1 2014-04-15 Richard Biener <rguenther@suse.de>
3 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4 Add struct function argument and adjust.
5 (find_func_aliases_for_call): Likewise.
6 (find_func_aliases): Likewise.
7 (find_func_clobbers): Likewise.
8 (intra_create_variable_infos): Likewise.
9 (compute_points_to_sets): Likewise.
10 (ipa_pta_execute): Adjust. Do not push/pop cfun.
12 2014-04-15 Richard Biener <rguenther@suse.de>
14 * tree.c (iterative_hash_expr): Use enum tree_code_class
15 to store TREE_CODE_CLASS.
16 (tree_block): Likewise.
17 (tree_set_block): Likewise.
18 * tree.h (fold_build_pointer_plus_loc): Use
19 convert_to_ptrofftype_loc.
21 2014-04-15 Jakub Jelinek <jakub@redhat.com>
24 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
27 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
29 * cfgloop.h (struct loop): Move force_vectorize down.
30 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
31 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
32 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
33 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
34 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
35 * tree-core.h (enum annot_expr_kind): Add new kind values.
36 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
37 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
39 * tree.def (ANNOTATE_EXPR): Tweak comment.
41 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
43 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
46 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
48 * tree.h (TYPE_IDENTIFIER): Declare.
49 * tree.c (subrange_type_for_debug_p): Use it.
50 * godump.c (go_format_type): Likewise.
51 * dwarf2out.c (is_cxx_auto, modified_type_die,
52 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
53 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
55 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
58 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
60 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
62 * config/i386/i386.c (examine_argument): Return bool. Return true if
63 parameter should be passed in memory.
64 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
65 (construct_container): Update calls to examine_argument.
66 (function_arg_advance_64): Ditto.
67 (return_in_memory_32): Merge with ix86_return_in_memory.
68 (return_in_memory_64): Ditto.
69 (return_in_memory_ms_64): Ditto.
71 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
73 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
74 * coverage.c (coverage_compute_profile_id): Handle externally visible
77 2014-04-14 Martin Jambor <mjambor@suse.cz>
79 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
80 DECL_DISREGARD_INLINE_LIMITS functions.
82 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
85 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
87 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
90 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
91 optimize_insn_for_speed_p instead of
92 optimize_function_for_speed_p.
94 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
96 * doc/invoke.texi (free): Document AArch64.
98 2014-04-14 Richard Biener <rguenther@suse.de>
100 PR tree-optimization/60042
101 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
102 (insert_into_preds_of_block): Do not prevent PHI insertion
103 for REFERENCE exprs here ...
104 (eliminate_dom_walker::before_dom_children): ... but prevent
105 their use here under similar conditions when applied to the
106 IL after PRE optimizations.
108 2014-04-14 Richard Biener <rguenther@suse.de>
110 * passes.def: Move early points-to after early SRA.
112 2014-04-14 Richard Biener <rguenther@suse.de>
114 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
115 check for which sign-changes we allow when forwarding
116 a converted value into a switch.
118 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
120 * stor-layout.c (place_field): Finalize non-constant offset for the
123 2014-04-14 Richard Biener <rguenther@suse.de>
125 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
127 (expand_switch_using_bit_tests_p): Likewise.
128 (process_switch): Compute and pass on speed_p based on the
130 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
131 optimize_bb_for_speed_p.
133 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
135 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
136 * function.h (struct function): Rename has_force_vect_loops into
137 has_force_vectorize_loops.
138 * lto-streamer-in.c (input_cfg): Adjust for renaming.
139 (input_struct_function_base): Likewise.
140 * lto-streamer-out.c (output_cfg): Likewise.
141 (output_struct_function_base): Likewise.
142 * omp-low.c (expand_omp_simd): Likewise.
143 * tree-cfg.c (move_sese_region_to_fn): Likewise.
144 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
145 (version_loop_for_if_conversion): Likewise.
146 (tree_if_conversion): Likewise.
147 (main_tree_if_conversion): Likewise.
148 (gate_tree_if_conversion): Likewise.
149 * tree-inline.c (copy_loops): Likewise.
150 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
151 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
152 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
153 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
154 * tree-vectorizer.c (vectorize_loops): Likewise.
155 * tree-vectorizer.h (unlimited_cost_model): Likewise.
157 2014-04-14 Richard Biener <rguenther@suse.de>
160 * lto-streamer-out.c (wrap_refs): New function.
161 (lto_output): Wrap symbol references in global initializes in
162 type-preserving MEM_REFs.
164 2014-04-14 Christian Bruel <christian.bruel@st.com>
166 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
168 2014-04-14 Christian Bruel <christian.bruel@st.com>
170 * config/sh/sh.md (setmemqi): New expand pattern.
171 * config/sh/sh.h (CLEAR_RATIO): Define.
172 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
173 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
175 2014-04-14 Richard Biener <rguenther@suse.de>
178 * fold-const.c (negate_expr_p): Don't negate directional rounding
180 (fold_negate_expr): Likewise.
182 2014-04-14 Richard Biener <rguenther@suse.de>
184 PR tree-optimization/59817
185 PR tree-optimization/60453
186 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
187 recursion to catch all CHRECs in the scalar evolution and restrict
188 the predicate for the remains appropriately.
190 2014-04-12 Catherine Moore <clm@codesourcery.com>
192 * config/mips/constraints.md: Add new register constraint "kb".
193 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
194 (*movhi_internal): Likewise.
195 (*movqi_internal): Likewise.
196 * config/mips/mips.h (M16_STORE_REGS): New register class.
197 (REG_CLASS_NAMES): Add M16_STORE_REGS.
198 (REG_CLASS_CONTENTS): Likewise.
199 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
201 2014-04-11 Tobias Burnus <burnus@net-b.de>
204 * doc/invoke.texi (-Wformat-signedness): Document it.
205 (Wformat=2): Mention that this enables -Wformat-signedness.
207 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
209 * common/config/epiphany/epiphany-common.c
210 (epiphany_option_optimization_table): Enable section anchors by
211 default at -O1 or higher.
212 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
213 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
214 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
215 carries no extra cost.
216 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
217 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
218 * config/epiphany/predicates.md (memclob_operand): New predicate.
219 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
220 Use memclob_operand predicate and X constraint for operand 3.
222 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
224 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
225 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
228 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
230 PR rtl-optimization/60651
231 * mode-switching.c (optimize_mode_switching): Make sure to emit
232 sets of a lower numbered entity before sets of a higher numbered
233 entity to a mode of the same or lower priority.
234 When creating a seginfo for a basic block that starts with a code
235 label, move the insertion point past the code label.
236 (new_seginfo): Document and enforce requirement that
237 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
238 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
239 * doc/tm.texi: Regenerate.
241 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
244 * config/arc/arc.c (arc_save_restore): Fix assert typo.
246 2013-04-11 Jakub Jelinek <jakub@redhat.com>
248 * BASE-VER: Set to 4.10.0.
250 2014-04-11 Tobias Burnus <burnus@net-b.de>
253 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
254 * doc/gcc.texi (Service): Update description in the @menu
255 * doc/invoke.texi (Option Summary): Remove misplaced and
258 2014-04-11 Steve Ellcey <sellcey@mips.com>
259 Jakub Jelinek <jakub@redhat.com>
262 * expr.c (convert_move): Use emit_store_flag_force instead of
263 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
266 2014-04-11 Richard Biener <rguenther@suse.de>
269 * varasm.c (assemble_alias): Avoid endless error reporting
270 recursion by setting TREE_ASM_WRITTEN.
272 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
274 * config/s390/s390.md: Add a splitter for NOT rtx.
276 2014-04-11 Jakub Jelinek <jakub@redhat.com>
278 PR rtl-optimization/60663
279 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
281 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
282 Jakub Jelinek <jakub@redhat.com>
285 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
286 flag from decl_node to node.
288 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
291 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
292 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
293 ameliorating the cases where it can be.
295 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
298 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
300 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
301 (loadsync_<mode>): Change mode.
302 (load_quadpti, store_quadpti): New.
303 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
304 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
305 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
307 2014-04-09 Cong Hou <congh@google.com>
310 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
313 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
315 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
316 instead of vnor to exploit possible fusion opportunity in the
318 (altivec_expand_vec_perm_const_le): Likewise.
320 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
322 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
323 (loadsync_<mode>): Change mode.
324 (load_quadpti, store_quadpti): New.
325 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
326 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
328 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
331 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
332 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
333 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
335 2014-04-08 Richard Biener <rguenther@suse.de>
338 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
339 a 64bit widest int print double-int similar to on HWI64 hosts.
341 2014-04-08 Richard Biener <rguenther@suse.de>
343 PR tree-optimization/60785
344 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
345 default defs properly.
347 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
349 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
352 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
354 * ipa-devirt.c (maybe_record_node): When node is not recorded,
355 set completep to false rather than true.
357 2014-04-07 Douglas B Rupp <rupp@adacore.com>
360 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
361 ARM_TARGET2_DWARF_FORMAT.
363 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
366 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
367 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
370 2014-04-07 Richard Biener <rguenther@suse.de>
372 PR tree-optimization/60766
373 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
374 (may_eliminate_iv): Convert cand_value_at result to desired type.
376 2014-04-07 Jason Merrill <jason@redhat.com>
379 * common.opt (-fno-gnu-unique): Add.
380 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
382 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
384 * haifa-sched.c: Fix outdated function reference and minor
385 grammar errors in introductory comment.
387 2014-04-07 Richard Biener <rguenther@suse.de>
390 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
392 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
394 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
397 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
398 size accounting for thunks.
399 (pa_asm_output_mi_thunk): Use final_start_function() and
400 final_end_function() to output function start and end directives.
402 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
404 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
405 device specific ISA/ feature information. Remove short_sp and
406 errata_skip ds. Add avr_device_specific_features enum to have device
408 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
409 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
410 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
411 updated device specific info.
412 * config/avr/avr-mcus.def: Merge device specific details to
414 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
416 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
417 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
418 assembler if RMW isa supported by current device.
419 * config/avr/genmultilib.awk: Update as device info structure changed.
420 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
422 2014-04-04 Cong Hou <congh@google.com>
424 PR tree-optimization/60656
425 * tree-vect-stmts.c (supportable_widening_operation):
426 Fix a bug that elements in a vector with vect_used_by_reduction
427 property are incorrectly reordered when the operation on it is not
428 consistant with the one in reduction operation.
430 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
432 PR rtl-optimization/60155
433 * gcse.c (record_set_data): New function.
434 (single_set_gcse): New function.
435 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
436 (hoist_code): Likewise.
437 (get_pressure_class_and_nregs): Likewise.
439 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
441 * explow.c (probe_stack_range): Emit a final optimization blockage.
443 2014-04-04 Anthony Green <green@moxielogic.com>
445 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
448 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
451 * lto-cgraph.c (input_overwrite_node): Check that partitioning
452 flags are set only during streaming.
453 * ipa.c (process_references, walk_polymorphic_call_targets,
454 symtab_remove_unreachable_nodes): Drop bodies of always inline
455 after early inlining.
456 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
458 2014-04-04 Jakub Jelinek <jakub@redhat.com>
459 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
462 * dwarf2out.c (const_ok_for_output_1): Reject expressions
465 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
468 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
470 (cortex_a53_fdivd): Likewise.
472 2014-04-04 Martin Jambor <mjambor@suse.cz>
475 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
477 * cgraph.c (clone_of_p): Also return true if thunks match.
478 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
479 cgraph_function_or_thunk_node and an obsolete comment.
480 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
482 (build_function_decl_skip_args): Likewise.
483 (set_new_clone_decl_and_node_flags): New function.
484 (duplicate_thunk_for_node): Likewise.
485 (redirect_edge_duplicating_thunks): Likewise.
486 (cgraph_clone_node): New parameter args_to_skip, pass it to
487 redirect_edge_duplicating_thunks which is called instead of
488 cgraph_redirect_edge_callee.
489 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
490 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
492 2014-04-04 Jeff Law <law@redhat.com>
495 * config/arm/predicates.md (const_int_I_operand): New predicate.
496 (const_int_M_operand): Similarly.
497 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
499 (insv_t2, extv_reg, extzv_t2): Likewise.
500 (load_multiple_with_writeback): Similarly for const_int_I_operand.
501 (pop_multiple_with_writeback_and_return): Likewise.
502 (vfp_pop_multiple_with_writeback): Likewise
504 2014-04-04 Richard Biener <rguenther@suse.de>
507 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
508 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
510 * gimplify.h (gimple_add_tmp_var_fn): Declare.
511 * gimplify.c (gimple_add_tmp_var_fn): New function.
512 * gimple-expr.h (create_tmp_reg_fn): Declare.
513 * gimple-expr.c (create_tmp_reg_fn): New function.
514 * gimple-low.c (record_vars_into): Don't change cfun.
515 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
516 code generation without cfun.
518 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
521 * Makefile.in (install-driver): Fix shell scripting.
523 2014-04-03 Cong Hou <congh@google.com>
525 PR tree-optimization/60505
526 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
527 threshold of number of iterations below which no vectorization
529 * tree-vect-loop.c (new_loop_vec_info):
530 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
531 * tree-vect-loop.c (vect_analyze_loop_operations):
532 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
533 * tree-vect-loop.c (vect_transform_loop):
534 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
535 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
536 of iterations of the loop and see if we should build the epilogue.
538 2014-04-03 Richard Biener <rguenther@suse.de>
540 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
541 (streamer_tree_cache_create): Adjust.
542 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
543 to allow optional nodes array.
544 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
545 (streamer_tree_cache_append): Likewise.
546 (streamer_tree_cache_create): Create nodes array optionally
547 as specified by parameter.
548 * lto-streamer-out.c (create_output_block): Avoid maintaining
549 the node array in the writer cache.
550 (DFS_write_tree): Remove assertion.
551 (produce_asm_for_decls): Free the out decl state hash table early.
552 * lto-streamer-in.c (lto_data_in_create): Adjust for
553 streamer_tree_cache_create prototype change.
555 2014-04-03 Richard Biener <rguenther@suse.de>
557 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
558 set TREE_CHAIN to NULL_TREE.
560 2014-04-03 Richard Biener <rguenther@suse.de>
562 PR tree-optimization/60740
563 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
564 over all GIMPLE_COND operands.
566 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
568 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
569 (Weffc++): Remove Scott's numbering, merge lists and reference
575 * class.c (accessible_nvdtor_p): New.
576 (check_bases): Don't check base destructor here ...
577 (check_bases_and_members): ... check them here. Trigger on
578 Wnon-virtual-dtor flag.
579 (finish_struct_1): Use accessible_nvdtor_p.
582 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
583 * g++.dg/warn/Wnvdtor-2.C: New.
584 * g++.dg/warn/Wnvdtor-3.C: New.
585 * g++.dg/warn/Wnvdtor-4.C: New.
586 * g++.dg/warn/Weff1.C: Delete.
587 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
588 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
590 2014-04-03 Nick Clifton <nickc@redhat.com>
592 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
595 2014-04-03 Martin Jambor <mjambor@suse.cz>
597 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
598 mention gcc_unreachable before failing.
599 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
602 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
605 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
606 inconsistent code and instead mark the context inconsistent.
607 (possible_polymorphic_call_targets): For inconsistent contexts
608 return empty complete list.
610 2014-04-02 Anthony Green <green@moxielogic.com>
612 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
613 (extendqisi2, extendhisi2): Define.
614 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
615 (WCHAR_TYPE): Change to unsigned int.
617 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
619 PR tree-optimization/60733
620 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
621 insertion point for PHI candidates to be the end of the feeding
622 block for the PHI argument.
624 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
626 PR rtl-optimization/60650
627 * lra-constraints.c (process_alt_operands): Decrease reject for
628 earlyclobber matching.
630 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
632 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
634 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
636 * config/spu/spu.c (pad_bb): Do not crash when the last
637 insn is CODE_FOR_blockage.
639 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
641 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
642 lies outside the target mode.
644 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
647 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
648 software floating point or no floating point registers, do not
649 allow any type in the FPRs. Eliminate a test for SPE SIMD types
650 in GPRs that occurs after we tested for GPRs that would never be
653 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
654 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
655 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
656 specifically allow DDmode, since that does not use the SPE SIMD
659 2014-04-02 Richard Biener <rguenther@suse.de>
662 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
663 MODE_INTs. Properly use negv_optab.
664 (expand_abs): Likewise.
666 2014-04-02 Richard Biener <rguenther@suse.de>
669 * Makefile.in (install-driver): Guard extra installs with special
672 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
674 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
677 2014-04-01 Richard Henderson <rth@redhat.com>
680 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
681 alternative enabled before register allocation.
683 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
685 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
686 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
688 (nios2_large_got_address): Remove unneeded 'sym' parameter.
689 (nios2_got_address): Update nios2_large_got_address call site.
690 (nios2_delegitimize_address): New function.
691 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
692 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
693 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
695 2014-04-01 Martin Husemann <martin@duskware.de>
697 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
700 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
702 PR rtl-optimization/60604
703 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
704 check from register_operand.
705 (register_operand): Redefine in terms of general_operand.
706 (nonmemory_operand): Use register_operand for the non-constant cases.
708 2014-04-01 Richard Biener <rguenther@suse.de>
710 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
712 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
714 * doc/invoke.texi (mapp-regs): Clarify.
716 2014-03-31 Ulrich Drepper <drepper@gmail.com>
718 * config/i386/avx512fintrin.h (__v32hi): Define type.
720 (_mm512_set1_epi8): Define.
721 (_mm512_set1_epi16): Define.
722 (_mm512_set4_epi32): Define.
723 (_mm512_set4_epi64): Define.
724 (_mm512_set4_pd): Define.
725 (_mm512_set4_ps): Define.
726 (_mm512_setr4_epi64): Define.
727 (_mm512_setr4_epi32): Define.
728 (_mm512_setr4_pd): Define.
729 (_mm512_setr4_ps): Define.
730 (_mm512_setzero_epi32): Define.
732 2014-03-31 Martin Jambor <mjambor@suse.cz>
735 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
736 callsite_arguments_match_p. Updated all callers. Also check types of
737 corresponding formal parameters and actual arguments.
738 (not_all_callers_have_enough_arguments_p) Renamed to
739 some_callers_have_mismatched_arguments_p.
741 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
743 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
745 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
748 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
751 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
753 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
754 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
756 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
757 Use FMAMODE_NOVF512 mode iterator.
758 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
759 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
760 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
761 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
763 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
764 Use VF_128_256 mode iterator.
765 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
768 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
770 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
771 static chain if needed.
773 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
776 * lra-constraints.c (index_part_to_reg): New.
777 (process_address): Use it.
779 2014-03-27 Jeff Law <law@redhat.com>
780 Jakub Jelinek <jakub@redhat.com>
783 * expr.c (do_tablejump): Use simplify_gen_binary rather than
784 gen_rtx_{PLUS,MULT} to build up the address expression.
786 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
787 creating non-canonical RTL.
789 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
792 * ipa-inline.c (want_inline_small_function_p): Short circuit large
793 functions; reorganize to make cheap checks first.
794 (inline_small_functions): Do not estimate growth when dumping;
796 * ipa-inline.h (inline_summary): Add min_size.
797 (growth_likely_positive): New function.
798 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
799 (set_cond_stmt_execution_predicate): Cleanup.
800 (estimate_edge_size_and_time): Compute min_size.
801 (estimate_calls_size_and_time): Likewise.
802 (estimate_node_size_and_time): Likewise.
803 (inline_update_overall_summary): Update min_size.
804 (do_estimate_edge_time): Likewise.
805 (do_estimate_edge_size): Update.
806 (do_estimate_edge_hints): Update.
807 (growth_likely_positive): New function.
809 2014-03-28 Jakub Jelinek <jakub@redhat.com>
812 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
813 also if addr has VOIDmode.
815 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
817 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
818 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
820 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
821 instructions as well as AdvancedSIMD loads.
823 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
825 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
826 Use crypto_aese type.
827 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
828 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
829 crypto_aese, crypto_aesmc. Move to types.md.
830 * config/arm/types.md (crypto_aes): Split into crypto_aese,
832 * config/arm/iterators.md (crypto_type): Likewise.
834 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
836 * cgraph.c: Include expr.h and tree-dfa.h.
837 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
840 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
843 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
844 regs from checking multi-reg pseudos.
