1 2014-04-14 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/60042
4 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
5 (insert_into_preds_of_block): Do not prevent PHI insertion
6 for REFERENCE exprs here ...
7 (eliminate_dom_walker::before_dom_children): ... but prevent
8 their use here under similar conditions when applied to the
9 IL after PRE optimizations.
11 2014-04-14 Richard Biener <rguenther@suse.de>
13 * passes.def: Move early points-to after early SRA.
15 2014-04-14 Richard Biener <rguenther@suse.de>
17 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
18 check for which sign-changes we allow when forwarding
19 a converted value into a switch.
21 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
23 * stor-layout.c (place_field): Finalize non-constant offset for the
26 2014-04-14 Richard Biener <rguenther@suse.de>
28 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
30 (expand_switch_using_bit_tests_p): Likewise.
31 (process_switch): Compute and pass on speed_p based on the
33 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
34 optimize_bb_for_speed_p.
36 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
38 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
39 * function.h (struct function): Rename has_force_vect_loops into
40 has_force_vectorize_loops.
41 * lto-streamer-in.c (input_cfg): Adjust for renaming.
42 (input_struct_function_base): Likewise.
43 * lto-streamer-out.c (output_cfg): Likewise.
44 (output_struct_function_base): Likewise.
45 * omp-low.c (expand_omp_simd): Likewise.
46 * tree-cfg.c (move_sese_region_to_fn): Likewise.
47 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
48 (version_loop_for_if_conversion): Likewise.
49 (tree_if_conversion): Likewise.
50 (main_tree_if_conversion): Likewise.
51 (gate_tree_if_conversion): Likewise.
52 * tree-inline.c (copy_loops): Likewise.
53 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
54 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
55 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
56 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
57 * tree-vectorizer.c (vectorize_loops): Likewise.
58 * tree-vectorizer.h (unlimited_cost_model): Likewise.
60 2014-04-14 Richard Biener <rguenther@suse.de>
63 * lto-streamer-out.c (wrap_refs): New function.
64 (lto_output): Wrap symbol references in global initializes in
65 type-preserving MEM_REFs.
67 2014-04-14 Christian Bruel <christian.bruel@st.com>
69 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
71 2014-04-14 Christian Bruel <christian.bruel@st.com>
73 * config/sh/sh.md (setmemqi): New expand pattern.
74 * config/sh/sh.h (CLEAR_RATIO): Define.
75 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
76 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
78 2014-04-14 Richard Biener <rguenther@suse.de>
81 * fold-const.c (negate_expr_p): Don't negate directional rounding
83 (fold_negate_expr): Likewise.
85 2014-04-14 Richard Biener <rguenther@suse.de>
87 PR tree-optimization/59817
88 PR tree-optimization/60453
89 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
90 recursion to catch all CHRECs in the scalar evolution and restrict
91 the predicate for the remains appropriately.
93 2014-04-12 Catherine Moore <clm@codesourcery.com>
95 * config/mips/constraints.md: Add new register constraint "kb".
96 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
97 (*movhi_internal): Likewise.
98 (*movqi_internal): Likewise.
99 * config/mips/mips.h (M16_STORE_REGS): New register class.
100 (REG_CLASS_NAMES): Add M16_STORE_REGS.
101 (REG_CLASS_CONTENTS): Likewise.
102 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
104 2014-04-11 Tobias Burnus <burnus@net-b.de>
107 * doc/invoke.texi (-Wformat-signedness): Document it.
108 (Wformat=2): Mention that this enables -Wformat-signedness.
110 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
112 * common/config/epiphany/epiphany-common.c
113 (epiphany_option_optimization_table): Enable section anchors by
114 default at -O1 or higher.
115 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
116 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
117 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
118 carries no extra cost.
119 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
120 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
121 * config/epiphany/predicates.md (memclob_operand): New predicate.
122 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
123 Use memclob_operand predicate and X constraint for operand 3.
125 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
127 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
128 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
131 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
133 PR rtl-optimization/60651
134 * mode-switching.c (optimize_mode_switching): Make sure to emit
135 sets of a lower numbered entity before sets of a higher numbered
136 entity to a mode of the same or lower priority.
137 When creating a seginfo for a basic block that starts with a code
138 label, move the insertion point past the code label.
139 (new_seginfo): Document and enforce requirement that
140 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
141 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
142 * doc/tm.texi: Regenerate.
144 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
147 * config/arc/arc.c (arc_save_restore): Fix assert typo.
149 2013-04-11 Jakub Jelinek <jakub@redhat.com>
151 * BASE-VER: Set to 4.10.0.
153 2014-04-11 Tobias Burnus <burnus@net-b.de>
156 * doc/bugreport.texi (Bugs): Remove nodes pointing to the
158 * doc/gcc.texi (Service): Update description in the @menu
159 * doc/invoke.texi (Option Summary): Remove misplaced and
162 2014-04-11 Steve Ellcey <sellcey@mips.com>
163 Jakub Jelinek <jakub@redhat.com>
166 * expr.c (convert_move): Use emit_store_flag_force instead of
167 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
170 2014-04-11 Richard Biener <rguenther@suse.de>
173 * varasm.c (assemble_alias): Avoid endless error reporting
174 recursion by setting TREE_ASM_WRITTEN.
176 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
178 * config/s390/s390.md: Add a splitter for NOT rtx.
180 2014-04-11 Jakub Jelinek <jakub@redhat.com>
182 PR rtl-optimization/60663
183 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
186 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
187 Jakub Jelinek <jakub@redhat.com>
190 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
193 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
196 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
197 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
198 ameliorating the cases where it can be.
200 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
203 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
205 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
206 (loadsync_<mode>): Change mode.
207 (load_quadpti, store_quadpti): New.
208 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
209 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
210 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
212 2014-04-09 Cong Hou <congh@google.com>
215 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
218 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
220 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
221 instead of vnor to exploit possible fusion opportunity in the
223 (altivec_expand_vec_perm_const_le): Likewise.
225 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
227 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
228 (loadsync_<mode>): Change mode.
229 (load_quadpti, store_quadpti): New.
230 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
231 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
233 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
236 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
237 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
238 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
240 2014-04-08 Richard Biener <rguenther@suse.de>
243 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
244 a 64bit widest int print double-int similar to on HWI64 hosts.
246 2014-04-08 Richard Biener <rguenther@suse.de>
248 PR tree-optimization/60785
249 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
250 default defs properly.
252 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
254 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
257 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
259 * ipa-devirt.c (maybe_record_node): When node is not recorded,
260 set completep to false rather than true.
262 2014-04-07 Douglas B Rupp <rupp@adacore.com>
265 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
266 ARM_TARGET2_DWARF_FORMAT.
268 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
271 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
272 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
275 2014-04-07 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/60766
278 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
279 (may_eliminate_iv): Convert cand_value_at result to desired type.
281 2014-04-07 Jason Merrill <jason@redhat.com>
284 * common.opt (-fno-gnu-unique): Add.
285 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
287 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
289 * haifa-sched.c: Fix outdated function reference and minor
290 grammar errors in introductory comment.
292 2014-04-07 Richard Biener <rguenther@suse.de>
295 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
297 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
299 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
302 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
303 size accounting for thunks.
304 (pa_asm_output_mi_thunk): Use final_start_function() and
305 final_end_function() to output function start and end directives.
307 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
309 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
310 specific ISA/ feature information. Remove short_sp and errata_skip ds.
311 Add avr_device_specific_features enum to have device specific info.
312 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
313 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
314 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
315 updated device specific info.
316 * config/avr/avr-mcus.def: Merge device specific details to
318 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
320 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
321 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
322 assembler if RMW isa supported by current device.
323 * config/avr/genmultilib.awk: Update as device info structure changed.
324 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
326 2014-04-04 Cong Hou <congh@google.com>
328 PR tree-optimization/60656
329 * tree-vect-stmts.c (supportable_widening_operation):
330 Fix a bug that elements in a vector with vect_used_by_reduction
331 property are incorrectly reordered when the operation on it is not
332 consistant with the one in reduction operation.
334 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
336 PR rtl-optimization/60155
337 * gcse.c (record_set_data): New function.
338 (single_set_gcse): New function.
339 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
340 (hoist_code): Likewise.
