Richard Biener <rguenther@suse.de>
12642 PR tree-optimization/80032
12643 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
12644 if set force the cleanup to happen unconditionally.
12645 (gimplify_target_expr): Push inserted clobbers with force_uncond
12646 to avoid them being removed by control-dependent DCE.
12648 2017-03-21 Richard Biener <rguenther@suse.de>
12650 PR tree-optimization/80122
12651 * tree-inline.c (copy_bb): Do not expans va-arg packs or
12652 va_arg_pack_len when the inlined call stmt requires pack
12654 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
12656 2017-03-21 Jakub Jelinek <jakub@redhat.com>
12659 * tsan.c (instrument_builtin_call): If the memory model argument
12660 is not a constant, assume it is valid.
12663 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
12666 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12668 PR rtl-optimization/79910
12669 * combine.c (can_combine_p): Do not allow combining an I0 or I1
12670 if its dest is used by an insn before I2 (other than the combined
12671 insns themselves, which are properly handled already).
12673 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
12676 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12678 * combine.c (record_used_regs): New static function.
12679 (try_combine): Handle situations where there is an additional
12680 instruction between I2 and I3 which needs to have a LOG_LINK
12684 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12686 * combine.c (try_combine): Delete redundant i1 test. Call
12687 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12689 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12692 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
12693 alternatives 13/14.
12695 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12697 PR tree-optimization/80054
12698 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
12699 the optimization if a PHI or any of its arguments is not dominated
12700 by the candidate's basis. Use gphi* rather than gimple* as
12702 (replace_profitable_candidates): Clean up a gimple* variable that
12703 should be a gphi* variable.
12705 2017-03-20 Martin Sebor <msebor@redhat.com>
12708 * doc/extend.texi (attribute constructor): Document present limitation.
12710 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12713 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
12714 __POWER9_VECTOR__ #ifdef control, change template definition to
12715 use Power9-specific built-in function.
12716 (vec_any_eq): Likewise.
12717 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
12718 to control outcomes from this test.
12719 (vector_ae_<mode>p): For VEC_F modes, likewise.
12721 2017-03-20 Ian Lance Taylor <iant@google.com>
12723 * config/i386/i386.c (ix86_function_regparm): Save an extra
12724 register for -fsplit-stack with DECL_STATIC_CHAIN.
12726 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12729 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
12730 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
12732 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
12734 * config/riscv/riscv.c (riscv_print_operand): Use "fence
12736 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
12739 2017-03-20 Marek Polacek <polacek@redhat.com>
12742 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
12744 2017-03-20 Richard Biener <rguenther@suse.de>
12746 PR tree-optimization/80113
12747 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
12748 allocate extra SSA name for PHI def.
12749 (add_close_phis_to_outer_loops): Likewise.
12750 (add_close_phis_to_merge_points): Likewise.
12751 (copy_loop_close_phi_args): Likewise.
12752 (copy_cond_phi_nodes): Likewise.
12754 2017-03-20 Martin Liska <mliska@suse.cz>
12756 PR middle-end/79753
12757 * tree-chkp.c (chkp_build_returned_bound): Do not build
12758 returned bounds for a LHS that's not a BOUNDED_P type.
12760 2017-03-20 Martin Liska <mliska@suse.cz>
12764 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
12765 COMPLEX_CST and VECTOR_CST.
12767 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12770 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
12771 target operand. A new splitter adds the clobber statement in case
12772 the target operand is dead anyway.
12774 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
12776 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
12777 to age-old versions of binutils and glibc.
12779 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
12781 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
12783 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12785 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
12787 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
12789 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
12790 requirement for binutils 2.13.
12792 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
12794 * combine.c (try_combine): Delete redundant i1 test. Call
12795 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
12797 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
12799 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
12800 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
12802 <riscv64-*-elf>: Re-arrange section
12803 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
12804 <riscv32-*-linux>: Likewise.
12805 <riscv64-*-elf>: Likewise
12806 <riscv64-*-linux>: Likewise.
12808 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
12811 * aarch64.opt(verbose-cost-dump): Fix typo.
12813 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
12816 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
12817 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
12819 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
12821 * reload.c (find_reloads): When reloading a nonoffsettable address,
12822 use RELOAD_OTHER for it and its address reloads.
12824 PR rtl-optimization/79910
12825 * combine.c (record_used_regs): New static function.
12826 (try_combine): Handle situations where there is an additional
12827 instruction between I2 and I3 which needs to have a LOG_LINK
12830 2017-03-17 Jeff Law <law@redhat.com>
12832 PR tree-optimization/71437
12833 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
12834 conditional in the hash table first.
12835 (vrp_dom_walker::before_dom_children): Extract condition from
12836 ASSERT_EXPR. Record condition, its inverion and any implied
12837 conditions as well.
12839 2017-03-17 Marek Polacek <polacek@redhat.com>
12840 Markus Trippelsdorf <markus@trippelsdorf.de>
12842 PR tree-optimization/80079
12843 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
12846 2017-03-17 Richard Biener <rguenther@suse.de>
12848 PR middle-end/80075
12849 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
12850 Properly verify the LHS before the RHS possibly claims to be
12852 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
12855 2017-03-17 Martin Jambor <mjambor@suse.cz>
12857 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
12858 (List of -O2 options): Likewise.
12859 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
12862 2017-03-17 Tom de Vries <tom@codesourcery.com>
12864 * gcov-dump.c (print_usage): Print bug_report_url.
12866 2017-03-17 Richard Biener <rguenther@suse.de>
12868 PR middle-end/80050
12869 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
12870 (parser::peek): Likewise.
12872 2017-03-17 Richard Biener <rguenther@suse.de>
12874 PR tree-optimization/80048
12875 * sese.c (free_sese_info): Properly release rename_map and
12876 copied_bb_map elements.
12878 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
12880 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
12881 Add linked-list forward and backlinks. Insert on
12882 construction, remove on destruction.
12883 (class pass_store_merging): Add m_stores_head field.
12884 (pass_store_merging::terminate_and_process_all_chains):
12885 Iterate over m_stores_head list.
12886 (pass_store_merging::terminate_all_aliasing_chains):
12888 (pass_store_merging::execute): Check for debug stmts first.
12889 Push new chains onto the m_stores_head stack.
12891 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
12894 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
12895 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
12896 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
12898 2017-03-16 Jeff Law <law@redhat.com>
12900 PR tree-optimization/71437
12901 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
12902 member function. Implementation moved into after_dom_children
12903 member function and into the threader's thread_outgoing_edges
12905 (dom_opt_dom_walker::after_dom_children): Simplify by moving
12906 some code into new thread_outgoing_edges.
12907 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
12908 definition. Simplify marker handling (do it here). Assume we always
12909 have the available expression and the const/copies tables.
12910 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
12912 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
12913 * tree-vrp.c (equiv_stack): No longer file scoped.
12914 (vrp_dom_walker): New class.
12915 (vrp_dom_walker::before_dom_children): New member function.
12916 (vrp_dom_walker::after_dom_children): Likewise.
12917 (identify_jump_threads): Setup domwalker. Use it rather than
12918 walking edges in a random order by hand. Simplify setup/finalization.
12919 (finalize_jump_threads): Remove.
12920 (vrp_finalize): Do not call identify_jump_threads here.
12921 (execute_vrp): Do it here instead and call thread_through_all_blocks
12924 PR tree-optimization/71437
12925 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
12927 (simplify_stmt_for_jump_threading): Add basic_block argument. All
12929 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
12930 (dom_opt_dom_walker::thread_across_edge): Remove
12931 handle_dominating_asserts argument. All callers changed.
12932 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
12933 changes. Remove calls to lhs_of_dominating_assert. Other
12934 uses of handle_dominating_asserts turn into unconditional code
12935 (simplify_control_stmt_condition_1): Likewise.
12936 (simplify_control_stmt_condition): Likewise.
12937 (thread_through_normal_block, thread_across_edge): Likewise.
12938 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
12939 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
12940 object if it is not an SSA_NAME.
12941 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
12942 before calling into the VRP specific simplifiers.
12943 (identify_jump_threads): Remove handle_dominating_asserts
12946 2017-03-16 Jakub Jelinek <jakub@redhat.com>
12949 * tree-diagnostic.c (default_tree_printer): No longer static.
12950 * tree-diagnostic.h (default_tree_printer): New prototype.
12952 2017-03-16 Tamar Christina <tamar.christina@arm.com>
12954 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
12955 Change ins into fmov.
12957 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12959 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
12960 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
12961 Use h_con constraint for operand 1.
12962 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
12963 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
12965 2017-03-15 Jeff Law <law@redhat.com>
12967 PR tree-optimization/71437
12968 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
12969 (record_temporary_equivalences): Use it.
12971 PR tree-optimization/71437
12972 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
12973 tree-ssa-scopedtables.
12974 (lookup_avail_expr, build_and_record_new_cond): Likewise.
12975 (record_conditions, record_cond, vuse_eq): Likewise.
12976 (record_edge_info): Adjust to API tweak of record_conditions.
12977 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
12978 (record_temporary_equivalences, optimize_stmt): Likewise.
12979 (eliminate_redundant_computations): Likewise.
12980 (record_equivalences_from_stmt): Likewise.
12981 * tree-ssa-scopedtables.c: Include options.h and params.h.
12982 (vuse_eq): New function, moved from tree-ssa-dom.c
12983 (build_and_record_new_cond): Likewise.
12984 (record_conditions): Likewise. Accept vector of conditions rather
12985 than edge_equivalence structure for first argument.
12986 for the first argument.
12987 (avail_exprs_stack::lookup_avail_expr): New member function, moved
12988 from tree-ssa-dom.c.
12989 (avail_exprs_stack::record_cond): Likewise.
12990 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
12991 from tree-ssa-dom.c.
12992 (avail_exprs_stack): Add new member functions lookup_avail_expr
12994 (record_conditions): Declare.
12996 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
12999 * lra-constraints.c (process_alt_operands): Increase reject for
13000 reloading an input/output operand.
13002 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13005 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13006 insns to convert from signed/unsigned char/short to IEEE 128-bit
13008 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13010 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13013 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13014 subreg of inner mode for values already in registers.
13016 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13018 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13019 iteration reg is used after the loop.
13021 2017-03-14 Martin Sebor <msebor@redhat.com>
13023 PR tree-optimization/79800
13024 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13025 precision in negative-positive range.
13026 (format_floating): Call non-const overload with adjusted precision.
13028 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13031 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13034 2017-03-14 Martin Sebor <msebor@redhat.com>
13036 PR middle-end/80020
13037 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13038 * builtins.def (aligned_alloc): Use it.
13041 * Makefile.in (GTFILES): Add calls.c.
13042 * calls.c: Include "gt-calls.h".
13044 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13046 PR rtl-optimization/79728
13047 * regs.h (struct target_regs): New field
13048 x_contains_allocatable_regs_of_mode.
13049 (contains_allocatable_regs_of_mode): New macro.
13050 * reginfo.c (init_reg_sets_1): Initialize it, and change
13051 contains_reg_of_mode so it includes global regs as well.
13052 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13053 rather than contains_regs_of_mode.
13055 2017-03-14 Martin Liska <mliska@suse.cz>
13057 * doc/invoke.texi: Document options that can't be combined with
13058 -fcheck-pointer-bounds.
13060 2017-03-14 Martin Liska <mliska@suse.cz>
13062 PR middle-end/79831
13063 * doc/invoke.texi (-Wchkp): Document the option.
13065 2017-03-14 Martin Liska <mliska@suse.cz>
13067 * Makefile.in: Install gcov-dump.
13069 2017-03-14 Martin Liska <mliska@suse.cz>
13071 * multiple_target.c (expand_target_clones): Bail out for
13072 an invalid attribute.
13074 2017-03-14 Richard Biener <rguenther@suse.de>
13076 * alias.c (struct alias_set_entry): Pack properly.
13077 * cfgloop.h (struct loop): Likewise.
13078 * cse.c (struct set): Likewise.
13079 * ipa-utils.c (struct searchc_env): Likewise.
13080 * loop-invariant.c (struct invariant): Likewise.
13081 * lra-remat.c (struct cand): Likewise.
13082 * recog.c (struct change_t): Likewise.
13083 * rtl.h (struct address_info): Likewise.
13084 * symbol-summary.h (function_summary): Likewise.
13085 * tree-loop-distribution.c (struct partition): Likewise.
13086 * tree-object-size.c (struct object_size_info): Likewise.
13087 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13088 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13089 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13090 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13091 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13092 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13093 (struct _stmt_vec_info): Likewise.
13095 2017-03-14 Martin Liska <mliska@suse.cz>
13098 * multiple_target.c (create_dispatcher_calls): Check that
13099 a target can create a function dispatcher.
13101 2017-03-14 Martin Liska <mliska@suse.cz>
13104 * multiple_target.c (expand_target_clones): Drop local.local
13105 flag for default implementation.
13107 2017-03-14 Richard Biener <rguenther@suse.de>
13109 PR tree-optimization/80030
13110 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13112 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13114 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13115 gcc_fallthrough() instead of __attribute__((fallthrough));
13117 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13119 * doc/gcc.texi: Remove "up" link to (DIR).
13120 * doc/gccint.texi: Ditto.
13122 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13124 * doc/install.texi (Specific) <avr>: Remove reference to
13127 2017-03-13 Jeff Law <law@redhat.com>
13129 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13130 attribute rather than comments.
13132 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13133 match_scratch operand is highest.
13135 2017-03-13 Martin Liska <mliska@suse.cz>
13137 PR middle-end/78339
13138 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13139 is a CHKP clone, use original declaration.
13141 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13143 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13144 (arc_conditional_register_usage): Use a different allocation order
13145 when optimizing for size.
13146 * common/config/arc/arc-common.c (arc_option_optimization_table):
13147 Section anchors default on when optimizing for size.
13149 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13151 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13153 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13155 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13156 * config/arc/arc.md (cpu_facility): Add cd variant.
13157 (*movqi_insn): Add code density variant.
13158 (*movhi_insn): Likewise.
13159 (*movqi_insn): Likewise.
13160 (*addsi3_mixed): Likewise.
13161 (subsi3_insn): Likewise.
13163 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13165 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13167 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13169 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13170 expressions with MINUS and UNARY ops.
13172 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13175 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13177 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13178 between vec_select and vector argument.
13179 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13180 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13181 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13182 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13183 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13184 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13186 2017-03-13 Richard Biener <rguenther@suse.de>
13189 * params.def (vect-max-peeling-for-alignment): Fix typo.
13191 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13193 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13194 issue that only occurred with binutils below 2.18.
13196 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13198 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13199 refer to binutils 2.11/2.12 minimum.
13201 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13203 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13204 ftp.kernel.org and simplify binutils requirement.
13206 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13208 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13210 (Optimize Options): Ditto. Also remove redundancy.
13212 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13214 PR translation/79848
13215 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13217 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13218 to G_ to avoid double translation.
13220 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13222 PR translation/79923
13223 * auto-profile.c (get_combined_location): Convert leading
13224 character of diagnostics to lower case and remove trailing period.
13225 (read_profile): Likewise for various diagnostics.
13226 * config/arm/arm.c (arm_option_override): Remove trailing period
13227 from various diagnostics.
13228 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13229 (msp430_expand_delay_cycles): Likewise.
13231 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13234 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13235 full command-line argument, rather than just "str".
13236 (aarch64_validate_march): Likewise.
13237 (aarch64_validate_mtune): Likewise.
13239 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13241 PR rtl-optimization/78911
13242 * lra-assigns.c (must_not_spill_p): New function.
13243 (spill_for): Use it.
13245 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13247 PR tree-optimization/79981
13248 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13249 ATOMIC_COMPARE_EXCHANGE ifn result.
13250 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13251 IFN_ATOMIC_COMPARE_EXCHANGE.
13253 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13256 * opts.c (parse_sanitizer_options): Add missing question mark to
13257 "did you mean" message.
13259 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13261 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13263 (VMULEUH_UNS): Likewise.
13264 (VMULOUB_UNS): Likewise.
13265 (VMULOUH_UNS): Likewise.
13266 * config/rs6000/rs6000.c (builtin_function_type): Remove
13267 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13269 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13272 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13273 x with result of extra_parsing_for_operand_code_0.
13274 (function_reader::extra_parsing_for_operand_code_0): Convert
13275 return type from void to rtx, returning x. When reading
13276 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13277 larger size containing struct block_symbol.
13279 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13281 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13282 -mfloat128-hardware without -m64.
13284 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13287 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13288 entries to the case statement that marks unsigned arguments to
13289 overloaded functions.
13291 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13293 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13294 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13296 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13299 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13300 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13302 2017-03-10 Martin Liska <mliska@suse.cz>
13306 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13307 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13310 2017-03-10 Olivier Hainque <hainque@adacore.com>
13312 * tree-switch-conversion (array_value_type): Start by resetting
13313 candidate type to it's main variant.
13315 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13317 PR rtl-optimization/79909
13318 * combine.c (try_combine): Use simplify_replace_rtx on individual
13319 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13320 of the whole CALL_INSN_FUNCTION_USAGE.
13322 PR tree-optimization/79972
13323 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13324 get_range_info on SSA_NAMEs. Formatting fixes.
13326 2017-03-10 Richard Biener <rguenther@suse.de>
13327 Jakub Jelinek <jakub@redhat.com>
13329 PR tree-optimization/77975
13330 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13331 edge to be constant.
13332 (get_val_for): For constant x return it. Formatting fix.
13333 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13334 would use the same bases as the current one.
13336 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13338 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13339 instead of vec_select for V1TImode.
13340 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13342 (VSX_LE_128): Add V1TI to this mode iterator.
13343 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13344 (*vsx_le_perm_store_<mode>): Likewise.
13345 (pre-reload splitter for VSX stores): Likewise.
13346 (post-reload splitter for VSX stores): Likewise.
13347 (*vsx_xxpermdi2_le_<mode>): Likewise.
13348 (*vsx_lxvd2x2_le_<mode>): Likewise.
13349 (*vsx_stxvd2x2_le_<mode>): Likewise.
13351 2017-03-09 Michael Eager <eager@eagercon.com>
13353 Correct failures with --enable-checking=yes,rtl.
13355 * config/microblaze/microblaze.c (microblaze_expand_shift):
13356 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13357 test for const0_rtx.
13358 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13359 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13361 2017-03-09 Richard Biener <rguenther@suse.de>
13363 PR tree-optimization/79977
13364 * graphite-scop-detection.c (scop_detection::merge_sese):
13365 Handle the case of extra exits to blocks dominating the entry.
13367 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13369 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13372 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13374 PR rtl-optimization/79949
13375 * lra-constraints.c (process_alt_operands): Check memory when
13376 trying to predict a cycle. Print about the overall increase.
13378 2017-03-09 Richard Biener <rguenther@suse.de>
13380 PR middle-end/79971
13381 * gimple-expr.c (useless_type_conversion_p): Preserve
13382 TYPE_SATURATING for fixed-point types.
13384 2017-03-09 Richard Biener <rguenther@suse.de>
13387 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13388 alignment of BLKmode params.
13390 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13393 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13394 (VALL_NO_V2Q): Likewise.
13396 * config/aarch64/aarch64-simd.md
13397 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13399 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13400 VALL_NO_V2Q mode iterator.
13401 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13403 2017-03-09 Martin Liska <mliska@suse.cz>
13405 PR tree-optimization/79631
13406 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13407 tree_int_cst_sign_bit just for INTEGER constants.
13409 2017-03-09 Martin Liska <mliska@suse.cz>
13413 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13416 2017-03-09 Marek Polacek <polacek@redhat.com>
13419 * tree.c (inchash::add_expr): Handle TREE_VEC.
13421 2017-03-09 Martin Liska <mliska@suse.cz>
13424 (chkp_narrow_size_and_offset): New function.
13425 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13426 (void chkp_parse_bit_field_ref): New function.
13427 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13428 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13430 2017-03-09 Martin Liska <mliska@suse.cz>
13433 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13434 (chkp_find_bounds_1): Remove gcc_unreachable.
13436 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13439 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13440 BUILT_IN_SYNC*, determine the access type from the size suffix and
13441 always build a MEM_REF with that type. Handle forgotten
13442 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13445 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13446 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13447 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13448 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13449 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13450 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13451 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13452 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13453 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13454 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13455 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13456 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13457 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13458 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13459 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13460 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13461 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13462 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13463 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13464 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13465 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13466 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13467 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13468 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13469 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13470 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13471 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13472 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13473 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13474 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13475 definitions outside of __OPTIMIZE__ guarded section.
13478 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13479 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13480 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13481 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13484 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13486 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13487 ("vfenez<mode>"): Add missing constraints.
13489 2017-03-08 Martin Sebor <msebor@redhat.com>
13492 * config/nds32/nds32.c (nds32_option_override):
13493 Fix misspelled diagnostic.
13495 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13498 * gimplify.c (gimplify_omp_for): Replace index var in outer
13499 taskloop statement with an artificial variable and add
13500 OMP_CLAUSE_PRIVATE clause for it.
13502 2017-03-08 Richard Biener <rguenther@suse.de>
13504 PR tree-optimization/79955
13505 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13506 for accesses that are completely outside of the variable.
13508 2017-03-08 Andrew Haley <aph@redhat.com>
13510 PR tree-optimization/79943
13511 * tree-ssa-loop-split.c (compute_new_first_bound): When
13512 calculating the new upper bound, (END-BEG) should be added, not
13515 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13517 * config/avr/avr.md (setmemhi): Make sure match_dup
13518 operand number comes before match_scratch.
13520 2017-03-08 Richard Biener <rguenther@suse.de>
13522 PR tree-optimization/79920
13523 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13524 with ncopies == 1 to ...