846 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
848 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
850 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
852 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
853 if it would clobber the stack pointer, even temporarily.
855 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
857 * mode-switching.c: Make small adjustments to the top comment.
859 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
861 * config/rs6000/constraints.md (wD constraint): New constraint to
862 match the constant integer to get the top DImode/DFmode out of a
863 vector in a VSX register.
865 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
866 match the constant integer to get the top DImode/DFmode out of a
867 vector in a VSX register.
869 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
872 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
875 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
876 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
878 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
879 Optimize vec_extract of 64-bit values, where the value being
880 extracted is in the top word, where we can use scalar
881 instructions. Add direct move and store support. Combine the big
882 endian/little endian vector select load support into a single insn.
883 (vsx_extract_<mode>_internal1): Likewise.
884 (vsx_extract_<mode>_internal2): Likewise.
885 (vsx_extract_<mode>_load): Likewise.
886 (vsx_extract_<mode>_store): Likewise.
887 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
888 combined into vsx_extract_<mode>_load.
889 (vsx_extract_<mode>_one_le): Likewise.
891 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
892 define the top 64-bit vector element.
894 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
897 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
898 Document vec_vbpermq builtin.
901 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
902 enable use of xxsldwi and xxpermdi builtin functions.
903 (vec_xxpermdi): Likewise.
905 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
906 Document use of vec_xxsldwi and vec_xxpermdi builtins.
908 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
910 PR rtl-optimization/60650
911 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
913 (find_spills_for): New.
914 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
915 Spill all pseudos on the second iteration.
917 2014-03-27 Marek Polacek <polacek@redhat.com>
920 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
923 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
925 * config/s390/s390.c (s390_can_use_return_insn): Check for
926 call-saved FPRs on 31 bit.
928 2014-03-27 Jakub Jelinek <jakub@redhat.com>
931 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
932 if they need regimplification, just drop them instead of
933 calling gimple_regimplify_operands on them.
935 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
938 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
939 (aarch64_frame_pointer_required): Adjust logic.
940 (aarch64_can_eliminate): Adjust logic.
941 (aarch64_override_options_after_change): Adjust logic.
943 2014-03-27 Dehao Chen <dehao@google.com>
945 * ipa-inline.c (early_inliner): Update node's inline info.
947 2014-03-26 Dehao Chen <dehao@google.com>
949 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
950 compiler inserted conditional jumps for NAN float check.
952 2014-03-26 Jakub Jelinek <jakub@redhat.com>
954 * ubsan.h (ubsan_create_data): Change second argument's type
955 to const location_t *.
956 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
958 (ubsan_create_data): Change second argument to const location_t *PLOC.
959 Create Loc field whenever PLOC is non-NULL.
960 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
961 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
965 * real.c (real_to_integer2): Change type of low to UHWI.
967 2014-03-26 Tobias Burnus <burnus@net-b.de>
969 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
970 (CILK_SELF_SPECS): New define.
971 (driver_self_specs): Use it.
973 2014-03-26 Richard Biener <rguenther@suse.de>
975 * tree-pretty-print.c (percent_K_format): Implement special
976 case for LTO and its stripped down BLOCK tree.
978 2014-03-26 Jakub Jelinek <jakub@redhat.com>
981 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
983 * tree-vrp.c (simplify_internal_call_using_ranges): If only
984 one range is range_int_cst_p, but not both, at least optimize
985 addition/subtraction of 0 and multiplication by 0 or 1.
986 * gimple-fold.c (gimple_fold_call): Fold
987 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
988 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
989 INTEGER_CSTs, try to fold at least x * 0 and y - y.
991 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
993 PR rtl-optimization/60452
994 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
995 <case REG>: Return 1 for invalid offsets from the frame pointer.
997 2014-03-26 Marek Polacek <polacek@redhat.com>
1000 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1003 2014-03-26 Marek Polacek <polacek@redhat.com>
1006 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1009 2014-03-26 Marek Polacek <polacek@redhat.com>
1012 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1013 multiplication in unsigned type.
1015 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1017 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
1019 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1021 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
1023 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1026 * cif-code.def (UNREACHABLE) New code.
1027 * ipa-inline.c (inline_small_functions): Skip edges to
1028 __builtlin_unreachable.
1029 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
1030 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
1031 predicate to __bulitin_unreachable.
1032 (set_cond_stmt_execution_predicate): Fix issue when
1033 invert_tree_comparison returns ERROR_MARK.
1034 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
1035 propagate to inline clones.
1036 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
1038 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
1039 * cgraphclones.c (cgraph_clone_node): If call destination is already
1040 ureachable, do not redirect it back.
1041 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
1044 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1046 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
1047 Do not modify inline clones.
1049 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1051 * config/i386/i386.md (general_sext_operand): New mode attr.
1052 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
1053 don't generate (sign_extend (const_int)).
1054 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
1055 operands[2]. Use We constraint instead of <i> and
1056 <general_sext_operand> predicate instead of <general_operand>.
1057 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
1058 * config/i386/constraints.md (We): New constraint.
1059 * config/i386/predicates.md (x86_64_sext_operand,
1060 sext_operand): New predicates.
1062 2014-03-25 Martin Jambor <mjambor@suse.cz>
1065 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1066 inconsistent devirtualizations to __builtin_unreachable.
1068 2014-03-25 Marek Polacek <polacek@redhat.com>
1071 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1073 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1075 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1076 order of elements for big-endian.
1078 2014-03-25 Richard Biener <rguenther@suse.de>
1081 * gimplify-me.c (gimple_regimplify_operands): Update the
1084 2014-03-25 Martin Jambor <mjambor@suse.cz>
1087 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1088 (lto_output_varpool_node): Likewise.
1089 (input_overwrite_node): Likewise.
1090 (input_varpool_node): Likewise.
1092 2014-03-25 Richard Biener <rguenther@suse.de>
1094 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1095 (run_gcc): Likewise.
1097 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1099 * combine.c (simplify_compare_const): Add MODE argument.
1100 Handle mode_width 0 as very large mode_width.
1101 (try_combine, simplify_comparison): Adjust callers.
1103 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1104 type to avoid signed integer overflow.
1105 * explow.c (plus_constant): Likewise.
1107 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1109 * doc/generic.texi: Correct typos.
1111 2014-03-24 Tobias Burnus <burnus@net-b.de>
1113 * doc/invoke.texi (-flto): Expand section about
1114 using static libraries with LTO.
1116 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1118 PR rtl-optimization/60501
1119 * optabs.def (addptr3_optab): New optab.
1120 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1121 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1122 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1124 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1126 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1128 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1130 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1133 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1134 (_mm256_undefined_ps): Define.
1135 (_mm256_undefined_pd): Define.
1136 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1137 (_mm_undefined_pd): Define.
1138 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1139 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1140 (_mm512_undefined_ps): Define.
1141 (_mm512_undefined_pd): Define.
1142 Use _mm*_undefined_*.
1143 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1145 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1147 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1148 (lshr_simd): DI mode added.
1149 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1150 (aarch64_ushr_simddi): Likewise.
1151 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1152 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1153 (vshrd_n_u64): Likewise.
1155 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1157 * Makefile.in (s-macro_list): Depend on cc1.
1159 2014-03-23 Teresa Johnson <tejohnson@google.com>
1161 * ipa-utils.c (ipa_print_order): Use specified dump file.
1163 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1165 PR rtl-optimization/60601
1166 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1168 * gcc.c (eval_spec_function): Initialize save_growing_value.
1170 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1173 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1174 code == MINUS_EXPR, never swap op0 with op1.
1176 * toplev.c (init_local_tick): Avoid signed integer multiplication
1178 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1179 shift by first operand's bitsize.
1181 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1184 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1186 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1187 TARGET_ISA_64BIT_P(x).
1189 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1191 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1192 pattern for vector nor instead of subtract from splat(-1).
1193 (altivec_expand_vec_perm_const_le): Likewise.
1195 2014-03-21 Richard Henderson <rth@twiddle.net>
1198 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1199 related insns after epilogue_completed.
1201 2014-03-21 Martin Jambor <mjambor@suse.cz>
1204 * cgraph.h (symtab_node): New flag body_removed.
1205 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1206 when removing bodies.
1207 * symtab.c (dump_symtab_base): Dump body_removed flag.
1208 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1209 had their bodies removed.
1211 2014-03-21 Martin Jambor <mjambor@suse.cz>
1214 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1217 2014-03-21 Richard Biener <rguenther@suse.de>
1219 PR tree-optimization/60577
1220 * tree-core.h (struct tree_base): Document nothrow_flag use
1221 in VAR_DECL_NONALIASED.
1222 * tree.h (VAR_DECL_NONALIASED): New.
1223 (may_be_aliased): Adjust.
1224 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1226 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1228 * expr.c (expand_expr_real_1): Remove outdated comment.
1230 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1233 * ira.c (adjust_cleared_regs): Call copy_rtx on
1234 *reg_equiv[REGNO (loc)].src_p before passing it to
1235 simplify_replace_fn_rtx.
1238 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1239 into CONST, put pic register as first operand of PLUS. Use
1240 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1242 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1244 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1246 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1248 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1249 around for store forwarding issue in the FPU on the UT699.
1250 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1251 loads and operations if -mfix-ut699 is specified.
1252 (divtf3_hq): Tweak attribute.
1253 (sqrttf2_hq): Likewise.
1255 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1257 * calls.c (store_one_arg): Remove incorrect const qualification on the
1258 type of the temporary.
1259 * cfgexpand.c (expand_return): Likewise.
1260 * expr.c (expand_constructor): Likewise.
1261 (expand_expr_real_1): Likewise.
1263 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1265 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1268 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1271 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1273 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1275 * config/arm/aarch-common-protos.h
1276 (alu_cost_table): Fix spelling of "extend".
1277 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1279 2014-03-19 Richard Biener <rguenther@suse.de>
1282 * tree-core.h (tree_type_common): Re-order pointer members
1283 to reduce recursion depth during GC walks.
1285 2014-03-19 Marek Polacek <polacek@redhat.com>
1288 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1289 before accessing it.
1291 2014-03-19 Richard Biener <rguenther@suse.de>
1294 * lto-streamer-in.c (input_function): In WPA stage do not drop
1297 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1299 PR tree-optimization/60559
1300 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1301 with build_zero_cst assignment.
1303 2014-03-18 Kai Tietz <ktietz@redhat.com>
1305 PR rtl-optimization/56356
1306 * sdbout.c (sdbout_parms): Verify that parms'
1307 incoming argument is valid.
1308 (sdbout_reg_parms): Likewise.
1310 2014-03-18 Richard Henderson <rth@redhat.com>
1313 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1314 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1315 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1317 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1319 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1320 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1321 Italicize plugin event names in description. Explain that
1322 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1323 Remind that no GCC functions should be called after PLUGIN_FINISH.
1324 Explain what pragmas with expansion are.
1326 2014-03-18 Martin Liska <mliska@suse.cz>
1328 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1329 gimple call statement is update.
1330 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1331 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1333 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1336 * ubsan.c (ubsan_instrument_unreachable): Call
1337 initialize_sanitizer_builtins.
1338 (ubsan_pass): Likewise.
1341 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1342 varpool_finalize_decl instead of rest_of_decl_compilation.
1344 2014-03-18 Richard Biener <rguenther@suse.de>
1346 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1347 by using bitmap_and_compl instead of bitmap_and_compl_into.
1348 (df_rd_transfer_function): Likewise.
1350 2014-03-18 Richard Biener <rguenther@suse.de>
1352 * doc/lto.texi (fresolution): Fix typo.
1354 2014-03-18 Richard Biener <rguenther@suse.de>
1356 * doc/invoke.texi (flto): Update for changes in 4.9.
1358 2014-03-18 Richard Biener <rguenther@suse.de>
1360 * doc/loop.texi: Remove section on the removed lambda framework.
1361 Update loop docs with recent changes in preserving loop structure.
1363 2014-03-18 Richard Biener <rguenther@suse.de>
1365 * doc/lto.texi (-fresolution): Document.
1367 2014-03-18 Richard Biener <rguenther@suse.de>
1369 * doc/contrib.texi: Adjust my name.
1371 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1374 * internal-fn.c: Include diagnostic-core.h.
1375 (expand_BUILTIN_EXPECT): New function.
1376 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1377 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1378 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1379 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1381 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1382 Revert 3 argument __builtin_expect code.
1383 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1384 * gimple-fold.c (gimple_fold_call): Likewise.
1385 * tree.h (fold_builtin_expect): New prototype.
1386 * builtins.c (build_builtin_expect_predicate): Add predictor
1387 argument, if non-NULL, create 3 argument __builtin_expect.
1388 (fold_builtin_expect): No longer static. Add ARG2 argument,
1389 pass it through to build_builtin_expect_predicate.
1390 (fold_builtin_2): Adjust caller.
1391 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1392 * internal-fn.def (BUILTIN_EXPECT): New.
1394 2014-03-18 Tobias Burnus <burnus@net-b.de>
1397 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1398 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1399 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1401 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1404 * predict.c (combine_predictions_for_bb): Fix up formatting.
1405 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1406 fill what it points to if non-NULL.
1407 (tree_predict_by_opcode): Adjust caller, use the predictor.
1408 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1410 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1412 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1413 proper constant for the store mode.
1415 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1417 * symtab.c (change_decl_assembler_name): Fix transparent alias
1420 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1422 * config/aarch64/aarch64.c: Correct the comments about the
1423 aarch64 stack layout.
1425 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1427 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1428 check for GF_OMP_FOR_KIND_FOR.
1430 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1432 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1433 ymm and zmm register names.
1435 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1438 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1439 note creation for the 2010-08-31 changes.
1441 2014-03-17 Marek Polacek <polacek@redhat.com>
1444 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1445 as -fno-tree-loop-vectorize.
1446 (expand_omp_simd): Likewise.
1448 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1450 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1451 (eligible_for_call_delay): New prototype.
1452 * config/sparc/sparc.c (tls_call_delay): Rename into...
1453 (eligible_for_call_delay): ...this. Return false if the instruction
1454 cannot be put in the delay slot of a branch.
1455 (eligible_for_restore_insn): Simplify.
1456 (eligible_for_return_delay): Return false if the instruction cannot be
1457 put in the delay slot of a branch and simplify.
1458 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1459 put in the delay slot of a branch.
1460 * config/sparc/sparc.md (fix_ut699): New attribute.
1461 (tls_call_delay): Delete.
1462 (in_call_delay): Reimplement.
1463 (eligible_for_sibcall_delay): Rename into...
1464 (in_sibcall_delay): ...this.
1465 (eligible_for_return_delay): Rename into...
1466 (in_return_delay): ...this.
1467 (in_branch_delay): Reimplement.
1468 (in_uncond_branch_delay): Delete.
1469 (in_annul_branch_delay): Delete.
1471 2014-03-14 Richard Henderson <rth@redhat.com>
1474 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1475 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1476 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1477 (floathi splitters): Remove.
1478 (float<SWI48x>xf2): New pattern.
1479 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1480 code that tried to handle DImode for 32-bit, but which was excluded
1481 by the pattern's condition. Drop allocation of stack temporary.
1482 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1483 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1484 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1485 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1486 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1487 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1488 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1489 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1490 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1491 (*float<SWI48x><X87MODEF>2_i387): Remove.
1492 (all float _with_temp splitters): Remove.
1493 (*float<SWI48x><MODEF>2_i387): New pattern.
1494 (*float<SWI48><MODEF>2_sse): New pattern.
1495 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1496 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1498 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1499 Marek Polacek <polacek@redhat.com>
1502 * common.opt (dump_base_name_prefixed): New Variable.
1503 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1504 if x_dump_base_name_prefixed is already set, set it at the end.
1506 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1508 PR rtl-optimization/60508
1509 * lra-constraints.c (get_reload_reg): Add new parameter
1511 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1512 Pass the new parameter values.
1514 2014-03-14 Richard Biener <rguenther@suse.de>
1516 * common.opt: Revert unintented changes from r205065.
1519 2014-03-14 Richard Biener <rguenther@suse.de>
1522 * cfghooks.c (split_block): Properly adjust all loops the
1523 block was a latch of.
1525 2014-03-14 Martin Jambor <mjambor@suse.cz>
1528 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1531 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1534 * config/avr/avr.c (avr_set_current_function): Pass function name
1535 through default_strip_name_encoding before sanity checking instead
1536 of skipping the first char of the assembler name.
1538 2014-03-13 Richard Henderson <rth@redhat.com>
1541 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1542 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1543 * config/i386/i386-protos.h: Likewise.
1544 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1545 in the expander instead of a splitter.
1546 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1547 any possibility of requiring a memory.
1548 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1549 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1550 (fp branch splitters): Update for ix86_split_fp_branch.
1551 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1552 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1553 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1554 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1555 (*fop_<MODEF>_3_i387): Likewise.
1556 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1557 (splitters for the fop_* register patterns): Remove.
1558 (fscalexf4_i387): Rename from *fscalexf4_i387.
1559 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1561 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1563 PR tree-optimization/59779
1564 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1565 type for bitsize and maxsize instead of HOST_WIDE_INT.
1567 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1569 PR rtl-optimization/57320
1570 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1571 the CFG after thread_prologue_and_epilogue_insns.
1573 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1575 PR rtl-optimization/57189
1576 * lra-constraints.c (process_alt_operands): Disfavor spilling
1579 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1581 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1583 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1585 PR tree-optimization/59025
1587 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1588 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1590 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1593 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1594 calls of avr_out_plus_1.
1596 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1598 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1599 BB's single pred and update the father loop's latch info later.
1601 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1603 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1607 (VEC_base): Likewise.
1608 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1609 registers, we need to swap double words in little endian mode.
1611 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1612 to be a container mode for 128-bit integer operations added in ISA
1613 2.07. Unlike TImode and PTImode, the preferred register set is
1614 the Altivec/VMX registers for the 128-bit operations.
1616 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1618 (rs6000_split_128bit_ok_p): Likewise.
1620 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1621 macros for creating ISA 2.07 normal and overloaded builtin
1622 functions with 3 arguments.
1623 (BU_P8V_OVERLOAD_3): Likewise.
1624 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1625 for use as overloaded functions.
1626 (VPERM_1TI_UNS): Likewise.
1627 (VSEL_1TI): Likewise.
1628 (VSEL_1TI_UNS): Likewise.
1629 (ST_INTERNAL_1ti): Likewise.
1630 (LD_INTERNAL_1ti): Likewise.
1631 (XXSEL_1TI): Likewise.
1632 (XXSEL_1TI_UNS): Likewise.
1633 (VPERM_1TI): Likewise.
1634 (VPERM_1TI_UNS): Likewise.
1635 (XXPERMDI_1TI): Likewise.
1636 (SET_1TI): Likewise.
1637 (LXVD2X_V1TI): Likewise.
1638 (STXVD2X_V1TI): Likewise.
1639 (VEC_INIT_V1TI): Likewise.
1640 (VEC_SET_V1TI): Likewise.
1641 (VEC_EXT_V1TI): Likewise.
1642 (EQV_V1TI): Likewise.
1643 (NAND_V1TI): Likewise.
1644 (ORC_V1TI): Likewise.
1645 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1646 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1648 (VADDUQM): Likewise.
1649 (VSUBCUQ): Likewise.
1650 (VADDEUQM): Likewise.
1651 (VADDECUQ): Likewise.
1652 (VSUBEUQM): Likewise.
1653 (VSUBECUQ): Likewise.
1655 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1656 __int128_t and __uint128_t types.
1657 (__uint128_type): Likewise.
1658 (altivec_categorize_keyword): Add support for vector __int128_t,
1659 vector __uint128_t, vector __int128, and vector unsigned __int128
1660 as a container type for TImode operations that need to be done in
1661 VSX/Altivec registers.
1662 (rs6000_macro_to_expand): Likewise.
1663 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1664 to support 128-bit integer instructions vaddcuq, vadduqm,
1665 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1666 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1668 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1669 for V1TImode, and set up preferences to use VSX/Altivec registers.
1670 Setup VSX reload handlers.
1671 (rs6000_debug_reg_global): Likewise.
1672 (rs6000_init_hard_regno_mode_ok): Likewise.
1673 (rs6000_preferred_simd_mode): Likewise.
1674 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1675 (easy_altivec_constant): Likewise.