341 (get_pressure_class_and_nregs): Likewise.
343 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
345 * explow.c (probe_stack_range): Emit a final optimization blockage.
347 2014-04-04 Anthony Green <green@moxielogic.com>
349 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
352 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
355 * lto-cgraph.c (input_overwrite_node): Check that partitioning
356 flags are set only during streaming.
357 * ipa.c (process_references, walk_polymorphic_call_targets,
358 symtab_remove_unreachable_nodes): Drop bodies of always inline
359 after early inlining.
360 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
362 2014-04-04 Jakub Jelinek <jakub@redhat.com>
363 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
366 * dwarf2out.c (const_ok_for_output_1): Reject expressions
369 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
372 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
374 (cortex_a53_fdivd): Likewise.
376 2014-04-04 Martin Jambor <mjambor@suse.cz>
379 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
381 * cgraph.c (clone_of_p): Also return true if thunks match.
382 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
383 cgraph_function_or_thunk_node and an obsolete comment.
384 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
386 (build_function_decl_skip_args): Likewise.
387 (set_new_clone_decl_and_node_flags): New function.
388 (duplicate_thunk_for_node): Likewise.
389 (redirect_edge_duplicating_thunks): Likewise.
390 (cgraph_clone_node): New parameter args_to_skip, pass it to
391 redirect_edge_duplicating_thunks which is called instead of
392 cgraph_redirect_edge_callee.
393 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
394 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
396 2014-04-04 Jeff Law <law@redhat.com>
399 * config/arm/predicates.md (const_int_I_operand): New predicate.
400 (const_int_M_operand): Similarly.
401 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
403 (insv_t2, extv_reg, extzv_t2): Likewise.
404 (load_multiple_with_writeback): Similarly for const_int_I_operand.
405 (pop_multiple_with_writeback_and_return): Likewise.
406 (vfp_pop_multiple_with_writeback): Likewise
408 2014-04-04 Richard Biener <rguenther@suse.de>
411 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
412 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
414 * gimplify.h (gimple_add_tmp_var_fn): Declare.
415 * gimplify.c (gimple_add_tmp_var_fn): New function.
416 * gimple-expr.h (create_tmp_reg_fn): Declare.
417 * gimple-expr.c (create_tmp_reg_fn): New function.
418 * gimple-low.c (record_vars_into): Don't change cfun.
419 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
420 code generation without cfun.
422 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
425 * Makefile.in (install-driver): Fix shell scripting.
427 2014-04-03 Cong Hou <congh@google.com>
429 PR tree-optimization/60505
430 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
431 threshold of number of iterations below which no vectorization will be
433 * tree-vect-loop.c (new_loop_vec_info):
434 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
435 * tree-vect-loop.c (vect_analyze_loop_operations):
436 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
437 * tree-vect-loop.c (vect_transform_loop):
438 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
439 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
440 of iterations of the loop and see if we should build the epilogue.
442 2014-04-03 Richard Biener <rguenther@suse.de>
444 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
446 (streamer_tree_cache_create): Adjust.
447 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
448 to allow optional nodes array.
449 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
450 (streamer_tree_cache_append): Likewise.
451 (streamer_tree_cache_create): Create nodes array optionally
452 as specified by parameter.
453 * lto-streamer-out.c (create_output_block): Avoid maintaining
454 the node array in the writer cache.
455 (DFS_write_tree): Remove assertion.
456 (produce_asm_for_decls): Free the out decl state hash table
458 * lto-streamer-in.c (lto_data_in_create): Adjust for
459 streamer_tree_cache_create prototype change.
461 2014-04-03 Richard Biener <rguenther@suse.de>
463 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
464 set TREE_CHAIN to NULL_TREE.
466 2014-04-03 Richard Biener <rguenther@suse.de>
468 PR tree-optimization/60740
469 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
470 over all GIMPLE_COND operands.
472 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
474 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
475 (Weffc++): Remove Scott's numbering, merge lists and reference
481 * class.c (accessible_nvdtor_p): New.
482 (check_bases): Don't check base destructor here ...
483 (check_bases_and_members): ... check them here. Trigger on
484 Wnon-virtual-dtor flag.
485 (finish_struct_1): Use accessible_nvdtor_p.
488 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
489 * g++.dg/warn/Wnvdtor-2.C: New.
490 * g++.dg/warn/Wnvdtor-3.C: New.
491 * g++.dg/warn/Wnvdtor-4.C: New.
492 * g++.dg/warn/Weff1.C: Delete.
493 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
494 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
496 2014-04-03 Nick Clifton <nickc@redhat.com>
498 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
501 2014-04-03 Martin Jambor <mjambor@suse.cz>
503 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
504 mention gcc_unreachable before failing.
505 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
508 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
511 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
512 code and instead mark the context inconsistent.
513 (possible_polymorphic_call_targets): For inconsistent contexts
514 return empty complete list.
516 2014-04-02 Anthony Green <green@moxielogic.com>
518 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
519 (extendqisi2, extendhisi2): Define.
520 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
521 (WCHAR_TYPE): Change to unsigned int.
523 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
525 PR tree-optimization/60733
526 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
527 insertion point for PHI candidates to be the end of the feeding
528 block for the PHI argument.
530 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
532 PR rtl-optimization/60650
533 * lra-constraints.c (process_alt_operands): Decrease reject for
534 earlyclobber matching.
536 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
538 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
540 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
542 * config/spu/spu.c (pad_bb): Do not crash when the last
543 insn is CODE_FOR_blockage.
545 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
547 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
548 lies outside the target mode.
550 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
553 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
554 software floating point or no floating point registers, do not
555 allow any type in the FPRs. Eliminate a test for SPE SIMD types
556 in GPRs that occurs after we tested for GPRs that would never be
559 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
560 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
561 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
562 specifically allow DDmode, since that does not use the SPE SIMD
565 2014-04-02 Richard Biener <rguenther@suse.de>
568 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
569 MODE_INTs. Properly use negv_optab.
570 (expand_abs): Likewise.
572 2014-04-02 Richard Biener <rguenther@suse.de>
575 * Makefile.in (install-driver): Guard extra installs with special
578 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
580 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
583 2014-04-01 Richard Henderson <rth@redhat.com>
586 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
587 alternative enabled before register allocation.
589 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
591 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
592 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
594 (nios2_large_got_address): Remove unneeded 'sym' parameter.
595 (nios2_got_address): Update nios2_large_got_address call site.
596 (nios2_delegitimize_address): New function.
597 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
598 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
599 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
601 2014-04-01 Martin Husemann <martin@duskware.de>
603 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
606 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
608 PR rtl-optimization/60604
609 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
610 check from register_operand.
611 (register_operand): Redefine in terms of general_operand.
612 (nonmemory_operand): Use register_operand for the non-constant cases.
614 2014-04-01 Richard Biener <rguenther@suse.de>
616 * gimple.h (struct gimple_statement_base): Align subcode to
619 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
621 * doc/invoke.texi (mapp-regs): Clarify.
623 2014-03-31 Ulrich Drepper <drepper@gmail.com>
625 * config/i386/avx512fintrin.h (__v32hi): Define type.
627 (_mm512_set1_epi8): Define.
628 (_mm512_set1_epi16): Define.
629 (_mm512_set4_epi32): Define.
630 (_mm512_set4_epi64): Define.
631 (_mm512_set4_pd): Define.
632 (_mm512_set4_ps): Define.
633 (_mm512_setr4_epi64): Define.
634 (_mm512_setr4_epi32): Define.
635 (_mm512_setr4_pd): Define.
636 (_mm512_setr4_ps): Define.
637 (_mm512_setzero_epi32): Define.
639 2014-03-31 Martin Jambor <mjambor@suse.cz>
642 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
643 callsite_arguments_match_p. Updated all callers. Also check types of
644 corresponding formal parameters and actual arguments.
645 (not_all_callers_have_enough_arguments_p) Renamed to
646 some_callers_have_mismatched_arguments_p.
648 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
650 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
652 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
655 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
658 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
660 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
661 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
663 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
664 Use FMAMODE_NOVF512 mode iterator.
665 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
666 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
667 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
668 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
670 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
671 Use VF_128_256 mode iterator.
672 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
675 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
677 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
678 static chain if needed.