13525 (vect_transform_slp_perm_load): ... here. Properly compute
13526 all element loads by iterating VF times over the group. Do
13527 not handle ncopies (computed in a broken way) in
13528 vect_create_mask_and_perm.
13530 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13533 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
13534 is a uniform vector, use uniform_vector_p return value instead of
13535 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
13537 2017-03-07 Marek Polacek <polacek@redhat.com>
13539 PR middle-end/79809
13540 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
13541 (alloca_call_type): Likewise.
13543 2017-03-07 Martin Liska <mliska@suse.cz>
13545 * gcov.c (process_args): Put comment to correct location.
13547 2017-03-07 Martin Liska <mliska@suse.cz>
13549 PR middle-end/68270
13550 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
13551 Use array_at_struct_end_p instead of DECL_CHAIN (field).
13552 (chkp_narrow_bounds_for_field): Likewise.
13553 (chkp_parse_array_and_component_ref): Pass one more argument to
13556 2017-03-07 Richard Biener <rguenther@suse.de>
13558 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
13561 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
13563 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
13564 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
13566 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13569 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
13570 to end of description.
13571 (PARAM_MAX_STORES_TO_MERGE): Likewise.
13573 2017-03-07 Jakub Jelinek <jakub@redhat.com>
13575 PR rtl-optimization/79901
13576 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
13578 (*avx512f_<code><mode>3<mask_name>): ... this.
13579 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
13580 iterator instead of VI8_AVX2_AVX512BW.
13582 PR rtl-optimization/79901
13583 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
13584 min/max expander, expand it using expand_vec_cond_expr.
13587 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
13590 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13593 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
13594 to void * for PCH reasons.
13595 * dwarf2out.c (output_loc_operands, output_die): Cast
13596 v.val_vec.array to unsigned char *.
13598 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
13601 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
13604 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
13606 PR rtl-optimization/79571
13607 * lra-constraints.c (process_alt_operands): Calculate static
13608 reject and subtract it from overall when only addresses will be
13611 2017-03-06 Julia Koval <julia.koval@intel.com>
13614 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
13615 incoming stack boundary to 128 for 64-bit targets.
13617 2017-03-06 Richard Biener <rguenther@suse.de>
13619 PR tree-optimization/79894
13620 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
13621 to NULL after folding it.
13623 2017-03-06 Richard Biener <rguenther@suse.de>
13625 PR tree-optimization/79824
13626 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
13627 check disabling peeling for gaps.
13629 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
13631 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13632 attributes): Document gettimeofday.
13634 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13636 * config/s390/s390.c (s390_option_override_internal): Set
13637 PARAM_MIN_VECT_LOOP_BOUND
13639 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
13641 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
13642 * config/s390/s390.md: Likewise.
13644 2017-03-06 Jakub Jelinek <jakub@redhat.com>
13647 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
13648 (<avx2_avx512>_perm<mode>): Rename to ...
13649 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
13651 (<avx512>_perm<mode>_mask): Rename to ...
13652 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
13654 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
13655 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
13656 instead of VI8F_256_512.
13657 (avx512f_perm<mode>): New define_expand.
13658 (avx512f_perm<mode>_mask): Likewise.
13659 (avx512f_perm<mode>_1<mask_name>): New define_insn.
13660 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
13662 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13664 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
13665 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
13667 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
13669 2017-03-06 Martin Liska <mliska@suse.cz>
13672 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
13673 when having a SSA NAME w/o VAR_DECL assigned to it.
13675 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13677 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
13678 msa_dpsub_<su>_d): Fix MODE for vec_select.
13680 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
13682 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
13684 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
13686 2017-03-06 Richard Biener <rguenther@suse.de>
13688 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
13689 * plugin.c (register_plugin_info): Likewise.
13690 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
13692 2017-03-05 Jakub Jelinek <jakub@redhat.com>
13694 * config/i386/sse.md (sse_storehps, sse_storelps,
13695 avx_<castmode><avxsizesuffix>_<castmode>,
13696 avx512f_<castmode><avxsizesuffix>_<castmode>,
13697 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
13698 in condition that at least one operand is not a MEM.
13700 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13702 PR middle-end/79805
13703 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
13704 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
13706 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
13707 gimple_call_nothrow_p flag based on whether original builtin can throw.
13708 If it can, emit following stmts on the fallthrough edge.
13709 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
13710 don't create new bb if inserting just debug stmts on the edge, try to
13711 insert them on the fallthru bb or just reset debug stmts.
13713 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
13716 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
13717 restore recog_data (including the operand rtxes inside it) around
13718 the call to get_insn_template.
13720 2017-03-03 Martin Sebor <msebor@redhat.com>
13722 PR tree-optimization/79699
13723 * context.c (context::~context): Free MPFR caches to avoid
13724 a memory leak on program exit.
13726 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13728 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
13729 Use wide_int::ulow () instead of .elt (0).
13731 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13733 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
13734 (*pushxf): Limit oF constraint to 32bit targets and add oC
13735 constraint for 64bit targets.
13736 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
13737 (*pushdf): Change rmF constraint to rmC.
13739 2017-03-03 Martin Liska <mliska@suse.cz>
13741 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
13742 Remove unused variable.
13744 2017-03-03 Jakub Jelinek <jakub@redhat.com>
13747 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
13748 is a memory operand, increase num_memory.
13749 (ix86_expand_args_builtin): Likewise.
13751 2017-03-03 Jan Hubicka <jh@suse.cz>
13754 * ipa-devirt.c (maybe_record_node): Properly handle
13755 __cxa_pure_virtual visibility.
13757 2017-03-03 Martin Liska <mliska@suse.cz>
13759 PR tree-optimization/79803
13760 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
13762 (pass_loop_prefetch::execute): Disabled optimization if an
13763 assumption about L1 cache size is not met.
13765 2017-03-03 Martin Liska <mliska@suse.cz>
13767 PR rtl-optimization/79574
13768 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
13769 (hash_scan_set): Likewise.
13770 (dump_hash_table): Likewise.
13771 (hoist_code): Likewise.
13773 2017-03-03 Richard Biener <rguenther@suse.de>
13775 * fixed-value.c (fixed_from_string): Restore use of elt (1)
13776 in place of uhigh ().
13777 (fixed_convert_from_real): Likewise.
13779 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
13782 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
13784 2017-03-03 Richard Biener <rguenther@suse.de>
13786 PR middle-end/79818
13787 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
13788 TYPE_OVERFLOW_UNDEFINED check.
13790 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13792 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
13794 (vector_ae_<mode>_p): Likewise.
13795 (vector_nez_<mode>_p): Likewise.
13796 (vector_ne_v2di_p): Likewise.
13797 (vector_ae_v2di_p): Likewise.
13798 (vector_ne_<mode>_p): Likewise.
13799 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
13801 (vsx_tsqrt<mode>2_fe): Likewise.
13803 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
13806 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
13808 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13810 PR rtl-optimization/79780
13811 * cprop.c (one_cprop_pass): When second and further conditional trap
13812 in a single basic block is turned into an unconditional trap, turn it
13813 into a deleted note to avoid RTL verification failures.
13815 2017-03-02 Richard Biener <rguenther@suse.de>
13817 * fold-const.c (const_binop): Use ulow () instead of elt (0).
13819 2017-03-02 Richard Biener <rguenther@suse.de>
13821 PR tree-optimization/79345
13823 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
13824 param and abort the walk, returning -1 if it is hit.
13825 (walk_aliased_vdefs): Take a limit param and pass it on.
13826 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
13827 defaulting to 0 and return a signed int.
13828 * tree-ssa-uninit.c (struct check_defs_data): New struct.
13829 (check_defs): New helper.
13830 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
13831 about uninitialized memory.
13832 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
13833 bogus uninitialized warning.
13834 (fixed_convert_from_real): Likewise.
13836 2017-03-02 Bin Cheng <bin.cheng@arm.com>
13838 PR tree-optimization/66768
13839 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
13840 iv_use if base object can't be determined.
13842 2017-03-02 Jakub Jelinek <jakub@redhat.com>
13844 PR tree-optimization/79345
13845 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
13846 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
13847 (get_pattern_stats): Initialize it.
13848 * genemit.c (gen_expand): Verify match_scratch numbers come after
13849 match_operand/match_dup numbers.
13850 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
13851 match_scratch numbers.
13852 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
13854 * config/s390/s390.md (trunctdsd2): Likewise.
13856 2017-03-02 Richard Biener <rguenther@suse.de>
13858 * wide-int.h (wide_int_storage::operator=): Implement in terms
13861 2017-03-02 Richard Biener <rguenther@suse.de>
13863 PR tree-optimization/79777
13864 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
13865 the to insert expression to sth existing.
13867 2017-03-01 Martin Sebor <msebor@redhat.com>
13869 PR middle-end/79692
13870 * gimple-ssa-sprintf.c
13871 (directive::known_width_and_precision): New function.
13872 (format_integer): Use it.
13873 (get_mpfr_format_length): Consider the full range of precision
13874 when computing %g output with the # flag. Set the likely byte
13875 count to 3 rather than 1 when precision is indeterminate.
13876 (format_floating): Correct the lower bound of precision.
13878 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13880 * doc/invoke.texi: Document default code model for 64-bit Linux.
13882 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13885 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
13886 udiv rather than div since input pattern is unsigned.
13888 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
13890 * config/i386/i386.c (print_reg): Warn for values of
13891 unsupported size in integer register.
13893 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
13896 * config/rs6000/predicates.md (current_file_function_operand): Do
13897 not allow self calls to be local if the function is replaceable.
13899 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
13902 * config/rs6000/altivec.h (vec_ctz and others): Change the
13903 preprocessor macro that controls conditional compilation from
13904 _ARCH_PWR9 to __POWER9_VECTOR__.
13905 (vec_all_ne): Change parameterization of __altivec_scalar_pred
13906 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
13907 control (instead of _ARCH_PWR9 control) so that template
13908 definition uses power9-specific function.
13909 (vec_any_eq): Likewise.
13910 (vec_all_ne): Change macro definition to use a power9-specific
13911 expansion under #ifdef __POWER9_VECTOR__ control (instead of
13912 _ARCH_PWR9 control).
13913 (vec_any_eq) Likewise.
13914 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
13915 expansion for CMPNEF to remove support for xvcmpnesp instruction.
13916 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
13917 support for xvcmpnedp instruction.
13918 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
13919 macro expansion so that Power9 implementation of vec_all_ne does
13920 not use the AltiVec predicate framework.
13921 (VCMPNEH_P): Likewise.
13922 (VCMPNEW_P): Likewise.
13923 (VCMPNED_P): Likewise.
13924 (VCMPNEFP_P): Likewise.
13925 (VCMPNEDP_P): Likewise.
13926 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
13927 implementation of vec_any_eq to not use AltiVec predicate
13929 (VCMPAEH_P): Likewise.
13930 (VCMPAEW_P): Likewise.
13931 (VCMPAED_P): Likewise.
13932 (VCMPAEFP_P): Likewise.
13933 (VCMPAEDP_P): Likewise.
13934 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
13935 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
13936 not use the AltiVec predicate framework.
13937 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
13938 of vec_any_eq to not use AltiVec predicate framework.
13939 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13940 support for predefined __POWER9_VECTOR__ macro to indicate that
13941 Power9 instruction selection is enabled.
13942 (altivec_overloaded_builtins): Remove extraneous
13943 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
13944 function argument types RS6000_BTI_bool_V16QI and
13945 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
13946 entry for overloaded function argument types RS6000_BTI_bool_V4SI
13947 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
13948 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
13949 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
13950 Power9 for implementations of vec_cmpne. Change the signature for
13951 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
13952 (representing vec_all_ne) to remove the previously described first
13953 argument of type RS6000_BTI_INTSI, as this was an artifact of
13954 reliance on the AltiVec predicate framework, which is no longer
13955 used in the implementation of these functions. Add
13956 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
13957 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
13958 since, unlike the AltiVec predicate framework implementation, we
13959 do not share function descriptors between vec_alle and vec_anyeq.
13960 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
13961 set of modes that receive special treatment even when
13962 TARGET_P9_VECTOR is true. The special treatment emits code that
13963 does not depend on Power9 instructions.
13964 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
13965 define_expand to not rely on AltiVec predicate framework.
13966 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13968 (vector_ne_v2di_p): Change this define_expand to not rely on
13969 AltiVec predicate framework.
13970 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
13972 (vector_ne_<mode>_p): Change this define_expand to not rely on
13973 AltiVec predicate framework.
13974 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
13976 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
13977 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
13978 define_insn pattern.
13979 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
13980 define_insn pattern because the xvcmpne<VSs>. instruction is not
13982 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
13983 instruction is not supported.
13985 2017-03-01 Jakub Jelinek <jakub@redhat.com>
13987 * config/nvptx/nvptx.c: Include intl.h.
13989 2017-03-01 Martin Jambor <mjambor@suse.cz>
13992 * ipa-prop.h (ipa_bits): Removed field known.
13993 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
13994 to pointers. Adjusted their comments to warn about their sharing.
13995 (ipcp_transformation_summary): Change bits to a vector of pointers.
13996 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
13997 (ipa_get_ipa_bits_for_value): Declare.
13998 * tree-vrp.h (value_range): Mark as GTY((for_user)).
13999 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14000 (ipa_bits_hash_table): Likewise.
14001 (ipa_vr_ggc_hash_traits): Likewise.
14002 (ipa_vr_hash_table): Likewise.
14003 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14004 being pointers and vr_known being removed.
14005 (ipa_set_jf_unknown): Likewise.
14006 (ipa_get_ipa_bits_for_value): New function.
14007 (ipa_set_jfunc_bits): Likewise.
14008 (ipa_get_value_range): New overloaded functions.
14009 (ipa_set_jfunc_vr): Likewise.
14010 (ipa_compute_jump_functions_for_edge): Use the above functions to
14011 construct bits and vr parts of jump functions.
14012 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14013 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14015 (ipcp_grow_transformations_if_necessary): Also allocate
14016 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14018 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14019 them. Fix too long lines.
14020 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14021 vr_known being removed.
14022 (ipa_read_jump_function): Use new setter functions to construct bits
14023 and vr parts of jump functions or set them to NULL.
14024 (write_ipcp_transformation_info): Adjust for bits being pointers.
14025 (read_ipcp_transformation_info): Likewise.
14026 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14028 Include gt-ipa-prop.h.
14029 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14031 (ipcp_store_bits_results): Likewise.
14032 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14033 Do not write to existing jump functions but use a temporary instead.
14035 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14038 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14039 attempt to use its first operand as BIT_FIELD_REF base.
14041 2017-03-01 Richard Biener <rguenther@suse.de>
14043 PR middle-end/79721
14044 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14045 interpolating formula in wrapping arithmetic.
14046 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14048 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14050 PR tree-optimization/79734
14051 * tree-vect-generic.c (expand_vector_condition): Optimize
14052 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14053 Handle VEC_COND_EXPR where comparison has different inner width from
14054 type's inner width.
14056 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14058 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14059 markup, and similar issues. Remove @opindex entries for things
14060 that aren't options. Add missing -mmpy-option entries.
14062 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14064 PR tree-optimization/79737
14065 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14066 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14067 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14068 instead of byte_size. Formatting fix.
14069 (shift_bytes_in_array_right): Formatting fix.
14071 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14074 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14075 condition on optimize for the leaf function test.
14077 2017-02-28 Martin Liska <mliska@suse.cz>
14080 * read-rtl-function.c (function_reader::handle_unknown_directive):
14081 Bail out when one uses -flto.
14083 2017-02-28 Martin Liska <mliska@suse.cz>
14085 * common.opt: Replace space with tabular for options of <number>
14087 * config/i386/i386.opt: Show <number> value for
14088 -mlarge-data-threshold.
14089 * opts.c (print_filtered_help): Do not display number in hexadecimal
14092 2017-02-28 Martin Liska <mliska@suse.cz>
14094 * common.opt: Fix --help=option -Q for options which are of
14097 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14099 * config/i386/i386.c (print_reg): Error out for values
14100 of 8-bit size in invalid integer register.
14102 2017-02-28 Martin Sebor <msebor@redhat.com>
14104 PR tree-optimization/79691
14105 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14107 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14110 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14111 gcc_unreachable with output_operand_lossage.
14113 2017-02-28 Richard Biener <rguenther@suse.de>
14115 PR tree-optimization/79740
14116 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14118 (visit_nary_op): Insert the nary into the hashtable if we
14119 pattern-matched sth.
14120 * tree-ssa-pre.c (eliminate_insert): Robustify.
14122 2017-02-28 Richard Biener <rguenther@suse.de>
14124 PR middle-end/79731
14125 * fold-const.c (decode_field_reference): Reject out-of-bound
14128 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14130 * config/i386/i386.c: Include intl.h.
14131 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14132 instead of just cond ? "..." : "...".
14133 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14134 * coverage.c (read_counts_file): Likewise.
14135 * omp-offload.c: Include intl.h.
14136 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14137 of just cond ? "..." : "...".
14138 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14139 of just cond ? "..." : "...".
14141 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14144 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14146 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14147 'tune for' CPU name.
14148 * config/arm/arm-cpu-data.h: Regenerated.
14150 2017-02-28 Richard Biener <rguenther@suse.de>
14152 PR tree-optimization/79732
14153 * tree-inline.c (expand_call_inline): Do not shadow var.
14155 2017-02-28 Richard Biener <rguenther@suse.de>
14157 PR tree-optimization/79723
14158 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14159 address-space properly.
14161 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14163 * doc/optinfo.texi (Optimization groups): Fix option used for
14165 * doc/invoke.texi (-fopt-info): Document "omp".
14166 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14167 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14168 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14169 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14170 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14172 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14174 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14175 OPTGROUP_OMP with "omp".
14177 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14180 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14181 for arithmetic shift of unsigned V2DI.
14183 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14185 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14186 arc/linux.h headers.
14187 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14188 (LINK_SPEC): Likewise.
14189 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14190 (EXTRA_SPECS): Likewise.
14191 (STARTFILE_SPEC): Likewise.
14192 (ENDFILE_SPEC): Likewise.
14193 (LIB_SPEC): Likewise.
14194 (TARGET_SDATA_DEFAULT): Likewise.
14195 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14196 (MULTILIB_DEFAULTS): Likewise.
14197 (DWARF2_UNWIND_INFO): Likewise.
14198 * config/arc/big.h: New file.
14199 * config/arc/elf.h: Likewise.
14200 * config/arc/linux.h: Likewise.
14201 * config/arc/t-uClibc: Remove.
14203 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14205 PR tree-optimization/77536
14206 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14207 (tree_transform_and_unroll_loop): Use above function to compute the
14208 estimated niter of unrolled loop and use it when scaling profile.
14209 Also use count info rather than frequency if it's non-zero.
14210 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14211 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14212 (vect_transform_loop): Call above function.
14214 2017-02-27 Richard Biener <rguenther@suse.de>
14216 PR tree-optimization/45397
14217 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14218 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14219 (visit_nary_op): Add pattern matching for CSEing sign-changed
14220 or truncated operations with wider ones.
14222 2017-02-27 Richard Biener <rguenther@suse.de>
14224 PR tree-optimization/79690
14225 * tree-vect-stmts.c (vectorizable_store): Use vector type
14226 built from the DR with address-space.
14228 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14230 * doc/invoke.texi (Optimize Options): Refine the description
14231 of asan-use-after-return.
14233 2017-02-25 Alan Modra <amodra@gmail.com>
14235 PR rtl-optimization/79584
14236 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14237 base, not ad->base_term, the reg within base. Remove assertion
14238 that ad->base == ad->base_term. Replace gen_int_mode using
14239 bogus mode with const0_rtx.
14241 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14243 PR middle-end/79396
14244 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14245 FMA_EXPR like tcc_binary or tcc_unary.
14247 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14250 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14252 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14253 (output_loc_operands): Handle DW_OP_call_ref and
14254 DW_OP_GNU_variable_value.
14255 (struct variable_value_struct): New type.
14256 (struct variable_value_hasher): Likewise.
14257 (variable_value_hash): New variable.
14258 (string_types): Remove.
14259 (copy_loc_descr): New function.
14260 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14261 (prepend_loc_descr_to_each): New function.
14262 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14263 instead of add_loc_descr_to_each if the first argument is single
14264 location list and the second has multiple.
14265 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14266 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14267 when looking for variable value which doesn't have other location info.
14268 (loc_list_from_tree): Formatting fix.
14269 (gen_array_type_die): Simplify DW_AT_string_length handling.
14270 (adjust_string_types): Remove.
14271 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14272 string_types. Call resolve_variable_values.
14273 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14274 (resolve_addr_in_expr): Likewise. Add A argument.
14275 (copy_deref_exprloc): Remove deref argument. Adjust for the
14276 original expression being DW_OP_GNU_variable_value with optionally
14277 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14278 optionally after it.
14279 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14280 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14281 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14282 (variable_value_hasher::hash, variable_value_hasher::equal): New
14284 (resolve_variable_value_in_expr, resolve_variable_value,
14285 resolve_variable_values, note_variable_value_in_expr,
14286 note_variable_value): New functions.
14287 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14290 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14293 * opts.h (handle_generated_option): Add GENERATED_P argument.
14294 * opts-common.c (handle_option): Adjust function comment.
14295 (handle_generated_option): Add GENERATED_P argument, pass it to
14297 (control_warning_option): Pass false to handle_generated_option
14299 * opts.c (maybe_default_option): Pass true to handle_generated_option
14301 * optc-gen.awk: Likewise.
14303 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14305 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14306 a REG, look at the REG it is a SUBREG of.
14307 (splitter for cmpeqsi_t): Ditto.
14309 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14311 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14312 the special USEs with the pattern of the insn, not the insn itself.