1676 (output_vec_const_move): Likewise.
1677 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1679 (rs6000_expand_vector_extract): Likewise.
1680 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1682 (rs6000_const_vec): Add support for V1TImode.
1683 (rs6000_emit_le_vsx_load): Swap double words when loading or
1684 storing TImode/V1TImode.
1685 (rs6000_emit_le_vsx_store): Likewise.
1686 (rs6000_emit_le_vsx_move): Likewise.
1687 (rs6000_emit_move): Add support for V1TImode.
1688 (altivec_expand_ld_builtin): Likewise.
1689 (altivec_expand_st_builtin): Likewise.
1690 (altivec_expand_vec_init_builtin): Likewise.
1691 (altivec_expand_builtin): Likewise.
1692 (rs6000_init_builtins): Add support for V1TImode type. Add
1693 support for ISA 2.07 128-bit integer builtins. Define type names
1694 for the VSX/Altivec vector types.
1695 (altivec_init_builtins): Add support for overloaded vector
1696 functions with V1TImode type.
1697 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1698 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1700 (rs6000_split_128bit_ok_p): Likewise.
1701 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1702 __int128_t and vector __uint128_t.
1704 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1705 and mode attributes.
1712 (VS_scalar): Likewise.
1713 (VS_double): Likewise.
1714 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1716 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1717 we support the ISA 2.07 128-bit integer arithmetic instructions.
1718 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1719 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1720 and TImode types for use with the builtin functions.
1721 (V1TI_type_node): Likewise.
1722 (unsigned_V1TI_type_node): Likewise.
1723 (intTI_type_internal_node): Likewise.
1724 (uintTI_type_internal_node): Likewise.
1726 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1727 128-bit builtin functions.
1728 (UNSPEC_VADDEUQM): Likewise.
1729 (UNSPEC_VADDECUQ): Likewise.
1730 (UNSPEC_VSUBCUQ): Likewise.
1731 (UNSPEC_VSUBEUQM): Likewise.
1732 (UNSPEC_VSUBECUQ): Likewise.
1733 (VM): Add V1TImode to vector mode iterators.
1735 (VI_unit): Likewise.
1736 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1737 (altivec_vaddcuq): Likewise.
1738 (altivec_vsubuqm): Likewise.
1739 (altivec_vsubcuq): Likewise.
1740 (altivec_vaddeuqm): Likewise.
1741 (altivec_vaddecuq): Likewise.
1742 (altivec_vsubeuqm): Likewise.
1743 (altivec_vsubecuq): Likewise.
1745 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1747 (BOOL_128): Likewise.
1748 (BOOL_REGS_OUTPUT): Likewise.
1749 (BOOL_REGS_OP1): Likewise.
1750 (BOOL_REGS_OP2): Likewise.
1751 (BOOL_REGS_UNARY): Likewise.
1752 (BOOL_REGS_AND_CR0): Likewise.
1754 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1755 128-bit integer builtin support.
1756 (vec_vadduqm): Likewise.
1757 (vec_vaddecuq): Likewise.
1758 (vec_vaddeuqm): Likewise.
1759 (vec_vsubecuq): Likewise.
1760 (vec_vsubeuqm): Likewise.
1761 (vec_vsubcuq): Likewise.
1762 (vec_vsubuqm): Likewise.
1764 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1765 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1766 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1767 128-bit integer add/subtract to ISA 2.07.
1769 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1771 * config/arc/arc.c (arc_predicate_delay_insns):
1772 Fix third argument passed to conditionalize_nonjump.
1774 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1776 * config/aarch64/aarch64-builtins.c
1777 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1778 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1779 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1780 instead of __builtin_lfloor.
1781 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1783 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1785 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1786 (tree_ssa_ifcombine_bb_1): New function.
1787 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1788 is an empty forwarder block to then_bb or vice versa and then_bb
1789 and else_bb are effectively swapped.
1791 2014-03-12 Christian Bruel <christian.bruel@st.com>
1794 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1795 REG_CFA_DEF_CFA note.
1796 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1797 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1799 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1801 PR tree-optimization/60454
1802 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1804 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1806 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1807 Do not define target_cpu_default2 to generic.
1808 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1809 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1810 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1812 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1813 Marc Glisse <marc.glisse@inria.fr>
1815 PR tree-optimization/60502
1816 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1817 instead of build_low_bits_mask.
1819 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1822 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1823 if there are multiple uses, but op doesn't live on E edge.
1824 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1825 clobber stmts before __builtin_unreachable.
1827 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1829 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1830 hard_frame_pointer_rtx.
1831 * cse.c (cse_insn): Remove volatile check.
1832 * cselib.c (cselib_process_insn): Likewise.
1833 * dse.c (scan_insn): Likewise.
1835 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1837 * config/arc/arc.c (conditionalize_nonjump): New function,
1839 (arc_ifcvt): ... this.
1840 (arc_predicate_delay_insns): Use it.
1842 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1844 * config/arc/predicates.md (extend_operand): During/after reload,
1845 allow const_int_operand.
1846 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1847 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1848 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1850 (umulsi3_highpart_i): Likewise.
1852 2014-03-11 Richard Biener <rguenther@suse.de>
1854 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1855 Add asserts to guard possible wrong-code bugs.
1857 2014-03-11 Richard Biener <rguenther@suse.de>
1859 PR tree-optimization/60429
1860 PR tree-optimization/60485
1861 * tree-ssa-structalias.c (set_union_with_increment): Properly
1862 take into account all fields that overlap the shifted vars.
1863 (do_sd_constraint): Likewise.
1864 (do_ds_constraint): Likewise.
1865 (get_constraint_for_ptr_offset): Likewise.
1867 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1869 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1870 (nios2_compute_frame_layout):
1871 Add calculation of cfun->machine->fp_save_offset.
1872 (nios2_expand_prologue): Correct setting of frame pointer register
1874 (nios2_expand_epilogue): Update recovery of stack pointer from
1875 frame pointer accordingly.
1876 (nios2_initial_elimination_offset): Update calculation of offset
1877 for eliminating to HARD_FRAME_POINTER_REGNUM.
1879 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1882 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1883 cgraph_get_create_node on VAR_DECLs.
1885 2014-03-10 Richard Biener <rguenther@suse.de>
1888 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1890 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1892 * config/vms/vms.opt (vms_float_format): New variable.
1894 2014-03-08 Tobias Burnus <burnus@net-b.de>
1896 * doc/invoke.texi (-fcilkplus): Update implementation status.
1898 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1899 Richard Biener <rguenther@suse.de>
1901 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1902 consistently accross all TUs.
1903 (run_gcc): Enable -fshort-double automatically at link at link-time
1904 and disallow override.
1906 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1909 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1910 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1911 if they can't be used.
1913 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1915 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1916 for Solaris 11/x86 ld.
1917 * configure: Regenerate.
1919 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1921 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1922 (LIB_TLS_SPEC): Save as ld_tls_libs.
1923 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1924 (HAVE_AS_IX86_TLSLDM): New test.
1925 * configure, config.in: Regenerate.
1926 * config/i386/i386.c (legitimize_tls_address): Fall back to
1927 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1928 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1929 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1930 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1932 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1934 * common.opt (fira-loop-pressure): Mark as optimization.
1936 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1938 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1939 an OpenMP mappable type.
1941 2014-03-06 Matthias Klose <doko@ubuntu.com>
1943 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1944 MULTILIB_OSDIRNAMES is not defined.
1946 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1947 Meador Inge <meadori@codesourcery.com>
1950 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1951 (arm_legitimize_address): Call legitimize_tls_address for any
1952 arm_tls_referenced_p expression, handle constant addend. Call it
1953 before testing for !TARGET_ARM.
1954 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1956 2014-03-06 Richard Biener <rguenther@suse.de>
1962 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1964 * tree-streamer.c (record_common_node): Assert we don't record
1965 nodes with type double.
1966 (preload_common_node): Skip type double, complex double and double
1967 pointer since it is now frontend dependent due to fshort-double option.
1969 2014-03-06 Richard Biener <rguenther@suse.de>
1971 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1972 or -fno-lto is specified and the linker has full plugin support.
1973 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1974 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1975 * lto-wrapper.c (merge_and_complain): Merge compile-time
1976 optimization levels.
1977 (run_gcc): And pass it through to the link options.
1979 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1983 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1985 * cselib.c (remove_useless_values): Skip to avoid quadratic
1986 behavior if the condition moved from...
1987 (cselib_process_insn): ... here holds.
1989 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1992 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1993 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1996 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1997 (TM_H): Add x86-tune.def.
1999 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2001 * config/aarch64/aarch64.c (generic_tunings):
2002 Use cortexa57_extra_costs.
2004 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2007 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2008 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2009 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2012 2014-03-04 Heiher <r@hev.cc>
2014 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
2015 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
2017 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
2019 * config/i386/predicates.md (const2356_operand): Change to ...
2020 (const2367_operand): ... this.
2021 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
2023 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2024 (*avx512pf_scatterpf<mode>sf): Ditto.
2025 (avx512pf_scatterpf<mode>df): Ditto.
2026 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2027 (*avx512pf_scatterpf<mode>df): Ditto.
2028 * config/i386/i386.c (ix86_expand_builtin): Update
2029 incorrect hint operand error message.
2031 2014-03-04 Richard Biener <rguenther@suse.de>
2033 * lto-section-in.c (lto_get_section_data): Fix const cast.
2035 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2037 * tree-streamer.c (record_common_node): Assert we don't record
2038 nodes with type double.
2039 (preload_common_node): Skip type double, complex double and double
2040 pointer since it is now frontend dependent due to fshort-double option.
2042 2014-03-04 Richard Biener <rguenther@suse.de>
2045 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
2046 (lto_input_toplevel_asms): Likewise.
2047 * lto-section-in.c (lto_get_section_data): Instead do it here
2050 2014-03-04 Richard Biener <rguenther@suse.de>
2052 PR tree-optimization/60382
2053 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
2054 dead PHIs a reduction.
2056 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
2058 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2060 (_mm_prefetch): Move out of GCC target("sse") pragma.
2061 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2062 GCC target("prfchw") pragma.
2063 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2065 * config/i386/i386.c (ix86_option_override_internal): Enable
2066 -mprfchw with -mprefetchwt1.
2068 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2070 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2073 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2075 * opts.h (CL_PCH_IGNORE): Define.
2076 * targhooks.c (option_affects_pch_p):
2077 Return false for options that have CL_PCH_IGNORE set.
2078 * opt-functions.awk: Process PchIgnore.
2079 * doc/options.texi: Document PchIgnore.
2081 * config/arc/arc.opt (misize): Add PchIgnore property.
2083 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2085 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2086 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2087 constraint on constants to permit them being loaded into
2088 GENERAL_REGS or BASE_REGS.
2090 2014-03-03 Nick Clifton <nickc@redhat.com>
2092 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2093 anti-cacnonical alternatives.
2094 (negandhi3_real): New pattern.
2095 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2097 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2099 * config/avr/avr-mcus.def: Remove atxmega16x1.
2100 * config/avr/avr-tables.opt: Regenerate.
2101 * config/avr/t-multilib: Regenerate.
2102 * doc/avr-mmcu.texi: Regenerate.
2104 2014-03-03 Tobias Grosser <tobias@grosser.es>
2105 Mircea Namolaru <mircea.namolaru@inria.fr>
2107 PR tree-optimization/58028
2108 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2111 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2113 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2114 not handled by recognizers.
2116 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2119 * function.c (expand_function_end): Don't emit
2120 clobber_return_register sequence if clobber_after is a BARRIER.
2121 * cfgexpand.c (construct_exit_block): Append instructions before
2122 return_label to prev_bb.
2124 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2126 * config/rs6000/constraints.md: Document reserved use of "wc".
2128 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2131 * ipa.c (function_and_variable_visibility): When dissolving comdat
2132 group, also set all symbols to local.
2134 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2139 2013-12-14 Jan Hubicka <jh@suse.cz>
2141 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2143 2014-03-02 Jon Beniston <jon@beniston.com>
2149 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2150 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2151 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2152 (simple_return, *simple_return): New patterns
2153 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2154 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2156 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2158 * dwarf2out.c (gen_subprogram_die): Tidy.
2160 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2163 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2164 (*mov_t_msb_neg_negc): ... this new insn.
2166 2014-02-28 Jason Merrill <jason@redhat.com>
2169 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2172 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2175 * dwarf2out.c (decltype_auto_die): New static.
2176 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2177 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2178 (is_cxx_auto): Likewise.
2180 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2182 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2183 we are not using general regs only.
2185 2014-02-28 Richard Biener <rguenther@suse.de>
2188 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2189 previous fix and only allow to remove trivial pre-headers
2190 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2191 (remove_forwarder_block): Properly update the latch of a loop.
2193 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2196 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2197 (cselib_preserved_hash_table): New.
2198 (preserve_constants_and_equivs): Move preserved vals to it.
2199 (cselib_find_slot): Look it up first.
2200 (cselib_init): Initialize it.
2201 (cselib_finish): Release it.
2202 (dump_cselib_table): Dump it.
2204 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2207 * cselib.c (remove_useless_values): Skip to avoid quadratic
2208 behavior if the condition moved from...
2209 (cselib_process_insn): ... here holds.
2211 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2214 * var-tracking.c (vt_initialize): Apply the same condition to
2215 preserve the CFA base value.
2217 2014-02-28 Joey Ye <joey.ye@arm.com>
2220 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2221 if reload in progress or completed.
2223 2014-02-28 Tobias Burnus <burnus@net-b.de>
2226 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2229 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2231 * doc/tm.texi.in (Condition Code Status): Update documention for
2232 relative locations of cc0-setter and cc0-user.
2234 2014-02-27 Jeff Law <law@redhat.com>
2236 PR rtl-optimization/52714
2237 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2238 into two independent simple sets, if I3 is a jump, ensure the
2239 pattern we place into I3 is a (set (pc) ...).
2241 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2242 Jeff Law <law@redhat.com>
2244 PR rtl-optimization/49847
2245 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2246 are in different blocks.
2247 * doc/tm.texi (Condition Code Status): Update documention for
2248 relative locations of cc0-setter and cc0-user.
2250 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2253 * lra.c (lra_emit_add): Check SUBREG too.
2255 2014-02-27 Andreas Schwab <schwab@suse.de>
2257 * config/m68k/m68k.c (m68k_option_override): Disable
2258 -flive-range-shrinkage for classic m68k.
2259 (m68k_override_options_after_change): Likewise.
2261 2014-02-27 Marek Polacek <polacek@redhat.com>
2264 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2265 -Wmaybe-uninitialized.
2267 2014-02-27 Alan Modra <amodra@gmail.com>
2270 * reload1.c (emit_input_reload_insns): When reload_override_in,
2271 set old to rl->in_reg when rl->in_reg is a subreg.
2273 2014-02-26 Richard Biener <rguenther@suse.de>
2276 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2278 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2280 * common/config/i386/predicates.md (const1256_operand): Remove.
2281 (const2356_operand): New.
2282 (const_1_to_2_operand): Remove.
2283 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2284 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2285 (*avx512pf_gatherpf<mode>sf): Ditto.
2286 (avx512pf_gatherpf<mode>df): Ditto.
2287 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2288 (*avx512pf_gatherpf<mode>df): Ditto.
2289 (avx512pf_scatterpf<mode>sf): Ditto.
2290 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2291 (*avx512pf_scatterpf<mode>sf): Ditto.
2292 (avx512pf_scatterpf<mode>df): Ditto.
2293 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2294 (*avx512pf_scatterpf<mode>df): Ditto.
2295 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2297 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2299 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2300 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2301 (_mm512_mask_testn_epi64_mask): Move to ...
2302 * config/i386/avx512cdintrin.h: Here.
2303 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2304 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2305 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2306 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2307 TARGET_AVX512F from TARGET_AVX512CD.
2309 2014-02-26 Richard Biener <rguenther@suse.de>
2312 * ipa.c (walk_polymorphic_call_targets): Properly guard
2313 call to inline_update_overall_summary.
2315 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2318 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2319 and latches only if requested. Fix latch if it is removed.
2320 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2321 LOOPS_HAVE_PREHEADERS.
2323 2014-02-25 Andrew Pinski <apinski@cavium.com>
2325 * builtins.c (expand_builtin_thread_pointer): Create a new target
2326 when the target is NULL.
2328 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2330 PR rtl-optimization/60317
2331 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2332 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2333 * lra-assigns.c: Include params.h.
2334 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2335 other reload pseudos considerations.
2337 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2339 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2340 to use canonical form for nor<mode>3.
2342 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2345 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2348 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2350 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2351 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2352 (ix86_handle_option): Handle OPT_mprefetchwt1.
2353 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2354 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2356 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2357 OPTION_MASK_ISA_PREFETCHWT1.
2358 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2359 (PTA_PREFETCHWT1): New.
2360 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2361 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2362 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2363 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2364 (*prefetch_avx512pf_<mode>_: Change into ...
2365 (*prefetch_prefetchwt1_<mode>: This.
2366 * config/i386/i386.opt (mprefetchwt1): New.
2367 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2368 (_mm_prefetch): Handle intent to write.
2369 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2371 2014-02-25 Richard Biener <rguenther@suse.de>
2374 * emit-rtl.c (mem_attrs_htab): Remove.
2375 (mem_attrs_htab_hash): Likewise.
2376 (mem_attrs_htab_eq): Likewise.
2377 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2378 (init_emit_once): Do not allocate mem_attrs_htab.
2380 2014-02-25 Richard Biener <rguenther@suse.de>
2383 * lto-opts.c (lto_write_options): Output non-explicit conservative
2384 -fwrapv, -fno-trapv and -fno-strict-overflow.
2385 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2386 (run_gcc): And pass them through.
2388 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2390 * sel-sched.c (calculate_new_fences): New parameter ptime.
2391 Calculate it as a maximum over all fence cycles.
2392 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2393 Print the final schedule timing when sched_verbose.
2395 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2397 PR rtl-optimization/60292
2398 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2399 bit fot the fence instruction.
2401 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2403 * calls.h: Fix typo in comment.
2405 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2407 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2408 adjusting offsetable addresses.
2410 2014-02-24 Guozhi Wei <carrot@google.com>
2412 * sparseset.h (sparseset_pop): Fix the wrong index.
2414 2014-02-24 Walter Lee <walt@tilera.com>
2416 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2417 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2419 * common/config/tilegx/tilegx-common.c
2420 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2421 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2423 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2424 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2425 (tilegx_gimplify_va_arg_expr): Handle big endian.
2426 (tilegx_expand_unaligned_load): Ditto.
2427 (tilegx_expand_unaligned_store): Ditto.
2428 (TARGET_RETURN_IN_MSB): New.
2429 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2430 (TARGET_ENDIAN_DEFAULT): New.
2431 (TARGET_BIG_ENDIAN): Handle big endian.
2432 (BYTES_BIG_ENDIAN): Ditto.
2433 (WORDS_BIG_ENDIAN): Ditto.
2434 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2437 * config/tilegx/tilegx.md (extv): Handle big endian.
2439 (insn_st<n>): Ditto.
2440 (insn_st<n>_add<bitsuffix>): Ditto.
2441 (insn_stnt<n>): Ditto.
2442 (insn_stnt<n>_add<bitsuffix>):Ditto.
2443 (vec_interleave_highv8qi): Handle big endian.
2444 (vec_interleave_highv8qi_be): New.
2445 (vec_interleave_highv8qi_le): New.
2446 (insn_v1int_h): Handle big endian.
2447 (vec_interleave_lowv8qi): Handle big endian.
2448 (vec_interleave_lowv8qi_be): New.
2449 (vec_interleave_lowv8qi_le): New.
2450 (insn_v1int_l): Handle big endian.
2451 (vec_interleave_highv4hi): Handle big endian.
2452 (vec_interleave_highv4hi_be): New.
2453 (vec_interleave_highv4hi_le): New.
2454 (insn_v2int_h): Handle big endian.
2455 (vec_interleave_lowv4hi): Handle big endian.
2456 (vec_interleave_lowv4hi_be): New.
2457 (vec_interleave_lowv4hi_le): New.
2458 (insn_v2int_l): Handle big endian.
2459 (vec_interleave_highv2si): Handle big endian.
2460 (vec_interleave_highv2si_be): New.
2461 (vec_interleave_highv2si_le): New.
