| blob | 5a7832bc02ccfbedb18759d69ae858a3f5095957 |
1 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
3 * doc/invoke.texi (free): Document AArch64.
5 2014-04-14 Richard Biener <rguenther@suse.de>
7 PR tree-optimization/60042
8 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
9 (insert_into_preds_of_block): Do not prevent PHI insertion
10 for REFERENCE exprs here ...
11 (eliminate_dom_walker::before_dom_children): ... but prevent
12 their use here under similar conditions when applied to the
13 IL after PRE optimizations.
15 2014-04-14 Richard Biener <rguenther@suse.de>
17 * passes.def: Move early points-to after early SRA.
19 2014-04-14 Richard Biener <rguenther@suse.de>
21 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
22 check for which sign-changes we allow when forwarding
23 a converted value into a switch.
25 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
27 * stor-layout.c (place_field): Finalize non-constant offset for the
28 field, if any.
30 2014-04-14 Richard Biener <rguenther@suse.de>
32 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
33 as argument.
34 (expand_switch_using_bit_tests_p): Likewise.
35 (process_switch): Compute and pass on speed_p based on the
36 switch stmt.
37 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
38 optimize_bb_for_speed_p.
40 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
42 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
43 * function.h (struct function): Rename has_force_vect_loops into
44 has_force_vectorize_loops.
45 * lto-streamer-in.c (input_cfg): Adjust for renaming.
46 (input_struct_function_base): Likewise.
47 * lto-streamer-out.c (output_cfg): Likewise.
48 (output_struct_function_base): Likewise.
49 * omp-low.c (expand_omp_simd): Likewise.
50 * tree-cfg.c (move_sese_region_to_fn): Likewise.
51 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
52 (version_loop_for_if_conversion): Likewise.
53 (tree_if_conversion): Likewise.
54 (main_tree_if_conversion): Likewise.
55 (gate_tree_if_conversion): Likewise.
56 * tree-inline.c (copy_loops): Likewise.
57 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
58 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
59 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
60 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
61 * tree-vectorizer.c (vectorize_loops): Likewise.
62 * tree-vectorizer.h (unlimited_cost_model): Likewise.
64 2014-04-14 Richard Biener <rguenther@suse.de>
66 PR lto/60720
67 * lto-streamer-out.c (wrap_refs): New function.
68 (lto_output): Wrap symbol references in global initializes in
69 type-preserving MEM_REFs.
71 2014-04-14 Christian Bruel <christian.bruel@st.com>
73 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
75 2014-04-14 Christian Bruel <christian.bruel@st.com>
77 * config/sh/sh.md (setmemqi): New expand pattern.
78 * config/sh/sh.h (CLEAR_RATIO): Define.
79 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
80 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
82 2014-04-14 Richard Biener <rguenther@suse.de>
84 PR middle-end/55022
85 * fold-const.c (negate_expr_p): Don't negate directional rounding
86 division.
87 (fold_negate_expr): Likewise.
89 2014-04-14 Richard Biener <rguenther@suse.de>
91 PR tree-optimization/59817
92 PR tree-optimization/60453
93 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
94 recursion to catch all CHRECs in the scalar evolution and restrict
95 the predicate for the remains appropriately.
97 2014-04-12 Catherine Moore <clm@codesourcery.com>
99 * config/mips/constraints.md: Add new register constraint "kb".
100 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
101 (*movhi_internal): Likewise.
102 (*movqi_internal): Likewise.
103 * config/mips/mips.h (M16_STORE_REGS): New register class.
104 (REG_CLASS_NAMES): Add M16_STORE_REGS.
105 (REG_CLASS_CONTENTS): Likewise.
106 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
108 2014-04-11 Tobias Burnus <burnus@net-b.de>
110 PR c/60194
111 * doc/invoke.texi (-Wformat-signedness): Document it.
112 (Wformat=2): Mention that this enables -Wformat-signedness.
114 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
116 * common/config/epiphany/epiphany-common.c
117 (epiphany_option_optimization_table): Enable section anchors by
118 default at -O1 or higher.
119 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
120 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
121 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
122 carries no extra cost.
123 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
124 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
125 * config/epiphany/predicates.md (memclob_operand): New predicate.
126 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
127 Use memclob_operand predicate and X constraint for operand 3.
129 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
131 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
132 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
133 its operands.
135 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
137 PR rtl-optimization/60651
138 * mode-switching.c (optimize_mode_switching): Make sure to emit
139 sets of a lower numbered entity before sets of a higher numbered
140 entity to a mode of the same or lower priority.
141 When creating a seginfo for a basic block that starts with a code
142 label, move the insertion point past the code label.
143 (new_seginfo): Document and enforce requirement that
144 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
145 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
146 * doc/tm.texi: Regenerate.
148 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
150 PR target/60811
151 * config/arc/arc.c (arc_save_restore): Fix assert typo.
153 2013-04-11 Jakub Jelinek <jakub@redhat.com>
155 * BASE-VER: Set to 4.10.0.
157 2014-04-11 Tobias Burnus <burnus@net-b.de>
159 PR other/59055
160 * doc/bugreport.texi (Bugs): Remove nodes pointing to the
161 nirvana.
162 * doc/gcc.texi (Service): Update description in the @menu
163 * doc/invoke.texi (Option Summary): Remove misplaced and
164 duplicated @menu.
166 2014-04-11 Steve Ellcey <sellcey@mips.com>
167 Jakub Jelinek <jakub@redhat.com>
169 PR middle-end/60556
170 * expr.c (convert_move): Use emit_store_flag_force instead of
171 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
172 argument to it.
174 2014-04-11 Richard Biener <rguenther@suse.de>
176 PR middle-end/60797
177 * varasm.c (assemble_alias): Avoid endless error reporting
178 recursion by setting TREE_ASM_WRITTEN.
180 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
182 * config/s390/s390.md: Add a splitter for NOT rtx.
184 2014-04-11 Jakub Jelinek <jakub@redhat.com>
186 PR rtl-optimization/60663
187 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
188 PARALLEL.
190 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
191 Jakub Jelinek <jakub@redhat.com>
193 PR lto/60567
194 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
195 decl_node to node.
197 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
199 PR debug/60655
200 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
201 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
202 ameliorating the cases where it can be.
204 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
206 Revert
207 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
209 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
210 (loadsync_<mode>): Change mode.
211 (load_quadpti, store_quadpti): New.
212 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
213 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
214 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
216 2014-04-09 Cong Hou <congh@google.com>
218 PR testsuite/60773
219 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
220 documentation.
222 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
224 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
225 instead of vnor to exploit possible fusion opportunity in the
226 future.
227 (altivec_expand_vec_perm_const_le): Likewise.
229 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
231 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
232 (loadsync_<mode>): Change mode.
233 (load_quadpti, store_quadpti): New.
234 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
235 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
237 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
239 PR target/60763
240 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
241 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
242 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
244 2014-04-08 Richard Biener <rguenther@suse.de>
246 PR middle-end/60706
247 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
248 a 64bit widest int print double-int similar to on HWI64 hosts.
250 2014-04-08 Richard Biener <rguenther@suse.de>
252 PR tree-optimization/60785
253 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
254 default defs properly.
256 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
258 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
259 (Weffc++): Likewise.
261 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
263 * ipa-devirt.c (maybe_record_node): When node is not recorded,
264 set completep to false rather than true.
266 2014-04-07 Douglas B Rupp <rupp@adacore.com>
268 PR target/60504
269 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
270 ARM_TARGET2_DWARF_FORMAT.
272 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
274 PR target/60609
275 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
276 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
277 ADDR_DIFF_VEC.
279 2014-04-07 Richard Biener <rguenther@suse.de>
281 PR tree-optimization/60766
282 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
283 (may_eliminate_iv): Convert cand_value_at result to desired type.
285 2014-04-07 Jason Merrill <jason@redhat.com>
287 PR c++/60731
288 * common.opt (-fno-gnu-unique): Add.
289 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
291 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
293 * haifa-sched.c: Fix outdated function reference and minor
294 grammar errors in introductory comment.
296 2014-04-07 Richard Biener <rguenther@suse.de>
298 PR middle-end/60750
299 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
300 for noreturn calls.
301 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
303 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
305 PR debug/55794
306 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
307 size accounting for thunks.
308 (pa_asm_output_mi_thunk): Use final_start_function() and
309 final_end_function() to output function start and end directives.
311 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
313 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
314 specific ISA/ feature information. Remove short_sp and errata_skip ds.
315 Add avr_device_specific_features enum to have device specific info.
316 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
317 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
318 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
319 updated device specific info.
320 * config/avr/avr-mcus.def: Merge device specific details to
321 dev_attribute field.
322 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
323 errata_skip.
324 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
325 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
326 assembler if RMW isa supported by current device.
327 * config/avr/genmultilib.awk: Update as device info structure changed.
328 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
330 2014-04-04 Cong Hou <congh@google.com>
332 PR tree-optimization/60656
333 * tree-vect-stmts.c (supportable_widening_operation):
334 Fix a bug that elements in a vector with vect_used_by_reduction
335 property are incorrectly reordered when the operation on it is not
336 consistant with the one in reduction operation.
338 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
340 PR rtl-optimization/60155
341 * gcse.c (record_set_data): New function.
342 (single_set_gcse): New function.
343 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
344 (hoist_code): Likewise.
345 (get_pressure_class_and_nregs): Likewise.
347 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
349 * explow.c (probe_stack_range): Emit a final optimization blockage.
351 2014-04-04 Anthony Green <green@moxielogic.com>
353 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
354 typos.
356 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
358 PR ipa/59626
359 * lto-cgraph.c (input_overwrite_node): Check that partitioning
360 flags are set only during streaming.
361 * ipa.c (process_references, walk_polymorphic_call_targets,
362 symtab_remove_unreachable_nodes): Drop bodies of always inline
363 after early inlining.
364 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
366 2014-04-04 Jakub Jelinek <jakub@redhat.com>
367 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
369 PR debug/60655
370 * dwarf2out.c (const_ok_for_output_1): Reject expressions
371 containing a NOT.
373 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
375 PR bootstrap/60743
376 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
377 duration.
378 (cortex_a53_fdivd): Likewise.
380 2014-04-04 Martin Jambor <mjambor@suse.cz>
382 PR ipa/60640
383 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
384 Adjust all callers.
385 * cgraph.c (clone_of_p): Also return true if thunks match.
386 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
387 cgraph_function_or_thunk_node and an obsolete comment.
388 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
389 file.
390 (build_function_decl_skip_args): Likewise.
391 (set_new_clone_decl_and_node_flags): New function.
392 (duplicate_thunk_for_node): Likewise.
393 (redirect_edge_duplicating_thunks): Likewise.
394 (cgraph_clone_node): New parameter args_to_skip, pass it to
395 redirect_edge_duplicating_thunks which is called instead of
396 cgraph_redirect_edge_callee.
397 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
398 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
400 2014-04-04 Jeff Law <law@redhat.com>
402 PR target/60657
403 * config/arm/predicates.md (const_int_I_operand): New predicate.
404 (const_int_M_operand): Similarly.
405 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
406 const_int_operand.
407 (insv_t2, extv_reg, extzv_t2): Likewise.
408 (load_multiple_with_writeback): Similarly for const_int_I_operand.
409 (pop_multiple_with_writeback_and_return): Likewise.
410 (vfp_pop_multiple_with_writeback): Likewise
412 2014-04-04 Richard Biener <rguenther@suse.de>
414 PR ipa/60746
415 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
416 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
417 non-GIMPLE_LABELs.
418 * gimplify.h (gimple_add_tmp_var_fn): Declare.
419 * gimplify.c (gimple_add_tmp_var_fn): New function.
420 * gimple-expr.h (create_tmp_reg_fn): Declare.
421 * gimple-expr.c (create_tmp_reg_fn): New function.
422 * gimple-low.c (record_vars_into): Don't change cfun.
423 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
424 code generation without cfun.
426 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
428 PR bootstrap/60719
429 * Makefile.in (install-driver): Fix shell scripting.
431 2014-04-03 Cong Hou <congh@google.com>
433 PR tree-optimization/60505
434 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
435 threshold of number of iterations below which no vectorization will be
436 done.
437 * tree-vect-loop.c (new_loop_vec_info):
438 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
439 * tree-vect-loop.c (vect_analyze_loop_operations):
440 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
441 * tree-vect-loop.c (vect_transform_loop):
442 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
443 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
444 of iterations of the loop and see if we should build the epilogue.
446 2014-04-03 Richard Biener <rguenther@suse.de>
448 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
449 member.
450 (streamer_tree_cache_create): Adjust.
451 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
452 to allow optional nodes array.
453 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
454 (streamer_tree_cache_append): Likewise.
455 (streamer_tree_cache_create): Create nodes array optionally
456 as specified by parameter.
457 * lto-streamer-out.c (create_output_block): Avoid maintaining
458 the node array in the writer cache.
459 (DFS_write_tree): Remove assertion.
460 (produce_asm_for_decls): Free the out decl state hash table
461 early.
462 * lto-streamer-in.c (lto_data_in_create): Adjust for
463 streamer_tree_cache_create prototype change.
465 2014-04-03 Richard Biener <rguenther@suse.de>
467 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
468 set TREE_CHAIN to NULL_TREE.
470 2014-04-03 Richard Biener <rguenther@suse.de>
472 PR tree-optimization/60740
473 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
474 over all GIMPLE_COND operands.
476 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
478 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
479 (Weffc++): Remove Scott's numbering, merge lists and reference
480 Wnon-virtual-dtor.
482 c-family/
484 cp/
485 * class.c (accessible_nvdtor_p): New.
486 (check_bases): Don't check base destructor here ...
487 (check_bases_and_members): ... check them here. Trigger on
488 Wnon-virtual-dtor flag.
489 (finish_struct_1): Use accessible_nvdtor_p.
491 testsuite/
492 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
493 * g++.dg/warn/Wnvdtor-2.C: New.
494 * g++.dg/warn/Wnvdtor-3.C: New.
495 * g++.dg/warn/Wnvdtor-4.C: New.
496 * g++.dg/warn/Weff1.C: Delete.
497 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
498 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
500 2014-04-03 Nick Clifton <nickc@redhat.com>
502 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
503 properly.
505 2014-04-03 Martin Jambor <mjambor@suse.cz>
507 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
508 mention gcc_unreachable before failing.
509 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
510 removed symbols.
512 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
514 PR ipa/60659
515 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
516 code and instead mark the context inconsistent.
517 (possible_polymorphic_call_targets): For inconsistent contexts
518 return empty complete list.
520 2014-04-02 Anthony Green <green@moxielogic.com>
522 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
523 (extendqisi2, extendhisi2): Define.
524 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
525 (WCHAR_TYPE): Change to unsigned int.
527 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
529 PR tree-optimization/60733
530 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
531 insertion point for PHI candidates to be the end of the feeding
532 block for the PHI argument.
534 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
536 PR rtl-optimization/60650
537 * lra-constraints.c (process_alt_operands): Decrease reject for
538 earlyclobber matching.
540 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
542 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
544 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
546 * config/spu/spu.c (pad_bb): Do not crash when the last
547 insn is CODE_FOR_blockage.
549 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
551 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
552 lies outside the target mode.
554 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
556 PR target/60735
557 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
558 software floating point or no floating point registers, do not
559 allow any type in the FPRs. Eliminate a test for SPE SIMD types
560 in GPRs that occurs after we tested for GPRs that would never be
561 true.
563 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
564 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
565 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
566 specifically allow DDmode, since that does not use the SPE SIMD
567 instructions.
569 2014-04-02 Richard Biener <rguenther@suse.de>
571 PR middle-end/60729
572 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
573 MODE_INTs. Properly use negv_optab.
574 (expand_abs): Likewise.
576 2014-04-02 Richard Biener <rguenther@suse.de>
578 PR bootstrap/60719
579 * Makefile.in (install-driver): Guard extra installs with special
580 names properly.
582 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
584 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
585 Document vec_vgbbd.
587 2014-04-01 Richard Henderson <rth@redhat.com>
589 PR target/60704
590 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
591 alternative enabled before register allocation.
593 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
595 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
596 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
597 typo.
598 (nios2_large_got_address): Remove unneeded 'sym' parameter.
599 (nios2_got_address): Update nios2_large_got_address call site.
600 (nios2_delegitimize_address): New function.
601 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
602 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
603 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
605 2014-04-01 Martin Husemann <martin@duskware.de>
607 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
608 for -mabi=32.
610 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
612 PR rtl-optimization/60604
613 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
614 check from register_operand.
615 (register_operand): Redefine in terms of general_operand.
616 (nonmemory_operand): Use register_operand for the non-constant cases.
618 2014-04-01 Richard Biener <rguenther@suse.de>
620 * gimple.h (struct gimple_statement_base): Align subcode to
621 16 bits.
623 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
625 * doc/invoke.texi (mapp-regs): Clarify.
627 2014-03-31 Ulrich Drepper <drepper@gmail.com>
629 * config/i386/avx512fintrin.h (__v32hi): Define type.
630 (__v64qi): Likewise.
631 (_mm512_set1_epi8): Define.
632 (_mm512_set1_epi16): Define.
633 (_mm512_set4_epi32): Define.
634 (_mm512_set4_epi64): Define.
635 (_mm512_set4_pd): Define.
636 (_mm512_set4_ps): Define.
637 (_mm512_setr4_epi64): Define.
638 (_mm512_setr4_epi32): Define.
639 (_mm512_setr4_pd): Define.
640 (_mm512_setr4_ps): Define.
641 (_mm512_setzero_epi32): Define.
643 2014-03-31 Martin Jambor <mjambor@suse.cz>
645 PR middle-end/60647
646 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
647 callsite_arguments_match_p. Updated all callers. Also check types of
648 corresponding formal parameters and actual arguments.
649 (not_all_callers_have_enough_arguments_p) Renamed to
650 some_callers_have_mismatched_arguments_p.
652 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
654 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
656 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
658 PR target/60034
659 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
660 section anchor.
662 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
664 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
665 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
666 Split out
667 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
668 Use FMAMODE_NOVF512 mode iterator.
669 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
670 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
671 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
672 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
673 Split out
674 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
675 Use VF_128_256 mode iterator.
676 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
677 Ditto.
679 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
681 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
682 static chain if needed.
684 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
686 PR target/60697
687 * lra-constraints.c (index_part_to_reg): New.
688 (process_address): Use it.
690 2014-03-27 Jeff Law <law@redhat.com>
691 Jakub Jelinek <jakub@redhat.com>
693 PR target/60648
694 * expr.c (do_tablejump): Use simplify_gen_binary rather than
695 gen_rtx_{PLUS,MULT} to build up the address expression.
697 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
698 creating non-canonical RTL.
700 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
702 PR ipa/60243
703 * ipa-inline.c (want_inline_small_function_p): Short circuit large
704 functions; reorganize to make cheap checks first.
705 (inline_small_functions): Do not estimate growth when dumping;
706 it is expensive.
707 * ipa-inline.h (inline_summary): Add min_size.
708 (growth_likely_positive): New function.
709 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
710 (set_cond_stmt_execution_predicate): Cleanup.
711 (estimate_edge_size_and_time): Compute min_size.
712 (estimate_calls_size_and_time): Likewise.
713 (estimate_node_size_and_time): Likewise.
714 (inline_update_overall_summary): Update min_size.
715 (do_estimate_edge_time): Likewise.
716 (do_estimate_edge_size): Update.
717 (do_estimate_edge_hints): Update.
718 (growth_likely_positive): New function.
720 2014-03-28 Jakub Jelinek <jakub@redhat.com>
722 PR target/60693
723 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
724 also if addr has VOIDmode.
726 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
728 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
729 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
730 Declare extern.
731 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
732 instructions as well as AdvancedSIMD loads.
734 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
736 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
737 Use crypto_aese type.
738 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
739 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
740 crypto_aese, crypto_aesmc. Move to types.md.
741 * config/arm/types.md (crypto_aes): Split into crypto_aese,
742 crypto_aesmc.
743 * config/arm/iterators.md (crypto_type): Likewise.
745 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
747 * cgraph.c: Include expr.h and tree-dfa.h.
748 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
749 remove LHS.
751 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
753 PR target/60675
754 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
755 regs from checking multi-reg pseudos.
757 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
759 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
761 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
763 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
764 if it would clobber the stack pointer, even temporarily.
766 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
768 * mode-switching.c: Make small adjustments to the top comment.
770 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
772 * config/rs6000/constraints.md (wD constraint): New constraint to
773 match the constant integer to get the top DImode/DFmode out of a
774 vector in a VSX register.
776 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
777 match the constant integer to get the top DImode/DFmode out of a
778 vector in a VSX register.
780 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
781 for ISA 2.07.
783 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
784 vbpermq builtins.
786 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
787 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
789 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
790 Optimize vec_extract of 64-bit values, where the value being
791 extracted is in the top word, where we can use scalar
792 instructions. Add direct move and store support. Combine the big
793 endian/little endian vector select load support into a single insn.
794 (vsx_extract_<mode>_internal1): Likewise.
795 (vsx_extract_<mode>_internal2): Likewise.
796 (vsx_extract_<mode>_load): Likewise.
797 (vsx_extract_<mode>_store): Likewise.
798 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
799 combined into vsx_extract_<mode>_load.
800 (vsx_extract_<mode>_one_le): Likewise.
802 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
803 define the top 64-bit vector element.
805 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
806 constraint.
808 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
809 Document vec_vbpermq builtin.
811 PR target/60672
812 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
813 enable use of xxsldwi and xxpermdi builtin functions.
814 (vec_xxpermdi): Likewise.
816 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
817 Document use of vec_xxsldwi and vec_xxpermdi builtins.
819 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
821 PR rtl-optimization/60650
822 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
823 first_p. Use it.