14314 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14317 * doc/invoke.texi: Document -mload-store-pairs.
14319 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14320 Sandra Loosemore <sandra@codesourcery.com>
14322 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14323 argument isn't a CONST_INT.
14324 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14325 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14326 (nios2_validate_compare): Bypass alternate compare logic if *op2
14327 is not a CONST_INT.
14328 (ldstwm_operation_p): Return false if first_base is not a REG or
14329 if first_offset is not a CONST_INT.
14331 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14333 * config/cris/cris.md: Use correct operand in a define_peephole2.
14335 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14337 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14339 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14341 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14342 this_insn if it is an INSN or JUMP_INSN.
14343 (force_offsettable): Look at base, not at addr.
14344 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14345 on things that aren't necessarily CONST_INTs.
14347 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14349 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14350 -mfpmath=sse is the default also for x86-32 targets with SSE2
14351 instruction set when @option{-ffast-math} is enabled
14353 2017-02-24 Jeff Law <law@redhat.com>
14355 PR rtl-optimizatoin/79286
14356 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14359 2017-02-24 Richard Biener <rguenther@suse.de>
14361 PR tree-optimization/79389
14362 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14365 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14367 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14368 function comment to reflect reality.
14369 (loop_exits_before_overflow): Fix typo in function description.
14371 2017-02-24 Richard Biener <rguenther@suse.de>
14373 PR tree-optimization/79389
14374 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14375 properly that a threading opportunity exists. Detect conditional
14376 copy/constant propagation opportunities.
14378 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14380 * config/visium/visium.md (type): Add trap.
14381 (b): New mode attribute.
14382 (*btst): Rename into...
14383 (*btst<mode>): ...this and adjust.
14384 (*cbranchsi4_btst_insn): Rename into...
14385 (*cbranch<mode>4_btst_insn): ...this and adjust.
14386 (trap): New define_insn.
14388 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14390 PR tree-optimization/79389
14391 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14392 (noce_reversed_cond_code): New function.
14393 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14394 reversed_comparison_code. Formatting fix.
14395 (noce_try_store_flag): Test rev_cond != NULL in addition to
14396 reversed_comparison_code.
14397 (noce_try_store_flag_constants): Likewise.
14398 (noce_try_store_flag_mask): Likewise.
14399 (noce_try_addcc): Use rev_cond if non-NULL instead of
14400 reversed_comparison_code.
14401 (noce_try_cmove_arith): Likewise. Formatting fixes.
14402 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14403 (noce_find_if_block): Initialize rev_cond.
14404 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14405 instead of false as last argument never attempt to reverse it
14408 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14410 PR tree-optimization/79663
14411 * tree-predcom.c (combine_chains): Process refs in reverse order
14412 only for ZERO length chains, and add explaining comment.
14414 2017-02-23 Jeff Law <law@redhat.com>
14416 PR tree-optimization/79578
14417 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14418 in call to operand_equal_p.
14420 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14423 * config/i386/cpuid.h: Fix another undefined behavior.
14425 2017-02-23 Richard Biener <rguenther@suse.de>
14427 PR tree-optimization/79683
14428 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14429 vector types for data-refs.
14431 2017-02-23 Martin Liska <mliska@suse.cz>
14433 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14435 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14437 PR middle-end/79665
14438 * internal-fn.c (get_range_pos_neg): Moved to ...
14439 * tree.c (get_range_pos_neg): ... here. No longer static.
14440 * tree.h (get_range_pos_neg): New prototype.
14441 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14442 are known to be in between 0 and signed maximum inclusive, try to
14443 expand both unsigned and signed divmod and use the cheaper one from
14446 2017-02-22 Jeff Law <law@redhat.com>
14448 PR tree-optimization/79578
14449 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14450 to compare base operands.
14452 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14455 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14456 gpc_reg_operand instead of fpr_reg_operand.
14458 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14460 * config/mips/mips.c (mips_return_in_memory): Force FP
14461 vector types to be returned in memory for o32 ABI.
14463 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14465 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14466 instead of DW_TAG_member for static data member declarations and don't
14467 set no_linkage_name for static inline data members.
14468 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14471 2017-02-22 Martin Liska <mliska@suse.cz>
14473 * doc/invoke.texi: Replace inequality signs with square brackets
14476 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14478 PR tree-optimization/68644
14479 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14481 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14484 * lra-constraints.c (simplify_operand_subreg): Handle
14485 WORD_REGISTER_OPERATIONS targets.
14487 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14490 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14491 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14492 elimination by swapping fld*.
14494 2017-02-22 Richard Biener <rguenther@suse.de>
14496 PR tree-optimization/79673
14497 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14498 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14499 irrelevant address-space qualifiers and avoiding a
14500 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14502 2017-02-22 Richard Biener <rguenther@suse.de>
14504 PR tree-optimization/79666
14505 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14506 to not symbolically negate if that may introduce undefined
14509 2017-02-22 Martin Liska <mliska@suse.cz>
14512 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14513 * data-streamer-out.c (streamer_write_gcov_count_stream):
14515 * value-prof.c (stream_out_histogram_value): Make assert more
14516 precise based on type of counter.
14518 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14521 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14522 Use nonimmediate_operand instead of memory_operand for operand 1.
14523 (float-extend standard_x87sse_constant_load splitter): Ditto.
14525 2017-02-21 Jeff Law <law@redhat.com>
14527 PR tree-optimization/79621
14528 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
14529 blocks with edges to themselves.
14531 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14534 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
14535 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
14536 Use gimple_call_builtin_p.
14539 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
14540 on temporarily removed DEBUG_INSNs.
14542 PR tree-optimization/79649
14543 * tree-loop-distribution.c (classify_partition): Give up on
14544 non-generic address space loads/stores.
14546 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
14548 * doc/loop.texi (Loop manipulation): Remove nonexistent
14549 tree_ssa_loop_version from the documentation.
14550 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
14552 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14555 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
14556 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
14557 * config/rs6000/rs6000.c: Include except.h.
14558 (rs6000_expand_split_stack_prologue): Call
14559 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
14561 2017-02-21 Martin Jambor <mjambor@suse.cz>
14564 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
14565 have been analyzed.
14567 2017-02-21 Martin Jambor <mjambor@suse.cz>
14569 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
14570 for backward compatibility only.
14571 * doc/invoke.texi (Option Summary): Remove all references to
14572 -fipa-cp-alignment.
14574 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
14578 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14580 * lra-constraints.c (curr_insn_transform): Handle
14581 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14583 2017-02-21 Martin Liska <mliska@suse.cz>
14585 * config/i386/i386.opt: Replace -masm-dialect with -masm.
14587 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
14589 PR translation/79638
14590 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
14592 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
14595 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
14596 (arm_function_ok_for_sibcall): Return false for an indirect call by
14597 descriptor if all the argument registers are used.
14598 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
14599 alignment of the function.
14601 2017-02-21 Jakub Jelinek <jakub@redhat.com>
14603 PR tree-optimization/61441
14604 * simplify-rtx.c (simplify_const_unary_operation): For
14605 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
14606 the sNaN unmodified.
14608 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14610 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14611 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14612 instead of SYSTEM_HEADER_DIR.
14614 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
14615 Martin Liška <mliska@suse.cz>
14617 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
14618 Fix typos and grammar, use active voice, and clarify.
14620 2017-02-20 Marek Polacek <polacek@redhat.com>
14622 PR middle-end/79537
14623 * gimplify.c (gimplify_expr): Handle unused *&&L;.
14626 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
14628 2017-02-20 Jakub Jelinek <jakub@redhat.com>
14631 * config/i386/i386.c (ix86_expand_builtin): Handle
14632 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
14633 ix86_builtins_isa[fcode].isa as a requirement of those
14634 flags and any other flag in the bitmask.
14635 (ix86_init_mmx_sse_builtins): Use 0 instead of
14636 ~OPTION_MASK_ISA_64BIT as mask.
14637 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
14638 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
14639 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
14640 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
14642 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14645 * lra-constraints.c (split_reg): Check requested split mode
14646 is supported by the register.
14648 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14650 * lra-constraints.c (simplify_operand_subreg): Remove early
14653 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14656 * lra-constraints.c (curr_insn_transform): Tighten condition
14657 for converting SUBREG reloads from OP_OUT to OP_INOUT.
14659 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
14662 * lra-constraints.c (curr_insn_transform): Handle
14663 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
14665 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
14668 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
14670 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
14672 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
14675 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14678 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14680 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
14683 2017-02-18 Jakub Jelinek <jakub@redhat.com>
14685 * final.c (last_columnnum, override_columnnum): New variables.
14686 (final_start_function): Set last_columnnum, pass it to begin_prologue
14687 hook and pass 0 to dwarf2out_begin_prologue.
14688 (final_scan_insn): Update override_columnnum. Pass last_columnnum
14689 to source_line debug hook.
14690 (notice_source_line): Compute last_columnnum and for debug_column_info
14691 return true on column changes.
14692 * debug.h (struct gcc_debug_hooks): Add column argument to
14693 source_line and begin_prologue hooks.
14694 (debug_nothing_int_charstar_int_bool): Remove prototype.
14695 (debug_nothing_int_int_charstar,
14696 debug_nothing_int_int_charstar_int_bool): New prototypes.
14697 (dwarf2out_begin_prologue): Add column argument.
14698 * debug.c (do_nothing_debug_hooks): Adjust source_line and
14699 begin_prologue hooks.
14700 (debug_nothing_int_charstar_int_bool): Remove.
14701 (debug_nothing_int_int_charstar,
14702 debug_nothing_int_int_charstar_int_bool): New functions.
14703 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
14704 through to dwarf2out_source_line.
14705 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
14706 (dwarf2out_source_line): Add column argument, emit it if requested.
14707 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
14709 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14710 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
14711 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
14712 through to dwarf2out_begin_prologue.
14713 (vmsdbgout_source_line): Add column argument, pass it through to
14714 dwarf2out_source_line.
14715 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
14716 dbxout_source_line caller.
14717 (dbxout_source_line): Add column argument.
14719 * common.opt (gno-column-info, gcolumn-info): New options.
14720 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
14721 (check_die): Also test for multiple DW_AT_decl_column attributes.
14722 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
14723 DW_AT_decl_column if requested.
14724 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
14726 (gen_variable_die): Likewise.
14727 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
14728 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
14731 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
14732 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
14733 (ix86_handle_option): Handle OPT_m3dnowa.
14734 * doc/invoke.texi (-m3dnowa): Document.
14735 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
14736 -m3dnowa instead of -m3dnow -march=athlon.
14739 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
14740 instead of gcc_assert for K, r and R code checks. Formatting fixes.
14742 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14745 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
14746 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
14747 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
14748 generator for vsx_xxpermdi_<mode>_be.
14749 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
14750 force big-endian semantics.
14751 (vsx_xxpermdi_<mode>_be): New define_expand with same
14752 implementation as previous version of vsx_xxpermdi_<mode>.
14754 2017-02-17 Jakub Jelinek <jakub@redhat.com>
14756 PR tree-optimization/79327
14757 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
14758 variable, its initialization and use.
14760 2017-02-17 Julia Koval <julia.koval@intel.com>
14762 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
14763 (OPTION_MASK_ISA_PKU_UNSET): New.
14764 (ix86_handle_option): Handle -mrdpid.
14765 * config/i386/cpuid.h (bit_RDPID): New.
14766 * config/i386/driver-i386.c (host_detect_local_cpu):
14767 Detect RDPID feature.
14768 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
14769 * config/i386/i386-c.c (ix86_target_macros_internal):
14771 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
14772 (ix86_valid_target_attribute_inner_p): Add "rdpid".
14773 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14774 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
14775 * config/i386/i386.md (define_insn "rdpid"): New.
14776 * config/i386/i386.opt Add -mrdpid.
14777 * config/i386/immintrin.h (_rdpid_u32): New.
14779 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
14781 PR rtl-optimization/79541
14782 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
14783 instead of transforming it into USE.
14785 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
14787 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
14788 If HONOR_SNANS (SFmode) force the input to a register.
14789 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
14790 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
14791 an frsp or similar insn.
14793 2017-02-17 Martin Liska <mliska@suse.cz>
14795 PR rtl-optimization/79577
14796 * params.def (selsched-max-sched-times): Increase minimum to 1.
14798 2017-02-17 Martin Liska <mliska@suse.cz>
14800 PR rtl-optimization/79574
14801 * gcse.c (want_to_gcse_p): Prevent integer overflow.
14803 2017-02-17 Martin Liska <mliska@suse.cz>
14805 PR tree-optimization/79529
14806 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
14807 ssa_defined_default_def_p to handle cases which are implicitly
14809 * tree-ssa.c (ssa_defined_default_def_p): New function.
14810 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
14811 which are implicitly defined.
14812 * tree-ssa.h (ssa_defined_default_def_p): Declare.
14814 2017-02-17 Richard Biener <rguenther@suse.de>
14816 PR middle-end/79576
14817 * params.def (max-ssa-name-query-depth): Limit to 10.
14819 2017-02-17 Richard Biener <rguenther@suse.de>
14821 PR tree-optimization/79552
14822 * tree-ssa-structalias.c (visit_loadstore): Properly verify
14825 2017-02-17 Richard Biener <rguenther@suse.de>
14828 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
14830 2017-02-17 Marek Polacek <polacek@redhat.com>
14832 PR middle-end/79536
14833 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
14834 (fold_negate_expr): New wrapper.
14836 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
14838 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
14839 Correct terminology and de-emphasize pre-standard behavior.
14841 2017-02-16 Alan Modra <amodra@gmail.com>
14843 PR rtl-optimization/79286
14844 * ira.c (def_dominates_uses): New function.
14845 (update_equiv_regs): Don't create an equivalence for insns that
14846 may trap where the register def does not dominate the use.
14848 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
14850 PR rtl-optimization/78127
14851 * lra.c (lra): Call lra_eliminate before finish the loop after
14854 2017-02-16 Richard Biener <rguenther@suse.de>
14856 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
14858 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
14859 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
14860 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
14861 (isl_val_int_from_wi): New function.
14862 (extract_affine_gmp): Rename to ...
14863 (extract_affine_wi): ... this, take a widest_int.
14864 (extract_affine_int): Just wrap extract_affine_wi.
14865 (add_param_constraints): Use isl_val_int_from_wi.
14866 (add_loop_constraints): Likewise, and extract_affine_wi.
14868 2017-02-15 Jeff Law <law@redhat.com>
14870 PR middle-end/79521
14871 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
14872 ira_init_register_move_cost_if_necessary.
14874 2017-02-15 Martin Sebor <msebor@redhat.com>
14876 PR middle-end/32003
14877 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
14878 removed in a prior commit.
14880 2017-02-15 Bin Cheng <bin.cheng@arm.com>
14882 PR tree-optimization/79347
14883 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
14884 counters during peeling.
14886 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
14888 * Makefile.in (site.exp): Remove "set ISLVER".
14890 2017-02-15 Jakub Jelinek <jakub@redhat.com>
14893 * real.c (real_from_integer): Call real_convert even for decimal.
14895 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
14898 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
14900 2017-02-14 Andrew Pinski <apinski@cavium.com>
14902 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
14903 cores and change the partno/implementer to be correct.
14904 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
14905 the 'B" as the implementer.
14906 * config/aarch64/aarch64-tune.md: Regenerate.
14908 2017-02-14 Carl Love <cel@us.ibm.com>
14910 * config/rs6000/rs6000.c: Add case statement entry to make the
14911 xvcvuxdsp built-in argument unsigned.
14912 * config/rs6000/vsx.md: Fix the source and return operand types so they
14913 match the instruction definitions from the ISA document. Fix typo
14914 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
14917 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
14920 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
14921 member early_clobber_alts.
14922 * lra-lives.c (reg_early_clobber_p): New.
14923 (process_bb_lives): Use it.
14924 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
14925 (debug_operand_data): Initialize early_clobber_alts.
14926 (setup_operand_alternative): Set up early_clobber_alts.
14927 (collect_non_operand_hard_regs): Ditto. Pass early clobber
14928 alternatives to new_insn_reg.
14929 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
14931 (lra_update_insn_regno_info): Pass the new arg.
14933 2017-02-14 Jakub Jelinek <jakub@redhat.com>
14935 PR middle-end/79505
14936 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
14937 (new_oacc_loop_raw): Don't clear already cleared fields.
14940 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
14941 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
14942 _mm512_prefetch_i64gather_ps): New inline functions and macros.
14944 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
14947 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
14949 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
14952 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
14953 the extra instruction to the right place to store 128-bit constant
14956 2017-02-14 Martin Sebor <msebor@redhat.com>
14958 PR middle-end/79448
14959 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
14960 warning for strings of unknown length.
14962 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
14964 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
14966 2017-02-14 Jeff Law <law@redhat.com>
14969 * ira-costs.c (scan_one_insn): Initialize register move costs
14970 for pseudos seen in USE/CLOBBER insns.
14972 PR tree-optimization/79095
14973 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
14974 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
14975 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
14976 if the operands are known to be not equal, then the resulting range
14978 (intersect_ranges): If the new range is ~[0,0] and the old range is
14979 wide, then prefer ~[0,0].
14980 * tree-vrp.c (overflow_comparison_p_1): New function.
14981 (overflow_comparison_p): New function.
14982 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
14983 if NAME is used in an overflow test.
14984 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
14985 overflow check that can be expressed as an equality test, then adjust
14986 ops to be that equality test.
14988 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14990 * config/s390/s390-builtin-types.def: Remove flags argument.
14991 * config/s390/s390.c (s390_init_builtins): Likewise.
14993 2017-02-14 Martin Liska <mliska@suse.cz>
14995 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
14996 vector. Fix trailing white spaces.
14998 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15000 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15003 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15005 PR rtl-optimization/68664
15006 * config/arm/arm.c (arm_sched_can_speculate_insn):
15007 New function. Declare prototype.
15008 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15010 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15012 PR rtl-optimization/68664
15013 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15015 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15017 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15019 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15020 max skip bytes for function, loop and jump.
15022 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15024 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15025 ABS_EXPR for gimple dump.
15027 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15030 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15032 PR tree-optimization/79408
15033 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15034 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15035 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15036 also if rhs1 is INTEGER_CST.
15038 2017-02-14 Richard Biener <rguenther@suse.de>
15040 PR middle-end/79432
15041 * tree-into-ssa.c (insert_phi_nodes): When the function can
15042 have abnormal edges rewrite SSA names with broken use-def
15043 dominance out of SSA and register them for PHI insertion.
15045 2017-02-13 Martin Sebor <msebor@redhat.com>
15047 PR middle-end/79496
15048 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15049 clearing info.nowrite flag when snprintf size argument is a range.
15051 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15053 * cprop.c (cprop_jump): Add missing space in string literal.
15054 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15055 (get_constraint_for_component_ref): Likewise.
15056 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15057 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15058 * lra-constraints.c (process_alt_operands): Likewise.
15059 * ipa-inline.c (inline_small_functions): Likewise.
15060 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15061 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15062 * trans-mem.c (diagnose_tm_1_op): Likewise.
15063 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15064 (grid_parallel_clauses_gridifiable): Likewise.
15066 * config/nvptx/mkoffload.c (process): Add space in between
15069 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15070 "MOD4_SSE_REGS" and "ALL_REGS".
15072 * spellcheck.c (test_data): Add , in between "foo" and "food".
15074 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15077 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15078 boundary crossing check and subsequent code generation agree.
15080 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15082 * config/aarch64/aarch64.c (has_memory_op): Delete.
15083 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15086 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15088 PR rtl-optimization/79388
15089 PR rtl-optimization/79450
15090 * combine.c (distribute_notes): When removing TEM_INSN for which
15091 corresponding dest has last value recorded, invalidate that last
15094 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15096 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15097 of explicit '@'. Add missing assembly comment marker on branch costs
15100 2017-02-13 Nathan Sidwell <nathan@acm.org>
15102 * gengtype-lex.l (<in_struct>): Add '/'.
15104 2017-02-13 Martin Liska <mliska@suse.cz>
15107 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15109 2017-02-13 Richard Biener <rguenther@suse.de>
15111 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15113 * configure: Re-generate.
15114 * config.in: Likewise.
15115 * graphite-dependences.c: Simplify as if
15116 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15117 * graphite-isl-ast-to-gimple.c: Likewise.
15118 * graphite-optimize-isl.c: Likewise.
15119 * graphite-poly.c: Likewise.
15120 * graphite-sese-to-poly.c: Likewise.
15121 * graphite.h: Likewise.
15122 * toplev.c: Include isl/version.h and use isl_version () for
15123 printing the ISL version.
15124 * doc/install.texi: Update ISL requirement.
15126 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15128 * doc/standards.texi (Standards): Update reference to
15131 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15133 * doc/extend.texi (Named Address Spaces): sourceware.org now
15135 * doc/install.texi (Binaries): Ditto.
15138 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15140 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15141 standard terminology "stringize"/"stringizing" throughout.
15142 * doc/cppinternals.texi: Likewise.
15144 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15146 * doc/extend.texi: Fix some spelling mistakes and typos.
15147 * doc/invoke.texi: Likewise.
15149 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15152 * params.def (inline-min-speedup) Change from 10 to 8.
15154 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15156 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15159 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15162 * ipa-inline-analysis.c (get_minimal_bb): New function.
15163 (record_modified): Use it.
15164 (remap_edge_change_prob): Handle also ancestor functions.
15166 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15168 * doc/contrib.texi (Contributors): Remove broken link into
15169 the Mauve CVS repository.
15171 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15173 PR middle-end/79454
15174 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15175 result computation whenever lhs doesn't have vector mode, not
15176 just when it has BLKmode.
15178 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15180 * doc/makefile.texi (profiledbootstrap): Refer to the
15181 installation instructions only in textual form.