680 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
683 * lra-constraints.c (index_part_to_reg): New.
684 (process_address): Use it.
686 2014-03-27 Jeff Law <law@redhat.com>
687 Jakub Jelinek <jakub@redhat.com>
690 * expr.c (do_tablejump): Use simplify_gen_binary rather than
691 gen_rtx_{PLUS,MULT} to build up the address expression.
693 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
694 creating non-canonical RTL.
696 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
699 * ipa-inline.c (want_inline_small_function_p): Short circuit large
700 functions; reorganize to make cheap checks first.
701 (inline_small_functions): Do not estimate growth when dumping;
703 * ipa-inline.h (inline_summary): Add min_size.
704 (growth_likely_positive): New function.
705 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
706 (set_cond_stmt_execution_predicate): Cleanup.
707 (estimate_edge_size_and_time): Compute min_size.
708 (estimate_calls_size_and_time): Likewise.
709 (estimate_node_size_and_time): Likewise.
710 (inline_update_overall_summary): Update min_size.
711 (do_estimate_edge_time): Likewise.
712 (do_estimate_edge_size): Update.
713 (do_estimate_edge_hints): Update.
714 (growth_likely_positive): New function.
716 2014-03-28 Jakub Jelinek <jakub@redhat.com>
719 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
720 also if addr has VOIDmode.
722 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
724 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
725 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
727 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
728 instructions as well as AdvancedSIMD loads.
730 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
732 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
733 Use crypto_aese type.
734 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
735 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
736 crypto_aese, crypto_aesmc. Move to types.md.
737 * config/arm/types.md (crypto_aes): Split into crypto_aese,
739 * config/arm/iterators.md (crypto_type): Likewise.
741 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
743 * cgraph.c: Include expr.h and tree-dfa.h.
744 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
747 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
750 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
751 regs from checking multi-reg pseudos.
753 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
755 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
757 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
759 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
760 if it would clobber the stack pointer, even temporarily.
762 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
764 * mode-switching.c: Make small adjustments to the top comment.
766 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
768 * config/rs6000/constraints.md (wD constraint): New constraint to
769 match the constant integer to get the top DImode/DFmode out of a
770 vector in a VSX register.
772 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
773 match the constant integer to get the top DImode/DFmode out of a
774 vector in a VSX register.
776 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
779 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
782 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
783 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
785 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
786 Optimize vec_extract of 64-bit values, where the value being
787 extracted is in the top word, where we can use scalar
788 instructions. Add direct move and store support. Combine the big
789 endian/little endian vector select load support into a single insn.
790 (vsx_extract_<mode>_internal1): Likewise.
791 (vsx_extract_<mode>_internal2): Likewise.
792 (vsx_extract_<mode>_load): Likewise.
793 (vsx_extract_<mode>_store): Likewise.
794 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
795 combined into vsx_extract_<mode>_load.
796 (vsx_extract_<mode>_one_le): Likewise.
798 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
799 define the top 64-bit vector element.
801 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
804 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
805 Document vec_vbpermq builtin.
808 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
809 enable use of xxsldwi and xxpermdi builtin functions.
810 (vec_xxpermdi): Likewise.
812 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
813 Document use of vec_xxsldwi and vec_xxpermdi builtins.
815 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
817 PR rtl-optimization/60650
818 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
820 (find_spills_for): New.
821 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
822 Spill all pseudos on the second iteration.
824 2014-03-27 Marek Polacek <polacek@redhat.com>
827 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
830 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
832 * config/s390/s390.c (s390_can_use_return_insn): Check for
833 call-saved FPRs on 31 bit.
835 2014-03-27 Jakub Jelinek <jakub@redhat.com>
838 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
839 if they need regimplification, just drop them instead of
840 calling gimple_regimplify_operands on them.
842 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
845 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
846 (aarch64_frame_pointer_required): Adjust logic.
847 (aarch64_can_eliminate): Adjust logic.
848 (aarch64_override_options_after_change): Adjust logic.
850 2014-03-27 Dehao Chen <dehao@google.com>
852 * ipa-inline.c (early_inliner): Update node's inline info.
854 2014-03-26 Dehao Chen <dehao@google.com>
856 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
857 compiler inserted conditional jumps for NAN float check.
859 2014-03-26 Jakub Jelinek <jakub@redhat.com>
861 * ubsan.h (ubsan_create_data): Change second argument's type
862 to const location_t *.
863 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
865 (ubsan_create_data): Change second argument to const location_t *PLOC.
866 Create Loc field whenever PLOC is non-NULL.
867 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
868 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
872 * real.c (real_to_integer2): Change type of low to UHWI.
874 2014-03-26 Tobias Burnus <burnus@net-b.de>
876 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
877 (CILK_SELF_SPECS): New define.
878 (driver_self_specs): Use it.
880 2014-03-26 Richard Biener <rguenther@suse.de>
882 * tree-pretty-print.c (percent_K_format): Implement special
883 case for LTO and its stripped down BLOCK tree.
885 2014-03-26 Jakub Jelinek <jakub@redhat.com>
888 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
890 * tree-vrp.c (simplify_internal_call_using_ranges): If only
891 one range is range_int_cst_p, but not both, at least optimize
892 addition/subtraction of 0 and multiplication by 0 or 1.
893 * gimple-fold.c (gimple_fold_call): Fold
894 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
895 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
896 INTEGER_CSTs, try to fold at least x * 0 and y - y.
898 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
900 PR rtl-optimization/60452
901 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
902 <case REG>: Return 1 for invalid offsets from the frame pointer.
904 2014-03-26 Marek Polacek <polacek@redhat.com>
907 * doc/extend.texi (C Extensions): Mention variable-length arrays in
910 2014-03-26 Marek Polacek <polacek@redhat.com>
913 * doc/extend.texi (Designated Inits): Describe what happens to omitted
916 2014-03-26 Marek Polacek <polacek@redhat.com>
919 * ira-color.c (update_conflict_hard_regno_costs): Perform the
920 multiplication in unsigned type.
922 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
924 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
926 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
928 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
930 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
933 * cif-code.def (UNREACHABLE) New code.
934 * ipa-inline.c (inline_small_functions): Skip edges to
935 __builtlin_unreachable.
936 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
937 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
938 predicate to __bulitin_unreachable.
939 (set_cond_stmt_execution_predicate): Fix issue when
940 invert_tree_comparison returns ERROR_MARK.
941 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
942 propagate to inline clones.
943 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
945 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
946 * cgraphclones.c (cgraph_clone_node): If call destination is already
947 ureachable, do not redirect it back.
948 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
951 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
953 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
954 Do not modify inline clones.
956 2014-03-25 Jakub Jelinek <jakub@redhat.com>
958 * config/i386/i386.md (general_sext_operand): New mode attr.
959 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
960 don't generate (sign_extend (const_int)).
961 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
962 operands[2]. Use We constraint instead of <i> and
963 <general_sext_operand> predicate instead of <general_operand>.
964 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
965 * config/i386/constraints.md (We): New constraint.
966 * config/i386/predicates.md (x86_64_sext_operand,
967 sext_operand): New predicates.
969 2014-03-25 Martin Jambor <mjambor@suse.cz>
972 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
973 inconsistent devirtualizations to __builtin_unreachable.
975 2014-03-25 Marek Polacek <polacek@redhat.com>
978 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
980 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
982 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
983 order of elements for big-endian.
985 2014-03-25 Richard Biener <rguenther@suse.de>
988 * gimplify-me.c (gimple_regimplify_operands): Update the
991 2014-03-25 Martin Jambor <mjambor@suse.cz>
994 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
995 (lto_output_varpool_node): Likewise.
996 (input_overwrite_node): Likewise.
997 (input_varpool_node): Likewise.
999 2014-03-25 Richard Biener <rguenther@suse.de>
1001 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1002 (run_gcc): Likewise.
1004 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1006 * combine.c (simplify_compare_const): Add MODE argument.
1007 Handle mode_width 0 as very large mode_width.
1008 (try_combine, simplify_comparison): Adjust callers.
1010 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1011 type to avoid signed integer overflow.
1012 * explow.c (plus_constant): Likewise.
1014 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1016 * doc/generic.texi: Correct typos.