824 (find_spills_for): New.
825 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
826 Spill all pseudos on the second iteration.
828 2014-03-27 Marek Polacek <polacek@redhat.com>
830 PR c/50347
831 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
832 types.
834 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
836 * config/s390/s390.c (s390_can_use_return_insn): Check for
837 call-saved FPRs on 31 bit.
839 2014-03-27 Jakub Jelinek <jakub@redhat.com>
841 PR middle-end/60682
842 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
843 if they need regimplification, just drop them instead of
844 calling gimple_regimplify_operands on them.
846 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
848 PR target/60580
849 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
850 (aarch64_frame_pointer_required): Adjust logic.
851 (aarch64_can_eliminate): Adjust logic.
852 (aarch64_override_options_after_change): Adjust logic.
854 2014-03-27 Dehao Chen <dehao@google.com>
856 * ipa-inline.c (early_inliner): Update node's inline info.
858 2014-03-26 Dehao Chen <dehao@google.com>
860 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
861 compiler inserted conditional jumps for NAN float check.
863 2014-03-26 Jakub Jelinek <jakub@redhat.com>
865 * ubsan.h (ubsan_create_data): Change second argument's type
866 to const location_t *.
867 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
868 _("<unknown>").
869 (ubsan_create_data): Change second argument to const location_t *PLOC.
870 Create Loc field whenever PLOC is non-NULL.
871 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
872 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
873 callers.
875 PR other/59545
876 * real.c (real_to_integer2): Change type of low to UHWI.
878 2014-03-26 Tobias Burnus <burnus@net-b.de>
880 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
881 (CILK_SELF_SPECS): New define.
882 (driver_self_specs): Use it.
884 2014-03-26 Richard Biener <rguenther@suse.de>
886 * tree-pretty-print.c (percent_K_format): Implement special
887 case for LTO and its stripped down BLOCK tree.
889 2014-03-26 Jakub Jelinek <jakub@redhat.com>
891 PR sanitizer/60636
892 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
894 * tree-vrp.c (simplify_internal_call_using_ranges): If only
895 one range is range_int_cst_p, but not both, at least optimize
896 addition/subtraction of 0 and multiplication by 0 or 1.
897 * gimple-fold.c (gimple_fold_call): Fold
898 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
899 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
900 INTEGER_CSTs, try to fold at least x * 0 and y - y.
902 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
904 PR rtl-optimization/60452
905 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
906 <case REG>: Return 1 for invalid offsets from the frame pointer.
908 2014-03-26 Marek Polacek <polacek@redhat.com>
910 PR c/37428
911 * doc/extend.texi (C Extensions): Mention variable-length arrays in
912 a structure/union.
914 2014-03-26 Marek Polacek <polacek@redhat.com>
916 PR c/39525
917 * doc/extend.texi (Designated Inits): Describe what happens to omitted
918 field members.
920 2014-03-26 Marek Polacek <polacek@redhat.com>
922 PR other/59545
923 * ira-color.c (update_conflict_hard_regno_costs): Perform the
924 multiplication in unsigned type.
926 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
928 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
930 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
932 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
934 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
936 PR ipa/60315
937 * cif-code.def (UNREACHABLE) New code.
938 * ipa-inline.c (inline_small_functions): Skip edges to
939 __builtlin_unreachable.
940 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
941 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
942 predicate to __bulitin_unreachable.
943 (set_cond_stmt_execution_predicate): Fix issue when
944 invert_tree_comparison returns ERROR_MARK.
945 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
946 propagate to inline clones.
947 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
948 to unreachable.
949 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
950 * cgraphclones.c (cgraph_clone_node): If call destination is already
951 ureachable, do not redirect it back.
952 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
953 unreachable.
955 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
957 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
958 Do not modify inline clones.
960 2014-03-25 Jakub Jelinek <jakub@redhat.com>
962 * config/i386/i386.md (general_sext_operand): New mode attr.
963 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
964 don't generate (sign_extend (const_int)).
965 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
966 operands[2]. Use We constraint instead of <i> and
967 <general_sext_operand> predicate instead of <general_operand>.
968 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
969 * config/i386/constraints.md (We): New constraint.
970 * config/i386/predicates.md (x86_64_sext_operand,
971 sext_operand): New predicates.
973 2014-03-25 Martin Jambor <mjambor@suse.cz>
975 PR ipa/60600
976 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
977 inconsistent devirtualizations to __builtin_unreachable.
979 2014-03-25 Marek Polacek <polacek@redhat.com>
981 PR c/35449
982 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
984 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
986 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
987 order of elements for big-endian.
989 2014-03-25 Richard Biener <rguenther@suse.de>
991 PR middle-end/60635
992 * gimplify-me.c (gimple_regimplify_operands): Update the
993 re-gimplifed stmt.
995 2014-03-25 Martin Jambor <mjambor@suse.cz>
997 PR ipa/59176
998 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
999 (lto_output_varpool_node): Likewise.
1000 (input_overwrite_node): Likewise.
1001 (input_varpool_node): Likewise.
1003 2014-03-25 Richard Biener <rguenther@suse.de>
1005 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1006 (run_gcc): Likewise.
1008 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1010 * combine.c (simplify_compare_const): Add MODE argument.
1011 Handle mode_width 0 as very large mode_width.
1012 (try_combine, simplify_comparison): Adjust callers.
1014 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1015 type to avoid signed integer overflow.
1016 * explow.c (plus_constant): Likewise.
1018 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1020 * doc/generic.texi: Correct typos.
1022 2014-03-24 Tobias Burnus <burnus@net-b.de>
1024 * doc/invoke.texi (-flto): Expand section about
1025 using static libraries with LTO.
1027 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1029 PR rtl-optimization/60501
1030 * optabs.def (addptr3_optab): New optab.
1031 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1032 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1033 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1035 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1037 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1039 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1041 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1042 _mm512_set1_pd.
1044 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1045 (_mm256_undefined_ps): Define.
1046 (_mm256_undefined_pd): Define.
1047 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1048 (_mm_undefined_pd): Define.
1049 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1050 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1051 (_mm512_undefined_ps): Define.
1052 (_mm512_undefined_pd): Define.
1053 Use _mm*_undefined_*.
1054 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1056 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1058 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1059 (lshr_simd): DI mode added.
1060 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1061 (aarch64_ushr_simddi): Likewise.
1062 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1063 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1064 (vshrd_n_u64): Likewise.
1066 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1068 * Makefile.in (s-macro_list): Depend on cc1.
1070 2014-03-23 Teresa Johnson <tejohnson@google.com>
1072 * ipa-utils.c (ipa_print_order): Use specified dump file.
1074 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1076 PR rtl-optimization/60601
1077 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1079 * gcc.c (eval_spec_function): Initialize save_growing_value.
1081 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1083 PR sanitizer/60613
1084 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1085 code == MINUS_EXPR, never swap op0 with op1.
1087 * toplev.c (init_local_tick): Avoid signed integer multiplication
1088 overflow.
1089 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1090 shift by first operand's bitsize.
1092 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1094 PR target/60610
1095 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1096 redefine to 1 or 0.
1097 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1098 TARGET_ISA_64BIT_P(x).
1100 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1102 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1103 pattern for vector nor instead of subtract from splat(-1).
1104 (altivec_expand_vec_perm_const_le): Likewise.
1106 2014-03-21 Richard Henderson <rth@twiddle.net>
1108 PR target/60598
1109 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1110 related insns after epilogue_completed.
1112 2014-03-21 Martin Jambor <mjambor@suse.cz>
1114 PR ipa/59176
1115 * cgraph.h (symtab_node): New flag body_removed.
1116 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1117 when removing bodies.
1118 * symtab.c (dump_symtab_base): Dump body_removed flag.
1119 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1120 had their bodies removed.
1122 2014-03-21 Martin Jambor <mjambor@suse.cz>
1124 PR ipa/60419
1125 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1126 in the border.
1128 2014-03-21 Richard Biener <rguenther@suse.de>
1130 PR tree-optimization/60577
1131 * tree-core.h (struct tree_base): Document nothrow_flag use
1132 in VAR_DECL_NONALIASED.
1133 * tree.h (VAR_DECL_NONALIASED): New.
1134 (may_be_aliased): Adjust.
1135 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1137 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1139 * expr.c (expand_expr_real_1): Remove outdated comment.
1141 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1143 PR middle-end/60597
1144 * ira.c (adjust_cleared_regs): Call copy_rtx on
1145 *reg_equiv[REGNO (loc)].src_p before passing it to
1146 simplify_replace_fn_rtx.
1148 PR target/60568
1149 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1150 into CONST, put pic register as first operand of PLUS. Use
1151 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1153 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1155 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1157 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1159 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1160 around for store forwarding issue in the FPU on the UT699.
1161 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1162 loads and operations if -mfix-ut699 is specified.
1163 (divtf3_hq): Tweak attribute.
1164 (sqrttf2_hq): Likewise.
1166 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1168 * calls.c (store_one_arg): Remove incorrect const qualification on the
1169 type of the temporary.
1170 * cfgexpand.c (expand_return): Likewise.
1171 * expr.c (expand_constructor): Likewise.
1172 (expand_expr_real_1): Likewise.
1174 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1176 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1177 of parts.
1179 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1181 PR target/60039
1182 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1184 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1186 * config/arm/aarch-common-protos.h
1187 (alu_cost_table): Fix spelling of "extend".
1188 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1190 2014-03-19 Richard Biener <rguenther@suse.de>
1192 PR middle-end/60553
1193 * tree-core.h (tree_type_common): Re-order pointer members
1194 to reduce recursion depth during GC walks.
1196 2014-03-19 Marek Polacek <polacek@redhat.com>
1198 PR sanitizer/60569
1199 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1200 before accessing it.
1202 2014-03-19 Richard Biener <rguenther@suse.de>
1204 PR lto/59543
1205 * lto-streamer-in.c (input_function): In WPA stage do not drop
1206 debug stmts.
1208 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1210 PR tree-optimization/60559
1211 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1212 with build_zero_cst assignment.
1214 2014-03-18 Kai Tietz <ktietz@redhat.com>
1216 PR rtl-optimization/56356
1217 * sdbout.c (sdbout_parms): Verify that parms'
1218 incoming argument is valid.
1219 (sdbout_reg_parms): Likewise.
1221 2014-03-18 Richard Henderson <rth@redhat.com>
1223 PR target/60562
1224 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1225 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1226 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1228 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1230 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1231 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1232 Italicize plugin event names in description. Explain that
1233 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1234 Remind that no GCC functions should be called after PLUGIN_FINISH.
1235 Explain what pragmas with expansion are.
1237 2014-03-18 Martin Liska <mliska@suse.cz>
1239 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1240 gimple call statement is update.
1241 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1242 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1244 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1246 PR sanitizer/60557
1247 * ubsan.c (ubsan_instrument_unreachable): Call
1248 initialize_sanitizer_builtins.
1249 (ubsan_pass): Likewise.
1251 PR sanitizer/60535
1252 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1253 varpool_finalize_decl instead of rest_of_decl_compilation.
1255 2014-03-18 Richard Biener <rguenther@suse.de>
1257 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1258 by using bitmap_and_compl instead of bitmap_and_compl_into.
1259 (df_rd_transfer_function): Likewise.
1261 2014-03-18 Richard Biener <rguenther@suse.de>
1263 * doc/lto.texi (fresolution): Fix typo.
1265 2014-03-18 Richard Biener <rguenther@suse.de>
1267 * doc/invoke.texi (flto): Update for changes in 4.9.
1269 2014-03-18 Richard Biener <rguenther@suse.de>
1271 * doc/loop.texi: Remove section on the removed lambda framework.
1272 Update loop docs with recent changes in preserving loop structure.
1274 2014-03-18 Richard Biener <rguenther@suse.de>
1276 * doc/lto.texi (-fresolution): Document.
1278 2014-03-18 Richard Biener <rguenther@suse.de>
1280 * doc/contrib.texi: Adjust my name.
1282 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1284 PR ipa/58721
1285 * internal-fn.c: Include diagnostic-core.h.
1286 (expand_BUILTIN_EXPECT): New function.
1287 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1288 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1289 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1290 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1291 IFN_BUILTIN_EXPECT.
1292 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1293 Revert 3 argument __builtin_expect code.
1294 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1295 * gimple-fold.c (gimple_fold_call): Likewise.
1296 * tree.h (fold_builtin_expect): New prototype.
1297 * builtins.c (build_builtin_expect_predicate): Add predictor
1298 argument, if non-NULL, create 3 argument __builtin_expect.
1299 (fold_builtin_expect): No longer static. Add ARG2 argument,
1300 pass it through to build_builtin_expect_predicate.
1301 (fold_builtin_2): Adjust caller.
1302 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1303 * internal-fn.def (BUILTIN_EXPECT): New.
1305 2014-03-18 Tobias Burnus <burnus@net-b.de>
1307 PR ipa/58721
1308 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1309 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1310 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1312 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1314 PR ipa/58721
1315 * predict.c (combine_predictions_for_bb): Fix up formatting.
1316 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1317 fill what it points to if non-NULL.
1318 (tree_predict_by_opcode): Adjust caller, use the predictor.
1319 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1321 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1323 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1324 proper constant for the store mode.
1326 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1328 * symtab.c (change_decl_assembler_name): Fix transparent alias
1329 chain construction.
1331 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1333 * config/aarch64/aarch64.c: Correct the comments about the
1334 aarch64 stack layout.
1336 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1338 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1339 check for GF_OMP_FOR_KIND_FOR.
1341 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1343 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1344 ymm and zmm register names.
1346 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1348 PR target/60516
1349 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1350 note creation for the 2010-08-31 changes.
1352 2014-03-17 Marek Polacek <polacek@redhat.com>
1354 PR middle-end/60534
1355 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1356 as -fno-tree-loop-vectorize.
1357 (expand_omp_simd): Likewise.
1359 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1361 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1362 (eligible_for_call_delay): New prototype.
1363 * config/sparc/sparc.c (tls_call_delay): Rename into...
1364 (eligible_for_call_delay): ...this. Return false if the instruction
1365 cannot be put in the delay slot of a branch.
1366 (eligible_for_restore_insn): Simplify.
1367 (eligible_for_return_delay): Return false if the instruction cannot be
1368 put in the delay slot of a branch and simplify.
1369 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1370 put in the delay slot of a branch.
1371 * config/sparc/sparc.md (fix_ut699): New attribute.
1372 (tls_call_delay): Delete.
1373 (in_call_delay): Reimplement.
1374 (eligible_for_sibcall_delay): Rename into...
1375 (in_sibcall_delay): ...this.
1376 (eligible_for_return_delay): Rename into...
1377 (in_return_delay): ...this.
1378 (in_branch_delay): Reimplement.
1379 (in_uncond_branch_delay): Delete.
1380 (in_annul_branch_delay): Delete.
1382 2014-03-14 Richard Henderson <rth@redhat.com>
1384 PR target/60525
1385 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1386 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1387 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1388 (floathi splitters): Remove.
1389 (float<SWI48x>xf2): New pattern.
1390 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1391 code that tried to handle DImode for 32-bit, but which was excluded
1392 by the pattern's condition. Drop allocation of stack temporary.
1393 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1394 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1395 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1396 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1397 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1398 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1399 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1400 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1401 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1402 (*float<SWI48x><X87MODEF>2_i387): Remove.
1403 (all float _with_temp splitters): Remove.
1404 (*float<SWI48x><MODEF>2_i387): New pattern.
1405 (*float<SWI48><MODEF>2_sse): New pattern.
1406 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1407 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1409 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1410 Marek Polacek <polacek@redhat.com>
1412 PR middle-end/60484
1413 * common.opt (dump_base_name_prefixed): New Variable.
1414 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1415 if x_dump_base_name_prefixed is already set, set it at the end.
1417 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1419 PR rtl-optimization/60508
1420 * lra-constraints.c (get_reload_reg): Add new parameter
1421 in_subreg_p.
1422 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1423 Pass the new parameter values.
1425 2014-03-14 Richard Biener <rguenther@suse.de>
1427 * common.opt: Revert unintented changes from r205065.
1428 * opts.c: Likewise.
1430 2014-03-14 Richard Biener <rguenther@suse.de>
1432 PR middle-end/60518
1433 * cfghooks.c (split_block): Properly adjust all loops the
1434 block was a latch of.
1436 2014-03-14 Martin Jambor <mjambor@suse.cz>
1438 PR lto/60461
1439 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1440 and simplify it.
1442 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1444 PR target/59396
1445 * config/avr/avr.c (avr_set_current_function): Pass function name
1446 through default_strip_name_encoding before sanity checking instead
1447 of skipping the first char of the assembler name.
1449 2014-03-13 Richard Henderson <rth@redhat.com>
1451 PR debug/60438
1452 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1453 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1454 * config/i386/i386-protos.h: Likewise.
1455 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1456 in the expander instead of a splitter.
1457 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1458 any possibility of requiring a memory.
1459 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1460 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1461 (fp branch splitters): Update for ix86_split_fp_branch.
1462 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1463 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1464 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1465 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1466 (*fop_<MODEF>_3_i387): Likewise.
1467 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1468 (splitters for the fop_* register patterns): Remove.
1469 (fscalexf4_i387): Rename from *fscalexf4_i387.
1470 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1472 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1474 PR tree-optimization/59779
1475 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1476 type for bitsize and maxsize instead of HOST_WIDE_INT.
1478 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1480 PR rtl-optimization/57320
1481 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1482 the CFG after thread_prologue_and_epilogue_insns.
1484 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1486 PR rtl-optimization/57189
1487 * lra-constraints.c (process_alt_operands): Disfavor spilling
1488 vector pseudos.
1490 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1492 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1494 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1496 PR tree-optimization/59025
1497 PR middle-end/60418
1498 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1499 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1501 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1503 PR target/60486
1504 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1505 calls of avr_out_plus_1.
1507 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1509 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1510 BB's single pred and update the father loop's latch info later.
1512 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1514 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1515 (VEC_M): Likewise.
1516 (VEC_N): Likewise.
1517 (VEC_R): Likewise.
1518 (VEC_base): Likewise.
1519 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1520 registers, we need to swap double words in little endian mode.
1522 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1523 to be a container mode for 128-bit integer operations added in ISA
1524 2.07. Unlike TImode and PTImode, the preferred register set is
1525 the Altivec/VMX registers for the 128-bit operations.
1527 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1528 declarations.
1529 (rs6000_split_128bit_ok_p): Likewise.
1531 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1532 macros for creating ISA 2.07 normal and overloaded builtin
1533 functions with 3 arguments.
1534 (BU_P8V_OVERLOAD_3): Likewise.
1535 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1536 for use as overloaded functions.
1537 (VPERM_1TI_UNS): Likewise.
1538 (VSEL_1TI): Likewise.
1539 (VSEL_1TI_UNS): Likewise.
1540 (ST_INTERNAL_1ti): Likewise.
1541 (LD_INTERNAL_1ti): Likewise.
1542 (XXSEL_1TI): Likewise.
1543 (XXSEL_1TI_UNS): Likewise.
1544 (VPERM_1TI): Likewise.
1545 (VPERM_1TI_UNS): Likewise.
1546 (XXPERMDI_1TI): Likewise.
1547 (SET_1TI): Likewise.
1548 (LXVD2X_V1TI): Likewise.
1549 (STXVD2X_V1TI): Likewise.
1550 (VEC_INIT_V1TI): Likewise.
1551 (VEC_SET_V1TI): Likewise.
1552 (VEC_EXT_V1TI): Likewise.
1553 (EQV_V1TI): Likewise.
1554 (NAND_V1TI): Likewise.
1555 (ORC_V1TI): Likewise.
1556 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1557 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1558 overloaded builtin.
1559 (VADDUQM): Likewise.
1560 (VSUBCUQ): Likewise.
1561 (VADDEUQM): Likewise.
1562 (VADDECUQ): Likewise.
1563 (VSUBEUQM): Likewise.
1564 (VSUBECUQ): Likewise.
1566 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1567 __int128_t and __uint128_t types.
1568 (__uint128_type): Likewise.
1569 (altivec_categorize_keyword): Add support for vector __int128_t,
1570 vector __uint128_t, vector __int128, and vector unsigned __int128
1571 as a container type for TImode operations that need to be done in
1572 VSX/Altivec registers.
1573 (rs6000_macro_to_expand): Likewise.
1574 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1575 to support 128-bit integer instructions vaddcuq, vadduqm,
1576 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1577 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1579 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1580 for V1TImode, and set up preferences to use VSX/Altivec registers.
1581 Setup VSX reload handlers.
1582 (rs6000_debug_reg_global): Likewise.
1583 (rs6000_init_hard_regno_mode_ok): Likewise.
1584 (rs6000_preferred_simd_mode): Likewise.
1585 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1586 (easy_altivec_constant): Likewise.
1587 (output_vec_const_move): Likewise.
1588 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1589 simple move.
1590 (rs6000_expand_vector_extract): Likewise.
1591 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1592 addressing.
1593 (rs6000_const_vec): Add support for V1TImode.
1594 (rs6000_emit_le_vsx_load): Swap double words when loading or
1595 storing TImode/V1TImode.
1596 (rs6000_emit_le_vsx_store): Likewise.
1597 (rs6000_emit_le_vsx_move): Likewise.
1598 (rs6000_emit_move): Add support for V1TImode.
1599 (altivec_expand_ld_builtin): Likewise.
1600 (altivec_expand_st_builtin): Likewise.
1601 (altivec_expand_vec_init_builtin): Likewise.
1602 (altivec_expand_builtin): Likewise.
1603 (rs6000_init_builtins): Add support for V1TImode type. Add
1604 support for ISA 2.07 128-bit integer builtins. Define type names
1605 for the VSX/Altivec vector types.