2462 (insn_v4int_h): Handle big endian.
2463 (vec_interleave_lowv2si): Handle big endian.
2464 (vec_interleave_lowv2si_be): New.
2465 (vec_interleave_lowv2si_le): New.
2466 (insn_v4int_l): Handle big endian.
2467 * config/tilegx/tilegx.opt (mbig-endian): New option.
2468 (mlittle-endian): New option.
2469 * doc/install.texi: Document tilegxbe-linux.
2470 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2472 2014-02-24 Martin Jambor <mjambor@suse.cz>
2475 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2476 there are no parameter descriptors.
2478 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2480 PR rtl-optimization/60268
2481 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2482 initialization to ...
2483 (sched_rgn_init): ... here.
2484 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2486 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2488 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2491 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2493 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2496 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2498 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2499 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2501 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2503 * config/microblaze/predicates.md: Add cmp_op predicate.
2504 * config/microblaze/microblaze.md: Add branch_compare instruction
2505 which uses cmp_op predicate and emits cmp insn before branch.
2506 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2507 to microblaze_expand_conditional_branch and consolidate logic.
2508 (microblaze_expand_conditional_branch): emit branch_compare
2509 insn instead of handling cmp op separate from branch insn.
2511 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2513 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2516 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2518 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2519 define_insn with define_expand and new define_insn
2520 *altivec_lve<VI_char>x_internal.
2521 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2522 and new define_insn *altivec_stve<VI_char>x_internal.
2523 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2525 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2527 (altivec_expand_stvex_be): New function.
2529 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2531 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2532 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2533 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2534 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2536 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2539 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2540 instead of emit_move_insn.
2542 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2544 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2546 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2547 gen_altivec_vsumsws.
2549 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2551 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2552 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2553 (altivec_lvxl_<mode>): New define_expand incorporating
2554 -maltivec=be semantics where needed.
2555 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2556 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2557 semantics where needed.
2558 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2559 (altivec_stvx_<mode>): New define_expand incorporating
2560 -maltivec=be semantics where needed.
2561 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2562 VM2 iterator instead of V4SI.
2563 (altivec_stvxl_<mode>): New define_expand incorporating
2564 -maltivec=be semantics where needed.
2565 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2566 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2567 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2568 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2569 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2570 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2571 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2572 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2573 ALTIVEC_BUILTIN_STVXL.
2574 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2575 (altivec_expand_stvx_be): Likewise.
2576 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2577 (altivec_expand_lvx_be): Likewise.
2578 (altivec_expand_stvx_be): Likewise.
2579 (altivec_expand_builtin): Add cases for
2580 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2581 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2582 (altivec_init_builtins): Add definitions for
2583 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2584 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2586 2014-02-21 Catherine Moore <clm@codesourcery.com>
2588 * doc/invoke.texi (mvirt, mno-virt): Document.
2589 * config/mips/mips.opt (mvirt): New option.
2590 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2592 2014-02-21 Richard Biener <rguenther@suse.de>
2594 PR tree-optimization/60276
2595 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2596 (STMT_VINFO_MIN_NEG_DIST): New macro.
2597 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2598 STMT_VINFO_MIN_NEG_DIST.
2599 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2600 made for negative dependence distances still hold.
2602 2014-02-21 Richard Biener <rguenther@suse.de>
2605 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2606 DECL_INITIAL for globals not in the current function context.
2608 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2610 PR tree-optimization/56490
2611 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2612 * tree-ssa-uninit.c: Include params.h.
2613 (compute_control_dep_chain): Add num_calls argument, return false
2614 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2615 num_calls to recursive call.
2616 (find_predicates): Change dep_chain into normal array,
2617 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2618 variable and adjust compute_control_dep_chain caller.
2619 (find_def_preds): Likewise.
2621 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2623 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2624 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2626 2014-02-21 Nick Clifton <nickc@redhat.com>
2628 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2629 (pushhi1): Likewise.
2630 (popqi1): Add mode to pre_dec.
2633 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2635 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2636 mode for mask of V8SFmode permutation.
2638 2014-02-20 Richard Henderson <rth@redhat.com>
2641 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2642 a new pseudo for OLDVAL.
2644 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2647 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2648 gen_reg_rtx if d->testing_p.
2649 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2650 if d->testing_p and we will certainly return true.
2651 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2654 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2656 * emit-rtl.c (gen_reg_rtx): Assert that
2657 crtl->emit.regno_pointer_align_length is non-zero.
2659 2014-02-20 Richard Henderson <rth@redhat.com>
2662 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2663 on failure the store back into EXPECT.
2665 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2666 Sandra Loosemore <sandra@codesourcery.com>
2668 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2669 * config/nios2/nios2.c (nios2_function_profiler): Add
2670 -fPIC (flag_pic == 2) support.
2671 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2672 (nios2_large_offset_p): New function.
2673 (nios2_unspec_reloc_p): Move up position, update to use
2674 nios2_large_offset_p.
2675 (nios2_unspec_address): Remove function.
2676 (nios2_unspec_offset): New function.
2677 (nios2_large_got_address): New function.
2678 (nios2_got_address): Add large offset support.
2679 (nios2_legitimize_tls_address): Update usage of removed and new
2681 (nios2_symbol_binds_local_p): New function.
2682 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2683 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2684 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2685 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2686 processing for (const (unspec ...)).
2687 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2689 2014-02-20 Richard Biener <rguenther@suse.de>
2691 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2692 doing the substitution.
2693 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2695 2014-02-20 Martin Jambor <mjambor@suse.cz>
2698 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2699 info when checking whether lattices are bottom.
2701 2014-02-20 Richard Biener <rguenther@suse.de>
2704 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2707 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2710 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2711 parameter specifying the scaling.
2712 (inline_call): Update.
2713 (want_inline_recursively): Guard division by zero.
2714 (recursive_inlining): Update.
2715 * ipa-inline.h (clone_inlined_nodes): Update.
2717 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2720 * config/i386/i386.c (classify_argument): Pass structures of size
2721 64 bytes or less in register.
2723 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2724 Kirill Yukhin <kirill.yukhin@intel.com>
2726 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2727 (_mm_rcp28_round_ss): Ditto.
2728 (_mm_rsqrt28_round_sd): Ditto.
2729 (_mm_rsqrt28_round_ss): Ditto.
2730 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2731 (_mm_rcp14_round_ss): Ditto.
2732 (_mm_rsqrt14_round_sd): Ditto.
2733 (_mm_rsqrt14_round_ss): Ditto.
2734 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2735 the first input operand, get rid of match_dup.
2736 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2738 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2740 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2741 operand as the first input operand, set type attribute.
2742 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2744 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2745 operand as the first input operand, set type attribute.
2747 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2749 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2752 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2755 * config/i386/i386.c (construct_container): Remove TFmode check
2756 for X86_64_INTEGER_CLASS.
2758 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2761 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2762 only when -Wpsabi is enabled.
2764 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2767 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2768 passing arrays in registers are the same as for structs, so remove the
2769 special case for them.
2771 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2773 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2774 destination type, extract only the valid bits if the source type is not
2775 integral and has a different mode.
2777 2014-02-19 Richard Biener <rguenther@suse.de>
2780 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2783 2014-02-19 Richard Biener <rguenther@suse.de>
2786 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2787 (ipa_modify_call_arguments): Emit an argument load explicitely and
2788 preserve virtual SSA form there and for the replacement call.
2789 Do not update SSA form nor free dominance info.
2791 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2793 * ipa.c (function_and_variable_visibility): Also clear WEAK
2794 flag when disolving COMDAT_GROUP.
2796 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2798 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2799 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2801 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2802 do more sanity checks.
2803 (detect_type_change): Return true when giving up early.
2804 (compute_complex_assign_jump_func): Fix type parameter of
2805 ipa_set_ancestor_jf.
2806 (compute_complex_ancestor_jump_func): Likewise.
2807 (update_jump_functions_after_inlining): Fix updating of
2809 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2811 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2813 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2814 inline clones when edge disappears.
2816 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2819 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2820 Split 64-bit moves into 2 patterns. Do not allow the use of
2821 direct move for TDmode in little endian, since the decimal value
2822 has little endian bytes within a word, but the 64-bit pieces are
2823 ordered in a big endian fashion, and normal subreg's of TDmode are
2825 (mov<mode>_64bit_dm): Likewise.
2826 (movtd_64bit_nodm): Likewise.
2828 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2830 PR tree-optimization/60174
2831 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2832 statement of an SSA_NAME that occurs in an abnormal PHI node.
2834 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2837 * final.c (SEEN_BB): Remove.
2838 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2839 (final_scan_insn): Don't force_source_line on second
2840 NOTE_INSN_BASIC_BLOCK.
2842 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2845 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2846 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2847 (type_natural_mode): Warn ABI change when %zmm register is not
2848 available for AVX512F vector value passing.
2850 2014-02-18 Kai Tietz <ktietz@redhat.com>
2853 * config/i386/i386.c (ix86_expand_prologue): Use value in
2854 rax register as displacement when restoring %r10 or %rax.
2855 Fix wrong offset when restoring both registers.
2857 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2859 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2860 assertion with conditional return.
2862 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2863 Uros Bizjak <ubizjak@gmail.com>
2866 * config/i386/driver-i386.c (host_detect_local_cpu): If
2867 YMM state is not saved by the OS, also clear has_f16c. Move
2868 CPUID 0x80000001 handling before YMM state saving checking.
2870 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2872 PR rtl-optimization/58960
2873 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2874 factored out from ...
2875 (sched_init): ... here.
2876 (free_global_sched_pressure_data): New, factored out from ...
2877 (sched_finish): ... here.
2878 * sched-int.h (free_global_sched_pressure_data): Declare.
2879 * sched-rgn.c (nr_regions_initial): New static global.
2880 (haifa_find_rgns): Initialize it.
2881 (schedule_region): Disable sched-pressure for the newly
2884 2014-02-17 Richard Biener <rguenther@suse.de>
2886 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2887 release SSA defs of pattern stmts.
2889 2014-02-17 Richard Biener <rguenther@suse.de>
2891 * tree-inline.c (expand_call_inline): Release the virtual
2892 operand defined by the call we are about to inline.
2894 2014-02-17 Richard Biener <rguenther@suse.de>
2896 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2898 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2899 Ilya Tocar <ilya.tocar@intel.com>
2901 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2902 arguments order in builtin.
2903 (_mm512_permutexvar_epi64): Ditto.
2904 (_mm512_mask_permutexvar_epi64): Ditto
2905 (_mm512_maskz_permutexvar_epi32): Ditto
2906 (_mm512_permutexvar_epi32): Ditto
2907 (_mm512_mask_permutexvar_epi32): Ditto
2909 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2911 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2912 (p8_vmrgow): Likewise.
2914 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2916 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2919 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2922 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2923 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2924 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2925 using direct move instructions on ISA 2.07. Also adjust
2926 instruction length for 64-bit.
2927 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2928 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2930 2014-02-15 Alan Modra <amodra@gmail.com>
2934 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2935 find_replacement on parts of insn rtl that might be reloaded.
2937 2014-02-15 Richard Biener <rguenther@suse.de>
2939 PR tree-optimization/60183
2940 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2941 (tree_ssa_phiprop): Calculate and free post-dominators.
2943 2014-02-14 Jeff Law <law@redhat.com>
2945 PR rtl-optimization/60131
2946 * ree.c (get_extended_src_reg): New function.
2947 (combine_reaching_defs): Use it rather than assuming location of REG.
2948 (find_and_remove_re): Verify first operand of extension is
2949 a REG before adding the insns to the copy list.
2951 2014-02-14 Roland McGrath <mcgrathr@google.com>
2953 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2954 * configure: Regenerated.
2955 * config.in: Regenerated.
2956 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2957 instead of ASM_SHORT.
2959 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2960 Richard Earnshaw <rearnsha@arm.com>
2962 PR rtl-optimization/59535
2963 * lra-constraints.c (process_alt_operands): Encourage alternative
2964 when unassigned pseudo class is superset of the alternative class.
2965 (inherit_reload_reg): Don't inherit when optimizing for code size.
2966 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2967 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2968 modes not less than 4 for Thumb1.
2970 2014-02-14 Kyle McMartin <kyle@redhat.com>
2973 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2975 2014-02-14 Richard Biener <rguenther@suse.de>
2977 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2978 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2979 Do not strip INDIRECT_REFs.
2981 2014-02-14 Richard Biener <rguenther@suse.de>
2984 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2985 DECL_FUNCTION_SPECIFIC_TARGET.
2986 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2987 * tree-streamer-out.c (pack_ts_target_option): Remove.
2988 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2989 (write_ts_function_decl_tree_pointers): Do not stream
2990 DECL_FUNCTION_SPECIFIC_TARGET.
2991 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2992 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2993 (lto_input_ts_function_decl_tree_pointers): Do not stream
2994 DECL_FUNCTION_SPECIFIC_TARGET.
2996 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2998 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2999 (get_initial_def_for_induction, vectorizable_induction): Ignore
3000 debug stmts when looking for exit_phi.
3001 (vectorizable_live_operation): Fix up condition.
3003 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3005 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3006 nreverse() because it changes the content of original tree list.
3008 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3010 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3011 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3013 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3015 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
3016 GNU coding standards.
3018 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3021 * dwarf2out.c (gen_subprogram_die): Don't call
3022 add_calling_convention_attribute if subr_die is old_die.
3024 2014-02-13 Sharad Singhai <singhai@google.com>
3026 * doc/optinfo.texi: Fix order of nodes.
3028 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3030 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
3031 operands[2], not operands[3].
3033 2014-02-13 Richard Biener <rguenther@suse.de>
3036 * doc/install.texi (ISL): Update recommended version to 0.12.2,
3037 mention the possibility of an in-tree build.
3038 (CLooG): Update recommended version to 0.18.1, mention the
3039 possibility of an in-tree build and clarify that the ISL
3040 bundled with CLooG does not work.
3042 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3045 * expr.c (compress_float_constant): If x is a hard register,
3046 extend into a pseudo and then move to x.
3048 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
3050 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
3051 caused by bad second argument to warning_at() with -mhotpatch and
3052 nested functions (e.g. with gfortran).
3054 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
3056 * opts.c (option_name): Remove "enabled by default" rider.
3058 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3060 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3062 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3063 Uros Bizjak <ubizjak@gmail.com>
3066 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3067 * configure: Regenerated.
3069 2014-02-12 Richard Biener <rguenther@suse.de>
3071 * vec.c (vec_prefix::calculate_allocation): Move as
3072 inline variant to vec.h.
3073 (vec_prefix::calculate_allocation_1): New out-of-line version.
3074 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3075 (vec_prefix::m_has_auto_buf): Rename to ...
3076 (vec_prefix::m_using_auto_storage): ... this.
3077 (vec_prefix::calculate_allocation): Inline the easy cases
3078 and dispatch to calculate_allocation_1 which doesn't need the
3080 (va_heap::reserve): Use gcc_checking_assert.
3081 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3082 m_using_auto_storage.
3083 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3085 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3086 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3087 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3089 2014-02-12 Richard Biener <rguenther@suse.de>
3091 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3092 when we found a dependence.
3094 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3096 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3098 (maybe_fold_stmt): ... into this new function.
3099 * omp-low.c (lower_omp): Update comment.
3101 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3104 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3107 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3109 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3110 identifiers in comments.
3111 (cortexa53_extra_costs): Likewise.
3112 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3113 (cortexa7_extra_costs): Likewise.
3114 (cortexa12_extra_costs): Likewise.
3115 (cortexa15_extra_costs): Likewise.
3116 (v7m_extra_costs): Likewise.
3118 2014-02-12 Richard Biener <rguenther@suse.de>
3121 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3122 of posix_memalign being successful.
3123 (lower_stmt): Restrict lowering of posix_memalign to when
3124 -ftree-bit-ccp is enabled.
3126 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3128 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3130 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3132 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3134 PR rtl-optimization/60116
3135 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3136 other_insn once the combination has been validated.
3138 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3141 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3143 * ipa-devirt.c: Include demangle.h
3144 (odr_violation_reported): New static variable.
3145 (add_type_duplicate): Update odr_violations.
3146 (maybe_record_node): Add completep parameter; update it.
3147 (record_target_from_binfo): Add COMPLETEP parameter;
3148 update it as needed.
3149 (possible_polymorphic_call_targets_1): Likewise.
3150 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3151 rename FINAL to COMPLETE.
3152 (record_targets_from_bases): Sanity check we found the binfo;
3153 fix COMPLETEP updating.
3154 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3155 parameter, fix computing of COMPLETEP.
3156 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3157 at LTO time do demangling.
3158 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3159 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3161 (gimple_get_virt_method_for_binfo): Likewise.
3162 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3163 gimple_get_virt_method_for_vtable): Update prototypes.
3165 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3168 * genautomata.c (add_presence_absence): Fix typo with
3169 {final_}presence_list.
3171 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3174 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3175 for VSX/Altivec vectors that land in GPR registers.
3177 2014-02-11 Richard Henderson <rth@redhat.com>
3178 Jakub Jelinek <jakub@redhat.com>
3181 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3182 around drhs if type conversion to lacc->type is not useless.
3184 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3186 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3188 (cortex-a57.cortex-a53): Likewise.
3189 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3191 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3193 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3196 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3198 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3199 add_options_for_arm_vfp3.
3201 2014-02-11 Jeff Law <law@redhat.com>
3204 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3205 object with an undesirable mode.
3207 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3210 * config/i386/sol2-9.h: New file.
3211 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3212 *-*-solaris2.9*): Use it.
3214 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3216 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3217 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3219 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3221 * config/microblaze/microblaze.c: Extend mcpu version format
3223 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3225 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3227 2014-02-10 Richard Henderson <rth@redhat.com>
3230 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3231 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3232 ms-abi vs -mno-accumulate-outgoing-args.
3233 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3234 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3237 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3240 * cfgexpand.c (expand_asm_operands): Attach source location to
3241 ASM_INPUT rtx objects.
3242 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3244 2014-02-10 Nick Clifton <nickc@redhat.com>
3246 * config/mn10300/mn10300.c (popcount): New function.
3247 (mn10300_expand_prologue): Include saved registers in stack usage
3250 2014-02-10 Jeff Law <law@redhat.com>
3253 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3254 when changing the SET_DEST of a prior insn to avoid an input reload.
3256 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3258 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3259 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3260 -mcall-openbsd, or -mcall-linux.
3261 (CC1_ENDIAN_BIG_SPEC): Remove.
3262 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3263 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3264 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3265 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3266 and %cc1_endian_default.
3267 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3269 2014-02-10 Richard Biener <rguenther@suse.de>
3271 PR tree-optimization/60115
3272 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3273 MEM_REF handling. Properly verify that the accesses are not
3274 out of the objects bound.
3276 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3278 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3281 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3283 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3284 proper constants and fix formatting.
3285 (possible_polymorphic_call_targets): Fix formatting.
3287 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3288 Ilya Tocar <ilya.tocar@intel.com>
3290 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3291 (_mm512_loadu_epi32): Renamed into...
3292 (_mm512_loadu_si512): This.
3293 (_mm512_storeu_epi32): Renamed into...
3294 (_mm512_storeu_si512): This.
3295 (_mm512_maskz_ceil_ps): Removed.
3296 (_mm512_maskz_ceil_pd): Ditto.
3297 (_mm512_maskz_floor_ps): Ditto.
3298 (_mm512_maskz_floor_pd): Ditto.
3299 (_mm512_floor_round_ps): Ditto.
3300 (_mm512_floor_round_pd): Ditto.
3301 (_mm512_ceil_round_ps): Ditto.
3302 (_mm512_ceil_round_pd): Ditto.
3303 (_mm512_mask_floor_round_ps): Ditto.
3304 (_mm512_mask_floor_round_pd): Ditto.
3305 (_mm512_mask_ceil_round_ps): Ditto.
3306 (_mm512_mask_ceil_round_pd): Ditto.
3307 (_mm512_maskz_floor_round_ps): Ditto.
3308 (_mm512_maskz_floor_round_pd): Ditto.
3309 (_mm512_maskz_ceil_round_ps): Ditto.
3310 (_mm512_maskz_ceil_round_pd): Ditto.
3311 (_mm512_expand_pd): Ditto.
3312 (_mm512_expand_ps): Ditto.