1018 2014-03-24 Tobias Burnus <burnus@net-b.de>
1020 * doc/invoke.texi (-flto): Expand section about
1021 using static libraries with LTO.
1023 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1025 PR rtl-optimization/60501
1026 * optabs.def (addptr3_optab): New optab.
1027 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1028 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1029 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1031 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1033 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1035 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1037 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1040 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1041 (_mm256_undefined_ps): Define.
1042 (_mm256_undefined_pd): Define.
1043 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1044 (_mm_undefined_pd): Define.
1045 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1046 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1047 (_mm512_undefined_ps): Define.
1048 (_mm512_undefined_pd): Define.
1049 Use _mm*_undefined_*.
1050 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1052 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1054 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1055 (lshr_simd): DI mode added.
1056 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1057 (aarch64_ushr_simddi): Likewise.
1058 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1059 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1060 (vshrd_n_u64): Likewise.
1062 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1064 * Makefile.in (s-macro_list): Depend on cc1.
1066 2014-03-23 Teresa Johnson <tejohnson@google.com>
1068 * ipa-utils.c (ipa_print_order): Use specified dump file.
1070 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1072 PR rtl-optimization/60601
1073 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1075 * gcc.c (eval_spec_function): Initialize save_growing_value.
1077 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1080 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1081 code == MINUS_EXPR, never swap op0 with op1.
1083 * toplev.c (init_local_tick): Avoid signed integer multiplication
1085 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1086 shift by first operand's bitsize.
1088 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1091 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1093 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1094 TARGET_ISA_64BIT_P(x).
1096 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1098 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1099 pattern for vector nor instead of subtract from splat(-1).
1100 (altivec_expand_vec_perm_const_le): Likewise.
1102 2014-03-21 Richard Henderson <rth@twiddle.net>
1105 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1106 related insns after epilogue_completed.
1108 2014-03-21 Martin Jambor <mjambor@suse.cz>
1111 * cgraph.h (symtab_node): New flag body_removed.
1112 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1113 when removing bodies.
1114 * symtab.c (dump_symtab_base): Dump body_removed flag.
1115 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1116 had their bodies removed.
1118 2014-03-21 Martin Jambor <mjambor@suse.cz>
1121 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1124 2014-03-21 Richard Biener <rguenther@suse.de>
1126 PR tree-optimization/60577
1127 * tree-core.h (struct tree_base): Document nothrow_flag use
1128 in VAR_DECL_NONALIASED.
1129 * tree.h (VAR_DECL_NONALIASED): New.
1130 (may_be_aliased): Adjust.
1131 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1133 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1135 * expr.c (expand_expr_real_1): Remove outdated comment.
1137 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1140 * ira.c (adjust_cleared_regs): Call copy_rtx on
1141 *reg_equiv[REGNO (loc)].src_p before passing it to
1142 simplify_replace_fn_rtx.
1145 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1146 into CONST, put pic register as first operand of PLUS. Use
1147 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1149 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1151 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1153 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1155 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1156 around for store forwarding issue in the FPU on the UT699.
1157 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1158 loads and operations if -mfix-ut699 is specified.
1159 (divtf3_hq): Tweak attribute.
1160 (sqrttf2_hq): Likewise.
1162 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1164 * calls.c (store_one_arg): Remove incorrect const qualification on the
1165 type of the temporary.
1166 * cfgexpand.c (expand_return): Likewise.
1167 * expr.c (expand_constructor): Likewise.
1168 (expand_expr_real_1): Likewise.
1170 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1172 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1175 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1178 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1180 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1182 * config/arm/aarch-common-protos.h
1183 (alu_cost_table): Fix spelling of "extend".
1184 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1186 2014-03-19 Richard Biener <rguenther@suse.de>
1189 * tree-core.h (tree_type_common): Re-order pointer members
1190 to reduce recursion depth during GC walks.
1192 2014-03-19 Marek Polacek <polacek@redhat.com>
1195 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1196 before accessing it.
1198 2014-03-19 Richard Biener <rguenther@suse.de>
1201 * lto-streamer-in.c (input_function): In WPA stage do not drop
1204 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1206 PR tree-optimization/60559
1207 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1208 with build_zero_cst assignment.
1210 2014-03-18 Kai Tietz <ktietz@redhat.com>
1212 PR rtl-optimization/56356
1213 * sdbout.c (sdbout_parms): Verify that parms'
1214 incoming argument is valid.
1215 (sdbout_reg_parms): Likewise.
1217 2014-03-18 Richard Henderson <rth@redhat.com>
1220 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1221 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1222 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1224 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1226 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1227 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1228 Italicize plugin event names in description. Explain that
1229 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1230 Remind that no GCC functions should be called after PLUGIN_FINISH.
1231 Explain what pragmas with expansion are.
1233 2014-03-18 Martin Liska <mliska@suse.cz>
1235 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1236 gimple call statement is update.
1237 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1238 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1240 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1243 * ubsan.c (ubsan_instrument_unreachable): Call
1244 initialize_sanitizer_builtins.
1245 (ubsan_pass): Likewise.
1248 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1249 varpool_finalize_decl instead of rest_of_decl_compilation.
1251 2014-03-18 Richard Biener <rguenther@suse.de>
1253 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1254 by using bitmap_and_compl instead of bitmap_and_compl_into.
1255 (df_rd_transfer_function): Likewise.
1257 2014-03-18 Richard Biener <rguenther@suse.de>
1259 * doc/lto.texi (fresolution): Fix typo.
1261 2014-03-18 Richard Biener <rguenther@suse.de>
1263 * doc/invoke.texi (flto): Update for changes in 4.9.
1265 2014-03-18 Richard Biener <rguenther@suse.de>
1267 * doc/loop.texi: Remove section on the removed lambda framework.
1268 Update loop docs with recent changes in preserving loop structure.
1270 2014-03-18 Richard Biener <rguenther@suse.de>
1272 * doc/lto.texi (-fresolution): Document.
1274 2014-03-18 Richard Biener <rguenther@suse.de>
1276 * doc/contrib.texi: Adjust my name.
1278 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1281 * internal-fn.c: Include diagnostic-core.h.
1282 (expand_BUILTIN_EXPECT): New function.
1283 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1284 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1285 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1286 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1288 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1289 Revert 3 argument __builtin_expect code.
1290 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1291 * gimple-fold.c (gimple_fold_call): Likewise.
1292 * tree.h (fold_builtin_expect): New prototype.
1293 * builtins.c (build_builtin_expect_predicate): Add predictor
1294 argument, if non-NULL, create 3 argument __builtin_expect.
1295 (fold_builtin_expect): No longer static. Add ARG2 argument,
1296 pass it through to build_builtin_expect_predicate.
1297 (fold_builtin_2): Adjust caller.
1298 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1299 * internal-fn.def (BUILTIN_EXPECT): New.
1301 2014-03-18 Tobias Burnus <burnus@net-b.de>
1304 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1305 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1306 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1308 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1311 * predict.c (combine_predictions_for_bb): Fix up formatting.
1312 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1313 fill what it points to if non-NULL.
1314 (tree_predict_by_opcode): Adjust caller, use the predictor.
1315 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1317 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1319 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1320 proper constant for the store mode.
1322 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1324 * symtab.c (change_decl_assembler_name): Fix transparent alias
1327 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1329 * config/aarch64/aarch64.c: Correct the comments about the
1330 aarch64 stack layout.
1332 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1334 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1335 check for GF_OMP_FOR_KIND_FOR.
1337 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1339 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1340 ymm and zmm register names.
1342 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1345 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1346 note creation for the 2010-08-31 changes.
1348 2014-03-17 Marek Polacek <polacek@redhat.com>
1351 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1352 as -fno-tree-loop-vectorize.
1353 (expand_omp_simd): Likewise.
1355 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1357 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1358 (eligible_for_call_delay): New prototype.
1359 * config/sparc/sparc.c (tls_call_delay): Rename into...
1360 (eligible_for_call_delay): ...this. Return false if the instruction
1361 cannot be put in the delay slot of a branch.
1362 (eligible_for_restore_insn): Simplify.
1363 (eligible_for_return_delay): Return false if the instruction cannot be
1364 put in the delay slot of a branch and simplify.
1365 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1366 put in the delay slot of a branch.