1606 (altivec_init_builtins): Add support for overloaded vector
1607 functions with V1TImode type.
1608 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1609 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1610 external function.
1611 (rs6000_split_128bit_ok_p): Likewise.
1612 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1613 __int128_t and vector __uint128_t.
1615 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1616 and mode attributes.
1617 (VSX_M): Likewise.
1618 (VSX_M2): Likewise.
1619 (VSm): Likewise.
1620 (VSs): Likewise.
1621 (VSr): Likewise.
1622 (VSv): Likewise.
1623 (VS_scalar): Likewise.
1624 (VS_double): Likewise.
1625 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1627 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1628 we support the ISA 2.07 128-bit integer arithmetic instructions.
1629 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1630 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1631 and TImode types for use with the builtin functions.
1632 (V1TI_type_node): Likewise.
1633 (unsigned_V1TI_type_node): Likewise.
1634 (intTI_type_internal_node): Likewise.
1635 (uintTI_type_internal_node): Likewise.
1637 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1638 128-bit builtin functions.
1639 (UNSPEC_VADDEUQM): Likewise.
1640 (UNSPEC_VADDECUQ): Likewise.
1641 (UNSPEC_VSUBCUQ): Likewise.
1642 (UNSPEC_VSUBEUQM): Likewise.
1643 (UNSPEC_VSUBECUQ): Likewise.
1644 (VM): Add V1TImode to vector mode iterators.
1645 (VM2): Likewise.
1646 (VI_unit): Likewise.
1647 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1648 (altivec_vaddcuq): Likewise.
1649 (altivec_vsubuqm): Likewise.
1650 (altivec_vsubcuq): Likewise.
1651 (altivec_vaddeuqm): Likewise.
1652 (altivec_vaddecuq): Likewise.
1653 (altivec_vsubeuqm): Likewise.
1654 (altivec_vsubecuq): Likewise.
1656 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1657 mode iterators.
1658 (BOOL_128): Likewise.
1659 (BOOL_REGS_OUTPUT): Likewise.
1660 (BOOL_REGS_OP1): Likewise.
1661 (BOOL_REGS_OP2): Likewise.
1662 (BOOL_REGS_UNARY): Likewise.
1663 (BOOL_REGS_AND_CR0): Likewise.
1665 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1666 128-bit integer builtin support.
1667 (vec_vadduqm): Likewise.
1668 (vec_vaddecuq): Likewise.
1669 (vec_vaddeuqm): Likewise.
1670 (vec_vsubecuq): Likewise.
1671 (vec_vsubeuqm): Likewise.
1672 (vec_vsubcuq): Likewise.
1673 (vec_vsubuqm): Likewise.
1675 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1676 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1677 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1678 128-bit integer add/subtract to ISA 2.07.
1680 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1682 * config/arc/arc.c (arc_predicate_delay_insns):
1683 Fix third argument passed to conditionalize_nonjump.
1685 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1687 * config/aarch64/aarch64-builtins.c
1688 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1689 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1690 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1691 instead of __builtin_lfloor.
1692 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1694 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1696 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1697 (tree_ssa_ifcombine_bb_1): New function.
1698 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1699 is an empty forwarder block to then_bb or vice versa and then_bb
1700 and else_bb are effectively swapped.
1702 2014-03-12 Christian Bruel <christian.bruel@st.com>
1704 PR target/60264
1705 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1706 REG_CFA_DEF_CFA note.
1707 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1708 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1710 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1712 PR tree-optimization/60454
1713 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1715 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1717 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1718 Do not define target_cpu_default2 to generic.
1719 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1720 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1721 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1723 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1724 Marc Glisse <marc.glisse@inria.fr>
1726 PR tree-optimization/60502
1727 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1728 instead of build_low_bits_mask.
1730 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1732 PR middle-end/60482
1733 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1734 if there are multiple uses, but op doesn't live on E edge.
1735 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1736 clobber stmts before __builtin_unreachable.
1738 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1740 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1741 hard_frame_pointer_rtx.
1742 * cse.c (cse_insn): Remove volatile check.
1743 * cselib.c (cselib_process_insn): Likewise.
1744 * dse.c (scan_insn): Likewise.
1746 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1748 * config/arc/arc.c (conditionalize_nonjump): New function,
1749 broken out of ...
1750 (arc_ifcvt): ... this.
1751 (arc_predicate_delay_insns): Use it.
1753 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1755 * config/arc/predicates.md (extend_operand): During/after reload,
1756 allow const_int_operand.
1757 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1758 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1759 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1760 to "i".
1761 (umulsi3_highpart_i): Likewise.
1763 2014-03-11 Richard Biener <rguenther@suse.de>
1765 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1766 Add asserts to guard possible wrong-code bugs.
1768 2014-03-11 Richard Biener <rguenther@suse.de>
1770 PR tree-optimization/60429
1771 PR tree-optimization/60485
1772 * tree-ssa-structalias.c (set_union_with_increment): Properly
1773 take into account all fields that overlap the shifted vars.
1774 (do_sd_constraint): Likewise.
1775 (do_ds_constraint): Likewise.
1776 (get_constraint_for_ptr_offset): Likewise.
1778 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1780 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1781 (nios2_compute_frame_layout):
1782 Add calculation of cfun->machine->fp_save_offset.
1783 (nios2_expand_prologue): Correct setting of frame pointer register
1784 in prologue.
1785 (nios2_expand_epilogue): Update recovery of stack pointer from
1786 frame pointer accordingly.
1787 (nios2_initial_elimination_offset): Update calculation of offset
1788 for eliminating to HARD_FRAME_POINTER_REGNUM.
1790 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1792 PR ipa/60457
1793 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1794 cgraph_get_create_node on VAR_DECLs.
1796 2014-03-10 Richard Biener <rguenther@suse.de>
1798 PR middle-end/60474
1799 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1801 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1803 * config/vms/vms.opt (vms_float_format): New variable.
1805 2014-03-08 Tobias Burnus <burnus@net-b.de>
1807 * doc/invoke.texi (-fcilkplus): Update implementation status.
1809 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1810 Richard Biener <rguenther@suse.de>
1812 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1813 consistently accross all TUs.
1814 (run_gcc): Enable -fshort-double automatically at link at link-time
1815 and disallow override.
1817 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1819 PR target/58271
1820 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1821 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1822 if they can't be used.
1824 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1826 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1827 for Solaris 11/x86 ld.
1828 * configure: Regenerate.
1830 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1832 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1833 (LIB_TLS_SPEC): Save as ld_tls_libs.
1834 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1835 (HAVE_AS_IX86_TLSLDM): New test.
1836 * configure, config.in: Regenerate.
1837 * config/i386/i386.c (legitimize_tls_address): Fall back to
1838 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1839 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1840 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1841 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1843 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1845 * common.opt (fira-loop-pressure): Mark as optimization.
1847 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1849 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1850 an OpenMP mappable type.
1852 2014-03-06 Matthias Klose <doko@ubuntu.com>
1854 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1855 MULTILIB_OSDIRNAMES is not defined.
1857 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1858 Meador Inge <meadori@codesourcery.com>
1860 PR target/58595
1861 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1862 (arm_legitimize_address): Call legitimize_tls_address for any
1863 arm_tls_referenced_p expression, handle constant addend. Call it
1864 before testing for !TARGET_ARM.
1865 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1867 2014-03-06 Richard Biener <rguenther@suse.de>
1869 PR middle-end/60445
1870 PR lto/60424
1871 PR lto/60427
1872 Revert
1873 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1875 * tree-streamer.c (record_common_node): Assert we don't record
1876 nodes with type double.
1877 (preload_common_node): Skip type double, complex double and double
1878 pointer since it is now frontend dependent due to fshort-double option.
1880 2014-03-06 Richard Biener <rguenther@suse.de>
1882 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1883 or -fno-lto is specified and the linker has full plugin support.
1884 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1885 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1886 * lto-wrapper.c (merge_and_complain): Merge compile-time
1887 optimization levels.
1888 (run_gcc): And pass it through to the link options.
1890 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1892 PR debug/60381
1893 Revert:
1894 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1895 PR debug/59992
1896 * cselib.c (remove_useless_values): Skip to avoid quadratic
1897 behavior if the condition moved from...
1898 (cselib_process_insn): ... here holds.
1900 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1902 PR plugins/59335
1903 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1904 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1906 PR plugins/59335
1907 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1908 (TM_H): Add x86-tune.def.
1910 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1912 * config/aarch64/aarch64.c (generic_tunings):
1913 Use cortexa57_extra_costs.
1915 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1917 PR lto/60404
1918 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1919 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1920 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1921 cost for in_lto_p.
1923 2014-03-04 Heiher <r@hev.cc>
1925 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1926 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1928 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1930 * config/i386/predicates.md (const2356_operand): Change to ...
1931 (const2367_operand): ... this.
1932 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1933 const2367_operand.
1934 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1935 (*avx512pf_scatterpf<mode>sf): Ditto.
1936 (avx512pf_scatterpf<mode>df): Ditto.
1937 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1938 (*avx512pf_scatterpf<mode>df): Ditto.
1939 * config/i386/i386.c (ix86_expand_builtin): Update
1940 incorrect hint operand error message.
1942 2014-03-04 Richard Biener <rguenther@suse.de>
1944 * lto-section-in.c (lto_get_section_data): Fix const cast.
1946 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1948 * tree-streamer.c (record_common_node): Assert we don't record
1949 nodes with type double.
1950 (preload_common_node): Skip type double, complex double and double
1951 pointer since it is now frontend dependent due to fshort-double option.
1953 2014-03-04 Richard Biener <rguenther@suse.de>
1955 PR lto/60405
1956 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1957 (lto_input_toplevel_asms): Likewise.
1958 * lto-section-in.c (lto_get_section_data): Instead do it here
1959 for every section.
1961 2014-03-04 Richard Biener <rguenther@suse.de>
1963 PR tree-optimization/60382
1964 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1965 dead PHIs a reduction.
1967 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1969 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1970 hint value.
1971 (_mm_prefetch): Move out of GCC target("sse") pragma.
1972 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1973 GCC target("prfchw") pragma.
1974 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1975 for locality <= 2.
1976 * config/i386/i386.c (ix86_option_override_internal): Enable
1977 -mprfchw with -mprefetchwt1.
1979 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1981 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1982 Mark as varying.
1984 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1986 * opts.h (CL_PCH_IGNORE): Define.
1987 * targhooks.c (option_affects_pch_p):
1988 Return false for options that have CL_PCH_IGNORE set.
1989 * opt-functions.awk: Process PchIgnore.
1990 * doc/options.texi: Document PchIgnore.
1992 * config/arc/arc.opt (misize): Add PchIgnore property.
1994 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1996 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1997 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1998 constraint on constants to permit them being loaded into
1999 GENERAL_REGS or BASE_REGS.
2001 2014-03-03 Nick Clifton <nickc@redhat.com>
2003 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2004 anti-cacnonical alternatives.
2005 (negandhi3_real): New pattern.
2006 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2008 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2010 * config/avr/avr-mcus.def: Remove atxmega16x1.
2011 * config/avr/avr-tables.opt: Regenerate.
2012 * config/avr/t-multilib: Regenerate.
2013 * doc/avr-mmcu.texi: Regenerate.
2015 2014-03-03 Tobias Grosser <tobias@grosser.es>
2016 Mircea Namolaru <mircea.namolaru@inria.fr>
2018 PR tree-optimization/58028
2019 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2020 scalar dimensions.
2022 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2024 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2025 not handled by recognizers.
2027 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2029 PR middle-end/60175
2030 * function.c (expand_function_end): Don't emit
2031 clobber_return_register sequence if clobber_after is a BARRIER.
2032 * cfgexpand.c (construct_exit_block): Append instructions before
2033 return_label to prev_bb.
2035 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2037 * config/rs6000/constraints.md: Document reserved use of "wc".
2039 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2041 PR ipa/60150
2042 * ipa.c (function_and_variable_visibility): When dissolving comdat
2043 group, also set all symbols to local.
2045 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2047 PR ipa/60306
2049 Revert:
2050 2013-12-14 Jan Hubicka <jh@suse.cz>
2051 PR middle-end/58477
2052 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2054 2014-03-02 Jon Beniston <jon@beniston.com>
2056 PR bootstrap/48230
2057 PR bootstrap/50927
2058 PR bootstrap/52466
2059 PR target/46898
2060 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2061 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2062 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2063 (simple_return, *simple_return): New patterns
2064 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2065 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2067 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2069 * dwarf2out.c (gen_subprogram_die): Tidy.
2071 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2073 PR target/60071
2074 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2075 (*mov_t_msb_neg_negc): ... this new insn.
2077 2014-02-28 Jason Merrill <jason@redhat.com>
2079 PR c++/58678
2080 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2081 function.
2083 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2085 PR c++/60314
2086 * dwarf2out.c (decltype_auto_die): New static.
2087 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2088 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2089 (is_cxx_auto): Likewise.
2091 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2093 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2094 we are not using general regs only.
2096 2014-02-28 Richard Biener <rguenther@suse.de>
2098 PR target/60280
2099 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2100 previous fix and only allow to remove trivial pre-headers
2101 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2102 (remove_forwarder_block): Properly update the latch of a loop.
2104 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2106 PR debug/59992
2107 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2108 (cselib_preserved_hash_table): New.
2109 (preserve_constants_and_equivs): Move preserved vals to it.
2110 (cselib_find_slot): Look it up first.
2111 (cselib_init): Initialize it.
2112 (cselib_finish): Release it.
2113 (dump_cselib_table): Dump it.
2115 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2117 PR debug/59992
2118 * cselib.c (remove_useless_values): Skip to avoid quadratic
2119 behavior if the condition moved from...
2120 (cselib_process_insn): ... here holds.
2122 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2124 PR debug/57232
2125 * var-tracking.c (vt_initialize): Apply the same condition to
2126 preserve the CFA base value.
2128 2014-02-28 Joey Ye <joey.ye@arm.com>
2130 PR target/PR60169
2131 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2132 if reload in progress or completed.
2134 2014-02-28 Tobias Burnus <burnus@net-b.de>
2136 PR middle-end/60147
2137 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2138 NAMELIST_DECL.
2140 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2142 * doc/tm.texi.in (Condition Code Status): Update documention for
2143 relative locations of cc0-setter and cc0-user.
2145 2014-02-27 Jeff Law <law@redhat.com>
2147 PR rtl-optimization/52714
2148 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2149 into two independent simple sets, if I3 is a jump, ensure the
2150 pattern we place into I3 is a (set (pc) ...).
2152 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2153 Jeff Law <law@redhat.com>
2155 PR rtl-optimization/49847
2156 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2157 are in different blocks.
2158 * doc/tm.texi (Condition Code Status): Update documention for
2159 relative locations of cc0-setter and cc0-user.
2161 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2163 PR target/59222
2164 * lra.c (lra_emit_add): Check SUBREG too.
2166 2014-02-27 Andreas Schwab <schwab@suse.de>
2168 * config/m68k/m68k.c (m68k_option_override): Disable
2169 -flive-range-shrinkage for classic m68k.
2170 (m68k_override_options_after_change): Likewise.
2172 2014-02-27 Marek Polacek <polacek@redhat.com>
2174 PR middle-end/59223
2175 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2176 -Wmaybe-uninitialized.
2178 2014-02-27 Alan Modra <amodra@gmail.com>
2180 PR target/57936
2181 * reload1.c (emit_input_reload_insns): When reload_override_in,
2182 set old to rl->in_reg when rl->in_reg is a subreg.
2184 2014-02-26 Richard Biener <rguenther@suse.de>
2186 PR bootstrap/60343
2187 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2189 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2191 * common/config/i386/predicates.md (const1256_operand): Remove.
2192 (const2356_operand): New.
2193 (const_1_to_2_operand): Remove.
2194 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2195 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2196 (*avx512pf_gatherpf<mode>sf): Ditto.
2197 (avx512pf_gatherpf<mode>df): Ditto.
2198 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2199 (*avx512pf_gatherpf<mode>df): Ditto.
2200 (avx512pf_scatterpf<mode>sf): Ditto.
2201 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2202 (*avx512pf_scatterpf<mode>sf): Ditto.
2203 (avx512pf_scatterpf<mode>df): Ditto.
2204 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2205 (*avx512pf_scatterpf<mode>df): Ditto.
2206 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2208 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2210 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2211 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2212 (_mm512_mask_testn_epi64_mask): Move to ...
2213 * config/i386/avx512cdintrin.h: Here.
2214 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2215 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2216 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2217 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2218 TARGET_AVX512F from TARGET_AVX512CD.
2220 2014-02-26 Richard Biener <rguenther@suse.de>
2222 PR ipa/60327
2223 * ipa.c (walk_polymorphic_call_targets): Properly guard
2224 call to inline_update_overall_summary.
2226 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2228 PR target/60280
2229 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2230 and latches only if requested. Fix latch if it is removed.
2231 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2232 LOOPS_HAVE_PREHEADERS.
2234 2014-02-25 Andrew Pinski <apinski@cavium.com>
2236 * builtins.c (expand_builtin_thread_pointer): Create a new target
2237 when the target is NULL.
2239 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2241 PR rtl-optimization/60317
2242 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2243 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2244 * lra-assigns.c: Include params.h.
2245 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2246 other reload pseudos considerations.
2248 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2250 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2251 to use canonical form for nor<mode>3.
2253 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2255 PR target/55426
2256 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2257 conversions.
2259 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2261 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2262 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2263 (ix86_handle_option): Handle OPT_mprefetchwt1.
2264 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2265 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2266 PREFETCHWT1 CPUID.
2267 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2268 OPTION_MASK_ISA_PREFETCHWT1.
2269 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2270 (PTA_PREFETCHWT1): New.
2271 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2272 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2273 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2274 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2275 (*prefetch_avx512pf_<mode>_: Change into ...
2276 (*prefetch_prefetchwt1_<mode>: This.
2277 * config/i386/i386.opt (mprefetchwt1): New.
2278 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2279 (_mm_prefetch): Handle intent to write.
2280 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2282 2014-02-25 Richard Biener <rguenther@suse.de>
2284 PR middle-end/60291
2285 * emit-rtl.c (mem_attrs_htab): Remove.
2286 (mem_attrs_htab_hash): Likewise.
2287 (mem_attrs_htab_eq): Likewise.
2288 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2289 (init_emit_once): Do not allocate mem_attrs_htab.
2291 2014-02-25 Richard Biener <rguenther@suse.de>
2293 PR lto/60319
2294 * lto-opts.c (lto_write_options): Output non-explicit conservative
2295 -fwrapv, -fno-trapv and -fno-strict-overflow.
2296 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2297 (run_gcc): And pass them through.
2299 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2301 * sel-sched.c (calculate_new_fences): New parameter ptime.
2302 Calculate it as a maximum over all fence cycles.
2303 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2304 Print the final schedule timing when sched_verbose.
2306 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2308 PR rtl-optimization/60292
2309 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2310 bit fot the fence instruction.
2312 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2314 * calls.h: Fix typo in comment.
2316 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2318 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2319 adjusting offsetable addresses.
2321 2014-02-24 Guozhi Wei <carrot@google.com>
2323 * sparseset.h (sparseset_pop): Fix the wrong index.
2325 2014-02-24 Walter Lee <walt@tilera.com>
2327 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2328 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2329 triplet.
2330 * common/config/tilegx/tilegx-common.c
2331 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2332 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2333 (LINK_SPEC): Ditto.
2334 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2335 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2336 (tilegx_gimplify_va_arg_expr): Handle big endian.
2337 (tilegx_expand_unaligned_load): Ditto.
2338 (tilegx_expand_unaligned_store): Ditto.
2339 (TARGET_RETURN_IN_MSB): New.
2340 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2341 (TARGET_ENDIAN_DEFAULT): New.
2342 (TARGET_BIG_ENDIAN): Handle big endian.
2343 (BYTES_BIG_ENDIAN): Ditto.
2344 (WORDS_BIG_ENDIAN): Ditto.
2345 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2346 (ENDIAN_SPEC): New.
2347 (EXTRA_SPECS): New.
2348 * config/tilegx/tilegx.md (extv): Handle big endian.
2349 (extzv): Ditto.
2350 (insn_st<n>): Ditto.
2351 (insn_st<n>_add<bitsuffix>): Ditto.
2352 (insn_stnt<n>): Ditto.
2353 (insn_stnt<n>_add<bitsuffix>):Ditto.
2354 (vec_interleave_highv8qi): Handle big endian.
2355 (vec_interleave_highv8qi_be): New.
2356 (vec_interleave_highv8qi_le): New.
2357 (insn_v1int_h): Handle big endian.
2358 (vec_interleave_lowv8qi): Handle big endian.
2359 (vec_interleave_lowv8qi_be): New.
2360 (vec_interleave_lowv8qi_le): New.
2361 (insn_v1int_l): Handle big endian.
2362 (vec_interleave_highv4hi): Handle big endian.
2363 (vec_interleave_highv4hi_be): New.
2364 (vec_interleave_highv4hi_le): New.
2365 (insn_v2int_h): Handle big endian.
2366 (vec_interleave_lowv4hi): Handle big endian.
2367 (vec_interleave_lowv4hi_be): New.
2368 (vec_interleave_lowv4hi_le): New.
2369 (insn_v2int_l): Handle big endian.
2370 (vec_interleave_highv2si): Handle big endian.
2371 (vec_interleave_highv2si_be): New.
2372 (vec_interleave_highv2si_le): New.
2373 (insn_v4int_h): Handle big endian.
2374 (vec_interleave_lowv2si): Handle big endian.
2375 (vec_interleave_lowv2si_be): New.
2376 (vec_interleave_lowv2si_le): New.
2377 (insn_v4int_l): Handle big endian.
