1 2014-04-15 Richard Biener <rguenther@suse.de>
3 * common.opt (lto_partition_model): New enum.
4 (flto-partition=): Merge separate options with a single with argument,
5 add -flto-partition=one support.
6 * flag-types.h (enum lto_partition_model): Declare.
7 * opts.c (finish_options): Remove duplicate -flto-partition=
9 * lto-wrapper.c (run_gcc): Adjust.
11 2014-04-15 Richard Biener <rguenther@suse.de>
13 * alias.c (ncr_compar): New function.
14 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
16 2014-04-15 Richard Biener <rguenther@suse.de>
18 * alias.c (record_component_aliases): Do not walk BINFOs.
20 2014-04-15 Richard Biener <rguenther@suse.de>
22 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23 Add struct function argument and adjust.
24 (find_func_aliases_for_call): Likewise.
25 (find_func_aliases): Likewise.
26 (find_func_clobbers): Likewise.
27 (intra_create_variable_infos): Likewise.
28 (compute_points_to_sets): Likewise.
29 (ipa_pta_execute): Adjust. Do not push/pop cfun.
31 2014-04-15 Richard Biener <rguenther@suse.de>
33 * tree.c (iterative_hash_expr): Use enum tree_code_class
34 to store TREE_CODE_CLASS.
35 (tree_block): Likewise.
36 (tree_set_block): Likewise.
37 * tree.h (fold_build_pointer_plus_loc): Use
38 convert_to_ptrofftype_loc.
40 2014-04-15 Jakub Jelinek <jakub@redhat.com>
43 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
46 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
48 * cfgloop.h (struct loop): Move force_vectorize down.
49 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
50 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
51 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
52 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
53 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
54 * tree-core.h (enum annot_expr_kind): Add new kind values.
55 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
56 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
58 * tree.def (ANNOTATE_EXPR): Tweak comment.
60 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
62 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
65 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
67 * tree.h (TYPE_IDENTIFIER): Declare.
68 * tree.c (subrange_type_for_debug_p): Use it.
69 * godump.c (go_format_type): Likewise.
70 * dwarf2out.c (is_cxx_auto, modified_type_die,
71 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
72 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
74 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
77 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
79 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
81 * config/i386/i386.c (examine_argument): Return bool. Return true if
82 parameter should be passed in memory.
83 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
84 (construct_container): Update calls to examine_argument.
85 (function_arg_advance_64): Ditto.
86 (return_in_memory_32): Merge with ix86_return_in_memory.
87 (return_in_memory_64): Ditto.
88 (return_in_memory_ms_64): Ditto.
90 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
92 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
93 * coverage.c (coverage_compute_profile_id): Handle externally visible
96 2014-04-14 Martin Jambor <mjambor@suse.cz>
98 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
99 DECL_DISREGARD_INLINE_LIMITS functions.
101 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
104 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
106 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
109 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
110 optimize_insn_for_speed_p instead of
111 optimize_function_for_speed_p.
113 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
115 * doc/invoke.texi (free): Document AArch64.
117 2014-04-14 Richard Biener <rguenther@suse.de>
119 PR tree-optimization/60042
120 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
121 (insert_into_preds_of_block): Do not prevent PHI insertion
122 for REFERENCE exprs here ...
123 (eliminate_dom_walker::before_dom_children): ... but prevent
124 their use here under similar conditions when applied to the
125 IL after PRE optimizations.
127 2014-04-14 Richard Biener <rguenther@suse.de>
129 * passes.def: Move early points-to after early SRA.
131 2014-04-14 Richard Biener <rguenther@suse.de>
133 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
134 check for which sign-changes we allow when forwarding
135 a converted value into a switch.
137 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
139 * stor-layout.c (place_field): Finalize non-constant offset for the
142 2014-04-14 Richard Biener <rguenther@suse.de>
144 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
146 (expand_switch_using_bit_tests_p): Likewise.
147 (process_switch): Compute and pass on speed_p based on the
149 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
150 optimize_bb_for_speed_p.
152 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
154 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
155 * function.h (struct function): Rename has_force_vect_loops into
156 has_force_vectorize_loops.
157 * lto-streamer-in.c (input_cfg): Adjust for renaming.
158 (input_struct_function_base): Likewise.
159 * lto-streamer-out.c (output_cfg): Likewise.
160 (output_struct_function_base): Likewise.
161 * omp-low.c (expand_omp_simd): Likewise.
162 * tree-cfg.c (move_sese_region_to_fn): Likewise.
163 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
164 (version_loop_for_if_conversion): Likewise.
165 (tree_if_conversion): Likewise.
166 (main_tree_if_conversion): Likewise.
167 (gate_tree_if_conversion): Likewise.
168 * tree-inline.c (copy_loops): Likewise.
169 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
170 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
171 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
172 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
173 * tree-vectorizer.c (vectorize_loops): Likewise.
174 * tree-vectorizer.h (unlimited_cost_model): Likewise.
176 2014-04-14 Richard Biener <rguenther@suse.de>
179 * lto-streamer-out.c (wrap_refs): New function.
180 (lto_output): Wrap symbol references in global initializes in
181 type-preserving MEM_REFs.
183 2014-04-14 Christian Bruel <christian.bruel@st.com>
185 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
187 2014-04-14 Christian Bruel <christian.bruel@st.com>
189 * config/sh/sh.md (setmemqi): New expand pattern.
190 * config/sh/sh.h (CLEAR_RATIO): Define.
191 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
192 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
194 2014-04-14 Richard Biener <rguenther@suse.de>
197 * fold-const.c (negate_expr_p): Don't negate directional rounding
199 (fold_negate_expr): Likewise.
201 2014-04-14 Richard Biener <rguenther@suse.de>
203 PR tree-optimization/59817
204 PR tree-optimization/60453
205 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
206 recursion to catch all CHRECs in the scalar evolution and restrict
207 the predicate for the remains appropriately.
209 2014-04-12 Catherine Moore <clm@codesourcery.com>
211 * config/mips/constraints.md: Add new register constraint "kb".
212 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
213 (*movhi_internal): Likewise.
214 (*movqi_internal): Likewise.
215 * config/mips/mips.h (M16_STORE_REGS): New register class.
216 (REG_CLASS_NAMES): Add M16_STORE_REGS.
217 (REG_CLASS_CONTENTS): Likewise.
218 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
220 2014-04-11 Tobias Burnus <burnus@net-b.de>
223 * doc/invoke.texi (-Wformat-signedness): Document it.
224 (Wformat=2): Mention that this enables -Wformat-signedness.
226 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
228 * common/config/epiphany/epiphany-common.c
229 (epiphany_option_optimization_table): Enable section anchors by
230 default at -O1 or higher.
231 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
232 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
233 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
234 carries no extra cost.
235 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
236 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
237 * config/epiphany/predicates.md (memclob_operand): New predicate.
238 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
239 Use memclob_operand predicate and X constraint for operand 3.
241 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
243 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
244 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
247 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
249 PR rtl-optimization/60651
250 * mode-switching.c (optimize_mode_switching): Make sure to emit
251 sets of a lower numbered entity before sets of a higher numbered
252 entity to a mode of the same or lower priority.
253 When creating a seginfo for a basic block that starts with a code
254 label, move the insertion point past the code label.
255 (new_seginfo): Document and enforce requirement that
256 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
257 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
258 * doc/tm.texi: Regenerate.
260 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
263 * config/arc/arc.c (arc_save_restore): Fix assert typo.
265 2013-04-11 Jakub Jelinek <jakub@redhat.com>
267 * BASE-VER: Set to 4.10.0.
269 2014-04-11 Tobias Burnus <burnus@net-b.de>
272 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
273 * doc/gcc.texi (Service): Update description in the @menu
274 * doc/invoke.texi (Option Summary): Remove misplaced and
277 2014-04-11 Steve Ellcey <sellcey@mips.com>
278 Jakub Jelinek <jakub@redhat.com>
281 * expr.c (convert_move): Use emit_store_flag_force instead of
282 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
285 2014-04-11 Richard Biener <rguenther@suse.de>
288 * varasm.c (assemble_alias): Avoid endless error reporting
289 recursion by setting TREE_ASM_WRITTEN.
291 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
293 * config/s390/s390.md: Add a splitter for NOT rtx.
295 2014-04-11 Jakub Jelinek <jakub@redhat.com>
297 PR rtl-optimization/60663
298 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
300 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
301 Jakub Jelinek <jakub@redhat.com>
304 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
305 flag from decl_node to node.
307 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
310 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
311 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
312 ameliorating the cases where it can be.
314 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
317 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
319 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
320 (loadsync_<mode>): Change mode.
321 (load_quadpti, store_quadpti): New.
322 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
323 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
324 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
326 2014-04-09 Cong Hou <congh@google.com>
329 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
332 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
334 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
335 instead of vnor to exploit possible fusion opportunity in the
337 (altivec_expand_vec_perm_const_le): Likewise.
339 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
341 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
342 (loadsync_<mode>): Change mode.
343 (load_quadpti, store_quadpti): New.
344 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
345 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
347 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
350 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
351 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
352 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
354 2014-04-08 Richard Biener <rguenther@suse.de>
357 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
358 a 64bit widest int print double-int similar to on HWI64 hosts.
360 2014-04-08 Richard Biener <rguenther@suse.de>
362 PR tree-optimization/60785
363 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
364 default defs properly.
366 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
368 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
371 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
373 * ipa-devirt.c (maybe_record_node): When node is not recorded,
374 set completep to false rather than true.
376 2014-04-07 Douglas B Rupp <rupp@adacore.com>
379 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
380 ARM_TARGET2_DWARF_FORMAT.
382 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
385 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
386 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
389 2014-04-07 Richard Biener <rguenther@suse.de>
391 PR tree-optimization/60766
392 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
393 (may_eliminate_iv): Convert cand_value_at result to desired type.
395 2014-04-07 Jason Merrill <jason@redhat.com>
398 * common.opt (-fno-gnu-unique): Add.
399 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
401 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
403 * haifa-sched.c: Fix outdated function reference and minor
404 grammar errors in introductory comment.
406 2014-04-07 Richard Biener <rguenther@suse.de>
409 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
411 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
413 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
416 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
417 size accounting for thunks.
418 (pa_asm_output_mi_thunk): Use final_start_function() and
419 final_end_function() to output function start and end directives.
421 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
423 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
424 device specific ISA/ feature information. Remove short_sp and
425 errata_skip ds. Add avr_device_specific_features enum to have device
427 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
428 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
429 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
430 updated device specific info.
431 * config/avr/avr-mcus.def: Merge device specific details to
433 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
435 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
436 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
437 assembler if RMW isa supported by current device.
438 * config/avr/genmultilib.awk: Update as device info structure changed.
439 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
441 2014-04-04 Cong Hou <congh@google.com>
443 PR tree-optimization/60656
444 * tree-vect-stmts.c (supportable_widening_operation):
445 Fix a bug that elements in a vector with vect_used_by_reduction
446 property are incorrectly reordered when the operation on it is not
447 consistant with the one in reduction operation.
449 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
451 PR rtl-optimization/60155
452 * gcse.c (record_set_data): New function.
453 (single_set_gcse): New function.
454 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
455 (hoist_code): Likewise.
456 (get_pressure_class_and_nregs): Likewise.
458 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
460 * explow.c (probe_stack_range): Emit a final optimization blockage.
462 2014-04-04 Anthony Green <green@moxielogic.com>
464 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
467 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
470 * lto-cgraph.c (input_overwrite_node): Check that partitioning
471 flags are set only during streaming.
472 * ipa.c (process_references, walk_polymorphic_call_targets,
473 symtab_remove_unreachable_nodes): Drop bodies of always inline
474 after early inlining.
475 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
477 2014-04-04 Jakub Jelinek <jakub@redhat.com>
478 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
481 * dwarf2out.c (const_ok_for_output_1): Reject expressions
484 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
487 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
489 (cortex_a53_fdivd): Likewise.
491 2014-04-04 Martin Jambor <mjambor@suse.cz>
494 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
496 * cgraph.c (clone_of_p): Also return true if thunks match.
497 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
498 cgraph_function_or_thunk_node and an obsolete comment.
499 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
501 (build_function_decl_skip_args): Likewise.
502 (set_new_clone_decl_and_node_flags): New function.
503 (duplicate_thunk_for_node): Likewise.
504 (redirect_edge_duplicating_thunks): Likewise.
505 (cgraph_clone_node): New parameter args_to_skip, pass it to
506 redirect_edge_duplicating_thunks which is called instead of
507 cgraph_redirect_edge_callee.
508 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
509 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
511 2014-04-04 Jeff Law <law@redhat.com>
514 * config/arm/predicates.md (const_int_I_operand): New predicate.
515 (const_int_M_operand): Similarly.
516 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
518 (insv_t2, extv_reg, extzv_t2): Likewise.
519 (load_multiple_with_writeback): Similarly for const_int_I_operand.
520 (pop_multiple_with_writeback_and_return): Likewise.
521 (vfp_pop_multiple_with_writeback): Likewise
523 2014-04-04 Richard Biener <rguenther@suse.de>
526 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
527 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
529 * gimplify.h (gimple_add_tmp_var_fn): Declare.
530 * gimplify.c (gimple_add_tmp_var_fn): New function.
531 * gimple-expr.h (create_tmp_reg_fn): Declare.
532 * gimple-expr.c (create_tmp_reg_fn): New function.
533 * gimple-low.c (record_vars_into): Don't change cfun.
534 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
535 code generation without cfun.
537 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
540 * Makefile.in (install-driver): Fix shell scripting.
542 2014-04-03 Cong Hou <congh@google.com>
544 PR tree-optimization/60505
545 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
546 threshold of number of iterations below which no vectorization
548 * tree-vect-loop.c (new_loop_vec_info):
549 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
550 * tree-vect-loop.c (vect_analyze_loop_operations):
551 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
552 * tree-vect-loop.c (vect_transform_loop):
553 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
554 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
555 of iterations of the loop and see if we should build the epilogue.
557 2014-04-03 Richard Biener <rguenther@suse.de>
559 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
560 (streamer_tree_cache_create): Adjust.
561 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
562 to allow optional nodes array.
563 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
564 (streamer_tree_cache_append): Likewise.
565 (streamer_tree_cache_create): Create nodes array optionally
566 as specified by parameter.
567 * lto-streamer-out.c (create_output_block): Avoid maintaining
568 the node array in the writer cache.
569 (DFS_write_tree): Remove assertion.
570 (produce_asm_for_decls): Free the out decl state hash table early.
571 * lto-streamer-in.c (lto_data_in_create): Adjust for
572 streamer_tree_cache_create prototype change.
574 2014-04-03 Richard Biener <rguenther@suse.de>
576 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
577 set TREE_CHAIN to NULL_TREE.
579 2014-04-03 Richard Biener <rguenther@suse.de>
581 PR tree-optimization/60740
582 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
583 over all GIMPLE_COND operands.
585 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
587 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
588 (Weffc++): Remove Scott's numbering, merge lists and reference
594 * class.c (accessible_nvdtor_p): New.
595 (check_bases): Don't check base destructor here ...
596 (check_bases_and_members): ... check them here. Trigger on
597 Wnon-virtual-dtor flag.
598 (finish_struct_1): Use accessible_nvdtor_p.
601 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
602 * g++.dg/warn/Wnvdtor-2.C: New.
603 * g++.dg/warn/Wnvdtor-3.C: New.
604 * g++.dg/warn/Wnvdtor-4.C: New.
605 * g++.dg/warn/Weff1.C: Delete.
606 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
607 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
609 2014-04-03 Nick Clifton <nickc@redhat.com>
611 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
614 2014-04-03 Martin Jambor <mjambor@suse.cz>
616 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
617 mention gcc_unreachable before failing.
618 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
621 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
624 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
625 inconsistent code and instead mark the context inconsistent.
626 (possible_polymorphic_call_targets): For inconsistent contexts
627 return empty complete list.
629 2014-04-02 Anthony Green <green@moxielogic.com>
631 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
632 (extendqisi2, extendhisi2): Define.
633 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
634 (WCHAR_TYPE): Change to unsigned int.
636 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
638 PR tree-optimization/60733
639 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
640 insertion point for PHI candidates to be the end of the feeding
641 block for the PHI argument.
643 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
645 PR rtl-optimization/60650
646 * lra-constraints.c (process_alt_operands): Decrease reject for
647 earlyclobber matching.
649 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
651 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
653 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
655 * config/spu/spu.c (pad_bb): Do not crash when the last
656 insn is CODE_FOR_blockage.
658 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
660 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
661 lies outside the target mode.
663 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
666 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
667 software floating point or no floating point registers, do not
668 allow any type in the FPRs. Eliminate a test for SPE SIMD types
669 in GPRs that occurs after we tested for GPRs that would never be
672 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
673 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
674 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
675 specifically allow DDmode, since that does not use the SPE SIMD
678 2014-04-02 Richard Biener <rguenther@suse.de>
681 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
682 MODE_INTs. Properly use negv_optab.
683 (expand_abs): Likewise.
685 2014-04-02 Richard Biener <rguenther@suse.de>
688 * Makefile.in (install-driver): Guard extra installs with special
691 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
693 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
696 2014-04-01 Richard Henderson <rth@redhat.com>
699 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
700 alternative enabled before register allocation.
702 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
704 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
705 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
707 (nios2_large_got_address): Remove unneeded 'sym' parameter.
708 (nios2_got_address): Update nios2_large_got_address call site.
709 (nios2_delegitimize_address): New function.
710 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
711 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
712 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
714 2014-04-01 Martin Husemann <martin@duskware.de>
716 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
719 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
721 PR rtl-optimization/60604
722 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
723 check from register_operand.
724 (register_operand): Redefine in terms of general_operand.
725 (nonmemory_operand): Use register_operand for the non-constant cases.
727 2014-04-01 Richard Biener <rguenther@suse.de>
729 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
731 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
733 * doc/invoke.texi (mapp-regs): Clarify.
735 2014-03-31 Ulrich Drepper <drepper@gmail.com>
737 * config/i386/avx512fintrin.h (__v32hi): Define type.
739 (_mm512_set1_epi8): Define.
740 (_mm512_set1_epi16): Define.
741 (_mm512_set4_epi32): Define.
742 (_mm512_set4_epi64): Define.
743 (_mm512_set4_pd): Define.
744 (_mm512_set4_ps): Define.
745 (_mm512_setr4_epi64): Define.
746 (_mm512_setr4_epi32): Define.
747 (_mm512_setr4_pd): Define.
748 (_mm512_setr4_ps): Define.
749 (_mm512_setzero_epi32): Define.
751 2014-03-31 Martin Jambor <mjambor@suse.cz>
754 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
755 callsite_arguments_match_p. Updated all callers. Also check types of
756 corresponding formal parameters and actual arguments.
757 (not_all_callers_have_enough_arguments_p) Renamed to
758 some_callers_have_mismatched_arguments_p.
760 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
762 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
764 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
767 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
770 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
772 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
773 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
775 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
776 Use FMAMODE_NOVF512 mode iterator.
777 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
778 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
779 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
780 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
782 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
783 Use VF_128_256 mode iterator.
784 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
787 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
789 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
790 static chain if needed.
792 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
795 * lra-constraints.c (index_part_to_reg): New.
796 (process_address): Use it.
798 2014-03-27 Jeff Law <law@redhat.com>
799 Jakub Jelinek <jakub@redhat.com>
802 * expr.c (do_tablejump): Use simplify_gen_binary rather than
803 gen_rtx_{PLUS,MULT} to build up the address expression.
805 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
806 creating non-canonical RTL.
808 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
811 * ipa-inline.c (want_inline_small_function_p): Short circuit large
812 functions; reorganize to make cheap checks first.
813 (inline_small_functions): Do not estimate growth when dumping;
815 * ipa-inline.h (inline_summary): Add min_size.
816 (growth_likely_positive): New function.
817 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
818 (set_cond_stmt_execution_predicate): Cleanup.
819 (estimate_edge_size_and_time): Compute min_size.
820 (estimate_calls_size_and_time): Likewise.
821 (estimate_node_size_and_time): Likewise.
822 (inline_update_overall_summary): Update min_size.
823 (do_estimate_edge_time): Likewise.
824 (do_estimate_edge_size): Update.
825 (do_estimate_edge_hints): Update.
826 (growth_likely_positive): New function.
828 2014-03-28 Jakub Jelinek <jakub@redhat.com>
831 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
832 also if addr has VOIDmode.