15183 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15186 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15188 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15190 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15191 (Specific): Update mingw-w64 reference.
15193 (Specific): Remove broken link to Renesas RX processor.
15195 2017-02-10 Richard Biener <rguenther@suse.de>
15197 * toplev.c (process_options): Do not mention obsolete graphite
15198 options when printing sorry message about missing graphite support.
15199 Mention -floop-nest-optimize.
15201 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15203 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15204 (vtst_p16): Likewise.
15205 (vtstq_p8): Likewise.
15206 (vtstq_p16): Likewise.
15208 (vtstq_p64): Likewise.
15209 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15210 (vset_lane_p64): New.
15211 (vsetq_lane_p64): New.
15213 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15215 PR tree-optimization/79411
15216 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15217 stmt operands are SSA_NAMEs used in abnormal phis.
15218 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15221 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15224 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15227 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15229 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15231 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15233 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15237 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15238 not arbitrary TREE_CONSTANT.
15241 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15242 "omp declare target link" attribute unless is_global_var.
15243 * omp-offload.c (find_link_var_op): Likewise.
15245 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15246 Chung-Lin Tang <cltang@codesourcery.com>
15248 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15250 (gimplify_adjust_omp_clauses): Don't delete TILE.
15251 (gimplify_omp_for): Deal with TILE.
15252 * internal-fn.c (expand_GOACC_TILE): New function.
15253 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15255 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15256 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15258 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15259 avoid DIV for outermost collapse var.
15260 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15261 Remove out of date comments, fix whitespace.
15262 * omp-general.c (omp_extract_for_data): Deal with tiling.
15263 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15264 adjust OLF_DIM_BASE value.
15265 (struct omp_for_data): Add tiling field.
15266 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15267 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15268 for auto loops. Remove default auto determining, moved to
15269 oacc_loop_fixed_partitions.
15270 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15271 stmts, add e_mask field.
15272 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15273 (oacc_thread_numbers): Use oacc_dim_call.
15274 (oacc_xform_tile): New.
15275 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15276 (finish_oacc_loop): Adjust for ifns vector.
15277 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15278 add case for GOACC_TILE fns.
15279 (oacc_loop_xform_loop): Delete.
15280 (oacc_loop_process): Iterate over call list directly, and add
15281 handling for GOACC_TILE fns.
15282 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15284 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15285 vector partitioning to outer loops. Assign 2 partitions to loops
15286 when available. Add TILE handling.
15287 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15288 (execite_oacc_device_lower): Process GOACC_TILE fns,
15289 ignore unknown specs.
15290 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15291 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15292 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15294 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15296 * configure.ac (ACX_BUGURL): Update.
15297 * configure: Regenerate.
15299 2017-02-09 Richard Biener <rguenther@suse.de>
15301 PR tree-optimization/69823
15302 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15303 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15305 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15307 * config/arc/arc-c.def: Add __NPS400__ definition.
15308 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15309 (TARGET_NPS400): Define.
15311 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15313 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15315 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15316 pointer, arch_info.
15317 (arc_cpu_types): Fill the arch_info field with a pointer into the
15318 arc_arch_types table.
15319 (arc_selected_cpu): Declare.
15320 * config/arc/arc.c (arc_selected_cpu): Make global.
15321 (arc_selected_arch): Delete.
15322 (arc_base_cpu): Delete.
15323 (arc_override_options): Remove references to deleted variables,
15324 update access to arch information.
15325 (ARC_OPT): Update access to arch information.
15326 (ARC_OPTX): Likewise.
15327 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15328 (TARGET_ARC600): Update access to arch information.
15329 (TARGET_ARC601): Likewise.
15330 (TARGET_ARC700): Likewise.
15331 (TARGET_EM): Likewise.
15332 (TARGET_HS): Likewise.
15333 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15336 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15339 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15340 condition/operands for integer GE/LE/GEU/LEU operations.
15342 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15344 PR translation/79397
15345 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15348 2017-02-08 Martin Jambor <mjambor@suse.cz>
15351 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15352 whether allocation happened.
15353 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15354 nothing was allocated.
15356 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15358 PR tree-optimization/79408
15359 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15360 constant, but SSA_NAME with a known integer range, use the minimum
15361 of that range instead of op1 to determine if modulo can be replaced
15362 with its first operand.
15364 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15366 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15368 2017-02-08 Richard Biener <rguenther@suse.de>
15370 PR tree-optimization/71824
15371 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15372 Check all loops contained in the merged region.
15374 2017-02-07 Andrew Pinski <apinski@cavium.com>
15376 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15378 2017-02-07 Andrew Pinski <apinski@cavium.com>
15380 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15381 (thunderxt88): Likewise.
15382 (thunderxt81): Disable LSE and change v8.1 to v8.
15383 (thunderxt83): Likewise.
15385 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15386 Richard Biener <rguenther@suse.de>
15388 PR middle-end/79399
15389 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15390 type from int to size_t.
15391 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15393 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15395 PR rtl-optimization/79386
15396 * cprop.c (bypass_conditional_jumps): Initialize
15397 bypass_last_basic_block already before splitting bbs after
15398 unconditional traps...
15399 (bypass_conditional_jumps): ... rather than here.
15402 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15403 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15404 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15405 fix -masm=intel patterns.
15407 2017-02-07 Richard Biener <rguenther@suse.de>
15409 PR tree-optimization/79256
15410 PR middle-end/79278
15411 * builtins.c (get_object_alignment_2): Use min_align_of_type
15412 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15413 and ADJUST_FIELD_ALIGN.
15415 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15417 * doc/tm.texi: Regenerate.
15418 * stor-layout.c (layout_decl): Adjust.
15419 (update_alignment_for_field): Likewise.
15420 (place_field): Likewise.
15421 (min_align_of_type): Likewise.
15422 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15423 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15424 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15425 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15426 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15427 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15428 * config/i386/i386.c (x86_field_alignment): Likewise.
15429 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15430 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15431 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15432 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15433 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15434 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15438 2017-01-30 Richard Biener <rguenther@suse.de>
15440 PR tree-optimization/79256
15441 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15442 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15445 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15447 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15448 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15449 builtins to SImode and emit a zero-extend, if necessary.
15451 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15453 * docs/invoke.texi (RISC-V Options): Alphabetize.
15455 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15457 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15460 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15462 * config/riscv/riscv.c: New file.
15463 * common/config/riscv/riscv-common.c: Likewise.
15464 * config.gcc: Likewise.
15465 * config/riscv/constraints.md: Likewise.
15466 * config/riscv/elf.h: Likewise.
15467 * config/riscv/generic.md: Likewise.
15468 * config/riscv/linux.h: Likewise.
15469 * config/riscv/multilib-generator: Likewise.
15470 * config/riscv/peephole.md: Likewise.
15471 * config/riscv/pic.md: Likewise.
15472 * config/riscv/predicates.md: Likewise.
15473 * config/riscv/riscv-builtins.c: Likewise.
15474 * config/riscv/riscv-c.c: Likewise.
15475 * config/riscv/riscv-ftypes.def: Likewise.
15476 * config/riscv/riscv-modes.def: Likewise.
15477 * config/riscv/riscv-opts.h: Likewise.
15478 * config/riscv/riscv-protos.h: Likewise.
15479 * config/riscv/riscv.h: Likewise.
15480 * config/riscv/riscv.md: Likewise.
15481 * config/riscv/riscv.opt: Likewise.
15482 * config/riscv/sync.md: Likewise.
15483 * config/riscv/t-elf-multilib: Likewise.
15484 * config/riscv/t-linux: Likewise.
15485 * config/riscv/t-linux-multilib: Likewise.
15486 * config/riscv/t-riscv: Likewise.
15487 * configure.ac: Likewise.
15488 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15489 Waterman as RISC-V maintainers.
15490 * doc/install.texi: Add RISC-V entries.
15491 * doc/invoke.texi: Add RISC-V options section.
15492 * doc/md.texi: Add RISC-V constraints section.
15493 * configure: Regenerated.
15495 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15498 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15499 false values to be constant vectors with all 0 or all 1 bits set.
15500 (vcondu<mode><mode>): Likewise.
15501 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15503 (fpmask_comparison_operator): Update comment.
15504 (vecint_comparison_operator): New predicate.
15505 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15506 vector conditionals when the true and false values are constant
15507 vectors with all 0 bits or all 1 bits set.
15509 2017-02-06 Martin Sebor <msebor@redhat.com>
15511 PR tree-optimization/79376
15512 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15514 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15516 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15517 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15518 to simplify split condition.
15520 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15522 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15523 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15526 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
15528 PR rtl-optimization/68664
15529 * target.def (can_speculate_insn): New hook.
15530 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
15531 * doc/tm.texi: Regenerate.
15532 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
15533 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
15534 (rs6000_sched_can_speculate_insn): New function.
15536 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15538 PR tree-optimization/79284
15539 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
15540 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
15541 vectorizable_mask_load_store, vectorizable_operation,
15542 vect_is_simple_cond, get_same_sized_vectype): Use it instead
15543 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
15544 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
15545 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
15546 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15547 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15548 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
15549 is_gimple_assign (stmt). Replace another such test with
15550 is_gimple_assign (stmt).
15552 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15555 * config/avr/avr.c (rtl-iter.h): Include it.
15556 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
15557 (avr_legitimate_combined_insn): ...and implementation.
15559 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
15561 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
15562 * config/s390/s390.c (s390_const_operand_ok)
15563 (s390_canonicalize_comparison, s390_extract_part)
15564 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
15565 (s390_contiguous_bitmask_p, s390_rtx_costs)
15566 (legitimize_pic_address): Likewise.
15567 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
15568 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
15569 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
15570 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
15571 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
15573 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
15575 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
15576 REGNO($0) == REGNO($1).
15578 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15580 * config/s390/linux.h(SIZE_TYPE): Add comment.
15582 2017-02-06 Julian Brown <julian@codesourcery.com>
15583 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15584 Virendra Pathak <virendra.pathak@broadcom.com>
15586 * config/aarch64/aarch64-cores.def: Change the scheduler
15588 * config/aarch64/aarch64.md: Include thunderx2t99.md.
15589 * config/aarch64/thunderx2t99.md: New file.
15591 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15593 * doc/standards.texi (Go Language): Update link to language
15596 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15598 * tree-eh.c (lower_resx): Sanitize profile.
15599 (cleanup_empty_eh_move_lp): Likewise.
15601 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
15604 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
15606 * cfgloopmanip.h (loop_version): Update prototype.
15607 * modulo-sched.c (sms_schedule): Update call of loop_version.
15608 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
15609 * tree-parloops.c (gen_parallel_loop): Likewise.
15610 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
15611 * tree-ssa-loop-split.c (split_loop): Likewise.
15612 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
15613 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15615 2017-02-05 Martin Liska <mliska@suse.cz>
15618 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
15620 (print_operand_address): Initialize a struct to zero.
15622 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15624 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
15625 garbage collector only in textual form.
15627 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
15629 * doc/extend.texi (x86 specific memory model extensions for
15630 transactional memory): Simplify a phrase.
15632 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
15635 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
15636 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
15637 (atomic_storedi_1): Likewise.
15639 2017-02-04 Jakub Jelinek <jakub@redhat.com>
15641 PR tree-optimization/79338
15642 * tree-parloops.c (gather_scalar_reductions): Don't call
15643 vect_analyze_loop_form for loop->inner before destroying loop's
15646 2017-02-03 Martin Sebor <msebor@redhat.com>
15648 PR tree-optimization/79327
15649 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
15650 when precision has resulted in leading zeros.
15651 (format_integer): Adjust the likely counter to assume an unknown
15652 argument that may be zero is non-zero.
15654 2017-02-03 Jason Merrill <jason@redhat.com>
15657 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
15658 avoid copying non-taken branch.
15660 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15662 PR tree-optimization/79340
15663 * tree-vect-loop.c (vectorizable_reduction): Release
15664 vec_defs elements after safe_splicing them into other vectors.
15667 PR tree-optimization/79327
15668 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
15669 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
15671 (format_integer): Use wide_int_to_tree instead of build_int_cst
15672 + to_?hwi. If argmin is NULL, just set argmin and argmax to
15673 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
15674 of shortest and longest sequence.
15676 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
15678 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
15679 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
15681 2017-02-03 Walter Lee <walt@tilera.com>
15684 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
15685 after initial stackframe link reg save.
15686 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
15688 2017-02-03 Jakub Jelinek <jakub@redhat.com>
15691 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
15692 wu for stxssp alternative.
15694 2017-02-03 Martin Sebor <msebor@redhat.com>
15696 PR tree-optimization/79352
15697 * gimple-fold.c (get_range_strlen): Add argument.
15698 (get_range_strlen): Change return type to bool.
15699 (get_maxval_strlen): Pass in a dummy argument.
15700 * gimple-fold.h (get_range_strlen): Change return type to bool.
15701 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
15702 * tree.h (array_at_struct_end_p): Add argument.
15703 * tree.c (array_at_struct_end_p): Handle it.
15705 2017-02-03 Martin Liska <mliska@suse.cz>
15708 * multiple_target.c (create_dispatcher_calls): Redirect edge
15709 from a caller of a dispatcher.
15710 (expand_target_clones): Make the clones local.
15711 (ipa_target_clone): Do both target clones and resolvers.
15712 (ipa_dispatcher_calls): Remove the pass.
15713 (pass_dispatcher_calls::gate): Likewise.
15714 (make_pass_dispatcher_calls): Likewise.
15715 * passes.def (pass_target_clone): Put as very first IPA early
15718 2017-02-03 Martin Liska <mliska@suse.cz>
15720 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
15721 in case of a function with ifunc attribute.
15723 2017-02-03 Martin Liska <mliska@suse.cz>
15725 * cgraph.c (cgraph_node::dump): Dump function version info.
15726 * symtab.c (symtab_node::dump_base): Add missing new line.
15728 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15730 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
15731 (ifcombine_ifandif): Use it.
15733 2017-02-03 Martin Liska <mliska@suse.cz>
15735 * doc/invoke.texi: Document default value for
15736 use-after-scope-direct-emission-threshold.
15738 2017-02-03 Martin Liska <mliska@suse.cz>
15740 PR tree-optimization/79339
15741 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
15742 (format_floating): Likewise.
15744 2017-02-03 Martin Liska <mliska@suse.cz>
15747 * ipa-prop.c (ipa_node_params_t::insert): Remove current
15749 (ipa_node_params_t::remove): Likewise.
15750 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
15751 initialization from removed ipa_node_params_t::insert.
15752 (ipa_node_params::~ipa_node_params): Move from removed
15753 ipa_node_params_t::release.
15754 * symbol-summary.h (symbol_summary::m_released): New member.
15755 Do not release a summary twice. Do not allow to call finalizer
15756 for types of a summary that live in GGC memory.
15758 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15760 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
15763 2017-02-02 Martin Sebor <msebor@redhat.com>
15765 PR middle-end/79275
15766 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
15767 (format_string): Tighten up the range of output for non-constant
15768 strings and correct the expected range for wide non-constant strings.
15770 2017-02-02 Martin Sebor <msebor@redhat.com>
15772 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
15774 PR middle-end/32003
15775 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
15777 (-fdump-tree-@var): Add to index and document how to come up
15778 with pass-specific option and dump file names.
15779 (-fdump-passes): Clarify where to look for output.
15781 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
15783 PR middle-end/77445
15784 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
15785 statistics of the analyzed path; allow threading for speed when
15786 any of BBs along the path are optimized for speed.
15788 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
15790 PR middle-end/78468
15791 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
15792 settings of the virtual registers.
15795 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
15797 * explow.c (get_dynamic_stack_size): Take known alignment of stack
15798 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
15801 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15803 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
15804 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
15806 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15808 * config/s390/s390.md: Add missing comments with the expanded
15810 * config/s390/vector.md: Likewise.
15811 * config/s390/vx-builtins.md: Likewise.
15813 2017-02-02 Jakub Jelinek <jakub@redhat.com>
15816 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
15817 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
15818 conditions on a single line.
15820 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15822 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15823 __S390_VX__ to __VX__.
15825 2017-02-01 Andrew Pinski <apinski@cavium.com>
15827 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
15828 stmt_info to record_stmt_cost.
15829 (vect_get_known_peeling_cost): Pass stmt_info if known to
15831 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
15832 cpu_vector_cost field into
15833 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
15834 field into vec_int_stmt_cost and vec_fp_stmt_cost.
15835 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
15836 splitting of scalar_stmt_cost and vec_stmt_cost.
15837 (thunderx_vector_cost): Likewise.
15838 (cortexa57_vector_cost): LIkewise.
15839 (exynosm1_vector_cost): Likewise.
15840 (xgene1_vector_cost): Likewise.
15841 (thunderx2t99_vector_cost): Improve after the splitting of the two
15843 (aarch64_builtin_vectorization_cost): Update for the splitting of
15844 scalar_stmt_cost and vec_stmt_cost.
15846 2017-02-01 Torvald Riegel <triegel@redhat.com>
15847 Richard Henderson <rth@redhat.com>
15849 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
15850 conditional on existance of a fast atomic load.
15851 * optabs-query.c (can_atomic_load_p): New function.
15852 * optabs-query.h (can_atomic_load_p): Declare it.
15853 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
15854 no fast atomic load is available for the particular size of access.
15855 (expand_atomic_compare_and_swap): Likewise.
15856 (expand_atomic_load): Likewise.
15857 (expand_atomic_store): Likewise.
15858 (expand_atomic_fetch_op): Likewise.
15859 * testsuite/lib/target-supports.exp
15860 (check_effective_target_sync_int_128): Remove x86 because it provides
15861 no fast atomic load.
15862 (check_effective_target_sync_int_128_runtime): Likewise.
15864 2017-02-01 Richard Biener <rguenther@suse.de>
15866 * graphite.c: Include tree-vectorizer.h for find_loop_location.
15867 (graphite_transform_loops): Provide opt-info for optimized nests.
15868 * tree-parloop.c (parallelize_loops): Provide opt-info for
15869 parallelized loops.
15871 2017-02-01 Richard Biener <rguenther@suse.de>
15873 PR middle-end/79315
15874 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
15875 was not set before.
15877 2017-02-01 Richard Biener <rguenther@suse.de>
15879 PR tree-optimization/71824
15880 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15881 Verify the loops are valid in the merged SESE region.
15882 (scop_detection::can_represent_loop_1): Check analyzing the
15883 evolution of the number of iterations in the region succeeds.
15885 2017-01-31 Ian Lance Taylor <iant@golang.org>
15887 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
15888 REG_ARGS_SIZE note to 32-bit push insns and call insn.
15890 2017-01-31 David Malcolm <dmalcolm@redhat.com>
15892 PR preprocessor/79210
15893 * input.c (get_substring_ranges_for_loc): Replace line_width
15894 assertion with error-handling.
15896 2017-01-31 Richard Biener <rguenther@suse.de>
15898 PR tree-optimization/77318
15899 * graphite-sese-to-poly.c (extract_affine): Fix assert.
15900 (create_pw_aff_from_tree): Take loop parameter.
15901 (add_condition_to_pbb): Pass loop of the condition to
15902 create_pw_aff_from_tree.
15904 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15906 * config/s390/s390.c (s390_asan_shadow_offset): New function.
15907 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
15909 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
15913 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
15915 (convert_int_to_float128): Likewise.
15916 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
15917 (convert_int_to_float128): Likewise.
15918 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
15919 (UNSPEC_IEEE128_CONVERT): Likewise.
15920 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
15921 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
15922 Use local variables for IBM extended format.
15923 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
15924 (fix_trunc<mode>si2_fprs): Likewise.
15925 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
15926 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
15927 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
15928 to know that we can now have integers of all sizes in vector
15930 (fix<uns>_<mode>di2_hw): Likewise.
15931 (float<uns>_<mode>si2_hw): Likewise.
15932 (fix_<mode>si2_hw): Likewise.
15933 (fixuns_<mode>si2_hw): Likewise.
15934 (float<uns>_<mode>di2_hw): Likewise.
15935 (float_<mode>di2_hw): Likewise.
15936 (float_<mode>si2_hw): Likewise.
15937 (floatuns_<mode>di2_hw): Likewise.
15938 (floatuns_<mode>si2_hw): Likewise.
15939 (xscvqp<su>wz_<mode>): Delete, no longer used.
15940 (xscvqp<su>dz_<mode>): Likewise.
15941 (xscv<su>dqp_<mode>): Likewise.
15942 (ieee128_mfvsrd_64bit): Likewise.
15943 (ieee128_mfvsrd_32bit): Likewise.
15944 (ieee128_mfvsrwz): Likewise.
15945 (ieee128_mtvsrw): Likewise.
15946 (ieee128_mtvsrd_64bit): Likewise.
15947 (ieee128_mtvsrd_32bit): Likewise.
15949 2017-01-31 Martin Liska <mliska@suse.cz>
15952 * ipa-prop.c (ipa_free_all_node_params): Call release method
15953 instead of ~sumbol_summary to not to trigger double times
15956 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
15958 PR tree-optimization/71691
15959 * bitmap.h (class auto_bitmap): New.
15960 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
15961 is_maybe_undefined instead of ssa_undefined_value_p.
15963 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15965 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
15966 __S390_ARCH_LEVEL__ to __ARCH__.
15968 2017-01-31 Jakub Jelinek <jakub@redhat.com>
15970 PR tree-optimization/79267
15971 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
15972 if should_remove_lhs_p is true.
15974 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
15977 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
15978 (add_alignment_attribute): New.
15979 (base_type_die): Add alignment attribute.
15980 (subrange_type_die): Likewise.
15981 (modified_type_die): Likewise.
15982 (gen_array_type_die): Likewise.