2378 * config/tilegx/tilegx.opt (mbig-endian): New option.
2379 (mlittle-endian): New option.
2380 * doc/install.texi: Document tilegxbe-linux.
2381 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2383 2014-02-24 Martin Jambor <mjambor@suse.cz>
2385 PR ipa/60266
2386 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2387 there are no parameter descriptors.
2389 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2391 PR rtl-optimization/60268
2392 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2393 initialization to ...
2394 (sched_rgn_init): ... here.
2395 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2397 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2399 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2400 names.
2402 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2404 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2405 definition.
2407 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2409 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2410 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2412 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2414 * config/microblaze/predicates.md: Add cmp_op predicate.
2415 * config/microblaze/microblaze.md: Add branch_compare instruction
2416 which uses cmp_op predicate and emits cmp insn before branch.
2417 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2418 to microblaze_expand_conditional_branch and consolidate logic.
2419 (microblaze_expand_conditional_branch): emit branch_compare
2420 insn instead of handling cmp op separate from branch insn.
2422 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2424 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2425 to permit subregs.
2427 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2429 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2430 define_insn with define_expand and new define_insn
2431 *altivec_lve<VI_char>x_internal.
2432 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2433 and new define_insn *altivec_stve<VI_char>x_internal.
2434 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2435 prototype.
2436 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2437 lve*x built-ins.
2438 (altivec_expand_stvex_be): New function.
2440 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2442 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2443 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2444 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2445 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2447 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2449 PR target/60298
2450 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2451 instead of emit_move_insn.
2453 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2455 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2456 vspltw with vsldoi.
2457 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2458 gen_altivec_vsumsws.
2460 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2462 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2463 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2464 (altivec_lvxl_<mode>): New define_expand incorporating
2465 -maltivec=be semantics where needed.
2466 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2467 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2468 semantics where needed.
2469 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2470 (altivec_stvx_<mode>): New define_expand incorporating
2471 -maltivec=be semantics where needed.
2472 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2473 VM2 iterator instead of V4SI.
2474 (altivec_stvxl_<mode>): New define_expand incorporating
2475 -maltivec=be semantics where needed.
2476 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2477 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2478 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2479 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2480 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2481 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2482 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2483 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2484 ALTIVEC_BUILTIN_STVXL.
2485 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2486 (altivec_expand_stvx_be): Likewise.
2487 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2488 (altivec_expand_lvx_be): Likewise.
2489 (altivec_expand_stvx_be): Likewise.
2490 (altivec_expand_builtin): Add cases for
2491 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2492 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2493 (altivec_init_builtins): Add definitions for
2494 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2495 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2497 2014-02-21 Catherine Moore <clm@codesourcery.com>
2499 * doc/invoke.texi (mvirt, mno-virt): Document.
2500 * config/mips/mips.opt (mvirt): New option.
2501 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2503 2014-02-21 Richard Biener <rguenther@suse.de>
2505 PR tree-optimization/60276
2506 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2507 (STMT_VINFO_MIN_NEG_DIST): New macro.
2508 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2509 STMT_VINFO_MIN_NEG_DIST.
2510 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2511 made for negative dependence distances still hold.
2513 2014-02-21 Richard Biener <rguenther@suse.de>
2515 PR middle-end/60291
2516 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2517 DECL_INITIAL for globals not in the current function context.
2519 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2521 PR tree-optimization/56490
2522 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2523 * tree-ssa-uninit.c: Include params.h.
2524 (compute_control_dep_chain): Add num_calls argument, return false
2525 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2526 num_calls to recursive call.
2527 (find_predicates): Change dep_chain into normal array,
2528 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2529 variable and adjust compute_control_dep_chain caller.
2530 (find_def_preds): Likewise.
2532 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2534 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2535 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2537 2014-02-21 Nick Clifton <nickc@redhat.com>
2539 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2540 (pushhi1): Likewise.
2541 (popqi1): Add mode to pre_dec.
2542 (pophi1): Likewise.
2544 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2546 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2547 mode for mask of V8SFmode permutation.
2549 2014-02-20 Richard Henderson <rth@redhat.com>
2551 PR c++/60272
2552 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2553 a new pseudo for OLDVAL.
2555 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2557 PR target/57896
2558 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2559 gen_reg_rtx if d->testing_p.
2560 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2561 if d->testing_p and we will certainly return true.
2562 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2563 if d->testing_p.
2565 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2567 * emit-rtl.c (gen_reg_rtx): Assert that
2568 crtl->emit.regno_pointer_align_length is non-zero.
2570 2014-02-20 Richard Henderson <rth@redhat.com>
2572 PR c++/60272
2573 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2574 on failure the store back into EXPECT.
2576 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2577 Sandra Loosemore <sandra@codesourcery.com>
2579 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2580 * config/nios2/nios2.c (nios2_function_profiler): Add
2581 -fPIC (flag_pic == 2) support.
2582 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2583 (nios2_large_offset_p): New function.
2584 (nios2_unspec_reloc_p): Move up position, update to use
2585 nios2_large_offset_p.
2586 (nios2_unspec_address): Remove function.
2587 (nios2_unspec_offset): New function.
2588 (nios2_large_got_address): New function.
2589 (nios2_got_address): Add large offset support.
2590 (nios2_legitimize_tls_address): Update usage of removed and new
2591 functions.
2592 (nios2_symbol_binds_local_p): New function.
2593 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2594 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2595 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2596 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2597 processing for (const (unspec ...)).
2598 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2600 2014-02-20 Richard Biener <rguenther@suse.de>
2602 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2603 doing the substitution.
2604 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2606 2014-02-20 Martin Jambor <mjambor@suse.cz>
2608 PR ipa/55260
2609 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2610 info when checking whether lattices are bottom.
2612 2014-02-20 Richard Biener <rguenther@suse.de>
2614 PR middle-end/60221
2615 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2616 regions at -O0.
2618 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2620 PR ipa/58555
2621 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2622 parameter specifying the scaling.
2623 (inline_call): Update.
2624 (want_inline_recursively): Guard division by zero.
2625 (recursive_inlining): Update.
2626 * ipa-inline.h (clone_inlined_nodes): Update.
2628 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2630 PR target/60204
2631 * config/i386/i386.c (classify_argument): Pass structures of size
2632 64 bytes or less in register.
2634 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2635 Kirill Yukhin <kirill.yukhin@intel.com>
2637 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2638 (_mm_rcp28_round_ss): Ditto.
2639 (_mm_rsqrt28_round_sd): Ditto.
2640 (_mm_rsqrt28_round_ss): Ditto.
2641 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2642 (_mm_rcp14_round_ss): Ditto.
2643 (_mm_rsqrt14_round_sd): Ditto.
2644 (_mm_rsqrt14_round_ss): Ditto.
2645 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2646 the first input operand, get rid of match_dup.
2647 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2648 attribute to sse.
2649 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2650 Ditto.
2651 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2652 operand as the first input operand, set type attribute.
2653 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2654 Set type attribute.
2655 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2656 operand as the first input operand, set type attribute.
2658 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2660 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2661 bit of zero.
2663 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2665 PR target/60207
2666 * config/i386/i386.c (construct_container): Remove TFmode check
2667 for X86_64_INTEGER_CLASS.
2669 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2671 PR target/59794
2672 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2673 only when -Wpsabi is enabled.
2675 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2677 PR target/59799
2678 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2679 passing arrays in registers are the same as for structs, so remove the
2680 special case for them.
2682 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2684 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2685 destination type, extract only the valid bits if the source type is not
2686 integral and has a different mode.
2688 2014-02-19 Richard Biener <rguenther@suse.de>
2690 PR ipa/60243
2691 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2692 for all calls.
2694 2014-02-19 Richard Biener <rguenther@suse.de>
2696 PR ipa/60243
2697 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2698 (ipa_modify_call_arguments): Emit an argument load explicitely and
2699 preserve virtual SSA form there and for the replacement call.
2700 Do not update SSA form nor free dominance info.
2702 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2704 * ipa.c (function_and_variable_visibility): Also clear WEAK
2705 flag when disolving COMDAT_GROUP.
2707 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2709 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2710 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2711 not devirtualizing.
2712 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2713 do more sanity checks.
2714 (detect_type_change): Return true when giving up early.
2715 (compute_complex_assign_jump_func): Fix type parameter of
2716 ipa_set_ancestor_jf.
2717 (compute_complex_ancestor_jump_func): Likewise.
2718 (update_jump_functions_after_inlining): Fix updating of
2719 ancestor function.
2720 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2722 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2724 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2725 inline clones when edge disappears.
2727 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2729 PR target/60203
2730 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2731 Split 64-bit moves into 2 patterns. Do not allow the use of
2732 direct move for TDmode in little endian, since the decimal value
2733 has little endian bytes within a word, but the 64-bit pieces are
2734 ordered in a big endian fashion, and normal subreg's of TDmode are
2735 not allowed.
2736 (mov<mode>_64bit_dm): Likewise.
2737 (movtd_64bit_nodm): Likewise.
2739 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2741 PR tree-optimization/60174
2742 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2743 statement of an SSA_NAME that occurs in an abnormal PHI node.
2745 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2747 PR sanitizer/60142
2748 * final.c (SEEN_BB): Remove.
2749 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2750 (final_scan_insn): Don't force_source_line on second
2751 NOTE_INSN_BASIC_BLOCK.
2753 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2755 PR target/60205
2756 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2757 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2758 (type_natural_mode): Warn ABI change when %zmm register is not
2759 available for AVX512F vector value passing.
2761 2014-02-18 Kai Tietz <ktietz@redhat.com>
2763 PR target/60193
2764 * config/i386/i386.c (ix86_expand_prologue): Use value in
2765 rax register as displacement when restoring %r10 or %rax.
2766 Fix wrong offset when restoring both registers.
2768 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2770 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2771 assertion with conditional return.
2773 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2774 Uros Bizjak <ubizjak@gmail.com>
2776 PR driver/60233
2777 * config/i386/driver-i386.c (host_detect_local_cpu): If
2778 YMM state is not saved by the OS, also clear has_f16c. Move
2779 CPUID 0x80000001 handling before YMM state saving checking.
2781 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2783 PR rtl-optimization/58960
2784 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2785 factored out from ...
2786 (sched_init): ... here.
2787 (free_global_sched_pressure_data): New, factored out from ...
2788 (sched_finish): ... here.
2789 * sched-int.h (free_global_sched_pressure_data): Declare.
2790 * sched-rgn.c (nr_regions_initial): New static global.
2791 (haifa_find_rgns): Initialize it.
2792 (schedule_region): Disable sched-pressure for the newly
2793 generated regions.
2795 2014-02-17 Richard Biener <rguenther@suse.de>
2797 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2798 release SSA defs of pattern stmts.
2800 2014-02-17 Richard Biener <rguenther@suse.de>
2802 * tree-inline.c (expand_call_inline): Release the virtual
2803 operand defined by the call we are about to inline.
2805 2014-02-17 Richard Biener <rguenther@suse.de>
2807 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2809 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2810 Ilya Tocar <ilya.tocar@intel.com>
2812 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2813 arguments order in builtin.
2814 (_mm512_permutexvar_epi64): Ditto.
2815 (_mm512_mask_permutexvar_epi64): Ditto
2816 (_mm512_maskz_permutexvar_epi32): Ditto
2817 (_mm512_permutexvar_epi32): Ditto
2818 (_mm512_mask_permutexvar_epi32): Ditto
2820 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2822 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2823 (p8_vmrgow): Likewise.
2825 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2827 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2828 endian targets.
2830 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2832 PR target/60203
2833 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2834 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2835 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2836 using direct move instructions on ISA 2.07. Also adjust
2837 instruction length for 64-bit.
2838 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2839 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2841 2014-02-15 Alan Modra <amodra@gmail.com>
2843 PR target/58675
2844 PR target/57935
2845 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2846 find_replacement on parts of insn rtl that might be reloaded.
2848 2014-02-15 Richard Biener <rguenther@suse.de>
2850 PR tree-optimization/60183
2851 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2852 (tree_ssa_phiprop): Calculate and free post-dominators.
2854 2014-02-14 Jeff Law <law@redhat.com>
2856 PR rtl-optimization/60131
2857 * ree.c (get_extended_src_reg): New function.
2858 (combine_reaching_defs): Use it rather than assuming location of REG.
2859 (find_and_remove_re): Verify first operand of extension is
2860 a REG before adding the insns to the copy list.
2862 2014-02-14 Roland McGrath <mcgrathr@google.com>
2864 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2865 * configure: Regenerated.
2866 * config.in: Regenerated.
2867 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2868 instead of ASM_SHORT.
2870 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2871 Richard Earnshaw <rearnsha@arm.com>
2873 PR rtl-optimization/59535
2874 * lra-constraints.c (process_alt_operands): Encourage alternative
2875 when unassigned pseudo class is superset of the alternative class.
2876 (inherit_reload_reg): Don't inherit when optimizing for code size.
2877 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2878 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2879 modes not less than 4 for Thumb1.
2881 2014-02-14 Kyle McMartin <kyle@redhat.com>
2883 PR pch/60010
2884 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2886 2014-02-14 Richard Biener <rguenther@suse.de>
2888 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2889 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2890 Do not strip INDIRECT_REFs.
2892 2014-02-14 Richard Biener <rguenther@suse.de>
2894 PR lto/60179
2895 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2896 DECL_FUNCTION_SPECIFIC_TARGET.
2897 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2898 * tree-streamer-out.c (pack_ts_target_option): Remove.
2899 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2900 (write_ts_function_decl_tree_pointers): Do not stream
2901 DECL_FUNCTION_SPECIFIC_TARGET.
2902 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2903 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2904 (lto_input_ts_function_decl_tree_pointers): Do not stream
2905 DECL_FUNCTION_SPECIFIC_TARGET.
2907 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2909 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2910 (get_initial_def_for_induction, vectorizable_induction): Ignore
2911 debug stmts when looking for exit_phi.
2912 (vectorizable_live_operation): Fix up condition.
2914 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2916 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2917 nreverse() because it changes the content of original tree list.
2919 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2921 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2922 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2924 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2926 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2927 GNU coding standards.
2929 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2931 PR debug/60152
2932 * dwarf2out.c (gen_subprogram_die): Don't call
2933 add_calling_convention_attribute if subr_die is old_die.
2935 2014-02-13 Sharad Singhai <singhai@google.com>
2937 * doc/optinfo.texi: Fix order of nodes.
2939 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2941 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2942 operands[2], not operands[3].
2944 2014-02-13 Richard Biener <rguenther@suse.de>
2946 PR bootstrap/59878
2947 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2948 mention the possibility of an in-tree build.
2949 (CLooG): Update recommended version to 0.18.1, mention the
2950 possibility of an in-tree build and clarify that the ISL
2951 bundled with CLooG does not work.
2953 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2955 PR target/43546
2956 * expr.c (compress_float_constant): If x is a hard register,
2957 extend into a pseudo and then move to x.
2959 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2961 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2962 caused by bad second argument to warning_at() with -mhotpatch and
2963 nested functions (e.g. with gfortran).
2965 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2967 * opts.c (option_name): Remove "enabled by default" rider.
2969 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2971 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2973 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2974 Uros Bizjak <ubizjak@gmail.com>
2976 PR target/60151
2977 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2978 * configure: Regenerated.
2980 2014-02-12 Richard Biener <rguenther@suse.de>
2982 * vec.c (vec_prefix::calculate_allocation): Move as
2983 inline variant to vec.h.
2984 (vec_prefix::calculate_allocation_1): New out-of-line version.
2985 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2986 (vec_prefix::m_has_auto_buf): Rename to ...
2987 (vec_prefix::m_using_auto_storage): ... this.
2988 (vec_prefix::calculate_allocation): Inline the easy cases
2989 and dispatch to calculate_allocation_1 which doesn't need the
2990 prefix address.
2991 (va_heap::reserve): Use gcc_checking_assert.
2992 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2993 m_using_auto_storage.
2994 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2995 member and adjust.
2996 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2997 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2998 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3000 2014-02-12 Richard Biener <rguenther@suse.de>
3002 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3003 when we found a dependence.
3005 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3007 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3008 common code...
3009 (maybe_fold_stmt): ... into this new function.
3010 * omp-low.c (lower_omp): Update comment.
3012 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3013 last use.
3015 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3016 dereference.
3018 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3020 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3021 identifiers in comments.
3022 (cortexa53_extra_costs): Likewise.
3023 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3024 (cortexa7_extra_costs): Likewise.
3025 (cortexa12_extra_costs): Likewise.
3026 (cortexa15_extra_costs): Likewise.
3027 (v7m_extra_costs): Likewise.
3029 2014-02-12 Richard Biener <rguenther@suse.de>
3031 PR middle-end/60092
3032 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3033 of posix_memalign being successful.
3034 (lower_stmt): Restrict lowering of posix_memalign to when
3035 -ftree-bit-ccp is enabled.
3037 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3039 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3040 arg_loc.
3041 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3043 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3045 PR rtl-optimization/60116
3046 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3047 other_insn once the combination has been validated.
3049 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3051 PR lto/59468
3052 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3053 and wrapper.
3054 * ipa-devirt.c: Include demangle.h
3055 (odr_violation_reported): New static variable.
3056 (add_type_duplicate): Update odr_violations.
3057 (maybe_record_node): Add completep parameter; update it.
3058 (record_target_from_binfo): Add COMPLETEP parameter;
3059 update it as needed.
3060 (possible_polymorphic_call_targets_1): Likewise.
3061 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3062 rename FINAL to COMPLETE.
3063 (record_targets_from_bases): Sanity check we found the binfo;
3064 fix COMPLETEP updating.
3065 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3066 parameter, fix computing of COMPLETEP.
3067 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3068 at LTO time do demangling.
3069 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3070 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3071 parameter.
3072 (gimple_get_virt_method_for_binfo): Likewise.
3073 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3074 gimple_get_virt_method_for_vtable): Update prototypes.
3076 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3078 PR target/49008
3079 * genautomata.c (add_presence_absence): Fix typo with
3080 {final_}presence_list.
3082 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3084 PR target/60137
3085 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3086 for VSX/Altivec vectors that land in GPR registers.
3088 2014-02-11 Richard Henderson <rth@redhat.com>
3089 Jakub Jelinek <jakub@redhat.com>
3091 PR debug/59776
3092 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3093 around drhs if type conversion to lacc->type is not useless.
3095 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3097 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3098 tuning struct.
3099 (cortex-a57.cortex-a53): Likewise.
3100 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3102 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3104 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3105 arm_restrict_it.
3107 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3109 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3110 add_options_for_arm_vfp3.
3112 2014-02-11 Jeff Law <law@redhat.com>
3114 PR middle-end/54041
3115 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3116 object with an undesirable mode.
3118 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3120 PR libgomp/60107
3121 * config/i386/sol2-9.h: New file.
3122 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3123 *-*-solaris2.9*): Use it.
3125 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3127 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3128 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3130 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3132 * config/microblaze/microblaze.c: Extend mcpu version format
3134 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3136 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3138 2014-02-10 Richard Henderson <rth@redhat.com>
3140 PR target/59927
3141 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3142 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3143 ms-abi vs -mno-accumulate-outgoing-args.
3144 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3145 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3146 respect to ms-abi.
3148 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3150 PR middle-end/60080
3151 * cfgexpand.c (expand_asm_operands): Attach source location to
3152 ASM_INPUT rtx objects.
3153 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3155 2014-02-10 Nick Clifton <nickc@redhat.com>
3157 * config/mn10300/mn10300.c (popcount): New function.
3158 (mn10300_expand_prologue): Include saved registers in stack usage
3159 count.
3161 2014-02-10 Jeff Law <law@redhat.com>
3163 PR middle-end/52306
3164 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3165 when changing the SET_DEST of a prior insn to avoid an input reload.
3167 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3169 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3170 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3171 -mcall-openbsd, or -mcall-linux.
3172 (CC1_ENDIAN_BIG_SPEC): Remove.
3173 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3174 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3175 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3176 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3177 and %cc1_endian_default.
3178 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3180 2014-02-10 Richard Biener <rguenther@suse.de>
3182 PR tree-optimization/60115
3183 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3184 MEM_REF handling. Properly verify that the accesses are not
3185 out of the objects bound.
3187 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3189 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3190 coretex to cortex.
3192 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3194 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3195 proper constants and fix formatting.
3196 (possible_polymorphic_call_targets): Fix formatting.
3198 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3199 Ilya Tocar <ilya.tocar@intel.com>
3201 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3202 (_mm512_loadu_epi32): Renamed into...
3203 (_mm512_loadu_si512): This.
3204 (_mm512_storeu_epi32): Renamed into...
3205 (_mm512_storeu_si512): This.
3206 (_mm512_maskz_ceil_ps): Removed.
3207 (_mm512_maskz_ceil_pd): Ditto.
3208 (_mm512_maskz_floor_ps): Ditto.
3209 (_mm512_maskz_floor_pd): Ditto.
3210 (_mm512_floor_round_ps): Ditto.
3211 (_mm512_floor_round_pd): Ditto.
3212 (_mm512_ceil_round_ps): Ditto.
3213 (_mm512_ceil_round_pd): Ditto.
3214 (_mm512_mask_floor_round_ps): Ditto.
3215 (_mm512_mask_floor_round_pd): Ditto.
3216 (_mm512_mask_ceil_round_ps): Ditto.
3217 (_mm512_mask_ceil_round_pd): Ditto.
3218 (_mm512_maskz_floor_round_ps): Ditto.
3219 (_mm512_maskz_floor_round_pd): Ditto.
3220 (_mm512_maskz_ceil_round_ps): Ditto.
3221 (_mm512_maskz_ceil_round_pd): Ditto.
3222 (_mm512_expand_pd): Ditto.
3223 (_mm512_expand_ps): Ditto.