834 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
836 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
837 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
839 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
840 instructions as well as AdvancedSIMD loads.
842 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
844 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
845 Use crypto_aese type.
846 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
847 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
848 crypto_aese, crypto_aesmc. Move to types.md.
849 * config/arm/types.md (crypto_aes): Split into crypto_aese,
851 * config/arm/iterators.md (crypto_type): Likewise.
853 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
855 * cgraph.c: Include expr.h and tree-dfa.h.
856 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
859 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
862 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
863 regs from checking multi-reg pseudos.
865 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
867 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
869 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
871 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
872 if it would clobber the stack pointer, even temporarily.
874 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
876 * mode-switching.c: Make small adjustments to the top comment.
878 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
880 * config/rs6000/constraints.md (wD constraint): New constraint to
881 match the constant integer to get the top DImode/DFmode out of a
882 vector in a VSX register.
884 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
885 match the constant integer to get the top DImode/DFmode out of a
886 vector in a VSX register.
888 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
891 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
894 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
895 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
897 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
898 Optimize vec_extract of 64-bit values, where the value being
899 extracted is in the top word, where we can use scalar
900 instructions. Add direct move and store support. Combine the big
901 endian/little endian vector select load support into a single insn.
902 (vsx_extract_<mode>_internal1): Likewise.
903 (vsx_extract_<mode>_internal2): Likewise.
904 (vsx_extract_<mode>_load): Likewise.
905 (vsx_extract_<mode>_store): Likewise.
906 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
907 combined into vsx_extract_<mode>_load.
908 (vsx_extract_<mode>_one_le): Likewise.
910 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
911 define the top 64-bit vector element.
913 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
916 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
917 Document vec_vbpermq builtin.
920 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
921 enable use of xxsldwi and xxpermdi builtin functions.
922 (vec_xxpermdi): Likewise.
924 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
925 Document use of vec_xxsldwi and vec_xxpermdi builtins.
927 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
929 PR rtl-optimization/60650
930 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
932 (find_spills_for): New.
933 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
934 Spill all pseudos on the second iteration.
936 2014-03-27 Marek Polacek <polacek@redhat.com>
939 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
942 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
944 * config/s390/s390.c (s390_can_use_return_insn): Check for
945 call-saved FPRs on 31 bit.
947 2014-03-27 Jakub Jelinek <jakub@redhat.com>
950 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
951 if they need regimplification, just drop them instead of
952 calling gimple_regimplify_operands on them.
954 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
957 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
958 (aarch64_frame_pointer_required): Adjust logic.
959 (aarch64_can_eliminate): Adjust logic.
960 (aarch64_override_options_after_change): Adjust logic.
962 2014-03-27 Dehao Chen <dehao@google.com>
964 * ipa-inline.c (early_inliner): Update node's inline info.
966 2014-03-26 Dehao Chen <dehao@google.com>
968 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
969 compiler inserted conditional jumps for NAN float check.
971 2014-03-26 Jakub Jelinek <jakub@redhat.com>
973 * ubsan.h (ubsan_create_data): Change second argument's type
974 to const location_t *.
975 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
977 (ubsan_create_data): Change second argument to const location_t *PLOC.
978 Create Loc field whenever PLOC is non-NULL.
979 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
980 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
984 * real.c (real_to_integer2): Change type of low to UHWI.
986 2014-03-26 Tobias Burnus <burnus@net-b.de>
988 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
989 (CILK_SELF_SPECS): New define.
990 (driver_self_specs): Use it.
992 2014-03-26 Richard Biener <rguenther@suse.de>
994 * tree-pretty-print.c (percent_K_format): Implement special
995 case for LTO and its stripped down BLOCK tree.
997 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1000 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
1002 * tree-vrp.c (simplify_internal_call_using_ranges): If only
1003 one range is range_int_cst_p, but not both, at least optimize
1004 addition/subtraction of 0 and multiplication by 0 or 1.
1005 * gimple-fold.c (gimple_fold_call): Fold
1006 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
1007 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
1008 INTEGER_CSTs, try to fold at least x * 0 and y - y.
1010 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
1012 PR rtl-optimization/60452
1013 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
1014 <case REG>: Return 1 for invalid offsets from the frame pointer.
1016 2014-03-26 Marek Polacek <polacek@redhat.com>
1019 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1022 2014-03-26 Marek Polacek <polacek@redhat.com>
1025 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1028 2014-03-26 Marek Polacek <polacek@redhat.com>
1031 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1032 multiplication in unsigned type.
1034 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1036 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
1038 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1040 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
1042 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1045 * cif-code.def (UNREACHABLE) New code.
1046 * ipa-inline.c (inline_small_functions): Skip edges to
1047 __builtlin_unreachable.
1048 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
1049 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
1050 predicate to __bulitin_unreachable.
1051 (set_cond_stmt_execution_predicate): Fix issue when
1052 invert_tree_comparison returns ERROR_MARK.
1053 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
1054 propagate to inline clones.
1055 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
1057 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
1058 * cgraphclones.c (cgraph_clone_node): If call destination is already
1059 ureachable, do not redirect it back.
1060 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
1063 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1065 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
1066 Do not modify inline clones.
1068 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1070 * config/i386/i386.md (general_sext_operand): New mode attr.
1071 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
1072 don't generate (sign_extend (const_int)).
1073 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
1074 operands[2]. Use We constraint instead of <i> and
1075 <general_sext_operand> predicate instead of <general_operand>.
1076 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
1077 * config/i386/constraints.md (We): New constraint.
1078 * config/i386/predicates.md (x86_64_sext_operand,
1079 sext_operand): New predicates.
1081 2014-03-25 Martin Jambor <mjambor@suse.cz>
1084 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1085 inconsistent devirtualizations to __builtin_unreachable.
1087 2014-03-25 Marek Polacek <polacek@redhat.com>
1090 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1092 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1094 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1095 order of elements for big-endian.
1097 2014-03-25 Richard Biener <rguenther@suse.de>
1100 * gimplify-me.c (gimple_regimplify_operands): Update the
1103 2014-03-25 Martin Jambor <mjambor@suse.cz>
1106 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1107 (lto_output_varpool_node): Likewise.
1108 (input_overwrite_node): Likewise.
1109 (input_varpool_node): Likewise.
1111 2014-03-25 Richard Biener <rguenther@suse.de>
1113 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1114 (run_gcc): Likewise.
1116 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1118 * combine.c (simplify_compare_const): Add MODE argument.
1119 Handle mode_width 0 as very large mode_width.
1120 (try_combine, simplify_comparison): Adjust callers.
1122 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1123 type to avoid signed integer overflow.
1124 * explow.c (plus_constant): Likewise.
1126 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1128 * doc/generic.texi: Correct typos.
1130 2014-03-24 Tobias Burnus <burnus@net-b.de>
1132 * doc/invoke.texi (-flto): Expand section about
1133 using static libraries with LTO.
1135 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1137 PR rtl-optimization/60501
1138 * optabs.def (addptr3_optab): New optab.
1139 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1140 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1141 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1143 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1145 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1147 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1149 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1152 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1153 (_mm256_undefined_ps): Define.
1154 (_mm256_undefined_pd): Define.
1155 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1156 (_mm_undefined_pd): Define.
1157 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1158 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1159 (_mm512_undefined_ps): Define.
1160 (_mm512_undefined_pd): Define.
1161 Use _mm*_undefined_*.
1162 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1164 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1166 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1167 (lshr_simd): DI mode added.
1168 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1169 (aarch64_ushr_simddi): Likewise.
1170 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1171 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1172 (vshrd_n_u64): Likewise.
1174 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1176 * Makefile.in (s-macro_list): Depend on cc1.
1178 2014-03-23 Teresa Johnson <tejohnson@google.com>
1180 * ipa-utils.c (ipa_print_order): Use specified dump file.
1182 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1184 PR rtl-optimization/60601
1185 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1187 * gcc.c (eval_spec_function): Initialize save_growing_value.
1189 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1192 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1193 code == MINUS_EXPR, never swap op0 with op1.
1195 * toplev.c (init_local_tick): Avoid signed integer multiplication
1197 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1198 shift by first operand's bitsize.
1200 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1203 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1205 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1206 TARGET_ISA_64BIT_P(x).
1208 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1210 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1211 pattern for vector nor instead of subtract from splat(-1).
1212 (altivec_expand_vec_perm_const_le): Likewise.
1214 2014-03-21 Richard Henderson <rth@twiddle.net>
1217 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1218 related insns after epilogue_completed.
1220 2014-03-21 Martin Jambor <mjambor@suse.cz>
1223 * cgraph.h (symtab_node): New flag body_removed.
1224 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1225 when removing bodies.
1226 * symtab.c (dump_symtab_base): Dump body_removed flag.
1227 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1228 had their bodies removed.
1230 2014-03-21 Martin Jambor <mjambor@suse.cz>
1233 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1236 2014-03-21 Richard Biener <rguenther@suse.de>
1238 PR tree-optimization/60577
1239 * tree-core.h (struct tree_base): Document nothrow_flag use
1240 in VAR_DECL_NONALIASED.
1241 * tree.h (VAR_DECL_NONALIASED): New.
1242 (may_be_aliased): Adjust.
1243 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1245 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1247 * expr.c (expand_expr_real_1): Remove outdated comment.
1249 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1252 * ira.c (adjust_cleared_regs): Call copy_rtx on
1253 *reg_equiv[REGNO (loc)].src_p before passing it to
1254 simplify_replace_fn_rtx.
1257 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1258 into CONST, put pic register as first operand of PLUS. Use
1259 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1261 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1263 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1265 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1267 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1268 around for store forwarding issue in the FPU on the UT699.
1269 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1270 loads and operations if -mfix-ut699 is specified.
1271 (divtf3_hq): Tweak attribute.
1272 (sqrttf2_hq): Likewise.
1274 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1276 * calls.c (store_one_arg): Remove incorrect const qualification on the
1277 type of the temporary.
1278 * cfgexpand.c (expand_return): Likewise.
1279 * expr.c (expand_constructor): Likewise.
1280 (expand_expr_real_1): Likewise.
1282 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1284 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1287 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1290 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1292 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1294 * config/arm/aarch-common-protos.h
1295 (alu_cost_table): Fix spelling of "extend".
1296 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1298 2014-03-19 Richard Biener <rguenther@suse.de>
1301 * tree-core.h (tree_type_common): Re-order pointer members
1302 to reduce recursion depth during GC walks.
1304 2014-03-19 Marek Polacek <polacek@redhat.com>
1307 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1308 before accessing it.
1310 2014-03-19 Richard Biener <rguenther@suse.de>
1313 * lto-streamer-in.c (input_function): In WPA stage do not drop
1316 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1318 PR tree-optimization/60559
1319 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1320 with build_zero_cst assignment.
1322 2014-03-18 Kai Tietz <ktietz@redhat.com>
1324 PR rtl-optimization/56356
1325 * sdbout.c (sdbout_parms): Verify that parms'
1326 incoming argument is valid.
1327 (sdbout_reg_parms): Likewise.
1329 2014-03-18 Richard Henderson <rth@redhat.com>
1332 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1333 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1334 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1336 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1338 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1339 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1340 Italicize plugin event names in description. Explain that
1341 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1342 Remind that no GCC functions should be called after PLUGIN_FINISH.
1343 Explain what pragmas with expansion are.
1345 2014-03-18 Martin Liska <mliska@suse.cz>
1347 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1348 gimple call statement is update.
1349 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1350 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1352 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1355 * ubsan.c (ubsan_instrument_unreachable): Call
1356 initialize_sanitizer_builtins.
1357 (ubsan_pass): Likewise.
1360 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1361 varpool_finalize_decl instead of rest_of_decl_compilation.
1363 2014-03-18 Richard Biener <rguenther@suse.de>
1365 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1366 by using bitmap_and_compl instead of bitmap_and_compl_into.
1367 (df_rd_transfer_function): Likewise.
1369 2014-03-18 Richard Biener <rguenther@suse.de>
1371 * doc/lto.texi (fresolution): Fix typo.
1373 2014-03-18 Richard Biener <rguenther@suse.de>
1375 * doc/invoke.texi (flto): Update for changes in 4.9.
1377 2014-03-18 Richard Biener <rguenther@suse.de>
1379 * doc/loop.texi: Remove section on the removed lambda framework.
1380 Update loop docs with recent changes in preserving loop structure.
1382 2014-03-18 Richard Biener <rguenther@suse.de>
1384 * doc/lto.texi (-fresolution): Document.
1386 2014-03-18 Richard Biener <rguenther@suse.de>
1388 * doc/contrib.texi: Adjust my name.
1390 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1393 * internal-fn.c: Include diagnostic-core.h.
1394 (expand_BUILTIN_EXPECT): New function.
1395 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1396 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1397 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1398 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1400 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1401 Revert 3 argument __builtin_expect code.
1402 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1403 * gimple-fold.c (gimple_fold_call): Likewise.
1404 * tree.h (fold_builtin_expect): New prototype.
1405 * builtins.c (build_builtin_expect_predicate): Add predictor
1406 argument, if non-NULL, create 3 argument __builtin_expect.
1407 (fold_builtin_expect): No longer static. Add ARG2 argument,
1408 pass it through to build_builtin_expect_predicate.
1409 (fold_builtin_2): Adjust caller.
1410 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1411 * internal-fn.def (BUILTIN_EXPECT): New.
1413 2014-03-18 Tobias Burnus <burnus@net-b.de>
1416 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1417 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1418 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1420 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1423 * predict.c (combine_predictions_for_bb): Fix up formatting.
1424 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1425 fill what it points to if non-NULL.
1426 (tree_predict_by_opcode): Adjust caller, use the predictor.
1427 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1429 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1431 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1432 proper constant for the store mode.
1434 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1436 * symtab.c (change_decl_assembler_name): Fix transparent alias
1439 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1441 * config/aarch64/aarch64.c: Correct the comments about the
1442 aarch64 stack layout.
1444 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1446 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1447 check for GF_OMP_FOR_KIND_FOR.
1449 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1451 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1452 ymm and zmm register names.
1454 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1457 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1458 note creation for the 2010-08-31 changes.
1460 2014-03-17 Marek Polacek <polacek@redhat.com>
1463 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1464 as -fno-tree-loop-vectorize.
1465 (expand_omp_simd): Likewise.
1467 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1469 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1470 (eligible_for_call_delay): New prototype.
1471 * config/sparc/sparc.c (tls_call_delay): Rename into...
1472 (eligible_for_call_delay): ...this. Return false if the instruction
1473 cannot be put in the delay slot of a branch.
1474 (eligible_for_restore_insn): Simplify.
1475 (eligible_for_return_delay): Return false if the instruction cannot be
1476 put in the delay slot of a branch and simplify.
1477 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1478 put in the delay slot of a branch.
1479 * config/sparc/sparc.md (fix_ut699): New attribute.
1480 (tls_call_delay): Delete.
1481 (in_call_delay): Reimplement.
1482 (eligible_for_sibcall_delay): Rename into...
1483 (in_sibcall_delay): ...this.
1484 (eligible_for_return_delay): Rename into...
1485 (in_return_delay): ...this.
1486 (in_branch_delay): Reimplement.
1487 (in_uncond_branch_delay): Delete.
1488 (in_annul_branch_delay): Delete.
1490 2014-03-14 Richard Henderson <rth@redhat.com>
1493 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1494 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1495 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1496 (floathi splitters): Remove.
1497 (float<SWI48x>xf2): New pattern.
1498 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1499 code that tried to handle DImode for 32-bit, but which was excluded
1500 by the pattern's condition. Drop allocation of stack temporary.
1501 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1502 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1503 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1504 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1505 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1506 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1507 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1508 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1509 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1510 (*float<SWI48x><X87MODEF>2_i387): Remove.
1511 (all float _with_temp splitters): Remove.
1512 (*float<SWI48x><MODEF>2_i387): New pattern.
1513 (*float<SWI48><MODEF>2_sse): New pattern.
1514 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1515 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1517 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1518 Marek Polacek <polacek@redhat.com>
1521 * common.opt (dump_base_name_prefixed): New Variable.
1522 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1523 if x_dump_base_name_prefixed is already set, set it at the end.
1525 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1527 PR rtl-optimization/60508
1528 * lra-constraints.c (get_reload_reg): Add new parameter
1530 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1531 Pass the new parameter values.
1533 2014-03-14 Richard Biener <rguenther@suse.de>
1535 * common.opt: Revert unintented changes from r205065.
1538 2014-03-14 Richard Biener <rguenther@suse.de>
1541 * cfghooks.c (split_block): Properly adjust all loops the
1542 block was a latch of.
1544 2014-03-14 Martin Jambor <mjambor@suse.cz>
1547 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1550 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1553 * config/avr/avr.c (avr_set_current_function): Pass function name
1554 through default_strip_name_encoding before sanity checking instead
1555 of skipping the first char of the assembler name.
1557 2014-03-13 Richard Henderson <rth@redhat.com>
1560 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1561 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1562 * config/i386/i386-protos.h: Likewise.
1563 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1564 in the expander instead of a splitter.
1565 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1566 any possibility of requiring a memory.
1567 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1568 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1569 (fp branch splitters): Update for ix86_split_fp_branch.
1570 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1571 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1572 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1573 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1574 (*fop_<MODEF>_3_i387): Likewise.
1575 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1576 (splitters for the fop_* register patterns): Remove.
1577 (fscalexf4_i387): Rename from *fscalexf4_i387.
1578 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1580 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1582 PR tree-optimization/59779
1583 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1584 type for bitsize and maxsize instead of HOST_WIDE_INT.
1586 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1588 PR rtl-optimization/57320
1589 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1590 the CFG after thread_prologue_and_epilogue_insns.
1592 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1594 PR rtl-optimization/57189
1595 * lra-constraints.c (process_alt_operands): Disfavor spilling
1598 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1600 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1602 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1604 PR tree-optimization/59025
1606 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1607 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1609 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1612 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1613 calls of avr_out_plus_1.
1615 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1617 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1618 BB's single pred and update the father loop's latch info later.
1620 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1622 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1626 (VEC_base): Likewise.
1627 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1628 registers, we need to swap double words in little endian mode.
1630 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1631 to be a container mode for 128-bit integer operations added in ISA
1632 2.07. Unlike TImode and PTImode, the preferred register set is
1633 the Altivec/VMX registers for the 128-bit operations.
1635 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1637 (rs6000_split_128bit_ok_p): Likewise.
1639 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1640 macros for creating ISA 2.07 normal and overloaded builtin
1641 functions with 3 arguments.
1642 (BU_P8V_OVERLOAD_3): Likewise.
1643 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1644 for use as overloaded functions.
1645 (VPERM_1TI_UNS): Likewise.
1646 (VSEL_1TI): Likewise.
1647 (VSEL_1TI_UNS): Likewise.
1648 (ST_INTERNAL_1ti): Likewise.
1649 (LD_INTERNAL_1ti): Likewise.
1650 (XXSEL_1TI): Likewise.
1651 (XXSEL_1TI_UNS): Likewise.
1652 (VPERM_1TI): Likewise.
1653 (VPERM_1TI_UNS): Likewise.
1654 (XXPERMDI_1TI): Likewise.
1655 (SET_1TI): Likewise.
1656 (LXVD2X_V1TI): Likewise.
1657 (STXVD2X_V1TI): Likewise.
1658 (VEC_INIT_V1TI): Likewise.
1659 (VEC_SET_V1TI): Likewise.
1660 (VEC_EXT_V1TI): Likewise.
1661 (EQV_V1TI): Likewise.
1662 (NAND_V1TI): Likewise.
1663 (ORC_V1TI): Likewise.
1664 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1665 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1667 (VADDUQM): Likewise.
1668 (VSUBCUQ): Likewise.
1669 (VADDEUQM): Likewise.
1670 (VADDECUQ): Likewise.
1671 (VSUBEUQM): Likewise.
1672 (VSUBECUQ): Likewise.
1674 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1675 __int128_t and __uint128_t types.
1676 (__uint128_type): Likewise.
1677 (altivec_categorize_keyword): Add support for vector __int128_t,
1678 vector __uint128_t, vector __int128, and vector unsigned __int128
1679 as a container type for TImode operations that need to be done in
1680 VSX/Altivec registers.
1681 (rs6000_macro_to_expand): Likewise.