3313 * config/i386/i386.c (ix86_builtins): Remove
3314 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3315 (bdesc_args): Ditto.
3316 * config/i386/predicates.md (const1256_operand): New.
3317 (const_1_to_2_operand): Ditto.
3318 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3319 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3320 (*avx512pf_gatherpf<mode>sf): Ditto.
3321 (avx512pf_gatherpf<mode>df): Ditto.
3322 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3323 (*avx512pf_gatherpf<mode>df): Ditto.
3324 (avx512pf_scatterpf<mode>sf): Ditto.
3325 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3326 (*avx512pf_scatterpf<mode>sf): Ditto.
3327 (avx512pf_scatterpf<mode>df): Ditto.
3328 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3329 (*avx512pf_scatterpf<mode>df): Ditto.
3330 (avx512f_expand<mode>): Removed.
3331 (<shift_insn><mode>3<mask_name>): Change predicate type.
3333 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3335 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3336 not at the end of datarefs vector use ordered_remove to avoid
3337 reordering datarefs vector.
3340 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3341 mark local addressable non-static vars as GOVD_PRIVATE
3342 instead of GOVD_LOCAL.
3343 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3344 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3348 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3349 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3350 assume_aligned or alloc_align attributes.
3351 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3352 arguments. Handle also assume_aligned and alloc_align attributes.
3353 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3354 calls to functions with assume_aligned or alloc_align attributes.
3355 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3357 2014-02-08 Terry Guo <terry.guo@arm.com>
3359 * doc/invoke.texi: Document ARM -march=armv7e-m.
3361 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3363 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3364 flag on __cilkrts_rethrow builtin.
3367 * ipa-cp.c (determine_versionability): Fail at -O0
3368 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3369 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3372 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3375 * tree-inline.c (copy_forbidden): Fail for
3376 __attribute__((optimize (0))) functions.
3378 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3380 * varpool.c: Include pointer-set.h.
3381 (varpool_remove_unreferenced_decls): Variables in other partitions
3382 will not be output; be however careful to not lose information
3385 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3387 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3388 lookup in the vtable constructor.
3390 2014-02-07 Jeff Law <law@redhat.com>
3393 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3394 define_insn_and_split.
3396 * ipa-inline.c (inline_small_functions): Fix typos.
3398 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3400 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3401 (s390_can_use_return_insn): Declare.
3402 * config/s390/s390.h (EPILOGUE_USES): Define.
3403 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3405 (s390_chunkify_start): Handle return JUMP_LABELs.
3406 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3407 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3408 (s390_can_use_return_insn): New functions.
3409 (s390_fix_long_loop_prediction): Handle conditional returns.
3410 (TARGET_SET_UP_BY_PROLOGUE): Define.
3411 * config/s390/s390.md (ANY_RETURN): New code iterator.
3412 (*creturn, *csimple_return, return, simple_return): New patterns.
3414 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3416 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3417 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3418 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3419 REG_CFA_RESTORE list when deciding not to restore a register.
3421 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3423 * config/s390/s390.c: Include tree-pass.h and context.h.
3424 (s390_early_mach): New function, split out from...
3425 (s390_emit_prologue): ...here.
3426 (pass_data_s390_early_mach): New pass structure.
3427 (pass_s390_early_mach): New class.
3428 (s390_option_override): Create and register early_mach pass.
3429 Move to end of file.
3431 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3433 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3434 to match for the exit block.
3436 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3438 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3439 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3440 Reject misaligned operands.
3442 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3444 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3446 2014-02-07 Richard Biener <rguenther@suse.de>
3449 * gimple-low.c (lower_builtin_posix_memalign): New function.
3450 (lower_stmt): Call it to lower posix_memalign in a way
3451 to make alignment info accessible.
3453 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3456 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3457 __builtin_setjmp_receiver.
3459 2014-02-07 Richard Biener <rguenther@suse.de>
3462 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3463 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3464 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3465 Handle BUILT_IN_POSIX_MEMALIGN.
3466 (find_func_clobbers): Likewise.
3467 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3468 (call_may_clobber_ref_p_1): Likewise.
3470 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3473 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3476 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3479 * lto-cgraph.c (lto_output_node): Use
3480 symtab_get_symbol_partitioning_class.
3481 (lto_output_varpool_node): likewise.
3482 (symtab_get_symbol_partitioning_class): Move here from
3484 * cgraph.h (symbol_partitioning_class): Likewise.
3485 (symtab_get_symbol_partitioning_class): Declare.
3487 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3489 * ggc.h (ggc_internal_cleared_alloc): New macro.
3490 * vec.h (vec_safe_copy): Handle memory stats.
3491 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3492 * target-globals.c (save_target_globals): Likewise.
3494 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3497 * expr.c (emit_move_resolve_push): Export; be bit more selective
3498 on when to clear alias set.
3499 * expr.h (emit_move_resolve_push): Declare.
3500 * function.h (struct function): Add tail_call_marked.
3501 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3502 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3503 * config/i386/i386.md (TImode move expander): De not call
3505 (FP push expanders): Preserve memory attributes.
3506 * config/i386/sse.md (push<mode>1): Remove.
3507 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3508 (ix86_expand_push): Remove.
3509 * config/i386/mmx.md (push<mode>1): Remove.
3511 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3513 PR rtl-optimization/60030
3514 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3515 lopart with paradoxical subreg before shifting it up by hprec.
3517 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3519 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3520 Remove extra newline at end of file.
3521 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3522 (arm_issue_rate): Handle cortexa57.
3523 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3524 (cortex-a57.cortex-a53): Likewise.
3526 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3529 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3530 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3532 (arm_expand_prologue): Adjust all callers.
3533 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3534 info, registers also at the lowest numbered registers side. Use
3535 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3539 * var-tracking.c (adjust_mems): Before adding a SET to
3540 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3542 2014-02-06 Alan Modra <amodra@gmail.com>
3545 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3546 change SDmode to DDmode when lra_in_progress.
3548 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3551 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3552 free_data_ref on the dr first, and before goto again also set dr
3553 to the next dr. For simd_lane_access, free old datarefs[i] before
3554 overwriting it. For get_vectype_for_scalar_type failure, don't
3555 free_data_ref if simd_lane_access.
3557 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3560 * tree.h (opts_for_fn): New inline function.
3561 (opt_for_fn): Define.
3562 * config/i386/i386.c (ix86_function_regparm): Use
3563 opt_for_fn (decl, optimize) instead of optimize.
3565 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3567 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3568 for SYMBOL_REF in large memory model.
3570 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3572 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3574 (cortex-a57): Likewise.
3575 (cortex-a57.cortex-a53): Likewise.
3577 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3578 Kugan Vivekanandarajah <kuganv@linaro.org>
3580 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3582 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3584 2014-02-06 Richard Biener <rguenther@suse.de>
3586 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3587 set_loop_copy and initialize_original_copy_tables.
3589 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3591 * config/aarch64/aarch64-simd.md
3592 (aarch64_ashr_simddi): Change QI to SI.
3594 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3595 Jakub Jelinek <jakub@redhat.com>
3598 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3601 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3603 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3604 CODE_FOR_altivec_vpku[hw]um to
3605 CODE_FOR_altivec_vpku[hw]um_direct.
3606 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3607 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3608 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3609 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3611 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3613 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3614 generation for -maltivec=be.
3615 (altivec_vsumsws): Simplify redundant test.
3617 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3619 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3620 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3621 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3622 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3623 gen_altivec_vpkuwum.
3624 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3626 (altivec_vpks<VI_char>ss): Likewise.
3627 (altivec_vpks<VI_char>us): Likewise.
3628 (altivec_vpku<VI_char>us): Likewise.
3629 (altivec_vpku<VI_char>um): Likewise.
3630 (altivec_vpku<VI_char>um_direct): New (copy of
3631 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3633 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3634 target is little endian and -maltivec=be is not specified.
3635 (*altivec_vupkhs<VU_char>_direct): New (copy of
3636 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3637 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3638 target is little endian and -maltivec=be is not specified.
3639 (*altivec_vupkls<VU_char>_direct): New (copy of
3640 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3641 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3642 little endian and -maltivec=be is not specified.
3643 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3644 little endian and -maltivec=be is not specified.
3646 2014-02-05 Richard Henderson <rth@redhat.com>
3649 * combine-stack-adj.c: Revert r206943.
3650 * sched-int.h (struct deps_desc): Add last_args_size.
3651 * sched-deps.c (init_deps): Initialize it.
3652 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3653 contain REG_ARGS_SIZE notes.
3655 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3657 * lto-cgraph.c (asm_nodes_output): Make global.
3658 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3659 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3660 (driver_handle_option): Handle OPT_fwpa.
3662 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3665 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3666 a comment typo and formatting issue. If odr_hash hasn't been
3667 created, return vNULL and set *completep to false.
3670 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3671 bb with no successors.
3673 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3676 * doc/invoke.texi (-march): Clarify documentation for ARM.
3680 2014-02-05 Richard Biener <rguenther@suse.de>
3682 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3683 when not vectorizing because of too many alias checks.
3684 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3685 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3687 2014-02-05 Nick Clifton <nickc@redhat.com>
3689 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3690 accept extended registers in any mode when compiling for the MN10300.
3692 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3694 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3695 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3696 sanitization attributes.
3697 (can_inline_edge_p): Likewise.
3698 (sanitize_attrs_match_for_inline_p): New function.
3700 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3702 * ipa-prop.c (detect_type_change): Shor circuit testing of
3703 type changes on THIS pointer.
3705 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3708 * config/pa/pa.c (legitimize_tls_address): Return original address
3709 if not passed a SYMBOL_REF rtx.
3710 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3712 (pa_emit_move_sequence): Simplify TLS source operands.
3713 (pa_legitimate_constant_p): Reject all TLS constants.
3714 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3715 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3717 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3719 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3720 groups when we know they are controlled by LTO.
3721 * varasm.c (default_binds_local_p_1): If object is in other partition,
3722 it will be resolved locally.
3724 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3726 * config/host-linux.c (linux_gt_pch_use_address): Don't
3727 use SSIZE_MAX because it is not always defined.
3729 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3732 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3733 threshold for pseudo splitting.
3734 (update_ebb_live_info): Process call argument hard registers and
3735 hard registers from insn definition too.
3736 (max_small_class_regs_num): New constant.
3737 (inherit_in_ebb): Update live hard regs through EBBs. Update
3738 reloads_num only for small register classes. Don't split for
3741 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3744 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3747 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3749 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3752 2014-02-04 Marek Polacek <polacek@redhat.com>
3754 * gdbinit.in (pel): Define.
3756 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3758 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3761 2014-02-04 Richard Biener <rguenther@suse.de>
3764 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3765 in function context local.
3766 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3767 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3768 similar to LTO_imported_decl_ref.
3770 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3772 PR tree-optimization/60002
3773 * cgraphclones.c (build_function_decl_skip_args): Clear
3776 PR tree-optimization/60023
3777 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3778 false to gsi_replace.
3779 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3780 has been in some EH region and vec_stmt could throw, add
3781 vec_stmt into the same EH region.
3782 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3783 has no lhs, ignore it.
3784 * internal-fn.c (expand_MASK_LOAD): Likewise.
3787 * tree-inline.c (copy_forbidden): Fail for
3788 __attribute__((optimize (0))) functions.
3791 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3792 is set, copy one less argument.
3793 (expand_simd_clones): Don't subtract clone_info->inbranch
3794 from simd_clone_struct_alloc argument.
3796 PR rtl-optimization/57915
3797 * recog.c (simplify_while_replacing): If all unary/binary/relational
3798 operation arguments are constant, attempt to simplify those.
3801 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3802 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3804 2014-02-04 Richard Biener <rguenther@suse.de>
3806 PR tree-optimization/60012
3807 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3808 TBAA disambiguation to all DDRs.
3810 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3813 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3814 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3816 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3819 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3821 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3823 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3824 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3826 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3829 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3830 to figure out targets of polymorphic calls with known decl.
3831 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3832 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3833 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3834 (get_polymorphic_call_info): ... here.
3835 (get_polymorphic_call_info_from_invariant): New function.
3837 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3839 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3840 lookup via vtable pointer; check for type consistency
3841 and turn inconsitent facts into UNREACHABLE.
3842 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3843 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3844 type inconsistent querries; return UNREACHABLE instead.
3846 2014-02-03 Richard Henderson <rth@twiddle.net>
3849 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3850 already processed this node.
3851 (normalize_one_pred_1): Pass along mark_set.
3852 (normalize_one_pred): Create and destroy a pointer_set_t.
3853 (normalize_one_pred_chain): Likewise.
3855 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3857 PR gcov-profile/58602
3858 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3860 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3863 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3864 -fno-devirtualize; try to devirtualize by the knowledge of
3865 virtual table pointer given by aggregate propagation.
3866 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3867 (ipa_print_node_jump_functions): Dump also offset that
3868 is relevant for polymorphic calls.
3869 (determine_known_aggregate_parts): Add arg_type parameter; use it
3870 instead of determining the type from pointer type.
3871 (ipa_compute_jump_functions_for_edge): Update call of
3872 determine_known_aggregate_parts.
3873 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3874 (gimple_get_virt_method_for_binfo): ... here; simplify using
3875 vtable_pointer_value_to_vtable.
3876 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3877 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3878 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3879 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3881 (vtable_pointer_value_to_binfo): ... here.
3882 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3884 2014-02-03 Teresa Johnson <tejohnson@google.com>
3886 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3887 redef of outer loop index variable.
3889 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3893 * doc/extend.texi (Function Attributes): Typo.
3895 2014-02-03 Cong Hou <congh@google.com>
3897 PR tree-optimization/60000
3898 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3899 if the vectorized statement is a store. A store statement can only
3900 appear at the end of pattern statements.
3902 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3904 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3905 (ix86_option_override_internal): Default long double to 64-bit for
3906 32-bit Bionic and to 128-bit for 64-bit Bionic.
3908 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3909 TARGET_LONG_DOUBLE_128 is true.
3910 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3912 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3913 (mlong-double-64): Negate -mlong-double-128.
3914 (mlong-double-128): New option.
3916 * config/i386/i386-c.c (ix86_target_macros): Define
3917 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3919 * doc/invoke.texi: Document -mlong-double-128.
3921 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3923 PR rtl-optimization/60024
3924 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3926 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3928 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3930 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3932 PR rtl-optimization/57662
3933 * sel-sched.c (code_motion_path_driver): Do not mark already not
3934 existing blocks in the visiting bitmap.
3936 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3938 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3939 on the insn being emitted.
3941 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3942 Will Deacon <will.deacon@arm.com>
3944 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3946 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3948 * config/arm/arm-tables.opt: Regenerate.
3950 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3952 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3953 for vector types other than V16QImode.
3954 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3955 define_expand, and call altivec_expand_vec_perm_le when producing
3956 code with little endian element order.
3957 (*altivec_vperm_<mode>_internal): New insn having previous
3958 behavior of altivec_vperm_<mode>.
3959 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3960 altivec_expand_vec_perm_le when producing code with little endian
3962 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3963 behavior of altivec_vperm_<mode>_uns.
3965 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3967 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3968 (altivec_vsumsws): Add handling for -maltivec=be with a little
3970 (altivec_vsumsws_direct): New.
3971 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3972 gen_altivec_vsumsws.
3974 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3976 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3977 vtable_pointer_value_to_binfo): New functions.
3978 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3979 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3981 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3983 * config/nios2/nios2.md (load_got_register): Initialize GOT
3984 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3985 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3987 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3989 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3990 preserverd by passthrough, do not propagate the type.
3992 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3994 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3995 (mips_atomic_assign_expand_fenv): New function.
3996 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3998 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4000 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4001 (__builtin_mips_set_fcsr): Likewise.
4002 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4003 MIPS_USI_FTYPE_VOID.
4004 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4005 (mips16_expand_set_fcsr): Likewise.
4006 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4007 (mips16_set_fcsr_stub): Likewise.
4008 (mips16_get_fcsr_one_only_stub): New class.
4009 (mips16_set_fcsr_one_only_stub): Likewise.
4010 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4011 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4012 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4013 (hard_float): New availability predicate.
4014 (mips_builtins): Add get_fcsr and set_fcsr.
4015 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
4016 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
4017 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
4018 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
4019 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
4022 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4024 * config/mips/mips.c (mips_one_only_stub): New class.
4025 (mips_need_mips16_rdhwr_p): Replace with...
4026 (mips16_rdhwr_stub): ...this new variable.
4027 (mips16_stub_call_address): New function.
4028 (mips16_rdhwr_one_only_stub): New class.
4029 (mips_expand_thread_pointer): Use mips16_stub_call_address.
4030 (mips_output_mips16_rdhwr): Delete.
4031 (mips_finish_stub): New function.
4032 (mips_code_end): Use it to handle rdhwr stubs.
4034 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
4037 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
4038 when calculating size of integer atomic types.
4040 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
4042 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
4044 2014-02-01 Jakub Jelinek <jakub@redhat.com>
4046 PR tree-optimization/60003
4047 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
4048 * profile.c (branch_prob): Use gimple_call_builtin_p
4049 to check for BUILT_IN_SETJMP_RECEIVER.
4050 * tree-inline.c (copy_bb): Call notice_special_calls.
4052 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
4055 * lra-constraints.c (process_alt_operands): Update reload_sum only
4058 2014-01-31 Richard Henderson <rth@redhat.com>
4061 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4062 until after else_eh is processed.
4064 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4066 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4067 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4068 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4069 in smmintrin.h, remove them.
4070 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4071 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4072 * config/i386/i386.md (ROUND_SAE): Fix value.
4073 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4074 (const48_operand): New.
4075 * config/i386/subst.md (round), (round_expand): Use
4076 const_4_or_8_to_11_operand.
4077 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4079 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4081 * config/i386/constraints.md (Yk): Swap meaning with k.
4082 * config/i386/i386.md (movhi_internal): Change Yk to k.
4083 (movqi_internal): Ditto.
4084 (*k<logic><mode>): Ditto.
4087 (kandn<mode>): Ditto.
4088 (*<code>hi_1): Ditto.
4089 (*<code>qi_1): Ditto.
4090 (kxnor<mode>): Ditto.
4091 (kortestzhi): Ditto.
4092 (kortestchi): Ditto.
4094 (*one_cmplhi2_1): Ditto.
4095 (*one_cmplqi2_1): Ditto.
4096 * config/i386/sse.md (): Change k to Yk.
4097 (avx512f_load<mode>_mask): Ditto.
4098 (avx512f_blendm<mode>): Ditto.
4099 (avx512f_store<mode>_mask): Ditto.
4100 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4101 (avx512f_storedqu<mode>_mask): Ditto.
4102 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4104 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4105 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4106 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4107 (avx512f_maskcmp<mode>3): Ditto.
4108 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4109 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4110 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4111 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4112 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4113 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4114 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4115 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4116 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4117 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4118 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4119 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4120 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4121 (vec_extract_lo_<mode>_maskm): Ditto.
4122 (vec_extract_hi_<mode>_maskm): Ditto.
4123 (avx512f_vternlog<mode>_mask): Ditto.
4124 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4125 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4126 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4127 (avx512f_<code>v8div16qi2_mask): Ditto.
4128 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4129 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4130 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4131 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4132 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4133 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4134 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4135 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4136 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4137 (avx512cd_maskb_vec_dupv8di): Ditto.
4138 (avx512cd_maskw_vec_dupv16si): Ditto.
4139 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4140 (avx512f_vpermi2var<mode>3_mask): Ditto.
4141 (avx512f_vpermi2var<mode>3_mask): Ditto.
4142 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4143 (*avx512f_gathersi<mode>): Ditto.
4144 (*avx512f_gathersi<mode>_2): Ditto.
4145 (*avx512f_gatherdi<mode>): Ditto.
4146 (*avx512f_gatherdi<mode>_2): Ditto.
4147 (*avx512f_scattersi<mode>): Ditto.
4148 (*avx512f_scatterdi<mode>): Ditto.
4149 (avx512f_compress<mode>_mask): Ditto.
4150 (avx512f_compressstore<mode>_mask): Ditto.
4151 (avx512f_expand<mode>_mask): Ditto.
4152 * config/i386/subst.md (mask): Change k to Yk.
4153 (mask_scalar_merge): Ditto.