3224 * config/i386/i386.c (ix86_builtins): Remove
3225 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3226 (bdesc_args): Ditto.
3227 * config/i386/predicates.md (const1256_operand): New.
3228 (const_1_to_2_operand): Ditto.
3229 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3230 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3231 (*avx512pf_gatherpf<mode>sf): Ditto.
3232 (avx512pf_gatherpf<mode>df): Ditto.
3233 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3234 (*avx512pf_gatherpf<mode>df): Ditto.
3235 (avx512pf_scatterpf<mode>sf): Ditto.
3236 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3237 (*avx512pf_scatterpf<mode>sf): Ditto.
3238 (avx512pf_scatterpf<mode>df): Ditto.
3239 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3240 (*avx512pf_scatterpf<mode>df): Ditto.
3241 (avx512f_expand<mode>): Removed.
3242 (<shift_insn><mode>3<mask_name>): Change predicate type.
3244 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3246 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3247 not at the end of datarefs vector use ordered_remove to avoid
3248 reordering datarefs vector.
3250 PR c/59984
3251 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3252 mark local addressable non-static vars as GOVD_PRIVATE
3253 instead of GOVD_LOCAL.
3254 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3255 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3256 than copying them.
3258 PR middle-end/60092
3259 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3260 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3261 assume_aligned or alloc_align attributes.
3262 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3263 arguments. Handle also assume_aligned and alloc_align attributes.
3264 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3265 calls to functions with assume_aligned or alloc_align attributes.
3266 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3268 2014-02-08 Terry Guo <terry.guo@arm.com>
3270 * doc/invoke.texi: Document ARM -march=armv7e-m.
3272 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3274 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3275 flag on __cilkrts_rethrow builtin.
3277 PR ipa/60026
3278 * ipa-cp.c (determine_versionability): Fail at -O0
3279 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3280 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3282 Revert:
3283 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3285 PR ipa/60026
3286 * tree-inline.c (copy_forbidden): Fail for
3287 __attribute__((optimize (0))) functions.
3289 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3291 * varpool.c: Include pointer-set.h.
3292 (varpool_remove_unreferenced_decls): Variables in other partitions
3293 will not be output; be however careful to not lose information
3294 about partitioning.
3296 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3298 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3299 lookup in the vtable constructor.
3301 2014-02-07 Jeff Law <law@redhat.com>
3303 PR target/40977
3304 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3305 define_insn_and_split.
3307 * ipa-inline.c (inline_small_functions): Fix typos.
3309 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3311 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3312 (s390_can_use_return_insn): Declare.
3313 * config/s390/s390.h (EPILOGUE_USES): Define.
3314 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3315 instructions.
3316 (s390_chunkify_start): Handle return JUMP_LABELs.
3317 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3318 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3319 (s390_can_use_return_insn): New functions.
3320 (s390_fix_long_loop_prediction): Handle conditional returns.
3321 (TARGET_SET_UP_BY_PROLOGUE): Define.
3322 * config/s390/s390.md (ANY_RETURN): New code iterator.
3323 (*creturn, *csimple_return, return, simple_return): New patterns.
3325 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3327 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3328 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3329 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3330 REG_CFA_RESTORE list when deciding not to restore a register.
3332 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3334 * config/s390/s390.c: Include tree-pass.h and context.h.
3335 (s390_early_mach): New function, split out from...
3336 (s390_emit_prologue): ...here.
3337 (pass_data_s390_early_mach): New pass structure.
3338 (pass_s390_early_mach): New class.
3339 (s390_option_override): Create and register early_mach pass.
3340 Move to end of file.
3342 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3344 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3345 to match for the exit block.
3347 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3349 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3350 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3351 Reject misaligned operands.
3353 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3355 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3357 2014-02-07 Richard Biener <rguenther@suse.de>
3359 PR middle-end/60092
3360 * gimple-low.c (lower_builtin_posix_memalign): New function.
3361 (lower_stmt): Call it to lower posix_memalign in a way
3362 to make alignment info accessible.
3364 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3366 PR c++/60082
3367 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3368 __builtin_setjmp_receiver.
3370 2014-02-07 Richard Biener <rguenther@suse.de>
3372 PR middle-end/60092
3373 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3374 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3375 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3376 Handle BUILT_IN_POSIX_MEMALIGN.
3377 (find_func_clobbers): Likewise.
3378 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3379 (call_may_clobber_ref_p_1): Likewise.
3381 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3383 PR ipa/59918
3384 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3385 sanity check.
3387 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3389 PR ipa/59469
3390 * lto-cgraph.c (lto_output_node): Use
3391 symtab_get_symbol_partitioning_class.
3392 (lto_output_varpool_node): likewise.
3393 (symtab_get_symbol_partitioning_class): Move here from
3394 lto/lto-partition.c
3395 * cgraph.h (symbol_partitioning_class): Likewise.
3396 (symtab_get_symbol_partitioning_class): Declare.
3398 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3400 * ggc.h (ggc_internal_cleared_alloc): New macro.
3401 * vec.h (vec_safe_copy): Handle memory stats.
3402 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3403 * target-globals.c (save_target_globals): Likewise.
3405 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3407 PR target/60077
3408 * expr.c (emit_move_resolve_push): Export; be bit more selective
3409 on when to clear alias set.
3410 * expr.h (emit_move_resolve_push): Declare.
3411 * function.h (struct function): Add tail_call_marked.
3412 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3413 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3414 * config/i386/i386.md (TImode move expander): De not call
3415 ix86_expand_push.
3416 (FP push expanders): Preserve memory attributes.
3417 * config/i386/sse.md (push<mode>1): Remove.
3418 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3419 (ix86_expand_push): Remove.
3420 * config/i386/mmx.md (push<mode>1): Remove.
3422 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3424 PR rtl-optimization/60030
3425 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3426 lopart with paradoxical subreg before shifting it up by hprec.
3428 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3430 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3431 Remove extra newline at end of file.
3432 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3433 (arm_issue_rate): Handle cortexa57.
3434 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3435 (cortex-a57.cortex-a53): Likewise.
3437 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3439 PR target/59575
3440 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3441 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3442 bitmask.
3443 (arm_expand_prologue): Adjust all callers.
3444 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3445 info, registers also at the lowest numbered registers side. Use
3446 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3447 XEXP.
3449 PR debug/59992
3450 * var-tracking.c (adjust_mems): Before adding a SET to
3451 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3453 2014-02-06 Alan Modra <amodra@gmail.com>
3455 PR target/60032
3456 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3457 change SDmode to DDmode when lra_in_progress.
3459 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3461 PR middle-end/59150
3462 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3463 free_data_ref on the dr first, and before goto again also set dr
3464 to the next dr. For simd_lane_access, free old datarefs[i] before
3465 overwriting it. For get_vectype_for_scalar_type failure, don't
3466 free_data_ref if simd_lane_access.
3468 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3470 PR target/60062
3471 * tree.h (opts_for_fn): New inline function.
3472 (opt_for_fn): Define.
3473 * config/i386/i386.c (ix86_function_regparm): Use
3474 opt_for_fn (decl, optimize) instead of optimize.
3476 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3478 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3479 for SYMBOL_REF in large memory model.
3481 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3483 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3484 and crypto support.
3485 (cortex-a57): Likewise.
3486 (cortex-a57.cortex-a53): Likewise.
3488 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3489 Kugan Vivekanandarajah <kuganv@linaro.org>
3491 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3492 unaligned_access.
3493 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3495 2014-02-06 Richard Biener <rguenther@suse.de>
3497 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3498 set_loop_copy and initialize_original_copy_tables.
3500 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3502 * config/aarch64/aarch64-simd.md
3503 (aarch64_ashr_simddi): Change QI to SI.
3505 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3506 Jakub Jelinek <jakub@redhat.com>
3508 PR middle-end/60013
3509 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3510 of the dataflow.
3512 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3514 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3515 CODE_FOR_altivec_vpku[hw]um to
3516 CODE_FOR_altivec_vpku[hw]um_direct.
3517 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3518 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3519 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3520 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3522 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3524 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3525 generation for -maltivec=be.
3526 (altivec_vsumsws): Simplify redundant test.
3528 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3530 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3531 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3532 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3533 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3534 gen_altivec_vpkuwum.
3535 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3536 BYTES_BIG_ENDIAN.
3537 (altivec_vpks<VI_char>ss): Likewise.
3538 (altivec_vpks<VI_char>us): Likewise.
3539 (altivec_vpku<VI_char>us): Likewise.
3540 (altivec_vpku<VI_char>um): Likewise.
3541 (altivec_vpku<VI_char>um_direct): New (copy of
3542 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3543 internal use).
3544 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3545 target is little endian and -maltivec=be is not specified.
3546 (*altivec_vupkhs<VU_char>_direct): New (copy of
3547 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3548 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3549 target is little endian and -maltivec=be is not specified.
3550 (*altivec_vupkls<VU_char>_direct): New (copy of
3551 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3552 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3553 little endian and -maltivec=be is not specified.
3554 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3555 little endian and -maltivec=be is not specified.
3557 2014-02-05 Richard Henderson <rth@redhat.com>
3559 PR debug/52727
3560 * combine-stack-adj.c: Revert r206943.
3561 * sched-int.h (struct deps_desc): Add last_args_size.
3562 * sched-deps.c (init_deps): Initialize it.
3563 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3564 contain REG_ARGS_SIZE notes.
3566 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3568 * lto-cgraph.c (asm_nodes_output): Make global.
3569 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3570 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3571 (driver_handle_option): Handle OPT_fwpa.
3573 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3575 PR ipa/59947
3576 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3577 a comment typo and formatting issue. If odr_hash hasn't been
3578 created, return vNULL and set *completep to false.
3580 PR middle-end/57499
3581 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3582 bb with no successors.
3584 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3586 PR target/59718
3587 * doc/invoke.texi (-march): Clarify documentation for ARM.
3588 (-mtune): Likewise.
3589 (-mcpu): Likewise.
3591 2014-02-05 Richard Biener <rguenther@suse.de>
3593 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3594 when not vectorizing because of too many alias checks.
3595 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3596 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3598 2014-02-05 Nick Clifton <nickc@redhat.com>
3600 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3601 accept extended registers in any mode when compiling for the MN10300.
3603 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3605 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3606 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3607 sanitization attributes.
3608 (can_inline_edge_p): Likewise.
3609 (sanitize_attrs_match_for_inline_p): New function.
3611 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3613 * ipa-prop.c (detect_type_change): Shor circuit testing of
3614 type changes on THIS pointer.
3616 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3618 PR target/59777
3619 * config/pa/pa.c (legitimize_tls_address): Return original address
3620 if not passed a SYMBOL_REF rtx.
3621 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3622 addresses.
3623 (pa_emit_move_sequence): Simplify TLS source operands.
3624 (pa_legitimate_constant_p): Reject all TLS constants.
3625 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3626 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3628 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3630 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3631 groups when we know they are controlled by LTO.
3632 * varasm.c (default_binds_local_p_1): If object is in other partition,
3633 it will be resolved locally.
3635 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3637 * config/host-linux.c (linux_gt_pch_use_address): Don't
3638 use SSIZE_MAX because it is not always defined.
3640 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3642 PR bootstrap/59913
3643 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3644 threshold for pseudo splitting.
3645 (update_ebb_live_info): Process call argument hard registers and
3646 hard registers from insn definition too.
3647 (max_small_class_regs_num): New constant.
3648 (inherit_in_ebb): Update live hard regs through EBBs. Update
3649 reloads_num only for small register classes. Don't split for
3650 outputs of jumps.
3652 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3654 PR ipa/60058
3655 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3656 is non-null.
3658 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3660 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3661 visibility is safe.
3663 2014-02-04 Marek Polacek <polacek@redhat.com>
3665 * gdbinit.in (pel): Define.
3667 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3669 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3670 behavior.
3672 2014-02-04 Richard Biener <rguenther@suse.de>
3674 PR lto/59723
3675 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3676 in function context local.
3677 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3678 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3679 similar to LTO_imported_decl_ref.
3681 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3683 PR tree-optimization/60002
3684 * cgraphclones.c (build_function_decl_skip_args): Clear
3685 DECL_LANG_SPECIFIC.
3687 PR tree-optimization/60023
3688 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3689 false to gsi_replace.
3690 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3691 has been in some EH region and vec_stmt could throw, add
3692 vec_stmt into the same EH region.
3693 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3694 has no lhs, ignore it.
3695 * internal-fn.c (expand_MASK_LOAD): Likewise.
3697 PR ipa/60026
3698 * tree-inline.c (copy_forbidden): Fail for
3699 __attribute__((optimize (0))) functions.
3701 PR other/58712
3702 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3703 is set, copy one less argument.
3704 (expand_simd_clones): Don't subtract clone_info->inbranch
3705 from simd_clone_struct_alloc argument.
3707 PR rtl-optimization/57915
3708 * recog.c (simplify_while_replacing): If all unary/binary/relational
3709 operation arguments are constant, attempt to simplify those.
3711 PR middle-end/59261
3712 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3713 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3715 2014-02-04 Richard Biener <rguenther@suse.de>
3717 PR tree-optimization/60012
3718 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3719 TBAA disambiguation to all DDRs.
3721 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3723 PR target/59788
3724 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3725 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3727 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3729 PR ipa/59882
3730 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3732 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3734 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3735 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3737 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3739 PR ipa/59831
3740 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3741 to figure out targets of polymorphic calls with known decl.
3742 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3743 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3744 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3745 (get_polymorphic_call_info): ... here.
3746 (get_polymorphic_call_info_from_invariant): New function.
3748 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3750 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3751 lookup via vtable pointer; check for type consistency
3752 and turn inconsitent facts into UNREACHABLE.
3753 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3754 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3755 type inconsistent querries; return UNREACHABLE instead.
3757 2014-02-03 Richard Henderson <rth@twiddle.net>
3759 PR tree-opt/59924
3760 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3761 already processed this node.
3762 (normalize_one_pred_1): Pass along mark_set.
3763 (normalize_one_pred): Create and destroy a pointer_set_t.
3764 (normalize_one_pred_chain): Likewise.
3766 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3768 PR gcov-profile/58602
3769 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3771 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3773 PR ipa/59831
3774 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3775 -fno-devirtualize; try to devirtualize by the knowledge of
3776 virtual table pointer given by aggregate propagation.
3777 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3778 (ipa_print_node_jump_functions): Dump also offset that
3779 is relevant for polymorphic calls.
3780 (determine_known_aggregate_parts): Add arg_type parameter; use it
3781 instead of determining the type from pointer type.
3782 (ipa_compute_jump_functions_for_edge): Update call of
3783 determine_known_aggregate_parts.
3784 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3785 (gimple_get_virt_method_for_binfo): ... here; simplify using
3786 vtable_pointer_value_to_vtable.
3787 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3788 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3789 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3790 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3791 POINTER_PLUS_EXPR.
3792 (vtable_pointer_value_to_binfo): ... here.
3793 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3795 2014-02-03 Teresa Johnson <tejohnson@google.com>
3797 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3798 redef of outer loop index variable.
3800 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3802 PR c++/53017
3803 PR c++/59211
3804 * doc/extend.texi (Function Attributes): Typo.
3806 2014-02-03 Cong Hou <congh@google.com>
3808 PR tree-optimization/60000
3809 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3810 if the vectorized statement is a store. A store statement can only
3811 appear at the end of pattern statements.
3813 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3815 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3816 (ix86_option_override_internal): Default long double to 64-bit for
3817 32-bit Bionic and to 128-bit for 64-bit Bionic.
3819 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3820 TARGET_LONG_DOUBLE_128 is true.
3821 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3823 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3824 (mlong-double-64): Negate -mlong-double-128.
3825 (mlong-double-128): New option.
3827 * config/i386/i386-c.c (ix86_target_macros): Define
3828 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3830 * doc/invoke.texi: Document -mlong-double-128.
3832 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3834 PR rtl-optimization/60024
3835 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3837 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3839 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3841 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3843 PR rtl-optimization/57662
3844 * sel-sched.c (code_motion_path_driver): Do not mark already not
3845 existing blocks in the visiting bitmap.
3847 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3849 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3850 on the insn being emitted.
3852 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3853 Will Deacon <will.deacon@arm.com>
3855 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3857 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3859 * config/arm/arm-tables.opt: Regenerate.
3861 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3863 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3864 for vector types other than V16QImode.
3865 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3866 define_expand, and call altivec_expand_vec_perm_le when producing
3867 code with little endian element order.
3868 (*altivec_vperm_<mode>_internal): New insn having previous
3869 behavior of altivec_vperm_<mode>.
3870 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3871 altivec_expand_vec_perm_le when producing code with little endian
3872 element order.
3873 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3874 behavior of altivec_vperm_<mode>_uns.
3876 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3878 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3879 (altivec_vsumsws): Add handling for -maltivec=be with a little
3880 endian target.
3881 (altivec_vsumsws_direct): New.
3882 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3883 gen_altivec_vsumsws.
3885 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3887 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3888 vtable_pointer_value_to_binfo): New functions.
3889 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3890 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3892 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3894 * config/nios2/nios2.md (load_got_register): Initialize GOT
3895 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3896 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3898 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3900 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3901 preserverd by passthrough, do not propagate the type.
3903 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3905 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3906 (mips_atomic_assign_expand_fenv): New function.
3907 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3909 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3911 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3912 (__builtin_mips_set_fcsr): Likewise.
3913 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3914 MIPS_USI_FTYPE_VOID.
3915 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3916 (mips16_expand_set_fcsr): Likewise.
3917 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3918 (mips16_set_fcsr_stub): Likewise.
3919 (mips16_get_fcsr_one_only_stub): New class.
3920 (mips16_set_fcsr_one_only_stub): Likewise.
3921 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3922 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3923 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3924 (hard_float): New availability predicate.
3925 (mips_builtins): Add get_fcsr and set_fcsr.
3926 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3927 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3928 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3929 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3930 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3931 patterns.
3933 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3935 * config/mips/mips.c (mips_one_only_stub): New class.
3936 (mips_need_mips16_rdhwr_p): Replace with...
3937 (mips16_rdhwr_stub): ...this new variable.
3938 (mips16_stub_call_address): New function.
3939 (mips16_rdhwr_one_only_stub): New class.
3940 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3941 (mips_output_mips16_rdhwr): Delete.
3942 (mips_finish_stub): New function.
3943 (mips_code_end): Use it to handle rdhwr stubs.
3945 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3947 PR target/60017
3948 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3949 when calculating size of integer atomic types.
3951 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3953 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3955 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3957 PR tree-optimization/60003
3958 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3959 * profile.c (branch_prob): Use gimple_call_builtin_p
3960 to check for BUILT_IN_SETJMP_RECEIVER.
3961 * tree-inline.c (copy_bb): Call notice_special_calls.
3963 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3965 PR bootstrap/59985
3966 * lra-constraints.c (process_alt_operands): Update reload_sum only
3967 on the first pass.
3969 2014-01-31 Richard Henderson <rth@redhat.com>
3971 PR middle-end/60004
3972 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3973 until after else_eh is processed.
3975 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3977 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3978 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3979 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3980 in smmintrin.h, remove them.
3981 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3982 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3983 * config/i386/i386.md (ROUND_SAE): Fix value.
3984 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3985 (const48_operand): New.
3986 * config/i386/subst.md (round), (round_expand): Use
3987 const_4_or_8_to_11_operand.
3988 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3990 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3992 * config/i386/constraints.md (Yk): Swap meaning with k.
3993 * config/i386/i386.md (movhi_internal): Change Yk to k.
3994 (movqi_internal): Ditto.
3995 (*k<logic><mode>): Ditto.
3996 (*andhi_1): Ditto.
3997 (*andqi_1): Ditto.
3998 (kandn<mode>): Ditto.
3999 (*<code>hi_1): Ditto.
4000 (*<code>qi_1): Ditto.
4001 (kxnor<mode>): Ditto.
4002 (kortestzhi): Ditto.
4003 (kortestchi): Ditto.
4004 (kunpckhi): Ditto.
4005 (*one_cmplhi2_1): Ditto.
4006 (*one_cmplqi2_1): Ditto.
4007 * config/i386/sse.md (): Change k to Yk.
4008 (avx512f_load<mode>_mask): Ditto.
4009 (avx512f_blendm<mode>): Ditto.
4010 (avx512f_store<mode>_mask): Ditto.
4011 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4012 (avx512f_storedqu<mode>_mask): Ditto.
4013 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4014 Ditto.
4015 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4016 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4017 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4018 (avx512f_maskcmp<mode>3): Ditto.
4019 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4020 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4021 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4022 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4023 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4024 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4025 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4026 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4027 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4028 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4029 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4030 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4031 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4032 (vec_extract_lo_<mode>_maskm): Ditto.
4033 (vec_extract_hi_<mode>_maskm): Ditto.
4034 (avx512f_vternlog<mode>_mask): Ditto.
4035 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4036 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4037 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4038 (avx512f_<code>v8div16qi2_mask): Ditto.
4039 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4040 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4041 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4042 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4043 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4044 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4045 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4046 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4047 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4048 (avx512cd_maskb_vec_dupv8di): Ditto.
4049 (avx512cd_maskw_vec_dupv16si): Ditto.
4050 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4051 (avx512f_vpermi2var<mode>3_mask): Ditto.
4052 (avx512f_vpermi2var<mode>3_mask): Ditto.
4053 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4054 (*avx512f_gathersi<mode>): Ditto.
4055 (*avx512f_gathersi<mode>_2): Ditto.
4056 (*avx512f_gatherdi<mode>): Ditto.
4057 (*avx512f_gatherdi<mode>_2): Ditto.
4058 (*avx512f_scattersi<mode>): Ditto.
4059 (*avx512f_scatterdi<mode>): Ditto.