15983 (gen_descr_array_type_die: Likewise.
15984 (gen_enumeration_type_die): Likewise.
15985 (gen_subprogram_die): Likewise.
15986 (gen_variable_die): Likewise.
15987 (gen_field_die): Likewise.
15988 (gen_ptr_to_mbr_type_die): Likewise.
15989 (gen_struct_or_union_type_die): Likewise.
15990 (gen_subroutine_type_die): Likewise.
15991 (gen_typedef_die): Likewise.
15992 (base_type_cmp): Compare alignment attribute.
15994 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15997 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
15998 (setb_unsigned) New pattern for setb with CCUNS.
15999 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16000 subfc./subfe sequence to avoid overflow problems. Generate a
16001 shorter sequence with cmpld/setb for power9.
16002 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16003 for generating subfc. instruction.
16004 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16005 now uses this instruction.
16007 2017-01-30 Ian Lance Taylor <iant@google.com>
16010 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16011 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16013 2017-01-30 Martin Sebor <msebor@redhat.com>
16015 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16016 Move constant to the right of a relational operator.
16017 (get_mpfr_format_length, format_character, format_string): Ditto.
16018 (should_warn_p, maybe_warn): Same.
16020 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16022 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16025 * asan.c (get_translation_unit_decl): Remove function.
16026 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16028 2017-01-30 Martin Liska <mliska@suse.cz>
16030 PR gcov-profile/79259
16031 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16032 -fprofile-generate.
16034 2017-01-30 Martin Liska <mliska@suse.cz>
16037 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16038 Initialize variables with NULL value.
16040 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16043 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16045 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16047 2017-01-30 Richard Biener <rguenther@suse.de>
16049 PR tree-optimization/79276
16050 * tree-vrp.c (process_assert_insertions): Properly adjust common
16051 when removing a duplicate.
16053 * gcc.dg/torture/pr79276.c: New testcase.
16055 2017-01-30 Richard Biener <rguenther@suse.de>
16057 PR tree-optimization/79256
16058 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16059 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16061 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16063 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16066 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16067 ("*r<noxa>sbg_<mode>_sll_bitmask")
16068 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16069 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16070 Use contiguous_bitmask_nowrap_operand.
16072 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16075 * config/rs6000/altivec.h (vec_xl): Revise #define.
16076 (vec_xst): Likewise.
16078 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16080 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16082 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16084 PR rtl-optimization/79194
16085 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16086 traps before call to bypass_conditional_jumps.
16088 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16090 PR tree-optimization/71374
16091 * lra-constraints.c (check_conflict_input_operands): New.
16092 (match_reload): Use it.
16094 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16097 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16098 account to calculate conflict_set.
16100 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16102 PR rtl-optimization/78559
16103 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16104 other_insn in combine.
16106 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16108 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16109 uint16_type_node for BT_UINT16.
16111 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16113 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16114 "RTL Tests" to menu.
16115 (GIMPLE Tests): New node.
16116 (RTL Tests): New node.
16118 2017-01-27 Richard Biener <rguenther@suse.de>
16120 PR tree-optimization/79245
16121 * tree-loop-distribution.c (distribute_loop): Apply cost
16122 modeling also to detected patterns.
16124 2017-01-27 Richard Biener <rguenther@suse.de>
16126 PR tree-optimization/71433
16127 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16128 (compare_assert_loc): New function.
16129 (process_assert_insertions): Sort and optimize assert locations
16130 to remove duplicates and push down identical assertions on
16131 edges to their destination block.
16133 2017-01-27 Richard Biener <rguenther@suse.de>
16135 PR tree-optimization/79244
16136 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16137 out SSA names even if abnormal.
16139 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16141 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16142 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16143 instead of MPFR_RNDN.
16145 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16148 * arm.c (arm_option_override): Don't call build_target_option_node
16149 until after doing all option overrides.
16150 (arm_valid_target_attribute_tree): Likewise.
16152 2017-01-27 Martin Liska <mliska@suse.cz>
16154 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16155 for {cd}tors and C++ {cd}tors.
16157 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16159 * config/s390/s390.md ("*setmem_long_and")
16160 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16162 2017-01-26 Martin Sebor <msebor@redhat.com>
16164 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16167 2017-01-26 Martin Sebor <msebor@redhat.com>
16169 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16170 HAVE_DFmode before using XFmode or DFmode.
16171 (parse_directive): Avoid using the z length modifier to avoid
16172 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16174 PR middle-end/78703
16175 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16176 to accept adjustment as an array.
16177 (get_int_range): New function.
16178 (struct directive): Make width and prec arrays.
16179 (directive::set_width, directive::set_precision): Call get_int_range.
16180 (format_integer, format_floating): Handle width and precision ranges.
16181 (format_string, parse_directive): Same.
16183 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16186 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16187 comdat_type_p set, just clone them, but keep the children in the
16191 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16192 which have direct callers with -fvar-tracking-assignments enabled
16194 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16195 inside of type units.
16197 2017-01-26 Martin Sebor <msebor@redhat.com>
16199 PR middle-end/78703
16200 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16202 (struct format_result): Replace number_chars, number_chars_min,
16203 and number_chars_max with a single member of struct result_range.
16205 (format_result::operator+=): Adjust.
16206 (struct fmtresult): Remove bounded. Handle likely and unlikely
16208 (fmtresult::adjust_for_width_or_precision): New function.
16209 (fmtresult:type_max_digits): New function.
16210 (bytes_remaining): Handle likely and unlikely counters.
16211 (min_bytes_remaining): Remove.
16212 (format_percent): Simplify.
16213 (format_integer, format_floating): Set likely and unlikely counters.
16214 (get_string_length, format_character, format_string): Same.
16215 (format_plain, should_warn_p): New function.
16216 (maybe_warn): Call should_warn_p. Update diagnostic messages
16217 and handle those for all directives, including plain strings.
16218 (format_directive): Handle likely and unlikely counters.
16219 Remove unnecessary quoting from diagnostics. Add an informational
16221 (add_bytes): Remove.
16222 (pass_sprintf_length::compute_format_length): Simplify.
16223 (try_substitute_return_value): Handle likely and unlikely counters.
16225 2017-01-26 Carl Love <cel@us.ibm.com>
16227 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16228 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16230 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16233 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16234 endianess for subregs into account.
16235 * lra-constraints.c (lra_constraints): Do risky transformations
16236 always on the first iteration.
16237 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16238 last_call_used_reg_set.
16239 (process_bb_lives): Define and use last_call_used_reg_set.
16240 * lra.c (lra): Always continue after lra_constraints on the first
16243 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16245 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16247 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16249 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16251 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16252 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16253 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16254 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16255 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16256 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16257 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16258 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16259 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16261 2017-01-26 Marek Polacek <polacek@redhat.com>
16264 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16265 for the third operand.
16267 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16269 PR middle-end/79236
16270 * omp-low.c (struct omp_context): Add simt_stmt field.
16271 (scan_omp_for): Return omp_context *.
16272 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16273 context to the _simt_ SIMD stmt.
16274 (lower_omp_for): For combined SIMD with sibling _simt_
16275 SIMD, make sure to use the same decls in _looptemp_
16276 clauses as in the sibling.
16278 2017-01-26 David Sherwood <david.sherwood@arm.com>
16280 PR middle-end/79212
16281 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16284 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16287 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16288 emit fld b; fld a; if possible.
16290 * brig-builtins.def: Update copyright years.
16291 * config/arm/arm_acle_builtins.def: Update copyright years.
16293 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16296 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16297 constraint instead of o for the stxsd instruction.
16299 2017-01-25 Carl Love <cel@us.ibm.com>
16301 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16302 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16304 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16306 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16308 2017-01-25 Richard Biener <rguenther@suse.de>
16310 PR tree-optimization/69264
16311 * target.def (vector_alignment_reachable): Improve documentation.
16312 * doc/tm.texi: Regenerate.
16313 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16315 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16316 earlier changes with respect to TYPE_USER_ALIGN.
16317 (vector_alignment_reachable_p): Likewise. Improve dumping.
16319 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16322 * config/arm/arm.md (xordi3): Force constant operand into a register
16325 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16327 * doc/invoke.texi (-fstore-merging): Correct default optimization
16328 levels at which it is enabled.
16329 (-O): Move -fstore-merging from list to...
16332 2017-01-25 Richard Biener <rguenther@suse.de>
16335 * omp-expand.c: Include debug.h.
16336 (expand_omp_taskreg): Make sure to generate early debug before
16337 outlining anything from a function.
16338 (expand_omp_target): Likewise.
16339 (grid_expand_target_grid_body): Likewise.
16341 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16344 * asan.c (get_translation_unit_decl): New function.
16345 (asan_add_global): Extract modules file name from globals
16346 TRANSLATION_UNIT_DECL name.
16348 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16351 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16352 for long calls with APCS frame and VFP.
16354 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16356 * cfg.c (original_copy_tables_initialized_p): New function.
16357 * cfg.h (original_copy_tables_initialized_p): New decl.
16358 * cfgrtl.c (relink_block_chain): Guard the call to
16359 free_original_copy_tables with a call to
16360 original_copy_tables_initialized_p.
16361 * cgraph.h (symtab_node::native_rtl_p): New decl.
16362 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16363 (symtab_node::needed_p): Don't assert for early assembly output
16364 for __RTL functions.
16365 (cgraph_node::finalize_function): Set "force_output" for __RTL
16367 (cgraph_node::analyze): Bail out early for __RTL functions.
16368 (analyze_functions): Update assertion to support __RTL functions.
16369 (cgraph_node::expand): Bail out early for __RTL functions.
16370 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16372 * function.h (struct function): Update comment for field
16374 * gimple-expr.c: Include "tree-pass.h".
16375 (gimple_has_body_p): Return false for __RTL functions.
16376 * Makefile.in (OBJS): Add run-rtl-passes.o.
16377 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16379 (gcc::pass_manager::get_clean_slate): New accessor.
16380 * passes.c: Include "insn-addr.h".
16381 (should_skip_pass_p): Add logging. Update logic for running
16382 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16383 property-provider override so it is only done for gimple passes.
16385 (skip_pass): New function.
16386 (execute_one_pass): Call skip_pass when skipping passes.
16387 * read-md.c (md_reader::read_char): Support filtering
16388 the input to a subset of line numbers.
16389 (md_reader::md_reader): Initialize fields
16390 m_first_line and m_last_line.
16391 (md_reader::read_file_fragment): New function.
16392 * read-md.h (md_reader::read_file_fragment): New decl.
16393 (md_reader::m_first_line): New field.
16394 (md_reader::m_last_line): New field.
16395 * read-rtl-function.c (function_reader::create_function): Only
16396 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16397 curr_properties. Set DECL_INITIAL to a dummy block.
16398 (read_rtl_function_body_from_file_range): New function.
16399 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16401 * run-rtl-passes.c: New file.
16402 * run-rtl-passes.h: New file.
16404 2017-01-24 Jeff Law <law@redhat.com>
16406 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16409 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16411 PR tree-optimization/79159
16412 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16413 (record_nonwrapping_iv): Improve boundary using above function if no
16414 value range information.
16416 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16417 Martin Jambor <mjambor@suse.cz>
16419 * brig-builtins.def: New file.
16420 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16421 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16422 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16423 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16424 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16425 * builtin-types.def (BT_INT8): New.
16426 (BT_INT16): Likewise.
16427 (BT_UINT8): Likewise.
16428 (BT_UINT16): Likewise.
16429 (BT_FN_ULONG): Likewise.
16430 (BT_FN_UINT_INT): Likewise.
16431 (BT_FN_UINT_ULONG): Likewise.
16432 (BT_FN_UINT_LONG): Likewise.
16433 (BT_FN_UINT_PTR): Likewise.
16434 (BT_FN_ULONG_PTR): Likewise.
16435 (BT_FN_INT8_FLOAT): Likewise.
16436 (BT_FN_INT16_FLOAT): Likewise.
16437 (BT_FN_UINT32_FLOAT): Likewise.
16438 (BT_FN_UINT16_FLOAT): Likewise.
16439 (BT_FN_UINT8_FLOAT): Likewise.
16440 (BT_FN_UINT64_FLOAT): Likewise.
16441 (BT_FN_UINT16_UINT32): Likewise.
16442 (BT_FN_UINT32_UINT16): Likewise.
16443 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16444 (BT_FN_INT_PTR_INT): Likewise.
16445 (BT_FN_UINT_PTR_UINT): Likewise.
16446 (BT_FN_LONG_PTR_LONG): Likewise.
16447 (BT_FN_ULONG_PTR_ULONG): Likewise.
16448 (BT_FN_VOID_UINT64_UINT64): Likewise.
16449 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16450 (BT_FN_INT8_INT8_INT8): Likewise.
16451 (BT_FN_INT16_INT16_INT16): Likewise.
16452 (BT_FN_INT_INT_INT): Likewise.
16453 (BT_FN_UINT_FLOAT_UINT): Likewise.
16454 (BT_FN_FLOAT_UINT_UINT): Likewise.
16455 (BT_FN_ULONG_UINT_UINT): Likewise.
16456 (BT_FN_ULONG_UINT_PTR): Likewise.
16457 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16458 (BT_FN_UINT_UINT_UINT): Likewise.
16459 (BT_FN_VOID_UINT_PTR): Likewise.
16460 (BT_FN_UINT_UINT_PTR: Likewise.
16461 (BT_FN_UINT32_UINT64_PTR): Likewise.
16462 (BT_FN_INT_INT_UINT_UINT): Likewise.
16463 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16464 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16465 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16466 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16467 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16468 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16469 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16470 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16471 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16472 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16473 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16474 * doc/frontends.texi: List BRIG FE.
16475 * doc/install.texi (Testing): Add BRIG tesring requirements.
16476 * doc/invoke.texi (Overall Options): Mention BRIG.
16477 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16479 2017-01-24 Richard Biener <rguenther@suse.de>
16481 PR translation/79208
16482 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16484 2017-01-24 Martin Jambor <mjambor@suse.cz>
16487 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16488 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16489 and known_contexts.
16491 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16493 PR middle-end/79123
16494 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16495 casts from signed to unsigned really don't have a range.
16497 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16499 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16500 GMP_RNDx for compatiblity.
16502 2017-01-24 Martin Liska <mliska@suse.cz>
16505 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16506 that would prevent us to call alloca with -1 as argument.
16508 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16510 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16511 Avoid -Wformat-security warning.
16513 2017-01-23 Andrew Pinski <apinski@cavium.com>
16515 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16518 2017-01-23 Martin Sebor <msebor@redhat.com>
16520 PR middle-end/78703
16521 * gimple-ssa-sprintf.c (warn_level): New global.
16522 (format_integer): Use it here and throughout the rest of the file.
16523 Use the same switch to compute sign as base.
16524 (maybe_warn): New function.
16525 (format_directive): Factor out warnings into maybe_warn.
16526 Add debugging output. Use warn_level.
16527 (add_bytes): Use warn_level.
16528 (pass_sprintf_length::compute_format_length): Add debugging output.
16529 (try_substitute_return_value): Same.
16530 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
16532 PR middle-end/78703
16533 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
16534 (struct fmtresult, format_integer, format_floating): Adjust.
16535 (fmtresult::fmtresult): Set max correctly in two argument ctor.
16536 (get_string_length, format_string,format_directive): Same.
16537 (pass_sprintf_length::compute_format_length): Same.
16538 (try_substitute_return_value): Simplify slightly.
16540 PR middle-end/78703
16541 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
16542 (fmtresult::operator+=): Outlined.
16543 (struct fmtresult): Add ctors.
16544 (struct conversion_spec): Rename...
16545 (struct directive): ...to this. Add and remove data members.
16546 (directive::set_width, directive::set_precision): New functions.
16547 (format_percent): Use fmtresult ctor.
16548 (get_width_and_precision): Remove.
16549 (format_integer): Make naming changes. Avoid computing width and
16551 (format_floating): Same. Adjust indentation.
16552 (format_character, format_none): New functions.
16553 (format_string): Moved character handling to format_character.
16554 (format_directive): Remove arguments, change return type.
16555 (parse_directive): New function.
16556 (pass_sprintf_length::compute_format_length): Move directive
16557 parsing to parse_directive.
16559 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16561 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
16562 (assign_assembler_name_if_needed): ... this.
16563 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
16564 (assign_assembler_name_if_needed): ... this.
16565 (free_lang_data_in_cgraph): Adjust callers.
16566 * cgraphunit.c (cgraph_node::analyze): Likewise.
16567 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
16570 2017-01-23 Richard Biener <rguenther@suse.de>
16572 PR tree-optimization/79088
16573 PR tree-optimization/79188
16574 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
16575 resetting loop bounds after last path deletion. Reset loop
16576 bounds of the target loop, make code match the comments.
16577 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
16578 Make sure loops need no fixups.
16580 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16582 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
16583 exponent support with double type for first argument.
16584 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
16585 type returned by __builtin_vec_extract_sig,
16586 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
16587 functions from "vector int" to "vector unsigned int" or from
16588 "vector long long int" to "vector unsigned long long int".
16589 Changed type returned by __builtin_vec_extract_exp,
16590 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
16591 functions from "vector int" to "vector unsigned int" or from
16592 "vector long long int" to "vector unsigned long long int".
16593 Changed return type of __builtin_vec_test_data_class,
16594 __builtin_vec_test_data_class_sp, and
16595 __builtin_vec_test_data_class_dp from "vector int" to
16596 "vector bool int" or from "vector long long int" to "vector bool
16597 long long int" and changed second argument type from "unsigned
16598 int" to "int". Added new overloaded function forms "vector float
16599 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
16600 "vector float __builtin_vec_insert_exp_sp (vector float, vector
16601 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
16602 double, vector unsigned long long int)" and "vector double
16603 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
16604 long int)". Changed return type of
16605 __builtin_scalar_test_data_class and
16606 __builtin_scalar_test_data_class_sp and
16607 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
16608 int" and changed second argument from "unsigned int" to "int".
16609 Changed type returned by __builtin_scalar_test_neg,
16610 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
16611 from "int" to "bool int". Added new overloaded function form
16612 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
16613 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
16614 exponent double-precision with floating point first argument.
16615 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16616 documentation of scalar_test_data_class, scalar_test_neg,
16617 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
16618 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
16619 vec_test_data_class built-in functions to reflect refinements in
16620 their type signatures.
16622 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
16624 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
16626 (aarch64_elf_asm_destructor): Likewise.
16628 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
16630 PR rtl-optimization/78634
16631 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
16632 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
16633 * ifcvt.c (noce_try_cmove): Add missing cost check.
16635 PR rtl-optimization/71724
16636 * combine.c (if_then_else_cond): Look for situations where it is
16637 beneficial to undo the work of one of the recursive calls.
16639 2017-01-23 Bin Cheng <bin.cheng@arm.com>
16641 PR tree-optimization/70754
16642 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
16643 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
16644 combined stmt before it if not NULL.
16645 (combine_chains): Process refs reversely and compute dominance point
16648 2017-01-23 Martin Liska <mliska@suse.cz>
16650 PR tree-optimization/79196
16651 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
16652 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
16654 (strlen_optimize_stmt): Call the renamed function.
16656 2017-01-23 Michael Matz <matz@suse.de>
16658 PR tree-optimization/78384
16659 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
16661 2017-01-23 Richard Biener <rguenther@suse.de>
16663 PR tree-optimization/79186
16664 * tree-vrp.c (register_new_assert_for): Make sure we've seen
16665 both incoming edges before moving an assert.
16667 2017-01-23 Martin Jambor <mjambor@suse.cz>
16669 * ipa-prop.c (load_from_param_1): Removed.
16670 (load_from_unmodified_param): Bits from load_from_param_1 put back
16672 (load_from_param): Removed.
16673 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
16674 with stmt. Reverted back to use of load_from_unmodified_param.
16676 2017-01-23 Martin Jambor <mjambor@suse.cz>
16679 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
16680 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
16681 field a pointer to garbage collected vector, mark lattices and
16682 ipcp_orig_node with GTY((skip)).
16683 (ipa_get_param_count): Adjust to descriptors being a pointer.
16684 (ipa_get_param): Likewise.
16685 (ipa_get_type): Likewise.
16686 (ipa_get_param_move_cost): Likewise.
16687 (ipa_set_param_used): Likewise.
16688 (ipa_get_controlled_uses): Likewise.
16689 (ipa_set_controlled_uses): Likewise.
16690 (ipa_is_param_used): Likewise.
16691 (ipa_node_params_t): Move into garbage collector. New methods insert
16693 (ipa_node_params_sum): Annotate wth GTY(()).
16694 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
16696 (ipa_load_from_parm_agg): Adjust declaration.
16697 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
16698 * ipa-profile.c (ipa_profile): Likewise.
16699 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
16700 (ipa_populate_param_decls): Make descriptors parameter garbage
16702 (ipa_dump_param): Adjust to descriptors being a pointer.
16703 (ipa_alloc_node_params): Likewise.
16704 (ipa_initialize_node_params): Likewise.
16705 (load_from_param_1): Make descriptors parameter garbage collected.
16706 (load_from_unmodified_param): Likewise.
16707 (load_from_param): Likewise.
16708 (ipa_load_from_parm_agg): Likewise.
16709 (ipa_node_params::~ipa_node_params): Removed.
16710 (ipa_free_all_node_params): Remove call to delete operator.
16711 (ipa_node_params_t::insert): New.
16712 (ipa_node_params_t::remove): Likewise.
16713 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
16714 copy known_csts and known_contexts vectors.
16715 (ipa_read_node_info): Adjust to descriptors being a pointer.
16716 (ipcp_modif_dom_walker): Make m_descriptors field garbage
16718 (ipcp_transform_function): Make descriptors variable garbage
16721 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
16723 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
16724 * config/i386/avx512dqintrin.h: Ditto.