1682 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1683 to support 128-bit integer instructions vaddcuq, vadduqm,
1684 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1685 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1687 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1688 for V1TImode, and set up preferences to use VSX/Altivec registers.
1689 Setup VSX reload handlers.
1690 (rs6000_debug_reg_global): Likewise.
1691 (rs6000_init_hard_regno_mode_ok): Likewise.
1692 (rs6000_preferred_simd_mode): Likewise.
1693 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1694 (easy_altivec_constant): Likewise.
1695 (output_vec_const_move): Likewise.
1696 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1698 (rs6000_expand_vector_extract): Likewise.
1699 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1701 (rs6000_const_vec): Add support for V1TImode.
1702 (rs6000_emit_le_vsx_load): Swap double words when loading or
1703 storing TImode/V1TImode.
1704 (rs6000_emit_le_vsx_store): Likewise.
1705 (rs6000_emit_le_vsx_move): Likewise.
1706 (rs6000_emit_move): Add support for V1TImode.
1707 (altivec_expand_ld_builtin): Likewise.
1708 (altivec_expand_st_builtin): Likewise.
1709 (altivec_expand_vec_init_builtin): Likewise.
1710 (altivec_expand_builtin): Likewise.
1711 (rs6000_init_builtins): Add support for V1TImode type. Add
1712 support for ISA 2.07 128-bit integer builtins. Define type names
1713 for the VSX/Altivec vector types.
1714 (altivec_init_builtins): Add support for overloaded vector
1715 functions with V1TImode type.
1716 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1717 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1719 (rs6000_split_128bit_ok_p): Likewise.
1720 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1721 __int128_t and vector __uint128_t.
1723 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1724 and mode attributes.
1731 (VS_scalar): Likewise.
1732 (VS_double): Likewise.
1733 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1735 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1736 we support the ISA 2.07 128-bit integer arithmetic instructions.
1737 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1738 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1739 and TImode types for use with the builtin functions.
1740 (V1TI_type_node): Likewise.
1741 (unsigned_V1TI_type_node): Likewise.
1742 (intTI_type_internal_node): Likewise.
1743 (uintTI_type_internal_node): Likewise.
1745 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1746 128-bit builtin functions.
1747 (UNSPEC_VADDEUQM): Likewise.
1748 (UNSPEC_VADDECUQ): Likewise.
1749 (UNSPEC_VSUBCUQ): Likewise.
1750 (UNSPEC_VSUBEUQM): Likewise.
1751 (UNSPEC_VSUBECUQ): Likewise.
1752 (VM): Add V1TImode to vector mode iterators.
1754 (VI_unit): Likewise.
1755 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1756 (altivec_vaddcuq): Likewise.
1757 (altivec_vsubuqm): Likewise.
1758 (altivec_vsubcuq): Likewise.
1759 (altivec_vaddeuqm): Likewise.
1760 (altivec_vaddecuq): Likewise.
1761 (altivec_vsubeuqm): Likewise.
1762 (altivec_vsubecuq): Likewise.
1764 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1766 (BOOL_128): Likewise.
1767 (BOOL_REGS_OUTPUT): Likewise.
1768 (BOOL_REGS_OP1): Likewise.
1769 (BOOL_REGS_OP2): Likewise.
1770 (BOOL_REGS_UNARY): Likewise.
1771 (BOOL_REGS_AND_CR0): Likewise.
1773 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1774 128-bit integer builtin support.
1775 (vec_vadduqm): Likewise.
1776 (vec_vaddecuq): Likewise.
1777 (vec_vaddeuqm): Likewise.
1778 (vec_vsubecuq): Likewise.
1779 (vec_vsubeuqm): Likewise.
1780 (vec_vsubcuq): Likewise.
1781 (vec_vsubuqm): Likewise.
1783 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1784 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1785 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1786 128-bit integer add/subtract to ISA 2.07.
1788 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1790 * config/arc/arc.c (arc_predicate_delay_insns):
1791 Fix third argument passed to conditionalize_nonjump.
1793 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1795 * config/aarch64/aarch64-builtins.c
1796 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1797 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1798 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1799 instead of __builtin_lfloor.
1800 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1802 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1804 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1805 (tree_ssa_ifcombine_bb_1): New function.
1806 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1807 is an empty forwarder block to then_bb or vice versa and then_bb
1808 and else_bb are effectively swapped.
1810 2014-03-12 Christian Bruel <christian.bruel@st.com>
1813 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1814 REG_CFA_DEF_CFA note.
1815 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1816 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1818 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1820 PR tree-optimization/60454
1821 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1823 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1825 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1826 Do not define target_cpu_default2 to generic.
1827 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1828 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1829 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1831 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1832 Marc Glisse <marc.glisse@inria.fr>
1834 PR tree-optimization/60502
1835 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1836 instead of build_low_bits_mask.
1838 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1841 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1842 if there are multiple uses, but op doesn't live on E edge.
1843 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1844 clobber stmts before __builtin_unreachable.
1846 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1848 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1849 hard_frame_pointer_rtx.
1850 * cse.c (cse_insn): Remove volatile check.
1851 * cselib.c (cselib_process_insn): Likewise.
1852 * dse.c (scan_insn): Likewise.
1854 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1856 * config/arc/arc.c (conditionalize_nonjump): New function,
1858 (arc_ifcvt): ... this.
1859 (arc_predicate_delay_insns): Use it.
1861 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1863 * config/arc/predicates.md (extend_operand): During/after reload,
1864 allow const_int_operand.
1865 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1866 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1867 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1869 (umulsi3_highpart_i): Likewise.
1871 2014-03-11 Richard Biener <rguenther@suse.de>
1873 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1874 Add asserts to guard possible wrong-code bugs.
1876 2014-03-11 Richard Biener <rguenther@suse.de>
1878 PR tree-optimization/60429
1879 PR tree-optimization/60485
1880 * tree-ssa-structalias.c (set_union_with_increment): Properly
1881 take into account all fields that overlap the shifted vars.
1882 (do_sd_constraint): Likewise.
1883 (do_ds_constraint): Likewise.
1884 (get_constraint_for_ptr_offset): Likewise.
1886 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1888 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1889 (nios2_compute_frame_layout):
1890 Add calculation of cfun->machine->fp_save_offset.
1891 (nios2_expand_prologue): Correct setting of frame pointer register
1893 (nios2_expand_epilogue): Update recovery of stack pointer from
1894 frame pointer accordingly.
1895 (nios2_initial_elimination_offset): Update calculation of offset
1896 for eliminating to HARD_FRAME_POINTER_REGNUM.
1898 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1901 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1902 cgraph_get_create_node on VAR_DECLs.
1904 2014-03-10 Richard Biener <rguenther@suse.de>
1907 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1909 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1911 * config/vms/vms.opt (vms_float_format): New variable.
1913 2014-03-08 Tobias Burnus <burnus@net-b.de>
1915 * doc/invoke.texi (-fcilkplus): Update implementation status.
1917 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1918 Richard Biener <rguenther@suse.de>
1920 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1921 consistently accross all TUs.
1922 (run_gcc): Enable -fshort-double automatically at link at link-time
1923 and disallow override.
1925 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1928 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1929 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1930 if they can't be used.
1932 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1934 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1935 for Solaris 11/x86 ld.
1936 * configure: Regenerate.
1938 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1940 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1941 (LIB_TLS_SPEC): Save as ld_tls_libs.
1942 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1943 (HAVE_AS_IX86_TLSLDM): New test.
1944 * configure, config.in: Regenerate.
1945 * config/i386/i386.c (legitimize_tls_address): Fall back to
1946 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1947 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1948 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1949 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1951 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1953 * common.opt (fira-loop-pressure): Mark as optimization.
1955 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1957 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1958 an OpenMP mappable type.
1960 2014-03-06 Matthias Klose <doko@ubuntu.com>
1962 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1963 MULTILIB_OSDIRNAMES is not defined.
1965 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1966 Meador Inge <meadori@codesourcery.com>
1969 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1970 (arm_legitimize_address): Call legitimize_tls_address for any
1971 arm_tls_referenced_p expression, handle constant addend. Call it
1972 before testing for !TARGET_ARM.
1973 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1975 2014-03-06 Richard Biener <rguenther@suse.de>
1981 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1983 * tree-streamer.c (record_common_node): Assert we don't record
1984 nodes with type double.
1985 (preload_common_node): Skip type double, complex double and double
1986 pointer since it is now frontend dependent due to fshort-double option.
1988 2014-03-06 Richard Biener <rguenther@suse.de>
1990 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1991 or -fno-lto is specified and the linker has full plugin support.
1992 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1993 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1994 * lto-wrapper.c (merge_and_complain): Merge compile-time
1995 optimization levels.
1996 (run_gcc): And pass it through to the link options.
1998 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
2002 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2004 * cselib.c (remove_useless_values): Skip to avoid quadratic
2005 behavior if the condition moved from...
2006 (cselib_process_insn): ... here holds.
2008 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2011 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
2012 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
2015 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
2016 (TM_H): Add x86-tune.def.
2018 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2020 * config/aarch64/aarch64.c (generic_tunings):
2021 Use cortexa57_extra_costs.
2023 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2026 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2027 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2028 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2031 2014-03-04 Heiher <r@hev.cc>
2033 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
2034 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
2036 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
2038 * config/i386/predicates.md (const2356_operand): Change to ...
2039 (const2367_operand): ... this.
2040 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
2042 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2043 (*avx512pf_scatterpf<mode>sf): Ditto.
2044 (avx512pf_scatterpf<mode>df): Ditto.
2045 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2046 (*avx512pf_scatterpf<mode>df): Ditto.
2047 * config/i386/i386.c (ix86_expand_builtin): Update
2048 incorrect hint operand error message.
2050 2014-03-04 Richard Biener <rguenther@suse.de>
2052 * lto-section-in.c (lto_get_section_data): Fix const cast.
2054 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2056 * tree-streamer.c (record_common_node): Assert we don't record
2057 nodes with type double.
2058 (preload_common_node): Skip type double, complex double and double
2059 pointer since it is now frontend dependent due to fshort-double option.
2061 2014-03-04 Richard Biener <rguenther@suse.de>
2064 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
2065 (lto_input_toplevel_asms): Likewise.
2066 * lto-section-in.c (lto_get_section_data): Instead do it here
2069 2014-03-04 Richard Biener <rguenther@suse.de>
2071 PR tree-optimization/60382
2072 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
2073 dead PHIs a reduction.
2075 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
2077 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2079 (_mm_prefetch): Move out of GCC target("sse") pragma.
2080 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2081 GCC target("prfchw") pragma.
2082 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2084 * config/i386/i386.c (ix86_option_override_internal): Enable
2085 -mprfchw with -mprefetchwt1.
2087 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2089 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2092 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2094 * opts.h (CL_PCH_IGNORE): Define.
2095 * targhooks.c (option_affects_pch_p):
2096 Return false for options that have CL_PCH_IGNORE set.
2097 * opt-functions.awk: Process PchIgnore.
2098 * doc/options.texi: Document PchIgnore.
2100 * config/arc/arc.opt (misize): Add PchIgnore property.
2102 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2104 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2105 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2106 constraint on constants to permit them being loaded into
2107 GENERAL_REGS or BASE_REGS.
2109 2014-03-03 Nick Clifton <nickc@redhat.com>
2111 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2112 anti-cacnonical alternatives.
2113 (negandhi3_real): New pattern.
2114 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2116 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2118 * config/avr/avr-mcus.def: Remove atxmega16x1.
2119 * config/avr/avr-tables.opt: Regenerate.
2120 * config/avr/t-multilib: Regenerate.
2121 * doc/avr-mmcu.texi: Regenerate.
2123 2014-03-03 Tobias Grosser <tobias@grosser.es>
2124 Mircea Namolaru <mircea.namolaru@inria.fr>
2126 PR tree-optimization/58028
2127 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2130 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2132 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2133 not handled by recognizers.
2135 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2138 * function.c (expand_function_end): Don't emit
2139 clobber_return_register sequence if clobber_after is a BARRIER.
2140 * cfgexpand.c (construct_exit_block): Append instructions before
2141 return_label to prev_bb.
2143 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2145 * config/rs6000/constraints.md: Document reserved use of "wc".
2147 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2150 * ipa.c (function_and_variable_visibility): When dissolving comdat
2151 group, also set all symbols to local.
2153 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2158 2013-12-14 Jan Hubicka <jh@suse.cz>
2160 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2162 2014-03-02 Jon Beniston <jon@beniston.com>
2168 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2169 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2170 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2171 (simple_return, *simple_return): New patterns
2172 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2173 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2175 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2177 * dwarf2out.c (gen_subprogram_die): Tidy.
2179 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2182 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2183 (*mov_t_msb_neg_negc): ... this new insn.
2185 2014-02-28 Jason Merrill <jason@redhat.com>
2188 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2191 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2194 * dwarf2out.c (decltype_auto_die): New static.
2195 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2196 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2197 (is_cxx_auto): Likewise.
2199 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2201 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2202 we are not using general regs only.
2204 2014-02-28 Richard Biener <rguenther@suse.de>
2207 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2208 previous fix and only allow to remove trivial pre-headers
2209 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2210 (remove_forwarder_block): Properly update the latch of a loop.
2212 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2215 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2216 (cselib_preserved_hash_table): New.
2217 (preserve_constants_and_equivs): Move preserved vals to it.
2218 (cselib_find_slot): Look it up first.
2219 (cselib_init): Initialize it.
2220 (cselib_finish): Release it.
2221 (dump_cselib_table): Dump it.
2223 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2226 * cselib.c (remove_useless_values): Skip to avoid quadratic
2227 behavior if the condition moved from...
2228 (cselib_process_insn): ... here holds.
2230 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2233 * var-tracking.c (vt_initialize): Apply the same condition to
2234 preserve the CFA base value.
2236 2014-02-28 Joey Ye <joey.ye@arm.com>
2239 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2240 if reload in progress or completed.
2242 2014-02-28 Tobias Burnus <burnus@net-b.de>
2245 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2248 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2250 * doc/tm.texi.in (Condition Code Status): Update documention for
2251 relative locations of cc0-setter and cc0-user.
2253 2014-02-27 Jeff Law <law@redhat.com>
2255 PR rtl-optimization/52714
2256 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2257 into two independent simple sets, if I3 is a jump, ensure the
2258 pattern we place into I3 is a (set (pc) ...).
2260 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2261 Jeff Law <law@redhat.com>
2263 PR rtl-optimization/49847
2264 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2265 are in different blocks.
2266 * doc/tm.texi (Condition Code Status): Update documention for
2267 relative locations of cc0-setter and cc0-user.
2269 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2272 * lra.c (lra_emit_add): Check SUBREG too.
2274 2014-02-27 Andreas Schwab <schwab@suse.de>
2276 * config/m68k/m68k.c (m68k_option_override): Disable
2277 -flive-range-shrinkage for classic m68k.
2278 (m68k_override_options_after_change): Likewise.
2280 2014-02-27 Marek Polacek <polacek@redhat.com>
2283 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2284 -Wmaybe-uninitialized.
2286 2014-02-27 Alan Modra <amodra@gmail.com>
2289 * reload1.c (emit_input_reload_insns): When reload_override_in,
2290 set old to rl->in_reg when rl->in_reg is a subreg.
2292 2014-02-26 Richard Biener <rguenther@suse.de>
2295 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2297 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2299 * common/config/i386/predicates.md (const1256_operand): Remove.
2300 (const2356_operand): New.
2301 (const_1_to_2_operand): Remove.
2302 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2303 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2304 (*avx512pf_gatherpf<mode>sf): Ditto.
2305 (avx512pf_gatherpf<mode>df): Ditto.
2306 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2307 (*avx512pf_gatherpf<mode>df): Ditto.
2308 (avx512pf_scatterpf<mode>sf): Ditto.
2309 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2310 (*avx512pf_scatterpf<mode>sf): Ditto.
2311 (avx512pf_scatterpf<mode>df): Ditto.
2312 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2313 (*avx512pf_scatterpf<mode>df): Ditto.
2314 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2316 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2318 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2319 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2320 (_mm512_mask_testn_epi64_mask): Move to ...
2321 * config/i386/avx512cdintrin.h: Here.
2322 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2323 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2324 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2325 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2326 TARGET_AVX512F from TARGET_AVX512CD.
2328 2014-02-26 Richard Biener <rguenther@suse.de>
2331 * ipa.c (walk_polymorphic_call_targets): Properly guard
2332 call to inline_update_overall_summary.
2334 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2337 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2338 and latches only if requested. Fix latch if it is removed.
2339 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2340 LOOPS_HAVE_PREHEADERS.
2342 2014-02-25 Andrew Pinski <apinski@cavium.com>
2344 * builtins.c (expand_builtin_thread_pointer): Create a new target
2345 when the target is NULL.
2347 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2349 PR rtl-optimization/60317
2350 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2351 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2352 * lra-assigns.c: Include params.h.
2353 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2354 other reload pseudos considerations.
2356 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2358 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2359 to use canonical form for nor<mode>3.
2361 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2364 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2367 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2369 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2370 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2371 (ix86_handle_option): Handle OPT_mprefetchwt1.
2372 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2373 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2375 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2376 OPTION_MASK_ISA_PREFETCHWT1.
2377 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2378 (PTA_PREFETCHWT1): New.
2379 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2380 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2381 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2382 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2383 (*prefetch_avx512pf_<mode>_: Change into ...
2384 (*prefetch_prefetchwt1_<mode>: This.
2385 * config/i386/i386.opt (mprefetchwt1): New.
2386 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2387 (_mm_prefetch): Handle intent to write.
2388 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2390 2014-02-25 Richard Biener <rguenther@suse.de>
2393 * emit-rtl.c (mem_attrs_htab): Remove.
2394 (mem_attrs_htab_hash): Likewise.
2395 (mem_attrs_htab_eq): Likewise.
2396 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2397 (init_emit_once): Do not allocate mem_attrs_htab.
2399 2014-02-25 Richard Biener <rguenther@suse.de>
2402 * lto-opts.c (lto_write_options): Output non-explicit conservative
2403 -fwrapv, -fno-trapv and -fno-strict-overflow.
2404 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2405 (run_gcc): And pass them through.
2407 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2409 * sel-sched.c (calculate_new_fences): New parameter ptime.
2410 Calculate it as a maximum over all fence cycles.
2411 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2412 Print the final schedule timing when sched_verbose.
2414 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2416 PR rtl-optimization/60292
2417 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2418 bit fot the fence instruction.
2420 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2422 * calls.h: Fix typo in comment.
2424 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2426 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2427 adjusting offsetable addresses.
2429 2014-02-24 Guozhi Wei <carrot@google.com>
2431 * sparseset.h (sparseset_pop): Fix the wrong index.
2433 2014-02-24 Walter Lee <walt@tilera.com>
2435 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2436 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2438 * common/config/tilegx/tilegx-common.c
2439 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2440 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2442 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2443 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2444 (tilegx_gimplify_va_arg_expr): Handle big endian.
2445 (tilegx_expand_unaligned_load): Ditto.
2446 (tilegx_expand_unaligned_store): Ditto.
2447 (TARGET_RETURN_IN_MSB): New.
2448 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2449 (TARGET_ENDIAN_DEFAULT): New.
2450 (TARGET_BIG_ENDIAN): Handle big endian.
2451 (BYTES_BIG_ENDIAN): Ditto.
2452 (WORDS_BIG_ENDIAN): Ditto.
2453 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2456 * config/tilegx/tilegx.md (extv): Handle big endian.
2458 (insn_st<n>): Ditto.
2459 (insn_st<n>_add<bitsuffix>): Ditto.
2460 (insn_stnt<n>): Ditto.
2461 (insn_stnt<n>_add<bitsuffix>):Ditto.
2462 (vec_interleave_highv8qi): Handle big endian.
2463 (vec_interleave_highv8qi_be): New.
2464 (vec_interleave_highv8qi_le): New.
2465 (insn_v1int_h): Handle big endian.
2466 (vec_interleave_lowv8qi): Handle big endian.
2467 (vec_interleave_lowv8qi_be): New.
2468 (vec_interleave_lowv8qi_le): New.
2469 (insn_v1int_l): Handle big endian.
2470 (vec_interleave_highv4hi): Handle big endian.
2471 (vec_interleave_highv4hi_be): New.
2472 (vec_interleave_highv4hi_le): New.
2473 (insn_v2int_h): Handle big endian.
2474 (vec_interleave_lowv4hi): Handle big endian.