4156 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4158 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4160 2014-01-31 Richard Biener <rguenther@suse.de>
4163 * builtins.c (fold_builtin_memory_op): Make sure to not
4164 use a floating-point mode or a boolean or enumeral type for
4167 2014-01-30 DJ Delorie <dj@redhat.com>
4169 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4170 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4171 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4172 whenever main() has an epilogue.
4174 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4176 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4177 unused variable "field".
4178 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4179 (vsx_mergeh_<mode>): Likewise.
4180 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4181 (altivec_vmrghh): Likewise.
4182 (altivec_vmrghw): Likewise.
4183 (altivec_vmrglb): Likewise.
4184 (altivec_vmrglh): Likewise.
4185 (altivec_vmrglw): Likewise.
4186 (altivec_vspltb): Add missing uses.
4187 (altivec_vsplth): Likewise.
4188 (altivec_vspltw): Likewise.
4189 (altivec_vspltsf): Likewise.
4191 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4194 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4195 frame related instructions.
4197 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4199 PR rtl-optimization/59959
4200 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4201 any reload of register whose subreg is invalid.
4203 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4205 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4206 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4207 Add missing return type - void.
4209 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4211 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4212 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4213 remove element index adjustment for endian (now handled in vsx.md
4215 (altivec_expand_vec_perm_const): Use
4216 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4217 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4218 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4219 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4220 define_expand and a new define_insn *altivec_vspltb_internal;
4221 adjust for -maltivec=be on a little endian target.
4222 (altivec_vspltb_direct): New.
4223 (altivec_vsplth): Divide into a define_expand and a new
4224 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4225 little endian target.
4226 (altivec_vsplth_direct): New.
4227 (altivec_vspltw): Divide into a define_expand and a new
4228 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4229 little endian target.
4230 (altivec_vspltw_direct): New.
4231 (altivec_vspltsf): Divide into a define_expand and a new
4232 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4233 a little endian target.
4235 2014-01-30 Richard Biener <rguenther@suse.de>
4237 PR tree-optimization/59993
4238 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4239 can propagate form the earlier stmt and avoid the transform
4240 when the intermediate result is needed.
4242 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4244 * README.Portability: Fix typo.
4246 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4248 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4249 comparison_operator with ordered_comparison_operator.
4251 2014-01-30 Nick Clifton <nickc@redhat.com>
4253 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4254 Rename to mn10300_store_multiple_regs.
4255 * config/mn10300/mn10300.c: Likewise.
4256 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4257 store_multiple_regs.
4258 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4259 Call mn10300_store_multiple_regs.
4261 2014-01-30 Nick Clifton <nickc@redhat.com>
4262 DJ Delorie <dj@redhat.com>
4264 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4265 %fp 2 to keep registers after it properly word-aligned.
4266 (rl78_alloc_physical_registers_umul): Handle the case where both
4267 input operands are the same.
4269 2014-01-30 Richard Biener <rguenther@suse.de>
4271 PR tree-optimization/59903
4272 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4275 2014-01-30 Jason Merrill <jason@redhat.com>
4278 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4281 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4283 2014-01-30 Richard Biener <rguenther@suse.de>
4285 PR tree-optimization/59951
4286 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4288 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4291 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4292 SFmode to DFmode case.
4294 2014-01-29 DJ Delorie <dj@redhat.com>
4296 * config/msp430/msp430.opt (-minrt): New.
4297 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4299 (ENDFILE_SPEC): Likewise.
4301 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4303 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4304 (estimate_function_body_sizes): Use it.
4306 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4309 * dwarf2out.c (is_cxx_auto): New.
4310 (is_base_type): Use it.
4311 (gen_type_die_with_usage): Likewise.
4313 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4315 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4316 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4317 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4318 -maltivec=be with LE targets.
4319 (vsx_mergeh_<mode>): Likewise.
4320 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4321 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4322 (altivec_vmrghb): Replace with define_expand and new
4323 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4324 (altivec_vmrghb_direct): New define_insn.
4325 (altivec_vmrghh): Replace with define_expand and new
4326 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4327 (altivec_vmrghh_direct): New define_insn.
4328 (altivec_vmrghw): Replace with define_expand and new
4329 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4330 (altivec_vmrghw_direct): New define_insn.
4331 (*altivec_vmrghsf): Adjust for endianness.
4332 (altivec_vmrglb): Replace with define_expand and new
4333 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4334 (altivec_vmrglb_direct): New define_insn.
4335 (altivec_vmrglh): Replace with define_expand and new
4336 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4337 (altivec_vmrglh_direct): New define_insn.
4338 (altivec_vmrglw): Replace with define_expand and new
4339 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4340 (altivec_vmrglw_direct): New define_insn.
4341 (*altivec_vmrglsf): Adjust for endianness.
4342 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4343 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4344 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4345 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4346 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4347 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4348 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4349 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4351 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4353 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4354 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4357 2014-01-29 Richard Biener <rguenther@suse.de>
4359 PR tree-optimization/58742
4360 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4361 associate_pointerplus_align.
4362 (associate_pointerplus_diff): New function.
4363 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4364 and associate_pointerplus_diff.
4366 2014-01-29 Richard Biener <rguenther@suse.de>
4368 * lto-streamer.h (LTO_major_version): Bump to 3.
4369 (LTO_minor_version): Reset to 0.
4371 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4373 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4374 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4375 (arm_file_start): Generate correct asm header for armv7ve.
4376 * config/arm/bpabi.h: Add multilib support for armv7ve.
4377 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4378 and cortex-a15 to armv7ve.
4379 * config/arm/t-aprofile: Add multilib support for armv7ve.
4380 * doc/invoke.texi: Document -march=armv7ve.
4382 2014-01-29 Richard Biener <rguenther@suse.de>
4384 PR tree-optimization/58742
4385 * tree-ssa-forwprop.c (associate_plusminus): Return true
4386 if we changed sth, defer EH cleanup to ...
4387 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4388 (simplify_mult): New function.
4390 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4393 PR tree-optimization/59920
4394 * tree.c (build_common_builtin_nodes): Remove
4395 __builtin_setjmp_dispatcher initialization.
4396 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4397 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4398 instead of gsi_after_labels + manually skipping debug stmts.
4399 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4400 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4401 * tree-inline.c (copy_edges_for_bb): Remove
4402 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4403 argument. Ignore computed_goto_p stmts. Don't call
4404 make_abnormal_goto_edges. If a call might need abnormal edges
4405 for non-local gotos, see if it already has an edge to
4406 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4407 with true argument, don't do anything then, otherwise add
4408 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4409 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4411 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4412 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4413 (lower_stmt): Don't set data->calls_builtin_setjmp.
4414 (lower_builtin_setjmp): Adjust comment.
4415 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4416 * tree-cfg.c (found_computed_goto): Remove.
4417 (factor_computed_gotos): Remove.
4418 (make_goto_expr_edges): Return bool, true for computed gotos.
4419 Don't call make_abnormal_goto_edges.
4420 (build_gimple_cfg): Don't set found_computed_goto, don't call
4421 factor_computed_gotos.
4422 (computed_goto_p): No longer static.
4423 (make_blocks): Don't set found_computed_goto.
4424 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4425 (make_edges): If make_goto_expr_edges returns true, push bb
4426 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4427 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4428 vector. Record mapping between bbs and OpenMP regions if there
4429 are any, adjust make_gimple_omp_edges caller. Call
4430 handle_abnormal_edges.
4431 (make_abnormal_goto_edges): Remove.
4432 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4433 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4434 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4435 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4436 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4437 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4438 filling *region also set *region_idx to (*region)->entry->index.
4441 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4442 For REGs set ORIGINAL_REGNO.
4444 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4446 * doc/md.texi: Mention that a target shouldn't implement
4447 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4450 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4452 PR tree-optimization/59594
4453 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4454 a copy of the datarefs vector rather than the vector itself.
4456 2014-01-28 Jason Merrill <jason@redhat.com>
4459 * dwarf2out.c (auto_die): New static.
4460 (gen_type_die_with_usage): Handle C++1y 'auto'.
4461 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4464 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4467 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4468 (SPEC_X32): Likewise.
4469 (SPEC_64): Likewise.
4470 * config/i386/i386.c (ix86_option_override_internal): Turn off
4471 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4473 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4474 * config/i386/i386.h (TARGET_16BIT): New macro.
4475 (TARGET_16BIT_P): Likewise.
4476 * config/i386/i386.opt: Add m16.
4477 * doc/invoke.texi: Document -m16.
4479 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4481 PR preprocessor/59935
4482 * input.c (location_get_source_line): Bail out on when line number
4483 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4485 2014-01-28 Richard Biener <rguenther@suse.de>
4487 PR tree-optimization/58742
4488 * tree-ssa-forwprop.c (associate_plusminus): Handle
4489 pointer subtraction of the form (T)(P + A) - (T)P.
4491 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4493 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4496 2014-01-28 Richard Biener <rguenther@suse.de>
4499 2014-01-28 Richard Biener <rguenther@suse.de>
4501 PR rtl-optimization/45364
4502 PR rtl-optimization/59890
4503 * var-tracking.c (local_get_addr_clear_given_value): Handle
4504 already cleared slot.
4505 (val_reset): Handle not allocated local_get_addr_cache.
4506 (vt_find_locations): Use post-order on the inverted CFG.
4508 2014-01-28 Richard Biener <rguenther@suse.de>
4510 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4512 2014-01-28 Richard Biener <rguenther@suse.de>
4514 PR rtl-optimization/45364
4515 PR rtl-optimization/59890
4516 * var-tracking.c (local_get_addr_clear_given_value): Handle
4517 already cleared slot.
4518 (val_reset): Handle not allocated local_get_addr_cache.
4519 (vt_find_locations): Use post-order on the inverted CFG.
4521 2014-01-28 Alan Modra <amodra@gmail.com>
4523 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4524 * configure.ac <recursive call for build != host>: Define
4525 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4526 and LD_FOR_BUILD too.
4527 * configure: Regenerate.
4529 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4531 * config/i386/i386.c (get_builtin_code_for_version): Separate
4532 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4533 Broadwell from Haswell.
4535 2014-01-27 Steve Ellcey <sellcey@mips.com>
4537 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4538 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4539 * config/mips/mips.c (mips_option_override): Change setting
4541 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4542 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4543 Change from Mask to Var.
4545 2014-01-27 Jeff Law <law@redhat.com>
4547 * ipa-inline.c (inline_small_functions): Fix typo.
4549 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4551 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4552 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4553 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4554 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4555 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4556 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4557 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4558 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4559 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4560 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4561 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4562 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4563 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4564 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4565 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4566 (_mm512_storeu_epi64): Ditto.
4567 (_mm512_cmpge_epi32_mask): Ditto.
4568 (_mm512_cmpge_epu32_mask): Ditto.
4569 (_mm512_cmpge_epi64_mask): Ditto.
4570 (_mm512_cmpge_epu64_mask): Ditto.
4571 (_mm512_cmple_epi32_mask): Ditto.
4572 (_mm512_cmple_epu32_mask): Ditto.
4573 (_mm512_cmple_epi64_mask): Ditto.
4574 (_mm512_cmple_epu64_mask): Ditto.
4575 (_mm512_cmplt_epi32_mask): Ditto.
4576 (_mm512_cmplt_epu32_mask): Ditto.
4577 (_mm512_cmplt_epi64_mask): Ditto.
4578 (_mm512_cmplt_epu64_mask): Ditto.
4579 (_mm512_cmpneq_epi32_mask): Ditto.
4580 (_mm512_cmpneq_epu32_mask): Ditto.
4581 (_mm512_cmpneq_epi64_mask): Ditto.
4582 (_mm512_cmpneq_epu64_mask): Ditto.
4583 (_mm512_expand_pd): Ditto.
4584 (_mm512_expand_ps): Ditto.
4585 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4586 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4587 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4588 * config/i386/i386.c (ix86_builtins): Add
4589 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4590 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4591 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4592 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4593 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4594 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4595 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4596 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4597 IX86_BUILTIN_PMOVUSQW512_MEM.
4598 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4599 __builtin_ia32_pmovsqd512mem_mask,
4600 __builtin_ia32_pmovqd512mem_mask,
4601 __builtin_ia32_pmovusqw512mem_mask,
4602 __builtin_ia32_pmovsqw512mem_mask,
4603 __builtin_ia32_pmovqw512mem_mask,
4604 __builtin_ia32_pmovusdw512mem_mask,
4605 __builtin_ia32_pmovsdw512mem_mask,
4606 __builtin_ia32_pmovdw512mem_mask,
4607 __builtin_ia32_pmovqb512mem_mask,
4608 __builtin_ia32_pmovusqb512mem_mask,
4609 __builtin_ia32_pmovsqb512mem_mask,
4610 __builtin_ia32_pmovusdb512mem_mask,
4611 __builtin_ia32_pmovsdb512mem_mask,
4612 __builtin_ia32_pmovdb512mem_mask.
4613 (bdesc_args): Add __builtin_ia32_expanddf512,
4614 __builtin_ia32_expandsf512.
4615 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4616 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4617 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4618 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4619 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4620 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4621 (avx512f_<code>v8div16qi2_mask_store): This.
4622 (avx512f_expand<mode>): New.
4624 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4626 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4628 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4629 (_mm512_prefetch_i32scatter_pd): Ditto.
4630 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4631 (_mm512_prefetch_i64scatter_pd): Ditto.
4632 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4633 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4634 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4635 (_mm512_prefetch_i32scatter_ps): Ditto.
4636 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4637 (_mm512_prefetch_i64scatter_ps): Ditto.
4638 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4639 * config/i386/i386-builtin-types.def: Define
4640 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4641 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4642 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4643 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4644 IX86_BUILTIN_SCATTERPFQPD.
4645 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4646 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4647 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4648 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4649 __builtin_ia32_scatterpfqps.
4650 (ix86_expand_builtin): Expand new built-ins.
4651 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4652 fix memory access data type.
4653 (*avx512pf_gatherpf<mode>_mask): Ditto.
4654 (*avx512pf_gatherpf<mode>): Ditto.
4655 (avx512pf_scatterpf<mode>): Ditto.
4656 (*avx512pf_scatterpf<mode>_mask): Ditto.
4657 (*avx512pf_scatterpf<mode>): Ditto.
4658 (GATHER_SCATTER_SF_MEM_MODE): New.
4659 (avx512pf_gatherpf<mode>df): Ditto.
4660 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4661 (*avx512pf_scatterpf<mode>df): Ditto.
4663 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4666 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4667 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4670 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4672 * common/config/arm/arm-common.c
4673 (arm_rewrite_mcpu): Handle multiple names.
4675 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4677 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4679 * gimple-builder.h (create_gimple_tmp): Delete.
4681 2014-01-27 Christian Bruel <christian.bruel@st.com>
4683 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4686 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4688 * config/pa/pa.md (call): Generate indirect long calls to non-local
4689 functions when outputing 32-bit code.
4690 (call_value): Likewise except for special call to buggy powf function.
4692 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4693 portable runtime and PIC indirect calls.
4694 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4695 and PIC call sequences. Use ldo instead of blr to set return register
4696 in PIC call sequence.
4698 2014-01-25 Walter Lee <walt@tilera.com>
4700 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4701 avoid clobbering a live register.
4703 2014-01-25 Walter Lee <walt@tilera.com>
4705 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4706 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4707 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4708 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4710 2014-01-25 Walter Lee <walt@tilera.com>
4712 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4713 arguments on even registers.
4714 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4716 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4717 (BIGGEST_ALIGNMENT): Ditto.
4718 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4720 2014-01-25 Walter Lee <walt@tilera.com>
4722 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4723 insns before bundling.
4724 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4726 2014-01-25 Walter Lee <walt@tilera.com>
4728 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4729 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4730 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4732 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4734 * config/mips/constraints.md (kl): Delete.
4735 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4736 define expands, using...
4737 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4738 instructions for MIPS16.
4739 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4740 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4742 2014-01-25 Walter Lee <walt@tilera.com>
4744 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4748 2014-01-25 Walter Lee <walt@tilera.com>
4750 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4751 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4753 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4755 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4758 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4760 * print-rtl.c (in_call_function_usage): New var.
4761 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4762 EXPR_LIST mode as mode and not as reg note name.
4765 * cfgloopmanip.c (copy_loop_info): If
4766 loop->warned_aggressive_loop_optimizations, make sure
4767 the flag is set in target loop too.
4769 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4771 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4773 * builtins.def: Likewise.
4774 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4775 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4776 * ira.c (ira_setup_eliminable_regset): Likewise.
4777 * omp-low.c (gate_expand_omp): Likewise.
4778 (execute_lower_omp): Likewise.
4779 (diagnose_sb_0): Likewise.
4780 (gate_diagnose_omp_blocks): Likewise.
4781 (simd_clone_clauses_extract): Likewise.
4784 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4786 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4787 correction for little endian...
4788 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4791 2014-01-24 Jeff Law <law@redhat.com>
4793 PR tree-optimization/59919
4794 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4795 for non-returning calls.
4797 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4799 * common/config/aarch64/aarch64-common.c
4800 (aarch64_rewrite_mcpu): Handle multiple names.
4801 * config/aarch64/aarch64.h
4802 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4804 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4806 * input.c (add_file_to_cache_tab): Handle the case where fopen
4809 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4812 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4813 from push operand if code of push isn't PRE_DEC.
4815 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4818 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4819 -mquad-memory-atomic. Update -mquad-memory documentation to say
4820 it is only used for non-atomic loads/stores.
4822 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4823 -mquad-memory or -mquad-memory-atomic switches.
4825 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4826 -mquad-memory-atomic to ISA 2.07 support.
4828 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4829 to separate support of normal quad word memory operations (ldq, stq)
4830 from the atomic quad word memory operations.
4832 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4833 support to separate non-atomic quad word operations from atomic
4834 quad word operations. Disable non-atomic quad word operations in
4835 little endian mode so that we don't have to swap words after the
4836 load and before the store.
4837 (quad_load_store_p): Add comment about atomic quad word support.
4838 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4839 options printed with -mdebug=reg.
4841 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4842 -mquad-memory-atomic as the test for whether we have quad word
4843 atomic instructions.
4844 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4845 or -mp8-vector are used, allow byte/half-word atomic operations.
4847 * config/rs6000/sync.md (load_lockedti): Insure that the address
4848 is a proper indexed or indirect address for the lqarx instruction.
4849 On little endian systems, swap the hi/lo registers after the lqarx
4851 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4852 insure the address is valid for the lqarx instruction.
4853 (store_conditionalti): Insure that the address is a proper indexed
4854 or indirect address for the stqcrx. instruction. On little endian
4855 systems, swap the hi/lo registers before doing the stqcrx.
4857 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4858 insure the address is valid for the stqcrx. instruction.
4860 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4861 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4862 type of quad memory support is available.
4864 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4867 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4868 there is a danger of looping.
4870 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4872 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4873 force flag_ira_loop_pressure if set via command line.
4875 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4877 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4878 (ashr_simd): New builtin handling DI mode.
4879 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4880 (aarch64_sshr_simddi): New match pattern.
4881 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4882 (vshrd_n_s64): Likewise.
4883 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4885 2014-01-23 Nick Clifton <nickc@redhat.com>
4887 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4888 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4889 favour of mcu specific scripts.
4890 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4893 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4894 Alex Velenko <Alex.Velenko@arm.com>
4896 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4897 (vaddv_s16): Likewise.
4898 (vaddv_s32): Likewise.
4899 (vaddv_u8): Likewise.
4900 (vaddv_u16): Likewise.
4901 (vaddv_u32): Likewise.
4902 (vaddvq_s8): Likewise.
4903 (vaddvq_s16): Likewise.
4904 (vaddvq_s32): Likewise.
4905 (vaddvq_s64): Likewise.
4906 (vaddvq_u8): Likewise.
4907 (vaddvq_u16): Likewise.
4908 (vaddvq_u32): Likewise.
4909 (vaddvq_u64): Likewise.
4910 (vaddv_f32): Likewise.
4911 (vaddvq_f32): Likewise.
4912 (vaddvq_f64): Likewise.
4913 (vmaxv_f32): Likewise.
4914 (vmaxv_s8): Likewise.
4915 (vmaxv_s16): Likewise.
4916 (vmaxv_s32): Likewise.
4917 (vmaxv_u8): Likewise.
4918 (vmaxv_u16): Likewise.