16725 * config/i386/avx512fintrin.h: Ditto.
16726 * config/i386/i386.c: Handle new builtins.
16727 * config/i386/i386-builtin.def: Add new builtins.
16728 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
16729 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
16731 2017-01-23 Jakub Jelinek <jakub@redhat.com>
16732 Martin Liska <mliska@suse.cz>
16734 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
16735 * asan.c (asan_expand_poison_ifn): Support stores and use
16736 appropriate ASAN report function.
16737 * internal-fn.c (expand_ASAN_POISON_USE): New function.
16738 * internal-fn.def (ASAN_POISON_USE): Declare.
16739 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
16740 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
16741 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
16742 ASAN_POISON calls w/o LHS.
16743 * tree-ssa.c (execute_update_addresses_taken): Create clobber
16744 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
16745 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
16746 * gimplify.c (asan_poison_variables): Add attribute
16747 use_after_scope_memory to variables that really needs to live
16749 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
16750 having the attribute.
16752 2017-01-23 Martin Liska <mliska@suse.cz>
16754 * asan.c (create_asan_shadow_var): New function.
16755 (asan_expand_poison_ifn): Likewise.
16756 * asan.h (asan_expand_poison_ifn): New declaration.
16757 * internal-fn.c (expand_ASAN_POISON): Likewise.
16758 * internal-fn.def (ASAN_POISON): New builtin.
16759 * sanopt.c (pass_sanopt::execute): Expand
16760 asan_expand_poison_ifn.
16761 * tree-inline.c (copy_decl_for_dup_finish): Make function
16763 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
16764 * tree-ssa.c (is_asan_mark_p): New function.
16765 (execute_update_addresses_taken): Rewrite local variables
16766 (identified just by use-after-scope as addressable) into SSA.
16768 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16770 * doc/install.texi (Specific): opensource.apple.com uses https
16771 now. Remove trailing slash.
16773 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
16775 * README.Portability: Remove note on an Irix compatibility issue.
16777 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
16779 * gcov.c (INCLUDE_ALGORITHM): Define.
16780 (INCLUDE_VECTOR): Define.
16781 No longer include <vector> and <algorithm> directly.
16783 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
16785 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
16787 * doc/invoke.texi (Code Gen Options): Ditto.
16789 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
16792 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
16794 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
16796 rtl-optimization/79125
16797 * cprop.c (local_cprop_pass): Handle cases where we make an
16798 unconditional trap.
16800 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
16804 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
16805 read from, for big endian.
16807 2017-01-20 Jiong Wang <jiong.wang@arm.com>
16809 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
16810 register pauth builtins for LP64 only.
16812 2017-01-20 Marek Polacek <polacek@redhat.com>
16815 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
16818 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16820 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
16822 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
16823 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
16824 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
16826 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16829 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
16830 in big-endian mode when they are not a single duplicated value.
16832 2017-01-20 Richard Biener <rguenther@suse.de>
16834 * BASE-VER: Bump to 7.0.1.
16836 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
16838 * omp-low.c (omplow_simd_context): New struct. Use it...
16839 (lower_rec_simd_input_clauses): ...here and...
16840 (lower_rec_input_clauses): ...here to hold common data. Adjust all
16841 references to idx, lane, max_vf, is_simt.
16843 2017-01-20 Graham Markall <graham.markall@embecosm.com>
16845 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
16848 2017-01-20 Martin Jambor <mjambor@suse.cz>
16850 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
16851 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
16853 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
16854 (GTFILES): Rename hsa.c to hsa-common.c.
16855 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
16856 * hsa-dump.c: Likewise.
16857 * hsa-gen.c: Likewise.
16858 * hsa-regalloc.c: Likewise.
16859 * ipa-hsa.c: Likewise.
16860 * omp-expand.c: Likewise.
16861 * omp-low.c: Likewise.
16862 * toplev.c: Likewise.
16864 2017-01-20 Marek Polacek <polacek@redhat.com>
16867 * doc/invoke.texi: Document -Wduplicated-branches.
16868 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
16869 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
16870 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
16871 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
16872 return 0 only when not OEP_LEXICOGRAPHIC.
16873 (fold_build_cleanup_point_expr): Use the expression
16874 location when building CLEANUP_POINT_EXPR.
16875 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
16876 * tree.c (add_expr): Handle error_mark_node.
16878 2017-01-20 Martin Liska <mliska@suse.cz>
16881 * tree-profile.c (init_ic_make_global_vars): Do not call
16883 (gimple_init_gcov_profiler): Likewise.
16885 2017-01-20 Martin Liska <mliska@suse.cz>
16888 * cgraph.h (maybe_create_reference): Remove argument and
16890 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
16892 * ipa-cp.c (create_specialized_node): Likewise.
16893 * symtab.c (symtab_node::maybe_create_reference): Handle
16894 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
16896 2017-01-20 Martin Liska <mliska@suse.cz>
16898 * read-rtl-function.c (function_reader::create_function): Use
16899 build_decl instread of build_decl_stat.
16901 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
16903 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
16904 * config/i386/avx512dqintrin.h: Ditto.
16905 * config/i386/avx512fintrin.h: Ditto.
16906 * config/i386/i386-builtin-types.def: Add new types.
16907 * config/i386/i386.c: Handle new types.
16908 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
16909 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
16910 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
16911 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
16912 (__builtin_ia32_kshiftridi): New.
16913 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
16915 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
16919 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
16920 define to rs6000_init_stack_protect_guard.
16921 (rs6000_init_stack_protect_guard): New function.
16923 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16924 Yunqiang Su <yunqiang.su@imgtec.com>
16926 * config.gcc (supported_defaults): Add madd4.
16927 (with_madd4): Add validation.
16928 (all_defaults): Add madd4.
16929 * config/mips/mips.opt (mmadd4): New option.
16930 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
16932 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
16934 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
16935 (ISA_HAS_FUSED_MADD4): Likewise.
16936 * doc/invoke.texi (-mmadd4): Document the new option.
16937 * doc/install.texi (--with-madd4): Document the new option.
16939 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16941 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
16942 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
16943 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
16944 (aarch64_init_pauth_hint_builtins): New.
16945 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
16946 (aarch64_expand_builtin): Expand new builtins.
16948 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16950 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
16951 * combine-stack-adj.c (no_unhandled_cfa): Handle
16952 REG_CFA_TOGGLE_RA_MANGLE.
16953 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
16954 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
16955 info for return address signing.
16956 (aarch64_expand_epilogue): Likewise.
16958 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16960 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
16961 * config/aarch64/aarch64-protos.h
16962 (aarch64_return_address_signing_enabled): New declaration.
16963 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
16965 (aarch64_expand_prologue): Sign return address before it's pushed onto
16967 (aarch64_expand_epilogue): Authenticate return address fetched from
16969 (aarch64_override_options): Sanity check for ILP32 and ISA level.
16970 (aarch64_attributes): New function attributes for "sign-return-address".
16971 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
16972 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
16973 ("*do_return"): Generate combined instructions according to key index.
16974 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
16975 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
16977 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
16978 * config/aarch64/aarch64.opt (msign-return-address=): New.
16979 * doc/extend.texi (AArch64 Function Attributes): Documents
16980 "sign-return-address=".
16981 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
16983 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
16985 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
16986 overall option summary.
16988 2017-01-19 Jiong Wang <jiong.wang@arm.com>
16990 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
16991 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
16992 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
16993 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
16995 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
16997 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
16998 -mpower9-minmax by default for -mcpu=power9.
16999 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17000 128-bit floating point.
17002 2017-01-20 Alan Modra <amodra@gmail.com>
17004 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17005 optimizing for size.
17007 2017-01-20 Alan Modra <amodra@gmail.com>
17010 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17011 for strcmp and strncmp from corresponding builtin decl.
17013 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17015 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17016 instead of i386/rtems-64.h.
17017 * config/i386/rtems-64.h: Remove.
17019 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17023 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17025 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17027 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17029 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17030 Change int to HOST_WIDE_INT.
17031 * config/aarch64/aarch64-protos.h
17032 (aarch64_simd_gen_const_vector_dup): Likewise.
17033 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17035 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17037 * langhooks-def.h (lhd_type_for_size): New decl.
17038 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17039 * langhooks.c (lhd_type_for_size): New function, taken from
17042 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17044 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17045 define_bypass for CR latency.
17046 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17047 (power9-alu2): Add define_bypass for CR latency.
17049 (power9-mul): Update insn latency.
17050 (power9-mul-compare): Update insn latency, bypass latency and remove
17053 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17055 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17057 * config/aarch64/aarch64.md
17058 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17059 aarch64_nopcrelative_literal_loads.
17060 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17062 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17064 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17065 TARGET_LOONGSON_3A.
17066 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17068 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17071 * config.gcc (supported_defaults): Add lxc1-sxc1.
17072 (with_lxc1_sxc1): Add validation.
17073 (all_defaults): Add lxc1-sxc1.
17074 * config/mips/mips.opt (mlxc1-sxc1): New option.
17075 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17077 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17078 __mips_no_lxc1_sxc1.
17079 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17080 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17081 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17083 2017-01-19 Richard Biener <rguenther@suse.de>
17085 PR tree-optimization/72488
17086 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17087 sure to restore SSA info.
17088 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17090 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17092 PR rtl-optimization/79121
17093 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17094 of the inner type when shifting an extended value.
17096 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17099 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17100 interposable aliases.
17102 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17105 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17106 Use the evmergelohi instruction.
17107 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17108 (mov_sitf_e500_subreg8_2_be): Likewise.
17109 (mov_sitf_e500_subreg12_2_le): Likewise.
17110 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17111 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17112 (mov_sitf_e500_subreg8_2_le): Likewise.
17113 (mov_sitf_e500_subreg12_2_be): Likewise.
17115 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17117 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17118 attribute from vecsimple to vecperm.
17119 (altivec_vbpermq2): Likewise.
17121 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17124 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17126 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17127 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17128 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17129 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17130 case where N arg is SIZE_MAX.
17131 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17132 (cmpstrsi): Add pattern.
17134 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17136 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17137 __builtin_vec_revb builtins.
17138 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17139 built-in functions to support generation of the ISA 3.0 XXBR<x>
17140 vector byte reverse instructions.
17141 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17142 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17143 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17144 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17145 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17146 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17147 (P9V_BUILTIN_VEC_REVB): Likewise.
17148 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17149 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17150 (p9_xxbrq_v16qi): Likewise.
17151 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17152 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17153 (p9_xxbrh_v8hi): Likewise.
17154 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17155 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17156 vec_revb built-in functions.
17158 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17160 PR rtl-optimization/78952
17161 * config/i386/i386.md (any_extract): New code iterator.
17162 (*insvqi_2): Use any_extract for source operand.
17163 (*insvqi_3): Use any_shiftrt for source operand.
17165 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17167 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17169 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17171 2017-01-18 Matthias Klose <doko@ubuntu.com>
17173 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17175 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17177 * config/rs6000/altivec.h (vec_bperm): Change #define.
17178 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17179 (altivec_vbpermq2): New define_insn.
17180 (altivec_vbpermd): Likewise.
17181 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17182 function interface.
17183 (VBPERMD): Likewise.
17184 (VBPERM): New polymorphic function interface.
17185 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17186 Add entries for P9V_BUILTIN_VEC_VBPERM.
17187 * doc/extend.texi: Add interfaces for vec_bperm.
17189 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17191 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17192 first letter of error messages.
17193 (s390_resolve_overloaded_builtin): Likewise.
17194 * config/s390/s390.c (s390_expand_builtin): Likewise.
17195 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17196 (s390_valid_target_attribute_inner_p): Likewise.
17197 * config/s390/s390.md ("tabort"): Likewise.
17199 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17201 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17202 (ISA_AVOID_DIV_HILO): New macro.
17203 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17204 (ISA_HAS_DDIV): Likewise.
17206 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17208 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17210 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17212 * doc/invoke.texi (fabi-version): Spelling fix.
17214 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17217 * doc/invoke.texi (fabi-version): Mention mangling fix for
17220 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17223 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17225 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17228 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17231 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17233 (stack_protect_test): Ditto.
17234 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17235 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17237 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17238 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17239 -mstack-protector-guard-offset=.
17240 (RS/6000 and PowerPC Options): Ditto.
17242 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17244 * config/i386/i386.h (MASK_CLASS_P): New define.
17245 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17246 there are no registers from different register sets also when
17247 mask registers are used. Update function comment.
17248 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17249 to (*k/*r) and (*k/*km) alternatives.
17251 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17253 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17254 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17255 (EH_RETURN_HANDLER_RTX): New define.
17256 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17257 Force frame pointer in EH return functions.
17258 (aarch64_expand_epilogue): Add barrier for eh_return.
17259 (aarch64_final_eh_return_addr): Remove.
17260 (aarch64_eh_return_handler_rtx): New function.
17261 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17263 (aarch64_eh_return_handler_rtx): New prototype.
17265 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17267 * config/rs6000/altivec.h (vec_rlmi): New #define.
17268 (vec_vrlnm): Likewise.
17269 (vec_rlnm): Likewise.
17270 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17271 (UNSPEC_VRLNM): Likewise.
17272 (VIlong): New mode iterator.
17273 (altivec_vrl<VI_char>mi): New define_insn.
17274 (altivec_vrl<VI_char>nm): Likewise.
17275 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17277 (VRLDNM): Likewise.
17278 (RLNM): New polymorphic function entry.
17279 (VRLWMI): New monomorphic function entry.
17280 (VRLDMI): Likewise.
17281 (RLMI): New polymorphic function entry.
17282 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17283 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17284 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17287 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17290 * dwarf2out.c (field_byte_offset): Restore the
17291 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17292 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17293 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17296 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17299 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17301 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17304 * dwarf2out.c (add_data_member_location_attribute): For constant
17305 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17306 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17307 DW_AT_byte_size attributes.
17309 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17311 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17312 after forcing to constant memory when the code model is medium.
17314 2017-01-17 Julia Koval <julia.koval@intel.com>
17317 * config/i386/avx512fintrin.h
17318 (_mm512_i32gather_ps): Change __addr type to void const*.
17319 (_mm512_mask_i32gather_ps): Ditto.
17320 (_mm512_i32gather_pd): Ditto.
17321 (_mm512_mask_i32gather_pd): Ditto.
17322 (_mm512_i64gather_ps): Ditto.
17323 (_mm512_mask_i64gather_ps): Ditto.
17324 (_mm512_i64gather_pd): Ditto.
17325 (_mm512_mask_i64gather_pd): Ditto.
17326 (_mm512_i32gather_epi32): Ditto.
17327 (_mm512_mask_i32gather_epi32): Ditto.
17328 (_mm512_i32gather_epi64): Ditto.
17329 (_mm512_mask_i32gather_epi64): Ditto.
17330 (_mm512_i64gather_epi32): Ditto.
17331 (_mm512_mask_i64gather_epi32): Ditto.
17332 (_mm512_i64gather_epi64): Ditto.
17333 (_mm512_mask_i64gather_epi64): Ditto.
17334 (_mm512_i32scatter_ps): Change __addr type to void*.
17335 (_mm512_mask_i32scatter_ps): Ditto.
17336 (_mm512_i32scatter_pd): Ditto.
17337 (_mm512_mask_i32scatter_pd): Ditto.
17338 (_mm512_i64scatter_ps): Ditto.
17339 (_mm512_mask_i64scatter_ps): Ditto.
17340 (_mm512_i64scatter_pd): Ditto.
17341 (_mm512_mask_i64scatter_pd): Ditto.
17342 (_mm512_i32scatter_epi32): Ditto.
17343 (_mm512_mask_i32scatter_epi32): Ditto.
17344 (_mm512_i32scatter_epi64): Ditto.
17345 (_mm512_mask_i32scatter_epi64): Ditto.
17346 (_mm512_i64scatter_epi32): Ditto.
17347 (_mm512_mask_i64scatter_epi32): Ditto.
17348 (_mm512_i64scatter_epi64): Ditto.
17349 (_mm512_mask_i64scatter_epi64): Ditto.
17350 * config/i386/avx512pfintrin.h
17351 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17352 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17353 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17354 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17355 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17356 (_mm512_prefetch_i32scatter_ps): Ditto.
17357 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17358 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17359 (_mm512_prefetch_i64scatter_pd): Ditto.
17360 (_mm512_prefetch_i64scatter_ps): Ditto.
17361 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17362 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17363 * config/i386/avx512vlintrin.h
17364 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17365 (_mm_mmask_i32gather_ps): Ditto.
17366 (_mm256_mmask_i32gather_pd): Ditto.
17367 (_mm_mmask_i32gather_pd): Ditto.
17368 (_mm256_mmask_i64gather_ps): Ditto.
17369 (_mm_mmask_i64gather_ps): Ditto.
17370 (_mm256_mmask_i64gather_pd): Ditto.
17371 (_mm_mmask_i64gather_pd): Ditto.
17372 (_mm256_mmask_i32gather_epi32): Ditto.
17373 (_mm_mmask_i32gather_epi32): Ditto.
17374 (_mm256_mmask_i32gather_epi64): Ditto.
17375 (_mm_mmask_i32gather_epi64): Ditto.
17376 (_mm256_mmask_i64gather_epi32): Ditto.
17377 (_mm_mmask_i64gather_epi32): Ditto.
17378 (_mm256_mmask_i64gather_epi64): Ditto.
17379 (_mm_mmask_i64gather_epi64): Ditto.
17380 (_mm256_i32scatter_ps): Change __addr type to void*.
17381 (_mm256_mask_i32scatter_ps): Ditto.
17382 (_mm_i32scatter_ps): Ditto.
17383 (_mm_mask_i32scatter_ps): Ditto.
17384 (_mm256_i32scatter_pd): Ditto.
17385 (_mm256_mask_i32scatter_pd): Ditto.
17386 (_mm_i32scatter_pd): Ditto.
17387 (_mm_mask_i32scatter_pd): Ditto.
17388 (_mm256_i64scatter_ps): Ditto.
17389 (_mm256_mask_i64scatter_ps): Ditto.
17390 (_mm_i64scatter_ps): Ditto.
17391 (_mm_mask_i64scatter_ps): Ditto.
17392 (_mm256_i64scatter_pd): Ditto.
17393 (_mm256_mask_i64scatter_pd): Ditto.
17394 (_mm_i64scatter_pd): Ditto.
17395 (_mm_mask_i64scatter_pd): Ditto.
17396 (_mm256_i32scatter_epi32): Ditto.
17397 (_mm256_mask_i32scatter_epi32): Ditto.
17398 (_mm_i32scatter_epi32): Ditto.
17399 (_mm_mask_i32scatter_epi32): Ditto.
17400 (_mm256_i32scatter_epi64): Ditto.
17401 (_mm256_mask_i32scatter_epi64): Ditto.
17402 (_mm_i32scatter_epi64): Ditto.
17403 (_mm_mask_i32scatter_epi64): Ditto.
17404 (_mm256_i64scatter_epi32): Ditto.
17405 (_mm256_mask_i64scatter_epi32): Ditto.
17406 (_mm_i64scatter_epi32): Ditto.
17407 (_mm_mask_i64scatter_epi32): Ditto.
17408 (_mm256_i64scatter_epi64): Ditto.
17409 (_mm256_mask_i64scatter_epi64): Ditto.
17410 (_mm_i64scatter_epi64): Ditto.
17411 (_mm_mask_i64scatter_epi64): Ditto.
17412 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17413 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17414 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17415 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17416 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17417 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17418 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17419 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17420 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17421 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17422 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17423 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17424 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17425 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17426 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17427 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17428 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17429 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17430 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17431 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17432 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17433 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17434 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17435 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17436 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17437 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17438 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17439 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17440 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17441 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17442 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17443 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17444 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17445 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17446 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17447 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17448 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17449 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17450 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17451 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17452 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17453 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17454 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17455 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17456 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17457 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17458 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17459 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17460 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17461 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17462 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17463 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17464 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17465 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17466 definitions accordingly.
17468 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17469 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17472 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17475 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17478 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17479 hard regs for inner regno.
17481 2017-01-17 Martin Liska <mliska@suse.cz>
17484 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17485 assumption and add comment.
17487 2017-01-17 Nathan Sidwell <nathan@acm.org>
17489 * ipa-visibility.c (localize_node): New function, broken out of ...
17490 (function_and_variable_visibility): ... here. Call it.
17492 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17494 PR middle-end/77445
17495 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17496 correctly set frequency of oudgoing edge.
17497 (duplicate_thread_path): Fix profile updating.
17499 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17502 * configure.ac: Add GCC_BASE_VER.
17503 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17504 version from BASE-VER file.
17505 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17506 (gcc.o): Depend on $(BASEVER).
17507 * common.opt (dumpfullversion): New option.
17508 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17509 * doc/invoke.texi: Document -dumpfullversion.
17510 * doc/install.texi: Document --with-gcc-major-version-only.
17511 * configure: Regenerated.
17513 2017-01-17 Richard Biener <rguenther@suse.de>
17515 PR tree-optimization/71433
17516 * tree-vrp.c (register_new_assert_for): Merge same asserts
17517 on all incoming edges.
17518 (process_assert_insertions_for): Handle insertions at the
17521 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17523 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17524 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17526 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
17529 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
17532 2017-01-17 Alan Modra <amodra@gmail.com>
17535 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
17536 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
17537 symbolic stack limit when pic.
17539 2017-01-16 Martin Sebor <msebor@redhat.com>
17541 PR tree-optimization/78608
17542 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
17544 2017-01-16 Jeff Law <law@redhat.com>
17547 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
17548 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
17549 for several include directories that may be relative to sysroot.
17550 * config/i386/x-mingw32 (gplus_includedir): Define.