2475 (vec_interleave_lowv4hi_be): New.
2476 (vec_interleave_lowv4hi_le): New.
2477 (insn_v2int_l): Handle big endian.
2478 (vec_interleave_highv2si): Handle big endian.
2479 (vec_interleave_highv2si_be): New.
2480 (vec_interleave_highv2si_le): New.
2481 (insn_v4int_h): Handle big endian.
2482 (vec_interleave_lowv2si): Handle big endian.
2483 (vec_interleave_lowv2si_be): New.
2484 (vec_interleave_lowv2si_le): New.
2485 (insn_v4int_l): Handle big endian.
2486 * config/tilegx/tilegx.opt (mbig-endian): New option.
2487 (mlittle-endian): New option.
2488 * doc/install.texi: Document tilegxbe-linux.
2489 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2491 2014-02-24 Martin Jambor <mjambor@suse.cz>
2494 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2495 there are no parameter descriptors.
2497 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2499 PR rtl-optimization/60268
2500 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2501 initialization to ...
2502 (sched_rgn_init): ... here.
2503 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2505 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2507 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2510 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2512 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2515 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2517 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2518 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2520 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2522 * config/microblaze/predicates.md: Add cmp_op predicate.
2523 * config/microblaze/microblaze.md: Add branch_compare instruction
2524 which uses cmp_op predicate and emits cmp insn before branch.
2525 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2526 to microblaze_expand_conditional_branch and consolidate logic.
2527 (microblaze_expand_conditional_branch): emit branch_compare
2528 insn instead of handling cmp op separate from branch insn.
2530 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2532 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2535 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2537 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2538 define_insn with define_expand and new define_insn
2539 *altivec_lve<VI_char>x_internal.
2540 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2541 and new define_insn *altivec_stve<VI_char>x_internal.
2542 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2544 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2546 (altivec_expand_stvex_be): New function.
2548 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2550 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2551 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2552 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2553 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2555 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2558 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2559 instead of emit_move_insn.
2561 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2563 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2565 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2566 gen_altivec_vsumsws.
2568 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2570 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2571 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2572 (altivec_lvxl_<mode>): New define_expand incorporating
2573 -maltivec=be semantics where needed.
2574 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2575 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2576 semantics where needed.
2577 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2578 (altivec_stvx_<mode>): New define_expand incorporating
2579 -maltivec=be semantics where needed.
2580 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2581 VM2 iterator instead of V4SI.
2582 (altivec_stvxl_<mode>): New define_expand incorporating
2583 -maltivec=be semantics where needed.
2584 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2585 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2586 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2587 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2588 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2589 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2590 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2591 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2592 ALTIVEC_BUILTIN_STVXL.
2593 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2594 (altivec_expand_stvx_be): Likewise.
2595 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2596 (altivec_expand_lvx_be): Likewise.
2597 (altivec_expand_stvx_be): Likewise.
2598 (altivec_expand_builtin): Add cases for
2599 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2600 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2601 (altivec_init_builtins): Add definitions for
2602 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2603 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2605 2014-02-21 Catherine Moore <clm@codesourcery.com>
2607 * doc/invoke.texi (mvirt, mno-virt): Document.
2608 * config/mips/mips.opt (mvirt): New option.
2609 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2611 2014-02-21 Richard Biener <rguenther@suse.de>
2613 PR tree-optimization/60276
2614 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2615 (STMT_VINFO_MIN_NEG_DIST): New macro.
2616 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2617 STMT_VINFO_MIN_NEG_DIST.
2618 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2619 made for negative dependence distances still hold.
2621 2014-02-21 Richard Biener <rguenther@suse.de>
2624 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2625 DECL_INITIAL for globals not in the current function context.
2627 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2629 PR tree-optimization/56490
2630 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2631 * tree-ssa-uninit.c: Include params.h.
2632 (compute_control_dep_chain): Add num_calls argument, return false
2633 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2634 num_calls to recursive call.
2635 (find_predicates): Change dep_chain into normal array,
2636 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2637 variable and adjust compute_control_dep_chain caller.
2638 (find_def_preds): Likewise.
2640 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2642 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2643 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2645 2014-02-21 Nick Clifton <nickc@redhat.com>
2647 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2648 (pushhi1): Likewise.
2649 (popqi1): Add mode to pre_dec.
2652 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2654 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2655 mode for mask of V8SFmode permutation.
2657 2014-02-20 Richard Henderson <rth@redhat.com>
2660 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2661 a new pseudo for OLDVAL.
2663 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2666 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2667 gen_reg_rtx if d->testing_p.
2668 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2669 if d->testing_p and we will certainly return true.
2670 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2673 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2675 * emit-rtl.c (gen_reg_rtx): Assert that
2676 crtl->emit.regno_pointer_align_length is non-zero.
2678 2014-02-20 Richard Henderson <rth@redhat.com>
2681 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2682 on failure the store back into EXPECT.
2684 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2685 Sandra Loosemore <sandra@codesourcery.com>
2687 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2688 * config/nios2/nios2.c (nios2_function_profiler): Add
2689 -fPIC (flag_pic == 2) support.
2690 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2691 (nios2_large_offset_p): New function.
2692 (nios2_unspec_reloc_p): Move up position, update to use
2693 nios2_large_offset_p.
2694 (nios2_unspec_address): Remove function.
2695 (nios2_unspec_offset): New function.
2696 (nios2_large_got_address): New function.
2697 (nios2_got_address): Add large offset support.
2698 (nios2_legitimize_tls_address): Update usage of removed and new
2700 (nios2_symbol_binds_local_p): New function.
2701 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2702 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2703 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2704 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2705 processing for (const (unspec ...)).
2706 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2708 2014-02-20 Richard Biener <rguenther@suse.de>
2710 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2711 doing the substitution.
2712 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2714 2014-02-20 Martin Jambor <mjambor@suse.cz>
2717 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2718 info when checking whether lattices are bottom.
2720 2014-02-20 Richard Biener <rguenther@suse.de>
2723 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2726 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2729 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2730 parameter specifying the scaling.
2731 (inline_call): Update.
2732 (want_inline_recursively): Guard division by zero.
2733 (recursive_inlining): Update.
2734 * ipa-inline.h (clone_inlined_nodes): Update.
2736 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2739 * config/i386/i386.c (classify_argument): Pass structures of size
2740 64 bytes or less in register.
2742 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2743 Kirill Yukhin <kirill.yukhin@intel.com>
2745 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2746 (_mm_rcp28_round_ss): Ditto.
2747 (_mm_rsqrt28_round_sd): Ditto.
2748 (_mm_rsqrt28_round_ss): Ditto.
2749 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2750 (_mm_rcp14_round_ss): Ditto.
2751 (_mm_rsqrt14_round_sd): Ditto.
2752 (_mm_rsqrt14_round_ss): Ditto.
2753 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2754 the first input operand, get rid of match_dup.
2755 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2757 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2759 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2760 operand as the first input operand, set type attribute.
2761 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2763 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2764 operand as the first input operand, set type attribute.
2766 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2768 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2771 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2774 * config/i386/i386.c (construct_container): Remove TFmode check
2775 for X86_64_INTEGER_CLASS.
2777 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2780 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2781 only when -Wpsabi is enabled.
2783 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2786 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2787 passing arrays in registers are the same as for structs, so remove the
2788 special case for them.
2790 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2792 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2793 destination type, extract only the valid bits if the source type is not
2794 integral and has a different mode.
2796 2014-02-19 Richard Biener <rguenther@suse.de>
2799 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2802 2014-02-19 Richard Biener <rguenther@suse.de>
2805 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2806 (ipa_modify_call_arguments): Emit an argument load explicitely and
2807 preserve virtual SSA form there and for the replacement call.
2808 Do not update SSA form nor free dominance info.
2810 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2812 * ipa.c (function_and_variable_visibility): Also clear WEAK
2813 flag when disolving COMDAT_GROUP.
2815 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2817 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2818 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2820 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2821 do more sanity checks.
2822 (detect_type_change): Return true when giving up early.
2823 (compute_complex_assign_jump_func): Fix type parameter of
2824 ipa_set_ancestor_jf.
2825 (compute_complex_ancestor_jump_func): Likewise.
2826 (update_jump_functions_after_inlining): Fix updating of
2828 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2830 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2832 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2833 inline clones when edge disappears.
2835 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2838 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2839 Split 64-bit moves into 2 patterns. Do not allow the use of
2840 direct move for TDmode in little endian, since the decimal value
2841 has little endian bytes within a word, but the 64-bit pieces are
2842 ordered in a big endian fashion, and normal subreg's of TDmode are
2844 (mov<mode>_64bit_dm): Likewise.
2845 (movtd_64bit_nodm): Likewise.
2847 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2849 PR tree-optimization/60174
2850 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2851 statement of an SSA_NAME that occurs in an abnormal PHI node.
2853 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2856 * final.c (SEEN_BB): Remove.
2857 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2858 (final_scan_insn): Don't force_source_line on second
2859 NOTE_INSN_BASIC_BLOCK.
2861 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2864 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2865 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2866 (type_natural_mode): Warn ABI change when %zmm register is not
2867 available for AVX512F vector value passing.
2869 2014-02-18 Kai Tietz <ktietz@redhat.com>
2872 * config/i386/i386.c (ix86_expand_prologue): Use value in
2873 rax register as displacement when restoring %r10 or %rax.
2874 Fix wrong offset when restoring both registers.
2876 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2878 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2879 assertion with conditional return.
2881 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2882 Uros Bizjak <ubizjak@gmail.com>
2885 * config/i386/driver-i386.c (host_detect_local_cpu): If
2886 YMM state is not saved by the OS, also clear has_f16c. Move
2887 CPUID 0x80000001 handling before YMM state saving checking.
2889 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2891 PR rtl-optimization/58960
2892 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2893 factored out from ...
2894 (sched_init): ... here.
2895 (free_global_sched_pressure_data): New, factored out from ...
2896 (sched_finish): ... here.
2897 * sched-int.h (free_global_sched_pressure_data): Declare.
2898 * sched-rgn.c (nr_regions_initial): New static global.
2899 (haifa_find_rgns): Initialize it.
2900 (schedule_region): Disable sched-pressure for the newly
2903 2014-02-17 Richard Biener <rguenther@suse.de>
2905 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2906 release SSA defs of pattern stmts.
2908 2014-02-17 Richard Biener <rguenther@suse.de>
2910 * tree-inline.c (expand_call_inline): Release the virtual
2911 operand defined by the call we are about to inline.
2913 2014-02-17 Richard Biener <rguenther@suse.de>
2915 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2917 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2918 Ilya Tocar <ilya.tocar@intel.com>
2920 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2921 arguments order in builtin.
2922 (_mm512_permutexvar_epi64): Ditto.
2923 (_mm512_mask_permutexvar_epi64): Ditto
2924 (_mm512_maskz_permutexvar_epi32): Ditto
2925 (_mm512_permutexvar_epi32): Ditto
2926 (_mm512_mask_permutexvar_epi32): Ditto
2928 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2930 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2931 (p8_vmrgow): Likewise.
2933 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2935 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2938 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2941 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2942 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2943 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2944 using direct move instructions on ISA 2.07. Also adjust
2945 instruction length for 64-bit.
2946 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2947 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2949 2014-02-15 Alan Modra <amodra@gmail.com>
2953 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2954 find_replacement on parts of insn rtl that might be reloaded.
2956 2014-02-15 Richard Biener <rguenther@suse.de>
2958 PR tree-optimization/60183
2959 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2960 (tree_ssa_phiprop): Calculate and free post-dominators.
2962 2014-02-14 Jeff Law <law@redhat.com>
2964 PR rtl-optimization/60131
2965 * ree.c (get_extended_src_reg): New function.
2966 (combine_reaching_defs): Use it rather than assuming location of REG.
2967 (find_and_remove_re): Verify first operand of extension is
2968 a REG before adding the insns to the copy list.
2970 2014-02-14 Roland McGrath <mcgrathr@google.com>
2972 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2973 * configure: Regenerated.
2974 * config.in: Regenerated.
2975 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2976 instead of ASM_SHORT.
2978 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2979 Richard Earnshaw <rearnsha@arm.com>
2981 PR rtl-optimization/59535
2982 * lra-constraints.c (process_alt_operands): Encourage alternative
2983 when unassigned pseudo class is superset of the alternative class.
2984 (inherit_reload_reg): Don't inherit when optimizing for code size.
2985 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2986 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2987 modes not less than 4 for Thumb1.
2989 2014-02-14 Kyle McMartin <kyle@redhat.com>
2992 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2994 2014-02-14 Richard Biener <rguenther@suse.de>
2996 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2997 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2998 Do not strip INDIRECT_REFs.
3000 2014-02-14 Richard Biener <rguenther@suse.de>
3003 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
3004 DECL_FUNCTION_SPECIFIC_TARGET.
3005 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
3006 * tree-streamer-out.c (pack_ts_target_option): Remove.
3007 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
3008 (write_ts_function_decl_tree_pointers): Do not stream
3009 DECL_FUNCTION_SPECIFIC_TARGET.
3010 * tree-streamer-in.c (unpack_ts_target_option): Remove.
3011 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
3012 (lto_input_ts_function_decl_tree_pointers): Do not stream
3013 DECL_FUNCTION_SPECIFIC_TARGET.
3015 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3017 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
3018 (get_initial_def_for_induction, vectorizable_induction): Ignore
3019 debug stmts when looking for exit_phi.
3020 (vectorizable_live_operation): Fix up condition.
3022 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3024 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3025 nreverse() because it changes the content of original tree list.
3027 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3029 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3030 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3032 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3034 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
3035 GNU coding standards.
3037 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3040 * dwarf2out.c (gen_subprogram_die): Don't call
3041 add_calling_convention_attribute if subr_die is old_die.
3043 2014-02-13 Sharad Singhai <singhai@google.com>
3045 * doc/optinfo.texi: Fix order of nodes.
3047 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3049 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
3050 operands[2], not operands[3].
3052 2014-02-13 Richard Biener <rguenther@suse.de>
3055 * doc/install.texi (ISL): Update recommended version to 0.12.2,
3056 mention the possibility of an in-tree build.
3057 (CLooG): Update recommended version to 0.18.1, mention the
3058 possibility of an in-tree build and clarify that the ISL
3059 bundled with CLooG does not work.
3061 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3064 * expr.c (compress_float_constant): If x is a hard register,
3065 extend into a pseudo and then move to x.
3067 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
3069 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
3070 caused by bad second argument to warning_at() with -mhotpatch and
3071 nested functions (e.g. with gfortran).
3073 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
3075 * opts.c (option_name): Remove "enabled by default" rider.
3077 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3079 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3081 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3082 Uros Bizjak <ubizjak@gmail.com>
3085 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3086 * configure: Regenerated.
3088 2014-02-12 Richard Biener <rguenther@suse.de>
3090 * vec.c (vec_prefix::calculate_allocation): Move as
3091 inline variant to vec.h.
3092 (vec_prefix::calculate_allocation_1): New out-of-line version.
3093 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3094 (vec_prefix::m_has_auto_buf): Rename to ...
3095 (vec_prefix::m_using_auto_storage): ... this.
3096 (vec_prefix::calculate_allocation): Inline the easy cases
3097 and dispatch to calculate_allocation_1 which doesn't need the
3099 (va_heap::reserve): Use gcc_checking_assert.
3100 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3101 m_using_auto_storage.
3102 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3104 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3105 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3106 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3108 2014-02-12 Richard Biener <rguenther@suse.de>
3110 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3111 when we found a dependence.
3113 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3115 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3117 (maybe_fold_stmt): ... into this new function.
3118 * omp-low.c (lower_omp): Update comment.
3120 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3123 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3126 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3128 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3129 identifiers in comments.
3130 (cortexa53_extra_costs): Likewise.
3131 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3132 (cortexa7_extra_costs): Likewise.
3133 (cortexa12_extra_costs): Likewise.
3134 (cortexa15_extra_costs): Likewise.
3135 (v7m_extra_costs): Likewise.
3137 2014-02-12 Richard Biener <rguenther@suse.de>
3140 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3141 of posix_memalign being successful.
3142 (lower_stmt): Restrict lowering of posix_memalign to when
3143 -ftree-bit-ccp is enabled.
3145 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3147 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3149 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3151 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3153 PR rtl-optimization/60116
3154 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3155 other_insn once the combination has been validated.
3157 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3160 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3162 * ipa-devirt.c: Include demangle.h
3163 (odr_violation_reported): New static variable.
3164 (add_type_duplicate): Update odr_violations.
3165 (maybe_record_node): Add completep parameter; update it.
3166 (record_target_from_binfo): Add COMPLETEP parameter;
3167 update it as needed.
3168 (possible_polymorphic_call_targets_1): Likewise.
3169 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3170 rename FINAL to COMPLETE.
3171 (record_targets_from_bases): Sanity check we found the binfo;
3172 fix COMPLETEP updating.
3173 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3174 parameter, fix computing of COMPLETEP.
3175 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3176 at LTO time do demangling.
3177 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3178 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3180 (gimple_get_virt_method_for_binfo): Likewise.
3181 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3182 gimple_get_virt_method_for_vtable): Update prototypes.
3184 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3187 * genautomata.c (add_presence_absence): Fix typo with
3188 {final_}presence_list.
3190 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3193 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3194 for VSX/Altivec vectors that land in GPR registers.
3196 2014-02-11 Richard Henderson <rth@redhat.com>
3197 Jakub Jelinek <jakub@redhat.com>
3200 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3201 around drhs if type conversion to lacc->type is not useless.
3203 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3205 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3207 (cortex-a57.cortex-a53): Likewise.
3208 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3210 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3212 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3215 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3217 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3218 add_options_for_arm_vfp3.
3220 2014-02-11 Jeff Law <law@redhat.com>
3223 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3224 object with an undesirable mode.
3226 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3229 * config/i386/sol2-9.h: New file.
3230 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3231 *-*-solaris2.9*): Use it.
3233 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3235 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3236 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3238 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3240 * config/microblaze/microblaze.c: Extend mcpu version format
3242 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3244 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3246 2014-02-10 Richard Henderson <rth@redhat.com>
3249 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3250 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3251 ms-abi vs -mno-accumulate-outgoing-args.
3252 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3253 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3256 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3259 * cfgexpand.c (expand_asm_operands): Attach source location to
3260 ASM_INPUT rtx objects.
3261 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3263 2014-02-10 Nick Clifton <nickc@redhat.com>
3265 * config/mn10300/mn10300.c (popcount): New function.
3266 (mn10300_expand_prologue): Include saved registers in stack usage
3269 2014-02-10 Jeff Law <law@redhat.com>
3272 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3273 when changing the SET_DEST of a prior insn to avoid an input reload.
3275 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3277 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3278 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3279 -mcall-openbsd, or -mcall-linux.
3280 (CC1_ENDIAN_BIG_SPEC): Remove.
3281 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3282 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3283 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3284 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3285 and %cc1_endian_default.
3286 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3288 2014-02-10 Richard Biener <rguenther@suse.de>
3290 PR tree-optimization/60115
3291 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3292 MEM_REF handling. Properly verify that the accesses are not
3293 out of the objects bound.
3295 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3297 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3300 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3302 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3303 proper constants and fix formatting.
3304 (possible_polymorphic_call_targets): Fix formatting.
3306 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3307 Ilya Tocar <ilya.tocar@intel.com>
3309 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3310 (_mm512_loadu_epi32): Renamed into...
3311 (_mm512_loadu_si512): This.
3312 (_mm512_storeu_epi32): Renamed into...
3313 (_mm512_storeu_si512): This.
3314 (_mm512_maskz_ceil_ps): Removed.
3315 (_mm512_maskz_ceil_pd): Ditto.
3316 (_mm512_maskz_floor_ps): Ditto.
3317 (_mm512_maskz_floor_pd): Ditto.
3318 (_mm512_floor_round_ps): Ditto.
3319 (_mm512_floor_round_pd): Ditto.
3320 (_mm512_ceil_round_ps): Ditto.
3321 (_mm512_ceil_round_pd): Ditto.
3322 (_mm512_mask_floor_round_ps): Ditto.
3323 (_mm512_mask_floor_round_pd): Ditto.
3324 (_mm512_mask_ceil_round_ps): Ditto.
3325 (_mm512_mask_ceil_round_pd): Ditto.
3326 (_mm512_maskz_floor_round_ps): Ditto.