1367 * config/sparc/sparc.md (fix_ut699): New attribute.
1368 (tls_call_delay): Delete.
1369 (in_call_delay): Reimplement.
1370 (eligible_for_sibcall_delay): Rename into...
1371 (in_sibcall_delay): ...this.
1372 (eligible_for_return_delay): Rename into...
1373 (in_return_delay): ...this.
1374 (in_branch_delay): Reimplement.
1375 (in_uncond_branch_delay): Delete.
1376 (in_annul_branch_delay): Delete.
1378 2014-03-14 Richard Henderson <rth@redhat.com>
1381 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1382 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1383 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1384 (floathi splitters): Remove.
1385 (float<SWI48x>xf2): New pattern.
1386 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1387 code that tried to handle DImode for 32-bit, but which was excluded
1388 by the pattern's condition. Drop allocation of stack temporary.
1389 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1390 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1391 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1392 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1393 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1394 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1395 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1396 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1397 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1398 (*float<SWI48x><X87MODEF>2_i387): Remove.
1399 (all float _with_temp splitters): Remove.
1400 (*float<SWI48x><MODEF>2_i387): New pattern.
1401 (*float<SWI48><MODEF>2_sse): New pattern.
1402 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1403 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1405 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1406 Marek Polacek <polacek@redhat.com>
1409 * common.opt (dump_base_name_prefixed): New Variable.
1410 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1411 if x_dump_base_name_prefixed is already set, set it at the end.
1413 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1415 PR rtl-optimization/60508
1416 * lra-constraints.c (get_reload_reg): Add new parameter
1418 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1419 Pass the new parameter values.
1421 2014-03-14 Richard Biener <rguenther@suse.de>
1423 * common.opt: Revert unintented changes from r205065.
1426 2014-03-14 Richard Biener <rguenther@suse.de>
1429 * cfghooks.c (split_block): Properly adjust all loops the
1430 block was a latch of.
1432 2014-03-14 Martin Jambor <mjambor@suse.cz>
1435 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1438 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1441 * config/avr/avr.c (avr_set_current_function): Pass function name
1442 through default_strip_name_encoding before sanity checking instead
1443 of skipping the first char of the assembler name.
1445 2014-03-13 Richard Henderson <rth@redhat.com>
1448 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1449 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1450 * config/i386/i386-protos.h: Likewise.
1451 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1452 in the expander instead of a splitter.
1453 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1454 any possibility of requiring a memory.
1455 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1456 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1457 (fp branch splitters): Update for ix86_split_fp_branch.
1458 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1459 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1460 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1461 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1462 (*fop_<MODEF>_3_i387): Likewise.
1463 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1464 (splitters for the fop_* register patterns): Remove.
1465 (fscalexf4_i387): Rename from *fscalexf4_i387.
1466 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1468 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1470 PR tree-optimization/59779
1471 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1472 type for bitsize and maxsize instead of HOST_WIDE_INT.
1474 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1476 PR rtl-optimization/57320
1477 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1478 the CFG after thread_prologue_and_epilogue_insns.
1480 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1482 PR rtl-optimization/57189
1483 * lra-constraints.c (process_alt_operands): Disfavor spilling
1486 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1488 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1490 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1492 PR tree-optimization/59025
1494 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1495 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1497 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1500 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1501 calls of avr_out_plus_1.
1503 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1505 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1506 BB's single pred and update the father loop's latch info later.
1508 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1510 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1514 (VEC_base): Likewise.
1515 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1516 registers, we need to swap double words in little endian mode.
1518 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1519 to be a container mode for 128-bit integer operations added in ISA
1520 2.07. Unlike TImode and PTImode, the preferred register set is
1521 the Altivec/VMX registers for the 128-bit operations.
1523 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1525 (rs6000_split_128bit_ok_p): Likewise.
1527 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1528 macros for creating ISA 2.07 normal and overloaded builtin
1529 functions with 3 arguments.
1530 (BU_P8V_OVERLOAD_3): Likewise.
1531 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1532 for use as overloaded functions.
1533 (VPERM_1TI_UNS): Likewise.
1534 (VSEL_1TI): Likewise.
1535 (VSEL_1TI_UNS): Likewise.
1536 (ST_INTERNAL_1ti): Likewise.
1537 (LD_INTERNAL_1ti): Likewise.
1538 (XXSEL_1TI): Likewise.
1539 (XXSEL_1TI_UNS): Likewise.
1540 (VPERM_1TI): Likewise.
1541 (VPERM_1TI_UNS): Likewise.
1542 (XXPERMDI_1TI): Likewise.
1543 (SET_1TI): Likewise.
1544 (LXVD2X_V1TI): Likewise.
1545 (STXVD2X_V1TI): Likewise.
1546 (VEC_INIT_V1TI): Likewise.
1547 (VEC_SET_V1TI): Likewise.
1548 (VEC_EXT_V1TI): Likewise.
1549 (EQV_V1TI): Likewise.
1550 (NAND_V1TI): Likewise.
1551 (ORC_V1TI): Likewise.
1552 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1553 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1555 (VADDUQM): Likewise.
1556 (VSUBCUQ): Likewise.
1557 (VADDEUQM): Likewise.
1558 (VADDECUQ): Likewise.
1559 (VSUBEUQM): Likewise.
1560 (VSUBECUQ): Likewise.
1562 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1563 __int128_t and __uint128_t types.
1564 (__uint128_type): Likewise.
1565 (altivec_categorize_keyword): Add support for vector __int128_t,
1566 vector __uint128_t, vector __int128, and vector unsigned __int128
1567 as a container type for TImode operations that need to be done in
1568 VSX/Altivec registers.
1569 (rs6000_macro_to_expand): Likewise.
1570 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1571 to support 128-bit integer instructions vaddcuq, vadduqm,
1572 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1573 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1575 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1576 for V1TImode, and set up preferences to use VSX/Altivec registers.
1577 Setup VSX reload handlers.
1578 (rs6000_debug_reg_global): Likewise.
1579 (rs6000_init_hard_regno_mode_ok): Likewise.
1580 (rs6000_preferred_simd_mode): Likewise.
1581 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1582 (easy_altivec_constant): Likewise.
1583 (output_vec_const_move): Likewise.
1584 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1586 (rs6000_expand_vector_extract): Likewise.
1587 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1589 (rs6000_const_vec): Add support for V1TImode.
1590 (rs6000_emit_le_vsx_load): Swap double words when loading or
1591 storing TImode/V1TImode.
1592 (rs6000_emit_le_vsx_store): Likewise.
1593 (rs6000_emit_le_vsx_move): Likewise.
1594 (rs6000_emit_move): Add support for V1TImode.
1595 (altivec_expand_ld_builtin): Likewise.
1596 (altivec_expand_st_builtin): Likewise.
1597 (altivec_expand_vec_init_builtin): Likewise.
1598 (altivec_expand_builtin): Likewise.
1599 (rs6000_init_builtins): Add support for V1TImode type. Add
1600 support for ISA 2.07 128-bit integer builtins. Define type names
1601 for the VSX/Altivec vector types.
1602 (altivec_init_builtins): Add support for overloaded vector
1603 functions with V1TImode type.
1604 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1605 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1607 (rs6000_split_128bit_ok_p): Likewise.
1608 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1609 __int128_t and vector __uint128_t.
1611 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1612 and mode attributes.
1619 (VS_scalar): Likewise.
1620 (VS_double): Likewise.
1621 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1623 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1624 we support the ISA 2.07 128-bit integer arithmetic instructions.
1625 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1626 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1627 and TImode types for use with the builtin functions.
1628 (V1TI_type_node): Likewise.
1629 (unsigned_V1TI_type_node): Likewise.
1630 (intTI_type_internal_node): Likewise.
1631 (uintTI_type_internal_node): Likewise.
1633 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1634 128-bit builtin functions.
1635 (UNSPEC_VADDEUQM): Likewise.
1636 (UNSPEC_VADDECUQ): Likewise.
1637 (UNSPEC_VSUBCUQ): Likewise.
1638 (UNSPEC_VSUBEUQM): Likewise.
1639 (UNSPEC_VSUBECUQ): Likewise.
1640 (VM): Add V1TImode to vector mode iterators.