4919 (vmaxv_u32): Likewise.
4920 (vmaxvq_f32): Likewise.
4921 (vmaxvq_f64): Likewise.
4922 (vmaxvq_s8): Likewise.
4923 (vmaxvq_s16): Likewise.
4924 (vmaxvq_s32): Likewise.
4925 (vmaxvq_u8): Likewise.
4926 (vmaxvq_u16): Likewise.
4927 (vmaxvq_u32): Likewise.
4928 (vmaxnmv_f32): Likewise.
4929 (vmaxnmvq_f32): Likewise.
4930 (vmaxnmvq_f64): Likewise.
4931 (vminv_f32): Likewise.
4932 (vminv_s8): Likewise.
4933 (vminv_s16): Likewise.
4934 (vminv_s32): Likewise.
4935 (vminv_u8): Likewise.
4936 (vminv_u16): Likewise.
4937 (vminv_u32): Likewise.
4938 (vminvq_f32): Likewise.
4939 (vminvq_f64): Likewise.
4940 (vminvq_s8): Likewise.
4941 (vminvq_s16): Likewise.
4942 (vminvq_s32): Likewise.
4943 (vminvq_u8): Likewise.
4944 (vminvq_u16): Likewise.
4945 (vminvq_u32): Likewise.
4946 (vminnmv_f32): Likewise.
4947 (vminnmvq_f32): Likewise.
4948 (vminnmvq_f64): Likewise.
4950 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4952 * config/aarch64/aarch64-simd.md
4953 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4954 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4955 (*aarch64_mul3_elt<mode>): Likewise.
4956 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4957 (*aarch64_mul3_elt_to_64v2df): Likewise.
4958 (*aarch64_mla_elt<mode>): Likewise.
4959 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4960 (*aarch64_mls_elt<mode>): Likewise.
4961 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4962 (*aarch64_fma4_elt<mode>): Likewise.
4963 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4964 (*aarch64_fma4_elt_to_64v2df): Likewise.
4965 (*aarch64_fnma4_elt<mode>): Likewise.
4966 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4967 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4968 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4969 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4970 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4971 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4972 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4973 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4974 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4976 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4978 * config/aarch64/aarch64-simd.md
4979 (aarch64_be_checked_get_lane<mode>): New define_expand.
4980 * config/aarch64/aarch64-simd-builtins.def
4981 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4982 New builtin definition.
4983 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4984 Use new safe be builtin.
4986 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4988 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4990 (aarch64_be_st1<mode>): Likewise.
4991 (aarch_ld1<VALL:mode>): Define_expand modified.
4992 (aarch_st1<VALL:mode>): Likewise.
4993 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4994 (UNSPEC_ST1): Likewise.
4996 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4998 * config/microblaze/microblaze.md: Add trap insn and attribute
5000 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5002 PR preprocessor/58580
5003 * input.h (location_get_source_line): Take an additional line_size
5005 (void diagnostics_file_cache_fini): Declare new function.
5006 * input.c (struct fcache): New type.
5007 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5008 New static constants.
5009 (diagnostic_file_cache_init, total_lines_num)
5010 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5011 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5012 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5013 (get_next_line, read_next_line, goto_next_line, read_line_num):
5014 New static function definitions.
5015 (diagnostic_file_cache_fini): New function.
5016 (location_get_source_line): Take an additional output line_len
5017 parameter. Re-write using lookup_or_add_file_to_cache_tab and
5019 * diagnostic.c (diagnostic_finish): Call
5020 diagnostic_file_cache_fini.
5021 (adjust_line): Take an additional input parameter for the length
5022 of the line, rather than calculating it with strlen.
5023 (diagnostic_show_locus): Adjust the use of
5024 location_get_source_line and adjust_line with respect to their new
5025 signature. While displaying a line now, do not stop at the first
5026 null byte. Rather, display the zero byte as a space and keep
5027 going until we reach the size of the line.
5028 * Makefile.in: Add vec.o to OBJS-libcommon
5030 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5031 Ilya Tocar <ilya.tocar@intel.com>
5033 * config/i386/avx512fintrin.h (_mm512_kmov): New.
5034 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
5035 (__builtin_ia32_kmov16): Ditto.
5036 * config/i386/i386.md (UNSPEC_KMOV): New.
5039 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5041 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
5042 (_mm512_storeu_si512): Ditto.
5044 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
5047 * rtl.h (get_referenced_operands): Declare.
5048 * recog.c (get_referenced_operands): New function.
5049 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
5050 operands have been referenced when recording LO_SUM references.
5052 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
5054 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
5056 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5058 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5059 Enable for generic and recent AMD targets.
5061 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5063 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5064 ARG_SIZE note when adjustment was eliminated.
5066 2014-01-22 Jeff Law <law@redhat.com>
5068 PR tree-optimization/59597
5069 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5070 in file. Accept new argument REGISTERING and use it to modify
5071 dump output appropriately.
5072 (register_jump_thread): Corresponding changes.
5073 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5074 thread paths involving joiner blocks. Add code to dump cancelled
5075 jump threading paths.
5077 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5079 PR rtl-optimization/59477
5080 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5081 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5083 2014-01-22 Tom Tromey <tromey@redhat.com>
5085 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5087 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5089 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5091 PR rtl-optimization/59896
5092 * lra-constraints.c (process_alt_operands): Check unused note for
5093 matched operands of insn with no output reloads.
5095 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5097 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5098 (mips_move_from_gpr_cost): Likewise.
5100 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5102 PR rtl-optimization/59858
5103 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5104 ira_class_hard_regs_num.
5105 (process_alt_operands): Increase reject for dying matched operand.
5107 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5110 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5111 smaller than size, perform several stores or loads and stores
5112 at dst + count - size to store or copy all of size bytes, rather
5113 than just last modesize bytes.
5115 2014-01-20 DJ Delorie <dj@redhat.com>
5117 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5118 that CLOBBERs are REGs before propogating their values.
5120 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5123 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5124 (cgraph_inline_failed_type): New function.
5125 * cgraph.h (DEFCIFCODE): Add type.
5126 (cgraph_inline_failed_type_t): New enum.
5127 (cgraph_inline_failed_type): New prototype.
5128 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5129 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5130 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5131 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5132 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5133 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5134 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5135 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5136 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5137 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5138 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5139 OPTIMIZATION_MISMATCH.
5140 * tree-inline.c (expand_call_inline): Emit errors during
5141 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5143 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5146 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5147 mode attribute in insn output.
5149 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5151 * output.h (output_constant): Delete.
5152 * varasm.c (output_constant): Make private.
5154 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5156 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5158 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5161 * tree.h (fold_builtin_strcat): New prototype.
5162 * builtins.c (fold_builtin_strcat): No longer static. Add len
5163 argument, if non-NULL, don't call c_strlen. Optimize
5164 directly into __builtin_memcpy instead of __builtin_strcpy.
5165 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5166 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5168 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5170 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5171 for SImode_address_operand operands, having only a REG argument.
5173 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5175 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5176 loader name using mbig-endian.
5177 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5179 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5181 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5185 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5187 * config/aarch64/aarch64-protos.h
5188 (aarch64_cannot_change_mode_class_ptr): Declare.
5189 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5190 aarch64_cannot_change_mode_class_ptr): New.
5191 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5192 backend hook aarch64_cannot_change_mode_class.
5194 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5196 * common/config/aarch64/aarch64-common.c
5197 (aarch64_handle_option): Don't handle any option order logic here.
5198 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5199 selected_cpu, warn on architecture version mismatch.
5200 (aarch64_override_options): Fix parsing order for option strings.
5202 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5203 Iain Sandoe <iain@codesourcery.com>
5206 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5207 warning. Amend comment to reflect current functionality.
5209 2014-01-20 Richard Biener <rguenther@suse.de>
5212 * builtins.c (fold_builtin_strcat): Remove case better handled
5213 by tree-ssa-strlen.c.
5215 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5217 * config/aarch64/aarch64.opt
5218 (mcpu, march, mtune): Make case-insensitive.
5220 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5223 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5224 if operands[1] is a REG or ZERO_EXTEND of a REG.
5226 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5228 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5230 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5232 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5233 long non-pic millicode calls.
5235 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5237 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5239 2014-01-19 Kito Cheng <kito@0xlab.org>
5241 * builtins.c (expand_movstr): Check movstr expand done or fail.
5243 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5244 H.J. Lu <hongjiu.lu@intel.com>
5247 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5248 to DImode for zero-extended addresses.
5250 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5252 PR rtl-optimization/57763
5253 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5254 on the new indirect jump_insn and increment LABEL_NUSES (label).
5256 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5260 * config.gcc (x86_archs): New variable.
5261 (x86_64_archs): Likewise.
5262 (x86_cpus): Likewise.
5263 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5264 --with-arch/--with-cpu= options.
5265 Support --with-arch=/--with-cpu={nehalem,westmere,
5266 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5268 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5270 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5271 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5273 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5275 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5277 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5280 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5281 clear cpp_get_options (parse_in)->warn_unused_macros for
5282 ix86_target_macros_internal with cpp_define.
5284 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5286 * jump.c (delete_related_insns): Keep (use (insn))s.
5287 * reorg.c (redundant_insn): Check for barriers too.
5289 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5291 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5293 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5295 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5296 call to $$dyncall when TARGET_LONG_CALLS is true.
5298 2014-01-17 Jeff Law <law@redhat.com>
5300 * ree.c (combine_set_extension): Temporarily disable test for
5301 changing number of hard registers.
5303 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5306 * ipa-inline-analysis.c (inline_free_summary):
5307 Do not free summary of aliases.
5309 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5312 * gimplify.c (gimplify_expr): Use create_tmp_var
5313 instead of create_tmp_var_raw. If cond doesn't have
5314 integral type, don't add the IFN_ANNOTATE builtin at all.
5316 2014-01-17 Martin Jambor <mjambor@suse.cz>
5319 * ipa-cp.c (prev_edge_clone): New variable.
5320 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5321 Also resize prev_edge_clone vector.
5322 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5323 (ipcp_edge_removal_hook): New function.
5324 (ipcp_driver): Register ipcp_edge_removal_hook.
5326 2014-01-17 Andrew Pinski <apinski@cavium.com>
5327 Steve Ellcey <sellcey@mips.com>
5330 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5333 2014-01-17 Jeff Law <law@redhat.com>
5336 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5337 so that pass_ccp runs first.
5339 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5341 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5342 (ix86_adjust_cost): Use !TARGET_XXX.
5343 (do_reorder_for_imul): Likewise.
5344 (swap_top_of_ready_list): Likewise.
5345 (ix86_sched_reorder): Likewise.
5347 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5349 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5350 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5351 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5352 (intel_memset): New. Duplicate slm_memset.
5353 (intel_cost): New. Duplicate slm_cost.
5354 (m_INTEL): New macro.
5355 (processor_target_table): Add "intel".
5356 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5357 with PROCESSOR_INTEL for "intel".
5358 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5359 PROCESSOR_SILVERMONT.
5360 (ix86_issue_rate): Likewise.
5361 (ix86_adjust_cost): Likewise.
5362 (ia32_multipass_dfa_lookahead): Likewise.
5363 (swap_top_of_ready_list): Likewise.
5364 (ix86_sched_reorder): Likewise.
5365 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5366 instead of TARGET_OPT_AGU.
5367 * config/i386/i386.h (TARGET_INTEL): New.
5368 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5369 (processor_type): Add PROCESSOR_INTEL.
5370 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5371 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5373 2014-01-17 Marek Polacek <polacek@redhat.com>
5376 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5379 2014-01-17 Richard Biener <rguenther@suse.de>
5381 PR tree-optimization/46590
5382 * opts.c (default_options_table): Add entries for
5383 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5384 all enabled at -O1 but not for -Og.
5385 * common.opt (fbranch-count-reg): Remove Init(1).
5386 (fmove-loop-invariants): Likewise.
5387 (ftree-pta): Likewise.
5389 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5391 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5392 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5393 decls to at least the GCC 4.8 used alignments.
5396 * tree-nested.c (convert_nonlocal_reference_stmt,
5397 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5398 of GIMPLE_BIND stmts, adjust associated decls.
5400 2014-01-17 Richard Biener <rguenther@suse.de>
5402 PR tree-optimization/46590
5403 * vec.h (vec<>::bseach): New member function implementing
5404 binary search according to C89 bsearch.
5405 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5406 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5407 bitmap pointer again. Make accesses_in_loop a flat array.
5408 (mem_ref_obstack): New global.
5409 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5410 (mark_ref_stored): Likewise.
5411 (ref_indep_loop_p_2): Likewise.
5412 (set_ref_stored_in_loop): New helper function.
5413 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5414 (memref_free): Adjust.
5415 (record_mem_ref_loc): Simplify.
5416 (gather_mem_refs_stmt): Adjust.
5417 (sort_locs_in_loop_postorder_cmp): New function.
5418 (analyze_memory_references): Sort accesses_in_loop after
5419 loop postorder number.
5420 (find_ref_loc_in_loop_cmp): New function.
5421 (for_all_locs_in_loop): Find relevant cluster of locs in
5422 accesses_in_loop and iterate without recursion.
5423 (execute_sm): Avoid uninit warning.
5424 (struct ref_always_accessed): Simplify.
5425 (ref_always_accessed::operator ()): Likewise.
5426 (ref_always_accessed_p): Likewise.
5427 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5428 loop postorder numbers here.
5429 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5432 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5435 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5436 on decls for which assemble_alias has been called.
5438 2014-01-17 Nick Clifton <nickc@redhat.com>
5440 * config/msp430/msp430.opt: (mcpu): New option.
5441 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5442 (msp430_option_override): Parse target_cpu. If the MCU name
5443 matches a generic string, clear target_mcu.
5444 (msp430_attr): Allow numeric interrupt values up to 63.
5445 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5446 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5448 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5450 * config/msp430/msp430.md (popm): Use %J rather than %I.
5451 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5452 (addhi_cy_i): Use immediate_operand for operand 2.
5453 * doc/invoke.texi: Document -mcpu option.
5455 2014-01-17 Richard Biener <rguenther@suse.de>
5457 PR rtl-optimization/38518
5458 * df.h (df_analyze_loop): Declare.
5459 * df-core.c: Include cfgloop.h.
5460 (df_analyze_1): Split out main part of df_analyze.
5461 (df_analyze): Adjust.
5462 (loop_inverted_post_order_compute): New function.
5463 (loop_post_order_compute): Likewise.
5464 (df_analyze_loop): New function avoiding whole-function
5466 * loop-invariant.c (find_defs): Use df_analyze_loop.
5467 (find_invariants): Adjust.
5468 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5470 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5472 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5473 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5475 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5477 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5478 traversal when removing references.
5480 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5483 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5485 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5488 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5489 pushing a reload for an autoinc when we had previously reloaded an
5490 inner part of the address.
5492 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5494 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5496 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5497 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5498 when not giving up or versioning for alias only because of
5500 (vect_analyze_data_ref_dependences): Set to true.
5501 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5503 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5504 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5508 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5511 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5512 operand 0 predicate for gathers, use a new pseudo as subtarget.
5514 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5517 * lra-constraints.c (process_alt_operands): Add printing debug info.
5518 Check absence of input/output reloads for matched operands too.
5520 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5522 PR rtl-optimization/59835
5523 * ira.c (ira_init_register_move_cost): Increase cost for
5526 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5528 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5530 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5533 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5534 non-register objects. Use gen_(high/low)part more consistently.
5537 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5540 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5541 endian support, remove tests for WORDS_BIG_ENDIAN.
5542 (p8_mfvsrd_3_<mode>): Likewise.
5543 (reload_gpr_from_vsx<mode>): Likewise.
5544 (reload_gpr_from_vsxsf): Likewise.
5545 (p8_mfvsrd_4_disf): Likewise.
5547 2014-01-16 Richard Biener <rguenther@suse.de>
5549 PR rtl-optimization/46590
5550 * lcm.c (compute_antinout_edge): Use postorder iteration.
5551 (compute_laterin): Use inverted postorder iteration.
5553 2014-01-16 Nick Clifton <nickc@redhat.com>
5556 * varasm.c (output_constant): Return the number of bytes actually
5558 (output_constructor_array_range): Update the field size with the
5559 number of bytes emitted by output_constant.
5560 (output_constructor_regular_field): Likewise. Also do not
5561 complain if the total number of bytes emitted is now greater
5562 than the expected fieldpos.
5563 * output.h (output_constant): Update prototype and descriptive comment.
5565 2014-01-16 Marek Polacek <polacek@redhat.com>
5568 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5569 it is error_mark_node.
5571 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5573 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5574 VALID_AVX256_REG_OR_OI_MODE.
5576 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5578 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5579 current procedure should be profiled.
5581 2014-01-15 Andrew Pinski <apinski@cavium.com>
5583 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5584 of moving from/to the STACK_REG register class.
5586 2014-01-15 Richard Henderson <rth@redhat.com>
5589 * reginfo.c (global_regs_decl): Globalize.
5590 * rtl.h (global_regs_decl): Declare.
5591 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5592 reserved via global_regs.
5594 2014-01-15 Teresa Johnson <tejohnson@google.com>
5596 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5598 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5600 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5601 and vmulosh rather than call gen_vec_widen_smult_*.
5602 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5603 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5604 (vec_widen_smult_even_v16qi): Likewise.
5605 (vec_widen_umult_even_v8hi): Likewise.
5606 (vec_widen_smult_even_v8hi): Likewise.
5607 (vec_widen_umult_odd_v16qi): Likewise.
5608 (vec_widen_smult_odd_v16qi): Likewise.
5609 (vec_widen_umult_odd_v8hi): Likewise.
5610 (vec_widen_smult_odd_v8hi): Likewise.
5611 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5612 vmuloub rather than call gen_vec_widen_umult_*.
5613 (vec_widen_umult_lo_v16qi): Likewise.
5614 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5615 vmulosb rather than call gen_vec_widen_smult_*.
5616 (vec_widen_smult_lo_v16qi): Likewise.
5617 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5618 rather than call gen_vec_widen_umult_*.
5619 (vec_widen_umult_lo_v8hi): Likewise.
5620 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5621 rather than call gen_vec_widen_smult_*.
5622 (vec_widen_smult_lo_v8hi): Likewise.
5624 2014-01-15 Jeff Law <law@redhat.com>
5626 PR tree-optimization/59747
5627 * ree.c (find_and_remove_re): Properly handle case where a second
5628 eliminated extension requires widening a copy created for elimination
5629 of a prior extension.
5630 (combine_set_extension): Ensure that the number of hard regs needed
5631 for a destination register does not change when we widen it.
5633 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5635 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5636 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5637 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5638 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5639 (avr-*-rtems*): Likewise.
5640 (bfin*-rtems*): Likewise.
5641 (moxie-*-rtems*): Likewise.
5642 (h8300-*-rtems*): Likewise.
5643 (i[34567]86-*-rtems*): Likewise.
5644 (lm32-*-rtems*): Likewise.
5645 (m32r-*-rtems*): Likewise.
5646 (m68k-*-rtems*): Likewise.
5647 (microblaze*-*-rtems*): Likewise.
5648 (mips*-*-rtems*): Likewise.
5649 (powerpc-*-rtems*): Likewise.
5650 (sh-*-rtems*): Likewise.
5651 (sparc-*-rtems*): Likewise.
5652 (sparc64-*-rtems*): Likewise.
5653 (v850-*-rtems*): Likewise.
5654 (m32c-*-rtems*): Likewise.
5656 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5658 PR rtl-optimization/59511
5659 * ira.c (ira_init_register_move_cost): Use memory costs for some
5660 cases of register move cost calculations.
5661 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5662 instead of BB frequency.
5663 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5664 * lra-assigns.c (find_hard_regno_for): Ditto.
5666 2014-01-15 Richard Biener <rguenther@suse.de>
5668 PR tree-optimization/59822
5669 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5670 (vectorizable_load): Use it to hoist defs of uses of invariant
5671 loads out of the loop.
5673 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5674 Kugan Vivekanandarajah <kuganv@linaro.org>
5677 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5680 2014-01-15 Richard Biener <rguenther@suse.de>
5682 PR rtl-optimization/59802
5683 * lcm.c (compute_available): Use inverted postorder to seed
5684 the initial worklist.
5686 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5689 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5690 ADDR_REGS for invalid symrefs in non-PIC code.