3327 (_mm512_maskz_floor_round_pd): Ditto.
3328 (_mm512_maskz_ceil_round_ps): Ditto.
3329 (_mm512_maskz_ceil_round_pd): Ditto.
3330 (_mm512_expand_pd): Ditto.
3331 (_mm512_expand_ps): Ditto.
3332 * config/i386/i386.c (ix86_builtins): Remove
3333 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3334 (bdesc_args): Ditto.
3335 * config/i386/predicates.md (const1256_operand): New.
3336 (const_1_to_2_operand): Ditto.
3337 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3338 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3339 (*avx512pf_gatherpf<mode>sf): Ditto.
3340 (avx512pf_gatherpf<mode>df): Ditto.
3341 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3342 (*avx512pf_gatherpf<mode>df): Ditto.
3343 (avx512pf_scatterpf<mode>sf): Ditto.
3344 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3345 (*avx512pf_scatterpf<mode>sf): Ditto.
3346 (avx512pf_scatterpf<mode>df): Ditto.
3347 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3348 (*avx512pf_scatterpf<mode>df): Ditto.
3349 (avx512f_expand<mode>): Removed.
3350 (<shift_insn><mode>3<mask_name>): Change predicate type.
3352 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3354 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3355 not at the end of datarefs vector use ordered_remove to avoid
3356 reordering datarefs vector.
3359 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3360 mark local addressable non-static vars as GOVD_PRIVATE
3361 instead of GOVD_LOCAL.
3362 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3363 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3367 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3368 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3369 assume_aligned or alloc_align attributes.
3370 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3371 arguments. Handle also assume_aligned and alloc_align attributes.
3372 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3373 calls to functions with assume_aligned or alloc_align attributes.
3374 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3376 2014-02-08 Terry Guo <terry.guo@arm.com>
3378 * doc/invoke.texi: Document ARM -march=armv7e-m.
3380 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3382 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3383 flag on __cilkrts_rethrow builtin.
3386 * ipa-cp.c (determine_versionability): Fail at -O0
3387 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3388 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3391 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3394 * tree-inline.c (copy_forbidden): Fail for
3395 __attribute__((optimize (0))) functions.
3397 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3399 * varpool.c: Include pointer-set.h.
3400 (varpool_remove_unreferenced_decls): Variables in other partitions
3401 will not be output; be however careful to not lose information
3404 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3406 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3407 lookup in the vtable constructor.
3409 2014-02-07 Jeff Law <law@redhat.com>
3412 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3413 define_insn_and_split.
3415 * ipa-inline.c (inline_small_functions): Fix typos.
3417 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3419 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3420 (s390_can_use_return_insn): Declare.
3421 * config/s390/s390.h (EPILOGUE_USES): Define.
3422 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3424 (s390_chunkify_start): Handle return JUMP_LABELs.
3425 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3426 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3427 (s390_can_use_return_insn): New functions.
3428 (s390_fix_long_loop_prediction): Handle conditional returns.
3429 (TARGET_SET_UP_BY_PROLOGUE): Define.
3430 * config/s390/s390.md (ANY_RETURN): New code iterator.
3431 (*creturn, *csimple_return, return, simple_return): New patterns.
3433 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3435 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3436 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3437 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3438 REG_CFA_RESTORE list when deciding not to restore a register.
3440 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3442 * config/s390/s390.c: Include tree-pass.h and context.h.
3443 (s390_early_mach): New function, split out from...
3444 (s390_emit_prologue): ...here.
3445 (pass_data_s390_early_mach): New pass structure.
3446 (pass_s390_early_mach): New class.
3447 (s390_option_override): Create and register early_mach pass.
3448 Move to end of file.
3450 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3452 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3453 to match for the exit block.
3455 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3457 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3458 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3459 Reject misaligned operands.
3461 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3463 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3465 2014-02-07 Richard Biener <rguenther@suse.de>
3468 * gimple-low.c (lower_builtin_posix_memalign): New function.
3469 (lower_stmt): Call it to lower posix_memalign in a way
3470 to make alignment info accessible.
3472 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3475 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3476 __builtin_setjmp_receiver.
3478 2014-02-07 Richard Biener <rguenther@suse.de>
3481 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3482 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3483 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3484 Handle BUILT_IN_POSIX_MEMALIGN.
3485 (find_func_clobbers): Likewise.
3486 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3487 (call_may_clobber_ref_p_1): Likewise.
3489 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3492 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3495 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3498 * lto-cgraph.c (lto_output_node): Use
3499 symtab_get_symbol_partitioning_class.
3500 (lto_output_varpool_node): likewise.
3501 (symtab_get_symbol_partitioning_class): Move here from
3503 * cgraph.h (symbol_partitioning_class): Likewise.
3504 (symtab_get_symbol_partitioning_class): Declare.
3506 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3508 * ggc.h (ggc_internal_cleared_alloc): New macro.
3509 * vec.h (vec_safe_copy): Handle memory stats.
3510 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3511 * target-globals.c (save_target_globals): Likewise.
3513 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3516 * expr.c (emit_move_resolve_push): Export; be bit more selective
3517 on when to clear alias set.
3518 * expr.h (emit_move_resolve_push): Declare.
3519 * function.h (struct function): Add tail_call_marked.
3520 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3521 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3522 * config/i386/i386.md (TImode move expander): De not call
3524 (FP push expanders): Preserve memory attributes.
3525 * config/i386/sse.md (push<mode>1): Remove.
3526 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3527 (ix86_expand_push): Remove.
3528 * config/i386/mmx.md (push<mode>1): Remove.
3530 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3532 PR rtl-optimization/60030
3533 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3534 lopart with paradoxical subreg before shifting it up by hprec.
3536 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3538 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3539 Remove extra newline at end of file.
3540 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3541 (arm_issue_rate): Handle cortexa57.
3542 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3543 (cortex-a57.cortex-a53): Likewise.
3545 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3548 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3549 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3551 (arm_expand_prologue): Adjust all callers.
3552 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3553 info, registers also at the lowest numbered registers side. Use
3554 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3558 * var-tracking.c (adjust_mems): Before adding a SET to
3559 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3561 2014-02-06 Alan Modra <amodra@gmail.com>
3564 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3565 change SDmode to DDmode when lra_in_progress.
3567 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3570 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3571 free_data_ref on the dr first, and before goto again also set dr
3572 to the next dr. For simd_lane_access, free old datarefs[i] before
3573 overwriting it. For get_vectype_for_scalar_type failure, don't
3574 free_data_ref if simd_lane_access.
3576 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3579 * tree.h (opts_for_fn): New inline function.
3580 (opt_for_fn): Define.
3581 * config/i386/i386.c (ix86_function_regparm): Use
3582 opt_for_fn (decl, optimize) instead of optimize.
3584 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3586 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3587 for SYMBOL_REF in large memory model.
3589 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3591 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3593 (cortex-a57): Likewise.
3594 (cortex-a57.cortex-a53): Likewise.
3596 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3597 Kugan Vivekanandarajah <kuganv@linaro.org>
3599 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3601 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3603 2014-02-06 Richard Biener <rguenther@suse.de>
3605 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3606 set_loop_copy and initialize_original_copy_tables.
3608 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3610 * config/aarch64/aarch64-simd.md
3611 (aarch64_ashr_simddi): Change QI to SI.
3613 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3614 Jakub Jelinek <jakub@redhat.com>
3617 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3620 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3622 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3623 CODE_FOR_altivec_vpku[hw]um to
3624 CODE_FOR_altivec_vpku[hw]um_direct.
3625 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3626 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3627 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3628 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3630 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3632 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3633 generation for -maltivec=be.
3634 (altivec_vsumsws): Simplify redundant test.
3636 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3638 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3639 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3640 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3641 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3642 gen_altivec_vpkuwum.
3643 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3645 (altivec_vpks<VI_char>ss): Likewise.
3646 (altivec_vpks<VI_char>us): Likewise.
3647 (altivec_vpku<VI_char>us): Likewise.
3648 (altivec_vpku<VI_char>um): Likewise.
3649 (altivec_vpku<VI_char>um_direct): New (copy of
3650 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3652 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3653 target is little endian and -maltivec=be is not specified.
3654 (*altivec_vupkhs<VU_char>_direct): New (copy of
3655 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3656 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3657 target is little endian and -maltivec=be is not specified.
3658 (*altivec_vupkls<VU_char>_direct): New (copy of
3659 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3660 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3661 little endian and -maltivec=be is not specified.
3662 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3663 little endian and -maltivec=be is not specified.
3665 2014-02-05 Richard Henderson <rth@redhat.com>
3668 * combine-stack-adj.c: Revert r206943.
3669 * sched-int.h (struct deps_desc): Add last_args_size.
3670 * sched-deps.c (init_deps): Initialize it.
3671 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3672 contain REG_ARGS_SIZE notes.
3674 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3676 * lto-cgraph.c (asm_nodes_output): Make global.
3677 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3678 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3679 (driver_handle_option): Handle OPT_fwpa.
3681 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3684 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3685 a comment typo and formatting issue. If odr_hash hasn't been
3686 created, return vNULL and set *completep to false.
3689 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3690 bb with no successors.
3692 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3695 * doc/invoke.texi (-march): Clarify documentation for ARM.
3699 2014-02-05 Richard Biener <rguenther@suse.de>
3701 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3702 when not vectorizing because of too many alias checks.
3703 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3704 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3706 2014-02-05 Nick Clifton <nickc@redhat.com>
3708 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3709 accept extended registers in any mode when compiling for the MN10300.
3711 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3713 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3714 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3715 sanitization attributes.
3716 (can_inline_edge_p): Likewise.
3717 (sanitize_attrs_match_for_inline_p): New function.
3719 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3721 * ipa-prop.c (detect_type_change): Shor circuit testing of
3722 type changes on THIS pointer.
3724 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3727 * config/pa/pa.c (legitimize_tls_address): Return original address
3728 if not passed a SYMBOL_REF rtx.
3729 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3731 (pa_emit_move_sequence): Simplify TLS source operands.
3732 (pa_legitimate_constant_p): Reject all TLS constants.
3733 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3734 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3736 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3738 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3739 groups when we know they are controlled by LTO.
3740 * varasm.c (default_binds_local_p_1): If object is in other partition,
3741 it will be resolved locally.
3743 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3745 * config/host-linux.c (linux_gt_pch_use_address): Don't
3746 use SSIZE_MAX because it is not always defined.
3748 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3751 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3752 threshold for pseudo splitting.
3753 (update_ebb_live_info): Process call argument hard registers and
3754 hard registers from insn definition too.
3755 (max_small_class_regs_num): New constant.
3756 (inherit_in_ebb): Update live hard regs through EBBs. Update
3757 reloads_num only for small register classes. Don't split for
3760 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3763 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3766 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3768 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3771 2014-02-04 Marek Polacek <polacek@redhat.com>
3773 * gdbinit.in (pel): Define.
3775 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3777 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3780 2014-02-04 Richard Biener <rguenther@suse.de>
3783 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3784 in function context local.
3785 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3786 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3787 similar to LTO_imported_decl_ref.
3789 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3791 PR tree-optimization/60002
3792 * cgraphclones.c (build_function_decl_skip_args): Clear
3795 PR tree-optimization/60023
3796 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3797 false to gsi_replace.
3798 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3799 has been in some EH region and vec_stmt could throw, add
3800 vec_stmt into the same EH region.
3801 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3802 has no lhs, ignore it.
3803 * internal-fn.c (expand_MASK_LOAD): Likewise.
3806 * tree-inline.c (copy_forbidden): Fail for
3807 __attribute__((optimize (0))) functions.
3810 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3811 is set, copy one less argument.
3812 (expand_simd_clones): Don't subtract clone_info->inbranch
3813 from simd_clone_struct_alloc argument.
3815 PR rtl-optimization/57915
3816 * recog.c (simplify_while_replacing): If all unary/binary/relational
3817 operation arguments are constant, attempt to simplify those.
3820 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3821 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3823 2014-02-04 Richard Biener <rguenther@suse.de>
3825 PR tree-optimization/60012
3826 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3827 TBAA disambiguation to all DDRs.
3829 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3832 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3833 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3835 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3838 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3840 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3842 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3843 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3845 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3848 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3849 to figure out targets of polymorphic calls with known decl.
3850 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3851 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3852 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3853 (get_polymorphic_call_info): ... here.
3854 (get_polymorphic_call_info_from_invariant): New function.
3856 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3858 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3859 lookup via vtable pointer; check for type consistency
3860 and turn inconsitent facts into UNREACHABLE.
3861 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3862 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3863 type inconsistent querries; return UNREACHABLE instead.
3865 2014-02-03 Richard Henderson <rth@twiddle.net>
3868 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3869 already processed this node.
3870 (normalize_one_pred_1): Pass along mark_set.
3871 (normalize_one_pred): Create and destroy a pointer_set_t.
3872 (normalize_one_pred_chain): Likewise.
3874 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3876 PR gcov-profile/58602
3877 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3879 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3882 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3883 -fno-devirtualize; try to devirtualize by the knowledge of
3884 virtual table pointer given by aggregate propagation.
3885 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3886 (ipa_print_node_jump_functions): Dump also offset that
3887 is relevant for polymorphic calls.
3888 (determine_known_aggregate_parts): Add arg_type parameter; use it
3889 instead of determining the type from pointer type.
3890 (ipa_compute_jump_functions_for_edge): Update call of
3891 determine_known_aggregate_parts.
3892 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3893 (gimple_get_virt_method_for_binfo): ... here; simplify using
3894 vtable_pointer_value_to_vtable.
3895 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3896 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3897 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3898 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3900 (vtable_pointer_value_to_binfo): ... here.
3901 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3903 2014-02-03 Teresa Johnson <tejohnson@google.com>
3905 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3906 redef of outer loop index variable.
3908 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3912 * doc/extend.texi (Function Attributes): Typo.
3914 2014-02-03 Cong Hou <congh@google.com>
3916 PR tree-optimization/60000
3917 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3918 if the vectorized statement is a store. A store statement can only
3919 appear at the end of pattern statements.
3921 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3923 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3924 (ix86_option_override_internal): Default long double to 64-bit for
3925 32-bit Bionic and to 128-bit for 64-bit Bionic.
3927 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3928 TARGET_LONG_DOUBLE_128 is true.
3929 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3931 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3932 (mlong-double-64): Negate -mlong-double-128.
3933 (mlong-double-128): New option.
3935 * config/i386/i386-c.c (ix86_target_macros): Define
3936 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3938 * doc/invoke.texi: Document -mlong-double-128.
3940 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3942 PR rtl-optimization/60024
3943 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3945 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3947 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3949 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3951 PR rtl-optimization/57662
3952 * sel-sched.c (code_motion_path_driver): Do not mark already not
3953 existing blocks in the visiting bitmap.
3955 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3957 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3958 on the insn being emitted.
3960 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3961 Will Deacon <will.deacon@arm.com>
3963 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3965 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3967 * config/arm/arm-tables.opt: Regenerate.
3969 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3971 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3972 for vector types other than V16QImode.
3973 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3974 define_expand, and call altivec_expand_vec_perm_le when producing
3975 code with little endian element order.
3976 (*altivec_vperm_<mode>_internal): New insn having previous
3977 behavior of altivec_vperm_<mode>.
3978 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3979 altivec_expand_vec_perm_le when producing code with little endian
3981 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3982 behavior of altivec_vperm_<mode>_uns.
3984 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3986 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3987 (altivec_vsumsws): Add handling for -maltivec=be with a little
3989 (altivec_vsumsws_direct): New.
3990 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3991 gen_altivec_vsumsws.
3993 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3995 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3996 vtable_pointer_value_to_binfo): New functions.
3997 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3998 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
4000 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
4002 * config/nios2/nios2.md (load_got_register): Initialize GOT
4003 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
4004 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
4006 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4008 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
4009 preserverd by passthrough, do not propagate the type.
4011 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4013 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
4014 (mips_atomic_assign_expand_fenv): New function.
4015 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
4017 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4019 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4020 (__builtin_mips_set_fcsr): Likewise.
4021 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4022 MIPS_USI_FTYPE_VOID.
4023 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4024 (mips16_expand_set_fcsr): Likewise.
4025 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4026 (mips16_set_fcsr_stub): Likewise.
4027 (mips16_get_fcsr_one_only_stub): New class.
4028 (mips16_set_fcsr_one_only_stub): Likewise.
4029 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4030 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4031 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4032 (hard_float): New availability predicate.
4033 (mips_builtins): Add get_fcsr and set_fcsr.
4034 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
4035 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
4036 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
4037 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
4038 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
4041 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4043 * config/mips/mips.c (mips_one_only_stub): New class.
4044 (mips_need_mips16_rdhwr_p): Replace with...
4045 (mips16_rdhwr_stub): ...this new variable.
4046 (mips16_stub_call_address): New function.
4047 (mips16_rdhwr_one_only_stub): New class.
4048 (mips_expand_thread_pointer): Use mips16_stub_call_address.
4049 (mips_output_mips16_rdhwr): Delete.
4050 (mips_finish_stub): New function.
4051 (mips_code_end): Use it to handle rdhwr stubs.
4053 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
4056 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
4057 when calculating size of integer atomic types.
4059 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
4061 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
4063 2014-02-01 Jakub Jelinek <jakub@redhat.com>
4065 PR tree-optimization/60003
4066 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
4067 * profile.c (branch_prob): Use gimple_call_builtin_p
4068 to check for BUILT_IN_SETJMP_RECEIVER.
4069 * tree-inline.c (copy_bb): Call notice_special_calls.
4071 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
4074 * lra-constraints.c (process_alt_operands): Update reload_sum only
4077 2014-01-31 Richard Henderson <rth@redhat.com>
4080 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4081 until after else_eh is processed.
4083 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4085 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4086 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4087 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4088 in smmintrin.h, remove them.
4089 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4090 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4091 * config/i386/i386.md (ROUND_SAE): Fix value.
4092 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4093 (const48_operand): New.
4094 * config/i386/subst.md (round), (round_expand): Use
4095 const_4_or_8_to_11_operand.
4096 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4098 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4100 * config/i386/constraints.md (Yk): Swap meaning with k.
4101 * config/i386/i386.md (movhi_internal): Change Yk to k.
4102 (movqi_internal): Ditto.
4103 (*k<logic><mode>): Ditto.
4106 (kandn<mode>): Ditto.
4107 (*<code>hi_1): Ditto.
4108 (*<code>qi_1): Ditto.
4109 (kxnor<mode>): Ditto.
4110 (kortestzhi): Ditto.
4111 (kortestchi): Ditto.
4113 (*one_cmplhi2_1): Ditto.
4114 (*one_cmplqi2_1): Ditto.
4115 * config/i386/sse.md (): Change k to Yk.
4116 (avx512f_load<mode>_mask): Ditto.
4117 (avx512f_blendm<mode>): Ditto.
4118 (avx512f_store<mode>_mask): Ditto.
4119 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4120 (avx512f_storedqu<mode>_mask): Ditto.
4121 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4123 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4124 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4125 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4126 (avx512f_maskcmp<mode>3): Ditto.
4127 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4128 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4129 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4130 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4131 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4132 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4133 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4134 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4135 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4136 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4137 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4138 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4139 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4140 (vec_extract_lo_<mode>_maskm): Ditto.
4141 (vec_extract_hi_<mode>_maskm): Ditto.
4142 (avx512f_vternlog<mode>_mask): Ditto.
4143 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4144 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4145 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4146 (avx512f_<code>v8div16qi2_mask): Ditto.
4147 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4148 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4149 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4150 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4151 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4152 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4153 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4154 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4155 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4156 (avx512cd_maskb_vec_dupv8di): Ditto.
4157 (avx512cd_maskw_vec_dupv16si): Ditto.
4158 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4159 (avx512f_vpermi2var<mode>3_mask): Ditto.
4160 (avx512f_vpermi2var<mode>3_mask): Ditto.
4161 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4162 (*avx512f_gathersi<mode>): Ditto.
4163 (*avx512f_gathersi<mode>_2): Ditto.
4164 (*avx512f_gatherdi<mode>): Ditto.
4165 (*avx512f_gatherdi<mode>_2): Ditto.
4166 (*avx512f_scattersi<mode>): Ditto.
4167 (*avx512f_scatterdi<mode>): Ditto.
4168 (avx512f_compress<mode>_mask): Ditto.
4169 (avx512f_compressstore<mode>_mask): Ditto.