4060 (avx512f_compress<mode>_mask): Ditto.
4061 (avx512f_compressstore<mode>_mask): Ditto.
4062 (avx512f_expand<mode>_mask): Ditto.
4063 * config/i386/subst.md (mask): Change k to Yk.
4064 (mask_scalar_merge): Ditto.
4065 (sd): Ditto.
4067 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4069 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4071 2014-01-31 Richard Biener <rguenther@suse.de>
4073 PR middle-end/59990
4074 * builtins.c (fold_builtin_memory_op): Make sure to not
4075 use a floating-point mode or a boolean or enumeral type for
4076 the copy operation.
4078 2014-01-30 DJ Delorie <dj@redhat.com>
4080 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4081 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4082 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4083 whenever main() has an epilogue.
4085 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4087 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4088 unused variable "field".
4089 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4090 (vsx_mergeh_<mode>): Likewise.
4091 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4092 (altivec_vmrghh): Likewise.
4093 (altivec_vmrghw): Likewise.
4094 (altivec_vmrglb): Likewise.
4095 (altivec_vmrglh): Likewise.
4096 (altivec_vmrglw): Likewise.
4097 (altivec_vspltb): Add missing uses.
4098 (altivec_vsplth): Likewise.
4099 (altivec_vspltw): Likewise.
4100 (altivec_vspltsf): Likewise.
4102 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4104 PR target/59923
4105 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4106 frame related instructions.
4108 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4110 PR rtl-optimization/59959
4111 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4112 any reload of register whose subreg is invalid.
4114 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4116 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4117 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4118 Add missing return type - void.
4120 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4122 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4123 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4124 remove element index adjustment for endian (now handled in vsx.md
4125 and altivec.md).
4126 (altivec_expand_vec_perm_const): Use
4127 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4128 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4129 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4130 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4131 define_expand and a new define_insn *altivec_vspltb_internal;
4132 adjust for -maltivec=be on a little endian target.
4133 (altivec_vspltb_direct): New.
4134 (altivec_vsplth): Divide into a define_expand and a new
4135 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4136 little endian target.
4137 (altivec_vsplth_direct): New.
4138 (altivec_vspltw): Divide into a define_expand and a new
4139 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4140 little endian target.
4141 (altivec_vspltw_direct): New.
4142 (altivec_vspltsf): Divide into a define_expand and a new
4143 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4144 a little endian target.
4146 2014-01-30 Richard Biener <rguenther@suse.de>
4148 PR tree-optimization/59993
4149 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4150 can propagate form the earlier stmt and avoid the transform
4151 when the intermediate result is needed.
4153 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4155 * README.Portability: Fix typo.
4157 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4159 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4160 comparison_operator with ordered_comparison_operator.
4162 2014-01-30 Nick Clifton <nickc@redhat.com>
4164 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4165 Rename to mn10300_store_multiple_regs.
4166 * config/mn10300/mn10300.c: Likewise.
4167 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4168 store_multiple_regs.
4169 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4170 Call mn10300_store_multiple_regs.
4172 2014-01-30 Nick Clifton <nickc@redhat.com>
4173 DJ Delorie <dj@redhat.com>
4175 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4176 %fp 2 to keep registers after it properly word-aligned.
4177 (rl78_alloc_physical_registers_umul): Handle the case where both
4178 input operands are the same.
4180 2014-01-30 Richard Biener <rguenther@suse.de>
4182 PR tree-optimization/59903
4183 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4184 check properly.
4186 2014-01-30 Jason Merrill <jason@redhat.com>
4188 PR c++/59633
4189 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4191 PR c++/59645
4192 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4194 2014-01-30 Richard Biener <rguenther@suse.de>
4196 PR tree-optimization/59951
4197 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4199 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4201 PR target/59784
4202 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4203 SFmode to DFmode case.
4205 2014-01-29 DJ Delorie <dj@redhat.com>
4207 * config/msp430/msp430.opt (-minrt): New.
4208 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4209 if -minrt given.
4210 (ENDFILE_SPEC): Likewise.
4212 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4214 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4215 (estimate_function_body_sizes): Use it.
4217 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4219 PR c++/58561
4220 * dwarf2out.c (is_cxx_auto): New.
4221 (is_base_type): Use it.
4222 (gen_type_die_with_usage): Likewise.
4224 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4226 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4227 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4228 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4229 -maltivec=be with LE targets.
4230 (vsx_mergeh_<mode>): Likewise.
4231 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4232 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4233 (altivec_vmrghb): Replace with define_expand and new
4234 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4235 (altivec_vmrghb_direct): New define_insn.
4236 (altivec_vmrghh): Replace with define_expand and new
4237 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4238 (altivec_vmrghh_direct): New define_insn.
4239 (altivec_vmrghw): Replace with define_expand and new
4240 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4241 (altivec_vmrghw_direct): New define_insn.
4242 (*altivec_vmrghsf): Adjust for endianness.
4243 (altivec_vmrglb): Replace with define_expand and new
4244 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4245 (altivec_vmrglb_direct): New define_insn.
4246 (altivec_vmrglh): Replace with define_expand and new
4247 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4248 (altivec_vmrglh_direct): New define_insn.
4249 (altivec_vmrglw): Replace with define_expand and new
4250 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4251 (altivec_vmrglw_direct): New define_insn.
4252 (*altivec_vmrglsf): Adjust for endianness.
4253 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4254 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4255 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4256 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4257 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4258 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4259 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4260 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4262 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4264 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4265 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4266 whitespace.
4268 2014-01-29 Richard Biener <rguenther@suse.de>
4270 PR tree-optimization/58742
4271 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4272 associate_pointerplus_align.
4273 (associate_pointerplus_diff): New function.
4274 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4275 and associate_pointerplus_diff.
4277 2014-01-29 Richard Biener <rguenther@suse.de>
4279 * lto-streamer.h (LTO_major_version): Bump to 3.
4280 (LTO_minor_version): Reset to 0.
4282 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4284 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4285 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4286 (arm_file_start): Generate correct asm header for armv7ve.
4287 * config/arm/bpabi.h: Add multilib support for armv7ve.
4288 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4289 and cortex-a15 to armv7ve.
4290 * config/arm/t-aprofile: Add multilib support for armv7ve.
4291 * doc/invoke.texi: Document -march=armv7ve.
4293 2014-01-29 Richard Biener <rguenther@suse.de>
4295 PR tree-optimization/58742
4296 * tree-ssa-forwprop.c (associate_plusminus): Return true
4297 if we changed sth, defer EH cleanup to ...
4298 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4299 (simplify_mult): New function.
4301 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4303 PR middle-end/59917
4304 PR tree-optimization/59920
4305 * tree.c (build_common_builtin_nodes): Remove
4306 __builtin_setjmp_dispatcher initialization.
4307 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4308 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4309 instead of gsi_after_labels + manually skipping debug stmts.
4310 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4311 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4312 * tree-inline.c (copy_edges_for_bb): Remove
4313 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4314 argument. Ignore computed_goto_p stmts. Don't call
4315 make_abnormal_goto_edges. If a call might need abnormal edges
4316 for non-local gotos, see if it already has an edge to
4317 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4318 with true argument, don't do anything then, otherwise add
4319 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4320 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4321 caller.
4322 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4323 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4324 (lower_stmt): Don't set data->calls_builtin_setjmp.
4325 (lower_builtin_setjmp): Adjust comment.
4326 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4327 * tree-cfg.c (found_computed_goto): Remove.
4328 (factor_computed_gotos): Remove.
4329 (make_goto_expr_edges): Return bool, true for computed gotos.
4330 Don't call make_abnormal_goto_edges.
4331 (build_gimple_cfg): Don't set found_computed_goto, don't call
4332 factor_computed_gotos.
4333 (computed_goto_p): No longer static.
4334 (make_blocks): Don't set found_computed_goto.
4335 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4336 (make_edges): If make_goto_expr_edges returns true, push bb
4337 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4338 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4339 vector. Record mapping between bbs and OpenMP regions if there
4340 are any, adjust make_gimple_omp_edges caller. Call
4341 handle_abnormal_edges.
4342 (make_abnormal_goto_edges): Remove.
4343 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4344 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4345 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4346 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4347 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4348 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4349 filling *region also set *region_idx to (*region)->entry->index.
4351 PR other/58712
4352 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4353 For REGs set ORIGINAL_REGNO.
4355 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4357 * doc/md.texi: Mention that a target shouldn't implement
4358 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4359 than hi/lo pair.
4361 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4363 PR tree-optimization/59594
4364 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4365 a copy of the datarefs vector rather than the vector itself.
4367 2014-01-28 Jason Merrill <jason@redhat.com>
4369 PR c++/53756
4370 * dwarf2out.c (auto_die): New static.
4371 (gen_type_die_with_usage): Handle C++1y 'auto'.
4372 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4373 on definition.
4375 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4377 PR target/59672
4378 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4379 (SPEC_X32): Likewise.
4380 (SPEC_64): Likewise.
4381 * config/i386/i386.c (ix86_option_override_internal): Turn off
4382 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4383 for TARGET_16BIT.
4384 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4385 * config/i386/i386.h (TARGET_16BIT): New macro.
4386 (TARGET_16BIT_P): Likewise.
4387 * config/i386/i386.opt: Add m16.
4388 * doc/invoke.texi: Document -m16.
4390 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4392 PR preprocessor/59935
4393 * input.c (location_get_source_line): Bail out on when line number
4394 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4396 2014-01-28 Richard Biener <rguenther@suse.de>
4398 PR tree-optimization/58742
4399 * tree-ssa-forwprop.c (associate_plusminus): Handle
4400 pointer subtraction of the form (T)(P + A) - (T)P.
4402 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4404 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4405 at const_int_cost.
4407 2014-01-28 Richard Biener <rguenther@suse.de>
4409 Revert
4410 2014-01-28 Richard Biener <rguenther@suse.de>
4412 PR rtl-optimization/45364
4413 PR rtl-optimization/59890
4414 * var-tracking.c (local_get_addr_clear_given_value): Handle
4415 already cleared slot.
4416 (val_reset): Handle not allocated local_get_addr_cache.
4417 (vt_find_locations): Use post-order on the inverted CFG.
4419 2014-01-28 Richard Biener <rguenther@suse.de>
4421 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4423 2014-01-28 Richard Biener <rguenther@suse.de>
4425 PR rtl-optimization/45364
4426 PR rtl-optimization/59890
4427 * var-tracking.c (local_get_addr_clear_given_value): Handle
4428 already cleared slot.
4429 (val_reset): Handle not allocated local_get_addr_cache.
4430 (vt_find_locations): Use post-order on the inverted CFG.
4432 2014-01-28 Alan Modra <amodra@gmail.com>
4434 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4435 * configure.ac <recursive call for build != host>: Define
4436 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4437 and LD_FOR_BUILD too.
4438 * configure: Regenerate.
4440 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4442 * config/i386/i386.c (get_builtin_code_for_version): Separate
4443 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4444 Broadwell from Haswell.
4446 2014-01-27 Steve Ellcey <sellcey@mips.com>
4448 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4449 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4450 * config/mips/mips.c (mips_option_override): Change setting
4451 of TARGET_DSP.
4452 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4453 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4454 Change from Mask to Var.
4456 2014-01-27 Jeff Law <law@redhat.com>
4458 * ipa-inline.c (inline_small_functions): Fix typo.
4460 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4462 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4463 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4464 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4465 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4466 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4467 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4468 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4469 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4470 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4471 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4472 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4473 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4474 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4475 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4476 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4477 (_mm512_storeu_epi64): Ditto.
4478 (_mm512_cmpge_epi32_mask): Ditto.
4479 (_mm512_cmpge_epu32_mask): Ditto.
4480 (_mm512_cmpge_epi64_mask): Ditto.
4481 (_mm512_cmpge_epu64_mask): Ditto.
4482 (_mm512_cmple_epi32_mask): Ditto.
4483 (_mm512_cmple_epu32_mask): Ditto.
4484 (_mm512_cmple_epi64_mask): Ditto.
4485 (_mm512_cmple_epu64_mask): Ditto.
4486 (_mm512_cmplt_epi32_mask): Ditto.
4487 (_mm512_cmplt_epu32_mask): Ditto.
4488 (_mm512_cmplt_epi64_mask): Ditto.
4489 (_mm512_cmplt_epu64_mask): Ditto.
4490 (_mm512_cmpneq_epi32_mask): Ditto.
4491 (_mm512_cmpneq_epu32_mask): Ditto.
4492 (_mm512_cmpneq_epi64_mask): Ditto.
4493 (_mm512_cmpneq_epu64_mask): Ditto.
4494 (_mm512_expand_pd): Ditto.
4495 (_mm512_expand_ps): Ditto.
4496 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4497 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4498 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4499 * config/i386/i386.c (ix86_builtins): Add
4500 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4501 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4502 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4503 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4504 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4505 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4506 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4507 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4508 IX86_BUILTIN_PMOVUSQW512_MEM.
4509 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4510 __builtin_ia32_pmovsqd512mem_mask,
4511 __builtin_ia32_pmovqd512mem_mask,
4512 __builtin_ia32_pmovusqw512mem_mask,
4513 __builtin_ia32_pmovsqw512mem_mask,
4514 __builtin_ia32_pmovqw512mem_mask,
4515 __builtin_ia32_pmovusdw512mem_mask,
4516 __builtin_ia32_pmovsdw512mem_mask,
4517 __builtin_ia32_pmovdw512mem_mask,
4518 __builtin_ia32_pmovqb512mem_mask,
4519 __builtin_ia32_pmovusqb512mem_mask,
4520 __builtin_ia32_pmovsqb512mem_mask,
4521 __builtin_ia32_pmovusdb512mem_mask,
4522 __builtin_ia32_pmovsdb512mem_mask,
4523 __builtin_ia32_pmovdb512mem_mask.
4524 (bdesc_args): Add __builtin_ia32_expanddf512,
4525 __builtin_ia32_expandsf512.
4526 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4527 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4528 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4529 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4530 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4531 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4532 (avx512f_<code>v8div16qi2_mask_store): This.
4533 (avx512f_expand<mode>): New.
4535 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4537 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4538 New.
4539 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4540 (_mm512_prefetch_i32scatter_pd): Ditto.
4541 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4542 (_mm512_prefetch_i64scatter_pd): Ditto.
4543 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4544 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4545 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4546 (_mm512_prefetch_i32scatter_ps): Ditto.
4547 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4548 (_mm512_prefetch_i64scatter_ps): Ditto.
4549 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4550 * config/i386/i386-builtin-types.def: Define
4551 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4552 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4553 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4554 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4555 IX86_BUILTIN_SCATTERPFQPD.
4556 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4557 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4558 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4559 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4560 __builtin_ia32_scatterpfqps.
4561 (ix86_expand_builtin): Expand new built-ins.
4562 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4563 fix memory access data type.
4564 (*avx512pf_gatherpf<mode>_mask): Ditto.
4565 (*avx512pf_gatherpf<mode>): Ditto.
4566 (avx512pf_scatterpf<mode>): Ditto.
4567 (*avx512pf_scatterpf<mode>_mask): Ditto.
4568 (*avx512pf_scatterpf<mode>): Ditto.
4569 (GATHER_SCATTER_SF_MEM_MODE): New.
4570 (avx512pf_gatherpf<mode>df): Ditto.
4571 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4572 (*avx512pf_scatterpf<mode>df): Ditto.
4574 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4576 PR bootstrap/59934
4577 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4578 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4579 reached.
4581 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4583 * common/config/arm/arm-common.c
4584 (arm_rewrite_mcpu): Handle multiple names.
4585 * config/arm/arm.h
4586 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4588 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4590 * gimple-builder.h (create_gimple_tmp): Delete.
4592 2014-01-27 Christian Bruel <christian.bruel@st.com>
4594 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4595 words comparisons.
4597 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4599 * config/pa/pa.md (call): Generate indirect long calls to non-local
4600 functions when outputing 32-bit code.
4601 (call_value): Likewise except for special call to buggy powf function.
4603 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4604 portable runtime and PIC indirect calls.
4605 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4606 and PIC call sequences. Use ldo instead of blr to set return register
4607 in PIC call sequence.
4609 2014-01-25 Walter Lee <walt@tilera.com>
4611 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4612 avoid clobbering a live register.
4614 2014-01-25 Walter Lee <walt@tilera.com>
4616 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4617 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4618 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4619 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4621 2014-01-25 Walter Lee <walt@tilera.com>
4623 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4624 arguments on even registers.
4625 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4626 STACK_BOUNDARY.
4627 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4628 (BIGGEST_ALIGNMENT): Ditto.
4629 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4631 2014-01-25 Walter Lee <walt@tilera.com>
4633 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4634 insns before bundling.
4635 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4637 2014-01-25 Walter Lee <walt@tilera.com>
4639 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4640 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4641 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4643 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4645 * config/mips/constraints.md (kl): Delete.
4646 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4647 define expands, using...
4648 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4649 instructions for MIPS16.
4650 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4651 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4653 2014-01-25 Walter Lee <walt@tilera.com>
4655 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4656 (clzdi2): Ditto.
4657 (ffsdi2): Ditto.
4659 2014-01-25 Walter Lee <walt@tilera.com>
4661 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4662 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4664 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4666 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4667 Handle XOR.
4669 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4671 * print-rtl.c (in_call_function_usage): New var.
4672 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4673 EXPR_LIST mode as mode and not as reg note name.
4675 PR middle-end/59561
4676 * cfgloopmanip.c (copy_loop_info): If
4677 loop->warned_aggressive_loop_optimizations, make sure
4678 the flag is set in target loop too.
4680 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4682 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4683 flag_cilkplus.
4684 * builtins.def: Likewise.
4685 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4686 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4687 * ira.c (ira_setup_eliminable_regset): Likewise.
4688 * omp-low.c (gate_expand_omp): Likewise.
4689 (execute_lower_omp): Likewise.
4690 (diagnose_sb_0): Likewise.
4691 (gate_diagnose_omp_blocks): Likewise.
4692 (simd_clone_clauses_extract): Likewise.
4693 (gate): Likewise.
4695 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4697 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4698 correction for little endian...
4699 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4700 here.
4702 2014-01-24 Jeff Law <law@redhat.com>
4704 PR tree-optimization/59919
4705 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4706 for non-returning calls.
4708 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4710 * common/config/aarch64/aarch64-common.c
4711 (aarch64_rewrite_mcpu): Handle multiple names.
4712 * config/aarch64/aarch64.h
4713 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4715 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4717 * input.c (add_file_to_cache_tab): Handle the case where fopen
4718 returns NULL.
4720 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4722 PR target/59929
4723 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4724 from push operand if code of push isn't PRE_DEC.
4726 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4728 PR target/59909
4729 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4730 -mquad-memory-atomic. Update -mquad-memory documentation to say
4731 it is only used for non-atomic loads/stores.
4733 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4734 -mquad-memory or -mquad-memory-atomic switches.
4736 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4737 -mquad-memory-atomic to ISA 2.07 support.
4739 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4740 to separate support of normal quad word memory operations (ldq, stq)
4741 from the atomic quad word memory operations.
4743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4744 support to separate non-atomic quad word operations from atomic
4745 quad word operations. Disable non-atomic quad word operations in
4746 little endian mode so that we don't have to swap words after the
4747 load and before the store.
4748 (quad_load_store_p): Add comment about atomic quad word support.
4749 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4750 options printed with -mdebug=reg.
4752 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4753 -mquad-memory-atomic as the test for whether we have quad word
4754 atomic instructions.
4755 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4756 or -mp8-vector are used, allow byte/half-word atomic operations.
4758 * config/rs6000/sync.md (load_lockedti): Insure that the address
4759 is a proper indexed or indirect address for the lqarx instruction.
4760 On little endian systems, swap the hi/lo registers after the lqarx
4761 instruction.
4762 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4763 insure the address is valid for the lqarx instruction.
4764 (store_conditionalti): Insure that the address is a proper indexed
4765 or indirect address for the stqcrx. instruction. On little endian
4766 systems, swap the hi/lo registers before doing the stqcrx.
4767 instruction.
4768 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4769 insure the address is valid for the stqcrx. instruction.
4771 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4772 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4773 type of quad memory support is available.
4775 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4777 PR regression/59915
4778 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4779 there is a danger of looping.
4781 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4783 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4784 force flag_ira_loop_pressure if set via command line.
4786 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4788 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4789 (ashr_simd): New builtin handling DI mode.
4790 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4791 (aarch64_sshr_simddi): New match pattern.
4792 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4793 (vshrd_n_s64): Likewise.
4794 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4796 2014-01-23 Nick Clifton <nickc@redhat.com>
4798 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4799 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4800 favour of mcu specific scripts.
4801 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4802 430x multilibs.
4804 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4805 Alex Velenko <Alex.Velenko@arm.com>
4807 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4808 (vaddv_s16): Likewise.
4809 (vaddv_s32): Likewise.
4810 (vaddv_u8): Likewise.
4811 (vaddv_u16): Likewise.
4812 (vaddv_u32): Likewise.
4813 (vaddvq_s8): Likewise.
4814 (vaddvq_s16): Likewise.
4815 (vaddvq_s32): Likewise.
4816 (vaddvq_s64): Likewise.
4817 (vaddvq_u8): Likewise.
4818 (vaddvq_u16): Likewise.
4819 (vaddvq_u32): Likewise.
4820 (vaddvq_u64): Likewise.
4821 (vaddv_f32): Likewise.
4822 (vaddvq_f32): Likewise.
4823 (vaddvq_f64): Likewise.
4824 (vmaxv_f32): Likewise.
4825 (vmaxv_s8): Likewise.
4826 (vmaxv_s16): Likewise.
4827 (vmaxv_s32): Likewise.
4828 (vmaxv_u8): Likewise.