5692 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5695 * builtins.c (determine_block_size): Initialize *probable_max_size
5696 even if len_rtx is CONST_INT.
5698 2014-01-14 Andrew Pinski <apinski@cavium.com>
5700 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5701 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5702 (cortexa53_tunings): Likewise.
5703 (aarch64_sched_issue_rate): New function.
5704 (TARGET_SCHED_ISSUE_RATE): Define.
5706 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5708 * ira-costs.c (find_costs_and_classes): Add missed
5709 ira_init_register_move_cost_if_necessary.
5711 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5714 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5716 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5719 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5720 to indicate if type is used for function return value. Warn ABI
5721 change if the vector mode isn't available for function return value.
5722 (ix86_function_arg_advance): Pass false to type_natural_mode.
5723 (ix86_function_arg): Likewise.
5724 (ix86_gimplify_va_arg): Likewise.
5725 (function_arg_32): Don't warn ABI change.
5726 (ix86_function_value): Pass true to type_natural_mode.
5727 (ix86_return_in_memory): Likewise.
5728 (ix86_struct_value_rtx): Removed.
5729 (TARGET_STRUCT_VALUE_RTX): Likewise.
5731 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5733 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5734 converting a conditional jump into a conditional return.
5736 2014-01-14 Richard Biener <rguenther@suse.de>
5738 PR tree-optimization/58921
5739 PR tree-optimization/59006
5740 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5741 hoisting invariant stmts.
5742 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5743 invariant loads on the preheader edge if possible.
5745 2014-01-14 Joey Ye <joey.ye@arm.com>
5747 * doc/plugin.texi (Building GCC plugins): Update to C++.
5749 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5751 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5752 (_mm_rcp28_round_ss): Ditto.
5753 (_mm_rsqrt28_round_sd): Ditto.
5754 (_mm_rsqrt28_round_ss): Ditto.
5755 (_mm_rcp28_sd): Ditto.
5756 (_mm_rcp28_ss): Ditto.
5757 (_mm_rsqrt28_sd): Ditto.
5758 (_mm_rsqrt28_ss): Ditto.
5759 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5760 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5761 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5762 (IX86_BUILTIN_RCP28SD): Ditto.
5763 (IX86_BUILTIN_RCP28SS): Ditto.
5764 (IX86_BUILTIN_RSQRT28SD): Ditto.
5765 (IX86_BUILTIN_RSQRT28SS): Ditto.
5766 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5767 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5768 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5769 (ix86_expand_special_args_builtin): Expand new FTYPE.
5770 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5771 (srcp14<mode>): Make insn unary.
5772 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5773 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5774 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5775 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5776 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5777 Fix rounding: make it SAE only.
5778 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5780 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5782 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5783 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5784 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5785 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5786 (round_saeonly_mask_scalar_operand4): Ditto.
5787 (round_saeonly_mask_scalar_op3): Ditto.
5788 (round_saeonly_mask_scalar_op4): Ditto.
5790 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5792 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5793 Implement -maltivec=be for vec_insert and vec_extract.
5795 2014-01-10 DJ Delorie <dj@redhat.com>
5797 * config/msp430/msp430.md (call_internal): Don't allow memory
5798 references with SP as the base register.
5799 (call_value_internal): Likewise.
5800 * config/msp430/constraints.md (Yc): New. For memory references
5801 that don't use SP as a base register.
5803 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5804 "an integer without a # prefix"
5805 * config/msp430/msp430.md (epilogue_helper): Use it.
5807 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5810 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5811 AVX512F gather builtins.
5812 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5813 on gather decls with INTEGER_TYPE masktype.
5814 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5815 directly into the builtin rather than hoisting it before loop.
5817 PR tree-optimization/59387
5818 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5819 (scev_const_prop): If folded_casts and type has undefined overflow,
5820 use force_gimple_operand instead of force_gimple_operand_gsi and
5821 for each added stmt if it is assign with
5822 arith_code_with_undefined_signed_overflow, call
5823 rewrite_to_defined_overflow.
5824 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5825 gimple-fold.h instead.
5826 (arith_code_with_undefined_signed_overflow,
5827 rewrite_to_defined_overflow): Moved to ...
5828 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5829 rewrite_to_defined_overflow): ... here. No longer static.
5830 Include gimplify-me.h.
5831 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5832 rewrite_to_defined_overflow): New prototypes.
5834 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5836 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5838 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5840 * builtins.c (get_object_alignment_2): Minor tweak.
5841 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5843 2014-01-13 Christian Bruel <christian.bruel@st.com>
5845 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5846 optimized non constant lengths.
5848 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5851 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5852 load as __atomic_load_N if possible.
5854 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5856 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5858 (rs6000_expand_builtin): Adjust call.
5860 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5863 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5864 * config/rs6000/rs6000.c: Include target-globals.h.
5865 (rs6000_set_current_function): Instead of doing target_reinit
5866 unconditionally, use save_target_globals_default_opts and
5867 restore_target_globals.
5869 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5871 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5872 (rs6000_expand_builtin): Handle mffs and mtfsf.
5873 (rs6000_init_builtins): Define mffs and mtfsf.
5874 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5875 (rs6000_mffs): New pattern.
5876 (rs6000_mtfsf): New pattern.
5878 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5880 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5881 Start narrowing with START. Apply candidate-use pair
5882 and check overall cost in narrowing.
5883 (iv_ca_prune): Pass new argument.
5885 2014-01-10 Jeff Law <law@redhat.com>
5888 * ree.c (combine_reaching_defs): Ensure the defining statement
5889 occurs before the extension when optimizing extensions with
5890 different source and destination hard registers.
5892 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5895 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5896 vtables into the type inheritance graph.
5898 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5900 PR rtl-optimization/59754
5901 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5902 modes in the REGNO != REGNO case.
5904 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5906 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5908 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5910 PR tree-optimization/59745
5911 * tree-predcom.c (tree_predictive_commoning_loop): Call
5912 free_affine_expand_cache if giving up because components is NULL.
5914 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5915 GC in payload of target_globals struct instead of allocating them on
5916 the heap and the larger structs separately using GC.
5917 * target-globals.h (struct target_globals): Make regs, hard_regs,
5918 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5919 of GTY((skip)) and change type to void *.
5920 (reset_target_globals): Cast loads from those fields to corresponding
5923 2014-01-10 Steve Ellcey <sellcey@mips.com>
5926 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5927 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5928 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5930 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5933 * aarch64-modes.def (CC_Zmode): New flags mode.
5934 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5935 represents an equality.
5936 (aarch64_get_condition_code): Handle CC_Zmode.
5937 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5939 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5941 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5942 extraction in good case.
5944 2014-01-10 Richard Biener <rguenther@suse.de>
5946 PR tree-optimization/59374
5947 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5948 checking after SLP discovery. Mark stmts not participating
5949 in any SLP instance properly.
5951 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5953 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5954 when handling a SET rtx.
5956 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5958 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5959 (cortex-a57): Likewise.
5960 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5962 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5964 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5965 non-iwmmxt builtins.
5967 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5971 * ipa-devirt.c record_target_from_binfo): Take as argument
5972 stack of binfos and lookup matching one for virtual inheritance.
5973 (possible_polymorphic_call_targets_1): Update.
5975 2014-01-10 Huacai Chen <chenhc@lemote.com>
5977 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5978 kernel strings for Loongson-2E/2F/3A.
5980 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5983 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5984 is_gimple_call before calling gimple_call_internal_p.
5986 2014-01-09 Steve Ellcey <sellcey@mips.com>
5988 * Makefile.in (TREE_FLOW_H): Remove.
5989 (TREE_SSA_H): Add file names from tree-flow.h.
5990 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5991 * tree.h: Remove tree-flow.h reference.
5992 * hash-table.h: Remove tree-flow.h reference.
5993 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5994 reference with tree-ssa-loop.h.
5996 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5998 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5999 default element-order behavior for -maltivec.
6000 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6001 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6002 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6003 when targeting big endian, at least for now.
6004 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6006 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6009 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6010 var satisfies use_register_for_decl, just take into account type
6011 alignment, rather than decl alignment.
6013 PR tree-optimization/59622
6014 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
6015 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
6016 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
6017 Don't devirtualize for inplace at all. For targets.length () == 1,
6018 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
6020 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6022 * config/i386/i386.md (cpu): Remove the unused btver1.
6024 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6026 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
6028 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6031 * tree-core.h (struct target_globals): New forward declaration.
6032 (struct tree_target_option): Add globals field.
6033 * tree.h (TREE_TARGET_GLOBALS): Define.
6034 (prepare_target_option_nodes_for_pch): New prototype.
6035 * target-globals.h (struct target_globals): Define even if
6037 * tree.c (prepare_target_option_node_for_pch,
6038 prepare_target_option_nodes_for_pch): New functions.
6039 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
6040 * config/i386/i386.c: Include target-globals.h.
6041 (ix86_set_current_function): Instead of doing target_reinit
6042 unconditionally, use save_target_globals_default_opts and
6043 restore_target_globals.
6045 2014-01-09 Richard Biener <rguenther@suse.de>
6047 PR tree-optimization/59715
6048 * tree-cfg.h (split_critical_edges): Declare.
6049 * tree-cfg.c (split_critical_edges): Export.
6050 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
6052 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
6054 * cfgexpand.c (expand_stack_vars): Optionally disable
6055 asan stack protection.
6056 (expand_used_vars): Likewise.
6057 (partition_stack_vars): Likewise.
6058 * asan.c (asan_emit_stack_protection): Optionally disable
6059 after return stack usage.
6060 (instrument_derefs): Optionally disable memory access instrumentation.
6061 (instrument_builtin_call): Likewise.
6062 (instrument_strlen_call): Likewise.
6063 (asan_protect_global): Optionally disable global variables protection.
6064 * doc/invoke.texi: Added doc for new options.
6065 * params.def: Added new options.
6066 * params.h: Likewise.
6068 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6070 PR rtl-optimization/59724
6071 * ifcvt.c (cond_exec_process_if_block): Don't call
6072 flow_find_head_matching_sequence with 0 longest_match.
6073 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6074 non-active insns if !stop_after.
6075 (try_head_merge_bb): Revert 2014-01-07 changes.
6077 2014-01-08 Jeff Law <law@redhat.com>
6079 * ree.c (get_sub_rtx): New function, extracted from...
6080 (merge_def_and_ext): Here.
6081 (combine_reaching_defs): Use get_sub_rtx.
6083 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6085 * cgraph.h (varpool_variable_node): Do not choke on null node.
6087 2014-01-08 Catherine Moore <clm@codesourcery.com>
6089 * config/mips/mips.md (simple_return): Attempt to use JRC
6091 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6093 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6095 PR rtl-optimization/59137
6096 * reorg.c (steal_delay_list_from_target): Call update_block for
6098 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6100 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6102 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6103 two duplicate entries.
6105 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6108 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6110 * config/mips/mips.c (mips_truncated_op_cost): New function.
6111 (mips_rtx_costs): Adjust test for BADDU.
6112 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6114 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6116 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6117 (*baddu_si): ...this new pattern.
6119 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6122 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6124 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6127 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6129 (expand_expr, expand_normal): Adjust.
6130 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6131 inner_reference_p. Use inner_reference_p to expand inner references.
6132 (store_expr): Adjust.
6133 * cfgexpand.c (expand_call_stmt): Adjust.
6135 2014-01-08 Rong Xu <xur@google.com>
6137 * gcov-io.c (gcov_var): Move from gcov-io.h.
6138 (gcov_position): Ditto.
6139 (gcov_is_error): Ditto.
6140 (gcov_rewrite): Ditto.
6141 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6142 only part to libgcc/libgcov.h.
6144 2014-01-08 Marek Polacek <polacek@redhat.com>
6147 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6149 2014-01-08 Marek Polacek <polacek@redhat.com>
6152 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6154 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6156 PR rtl-optimization/59649
6157 * stor-layout.c (get_mode_bounds): For BImode return
6158 0 and STORE_FLAG_VALUE.
6160 2014-01-08 Richard Biener <rguenther@suse.de>
6163 * gimple.h (is_gimple_builtin_call): Remove.
6164 (gimple_builtin_call_types_compatible_p): New.
6165 (gimple_call_builtin_p): New overload.
6166 * gimple.c (is_gimple_builtin_call): Remove.
6167 (validate_call): Rename to ...
6168 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6170 (validate_type): New static function.
6171 (gimple_call_builtin_p): New overload and adjust.
6172 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6173 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6174 (gimple_fold_stmt_to_constant_1): Likewise.
6175 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6177 2014-01-08 Richard Biener <rguenther@suse.de>
6180 * gimplify.c (gimplify_expr): Gimplify register-register type
6181 VIEW_CONVERT_EXPRs to separate stmts.
6183 2014-01-07 Jeff Law <law@redhat.com>
6186 * ree.c (combine_set_extension): Handle case where source
6187 and destination registers in an extension insn are different.
6188 (combine_reaching_defs): Allow source and destination registers
6189 in extension to be different under limited circumstances.
6190 (add_removable_extension): Remove restriction that the
6191 source and destination registers in the extension are the same.
6192 (find_and_remove_re): Emit a copy from the extension's
6193 destination to its source after the defining insn if
6194 the source and destination registers are different.
6197 * ifcvt.c (merge_if_block): If we are merging a block with more than
6198 one successor with a block with no successors, remove any BARRIER
6199 after the second block.
6201 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6203 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6205 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6208 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6209 for 14-bit register offsets when INT14_OK_STRICT is false.
6211 2014-01-07 Roland Stigge <stigge@antcom.de>
6212 Michael Meissner <meissner@linux.vnet.ibm.com>
6215 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6216 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6218 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6220 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6223 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6225 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6226 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6227 rtx is const0_rtx or not.
6229 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6232 * target-globals.c (save_target_globals): Remove this_fn_optab
6234 * toplev.c: Include optabs.h.
6235 (target_reinit): Temporarily restore the global options if another
6236 set of options are in force.
6238 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6240 PR rtl-optimization/58668
6241 * cfgcleanup.c (flow_find_cross_jump): Don't count
6242 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6243 to determine what is counted.
6244 (flow_find_head_matching_sequence): Use active_insn_p to determine
6246 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6248 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6249 determine what is counted.
6251 PR tree-optimization/59643
6252 * tree-predcom.c (split_data_refs_to_components): If one dr is
6253 read and one write, determine_offset fails and the write isn't
6254 in the bad component, just put the read into the bad component.
6256 2014-01-07 Mike Stump <mikestump@comcast.net>
6257 Jakub Jelinek <jakub@redhat.com>
6260 * tree-core.h (struct tree_optimization_option): Change optabs
6261 type from unsigned char * to void *.
6262 * optabs.c (init_tree_optimization_optabs): Adjust
6263 TREE_OPTIMIZATION_OPTABS initialization.
6265 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6268 * config/i386/i386.h (struct machine_function): Add
6269 no_drap_save_restore field.
6270 * config/i386/i386.c (ix86_save_reg): Use
6271 !cfun->machine->no_drap_save_restore instead of
6272 crtl->stack_realign_needed.
6273 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6274 this function clears frame_pointer_needed. Set
6275 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6276 and DRAP reg is needed.
6278 2014-01-06 Marek Polacek <polacek@redhat.com>
6281 * doc/implement-c.texi: Mention that other integer types are
6282 permitted as bit-field types in strictly conforming mode.
6284 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6286 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6289 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6291 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6293 2014-01-06 Martin Jambor <mjambor@suse.cz>
6296 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6298 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6300 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6304 * var-tracking.c (add_stores): Preserve the value of the source even if
6305 we don't record the store.
6307 2014-01-06 Terry Guo <terry.guo@arm.com>
6309 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6311 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6314 * config/darwin.c (darwin_function_section): Adjust return values to
6315 correspond to optimisation changes made in r206070.
6317 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6319 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6320 from prefetch_block tune setting.
6321 (nocona_cost): Correct size of prefetch block to 64.
6323 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6325 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6326 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6327 used to save the static chain register in the computation of the offset
6328 from which the FP registers need to be restored.
6330 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6332 PR tree-optimization/59519
6333 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6334 ICE if get_current_def (current_new_name) is already non-NULL, as long
6335 as it is a phi result of some other phi in *new_exit_bb that has
6338 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6339 or vmovdqu* for misaligned_operand.
6340 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6341 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6342 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6343 aligned_mem for AVX512F masked aligned load and store builtins and for
6346 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6348 PR tree-optimization/59651
6349 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6350 Address range for negative step should be added by TYPE_SIZE_UNIT.
6352 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6354 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6355 overlapping registers also for registers other than the stack pointer.
6357 2014-01-03 Marek Polacek <polacek@redhat.com>
6360 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6363 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6366 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6369 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6370 (push splitter): Use <P:MODE_SIZE> instead of
6371 GET_MODE_SIZE (<P:MODE>mode).
6372 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6373 (mov -1, reg peephole2): Likewise.
6374 * config/i386/sse.md (*mov<mode>_internal,
6375 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6376 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6377 *<code><mode>3, *andnot<mode>3<mask_name>,
6378 <mask_codefor><code><mode>3<mask_name>): Likewise.
6379 * config/i386/subst.md (mask_mode512bit_condition,
6380 sd_mask_mode512bit_condition): Likewise.
6382 2014-01-02 Xinliang David Li <davidxl@google.com>
6384 PR tree-optimization/59303
6385 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6386 (dump_predicates): Better output format.
6387 (pred_equal_p): New function.
6388 (is_neq_relop_p): Ditto.
6389 (is_neq_zero_form_p): Ditto.
6390 (pred_expr_equal_p): Ditto.
6391 (pred_neg_p): Ditto.
6392 (simplify_pred): Ditto.
6393 (simplify_preds_2): Ditto.
6394 (simplify_preds_3): Ditto.
6395 (simplify_preds_4): Ditto.
6396 (simplify_preds): Ditto.
6398 (push_to_worklist): Ditto.
6399 (get_pred_info_from_cmp): Ditto.
6400 (is_degenerated_phi): Ditto.
6401 (normalize_one_pred_1): Ditto.
6402 (normalize_one_pred): Ditto.
6403 (normalize_one_pred_chain): Ditto.
6404 (normalize_preds): Ditto.
6405 (normalize_cond_1): Remove function.
6406 (normalize_cond): Ditto.
6407 (is_gcond_subset_of): Ditto.
6408 (is_subset_of_any): Ditto.
6409 (is_or_set_subset_of): Ditto.
6410 (is_and_set_subset_of): Ditto.
6411 (is_norm_cond_subset_of): Ditto.
6412 (pred_chain_length_cmp): Ditto.
6413 (convert_control_dep_chain_into_preds): Type change.
6414 (find_predicates): Ditto.
6415 (find_def_preds): Ditto.
6416 (destroy_predicates_vecs): Ditto.
6417 (find_matching_predicates_in_rest_chains): Ditto.
6418 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6419 (is_pred_expr_subset): Ditto.
6420 (is_pred_chain_subset_of): Ditto.
6421 (is_included_in): Ditto.
6422 (is_superset_of): Ditto.
6424 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6426 Update copyright years.
6428 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6430 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6431 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6432 config/arc/arc.md, config/arc/arc.opt,
6433 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6434 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6435 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6436 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6437 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6438 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6439 vtable-verify.c, vtable-verify.h: Use the standard form for the
6442 2014-01-02 Tobias Burnus <burnus@net-b.de>
6444 * gcc.c (process_command): Update copyright notice dates.
6445 * gcov-dump.c: Ditto.
6447 * doc/cpp.texi: Bump @copying's copyright year.
6448 * doc/cppinternals.texi: Ditto.
6449 * doc/gcc.texi: Ditto.
6450 * doc/gccint.texi: Ditto.
6451 * doc/gcov.texi: Ditto.
6452 * doc/install.texi: Ditto.
6453 * doc/invoke.texi: Ditto.
6455 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6457 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6459 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6461 * config/i386/sse.md (*mov<mode>_internal): Guard
6462 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6464 PR rtl-optimization/59647
6465 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6466 new_rtx into UNSIGNED_FLOAT rtxes.
6468 Copyright (C) 2014 Free Software Foundation, Inc.
6470 Copying and distribution of this file, with or without modification,
6471 are permitted in any medium without royalty provided the copyright
6472 notice and this notice are preserved.