1642 (VI_unit): Likewise.
1643 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1644 (altivec_vaddcuq): Likewise.
1645 (altivec_vsubuqm): Likewise.
1646 (altivec_vsubcuq): Likewise.
1647 (altivec_vaddeuqm): Likewise.
1648 (altivec_vaddecuq): Likewise.
1649 (altivec_vsubeuqm): Likewise.
1650 (altivec_vsubecuq): Likewise.
1652 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1654 (BOOL_128): Likewise.
1655 (BOOL_REGS_OUTPUT): Likewise.
1656 (BOOL_REGS_OP1): Likewise.
1657 (BOOL_REGS_OP2): Likewise.
1658 (BOOL_REGS_UNARY): Likewise.
1659 (BOOL_REGS_AND_CR0): Likewise.
1661 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1662 128-bit integer builtin support.
1663 (vec_vadduqm): Likewise.
1664 (vec_vaddecuq): Likewise.
1665 (vec_vaddeuqm): Likewise.
1666 (vec_vsubecuq): Likewise.
1667 (vec_vsubeuqm): Likewise.
1668 (vec_vsubcuq): Likewise.
1669 (vec_vsubuqm): Likewise.
1671 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1672 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1673 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1674 128-bit integer add/subtract to ISA 2.07.
1676 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1678 * config/arc/arc.c (arc_predicate_delay_insns):
1679 Fix third argument passed to conditionalize_nonjump.
1681 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1683 * config/aarch64/aarch64-builtins.c
1684 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1685 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1686 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1687 instead of __builtin_lfloor.
1688 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1690 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1692 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1693 (tree_ssa_ifcombine_bb_1): New function.
1694 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1695 is an empty forwarder block to then_bb or vice versa and then_bb
1696 and else_bb are effectively swapped.
1698 2014-03-12 Christian Bruel <christian.bruel@st.com>
1701 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1702 REG_CFA_DEF_CFA note.
1703 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1704 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1706 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1708 PR tree-optimization/60454
1709 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1711 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1713 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1714 Do not define target_cpu_default2 to generic.
1715 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1716 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1717 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1719 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1720 Marc Glisse <marc.glisse@inria.fr>
1722 PR tree-optimization/60502
1723 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1724 instead of build_low_bits_mask.
1726 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1729 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1730 if there are multiple uses, but op doesn't live on E edge.
1731 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1732 clobber stmts before __builtin_unreachable.
1734 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1736 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1737 hard_frame_pointer_rtx.
1738 * cse.c (cse_insn): Remove volatile check.
1739 * cselib.c (cselib_process_insn): Likewise.
1740 * dse.c (scan_insn): Likewise.
1742 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1744 * config/arc/arc.c (conditionalize_nonjump): New function,
1746 (arc_ifcvt): ... this.
1747 (arc_predicate_delay_insns): Use it.
1749 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1751 * config/arc/predicates.md (extend_operand): During/after reload,
1752 allow const_int_operand.
1753 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1754 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1755 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1757 (umulsi3_highpart_i): Likewise.
1759 2014-03-11 Richard Biener <rguenther@suse.de>
1761 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1762 Add asserts to guard possible wrong-code bugs.
1764 2014-03-11 Richard Biener <rguenther@suse.de>
1766 PR tree-optimization/60429
1767 PR tree-optimization/60485
1768 * tree-ssa-structalias.c (set_union_with_increment): Properly
1769 take into account all fields that overlap the shifted vars.
1770 (do_sd_constraint): Likewise.
1771 (do_ds_constraint): Likewise.
1772 (get_constraint_for_ptr_offset): Likewise.
1774 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1776 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1777 (nios2_compute_frame_layout):
1778 Add calculation of cfun->machine->fp_save_offset.
1779 (nios2_expand_prologue): Correct setting of frame pointer register
1781 (nios2_expand_epilogue): Update recovery of stack pointer from
1782 frame pointer accordingly.
1783 (nios2_initial_elimination_offset): Update calculation of offset
1784 for eliminating to HARD_FRAME_POINTER_REGNUM.
1786 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1789 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1790 cgraph_get_create_node on VAR_DECLs.
1792 2014-03-10 Richard Biener <rguenther@suse.de>
1795 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1797 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1799 * config/vms/vms.opt (vms_float_format): New variable.
1801 2014-03-08 Tobias Burnus <burnus@net-b.de>
1803 * doc/invoke.texi (-fcilkplus): Update implementation status.
1805 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1806 Richard Biener <rguenther@suse.de>
1808 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1809 consistently accross all TUs.
1810 (run_gcc): Enable -fshort-double automatically at link at link-time
1811 and disallow override.
1813 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1816 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1817 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1818 if they can't be used.
1820 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1822 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1823 for Solaris 11/x86 ld.
1824 * configure: Regenerate.
1826 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1828 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1829 (LIB_TLS_SPEC): Save as ld_tls_libs.
1830 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1831 (HAVE_AS_IX86_TLSLDM): New test.
1832 * configure, config.in: Regenerate.
1833 * config/i386/i386.c (legitimize_tls_address): Fall back to
1834 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1835 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1836 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1837 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1839 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1841 * common.opt (fira-loop-pressure): Mark as optimization.
1843 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1845 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1846 an OpenMP mappable type.
1848 2014-03-06 Matthias Klose <doko@ubuntu.com>
1850 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1851 MULTILIB_OSDIRNAMES is not defined.
1853 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1854 Meador Inge <meadori@codesourcery.com>
1857 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1858 (arm_legitimize_address): Call legitimize_tls_address for any
1859 arm_tls_referenced_p expression, handle constant addend. Call it
1860 before testing for !TARGET_ARM.
1861 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1863 2014-03-06 Richard Biener <rguenther@suse.de>
1869 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1871 * tree-streamer.c (record_common_node): Assert we don't record
1872 nodes with type double.
1873 (preload_common_node): Skip type double, complex double and double
1874 pointer since it is now frontend dependent due to fshort-double option.
1876 2014-03-06 Richard Biener <rguenther@suse.de>
1878 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1879 or -fno-lto is specified and the linker has full plugin support.
1880 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1881 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1882 * lto-wrapper.c (merge_and_complain): Merge compile-time
1883 optimization levels.
1884 (run_gcc): And pass it through to the link options.
1886 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1890 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1892 * cselib.c (remove_useless_values): Skip to avoid quadratic
1893 behavior if the condition moved from...
1894 (cselib_process_insn): ... here holds.
1896 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1899 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1900 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1903 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1904 (TM_H): Add x86-tune.def.
1906 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1908 * config/aarch64/aarch64.c (generic_tunings):
1909 Use cortexa57_extra_costs.
1911 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1914 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1915 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1916 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1919 2014-03-04 Heiher <r@hev.cc>
1921 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1922 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1924 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1926 * config/i386/predicates.md (const2356_operand): Change to ...
1927 (const2367_operand): ... this.
1928 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1930 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1931 (*avx512pf_scatterpf<mode>sf): Ditto.
1932 (avx512pf_scatterpf<mode>df): Ditto.
1933 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1934 (*avx512pf_scatterpf<mode>df): Ditto.
1935 * config/i386/i386.c (ix86_expand_builtin): Update
1936 incorrect hint operand error message.
1938 2014-03-04 Richard Biener <rguenther@suse.de>
1940 * lto-section-in.c (lto_get_section_data): Fix const cast.
1942 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1944 * tree-streamer.c (record_common_node): Assert we don't record
1945 nodes with type double.
1946 (preload_common_node): Skip type double, complex double and double
1947 pointer since it is now frontend dependent due to fshort-double option.
1949 2014-03-04 Richard Biener <rguenther@suse.de>
1952 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1953 (lto_input_toplevel_asms): Likewise.
1954 * lto-section-in.c (lto_get_section_data): Instead do it here
1957 2014-03-04 Richard Biener <rguenther@suse.de>
1959 PR tree-optimization/60382
1960 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1961 dead PHIs a reduction.
1963 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1965 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1967 (_mm_prefetch): Move out of GCC target("sse") pragma.
1968 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1969 GCC target("prfchw") pragma.
1970 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1972 * config/i386/i386.c (ix86_option_override_internal): Enable
1973 -mprfchw with -mprefetchwt1.