17551 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
17552 (native_system_includedir): Likewise.
17553 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
17554 override if TARGET_SYSTEM_ROOT is defined.
17555 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
17557 PR tree-optimization/79090
17558 PR tree-optimization/33562
17559 PR tree-optimization/61912
17560 PR tree-optimization/77485
17561 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
17562 and computed trims into the dump file.
17564 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17566 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
17568 2017-01-16 Jakub Jelinek <jakub@redhat.com>
17571 * gimplify.c (gimplify_init_constructor): If want_value and
17572 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
17576 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
17577 sequence. Formatting fixes.
17578 (doloop_optimize): Formatting fixes.
17581 * gcc.c (debug_level_greater_than_spec_func): New function.
17582 (static_spec_functions): Add debug-level-gt spec function.
17583 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
17585 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17586 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
17587 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
17588 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
17589 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
17590 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
17592 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
17594 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
17595 QImode fixups to general and mask registers only.
17597 2017-01-16 Carl Love <cel@us.ibm.com>
17599 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
17600 for built-in functions
17601 vector signed char vec_nabs (vector signed char)
17602 vector signed short vec_nabs (vector signed short)
17603 vector signed int vec_nabs (vector signed int)
17604 vector signed long long vec_nabs (vector signed long long)
17605 vector float vec_nabs (vector float)
17606 vector double vec_nabs (vector double)
17607 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
17609 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
17610 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
17611 * doc/extend.texi: Update the documentation file for the new built-in
17614 2017-01-16 Martin Sebor <msebor@redhat.com>
17616 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
17619 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17621 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
17622 UNSPEC_VSX__XXSPLTD to require special splat handling.
17624 2017-01-16 David Malcolm <dmalcolm@redhat.com>
17627 * system.h: Poison strndup.
17629 2017-01-16 Alan Modra <amodra@gmail.com>
17632 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
17635 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
17637 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
17639 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
17641 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
17642 call recog here. Assert that INSN_CODE (insn) is non-negative.
17644 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17647 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
17649 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
17650 in the currently scheduled RTL fragment.
17652 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
17654 PR rtl-optimization/78751
17655 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
17658 2017-01-14 Jeff Law <law@redhat.com>
17660 PR tree-optimization/79090
17661 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
17662 variable length stores.
17663 (compute_trims): Delete dead assignment to *trim_tail.
17664 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
17667 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
17669 PR rtl-optimization/78626
17670 PR rtl-optimization/78727
17671 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
17672 of a block, and split such blocks after everything else is finished.
17674 2017-01-14 Alan Modra <amodra@gmail.com>
17677 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
17678 target legitimate_combined_insn.
17679 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
17680 (rs6000_legitimate_combined_insn): New function.
17681 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
17683 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
17684 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
17685 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
17687 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
17689 * doc/frontends.texi (G++ and GCC): Remove references to Java.
17691 2017-01-13 Jeff Law <law@redhat.com>
17693 PR tree-optimization/33562
17694 PR tree-optimization/61912
17695 PR tree-optimization/77485
17696 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
17698 (delete_dead_assignment): Likewise.
17699 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
17700 statement to delete_dead_call and delete_dead_assignment.
17702 2017-01-13 David Malcolm <dmalcolm@redhat.com>
17705 * substring-locations.c (format_warning_va): Strengthen case 1 so
17706 that both endpoints of the substring must be within the format
17707 range for just the substring to be printed.
17709 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
17711 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
17712 * config/i386/i386.c (ix86_target_string): Add missing options
17713 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
17714 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
17715 flags_other and ix86_target_other to flags2_other. Display unknown
17717 (ix86_valid_target_attribute_inner_p): Add missing options and
17718 reorder options by implied ISAs, as in ix86_target_string.
17720 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17722 * hash-table.h (hash_table::too_empty_p): New function.
17723 (hash_table::expand): Use it.
17724 (hash_table::traverse): Likewise.
17725 (hash_table::empty_slot): Use sizeof (value_type) instead of
17726 sizeof (PTR) to convert bytes to elements. Shrink the table
17727 if the current size is excessive for the current number of
17730 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
17732 * ira-costs.c (record_reg_classes): Break from the inner loop
17733 early once alt_fail is known to be true. Update outer loop
17734 handling accordingly.
17736 2017-01-13 Jeff Law <law@redhat.com>
17738 * tree-ssa-dse.c (decrement_count): New function.
17739 (increment_start_addr, maybe_trim_memstar_call): Likewise.
17740 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
17741 when we know the partially dead statement is a mem* function.
17743 PR tree-optimization/61912
17744 PR tree-optimization/77485
17745 * tree-ssa-dse.c: Include expr.h.
17746 (maybe_trim_constructor_store): New function.
17747 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
17749 PR tree-optimization/33562
17750 PR tree-optimization/61912
17751 PR tree-optimization/77485
17752 * doc/invoke.texi: Document new dse-max-object-size param.
17753 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
17754 * tree-ssa-dse.c: Include params.h.
17755 (dse_store_status): New enum.
17756 (initialize_ao_ref_for_dse): New, partially extracted from
17758 (valid_ao_ref_for_dse, normalize_ref): New.
17759 (setup_live_bytes_from_ref, compute_trims): Likewise.
17760 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
17761 (maybe_trim_partially_dead_store): Likewise.
17762 (maybe_trim_complex_store): Likewise.
17763 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
17764 Track what bytes live from the original store. Return tri-state
17765 for dead, partially dead or live.
17766 (dse_dom_walker): Add constructor, destructor and new private members.
17767 (delete_dead_call, delete_dead_assignment): New extracted from
17769 (dse_optimize_stmt): Make a member of dse_dom_walker.
17770 Use initialize_ao_ref_for_dse.
17772 PR tree-optimization/33562
17773 PR tree-optimization/61912
17774 PR tree-optimization/77485
17775 * sbitmap.h (bitmap_count_bits): Prototype.
17776 (bitmap_clear_range, bitmap_set_range): Likewise.
17777 * sbitmap.c (bitmap_clear_range): New function.
17778 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
17780 2017-01-13 Martin Liska <mliska@suse.cz>
17783 * function.c (set_cfun): Add new argument force.
17784 * function.h (set_cfun): Likewise.
17785 * ipa-inline-transform.c (inline_call): Use the function when
17786 strict alising from is dropped for function we inline to.
17788 2017-01-13 Richard Biener <rguenther@suse.de>
17790 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
17791 for dumping GIMPLE INTEGER_CSTs.
17793 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17795 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
17796 to 201112L since C++17.
17798 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17801 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
17802 if -fsanitize=kernel-address is present.
17804 2017-01-13 Richard Biener <rguenther@suse.de>
17806 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
17807 as _Literal ( type ) number in case usual suffixes do not
17808 preserve all information.
17810 2017-01-13 Richard Biener <rguenther@suse.de>
17812 PR tree-optimization/77283
17813 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
17814 and ssa-iterators.h.
17815 (is_feasible_trace): Implement a cost model based on joiner
17818 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17821 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
17822 char or short to __float128/_Float128 directly.
17824 2017-01-12 Martin Sebor <msebor@redhat.com>
17826 to -Wformat-overflow.
17827 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
17828 (min_bytes_remaining): Same.
17829 (get_string_length): Same.
17830 (format_string): Same.
17831 (format_directive): Same.
17833 (pass_sprintf_length::handle_gimple_call): Same.
17835 2017-01-12 Jakub Jelinek <jakub@redhat.com>
17837 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
17838 info.nowrite calls with no lhs that can't throw. Return bool
17839 whether gsi_remove has been called or not.
17840 (pass_sprintf_length::handle_gimple_call): Return bool whether
17841 try_substitute_return_value called gsi_remove. Formatting fix.
17842 (pass_sprintf_length::execute): Don't use gsi_remove if
17843 handle_gimple_call returned true.
17846 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
17847 be removed due to side-effects, don't remove following barrier nor
17848 turn the successor edge into fallthru edge.
17850 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17853 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
17854 element-reversing loads and stores as not swappable.
17856 2017-01-12 Nathan Sidwell <nathan@acm.org>
17857 Nicolai Stange <nicstange@gmail.com>
17859 * combine.c (try_combine): Don't ignore result of overlap checking
17860 loop. Combine overlap & asm check into single loop.
17862 2017-01-12 Richard Biener <rguenther@suse.de>
17864 * tree-pretty-print.c (dump_generic_node): Provide -gimple
17865 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
17867 2017-01-12 Richard Biener <rguenther@suse.de>
17869 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
17870 and TS_TARGET_OPTION directly derive from TS_BASE.
17871 * tree-core.h (tree_optimization_option): Derive from tree_base.
17872 (tree_target_option): Likewise.
17874 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
17876 * config/i386/i386.c (memory_address_length): Increase len
17877 only when rip_relative_addr_p returns false.
17879 2017-01-11 Julia Koval <julia.koval@intel.com>
17881 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
17882 (OPTION_MASK_ISA_SGX_SET): New.
17883 (ix86_handle_option): Handle OPT_msgx.
17884 * config.gcc: Added sgxintrin.h.
17885 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
17886 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
17887 * config/i386/i386.c (ix86_target_string): Add -msgx.
17889 (ix86_option_override_internal): Handle new options.
17890 (ix86_valid_target_attribute_inner_p): Add sgx.
17891 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
17892 * config/i386/i386.opt: Add msgx.
17893 * config/i386/sgxintrin.h: New file.
17894 * config/i386/x86intrin.h: Add sgxintrin.h.
17896 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17899 * fold-const.c (maybe_nonzero_address): Return 1 for function
17901 (tree_single_nonzero_warnv_p): Don't handle function local objects
17905 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
17908 2017-01-11 David Malcolm <dmalcolm@redhat.com>
17911 * opts.c: Include "spellcheck.h"
17912 (struct string_fragment): New struct.
17913 (struct edit_distance_traits<const string_fragment &>): New
17915 (get_closest_sanitizer_option): New function.
17916 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
17918 2017-01-11 Jakub Jelinek <jakub@redhat.com>
17920 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
17922 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
17923 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
17924 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
17925 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
17926 for initial die_offset if dwarf_split_debug_info.
17927 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
17928 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
17930 (output_skeleton_debug_sections): Formatting fix. Use
17931 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
17932 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
17934 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
17936 * config/arm/cortex-a53.md: Add bypasses for
17937 cortex_a53_r2f_cvt.
17938 (cortex_a53_r2f): Only use for transfers.
17939 (cortex_a53_f2r): Likewise.
17940 (cortex_a53_r2f_cvt): Add reservation for conversions.
17941 (cortex_a53_f2r_cvt): Likewise.
17943 2017-01-11 Tamar Christina <tamar.christina@arm.com>
17945 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
17946 to all inlined functions, change static to extern.
17948 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
17951 * config/arm/arm.c (legitimize_pic_address): Handle reference to
17953 (arm_assemble_integer): Likewise.
17955 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
17957 * config.gcc: Use new awk script to check CPU, FPU and architecture
17958 parameters for --with-... options.
17959 * config/arm/parsecpu.awk: New file
17960 * config/arm/arm-cpus.in: New file.
17961 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
17963 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
17965 * config/arm/t-arm: Update dependency rules.
17966 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
17967 of processing .def files.
17968 * config/arm/genopt.sh: Deleted.
17969 * config/arm/gentune.sh: Deleted.
17970 * config/arm/arm-cores.def: Deleted.
17971 * config/arm/arm-arches.def: Deleted.
17972 * config/arm/arm-fpus.def: Deleted.
17973 * config/arm/arm-tune.md: Regenerated.
17974 * config/arm/arm-tables.opt: Regenerated.
17975 * config/arm/arm-cpu.h: New generated file.
17976 * config/arm/arm-cpu-data.h: New generated file.
17977 * config/arm/arm-cpu-cdata.h: New generated file.
17979 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
17982 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
17984 (input_varpool_node): Unpack dynamically_initialized bit.
17986 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
17988 PR rtl-optimization/79032
17989 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
17990 the alignment of the adjusted memory reference against that of MODE,
17991 instead of the alignment of the original memory reference.
17993 2017-01-11 Martin Jambor <mjambor@suse.cz>
17995 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
17997 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
17998 decorated functions.
18000 2017-01-11 Richard Biener <rguenther@suse.de>
18002 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18003 set range/nonnull info for PHI results. Do not set it on
18004 stmts marked for removal.
18006 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18008 * expr.c (store_field): In the bitfield case, fetch the return value
18009 from the registers before applying a single big-endian adjustment.
18010 Always do a final load for a BLKmode value not larger than a word.
18012 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18015 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18016 that we correctly handle column numbers greater than
18017 LINE_MAP_MAX_COLUMN_NUMBER.
18019 2017-01-10 Martin Sebor <msebor@redhat.com>
18021 PR middle-end/78245
18022 * gimple-ssa-sprintf.c (get_destination_size): Call
18023 {init,fini}object_sizes.
18024 * tree-object-size.c (addr_object_size): Adjust.
18025 (pass_through_call): Adjust.
18026 (pass_object_sizes::execute): Adjust.
18027 * tree-object-size.h (fini_object_sizes): Declare.
18029 2017-01-10 Martin Sebor <msebor@redhat.com>
18031 PR tree-optimization/78775
18032 * builtins.c (get_size_range): Move...
18033 * calls.c: ...to here.
18034 (alloc_max_size): Accept zero argument.
18035 (operand_signed_p): Remove.
18036 (maybe_warn_alloc_args_overflow): Call get_size_range.
18037 * calls.h (get_size_range): Declare.
18039 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18041 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18042 from TI's devices.csv file as of September 2016.
18043 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18045 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18047 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18048 * doc/invoke.texi: Likewise.
18049 * doc/md.texi: Likewise.
18050 * doc/objc.texi: Likewise.
18052 2017-01-10 Joshua Conner <joshconner@google.com>
18054 * config/arm/fuchsia-elf.h: New file.
18055 * config/fuchsia.h: New file.
18056 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18057 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18059 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18061 2016-01-10 Richard Biener <rguenther@suse.de>
18063 PR tree-optimization/79034
18064 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18065 Propagate out degenerate PHIs in the joiner.
18067 2017-01-10 Martin Liska <mliska@suse.cz>
18069 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18070 (sort_congruence_classes_by_decl_uid): Likewise.
18071 (sort_congruence_class_groups_by_decl_uid): Likewise.
18072 (sem_item_optimizer::merge_classes): Sort class, groups in these
18073 classes and members in the groups by DECL_UID of declarations.
18074 This would make merge operations stable.
18076 2017-01-10 Martin Liska <mliska@suse.cz>
18078 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18079 usage of m_classes_vec.
18080 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18081 (sem_item_optimizer::get_group_by_hash): Likewise.
18082 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18083 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18084 (sem_item_optimizer::verify_classes): Likewise.
18085 (sem_item_optimizer::process_cong_reduction): Likewise.
18086 (sem_item_optimizer::dump_cong_classes): Likewise.
18087 (sem_item_optimizer::merge_classes): Likewise.
18088 * ipa-icf.h (congruence_class_hash): Rename from
18089 congruence_class_group_hash. Remove declaration of m_classes_vec.
18091 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18093 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18094 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18095 * config.gcc: Add avx512vpopcntdqintrin.h.
18096 * config/i386/avx512vpopcntdqintrin.h: New.
18097 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18098 * config/i386/i386-builtin-types.def: Add new types.
18099 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18100 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18101 __builtin_ia32_vpopcountq_v8di_mask): New.
18102 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18103 __AVX512VPOPCNTDQ__.
18104 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18105 (PTA_AVX512VPOPCNTDQ): Define.
18106 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18107 TARGET_AVX512VPOPCNTDQ_P): Define.
18108 * config/i386/i386.opt: Add mavx512vpopcntdq.
18109 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18110 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18112 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18114 PR middle-end/77484
18115 * predict.def (PRED_CALL): Set to 67.
18117 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18119 * expr.c (store_field): In the bitfield case, if the value comes from
18120 a function call and is of an aggregate type returned in registers, do
18121 not modify the field mode; extract the value in all cases if the mode
18122 is BLKmode and the size is not larger than a word.
18124 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18127 * config/i386/cpuid.h: Fix undefined behavior.
18129 2017-01-04 Jeff Law <law@redhat.com>
18131 PR tree-optimization/79007
18132 PR tree-optimization/67955
18133 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18134 conservative for pt.null when flag_non_call_exceptions is on.
18136 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18138 PR translation/79019
18139 PR translation/79020
18140 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18141 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18142 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18144 * config/avr/avr.opt (maccumulate-args): Likewise.
18145 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18146 * common.opt (freport-bug): Likewise.
18147 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18148 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18149 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18150 translatable string.
18151 * config/i386/i386.c (function_value_32): Likewise.
18152 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18153 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18155 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18156 * common/config/msp430/msp430-common.c (msp430_handle_option):
18158 * symtab.c (symtab_node::verify_base): Likewise.
18159 * opts.c (set_debug_level): Likewise.
18160 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18161 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18162 missing whitespace to translatable strings.
18163 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18164 * config/sh/superh.h: Likewise.
18165 * config/i386/xopintrin.h: Likewise.
18166 * config/i386/znver1.md: Likewise.
18167 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18168 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18169 * double-int.h (struct double_int): Likewise.
18170 * double-int.c (div_and_round_double): Likewise.
18171 * wide-int.cc: Likewise.
18172 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18173 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18174 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18175 (crossjumps_occurred): ... this.
18176 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18179 PR tree-optimization/78899
18180 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18181 returning bool return struct loop *, NULL for failure and the new
18183 (versionable_outer_loop_p): Don't version outer loop if it has
18184 dont_vectorized bit set.
18185 (tree_if_conversion): When versioning outer loop, ensure
18186 tree_if_conversion is performed also on the inner loop of the
18187 non-vectorizable outer loop copy.
18188 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18189 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18190 prevent vectorization of it.
18191 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18192 the outer loop vectorization of the non-scalar version is attempted
18193 before vectorization of the inner loop in scalar version. If
18194 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18195 vectorization of its inner loop.
18196 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18197 has 2 inner loops, rename also on edges from bb whose single pred
18198 is outer_loop->header. Fix typo in function comment.
18200 2017-01-09 Martin Sebor <msebor@redhat.com>
18203 * asan.c (asan_emit_stack_protection): Increase local buffer size
18204 to avoid snprintf truncation warning.
18206 2017-01-09 Andrew Pinski <apinski@cavium.com>
18208 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18209 to reference thunderx2t99 for the tuning structure
18210 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18212 (thunderx2t99_extra_costs): This.
18213 * config/aarch64/aarch64-tune.md: Regenerate.
18214 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18215 (vulcan_addrcost_table): This.
18216 (vulcan_regmove_cost): Rename to ...
18217 (thunderx2t99_regmove_cost): This.
18218 (vulcan_vector_cost): Rename to ...
18219 (thunderx2t99_vector_cost): this.
18220 (vulcan_branch_cost): Rename to ...
18221 (thunderx2t99_branch_cost): This.
18222 (vulcan_tunings): Rename to ...
18223 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18224 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18226 2017-01-09 Martin Jambor <mjambor@suse.cz>
18230 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18231 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18232 (propagate_vr_accross_jump_function): Use the above function for all
18233 value range computations for pass-through jump functions and type
18234 converasion from explicit value range values.
18235 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18236 parameters from TYPE_ARG_TYPES.
18237 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18238 (ipa_write_node_info): Stream type of the actual argument.
18239 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18241 2017-01-09 Martin Liska <mliska@suse.cz>
18244 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18245 (lookup_compiler): Do not show error message with have_E.
18247 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18249 PR tree-optimization/78938
18250 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18251 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18252 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18253 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18256 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18258 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18261 2017-01-09 Richard Biener <rguenther@suse.de>
18263 PR tree-optimization/78997
18264 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18265 name condition properly.
18267 2017-01-09 Richard Biener <rguenther@suse.de>
18270 * dwarf2out.c (is_cxx): New overload with context.
18271 (is_naming_typedef_decl): Use it.
18273 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18275 * invoke.texi (Option Summary): Correct spacing in option lists
18276 and add line breaks to fix over-long lines.
18278 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18280 PR middle-end/17660
18282 * extend.texi (Common Variable Attributes): Add xref to GCC
18283 Internals manual to explain mode attribute keywords.
18285 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18288 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18289 and Preprocessor Options.
18290 (Options for Linking): Document -pthread here....
18291 (RS/6000 and PowerPC Options): ...not here.
18292 (Solaris 2 Options): ...or here.
18293 * doc/cppopts.texi: Document -pthread.
18295 2017-01-08 Martin Sebor <msebor@redhat.com>
18297 PR middle-end/77708
18298 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18299 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18300 New member functions.
18301 (format_directive): Used them.
18303 (pass_sprintf_length::handle_gimple_call): Same.
18304 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18305 to avoid truncation for any argument.
18306 (extract_affine_mul): Same.
18307 * tree.c (get_file_function_name): Same.
18309 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18311 PR middle-end/77484
18312 * predict.def (PRED_INDIR_CALL): Set to 86.
18314 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18316 PR preprocessor/54124
18317 * doc/cppopts.texi: Reformat -d subtable to list the full name
18318 of the options. Add cross-reference to the docs for the general
18319 compiler -d options.
18320 * doc/invoke.texi (Developer Options): Add cross-reference to the
18321 preprocessor-specific -d option documentation.
18323 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18325 PR preprocessor/13498
18326 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18327 redudant material, and reflect new command-line options.
18328 (System Headers): Likewise.
18330 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18332 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18333 -isystem, and -idirafter. Copy-edit.
18334 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18335 default for -ftrack-macro-expansion. Delete obsolete and
18336 badly-formatted implementation details about -fdebug-cpp output.
18337 * doc/cppwarnopts.texi: Copy-edit.