4829 (vmaxv_u16): Likewise.
4830 (vmaxv_u32): Likewise.
4831 (vmaxvq_f32): Likewise.
4832 (vmaxvq_f64): Likewise.
4833 (vmaxvq_s8): Likewise.
4834 (vmaxvq_s16): Likewise.
4835 (vmaxvq_s32): Likewise.
4836 (vmaxvq_u8): Likewise.
4837 (vmaxvq_u16): Likewise.
4838 (vmaxvq_u32): Likewise.
4839 (vmaxnmv_f32): Likewise.
4840 (vmaxnmvq_f32): Likewise.
4841 (vmaxnmvq_f64): Likewise.
4842 (vminv_f32): Likewise.
4843 (vminv_s8): Likewise.
4844 (vminv_s16): Likewise.
4845 (vminv_s32): Likewise.
4846 (vminv_u8): Likewise.
4847 (vminv_u16): Likewise.
4848 (vminv_u32): Likewise.
4849 (vminvq_f32): Likewise.
4850 (vminvq_f64): Likewise.
4851 (vminvq_s8): Likewise.
4852 (vminvq_s16): Likewise.
4853 (vminvq_s32): Likewise.
4854 (vminvq_u8): Likewise.
4855 (vminvq_u16): Likewise.
4856 (vminvq_u32): Likewise.
4857 (vminnmv_f32): Likewise.
4858 (vminnmvq_f32): Likewise.
4859 (vminnmvq_f64): Likewise.
4861 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4863 * config/aarch64/aarch64-simd.md
4864 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4865 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4866 (*aarch64_mul3_elt<mode>): Likewise.
4867 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4868 (*aarch64_mul3_elt_to_64v2df): Likewise.
4869 (*aarch64_mla_elt<mode>): Likewise.
4870 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4871 (*aarch64_mls_elt<mode>): Likewise.
4872 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4873 (*aarch64_fma4_elt<mode>): Likewise.
4874 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4875 (*aarch64_fma4_elt_to_64v2df): Likewise.
4876 (*aarch64_fnma4_elt<mode>): Likewise.
4877 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4878 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4879 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4880 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4881 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4882 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4883 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4884 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4885 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4887 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4889 * config/aarch64/aarch64-simd.md
4890 (aarch64_be_checked_get_lane<mode>): New define_expand.
4891 * config/aarch64/aarch64-simd-builtins.def
4892 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4893 New builtin definition.
4894 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4895 Use new safe be builtin.
4897 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4899 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4900 New define_insn.
4901 (aarch64_be_st1<mode>): Likewise.
4902 (aarch_ld1<VALL:mode>): Define_expand modified.
4903 (aarch_st1<VALL:mode>): Likewise.
4904 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4905 (UNSPEC_ST1): Likewise.
4907 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4909 * config/microblaze/microblaze.md: Add trap insn and attribute
4911 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4913 PR preprocessor/58580
4914 * input.h (location_get_source_line): Take an additional line_size
4915 parameter.
4916 (void diagnostics_file_cache_fini): Declare new function.
4917 * input.c (struct fcache): New type.
4918 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4919 New static constants.
4920 (diagnostic_file_cache_init, total_lines_num)
4921 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4922 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4923 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4924 (get_next_line, read_next_line, goto_next_line, read_line_num):
4925 New static function definitions.
4926 (diagnostic_file_cache_fini): New function.
4927 (location_get_source_line): Take an additional output line_len
4928 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4929 read_line_num.
4930 * diagnostic.c (diagnostic_finish): Call
4931 diagnostic_file_cache_fini.
4932 (adjust_line): Take an additional input parameter for the length
4933 of the line, rather than calculating it with strlen.
4934 (diagnostic_show_locus): Adjust the use of
4935 location_get_source_line and adjust_line with respect to their new
4936 signature. While displaying a line now, do not stop at the first
4937 null byte. Rather, display the zero byte as a space and keep
4938 going until we reach the size of the line.
4939 * Makefile.in: Add vec.o to OBJS-libcommon
4941 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4942 Ilya Tocar <ilya.tocar@intel.com>
4944 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4945 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4946 (__builtin_ia32_kmov16): Ditto.
4947 * config/i386/i386.md (UNSPEC_KMOV): New.
4948 (kmovw): Ditto.
4950 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4952 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4953 (_mm512_storeu_si512): Ditto.
4955 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4957 PR target/52125
4958 * rtl.h (get_referenced_operands): Declare.
4959 * recog.c (get_referenced_operands): New function.
4960 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4961 operands have been referenced when recording LO_SUM references.
4963 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4965 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4967 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4969 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4970 Enable for generic and recent AMD targets.
4972 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4974 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4975 ARG_SIZE note when adjustment was eliminated.
4977 2014-01-22 Jeff Law <law@redhat.com>
4979 PR tree-optimization/59597
4980 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4981 in file. Accept new argument REGISTERING and use it to modify
4982 dump output appropriately.
4983 (register_jump_thread): Corresponding changes.
4984 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4985 thread paths involving joiner blocks. Add code to dump cancelled
4986 jump threading paths.
4988 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4990 PR rtl-optimization/59477
4991 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4992 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4994 2014-01-22 Tom Tromey <tromey@redhat.com>
4996 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4997 PARAMS.
4998 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5000 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5002 PR rtl-optimization/59896
5003 * lra-constraints.c (process_alt_operands): Check unused note for
5004 matched operands of insn with no output reloads.
5006 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5008 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5009 (mips_move_from_gpr_cost): Likewise.
5011 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5013 PR rtl-optimization/59858
5014 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5015 ira_class_hard_regs_num.
5016 (process_alt_operands): Increase reject for dying matched operand.
5018 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5020 PR target/59003
5021 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5022 smaller than size, perform several stores or loads and stores
5023 at dst + count - size to store or copy all of size bytes, rather
5024 than just last modesize bytes.
5026 2014-01-20 DJ Delorie <dj@redhat.com>
5028 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5029 that CLOBBERs are REGs before propogating their values.
5031 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5033 PR middle-end/59789
5034 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5035 (cgraph_inline_failed_type): New function.
5036 * cgraph.h (DEFCIFCODE): Add type.
5037 (cgraph_inline_failed_type_t): New enum.
5038 (cgraph_inline_failed_type): New prototype.
5039 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5040 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5041 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5042 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5043 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5044 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5045 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5046 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5047 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5048 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5049 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5050 OPTIMIZATION_MISMATCH.
5051 * tree-inline.c (expand_call_inline): Emit errors during
5052 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5054 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5056 PR target/59685
5057 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5058 mode attribute in insn output.
5060 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5062 * output.h (output_constant): Delete.
5063 * varasm.c (output_constant): Make private.
5065 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5067 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5069 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5071 PR middle-end/59860
5072 * tree.h (fold_builtin_strcat): New prototype.
5073 * builtins.c (fold_builtin_strcat): No longer static. Add len
5074 argument, if non-NULL, don't call c_strlen. Optimize
5075 directly into __builtin_memcpy instead of __builtin_strcpy.
5076 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5077 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5079 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5081 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5082 for SImode_address_operand operands, having only a REG argument.
5084 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5086 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5087 loader name using mbig-endian.
5088 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5090 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5092 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5093 (-mtune): Likewise.
5094 (-mcpu): Likewise.
5096 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5098 * config/aarch64/aarch64-protos.h
5099 (aarch64_cannot_change_mode_class_ptr): Declare.
5100 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5101 aarch64_cannot_change_mode_class_ptr): New.
5102 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5103 backend hook aarch64_cannot_change_mode_class.
5105 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5107 * common/config/aarch64/aarch64-common.c
5108 (aarch64_handle_option): Don't handle any option order logic here.
5109 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5110 selected_cpu, warn on architecture version mismatch.
5111 (aarch64_override_options): Fix parsing order for option strings.
5113 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5114 Iain Sandoe <iain@codesourcery.com>
5116 PR bootstrap/59496
5117 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5118 warning. Amend comment to reflect current functionality.
5120 2014-01-20 Richard Biener <rguenther@suse.de>
5122 PR middle-end/59860
5123 * builtins.c (fold_builtin_strcat): Remove case better handled
5124 by tree-ssa-strlen.c.
5126 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5128 * config/aarch64/aarch64.opt
5129 (mcpu, march, mtune): Make case-insensitive.
5131 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5133 PR target/59880
5134 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5135 if operands[1] is a REG or ZERO_EXTEND of a REG.
5137 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5139 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5141 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5143 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5144 long non-pic millicode calls.
5146 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5148 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5150 2014-01-19 Kito Cheng <kito@0xlab.org>
5152 * builtins.c (expand_movstr): Check movstr expand done or fail.
5154 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5155 H.J. Lu <hongjiu.lu@intel.com>
5157 PR target/59379
5158 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5159 to DImode for zero-extended addresses.
5161 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5163 PR rtl-optimization/57763
5164 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5165 on the new indirect jump_insn and increment LABEL_NUSES (label).
5167 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5169 PR bootstrap/59580
5170 PR bootstrap/59583
5171 * config.gcc (x86_archs): New variable.
5172 (x86_64_archs): Likewise.
5173 (x86_cpus): Likewise.
5174 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5175 --with-arch/--with-cpu= options.
5176 Support --with-arch=/--with-cpu={nehalem,westmere,
5177 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5179 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5181 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5182 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5184 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5186 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5188 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5190 PR target/58944
5191 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5192 clear cpp_get_options (parse_in)->warn_unused_macros for
5193 ix86_target_macros_internal with cpp_define.
5195 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5197 * jump.c (delete_related_insns): Keep (use (insn))s.
5198 * reorg.c (redundant_insn): Check for barriers too.
5200 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5202 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5204 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5206 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5207 call to $$dyncall when TARGET_LONG_CALLS is true.
5209 2014-01-17 Jeff Law <law@redhat.com>
5211 * ree.c (combine_set_extension): Temporarily disable test for
5212 changing number of hard registers.
5214 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5216 PR middle-end/58125
5217 * ipa-inline-analysis.c (inline_free_summary):
5218 Do not free summary of aliases.
5220 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5222 PR middle-end/59706
5223 * gimplify.c (gimplify_expr): Use create_tmp_var
5224 instead of create_tmp_var_raw. If cond doesn't have
5225 integral type, don't add the IFN_ANNOTATE builtin at all.
5227 2014-01-17 Martin Jambor <mjambor@suse.cz>
5229 PR ipa/59736
5230 * ipa-cp.c (prev_edge_clone): New variable.
5231 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5232 Also resize prev_edge_clone vector.
5233 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5234 (ipcp_edge_removal_hook): New function.
5235 (ipcp_driver): Register ipcp_edge_removal_hook.
5237 2014-01-17 Andrew Pinski <apinski@cavium.com>
5238 Steve Ellcey <sellcey@mips.com>
5240 PR target/59462
5241 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5242 of operator mode.
5244 2014-01-17 Jeff Law <law@redhat.com>
5246 PR middle-end/57904
5247 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5248 so that pass_ccp runs first.
5250 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5252 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5253 (ix86_adjust_cost): Use !TARGET_XXX.
5254 (do_reorder_for_imul): Likewise.
5255 (swap_top_of_ready_list): Likewise.
5256 (ix86_sched_reorder): Likewise.
5258 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5260 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5261 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5262 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5263 (intel_memset): New. Duplicate slm_memset.
5264 (intel_cost): New. Duplicate slm_cost.
5265 (m_INTEL): New macro.
5266 (processor_target_table): Add "intel".
5267 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5268 with PROCESSOR_INTEL for "intel".
5269 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5270 PROCESSOR_SILVERMONT.
5271 (ix86_issue_rate): Likewise.
5272 (ix86_adjust_cost): Likewise.
5273 (ia32_multipass_dfa_lookahead): Likewise.
5274 (swap_top_of_ready_list): Likewise.
5275 (ix86_sched_reorder): Likewise.
5276 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5277 instead of TARGET_OPT_AGU.
5278 * config/i386/i386.h (TARGET_INTEL): New.
5279 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5280 (processor_type): Add PROCESSOR_INTEL.
5281 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5282 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5284 2014-01-17 Marek Polacek <polacek@redhat.com>
5286 PR c/58346
5287 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5288 size is zero.
5290 2014-01-17 Richard Biener <rguenther@suse.de>
5292 PR tree-optimization/46590
5293 * opts.c (default_options_table): Add entries for
5294 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5295 all enabled at -O1 but not for -Og.
5296 * common.opt (fbranch-count-reg): Remove Init(1).
5297 (fmove-loop-invariants): Likewise.
5298 (ftree-pta): Likewise.
5300 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5302 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5303 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5304 decls to at least the GCC 4.8 used alignments.
5306 PR fortran/59440
5307 * tree-nested.c (convert_nonlocal_reference_stmt,
5308 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5309 of GIMPLE_BIND stmts, adjust associated decls.
5311 2014-01-17 Richard Biener <rguenther@suse.de>
5313 PR tree-optimization/46590
5314 * vec.h (vec<>::bseach): New member function implementing
5315 binary search according to C89 bsearch.
5316 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5317 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5318 bitmap pointer again. Make accesses_in_loop a flat array.
5319 (mem_ref_obstack): New global.
5320 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5321 (mark_ref_stored): Likewise.
5322 (ref_indep_loop_p_2): Likewise.
5323 (set_ref_stored_in_loop): New helper function.
5324 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5325 (memref_free): Adjust.
5326 (record_mem_ref_loc): Simplify.
5327 (gather_mem_refs_stmt): Adjust.
5328 (sort_locs_in_loop_postorder_cmp): New function.
5329 (analyze_memory_references): Sort accesses_in_loop after
5330 loop postorder number.
5331 (find_ref_loc_in_loop_cmp): New function.
5332 (for_all_locs_in_loop): Find relevant cluster of locs in
5333 accesses_in_loop and iterate without recursion.
5334 (execute_sm): Avoid uninit warning.
5335 (struct ref_always_accessed): Simplify.
5336 (ref_always_accessed::operator ()): Likewise.
5337 (ref_always_accessed_p): Likewise.
5338 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5339 loop postorder numbers here.
5340 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5341 numbers.
5343 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5345 PR c++/57945
5346 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5347 on decls for which assemble_alias has been called.
5349 2014-01-17 Nick Clifton <nickc@redhat.com>
5351 * config/msp430/msp430.opt: (mcpu): New option.
5352 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5353 (msp430_option_override): Parse target_cpu. If the MCU name
5354 matches a generic string, clear target_mcu.
5355 (msp430_attr): Allow numeric interrupt values up to 63.
5356 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5357 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5358 option.
5359 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5360 Add mcpu matches.
5361 * config/msp430/msp430.md (popm): Use %J rather than %I.
5362 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5363 (addhi_cy_i): Use immediate_operand for operand 2.
5364 * doc/invoke.texi: Document -mcpu option.
5366 2014-01-17 Richard Biener <rguenther@suse.de>
5368 PR rtl-optimization/38518
5369 * df.h (df_analyze_loop): Declare.
5370 * df-core.c: Include cfgloop.h.
5371 (df_analyze_1): Split out main part of df_analyze.
5372 (df_analyze): Adjust.
5373 (loop_inverted_post_order_compute): New function.
5374 (loop_post_order_compute): Likewise.
5375 (df_analyze_loop): New function avoiding whole-function
5376 postorder computes.
5377 * loop-invariant.c (find_defs): Use df_analyze_loop.
5378 (find_invariants): Adjust.
5379 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5381 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5383 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5384 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5386 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5388 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5389 traversal when removing references.
5391 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5393 PR ipa/59775
5394 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5396 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5398 PR middle-end/56791
5399 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5400 pushing a reload for an autoinc when we had previously reloaded an
5401 inner part of the address.
5403 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5405 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5406 field.
5407 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5408 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5409 when not giving up or versioning for alias only because of
5410 loop->safelen.
5411 (vect_analyze_data_ref_dependences): Set to true.
5412 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5413 is a GIMPLE_PHI.
5414 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5415 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5416 to the condition.
5418 PR middle-end/58344
5419 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5421 PR target/59839
5422 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5423 operand 0 predicate for gathers, use a new pseudo as subtarget.
5425 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5427 PR middle-end/59609
5428 * lra-constraints.c (process_alt_operands): Add printing debug info.
5429 Check absence of input/output reloads for matched operands too.
5431 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5433 PR rtl-optimization/59835
5434 * ira.c (ira_init_register_move_cost): Increase cost for
5435 impossible modes.
5437 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5439 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5441 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5443 PR target/59780
5444 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5445 non-register objects. Use gen_(high/low)part more consistently.
5446 Fix assertions.
5448 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5450 PR target/59844
5451 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5452 endian support, remove tests for WORDS_BIG_ENDIAN.
5453 (p8_mfvsrd_3_<mode>): Likewise.
5454 (reload_gpr_from_vsx<mode>): Likewise.
5455 (reload_gpr_from_vsxsf): Likewise.
5456 (p8_mfvsrd_4_disf): Likewise.
5458 2014-01-16 Richard Biener <rguenther@suse.de>
5460 PR rtl-optimization/46590
5461 * lcm.c (compute_antinout_edge): Use postorder iteration.
5462 (compute_laterin): Use inverted postorder iteration.
5464 2014-01-16 Nick Clifton <nickc@redhat.com>
5466 PR middle-end/28865
5467 * varasm.c (output_constant): Return the number of bytes actually
5468 emitted.
5469 (output_constructor_array_range): Update the field size with the
5470 number of bytes emitted by output_constant.
5471 (output_constructor_regular_field): Likewise. Also do not
5472 complain if the total number of bytes emitted is now greater
5473 than the expected fieldpos.
5474 * output.h (output_constant): Update prototype and descriptive comment.
5476 2014-01-16 Marek Polacek <polacek@redhat.com>
5478 PR middle-end/59827
5479 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5480 it is error_mark_node.
5482 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5484 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5485 VALID_AVX256_REG_OR_OI_MODE.
5487 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5489 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5490 current procedure should be profiled.
5492 2014-01-15 Andrew Pinski <apinski@cavium.com>
5494 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5495 of moving from/to the STACK_REG register class.
5497 2014-01-15 Richard Henderson <rth@redhat.com>
5499 PR debug/54694
5500 * reginfo.c (global_regs_decl): Globalize.
5501 * rtl.h (global_regs_decl): Declare.
5502 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5503 reserved via global_regs.
5505 2014-01-15 Teresa Johnson <tejohnson@google.com>
5507 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5509 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5511 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5512 and vmulosh rather than call gen_vec_widen_smult_*.
5513 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5514 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5515 (vec_widen_smult_even_v16qi): Likewise.
5516 (vec_widen_umult_even_v8hi): Likewise.
5517 (vec_widen_smult_even_v8hi): Likewise.
5518 (vec_widen_umult_odd_v16qi): Likewise.
5519 (vec_widen_smult_odd_v16qi): Likewise.
5520 (vec_widen_umult_odd_v8hi): Likewise.
5521 (vec_widen_smult_odd_v8hi): Likewise.
5522 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5523 vmuloub rather than call gen_vec_widen_umult_*.
5524 (vec_widen_umult_lo_v16qi): Likewise.
5525 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5526 vmulosb rather than call gen_vec_widen_smult_*.
5527 (vec_widen_smult_lo_v16qi): Likewise.
5528 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5529 rather than call gen_vec_widen_umult_*.
5530 (vec_widen_umult_lo_v8hi): Likewise.
5531 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5532 rather than call gen_vec_widen_smult_*.
5533 (vec_widen_smult_lo_v8hi): Likewise.
5535 2014-01-15 Jeff Law <law@redhat.com>
5537 PR tree-optimization/59747
5538 * ree.c (find_and_remove_re): Properly handle case where a second
5539 eliminated extension requires widening a copy created for elimination
5540 of a prior extension.
5541 (combine_set_extension): Ensure that the number of hard regs needed
5542 for a destination register does not change when we widen it.
5544 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5546 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5547 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5548 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5549 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5550 (avr-*-rtems*): Likewise.
5551 (bfin*-rtems*): Likewise.
5552 (moxie-*-rtems*): Likewise.
5553 (h8300-*-rtems*): Likewise.
5554 (i[34567]86-*-rtems*): Likewise.
5555 (lm32-*-rtems*): Likewise.
5556 (m32r-*-rtems*): Likewise.
5557 (m68k-*-rtems*): Likewise.
5558 (microblaze*-*-rtems*): Likewise.
5559 (mips*-*-rtems*): Likewise.
5560 (powerpc-*-rtems*): Likewise.
5561 (sh-*-rtems*): Likewise.
5562 (sparc-*-rtems*): Likewise.
5563 (sparc64-*-rtems*): Likewise.
5564 (v850-*-rtems*): Likewise.
5565 (m32c-*-rtems*): Likewise.
5567 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5569 PR rtl-optimization/59511
5570 * ira.c (ira_init_register_move_cost): Use memory costs for some
5571 cases of register move cost calculations.
5572 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5573 instead of BB frequency.
5574 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5575 * lra-assigns.c (find_hard_regno_for): Ditto.
5577 2014-01-15 Richard Biener <rguenther@suse.de>
5579 PR tree-optimization/59822
5580 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5581 (vectorizable_load): Use it to hoist defs of uses of invariant
5582 loads out of the loop.
5584 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5585 Kugan Vivekanandarajah <kuganv@linaro.org>
5587 PR target/59695
5588 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5589 truncation.
5591 2014-01-15 Richard Biener <rguenther@suse.de>
5593 PR rtl-optimization/59802
5594 * lcm.c (compute_available): Use inverted postorder to seed
5595 the initial worklist.
5597 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5599 PR target/59803
5600 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5601 ADDR_REGS for invalid symrefs in non-PIC code.
5603 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5605 PR other/58712
5606 * builtins.c (determine_block_size): Initialize *probable_max_size
5607 even if len_rtx is CONST_INT.