4170 (avx512f_expand<mode>_mask): Ditto.
4171 * config/i386/subst.md (mask): Change k to Yk.
4172 (mask_scalar_merge): Ditto.
4175 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4177 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4179 2014-01-31 Richard Biener <rguenther@suse.de>
4182 * builtins.c (fold_builtin_memory_op): Make sure to not
4183 use a floating-point mode or a boolean or enumeral type for
4186 2014-01-30 DJ Delorie <dj@redhat.com>
4188 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4189 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4190 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4191 whenever main() has an epilogue.
4193 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4195 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4196 unused variable "field".
4197 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4198 (vsx_mergeh_<mode>): Likewise.
4199 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4200 (altivec_vmrghh): Likewise.
4201 (altivec_vmrghw): Likewise.
4202 (altivec_vmrglb): Likewise.
4203 (altivec_vmrglh): Likewise.
4204 (altivec_vmrglw): Likewise.
4205 (altivec_vspltb): Add missing uses.
4206 (altivec_vsplth): Likewise.
4207 (altivec_vspltw): Likewise.
4208 (altivec_vspltsf): Likewise.
4210 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4213 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4214 frame related instructions.
4216 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4218 PR rtl-optimization/59959
4219 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4220 any reload of register whose subreg is invalid.
4222 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4224 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4225 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4226 Add missing return type - void.
4228 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4230 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4231 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4232 remove element index adjustment for endian (now handled in vsx.md
4234 (altivec_expand_vec_perm_const): Use
4235 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4236 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4237 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4238 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4239 define_expand and a new define_insn *altivec_vspltb_internal;
4240 adjust for -maltivec=be on a little endian target.
4241 (altivec_vspltb_direct): New.
4242 (altivec_vsplth): Divide into a define_expand and a new
4243 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4244 little endian target.
4245 (altivec_vsplth_direct): New.
4246 (altivec_vspltw): Divide into a define_expand and a new
4247 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4248 little endian target.
4249 (altivec_vspltw_direct): New.
4250 (altivec_vspltsf): Divide into a define_expand and a new
4251 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4252 a little endian target.
4254 2014-01-30 Richard Biener <rguenther@suse.de>
4256 PR tree-optimization/59993
4257 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4258 can propagate form the earlier stmt and avoid the transform
4259 when the intermediate result is needed.
4261 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4263 * README.Portability: Fix typo.
4265 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4267 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4268 comparison_operator with ordered_comparison_operator.
4270 2014-01-30 Nick Clifton <nickc@redhat.com>
4272 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4273 Rename to mn10300_store_multiple_regs.
4274 * config/mn10300/mn10300.c: Likewise.
4275 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4276 store_multiple_regs.
4277 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4278 Call mn10300_store_multiple_regs.
4280 2014-01-30 Nick Clifton <nickc@redhat.com>
4281 DJ Delorie <dj@redhat.com>
4283 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4284 %fp 2 to keep registers after it properly word-aligned.
4285 (rl78_alloc_physical_registers_umul): Handle the case where both
4286 input operands are the same.
4288 2014-01-30 Richard Biener <rguenther@suse.de>
4290 PR tree-optimization/59903
4291 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4294 2014-01-30 Jason Merrill <jason@redhat.com>
4297 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4300 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4302 2014-01-30 Richard Biener <rguenther@suse.de>
4304 PR tree-optimization/59951
4305 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4307 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4310 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4311 SFmode to DFmode case.
4313 2014-01-29 DJ Delorie <dj@redhat.com>
4315 * config/msp430/msp430.opt (-minrt): New.
4316 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4318 (ENDFILE_SPEC): Likewise.
4320 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4322 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4323 (estimate_function_body_sizes): Use it.
4325 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4328 * dwarf2out.c (is_cxx_auto): New.
4329 (is_base_type): Use it.
4330 (gen_type_die_with_usage): Likewise.
4332 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4334 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4335 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4336 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4337 -maltivec=be with LE targets.
4338 (vsx_mergeh_<mode>): Likewise.
4339 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4340 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4341 (altivec_vmrghb): Replace with define_expand and new
4342 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4343 (altivec_vmrghb_direct): New define_insn.
4344 (altivec_vmrghh): Replace with define_expand and new
4345 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4346 (altivec_vmrghh_direct): New define_insn.
4347 (altivec_vmrghw): Replace with define_expand and new
4348 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4349 (altivec_vmrghw_direct): New define_insn.
4350 (*altivec_vmrghsf): Adjust for endianness.
4351 (altivec_vmrglb): Replace with define_expand and new
4352 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4353 (altivec_vmrglb_direct): New define_insn.
4354 (altivec_vmrglh): Replace with define_expand and new
4355 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4356 (altivec_vmrglh_direct): New define_insn.
4357 (altivec_vmrglw): Replace with define_expand and new
4358 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4359 (altivec_vmrglw_direct): New define_insn.
4360 (*altivec_vmrglsf): Adjust for endianness.
4361 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4362 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4363 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4364 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4365 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4366 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4367 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4368 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4370 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4372 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4373 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4376 2014-01-29 Richard Biener <rguenther@suse.de>
4378 PR tree-optimization/58742
4379 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4380 associate_pointerplus_align.
4381 (associate_pointerplus_diff): New function.
4382 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4383 and associate_pointerplus_diff.
4385 2014-01-29 Richard Biener <rguenther@suse.de>
4387 * lto-streamer.h (LTO_major_version): Bump to 3.
4388 (LTO_minor_version): Reset to 0.
4390 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4392 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4393 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4394 (arm_file_start): Generate correct asm header for armv7ve.
4395 * config/arm/bpabi.h: Add multilib support for armv7ve.
4396 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4397 and cortex-a15 to armv7ve.
4398 * config/arm/t-aprofile: Add multilib support for armv7ve.
4399 * doc/invoke.texi: Document -march=armv7ve.
4401 2014-01-29 Richard Biener <rguenther@suse.de>
4403 PR tree-optimization/58742
4404 * tree-ssa-forwprop.c (associate_plusminus): Return true
4405 if we changed sth, defer EH cleanup to ...
4406 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4407 (simplify_mult): New function.
4409 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4412 PR tree-optimization/59920
4413 * tree.c (build_common_builtin_nodes): Remove
4414 __builtin_setjmp_dispatcher initialization.
4415 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4416 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4417 instead of gsi_after_labels + manually skipping debug stmts.
4418 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4419 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4420 * tree-inline.c (copy_edges_for_bb): Remove
4421 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4422 argument. Ignore computed_goto_p stmts. Don't call
4423 make_abnormal_goto_edges. If a call might need abnormal edges
4424 for non-local gotos, see if it already has an edge to
4425 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4426 with true argument, don't do anything then, otherwise add
4427 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4428 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4430 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4431 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4432 (lower_stmt): Don't set data->calls_builtin_setjmp.
4433 (lower_builtin_setjmp): Adjust comment.
4434 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4435 * tree-cfg.c (found_computed_goto): Remove.
4436 (factor_computed_gotos): Remove.
4437 (make_goto_expr_edges): Return bool, true for computed gotos.
4438 Don't call make_abnormal_goto_edges.
4439 (build_gimple_cfg): Don't set found_computed_goto, don't call
4440 factor_computed_gotos.
4441 (computed_goto_p): No longer static.
4442 (make_blocks): Don't set found_computed_goto.
4443 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4444 (make_edges): If make_goto_expr_edges returns true, push bb
4445 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4446 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4447 vector. Record mapping between bbs and OpenMP regions if there
4448 are any, adjust make_gimple_omp_edges caller. Call
4449 handle_abnormal_edges.
4450 (make_abnormal_goto_edges): Remove.
4451 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4452 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4453 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4454 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4455 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4456 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4457 filling *region also set *region_idx to (*region)->entry->index.
4460 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4461 For REGs set ORIGINAL_REGNO.
4463 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4465 * doc/md.texi: Mention that a target shouldn't implement
4466 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4469 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4471 PR tree-optimization/59594
4472 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4473 a copy of the datarefs vector rather than the vector itself.
4475 2014-01-28 Jason Merrill <jason@redhat.com>
4478 * dwarf2out.c (auto_die): New static.
4479 (gen_type_die_with_usage): Handle C++1y 'auto'.
4480 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4483 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4486 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4487 (SPEC_X32): Likewise.
4488 (SPEC_64): Likewise.
4489 * config/i386/i386.c (ix86_option_override_internal): Turn off
4490 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4492 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4493 * config/i386/i386.h (TARGET_16BIT): New macro.
4494 (TARGET_16BIT_P): Likewise.
4495 * config/i386/i386.opt: Add m16.
4496 * doc/invoke.texi: Document -m16.
4498 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4500 PR preprocessor/59935
4501 * input.c (location_get_source_line): Bail out on when line number
4502 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4504 2014-01-28 Richard Biener <rguenther@suse.de>
4506 PR tree-optimization/58742
4507 * tree-ssa-forwprop.c (associate_plusminus): Handle
4508 pointer subtraction of the form (T)(P + A) - (T)P.
4510 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4512 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4515 2014-01-28 Richard Biener <rguenther@suse.de>
4518 2014-01-28 Richard Biener <rguenther@suse.de>
4520 PR rtl-optimization/45364
4521 PR rtl-optimization/59890
4522 * var-tracking.c (local_get_addr_clear_given_value): Handle
4523 already cleared slot.
4524 (val_reset): Handle not allocated local_get_addr_cache.
4525 (vt_find_locations): Use post-order on the inverted CFG.
4527 2014-01-28 Richard Biener <rguenther@suse.de>
4529 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4531 2014-01-28 Richard Biener <rguenther@suse.de>
4533 PR rtl-optimization/45364
4534 PR rtl-optimization/59890
4535 * var-tracking.c (local_get_addr_clear_given_value): Handle
4536 already cleared slot.
4537 (val_reset): Handle not allocated local_get_addr_cache.
4538 (vt_find_locations): Use post-order on the inverted CFG.
4540 2014-01-28 Alan Modra <amodra@gmail.com>
4542 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4543 * configure.ac <recursive call for build != host>: Define
4544 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4545 and LD_FOR_BUILD too.
4546 * configure: Regenerate.
4548 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4550 * config/i386/i386.c (get_builtin_code_for_version): Separate
4551 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4552 Broadwell from Haswell.
4554 2014-01-27 Steve Ellcey <sellcey@mips.com>
4556 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4557 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4558 * config/mips/mips.c (mips_option_override): Change setting
4560 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4561 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4562 Change from Mask to Var.
4564 2014-01-27 Jeff Law <law@redhat.com>
4566 * ipa-inline.c (inline_small_functions): Fix typo.
4568 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4570 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4571 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4572 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4573 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4574 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4575 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4576 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4577 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4578 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4579 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4580 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4581 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4582 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4583 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4584 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4585 (_mm512_storeu_epi64): Ditto.
4586 (_mm512_cmpge_epi32_mask): Ditto.
4587 (_mm512_cmpge_epu32_mask): Ditto.
4588 (_mm512_cmpge_epi64_mask): Ditto.
4589 (_mm512_cmpge_epu64_mask): Ditto.
4590 (_mm512_cmple_epi32_mask): Ditto.
4591 (_mm512_cmple_epu32_mask): Ditto.
4592 (_mm512_cmple_epi64_mask): Ditto.
4593 (_mm512_cmple_epu64_mask): Ditto.
4594 (_mm512_cmplt_epi32_mask): Ditto.
4595 (_mm512_cmplt_epu32_mask): Ditto.
4596 (_mm512_cmplt_epi64_mask): Ditto.
4597 (_mm512_cmplt_epu64_mask): Ditto.
4598 (_mm512_cmpneq_epi32_mask): Ditto.
4599 (_mm512_cmpneq_epu32_mask): Ditto.
4600 (_mm512_cmpneq_epi64_mask): Ditto.
4601 (_mm512_cmpneq_epu64_mask): Ditto.
4602 (_mm512_expand_pd): Ditto.
4603 (_mm512_expand_ps): Ditto.
4604 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4605 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4606 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4607 * config/i386/i386.c (ix86_builtins): Add
4608 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4609 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4610 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4611 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4612 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4613 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4614 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4615 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4616 IX86_BUILTIN_PMOVUSQW512_MEM.
4617 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4618 __builtin_ia32_pmovsqd512mem_mask,
4619 __builtin_ia32_pmovqd512mem_mask,
4620 __builtin_ia32_pmovusqw512mem_mask,
4621 __builtin_ia32_pmovsqw512mem_mask,
4622 __builtin_ia32_pmovqw512mem_mask,
4623 __builtin_ia32_pmovusdw512mem_mask,
4624 __builtin_ia32_pmovsdw512mem_mask,
4625 __builtin_ia32_pmovdw512mem_mask,
4626 __builtin_ia32_pmovqb512mem_mask,
4627 __builtin_ia32_pmovusqb512mem_mask,
4628 __builtin_ia32_pmovsqb512mem_mask,
4629 __builtin_ia32_pmovusdb512mem_mask,
4630 __builtin_ia32_pmovsdb512mem_mask,
4631 __builtin_ia32_pmovdb512mem_mask.
4632 (bdesc_args): Add __builtin_ia32_expanddf512,
4633 __builtin_ia32_expandsf512.
4634 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4635 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4636 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4637 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4638 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4639 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4640 (avx512f_<code>v8div16qi2_mask_store): This.
4641 (avx512f_expand<mode>): New.
4643 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4645 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4647 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4648 (_mm512_prefetch_i32scatter_pd): Ditto.
4649 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4650 (_mm512_prefetch_i64scatter_pd): Ditto.
4651 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4652 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4653 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4654 (_mm512_prefetch_i32scatter_ps): Ditto.
4655 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4656 (_mm512_prefetch_i64scatter_ps): Ditto.
4657 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4658 * config/i386/i386-builtin-types.def: Define
4659 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4660 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4661 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4662 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4663 IX86_BUILTIN_SCATTERPFQPD.
4664 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4665 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4666 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4667 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4668 __builtin_ia32_scatterpfqps.
4669 (ix86_expand_builtin): Expand new built-ins.
4670 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4671 fix memory access data type.
4672 (*avx512pf_gatherpf<mode>_mask): Ditto.
4673 (*avx512pf_gatherpf<mode>): Ditto.
4674 (avx512pf_scatterpf<mode>): Ditto.
4675 (*avx512pf_scatterpf<mode>_mask): Ditto.
4676 (*avx512pf_scatterpf<mode>): Ditto.
4677 (GATHER_SCATTER_SF_MEM_MODE): New.
4678 (avx512pf_gatherpf<mode>df): Ditto.
4679 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4680 (*avx512pf_scatterpf<mode>df): Ditto.
4682 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4685 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4686 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4689 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4691 * common/config/arm/arm-common.c
4692 (arm_rewrite_mcpu): Handle multiple names.
4694 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4696 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4698 * gimple-builder.h (create_gimple_tmp): Delete.
4700 2014-01-27 Christian Bruel <christian.bruel@st.com>
4702 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4705 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4707 * config/pa/pa.md (call): Generate indirect long calls to non-local
4708 functions when outputing 32-bit code.
4709 (call_value): Likewise except for special call to buggy powf function.
4711 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4712 portable runtime and PIC indirect calls.
4713 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4714 and PIC call sequences. Use ldo instead of blr to set return register
4715 in PIC call sequence.
4717 2014-01-25 Walter Lee <walt@tilera.com>
4719 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4720 avoid clobbering a live register.
4722 2014-01-25 Walter Lee <walt@tilera.com>
4724 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4725 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4726 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4727 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4729 2014-01-25 Walter Lee <walt@tilera.com>
4731 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4732 arguments on even registers.
4733 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4735 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4736 (BIGGEST_ALIGNMENT): Ditto.
4737 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4739 2014-01-25 Walter Lee <walt@tilera.com>
4741 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4742 insns before bundling.
4743 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4745 2014-01-25 Walter Lee <walt@tilera.com>
4747 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4748 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4749 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4751 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4753 * config/mips/constraints.md (kl): Delete.
4754 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4755 define expands, using...
4756 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4757 instructions for MIPS16.
4758 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4759 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4761 2014-01-25 Walter Lee <walt@tilera.com>
4763 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4767 2014-01-25 Walter Lee <walt@tilera.com>
4769 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4770 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4772 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4774 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4777 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4779 * print-rtl.c (in_call_function_usage): New var.
4780 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4781 EXPR_LIST mode as mode and not as reg note name.
4784 * cfgloopmanip.c (copy_loop_info): If
4785 loop->warned_aggressive_loop_optimizations, make sure
4786 the flag is set in target loop too.
4788 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4790 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4792 * builtins.def: Likewise.
4793 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4794 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4795 * ira.c (ira_setup_eliminable_regset): Likewise.
4796 * omp-low.c (gate_expand_omp): Likewise.
4797 (execute_lower_omp): Likewise.
4798 (diagnose_sb_0): Likewise.
4799 (gate_diagnose_omp_blocks): Likewise.
4800 (simd_clone_clauses_extract): Likewise.
4803 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4805 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4806 correction for little endian...
4807 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4810 2014-01-24 Jeff Law <law@redhat.com>
4812 PR tree-optimization/59919
4813 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4814 for non-returning calls.
4816 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4818 * common/config/aarch64/aarch64-common.c
4819 (aarch64_rewrite_mcpu): Handle multiple names.
4820 * config/aarch64/aarch64.h
4821 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4823 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4825 * input.c (add_file_to_cache_tab): Handle the case where fopen
4828 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4831 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4832 from push operand if code of push isn't PRE_DEC.
4834 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4837 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4838 -mquad-memory-atomic. Update -mquad-memory documentation to say
4839 it is only used for non-atomic loads/stores.
4841 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4842 -mquad-memory or -mquad-memory-atomic switches.
4844 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4845 -mquad-memory-atomic to ISA 2.07 support.
4847 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4848 to separate support of normal quad word memory operations (ldq, stq)
4849 from the atomic quad word memory operations.
4851 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4852 support to separate non-atomic quad word operations from atomic
4853 quad word operations. Disable non-atomic quad word operations in
4854 little endian mode so that we don't have to swap words after the
4855 load and before the store.
4856 (quad_load_store_p): Add comment about atomic quad word support.
4857 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4858 options printed with -mdebug=reg.
4860 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4861 -mquad-memory-atomic as the test for whether we have quad word
4862 atomic instructions.
4863 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4864 or -mp8-vector are used, allow byte/half-word atomic operations.
4866 * config/rs6000/sync.md (load_lockedti): Insure that the address
4867 is a proper indexed or indirect address for the lqarx instruction.
4868 On little endian systems, swap the hi/lo registers after the lqarx
4870 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4871 insure the address is valid for the lqarx instruction.
4872 (store_conditionalti): Insure that the address is a proper indexed
4873 or indirect address for the stqcrx. instruction. On little endian
4874 systems, swap the hi/lo registers before doing the stqcrx.
4876 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4877 insure the address is valid for the stqcrx. instruction.
4879 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4880 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4881 type of quad memory support is available.
4883 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4886 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4887 there is a danger of looping.
4889 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4891 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4892 force flag_ira_loop_pressure if set via command line.
4894 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4896 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4897 (ashr_simd): New builtin handling DI mode.
4898 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4899 (aarch64_sshr_simddi): New match pattern.
4900 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4901 (vshrd_n_s64): Likewise.
4902 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4904 2014-01-23 Nick Clifton <nickc@redhat.com>
4906 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4907 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4908 favour of mcu specific scripts.
4909 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4912 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4913 Alex Velenko <Alex.Velenko@arm.com>
4915 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4916 (vaddv_s16): Likewise.
4917 (vaddv_s32): Likewise.
4918 (vaddv_u8): Likewise.
4919 (vaddv_u16): Likewise.
4920 (vaddv_u32): Likewise.
4921 (vaddvq_s8): Likewise.
4922 (vaddvq_s16): Likewise.
4923 (vaddvq_s32): Likewise.
4924 (vaddvq_s64): Likewise.
4925 (vaddvq_u8): Likewise.
4926 (vaddvq_u16): Likewise.
4927 (vaddvq_u32): Likewise.
4928 (vaddvq_u64): Likewise.
4929 (vaddv_f32): Likewise.
4930 (vaddvq_f32): Likewise.
4931 (vaddvq_f64): Likewise.
4932 (vmaxv_f32): Likewise.
4933 (vmaxv_s8): Likewise.
4934 (vmaxv_s16): Likewise.