1975 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1977 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1980 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1982 * opts.h (CL_PCH_IGNORE): Define.
1983 * targhooks.c (option_affects_pch_p):
1984 Return false for options that have CL_PCH_IGNORE set.
1985 * opt-functions.awk: Process PchIgnore.
1986 * doc/options.texi: Document PchIgnore.
1988 * config/arc/arc.opt (misize): Add PchIgnore property.
1990 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1992 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1993 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1994 constraint on constants to permit them being loaded into
1995 GENERAL_REGS or BASE_REGS.
1997 2014-03-03 Nick Clifton <nickc@redhat.com>
1999 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2000 anti-cacnonical alternatives.
2001 (negandhi3_real): New pattern.
2002 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2004 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2006 * config/avr/avr-mcus.def: Remove atxmega16x1.
2007 * config/avr/avr-tables.opt: Regenerate.
2008 * config/avr/t-multilib: Regenerate.
2009 * doc/avr-mmcu.texi: Regenerate.
2011 2014-03-03 Tobias Grosser <tobias@grosser.es>
2012 Mircea Namolaru <mircea.namolaru@inria.fr>
2014 PR tree-optimization/58028
2015 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2018 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2020 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2021 not handled by recognizers.
2023 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2026 * function.c (expand_function_end): Don't emit
2027 clobber_return_register sequence if clobber_after is a BARRIER.
2028 * cfgexpand.c (construct_exit_block): Append instructions before
2029 return_label to prev_bb.
2031 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2033 * config/rs6000/constraints.md: Document reserved use of "wc".
2035 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2038 * ipa.c (function_and_variable_visibility): When dissolving comdat
2039 group, also set all symbols to local.
2041 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2046 2013-12-14 Jan Hubicka <jh@suse.cz>
2048 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2050 2014-03-02 Jon Beniston <jon@beniston.com>
2056 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2057 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2058 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2059 (simple_return, *simple_return): New patterns
2060 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2061 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2063 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2065 * dwarf2out.c (gen_subprogram_die): Tidy.
2067 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2070 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2071 (*mov_t_msb_neg_negc): ... this new insn.
2073 2014-02-28 Jason Merrill <jason@redhat.com>
2076 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2079 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2082 * dwarf2out.c (decltype_auto_die): New static.
2083 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2084 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2085 (is_cxx_auto): Likewise.
2087 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2089 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2090 we are not using general regs only.
2092 2014-02-28 Richard Biener <rguenther@suse.de>
2095 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2096 previous fix and only allow to remove trivial pre-headers
2097 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2098 (remove_forwarder_block): Properly update the latch of a loop.
2100 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2103 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2104 (cselib_preserved_hash_table): New.
2105 (preserve_constants_and_equivs): Move preserved vals to it.
2106 (cselib_find_slot): Look it up first.
2107 (cselib_init): Initialize it.
2108 (cselib_finish): Release it.
2109 (dump_cselib_table): Dump it.
2111 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2114 * cselib.c (remove_useless_values): Skip to avoid quadratic
2115 behavior if the condition moved from...
2116 (cselib_process_insn): ... here holds.
2118 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2121 * var-tracking.c (vt_initialize): Apply the same condition to
2122 preserve the CFA base value.
2124 2014-02-28 Joey Ye <joey.ye@arm.com>
2127 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2128 if reload in progress or completed.
2130 2014-02-28 Tobias Burnus <burnus@net-b.de>
2133 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2136 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2138 * doc/tm.texi.in (Condition Code Status): Update documention for
2139 relative locations of cc0-setter and cc0-user.
2141 2014-02-27 Jeff Law <law@redhat.com>
2143 PR rtl-optimization/52714
2144 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2145 into two independent simple sets, if I3 is a jump, ensure the
2146 pattern we place into I3 is a (set (pc) ...).
2148 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2149 Jeff Law <law@redhat.com>
2151 PR rtl-optimization/49847
2152 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2153 are in different blocks.
2154 * doc/tm.texi (Condition Code Status): Update documention for
2155 relative locations of cc0-setter and cc0-user.
2157 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2160 * lra.c (lra_emit_add): Check SUBREG too.
2162 2014-02-27 Andreas Schwab <schwab@suse.de>
2164 * config/m68k/m68k.c (m68k_option_override): Disable
2165 -flive-range-shrinkage for classic m68k.
2166 (m68k_override_options_after_change): Likewise.
2168 2014-02-27 Marek Polacek <polacek@redhat.com>
2171 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2172 -Wmaybe-uninitialized.
2174 2014-02-27 Alan Modra <amodra@gmail.com>
2177 * reload1.c (emit_input_reload_insns): When reload_override_in,
2178 set old to rl->in_reg when rl->in_reg is a subreg.
2180 2014-02-26 Richard Biener <rguenther@suse.de>
2183 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2185 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2187 * common/config/i386/predicates.md (const1256_operand): Remove.
2188 (const2356_operand): New.
2189 (const_1_to_2_operand): Remove.
2190 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2191 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2192 (*avx512pf_gatherpf<mode>sf): Ditto.
2193 (avx512pf_gatherpf<mode>df): Ditto.
2194 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2195 (*avx512pf_gatherpf<mode>df): Ditto.
2196 (avx512pf_scatterpf<mode>sf): Ditto.
2197 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2198 (*avx512pf_scatterpf<mode>sf): Ditto.
2199 (avx512pf_scatterpf<mode>df): Ditto.
2200 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2201 (*avx512pf_scatterpf<mode>df): Ditto.
2202 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2204 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2206 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2207 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2208 (_mm512_mask_testn_epi64_mask): Move to ...
2209 * config/i386/avx512cdintrin.h: Here.
2210 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2211 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2212 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2213 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2214 TARGET_AVX512F from TARGET_AVX512CD.
2216 2014-02-26 Richard Biener <rguenther@suse.de>
2219 * ipa.c (walk_polymorphic_call_targets): Properly guard
2220 call to inline_update_overall_summary.
2222 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2225 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2226 and latches only if requested. Fix latch if it is removed.
2227 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2228 LOOPS_HAVE_PREHEADERS.
2230 2014-02-25 Andrew Pinski <apinski@cavium.com>
2232 * builtins.c (expand_builtin_thread_pointer): Create a new target
2233 when the target is NULL.
2235 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2237 PR rtl-optimization/60317
2238 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2239 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2240 * lra-assigns.c: Include params.h.
2241 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2242 other reload pseudos considerations.
2244 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2246 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2247 to use canonical form for nor<mode>3.
2249 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2252 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2255 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2258 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2259 (ix86_handle_option): Handle OPT_mprefetchwt1.
2260 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2261 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2263 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2264 OPTION_MASK_ISA_PREFETCHWT1.
2265 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2266 (PTA_PREFETCHWT1): New.
2267 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2268 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2269 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2270 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2271 (*prefetch_avx512pf_<mode>_: Change into ...
2272 (*prefetch_prefetchwt1_<mode>: This.
2273 * config/i386/i386.opt (mprefetchwt1): New.
2274 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2275 (_mm_prefetch): Handle intent to write.
2276 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2278 2014-02-25 Richard Biener <rguenther@suse.de>
2281 * emit-rtl.c (mem_attrs_htab): Remove.
2282 (mem_attrs_htab_hash): Likewise.
2283 (mem_attrs_htab_eq): Likewise.
2284 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2285 (init_emit_once): Do not allocate mem_attrs_htab.
2287 2014-02-25 Richard Biener <rguenther@suse.de>
2290 * lto-opts.c (lto_write_options): Output non-explicit conservative
2291 -fwrapv, -fno-trapv and -fno-strict-overflow.
2292 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2293 (run_gcc): And pass them through.
2295 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2297 * sel-sched.c (calculate_new_fences): New parameter ptime.
2298 Calculate it as a maximum over all fence cycles.
2299 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2300 Print the final schedule timing when sched_verbose.
2302 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2304 PR rtl-optimization/60292
2305 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2306 bit fot the fence instruction.
2308 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2310 * calls.h: Fix typo in comment.
2312 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2314 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2315 adjusting offsetable addresses.
2317 2014-02-24 Guozhi Wei <carrot@google.com>
2319 * sparseset.h (sparseset_pop): Fix the wrong index.