18339 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18342 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18343 that the transition from a max line width >= 1<<10 to narrower
18344 lines works correctly.
18346 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18348 * doc/options.texi (PerFunction): New.
18349 * opt-functions.awk (switch_flags): Map both Optimization and
18350 PerFunction to CL_OPTIMIZATION.
18351 * opth-gen.awk: Test for PerFunction flag along with
18353 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18354 it only when the latter is present. Skip those that don't in
18355 the hash function generator.
18356 * common.opt (fvar-tracking): Mark as PerFunction instead of
18358 (fvar-tracking-assignments): Likewise.
18359 (fvar-tracking-assignments-toggle): Likewise.
18360 (fvar-tracking-uninit): Likewise.
18362 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18364 PR translation/79018
18365 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18368 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18371 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18372 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18373 TARGET_LONG_JUMP_TABLE_OFFSETS.
18374 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18375 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18376 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18377 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18378 * config/m68k/m68k.md (tablejump expander): Likewise.
18379 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18380 TARGET_LONG_JUMP_TABLE_OFFSETS.
18381 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18382 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18384 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18385 David Holsgrove <david.holsgrove@xilinx.com>
18387 * common/config/microblaze/microblaze-common.c
18388 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18389 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18391 * config/microblaze/microblaze.c (microblaze_must_save_register)
18392 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18394 (microblaze_eh_return): New function.
18395 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18396 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18397 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18398 * config/microblaze/microblaze.md (eh_return): New pattern.
18400 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18402 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18403 GCC_DIAGNOSTIC_STRINGIFY): Define.
18405 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18407 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18409 * config/arm/arm.md (<mcrr>): New.
18411 * config/arm/arm.c (arm_arch5te): New.
18412 (arm_option_override): Set arm_arch5te.
18413 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18415 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18416 (arm_mcrr_qualifiers): ... this. New.
18417 (MRRC_QUALIFIERS): Define to...
18418 (arm_mrrc_qualifiers): ... this. New.
18419 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18421 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18422 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18423 (MRRCI, mrrc, MRRC): New.
18424 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18425 VUNSPEC_MRRC2): New.
18427 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18429 * config/arm/arm.md (<mcr>): New.
18431 * config/arm/arm.c (arm_coproc_builtin_available): Add
18432 support for mcr, mrc, mcr2 and mrc2.
18433 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18434 (arm_mcr_qualifiers): ... this. New.
18435 (MRC_QUALIFIERS): Define to ...
18436 (arm_mrc_qualifiers): ... this. New.
18437 (MCR_QUALIFIERS): Define to ...
18438 (arm_mcr_qualifiers): ... this. New.
18439 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18441 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18442 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18443 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18444 VUNSPEC_MRC2): New.
18446 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18448 * config/arm/arm.md (*ldc): New.
18452 * config/arm/arm.c (arm_coproc_builtin_available): Add
18453 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18454 (arm_coproc_ldc_stc_legitimate_address): New.
18455 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18456 'qualifier_const_pointer'.
18457 (LDC_QUALIFIERS): Define to...
18458 (arm_ldc_qualifiers): ... this. New.
18459 (STC_QUALIFIERS): Define to...
18460 (arm_stc_qualifiers): ... this. New.
18461 * config/arm/arm-protos.h
18462 (arm_coproc_ldc_stc_legitimate_address): New.
18463 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18464 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18465 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18466 stc2, stcl, stc2l): New.
18467 * config/arm/constraints.md (Uz): New.
18468 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18469 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18470 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18471 VUNSPEC_STC2L): New.
18473 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18475 * config/arm/arm.md (<cdp>): New.
18476 * config/arm/arm.c (neon_const_bounds): Rename this ...
18477 (arm_const_bounds): ... this.
18478 (arm_coproc_builtin_available): New.
18479 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18480 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18481 (CDP_QUALIFIERS): Define to...
18482 (arm_cdp_qualifiers): ... this. New.
18484 (arm_expand_builtin_args): Add case for 6 arguments.
18485 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18486 (arm_const_bounds): ... this.
18487 (arm_coproc_builtin_available): New.
18488 * config/arm/arm_acle.h (__arm_cdp): New.
18490 * config/arm/arm_acle_builtins.def (cdp): New.
18492 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18493 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18494 'arm_const_bounds'.
18495 * config/arm/types.md (coproc): New.
18496 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18497 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18498 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18499 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18501 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18503 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18504 (UBINOP_QUALIFIERS): New.
18506 (acle_builtin_data): New. Change comment.
18507 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18508 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18509 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18510 arm_acle_builtins.def.
18511 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18512 (arm_init_acle_builtins): New.
18513 (CRC32_BUILTIN): Remove.
18514 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18515 crc32cb, crc32ch and crc32cw.
18516 (arm_init_crc32_builtins): Remove.
18517 (arm_init_builtins): Use arm_init_acle_builtins rather
18518 than arm_init_crc32_builtins.
18519 (arm_expand_acle_builtin): New.
18520 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18521 * config/arm/arm_acle_builtins.def: New.
18523 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18525 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18526 (arm_builtin_datum): ... this.
18527 (arm_init_neon_builtin): Rename to ...
18528 (arm_init_builtin): ... this. Add a new parameters PREFIX
18529 and USE_SIG_IN_NAME.
18530 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
18531 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
18532 'arm_builtin_datum'.
18533 (arm_init_vfp_builtins): Likewise.
18534 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
18535 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
18536 (arm_expand_neon_args): Rename to ...
18537 (arm_expand_builtin_args): ... this. Rename builtin_arg
18538 enum values and differentiate between ARG_BUILTIN_MEMORY
18539 and ARG_BUILTIN_NEON_MEMORY.
18540 (arm_expand_neon_builtin_1): Rename to ...
18541 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
18542 values, arm_expand_builtin_args and add bool parameter NEON.
18543 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
18544 (arm_expand_vfp_builtin): Likewise.
18545 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
18547 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18549 PR middle-end/77484
18550 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
18551 * predict.c (tree_estimate_probability_bb): Reverse direction of
18552 polymorphic call predictor.
18554 2017-01-06 David Malcolm <dmalcolm@redhat.com>
18556 * passes.c (execute_one_pass): Split out pass-skipping logic into...
18557 (determine_pass_name_match): ...this new function and...
18558 (should_skip_pass_p): ...this new function.
18560 2017-01-06 Nathan Sidwell <nathan@acm.org>
18562 * ipa-visibility.c (function_and_variable_visibility): Reformat
18563 comments and long lines. Remove extrneous if.
18564 * symtab.c (symtab_node::make_decl_local): Fix code format.
18565 (symtab_node::set_section_for_node): Fix comment typo.
18567 2017-01-06 Martin Liska <mliska@suse.cz>
18570 * lra-constraints.c: Rename invariant to lra_invariant.
18571 * predict.c (set_even_probabilities): Initialize e to NULL.
18573 2017-01-05 Martin Sebor <msebor@redhat.com>
18575 PR tree-optimization/78910
18576 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
18577 (format_integer): Correct off-by-one error in the handling
18578 of precision with negative numbers in signed conversions..
18580 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
18582 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
18584 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18586 PR tree-optimization/71016
18587 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
18588 factor_out_conditional_conversion. Formatting fix.
18589 (factor_out_conditional_conversion): Add cond_stmt argument.
18590 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
18591 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
18594 2017-01-05 David Malcolm <dmalcolm@redhat.com>
18596 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
18597 read-rtl-function.o, and selftest-rtl.o.
18598 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
18599 (selftest::aarch64_test_loading_full_dump): New function.
18600 (selftest::aarch64_run_selftests): New function.
18601 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
18602 selftest::aarch64_run_selftests.
18603 * config/i386/i386.c
18604 (selftest::ix86_test_loading_dump_fragment_1): New function.
18605 (selftest::ix86_test_loading_call_insn): New function.
18606 (selftest::ix86_test_loading_full_dump): New function.
18607 (selftest::ix86_test_loading_unspec): New function.
18608 (selftest::ix86_run_selftests): Call the new functions.
18609 * emit-rtl.c (maybe_set_max_label_num): New function.
18610 * emit-rtl.h (maybe_set_max_label_num): New decl.
18611 * function.c (instantiate_decls): Guard call to
18612 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
18613 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
18615 * gensupport.c (gen_reader::gen_reader): Pass "false"
18616 for new "compact" param of rtx_reader.
18617 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
18618 rather than an empty string for NULL strings.
18619 * read-md.c: Potentially include config.h rather than bconfig.h.
18620 Wrap include of errors.h with #ifdef GENERATOR_FILE.
18621 (have_error): New global, copied from errors.c.
18622 (md_reader::read_name): Rename to...
18623 (md_reader::read_name_1): ...this, adding "out_loc" param,
18624 and converting "missing name or number" to returning false, rather
18626 (md_reader::read_name): Reimplement in terms of read_name_1.
18627 (md_reader::read_name_or_nil): New function.
18628 (md_reader::read_string): Handle "(nil)" by returning NULL.
18629 (md_reader::md_reader): Add new param "compact".
18630 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
18631 (md_reader::read_file): New method.
18632 * read-md.h (md_reader::md_reader): Add new param "compact".
18633 (md_reader::read_file): New method.
18634 (md_reader::is_compact): New accessor.
18635 (md_reader::read_name): Convert return type from void to file_location.
18636 (md_reader::read_name_or_nil): New decl.
18637 (md_reader::read_name_1): New decl.
18638 (md_reader::m_compact): New field.
18639 (noop_reader::noop_reader): Pass "false" for new "compact" param
18641 (rtx_reader::rtx_reader): Add new "compact" param.
18642 (rtx_reader::read_rtx_operand): Make virtual and convert return
18643 type from void to rtx.
18644 (rtx_reader::read_until): New decl.
18645 (rtx_reader::handle_any_trailing_information): New virtual function.
18646 (rtx_reader::postprocess): New virtual function.
18647 (rtx_reader::finalize_string): New virtual function.
18648 (rtx_reader::m_in_call_function_usage): New field.
18649 (rtx_reader::m_reuse_rtx_by_id): New field.
18650 * read-rtl-function.c: New file.
18651 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
18652 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
18653 (selftest::verify_three_block_rtl_cfg): New decl.
18654 * read-rtl-function.h: New file.
18655 * read-rtl.c: Potentially include config.h rather than bconfig.h.
18656 For host, include function.h, memmodel.h, and emit-rtl.h.
18657 (one_time_initialization): New function.
18658 (struct compact_insn_name): New struct.
18659 (compact_insn_names): New array.
18660 (find_code): Handle insn codes in compact dumps.
18661 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
18662 (bind_subst_iter_and_attr): Likewise.
18663 (add_condition_to_string): Likewise.
18664 (add_condition_to_rtx): Likewise.
18665 (apply_attribute_uses): Likewise.
18666 (add_current_iterators): Likewise.
18667 (apply_iterators): Likewise.
18668 (initialize_iterators): Guard usage of apply_subst_iterator with
18669 #ifdef GENERATOR_FILE.
18670 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
18671 (md_reader::read_mapping): Likewise.
18672 (add_define_attr_for_define_subst): Likewise.
18673 (add_define_subst_attr): Likewise.
18674 (read_subst_mapping): Likewise.
18675 (check_code_iterator): Likewise.
18676 (rtx_reader::read_rtx): Likewise. Move one-time initialization
18678 (one_time_initialization): New function.
18679 (rtx_reader::read_until): New method.
18680 (read_flags): New function.
18681 (parse_reg_note_name): New function.
18682 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
18683 Handle reuse_rtx ids.
18684 Wrap iterator lookup within #ifdef GENERATOR_FILE.
18685 Add parsing support for RTL dumps, mirroring the special-cases in
18686 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
18687 values, and calling handle_any_trailing_information.
18688 (rtx_reader::read_rtx_operand): Convert return type from void
18689 to rtx, returning return_rtx. Handle case 'e'. Call
18690 finalize_string on XSTR and XTMPL fields.
18691 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
18692 "(nil)" values were omitted. Call the postprocess vfunc on the
18694 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
18695 class ctor. Initialize m_in_call_function_usage. Call
18696 one_time_initialization.
18697 * rtl-tests.c (selftest::test_uncond_jump): Call
18698 set_new_first_and_last_insn.
18699 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
18700 * selftest-rtl.c: New file.
18701 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
18702 (selftest::get_insn_by_uid): New decl.
18703 * selftest-run-tests.c (selftest::run_tests): Call
18704 read_rtl_function_c_tests.
18705 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
18706 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
18709 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
18711 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
18712 operands in a special way. Assert that pos+len <= mode precision.
18714 2017-01-05 Jakub Jelinek <jakub@redhat.com>
18716 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
18717 3 argument Alias with unlimited for the negative form.
18718 (fno-vect-cost-model): Removed.
18720 2017-01-05 Martin Liska <mliska@suse.cz>
18722 * hsa-gen.c (gen_hsa_divmod): New function.
18723 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
18725 2017-01-05 Martin Liska <mliska@suse.cz>
18728 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
18731 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18733 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
18734 small constant length operands.
18736 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18738 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
18739 between loop iterations.
18741 2017-01-05 Martin Liska <mliska@suse.cz>
18744 * gimplify.c (gimplify_decl_expr): Compare to
18745 asan_poisoned_variables instread of checking flags.
18746 (gimplify_target_expr): Likewise.
18747 (gimplify_expr): Likewise.
18748 (gimplify_function_tree): Conditionally initialize
18749 asan_poisoned_variables.
18751 2017-01-04 Jeff Law <law@redhat.com>
18753 PR tree-optimizatin/78812
18754 * rtl.h (contains_mem_rtx_p): Prototype.
18755 * ifcvt.c (containts_mem_rtx_p): Move from here to...
18756 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
18757 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
18758 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
18759 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
18761 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18763 * input.c (assert_char_at_range): Default-initialize actual_range.
18765 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18767 * df-scan.c (df_ref_create_structure): Make regno unsigned,
18768 to match the caller.
18770 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18772 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
18773 insns after final jump in test to emit dummy move.
18775 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18777 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
18778 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
18780 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
18782 * multiple_target.c (create_dispatcher_calls): Init e_next.
18783 * tree-ssa-loop-split.c (split_loop): Init border.
18784 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
18787 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
18792 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
18793 (altivec_register_operand): Do not return true if the operand
18794 contains a SUBREG mixing SImode and SFmode.
18795 (vsx_register_operand): Likewise.
18796 (vsx_reg_sfsubreg_ok): New predicate.
18797 (vfloat_operand): Do not return true if the operand contains a
18798 SUBREG mixing SImode and SFmode.
18799 (vint_operand): Likewise.
18800 (vlogical_operand): Likewise.
18801 (gpc_reg_operand): Likewise.
18802 (int_reg_operand): Likewise.
18803 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
18804 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
18805 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
18807 (rs6000_emit_move_si_sf_subreg): New helper function.
18808 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
18809 fixup SUBREGs involving SImode and SFmode.
18810 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
18811 numbers for the new peephole2 optimization.
18812 (peephole2 for SFmode unions): New peephole2 to optimize cases in
18813 the GLIBC math library that do AND/IOR/XOR operations on single
18814 precision floating point.
18815 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
18816 target macros to say whether we need to avoid SUBREGs mixing
18818 (TARGET_ALLOW_SF_SUBREG): Likewise.
18819 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
18820 (UNSPEC_SI_FROM_SF): Likewise.
18821 (iorxor): Change spacing.
18822 (and_ior_xor): New iterator for AND, IOR, and XOR.
18823 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
18824 (movdi_from_sf_zero_ext): Likewise.
18825 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
18826 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
18827 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
18828 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
18829 (fms<mode>4): Likewise.
18830 (fnma<mode>4): Likewise.
18831 (fnms<mode>4): Likewise.
18832 (nfma<mode>4): Likewise.
18833 (nfms<mode>4): Likewise.
18835 2017-01-04 Marek Polacek <polacek@redhat.com>
18838 * doc/invoke.texi: Document -Wpointer-compare.
18840 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18842 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
18845 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
18846 descriptions for -gdwarf-5 and emit them as uleb128 instead of
18849 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
18852 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
18853 documentation of the powerpc_popcntb_ok attribute.
18854 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18855 code to issue warning messages if a requested CPU configuration is
18856 not supported by the binary (assembler and loader) toolchain.
18857 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
18858 made to define a built-in function that has been disabled.
18859 (paired_init_builtins): Add assertion to prevent ICE if attempt is
18860 made to define a built-in function that has been disabled.
18861 (altivec_init_builtins): Add comment explaining why definition
18862 of the DST built-in functions is not preceded by an assertion
18863 check. Add assertions to prevent ICE if attempts are made to
18864 define an altivec predicate or an abs* built-in function that has
18866 (htm_init_builtins): Add comment explaining why definition of the
18867 htm built-in functions is not preceded by an assertion check.
18869 2017-01-04 Jeff Law <law@redhat.com>
18871 PR tree-optimizatin/67955
18872 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
18873 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
18874 the points-to solution does not include pt_null. Use DECL_PT_UID
18877 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
18879 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
18880 Use gen_int_mode instead of gen_lopwart for const_int operands.
18882 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18884 PR tree-optimization/71563
18885 * match.pd: Simplify X << Y into X if Y is known to be 0 or
18886 out of range value - has low bits known to be zero.
18888 2017-01-04 Alan Modra <amodra@gmail.com>
18890 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
18891 * configure: Regenerate.
18892 * config.in: Regenerate.
18894 2017-01-04 Jakub Jelinek <jakub@redhat.com>
18897 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
18898 a substring of the message, but strcmp with the whole message. Ifdef
18899 ENABLE_NLS, translate the message first using dgettext.
18901 2017-01-03 Jeff Law <law@redhat.com>
18903 PR tree-optimizatin/78856
18904 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
18905 (mark_threaded_blocks): Remove code to truncate thread paths that
18906 cross multiple loop headers. Instead invalidate the cached loop
18907 iteration information and handle case of a thread path walking
18908 into an irreducible region.
18910 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
18913 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
18914 assertions. Add support for doing the signbit if the IEEE 128-bit
18915 floating point value is in a GPR.
18916 * config/rs6000/rs6000.md (Fsignbit): Delete.
18917 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
18918 Update the length attribute if the value is in a GPR.
18919 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
18920 the sign or zero extension instruction, since the value is always 0/1.
18921 (signbit<mode>2_dm2): Delete using <Fsignbit>.
18924 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
18925 extracting SImode to a GPR register so that we can generate a
18926 store, limit the vector to be in a traditional Altivec register
18927 for the vextuwrx instruction.
18929 2017-01-03 Ian Lance Taylor <iant@google.com>
18931 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
18933 2017-01-03 Martin Sebor <msebor@redhat.com>
18935 PR tree-optimization/78696
18936 * gimple-ssa-sprintf.c (format_floating): Correct handling of
18937 precision. Use MPFR for %f for greater fidelity. Correct handling
18939 (pass_sprintf_length::compute_format_length): Set width and precision
18940 specified by asrerisk to void_node for vararg functions.
18941 (try_substitute_return_value): Adjust dump output.
18943 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
18945 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
18947 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
18949 * doc/invoke.texi (SPARC options): Document -mlra as the default.
18950 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
18951 -mlra/-mno-lra was passed to the compiler.
18953 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
18955 PR rtl-optimization/65618
18956 * emit-rtl.c (try_split): Move initialization of "before" and
18957 "after" to just before the call to emit_insn_after_setloc.
18959 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
18961 * doc/md.texi (Standard Names): Remove reference to Java frontend.
18963 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
18965 * dwarf2out.c (gen_enumeration_type_die): When
18966 -gno-strict-dwarf, add a DW_AT_encoding attribute.
18968 2017-01-03 Jakub Jelinek <jakub@redhat.com>
18970 PR tree-optimization/78965
18971 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
18972 Change first argument from const call_info & to call_info &. For %n
18973 set info.nowrite to false.
18975 PR middle-end/78901
18976 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
18977 possibly throwing calls.
18979 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
18980 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
18981 and fns handling, rather than in a separate case SSA_NAME.
18983 2017-01-02 Jeff Law <law@redhat.com>
18985 * config/darwin-driver.c (darwin_driver_init): Const-correctness
18986 fixes for first_period and second_period variables.
18988 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
18991 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
18992 (*insvqi_1): New insn pattern.
18993 (*insvqi_1_mem_rex64): Ditto.
18994 (*insvqi_2): Ditto.
18995 (*insvqi_3): Rename from *insvqi.
18997 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
18999 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19001 * doc/cfg.texi (Edges): Remove reference to Java.
19002 (Maintaining the CFG): Ditto.
19004 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19006 PR middle-end/77674
19007 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19008 transparent aliases.
19010 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19012 PR middle-end/77484
19013 * predict.def (PRED_CALL): Update hitrate.
19014 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19015 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19016 into direct/indirect/polymorphic variants.
19018 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19020 Update copyright years.
19022 * gcc.c (process_command): Update copyright notice dates.
19023 * gcov-dump.c (print_version): Ditto.
19024 * gcov.c (print_version): Ditto.
19025 * gcov-tool.c (print_version): Ditto.
19026 * gengtype.c (create_file): Ditto.
19027 * doc/cpp.texi: Bump @copying's copyright year.
19028 * doc/cppinternals.texi: Ditto.
19029 * doc/gcc.texi: Ditto.
19030 * doc/gccint.texi: Ditto.
19031 * doc/gcov.texi: Ditto.
19032 * doc/install.texi: Ditto.
19033 * doc/invoke.texi: Ditto.
19035 Copyright (C) 2017 Free Software Foundation, Inc.
19037 Copying and distribution of this file, with or without modification,
19038 are permitted in any medium without royalty provided the copyright
19039 notice and this notice are preserved.