4935 (vmaxv_s32): Likewise.
4936 (vmaxv_u8): Likewise.
4937 (vmaxv_u16): Likewise.
4938 (vmaxv_u32): Likewise.
4939 (vmaxvq_f32): Likewise.
4940 (vmaxvq_f64): Likewise.
4941 (vmaxvq_s8): Likewise.
4942 (vmaxvq_s16): Likewise.
4943 (vmaxvq_s32): Likewise.
4944 (vmaxvq_u8): Likewise.
4945 (vmaxvq_u16): Likewise.
4946 (vmaxvq_u32): Likewise.
4947 (vmaxnmv_f32): Likewise.
4948 (vmaxnmvq_f32): Likewise.
4949 (vmaxnmvq_f64): Likewise.
4950 (vminv_f32): Likewise.
4951 (vminv_s8): Likewise.
4952 (vminv_s16): Likewise.
4953 (vminv_s32): Likewise.
4954 (vminv_u8): Likewise.
4955 (vminv_u16): Likewise.
4956 (vminv_u32): Likewise.
4957 (vminvq_f32): Likewise.
4958 (vminvq_f64): Likewise.
4959 (vminvq_s8): Likewise.
4960 (vminvq_s16): Likewise.
4961 (vminvq_s32): Likewise.
4962 (vminvq_u8): Likewise.
4963 (vminvq_u16): Likewise.
4964 (vminvq_u32): Likewise.
4965 (vminnmv_f32): Likewise.
4966 (vminnmvq_f32): Likewise.
4967 (vminnmvq_f64): Likewise.
4969 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4971 * config/aarch64/aarch64-simd.md
4972 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4973 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4974 (*aarch64_mul3_elt<mode>): Likewise.
4975 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4976 (*aarch64_mul3_elt_to_64v2df): Likewise.
4977 (*aarch64_mla_elt<mode>): Likewise.
4978 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4979 (*aarch64_mls_elt<mode>): Likewise.
4980 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4981 (*aarch64_fma4_elt<mode>): Likewise.
4982 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4983 (*aarch64_fma4_elt_to_64v2df): Likewise.
4984 (*aarch64_fnma4_elt<mode>): Likewise.
4985 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4986 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4987 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4988 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4989 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4990 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4991 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4992 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4993 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4995 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4997 * config/aarch64/aarch64-simd.md
4998 (aarch64_be_checked_get_lane<mode>): New define_expand.
4999 * config/aarch64/aarch64-simd-builtins.def
5000 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
5001 New builtin definition.
5002 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
5003 Use new safe be builtin.
5005 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5007 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
5009 (aarch64_be_st1<mode>): Likewise.
5010 (aarch_ld1<VALL:mode>): Define_expand modified.
5011 (aarch_st1<VALL:mode>): Likewise.
5012 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
5013 (UNSPEC_ST1): Likewise.
5015 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
5017 * config/microblaze/microblaze.md: Add trap insn and attribute
5019 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5021 PR preprocessor/58580
5022 * input.h (location_get_source_line): Take an additional line_size
5024 (void diagnostics_file_cache_fini): Declare new function.
5025 * input.c (struct fcache): New type.
5026 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5027 New static constants.
5028 (diagnostic_file_cache_init, total_lines_num)
5029 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5030 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5031 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5032 (get_next_line, read_next_line, goto_next_line, read_line_num):
5033 New static function definitions.
5034 (diagnostic_file_cache_fini): New function.
5035 (location_get_source_line): Take an additional output line_len
5036 parameter. Re-write using lookup_or_add_file_to_cache_tab and
5038 * diagnostic.c (diagnostic_finish): Call
5039 diagnostic_file_cache_fini.
5040 (adjust_line): Take an additional input parameter for the length
5041 of the line, rather than calculating it with strlen.
5042 (diagnostic_show_locus): Adjust the use of
5043 location_get_source_line and adjust_line with respect to their new
5044 signature. While displaying a line now, do not stop at the first
5045 null byte. Rather, display the zero byte as a space and keep
5046 going until we reach the size of the line.
5047 * Makefile.in: Add vec.o to OBJS-libcommon
5049 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5050 Ilya Tocar <ilya.tocar@intel.com>
5052 * config/i386/avx512fintrin.h (_mm512_kmov): New.
5053 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
5054 (__builtin_ia32_kmov16): Ditto.
5055 * config/i386/i386.md (UNSPEC_KMOV): New.
5058 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5060 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
5061 (_mm512_storeu_si512): Ditto.
5063 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
5066 * rtl.h (get_referenced_operands): Declare.
5067 * recog.c (get_referenced_operands): New function.
5068 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
5069 operands have been referenced when recording LO_SUM references.
5071 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
5073 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
5075 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5077 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5078 Enable for generic and recent AMD targets.
5080 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5082 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5083 ARG_SIZE note when adjustment was eliminated.
5085 2014-01-22 Jeff Law <law@redhat.com>
5087 PR tree-optimization/59597
5088 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5089 in file. Accept new argument REGISTERING and use it to modify
5090 dump output appropriately.
5091 (register_jump_thread): Corresponding changes.
5092 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5093 thread paths involving joiner blocks. Add code to dump cancelled
5094 jump threading paths.
5096 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5098 PR rtl-optimization/59477
5099 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5100 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5102 2014-01-22 Tom Tromey <tromey@redhat.com>
5104 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5106 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5108 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5110 PR rtl-optimization/59896
5111 * lra-constraints.c (process_alt_operands): Check unused note for
5112 matched operands of insn with no output reloads.
5114 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5116 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5117 (mips_move_from_gpr_cost): Likewise.
5119 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5121 PR rtl-optimization/59858
5122 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5123 ira_class_hard_regs_num.
5124 (process_alt_operands): Increase reject for dying matched operand.
5126 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5129 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5130 smaller than size, perform several stores or loads and stores
5131 at dst + count - size to store or copy all of size bytes, rather
5132 than just last modesize bytes.
5134 2014-01-20 DJ Delorie <dj@redhat.com>
5136 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5137 that CLOBBERs are REGs before propogating their values.
5139 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5142 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5143 (cgraph_inline_failed_type): New function.
5144 * cgraph.h (DEFCIFCODE): Add type.
5145 (cgraph_inline_failed_type_t): New enum.
5146 (cgraph_inline_failed_type): New prototype.
5147 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5148 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5149 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5150 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5151 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5152 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5153 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5154 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5155 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5156 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5157 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5158 OPTIMIZATION_MISMATCH.
5159 * tree-inline.c (expand_call_inline): Emit errors during
5160 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5162 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5165 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5166 mode attribute in insn output.
5168 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5170 * output.h (output_constant): Delete.
5171 * varasm.c (output_constant): Make private.
5173 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5175 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5177 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5180 * tree.h (fold_builtin_strcat): New prototype.
5181 * builtins.c (fold_builtin_strcat): No longer static. Add len
5182 argument, if non-NULL, don't call c_strlen. Optimize
5183 directly into __builtin_memcpy instead of __builtin_strcpy.
5184 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5185 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5187 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5189 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5190 for SImode_address_operand operands, having only a REG argument.
5192 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5194 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5195 loader name using mbig-endian.
5196 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5198 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5200 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5204 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5206 * config/aarch64/aarch64-protos.h
5207 (aarch64_cannot_change_mode_class_ptr): Declare.
5208 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5209 aarch64_cannot_change_mode_class_ptr): New.
5210 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5211 backend hook aarch64_cannot_change_mode_class.
5213 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5215 * common/config/aarch64/aarch64-common.c
5216 (aarch64_handle_option): Don't handle any option order logic here.
5217 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5218 selected_cpu, warn on architecture version mismatch.
5219 (aarch64_override_options): Fix parsing order for option strings.
5221 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5222 Iain Sandoe <iain@codesourcery.com>
5225 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5226 warning. Amend comment to reflect current functionality.
5228 2014-01-20 Richard Biener <rguenther@suse.de>
5231 * builtins.c (fold_builtin_strcat): Remove case better handled
5232 by tree-ssa-strlen.c.
5234 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5236 * config/aarch64/aarch64.opt
5237 (mcpu, march, mtune): Make case-insensitive.
5239 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5242 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5243 if operands[1] is a REG or ZERO_EXTEND of a REG.
5245 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5247 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5249 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5251 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5252 long non-pic millicode calls.
5254 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5256 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5258 2014-01-19 Kito Cheng <kito@0xlab.org>
5260 * builtins.c (expand_movstr): Check movstr expand done or fail.
5262 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5263 H.J. Lu <hongjiu.lu@intel.com>
5266 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5267 to DImode for zero-extended addresses.
5269 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5271 PR rtl-optimization/57763
5272 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5273 on the new indirect jump_insn and increment LABEL_NUSES (label).
5275 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5279 * config.gcc (x86_archs): New variable.
5280 (x86_64_archs): Likewise.
5281 (x86_cpus): Likewise.
5282 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5283 --with-arch/--with-cpu= options.
5284 Support --with-arch=/--with-cpu={nehalem,westmere,
5285 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5287 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5289 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5290 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5292 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5294 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5296 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5299 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5300 clear cpp_get_options (parse_in)->warn_unused_macros for
5301 ix86_target_macros_internal with cpp_define.
5303 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5305 * jump.c (delete_related_insns): Keep (use (insn))s.
5306 * reorg.c (redundant_insn): Check for barriers too.
5308 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5310 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5312 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5314 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5315 call to $$dyncall when TARGET_LONG_CALLS is true.
5317 2014-01-17 Jeff Law <law@redhat.com>
5319 * ree.c (combine_set_extension): Temporarily disable test for
5320 changing number of hard registers.
5322 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5325 * ipa-inline-analysis.c (inline_free_summary):
5326 Do not free summary of aliases.
5328 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5331 * gimplify.c (gimplify_expr): Use create_tmp_var
5332 instead of create_tmp_var_raw. If cond doesn't have
5333 integral type, don't add the IFN_ANNOTATE builtin at all.
5335 2014-01-17 Martin Jambor <mjambor@suse.cz>
5338 * ipa-cp.c (prev_edge_clone): New variable.
5339 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5340 Also resize prev_edge_clone vector.
5341 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5342 (ipcp_edge_removal_hook): New function.
5343 (ipcp_driver): Register ipcp_edge_removal_hook.
5345 2014-01-17 Andrew Pinski <apinski@cavium.com>
5346 Steve Ellcey <sellcey@mips.com>
5349 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5352 2014-01-17 Jeff Law <law@redhat.com>
5355 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5356 so that pass_ccp runs first.
5358 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5360 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5361 (ix86_adjust_cost): Use !TARGET_XXX.
5362 (do_reorder_for_imul): Likewise.
5363 (swap_top_of_ready_list): Likewise.
5364 (ix86_sched_reorder): Likewise.
5366 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5368 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5369 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5370 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5371 (intel_memset): New. Duplicate slm_memset.
5372 (intel_cost): New. Duplicate slm_cost.
5373 (m_INTEL): New macro.
5374 (processor_target_table): Add "intel".
5375 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5376 with PROCESSOR_INTEL for "intel".
5377 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5378 PROCESSOR_SILVERMONT.
5379 (ix86_issue_rate): Likewise.
5380 (ix86_adjust_cost): Likewise.
5381 (ia32_multipass_dfa_lookahead): Likewise.
5382 (swap_top_of_ready_list): Likewise.
5383 (ix86_sched_reorder): Likewise.
5384 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5385 instead of TARGET_OPT_AGU.
5386 * config/i386/i386.h (TARGET_INTEL): New.
5387 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5388 (processor_type): Add PROCESSOR_INTEL.
5389 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5390 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5392 2014-01-17 Marek Polacek <polacek@redhat.com>
5395 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5398 2014-01-17 Richard Biener <rguenther@suse.de>
5400 PR tree-optimization/46590
5401 * opts.c (default_options_table): Add entries for
5402 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5403 all enabled at -O1 but not for -Og.
5404 * common.opt (fbranch-count-reg): Remove Init(1).
5405 (fmove-loop-invariants): Likewise.
5406 (ftree-pta): Likewise.
5408 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5410 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5411 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5412 decls to at least the GCC 4.8 used alignments.
5415 * tree-nested.c (convert_nonlocal_reference_stmt,
5416 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5417 of GIMPLE_BIND stmts, adjust associated decls.
5419 2014-01-17 Richard Biener <rguenther@suse.de>
5421 PR tree-optimization/46590
5422 * vec.h (vec<>::bseach): New member function implementing
5423 binary search according to C89 bsearch.
5424 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5425 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5426 bitmap pointer again. Make accesses_in_loop a flat array.
5427 (mem_ref_obstack): New global.
5428 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5429 (mark_ref_stored): Likewise.
5430 (ref_indep_loop_p_2): Likewise.
5431 (set_ref_stored_in_loop): New helper function.
5432 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5433 (memref_free): Adjust.
5434 (record_mem_ref_loc): Simplify.
5435 (gather_mem_refs_stmt): Adjust.
5436 (sort_locs_in_loop_postorder_cmp): New function.
5437 (analyze_memory_references): Sort accesses_in_loop after
5438 loop postorder number.
5439 (find_ref_loc_in_loop_cmp): New function.
5440 (for_all_locs_in_loop): Find relevant cluster of locs in
5441 accesses_in_loop and iterate without recursion.
5442 (execute_sm): Avoid uninit warning.
5443 (struct ref_always_accessed): Simplify.
5444 (ref_always_accessed::operator ()): Likewise.
5445 (ref_always_accessed_p): Likewise.
5446 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5447 loop postorder numbers here.
5448 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5451 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5454 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5455 on decls for which assemble_alias has been called.
5457 2014-01-17 Nick Clifton <nickc@redhat.com>
5459 * config/msp430/msp430.opt: (mcpu): New option.
5460 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5461 (msp430_option_override): Parse target_cpu. If the MCU name
5462 matches a generic string, clear target_mcu.
5463 (msp430_attr): Allow numeric interrupt values up to 63.
5464 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5465 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5467 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5469 * config/msp430/msp430.md (popm): Use %J rather than %I.
5470 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5471 (addhi_cy_i): Use immediate_operand for operand 2.
5472 * doc/invoke.texi: Document -mcpu option.
5474 2014-01-17 Richard Biener <rguenther@suse.de>
5476 PR rtl-optimization/38518
5477 * df.h (df_analyze_loop): Declare.
5478 * df-core.c: Include cfgloop.h.
5479 (df_analyze_1): Split out main part of df_analyze.
5480 (df_analyze): Adjust.
5481 (loop_inverted_post_order_compute): New function.
5482 (loop_post_order_compute): Likewise.
5483 (df_analyze_loop): New function avoiding whole-function
5485 * loop-invariant.c (find_defs): Use df_analyze_loop.
5486 (find_invariants): Adjust.
5487 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5489 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5491 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5492 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5494 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5496 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5497 traversal when removing references.
5499 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5502 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5504 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5507 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5508 pushing a reload for an autoinc when we had previously reloaded an
5509 inner part of the address.
5511 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5513 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5515 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5516 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5517 when not giving up or versioning for alias only because of
5519 (vect_analyze_data_ref_dependences): Set to true.
5520 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5522 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5523 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5527 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5530 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5531 operand 0 predicate for gathers, use a new pseudo as subtarget.
5533 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5536 * lra-constraints.c (process_alt_operands): Add printing debug info.
5537 Check absence of input/output reloads for matched operands too.
5539 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5541 PR rtl-optimization/59835
5542 * ira.c (ira_init_register_move_cost): Increase cost for
5545 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5547 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5549 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5552 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5553 non-register objects. Use gen_(high/low)part more consistently.
5556 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5559 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5560 endian support, remove tests for WORDS_BIG_ENDIAN.
5561 (p8_mfvsrd_3_<mode>): Likewise.
5562 (reload_gpr_from_vsx<mode>): Likewise.
5563 (reload_gpr_from_vsxsf): Likewise.
5564 (p8_mfvsrd_4_disf): Likewise.
5566 2014-01-16 Richard Biener <rguenther@suse.de>
5568 PR rtl-optimization/46590
5569 * lcm.c (compute_antinout_edge): Use postorder iteration.
5570 (compute_laterin): Use inverted postorder iteration.
5572 2014-01-16 Nick Clifton <nickc@redhat.com>
5575 * varasm.c (output_constant): Return the number of bytes actually
5577 (output_constructor_array_range): Update the field size with the
5578 number of bytes emitted by output_constant.
5579 (output_constructor_regular_field): Likewise. Also do not
5580 complain if the total number of bytes emitted is now greater
5581 than the expected fieldpos.
5582 * output.h (output_constant): Update prototype and descriptive comment.
5584 2014-01-16 Marek Polacek <polacek@redhat.com>
5587 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5588 it is error_mark_node.
5590 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5592 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5593 VALID_AVX256_REG_OR_OI_MODE.
5595 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5597 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5598 current procedure should be profiled.
5600 2014-01-15 Andrew Pinski <apinski@cavium.com>
5602 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5603 of moving from/to the STACK_REG register class.
5605 2014-01-15 Richard Henderson <rth@redhat.com>
5608 * reginfo.c (global_regs_decl): Globalize.
5609 * rtl.h (global_regs_decl): Declare.
5610 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5611 reserved via global_regs.
5613 2014-01-15 Teresa Johnson <tejohnson@google.com>
5615 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5617 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5619 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5620 and vmulosh rather than call gen_vec_widen_smult_*.
5621 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5622 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5623 (vec_widen_smult_even_v16qi): Likewise.
5624 (vec_widen_umult_even_v8hi): Likewise.
5625 (vec_widen_smult_even_v8hi): Likewise.
5626 (vec_widen_umult_odd_v16qi): Likewise.
5627 (vec_widen_smult_odd_v16qi): Likewise.
5628 (vec_widen_umult_odd_v8hi): Likewise.
5629 (vec_widen_smult_odd_v8hi): Likewise.
5630 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5631 vmuloub rather than call gen_vec_widen_umult_*.
5632 (vec_widen_umult_lo_v16qi): Likewise.
5633 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5634 vmulosb rather than call gen_vec_widen_smult_*.
5635 (vec_widen_smult_lo_v16qi): Likewise.
5636 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5637 rather than call gen_vec_widen_umult_*.
5638 (vec_widen_umult_lo_v8hi): Likewise.
5639 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5640 rather than call gen_vec_widen_smult_*.
5641 (vec_widen_smult_lo_v8hi): Likewise.
5643 2014-01-15 Jeff Law <law@redhat.com>
5645 PR tree-optimization/59747
5646 * ree.c (find_and_remove_re): Properly handle case where a second
5647 eliminated extension requires widening a copy created for elimination
5648 of a prior extension.
5649 (combine_set_extension): Ensure that the number of hard regs needed
5650 for a destination register does not change when we widen it.
5652 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5654 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5655 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5656 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5657 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5658 (avr-*-rtems*): Likewise.
5659 (bfin*-rtems*): Likewise.
5660 (moxie-*-rtems*): Likewise.
5661 (h8300-*-rtems*): Likewise.
5662 (i[34567]86-*-rtems*): Likewise.
5663 (lm32-*-rtems*): Likewise.
5664 (m32r-*-rtems*): Likewise.
5665 (m68k-*-rtems*): Likewise.
5666 (microblaze*-*-rtems*): Likewise.
5667 (mips*-*-rtems*): Likewise.
5668 (powerpc-*-rtems*): Likewise.
5669 (sh-*-rtems*): Likewise.
5670 (sparc-*-rtems*): Likewise.
5671 (sparc64-*-rtems*): Likewise.
5672 (v850-*-rtems*): Likewise.
5673 (m32c-*-rtems*): Likewise.
5675 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5677 PR rtl-optimization/59511
5678 * ira.c (ira_init_register_move_cost): Use memory costs for some
5679 cases of register move cost calculations.
5680 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5681 instead of BB frequency.
5682 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5683 * lra-assigns.c (find_hard_regno_for): Ditto.
5685 2014-01-15 Richard Biener <rguenther@suse.de>
5687 PR tree-optimization/59822
5688 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5689 (vectorizable_load): Use it to hoist defs of uses of invariant
5690 loads out of the loop.
5692 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5693 Kugan Vivekanandarajah <kuganv@linaro.org>
5696 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5699 2014-01-15 Richard Biener <rguenther@suse.de>
5701 PR rtl-optimization/59802
5702 * lcm.c (compute_available): Use inverted postorder to seed
5703 the initial worklist.