5609 2014-01-14 Andrew Pinski <apinski@cavium.com>
5611 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5612 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5613 (cortexa53_tunings): Likewise.
5614 (aarch64_sched_issue_rate): New function.
5615 (TARGET_SCHED_ISSUE_RATE): Define.
5617 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5619 * ira-costs.c (find_costs_and_classes): Add missed
5620 ira_init_register_move_cost_if_necessary.
5622 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5624 PR target/59787
5625 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5627 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5629 PR target/59794
5630 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5631 to indicate if type is used for function return value. Warn ABI
5632 change if the vector mode isn't available for function return value.
5633 (ix86_function_arg_advance): Pass false to type_natural_mode.
5634 (ix86_function_arg): Likewise.
5635 (ix86_gimplify_va_arg): Likewise.
5636 (function_arg_32): Don't warn ABI change.
5637 (ix86_function_value): Pass true to type_natural_mode.
5638 (ix86_return_in_memory): Likewise.
5639 (ix86_struct_value_rtx): Removed.
5640 (TARGET_STRUCT_VALUE_RTX): Likewise.
5642 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5644 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5645 converting a conditional jump into a conditional return.
5647 2014-01-14 Richard Biener <rguenther@suse.de>
5649 PR tree-optimization/58921
5650 PR tree-optimization/59006
5651 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5652 hoisting invariant stmts.
5653 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5654 invariant loads on the preheader edge if possible.
5656 2014-01-14 Joey Ye <joey.ye@arm.com>
5658 * doc/plugin.texi (Building GCC plugins): Update to C++.
5660 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5662 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5663 (_mm_rcp28_round_ss): Ditto.
5664 (_mm_rsqrt28_round_sd): Ditto.
5665 (_mm_rsqrt28_round_ss): Ditto.
5666 (_mm_rcp28_sd): Ditto.
5667 (_mm_rcp28_ss): Ditto.
5668 (_mm_rsqrt28_sd): Ditto.
5669 (_mm_rsqrt28_ss): Ditto.
5670 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5671 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5672 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5673 (IX86_BUILTIN_RCP28SD): Ditto.
5674 (IX86_BUILTIN_RCP28SS): Ditto.
5675 (IX86_BUILTIN_RSQRT28SD): Ditto.
5676 (IX86_BUILTIN_RSQRT28SS): Ditto.
5677 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5678 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5679 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5680 (ix86_expand_special_args_builtin): Expand new FTYPE.
5681 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5682 (srcp14<mode>): Make insn unary.
5683 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5684 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5685 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5686 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5687 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5688 Fix rounding: make it SAE only.
5689 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5690 Ditto.
5691 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5692 Ditto.
5693 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5694 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5695 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5696 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5697 (round_saeonly_mask_scalar_operand4): Ditto.
5698 (round_saeonly_mask_scalar_op3): Ditto.
5699 (round_saeonly_mask_scalar_op4): Ditto.
5701 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5703 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5704 Implement -maltivec=be for vec_insert and vec_extract.
5706 2014-01-10 DJ Delorie <dj@redhat.com>
5708 * config/msp430/msp430.md (call_internal): Don't allow memory
5709 references with SP as the base register.
5710 (call_value_internal): Likewise.
5711 * config/msp430/constraints.md (Yc): New. For memory references
5712 that don't use SP as a base register.
5714 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5715 "an integer without a # prefix"
5716 * config/msp430/msp430.md (epilogue_helper): Use it.
5718 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5720 PR target/59617
5721 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5722 AVX512F gather builtins.
5723 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5724 on gather decls with INTEGER_TYPE masktype.
5725 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5726 directly into the builtin rather than hoisting it before loop.
5728 PR tree-optimization/59387
5729 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5730 (scev_const_prop): If folded_casts and type has undefined overflow,
5731 use force_gimple_operand instead of force_gimple_operand_gsi and
5732 for each added stmt if it is assign with
5733 arith_code_with_undefined_signed_overflow, call
5734 rewrite_to_defined_overflow.
5735 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5736 gimple-fold.h instead.
5737 (arith_code_with_undefined_signed_overflow,
5738 rewrite_to_defined_overflow): Moved to ...
5739 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5740 rewrite_to_defined_overflow): ... here. No longer static.
5741 Include gimplify-me.h.
5742 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5743 rewrite_to_defined_overflow): New prototypes.
5745 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5747 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5749 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5751 * builtins.c (get_object_alignment_2): Minor tweak.
5752 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5754 2014-01-13 Christian Bruel <christian.bruel@st.com>
5756 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5757 optimized non constant lengths.
5759 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5761 PR libgomp/59194
5762 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5763 load as __atomic_load_N if possible.
5765 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5767 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5768 target parameter.
5769 (rs6000_expand_builtin): Adjust call.
5771 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5773 PR target/58115
5774 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5775 * config/rs6000/rs6000.c: Include target-globals.h.
5776 (rs6000_set_current_function): Instead of doing target_reinit
5777 unconditionally, use save_target_globals_default_opts and
5778 restore_target_globals.
5780 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5781 FPSCR.
5782 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5783 (rs6000_expand_builtin): Handle mffs and mtfsf.
5784 (rs6000_init_builtins): Define mffs and mtfsf.
5785 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5786 (rs6000_mffs): New pattern.
5787 (rs6000_mtfsf): New pattern.
5789 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5791 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5792 Start narrowing with START. Apply candidate-use pair
5793 and check overall cost in narrowing.
5794 (iv_ca_prune): Pass new argument.
5796 2014-01-10 Jeff Law <law@redhat.com>
5798 PR middle-end/59743
5799 * ree.c (combine_reaching_defs): Ensure the defining statement
5800 occurs before the extension when optimizing extensions with
5801 different source and destination hard registers.
5803 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5805 PR ipa/58585
5806 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5807 vtables into the type inheritance graph.
5809 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5811 PR rtl-optimization/59754
5812 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5813 modes in the REGNO != REGNO case.
5815 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5817 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5819 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5821 PR tree-optimization/59745
5822 * tree-predcom.c (tree_predictive_commoning_loop): Call
5823 free_affine_expand_cache if giving up because components is NULL.
5825 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5826 GC in payload of target_globals struct instead of allocating them on
5827 the heap and the larger structs separately using GC.
5828 * target-globals.h (struct target_globals): Make regs, hard_regs,
5829 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5830 of GTY((skip)) and change type to void *.
5831 (reset_target_globals): Cast loads from those fields to corresponding
5832 types.
5834 2014-01-10 Steve Ellcey <sellcey@mips.com>
5836 PR plugins/59335
5837 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5838 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5839 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5841 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5843 PR target/59744
5844 * aarch64-modes.def (CC_Zmode): New flags mode.
5845 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5846 represents an equality.
5847 (aarch64_get_condition_code): Handle CC_Zmode.
5848 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5850 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5852 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5853 extraction in good case.
5855 2014-01-10 Richard Biener <rguenther@suse.de>
5857 PR tree-optimization/59374
5858 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5859 checking after SLP discovery. Mark stmts not participating
5860 in any SLP instance properly.
5862 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5864 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5865 when handling a SET rtx.
5867 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5869 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5870 (cortex-a57): Likewise.
5871 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5873 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5875 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5876 non-iwmmxt builtins.
5878 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5880 PR ipa/58252
5881 PR ipa/59226
5882 * ipa-devirt.c record_target_from_binfo): Take as argument
5883 stack of binfos and lookup matching one for virtual inheritance.
5884 (possible_polymorphic_call_targets_1): Update.
5886 2014-01-10 Huacai Chen <chenhc@lemote.com>
5888 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5889 kernel strings for Loongson-2E/2F/3A.
5891 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5893 PR middle-end/59670
5894 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5895 is_gimple_call before calling gimple_call_internal_p.
5897 2014-01-09 Steve Ellcey <sellcey@mips.com>
5899 * Makefile.in (TREE_FLOW_H): Remove.
5900 (TREE_SSA_H): Add file names from tree-flow.h.
5901 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5902 * tree.h: Remove tree-flow.h reference.
5903 * hash-table.h: Remove tree-flow.h reference.
5904 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5905 reference with tree-ssa-loop.h.
5907 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5909 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5910 default element-order behavior for -maltivec.
5911 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5912 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5913 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5914 when targeting big endian, at least for now.
5915 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5917 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5919 PR middle-end/47735
5920 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5921 var satisfies use_register_for_decl, just take into account type
5922 alignment, rather than decl alignment.
5924 PR tree-optimization/59622
5925 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5926 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5927 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5928 Don't devirtualize for inplace at all. For targets.length () == 1,
5929 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5931 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5933 * config/i386/i386.md (cpu): Remove the unused btver1.
5935 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5937 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5939 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5941 PR target/58115
5942 * tree-core.h (struct target_globals): New forward declaration.
5943 (struct tree_target_option): Add globals field.
5944 * tree.h (TREE_TARGET_GLOBALS): Define.
5945 (prepare_target_option_nodes_for_pch): New prototype.
5946 * target-globals.h (struct target_globals): Define even if
5947 !SWITCHABLE_TARGET.
5948 * tree.c (prepare_target_option_node_for_pch,
5949 prepare_target_option_nodes_for_pch): New functions.
5950 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5951 * config/i386/i386.c: Include target-globals.h.
5952 (ix86_set_current_function): Instead of doing target_reinit
5953 unconditionally, use save_target_globals_default_opts and
5954 restore_target_globals.
5956 2014-01-09 Richard Biener <rguenther@suse.de>
5958 PR tree-optimization/59715
5959 * tree-cfg.h (split_critical_edges): Declare.
5960 * tree-cfg.c (split_critical_edges): Export.
5961 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5963 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5965 * cfgexpand.c (expand_stack_vars): Optionally disable
5966 asan stack protection.
5967 (expand_used_vars): Likewise.
5968 (partition_stack_vars): Likewise.
5969 * asan.c (asan_emit_stack_protection): Optionally disable
5970 after return stack usage.
5971 (instrument_derefs): Optionally disable memory access instrumentation.
5972 (instrument_builtin_call): Likewise.
5973 (instrument_strlen_call): Likewise.
5974 (asan_protect_global): Optionally disable global variables protection.
5975 * doc/invoke.texi: Added doc for new options.
5976 * params.def: Added new options.
5977 * params.h: Likewise.
5979 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5981 PR rtl-optimization/59724
5982 * ifcvt.c (cond_exec_process_if_block): Don't call
5983 flow_find_head_matching_sequence with 0 longest_match.
5984 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5985 non-active insns if !stop_after.
5986 (try_head_merge_bb): Revert 2014-01-07 changes.
5988 2014-01-08 Jeff Law <law@redhat.com>
5990 * ree.c (get_sub_rtx): New function, extracted from...
5991 (merge_def_and_ext): Here.
5992 (combine_reaching_defs): Use get_sub_rtx.
5994 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5996 * cgraph.h (varpool_variable_node): Do not choke on null node.
5998 2014-01-08 Catherine Moore <clm@codesourcery.com>
6000 * config/mips/mips.md (simple_return): Attempt to use JRC
6001 for microMIPS.
6002 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6004 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6006 PR rtl-optimization/59137
6007 * reorg.c (steal_delay_list_from_target): Call update_block for
6008 elided insns.
6009 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6011 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6013 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6014 two duplicate entries.
6016 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6018 Revert:
6019 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6021 * config/mips/mips.c (mips_truncated_op_cost): New function.
6022 (mips_rtx_costs): Adjust test for BADDU.
6023 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6025 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6027 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6028 (*baddu_si): ...this new pattern.
6030 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6032 PR ipa/59722
6033 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6035 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6037 PR middle-end/57748
6038 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6039 inner_reference_p.
6040 (expand_expr, expand_normal): Adjust.
6041 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6042 inner_reference_p. Use inner_reference_p to expand inner references.
6043 (store_expr): Adjust.
6044 * cfgexpand.c (expand_call_stmt): Adjust.
6046 2014-01-08 Rong Xu <xur@google.com>
6048 * gcov-io.c (gcov_var): Move from gcov-io.h.
6049 (gcov_position): Ditto.
6050 (gcov_is_error): Ditto.
6051 (gcov_rewrite): Ditto.
6052 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6053 only part to libgcc/libgcov.h.
6055 2014-01-08 Marek Polacek <polacek@redhat.com>
6057 PR middle-end/59669
6058 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6060 2014-01-08 Marek Polacek <polacek@redhat.com>
6062 PR sanitizer/59667
6063 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6065 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6067 PR rtl-optimization/59649
6068 * stor-layout.c (get_mode_bounds): For BImode return
6069 0 and STORE_FLAG_VALUE.
6071 2014-01-08 Richard Biener <rguenther@suse.de>
6073 PR middle-end/59630
6074 * gimple.h (is_gimple_builtin_call): Remove.
6075 (gimple_builtin_call_types_compatible_p): New.
6076 (gimple_call_builtin_p): New overload.
6077 * gimple.c (is_gimple_builtin_call): Remove.
6078 (validate_call): Rename to ...
6079 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6080 check return types.
6081 (validate_type): New static function.
6082 (gimple_call_builtin_p): New overload and adjust.
6083 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6084 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6085 (gimple_fold_stmt_to_constant_1): Likewise.
6086 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6088 2014-01-08 Richard Biener <rguenther@suse.de>
6090 PR middle-end/59471
6091 * gimplify.c (gimplify_expr): Gimplify register-register type
6092 VIEW_CONVERT_EXPRs to separate stmts.
6094 2014-01-07 Jeff Law <law@redhat.com>
6096 PR middle-end/53623
6097 * ree.c (combine_set_extension): Handle case where source
6098 and destination registers in an extension insn are different.
6099 (combine_reaching_defs): Allow source and destination registers
6100 in extension to be different under limited circumstances.
6101 (add_removable_extension): Remove restriction that the
6102 source and destination registers in the extension are the same.
6103 (find_and_remove_re): Emit a copy from the extension's
6104 destination to its source after the defining insn if
6105 the source and destination registers are different.
6107 PR middle-end/59285
6108 * ifcvt.c (merge_if_block): If we are merging a block with more than
6109 one successor with a block with no successors, remove any BARRIER
6110 after the second block.
6112 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6114 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6116 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6118 PR target/59652
6119 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6120 for 14-bit register offsets when INT14_OK_STRICT is false.
6122 2014-01-07 Roland Stigge <stigge@antcom.de>
6123 Michael Meissner <meissner@linux.vnet.ibm.com>
6125 PR 57386/target
6126 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6127 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6129 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6131 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6132 -mcpu.
6134 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6136 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6137 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6138 rtx is const0_rtx or not.
6140 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6142 PR target/58115
6143 * target-globals.c (save_target_globals): Remove this_fn_optab
6144 handling.
6145 * toplev.c: Include optabs.h.
6146 (target_reinit): Temporarily restore the global options if another
6147 set of options are in force.
6149 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6151 PR rtl-optimization/58668
6152 * cfgcleanup.c (flow_find_cross_jump): Don't count
6153 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6154 to determine what is counted.
6155 (flow_find_head_matching_sequence): Use active_insn_p to determine
6156 what is counted.
6157 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6158 counting change.
6159 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6160 determine what is counted.
6162 PR tree-optimization/59643
6163 * tree-predcom.c (split_data_refs_to_components): If one dr is
6164 read and one write, determine_offset fails and the write isn't
6165 in the bad component, just put the read into the bad component.
6167 2014-01-07 Mike Stump <mikestump@comcast.net>
6168 Jakub Jelinek <jakub@redhat.com>
6170 PR pch/59436
6171 * tree-core.h (struct tree_optimization_option): Change optabs
6172 type from unsigned char * to void *.
6173 * optabs.c (init_tree_optimization_optabs): Adjust
6174 TREE_OPTIMIZATION_OPTABS initialization.
6176 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6178 PR target/59644
6179 * config/i386/i386.h (struct machine_function): Add
6180 no_drap_save_restore field.
6181 * config/i386/i386.c (ix86_save_reg): Use
6182 !cfun->machine->no_drap_save_restore instead of
6183 crtl->stack_realign_needed.
6184 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6185 this function clears frame_pointer_needed. Set
6186 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6187 and DRAP reg is needed.
6189 2014-01-06 Marek Polacek <polacek@redhat.com>
6191 PR c/57773
6192 * doc/implement-c.texi: Mention that other integer types are
6193 permitted as bit-field types in strictly conforming mode.
6195 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6197 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6198 is newly allocated.
6200 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6202 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6204 2014-01-06 Martin Jambor <mjambor@suse.cz>
6206 PR ipa/59008
6207 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6208 to int.
6209 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6211 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6213 PR debug/59350
6214 PR debug/59510
6215 * var-tracking.c (add_stores): Preserve the value of the source even if
6216 we don't record the store.
6218 2014-01-06 Terry Guo <terry.guo@arm.com>
6220 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6222 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6224 PR bootstrap/59541
6225 * config/darwin.c (darwin_function_section): Adjust return values to
6226 correspond to optimisation changes made in r206070.
6228 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6230 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6231 from prefetch_block tune setting.
6232 (nocona_cost): Correct size of prefetch block to 64.
6234 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6236 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6237 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6238 used to save the static chain register in the computation of the offset
6239 from which the FP registers need to be restored.
6241 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6243 PR tree-optimization/59519
6244 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6245 ICE if get_current_def (current_new_name) is already non-NULL, as long
6246 as it is a phi result of some other phi in *new_exit_bb that has
6247 the same argument.
6249 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6250 or vmovdqu* for misaligned_operand.
6251 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6252 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6253 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6254 aligned_mem for AVX512F masked aligned load and store builtins and for
6255 non-temporal moves.
6257 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6259 PR tree-optimization/59651
6260 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6261 Address range for negative step should be added by TYPE_SIZE_UNIT.
6263 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6265 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6266 overlapping registers also for registers other than the stack pointer.
6268 2014-01-03 Marek Polacek <polacek@redhat.com>
6270 PR other/59661
6271 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6272 __builtin_FILE.
6274 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6276 PR target/59625
6277 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6278 asm goto as jump.
6280 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6281 (push splitter): Use <P:MODE_SIZE> instead of
6282 GET_MODE_SIZE (<P:MODE>mode).
6283 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6284 (mov -1, reg peephole2): Likewise.
6285 * config/i386/sse.md (*mov<mode>_internal,
6286 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6287 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6288 *<code><mode>3, *andnot<mode>3<mask_name>,
6289 <mask_codefor><code><mode>3<mask_name>): Likewise.
6290 * config/i386/subst.md (mask_mode512bit_condition,
6291 sd_mask_mode512bit_condition): Likewise.
6293 2014-01-02 Xinliang David Li <davidxl@google.com>
6295 PR tree-optimization/59303
6296 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6297 (dump_predicates): Better output format.
6298 (pred_equal_p): New function.
6299 (is_neq_relop_p): Ditto.
6300 (is_neq_zero_form_p): Ditto.
6301 (pred_expr_equal_p): Ditto.
6302 (pred_neg_p): Ditto.
6303 (simplify_pred): Ditto.
6304 (simplify_preds_2): Ditto.
6305 (simplify_preds_3): Ditto.
6306 (simplify_preds_4): Ditto.
6307 (simplify_preds): Ditto.
6308 (push_pred): Ditto.
6309 (push_to_worklist): Ditto.
6310 (get_pred_info_from_cmp): Ditto.
6311 (is_degenerated_phi): Ditto.
6312 (normalize_one_pred_1): Ditto.
6313 (normalize_one_pred): Ditto.
6314 (normalize_one_pred_chain): Ditto.
6315 (normalize_preds): Ditto.
6316 (normalize_cond_1): Remove function.
6317 (normalize_cond): Ditto.
6318 (is_gcond_subset_of): Ditto.
6319 (is_subset_of_any): Ditto.
6320 (is_or_set_subset_of): Ditto.
6321 (is_and_set_subset_of): Ditto.
6322 (is_norm_cond_subset_of): Ditto.
6323 (pred_chain_length_cmp): Ditto.
6324 (convert_control_dep_chain_into_preds): Type change.
6325 (find_predicates): Ditto.
6326 (find_def_preds): Ditto.
6327 (destroy_predicates_vecs): Ditto.
6328 (find_matching_predicates_in_rest_chains): Ditto.
6329 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6330 (is_pred_expr_subset): Ditto.
6331 (is_pred_chain_subset_of): Ditto.
6332 (is_included_in): Ditto.
6333 (is_superset_of): Ditto.
6335 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6337 Update copyright years.
6339 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6341 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6342 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6343 config/arc/arc.md, config/arc/arc.opt,
6344 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6345 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6346 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6347 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6348 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6349 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6350 vtable-verify.c, vtable-verify.h: Use the standard form for the
6351 copyright notice.
6353 2014-01-02 Tobias Burnus <burnus@net-b.de>
6355 * gcc.c (process_command): Update copyright notice dates.
6356 * gcov-dump.c: Ditto.
6357 * gcov.c: Ditto.
6358 * doc/cpp.texi: Bump @copying's copyright year.
6359 * doc/cppinternals.texi: Ditto.
6360 * doc/gcc.texi: Ditto.
6361 * doc/gccint.texi: Ditto.
6362 * doc/gcov.texi: Ditto.
6363 * doc/install.texi: Ditto.
6364 * doc/invoke.texi: Ditto.
6366 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6368 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6370 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6372 * config/i386/sse.md (*mov<mode>_internal): Guard
6373 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6375 PR rtl-optimization/59647
6376 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6377 new_rtx into UNSIGNED_FLOAT rtxes.
6378 \f
6379 Copyright (C) 2014 Free Software Foundation, Inc.
6381 Copying and distribution of this file, with or without modification,
6382 are permitted in any medium without royalty provided the copyright
6383 notice and this notice are preserved.