5705 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5708 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5709 ADDR_REGS for invalid symrefs in non-PIC code.
5711 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5714 * builtins.c (determine_block_size): Initialize *probable_max_size
5715 even if len_rtx is CONST_INT.
5717 2014-01-14 Andrew Pinski <apinski@cavium.com>
5719 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5720 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5721 (cortexa53_tunings): Likewise.
5722 (aarch64_sched_issue_rate): New function.
5723 (TARGET_SCHED_ISSUE_RATE): Define.
5725 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5727 * ira-costs.c (find_costs_and_classes): Add missed
5728 ira_init_register_move_cost_if_necessary.
5730 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5733 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5735 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5738 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5739 to indicate if type is used for function return value. Warn ABI
5740 change if the vector mode isn't available for function return value.
5741 (ix86_function_arg_advance): Pass false to type_natural_mode.
5742 (ix86_function_arg): Likewise.
5743 (ix86_gimplify_va_arg): Likewise.
5744 (function_arg_32): Don't warn ABI change.
5745 (ix86_function_value): Pass true to type_natural_mode.
5746 (ix86_return_in_memory): Likewise.
5747 (ix86_struct_value_rtx): Removed.
5748 (TARGET_STRUCT_VALUE_RTX): Likewise.
5750 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5752 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5753 converting a conditional jump into a conditional return.
5755 2014-01-14 Richard Biener <rguenther@suse.de>
5757 PR tree-optimization/58921
5758 PR tree-optimization/59006
5759 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5760 hoisting invariant stmts.
5761 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5762 invariant loads on the preheader edge if possible.
5764 2014-01-14 Joey Ye <joey.ye@arm.com>
5766 * doc/plugin.texi (Building GCC plugins): Update to C++.
5768 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5770 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5771 (_mm_rcp28_round_ss): Ditto.
5772 (_mm_rsqrt28_round_sd): Ditto.
5773 (_mm_rsqrt28_round_ss): Ditto.
5774 (_mm_rcp28_sd): Ditto.
5775 (_mm_rcp28_ss): Ditto.
5776 (_mm_rsqrt28_sd): Ditto.
5777 (_mm_rsqrt28_ss): Ditto.
5778 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5779 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5780 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5781 (IX86_BUILTIN_RCP28SD): Ditto.
5782 (IX86_BUILTIN_RCP28SS): Ditto.
5783 (IX86_BUILTIN_RSQRT28SD): Ditto.
5784 (IX86_BUILTIN_RSQRT28SS): Ditto.
5785 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5786 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5787 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5788 (ix86_expand_special_args_builtin): Expand new FTYPE.
5789 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5790 (srcp14<mode>): Make insn unary.
5791 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5792 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5793 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5794 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5795 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5796 Fix rounding: make it SAE only.
5797 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5799 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5801 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5802 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5803 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5804 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5805 (round_saeonly_mask_scalar_operand4): Ditto.
5806 (round_saeonly_mask_scalar_op3): Ditto.
5807 (round_saeonly_mask_scalar_op4): Ditto.
5809 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5811 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5812 Implement -maltivec=be for vec_insert and vec_extract.
5814 2014-01-10 DJ Delorie <dj@redhat.com>
5816 * config/msp430/msp430.md (call_internal): Don't allow memory
5817 references with SP as the base register.
5818 (call_value_internal): Likewise.
5819 * config/msp430/constraints.md (Yc): New. For memory references
5820 that don't use SP as a base register.
5822 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5823 "an integer without a # prefix"
5824 * config/msp430/msp430.md (epilogue_helper): Use it.
5826 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5829 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5830 AVX512F gather builtins.
5831 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5832 on gather decls with INTEGER_TYPE masktype.
5833 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5834 directly into the builtin rather than hoisting it before loop.
5836 PR tree-optimization/59387
5837 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5838 (scev_const_prop): If folded_casts and type has undefined overflow,
5839 use force_gimple_operand instead of force_gimple_operand_gsi and
5840 for each added stmt if it is assign with
5841 arith_code_with_undefined_signed_overflow, call
5842 rewrite_to_defined_overflow.
5843 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5844 gimple-fold.h instead.
5845 (arith_code_with_undefined_signed_overflow,
5846 rewrite_to_defined_overflow): Moved to ...
5847 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5848 rewrite_to_defined_overflow): ... here. No longer static.
5849 Include gimplify-me.h.
5850 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5851 rewrite_to_defined_overflow): New prototypes.
5853 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5855 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5857 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5859 * builtins.c (get_object_alignment_2): Minor tweak.
5860 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5862 2014-01-13 Christian Bruel <christian.bruel@st.com>
5864 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5865 optimized non constant lengths.
5867 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5870 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5871 load as __atomic_load_N if possible.
5873 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5875 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5877 (rs6000_expand_builtin): Adjust call.
5879 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5882 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5883 * config/rs6000/rs6000.c: Include target-globals.h.
5884 (rs6000_set_current_function): Instead of doing target_reinit
5885 unconditionally, use save_target_globals_default_opts and
5886 restore_target_globals.
5888 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5890 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5891 (rs6000_expand_builtin): Handle mffs and mtfsf.
5892 (rs6000_init_builtins): Define mffs and mtfsf.
5893 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5894 (rs6000_mffs): New pattern.
5895 (rs6000_mtfsf): New pattern.
5897 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5899 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5900 Start narrowing with START. Apply candidate-use pair
5901 and check overall cost in narrowing.
5902 (iv_ca_prune): Pass new argument.
5904 2014-01-10 Jeff Law <law@redhat.com>
5907 * ree.c (combine_reaching_defs): Ensure the defining statement
5908 occurs before the extension when optimizing extensions with
5909 different source and destination hard registers.
5911 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5914 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5915 vtables into the type inheritance graph.
5917 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5919 PR rtl-optimization/59754
5920 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5921 modes in the REGNO != REGNO case.
5923 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5925 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5927 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5929 PR tree-optimization/59745
5930 * tree-predcom.c (tree_predictive_commoning_loop): Call
5931 free_affine_expand_cache if giving up because components is NULL.
5933 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5934 GC in payload of target_globals struct instead of allocating them on
5935 the heap and the larger structs separately using GC.
5936 * target-globals.h (struct target_globals): Make regs, hard_regs,
5937 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5938 of GTY((skip)) and change type to void *.
5939 (reset_target_globals): Cast loads from those fields to corresponding
5942 2014-01-10 Steve Ellcey <sellcey@mips.com>
5945 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5946 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5947 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5949 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5952 * aarch64-modes.def (CC_Zmode): New flags mode.
5953 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5954 represents an equality.
5955 (aarch64_get_condition_code): Handle CC_Zmode.
5956 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5958 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5960 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5961 extraction in good case.
5963 2014-01-10 Richard Biener <rguenther@suse.de>
5965 PR tree-optimization/59374
5966 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5967 checking after SLP discovery. Mark stmts not participating
5968 in any SLP instance properly.
5970 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5972 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5973 when handling a SET rtx.
5975 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5977 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5978 (cortex-a57): Likewise.
5979 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5981 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5983 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5984 non-iwmmxt builtins.
5986 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5990 * ipa-devirt.c record_target_from_binfo): Take as argument
5991 stack of binfos and lookup matching one for virtual inheritance.
5992 (possible_polymorphic_call_targets_1): Update.
5994 2014-01-10 Huacai Chen <chenhc@lemote.com>
5996 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5997 kernel strings for Loongson-2E/2F/3A.
5999 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6002 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
6003 is_gimple_call before calling gimple_call_internal_p.
6005 2014-01-09 Steve Ellcey <sellcey@mips.com>
6007 * Makefile.in (TREE_FLOW_H): Remove.
6008 (TREE_SSA_H): Add file names from tree-flow.h.
6009 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
6010 * tree.h: Remove tree-flow.h reference.
6011 * hash-table.h: Remove tree-flow.h reference.
6012 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
6013 reference with tree-ssa-loop.h.
6015 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6017 * doc/invoke.texi: Add -maltivec={be,le} options, and document
6018 default element-order behavior for -maltivec.
6019 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6020 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6021 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6022 when targeting big endian, at least for now.
6023 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6025 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6028 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6029 var satisfies use_register_for_decl, just take into account type
6030 alignment, rather than decl alignment.
6032 PR tree-optimization/59622
6033 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
6034 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
6035 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
6036 Don't devirtualize for inplace at all. For targets.length () == 1,
6037 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
6039 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6041 * config/i386/i386.md (cpu): Remove the unused btver1.
6043 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6045 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
6047 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6050 * tree-core.h (struct target_globals): New forward declaration.
6051 (struct tree_target_option): Add globals field.
6052 * tree.h (TREE_TARGET_GLOBALS): Define.
6053 (prepare_target_option_nodes_for_pch): New prototype.
6054 * target-globals.h (struct target_globals): Define even if
6056 * tree.c (prepare_target_option_node_for_pch,
6057 prepare_target_option_nodes_for_pch): New functions.
6058 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
6059 * config/i386/i386.c: Include target-globals.h.
6060 (ix86_set_current_function): Instead of doing target_reinit
6061 unconditionally, use save_target_globals_default_opts and
6062 restore_target_globals.
6064 2014-01-09 Richard Biener <rguenther@suse.de>
6066 PR tree-optimization/59715
6067 * tree-cfg.h (split_critical_edges): Declare.
6068 * tree-cfg.c (split_critical_edges): Export.
6069 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
6071 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
6073 * cfgexpand.c (expand_stack_vars): Optionally disable
6074 asan stack protection.
6075 (expand_used_vars): Likewise.
6076 (partition_stack_vars): Likewise.
6077 * asan.c (asan_emit_stack_protection): Optionally disable
6078 after return stack usage.
6079 (instrument_derefs): Optionally disable memory access instrumentation.
6080 (instrument_builtin_call): Likewise.
6081 (instrument_strlen_call): Likewise.
6082 (asan_protect_global): Optionally disable global variables protection.
6083 * doc/invoke.texi: Added doc for new options.
6084 * params.def: Added new options.
6085 * params.h: Likewise.
6087 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6089 PR rtl-optimization/59724
6090 * ifcvt.c (cond_exec_process_if_block): Don't call
6091 flow_find_head_matching_sequence with 0 longest_match.
6092 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6093 non-active insns if !stop_after.
6094 (try_head_merge_bb): Revert 2014-01-07 changes.
6096 2014-01-08 Jeff Law <law@redhat.com>
6098 * ree.c (get_sub_rtx): New function, extracted from...
6099 (merge_def_and_ext): Here.
6100 (combine_reaching_defs): Use get_sub_rtx.
6102 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6104 * cgraph.h (varpool_variable_node): Do not choke on null node.
6106 2014-01-08 Catherine Moore <clm@codesourcery.com>
6108 * config/mips/mips.md (simple_return): Attempt to use JRC
6110 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6112 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6114 PR rtl-optimization/59137
6115 * reorg.c (steal_delay_list_from_target): Call update_block for
6117 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6119 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6121 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6122 two duplicate entries.
6124 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6127 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6129 * config/mips/mips.c (mips_truncated_op_cost): New function.
6130 (mips_rtx_costs): Adjust test for BADDU.
6131 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6133 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6135 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6136 (*baddu_si): ...this new pattern.
6138 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6141 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6143 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6146 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6148 (expand_expr, expand_normal): Adjust.
6149 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6150 inner_reference_p. Use inner_reference_p to expand inner references.
6151 (store_expr): Adjust.
6152 * cfgexpand.c (expand_call_stmt): Adjust.
6154 2014-01-08 Rong Xu <xur@google.com>
6156 * gcov-io.c (gcov_var): Move from gcov-io.h.
6157 (gcov_position): Ditto.
6158 (gcov_is_error): Ditto.
6159 (gcov_rewrite): Ditto.
6160 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6161 only part to libgcc/libgcov.h.
6163 2014-01-08 Marek Polacek <polacek@redhat.com>
6166 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6168 2014-01-08 Marek Polacek <polacek@redhat.com>
6171 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6173 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6175 PR rtl-optimization/59649
6176 * stor-layout.c (get_mode_bounds): For BImode return
6177 0 and STORE_FLAG_VALUE.
6179 2014-01-08 Richard Biener <rguenther@suse.de>
6182 * gimple.h (is_gimple_builtin_call): Remove.
6183 (gimple_builtin_call_types_compatible_p): New.
6184 (gimple_call_builtin_p): New overload.
6185 * gimple.c (is_gimple_builtin_call): Remove.
6186 (validate_call): Rename to ...
6187 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6189 (validate_type): New static function.
6190 (gimple_call_builtin_p): New overload and adjust.
6191 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6192 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6193 (gimple_fold_stmt_to_constant_1): Likewise.
6194 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6196 2014-01-08 Richard Biener <rguenther@suse.de>
6199 * gimplify.c (gimplify_expr): Gimplify register-register type
6200 VIEW_CONVERT_EXPRs to separate stmts.
6202 2014-01-07 Jeff Law <law@redhat.com>
6205 * ree.c (combine_set_extension): Handle case where source
6206 and destination registers in an extension insn are different.
6207 (combine_reaching_defs): Allow source and destination registers
6208 in extension to be different under limited circumstances.
6209 (add_removable_extension): Remove restriction that the
6210 source and destination registers in the extension are the same.
6211 (find_and_remove_re): Emit a copy from the extension's
6212 destination to its source after the defining insn if
6213 the source and destination registers are different.
6216 * ifcvt.c (merge_if_block): If we are merging a block with more than
6217 one successor with a block with no successors, remove any BARRIER
6218 after the second block.
6220 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6222 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6224 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6227 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6228 for 14-bit register offsets when INT14_OK_STRICT is false.
6230 2014-01-07 Roland Stigge <stigge@antcom.de>
6231 Michael Meissner <meissner@linux.vnet.ibm.com>
6234 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6235 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6237 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6239 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6242 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6244 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6245 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6246 rtx is const0_rtx or not.
6248 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6251 * target-globals.c (save_target_globals): Remove this_fn_optab
6253 * toplev.c: Include optabs.h.
6254 (target_reinit): Temporarily restore the global options if another
6255 set of options are in force.
6257 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6259 PR rtl-optimization/58668
6260 * cfgcleanup.c (flow_find_cross_jump): Don't count
6261 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6262 to determine what is counted.
6263 (flow_find_head_matching_sequence): Use active_insn_p to determine
6265 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6267 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6268 determine what is counted.
6270 PR tree-optimization/59643
6271 * tree-predcom.c (split_data_refs_to_components): If one dr is
6272 read and one write, determine_offset fails and the write isn't
6273 in the bad component, just put the read into the bad component.
6275 2014-01-07 Mike Stump <mikestump@comcast.net>
6276 Jakub Jelinek <jakub@redhat.com>
6279 * tree-core.h (struct tree_optimization_option): Change optabs
6280 type from unsigned char * to void *.
6281 * optabs.c (init_tree_optimization_optabs): Adjust
6282 TREE_OPTIMIZATION_OPTABS initialization.
6284 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6287 * config/i386/i386.h (struct machine_function): Add
6288 no_drap_save_restore field.
6289 * config/i386/i386.c (ix86_save_reg): Use
6290 !cfun->machine->no_drap_save_restore instead of
6291 crtl->stack_realign_needed.
6292 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6293 this function clears frame_pointer_needed. Set
6294 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6295 and DRAP reg is needed.
6297 2014-01-06 Marek Polacek <polacek@redhat.com>
6300 * doc/implement-c.texi: Mention that other integer types are
6301 permitted as bit-field types in strictly conforming mode.
6303 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6305 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6308 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6310 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6312 2014-01-06 Martin Jambor <mjambor@suse.cz>
6315 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6317 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6319 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6323 * var-tracking.c (add_stores): Preserve the value of the source even if
6324 we don't record the store.
6326 2014-01-06 Terry Guo <terry.guo@arm.com>
6328 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6330 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6333 * config/darwin.c (darwin_function_section): Adjust return values to
6334 correspond to optimisation changes made in r206070.
6336 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6338 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6339 from prefetch_block tune setting.
6340 (nocona_cost): Correct size of prefetch block to 64.
6342 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6344 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6345 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6346 used to save the static chain register in the computation of the offset
6347 from which the FP registers need to be restored.
6349 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6351 PR tree-optimization/59519
6352 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6353 ICE if get_current_def (current_new_name) is already non-NULL, as long
6354 as it is a phi result of some other phi in *new_exit_bb that has
6357 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6358 or vmovdqu* for misaligned_operand.
6359 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6360 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6361 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6362 aligned_mem for AVX512F masked aligned load and store builtins and for
6365 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6367 PR tree-optimization/59651
6368 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6369 Address range for negative step should be added by TYPE_SIZE_UNIT.
6371 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6373 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6374 overlapping registers also for registers other than the stack pointer.
6376 2014-01-03 Marek Polacek <polacek@redhat.com>
6379 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6382 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6385 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6388 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6389 (push splitter): Use <P:MODE_SIZE> instead of
6390 GET_MODE_SIZE (<P:MODE>mode).
6391 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6392 (mov -1, reg peephole2): Likewise.
6393 * config/i386/sse.md (*mov<mode>_internal,
6394 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6395 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6396 *<code><mode>3, *andnot<mode>3<mask_name>,
6397 <mask_codefor><code><mode>3<mask_name>): Likewise.
6398 * config/i386/subst.md (mask_mode512bit_condition,
6399 sd_mask_mode512bit_condition): Likewise.
6401 2014-01-02 Xinliang David Li <davidxl@google.com>
6403 PR tree-optimization/59303
6404 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6405 (dump_predicates): Better output format.
6406 (pred_equal_p): New function.
6407 (is_neq_relop_p): Ditto.
6408 (is_neq_zero_form_p): Ditto.
6409 (pred_expr_equal_p): Ditto.
6410 (pred_neg_p): Ditto.
6411 (simplify_pred): Ditto.
6412 (simplify_preds_2): Ditto.
6413 (simplify_preds_3): Ditto.
6414 (simplify_preds_4): Ditto.
6415 (simplify_preds): Ditto.
6417 (push_to_worklist): Ditto.
6418 (get_pred_info_from_cmp): Ditto.
6419 (is_degenerated_phi): Ditto.
6420 (normalize_one_pred_1): Ditto.
6421 (normalize_one_pred): Ditto.
6422 (normalize_one_pred_chain): Ditto.
6423 (normalize_preds): Ditto.
6424 (normalize_cond_1): Remove function.
6425 (normalize_cond): Ditto.
6426 (is_gcond_subset_of): Ditto.
6427 (is_subset_of_any): Ditto.
6428 (is_or_set_subset_of): Ditto.
6429 (is_and_set_subset_of): Ditto.
6430 (is_norm_cond_subset_of): Ditto.
6431 (pred_chain_length_cmp): Ditto.
6432 (convert_control_dep_chain_into_preds): Type change.
6433 (find_predicates): Ditto.
6434 (find_def_preds): Ditto.
6435 (destroy_predicates_vecs): Ditto.
6436 (find_matching_predicates_in_rest_chains): Ditto.
6437 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6438 (is_pred_expr_subset): Ditto.
6439 (is_pred_chain_subset_of): Ditto.
6440 (is_included_in): Ditto.
6441 (is_superset_of): Ditto.
6443 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6445 Update copyright years.
6447 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6449 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6450 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6451 config/arc/arc.md, config/arc/arc.opt,
6452 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6453 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6454 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6455 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6456 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6457 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6458 vtable-verify.c, vtable-verify.h: Use the standard form for the
6461 2014-01-02 Tobias Burnus <burnus@net-b.de>
6463 * gcc.c (process_command): Update copyright notice dates.
6464 * gcov-dump.c: Ditto.
6466 * doc/cpp.texi: Bump @copying's copyright year.
6467 * doc/cppinternals.texi: Ditto.
6468 * doc/gcc.texi: Ditto.
6469 * doc/gccint.texi: Ditto.
6470 * doc/gcov.texi: Ditto.
6471 * doc/install.texi: Ditto.
6472 * doc/invoke.texi: Ditto.
6474 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6476 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6478 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6480 * config/i386/sse.md (*mov<mode>_internal): Guard
6481 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6483 PR rtl-optimization/59647
6484 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6485 new_rtx into UNSIGNED_FLOAT rtxes.
6487 Copyright (C) 2014 Free Software Foundation, Inc.
6489 Copying and distribution of this file, with or without modification,
6490 are permitted in any medium without royalty provided the copyright
6491 notice and this notice are preserved.