2321 2014-02-24 Walter Lee <walt@tilera.com>
2323 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2324 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2326 * common/config/tilegx/tilegx-common.c
2327 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2328 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2330 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2331 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2332 (tilegx_gimplify_va_arg_expr): Handle big endian.
2333 (tilegx_expand_unaligned_load): Ditto.
2334 (tilegx_expand_unaligned_store): Ditto.
2335 (TARGET_RETURN_IN_MSB): New.
2336 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2337 (TARGET_ENDIAN_DEFAULT): New.
2338 (TARGET_BIG_ENDIAN): Handle big endian.
2339 (BYTES_BIG_ENDIAN): Ditto.
2340 (WORDS_BIG_ENDIAN): Ditto.
2341 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2344 * config/tilegx/tilegx.md (extv): Handle big endian.
2346 (insn_st<n>): Ditto.
2347 (insn_st<n>_add<bitsuffix>): Ditto.
2348 (insn_stnt<n>): Ditto.
2349 (insn_stnt<n>_add<bitsuffix>):Ditto.
2350 (vec_interleave_highv8qi): Handle big endian.
2351 (vec_interleave_highv8qi_be): New.
2352 (vec_interleave_highv8qi_le): New.
2353 (insn_v1int_h): Handle big endian.
2354 (vec_interleave_lowv8qi): Handle big endian.
2355 (vec_interleave_lowv8qi_be): New.
2356 (vec_interleave_lowv8qi_le): New.
2357 (insn_v1int_l): Handle big endian.
2358 (vec_interleave_highv4hi): Handle big endian.
2359 (vec_interleave_highv4hi_be): New.
2360 (vec_interleave_highv4hi_le): New.
2361 (insn_v2int_h): Handle big endian.
2362 (vec_interleave_lowv4hi): Handle big endian.
2363 (vec_interleave_lowv4hi_be): New.
2364 (vec_interleave_lowv4hi_le): New.
2365 (insn_v2int_l): Handle big endian.
2366 (vec_interleave_highv2si): Handle big endian.
2367 (vec_interleave_highv2si_be): New.
2368 (vec_interleave_highv2si_le): New.
2369 (insn_v4int_h): Handle big endian.
2370 (vec_interleave_lowv2si): Handle big endian.
2371 (vec_interleave_lowv2si_be): New.
2372 (vec_interleave_lowv2si_le): New.
2373 (insn_v4int_l): Handle big endian.
2374 * config/tilegx/tilegx.opt (mbig-endian): New option.
2375 (mlittle-endian): New option.
2376 * doc/install.texi: Document tilegxbe-linux.
2377 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2379 2014-02-24 Martin Jambor <mjambor@suse.cz>
2382 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2383 there are no parameter descriptors.
2385 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2387 PR rtl-optimization/60268
2388 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2389 initialization to ...
2390 (sched_rgn_init): ... here.
2391 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2393 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2395 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2398 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2400 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2403 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2405 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2406 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2408 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2410 * config/microblaze/predicates.md: Add cmp_op predicate.
2411 * config/microblaze/microblaze.md: Add branch_compare instruction
2412 which uses cmp_op predicate and emits cmp insn before branch.
2413 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2414 to microblaze_expand_conditional_branch and consolidate logic.
2415 (microblaze_expand_conditional_branch): emit branch_compare
2416 insn instead of handling cmp op separate from branch insn.
2418 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2420 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2423 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2425 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2426 define_insn with define_expand and new define_insn
2427 *altivec_lve<VI_char>x_internal.
2428 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2429 and new define_insn *altivec_stve<VI_char>x_internal.
2430 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2432 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2434 (altivec_expand_stvex_be): New function.
2436 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2438 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2439 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2440 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2441 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2443 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2446 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2447 instead of emit_move_insn.
2449 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2451 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2453 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2454 gen_altivec_vsumsws.
2456 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2458 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2459 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2460 (altivec_lvxl_<mode>): New define_expand incorporating
2461 -maltivec=be semantics where needed.
2462 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2463 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2464 semantics where needed.
2465 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2466 (altivec_stvx_<mode>): New define_expand incorporating
2467 -maltivec=be semantics where needed.
2468 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2469 VM2 iterator instead of V4SI.
2470 (altivec_stvxl_<mode>): New define_expand incorporating
2471 -maltivec=be semantics where needed.
2472 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2473 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2474 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2475 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2476 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2477 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2478 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2479 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2480 ALTIVEC_BUILTIN_STVXL.
2481 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2482 (altivec_expand_stvx_be): Likewise.
2483 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2484 (altivec_expand_lvx_be): Likewise.
2485 (altivec_expand_stvx_be): Likewise.
2486 (altivec_expand_builtin): Add cases for
2487 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2488 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2489 (altivec_init_builtins): Add definitions for
2490 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2491 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2493 2014-02-21 Catherine Moore <clm@codesourcery.com>
2495 * doc/invoke.texi (mvirt, mno-virt): Document.
2496 * config/mips/mips.opt (mvirt): New option.
2497 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2499 2014-02-21 Richard Biener <rguenther@suse.de>
2501 PR tree-optimization/60276
2502 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2503 (STMT_VINFO_MIN_NEG_DIST): New macro.
2504 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2505 STMT_VINFO_MIN_NEG_DIST.
2506 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2507 made for negative dependence distances still hold.
2509 2014-02-21 Richard Biener <rguenther@suse.de>
2512 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2513 DECL_INITIAL for globals not in the current function context.
2515 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2517 PR tree-optimization/56490
2518 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2519 * tree-ssa-uninit.c: Include params.h.
2520 (compute_control_dep_chain): Add num_calls argument, return false
2521 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2522 num_calls to recursive call.
2523 (find_predicates): Change dep_chain into normal array,
2524 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2525 variable and adjust compute_control_dep_chain caller.
2526 (find_def_preds): Likewise.
2528 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2530 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2531 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2533 2014-02-21 Nick Clifton <nickc@redhat.com>
2535 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2536 (pushhi1): Likewise.
2537 (popqi1): Add mode to pre_dec.
2540 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2542 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2543 mode for mask of V8SFmode permutation.
2545 2014-02-20 Richard Henderson <rth@redhat.com>
2548 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2549 a new pseudo for OLDVAL.
2551 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2554 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2555 gen_reg_rtx if d->testing_p.
2556 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2557 if d->testing_p and we will certainly return true.
2558 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2561 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2563 * emit-rtl.c (gen_reg_rtx): Assert that
2564 crtl->emit.regno_pointer_align_length is non-zero.
2566 2014-02-20 Richard Henderson <rth@redhat.com>
2569 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2570 on failure the store back into EXPECT.
2572 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2573 Sandra Loosemore <sandra@codesourcery.com>
2575 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2576 * config/nios2/nios2.c (nios2_function_profiler): Add
2577 -fPIC (flag_pic == 2) support.
2578 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2579 (nios2_large_offset_p): New function.
2580 (nios2_unspec_reloc_p): Move up position, update to use
2581 nios2_large_offset_p.
2582 (nios2_unspec_address): Remove function.
2583 (nios2_unspec_offset): New function.
2584 (nios2_large_got_address): New function.
2585 (nios2_got_address): Add large offset support.
2586 (nios2_legitimize_tls_address): Update usage of removed and new
2588 (nios2_symbol_binds_local_p): New function.
2589 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2590 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2591 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2592 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2593 processing for (const (unspec ...)).
2594 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2596 2014-02-20 Richard Biener <rguenther@suse.de>
2598 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2599 doing the substitution.
2600 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2602 2014-02-20 Martin Jambor <mjambor@suse.cz>
2605 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2606 info when checking whether lattices are bottom.
2608 2014-02-20 Richard Biener <rguenther@suse.de>
2611 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2614 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2617 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2618 parameter specifying the scaling.
2619 (inline_call): Update.
2620 (want_inline_recursively): Guard division by zero.
2621 (recursive_inlining): Update.
2622 * ipa-inline.h (clone_inlined_nodes): Update.
2624 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2627 * config/i386/i386.c (classify_argument): Pass structures of size
2628 64 bytes or less in register.
2630 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2631 Kirill Yukhin <kirill.yukhin@intel.com>
2633 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2634 (_mm_rcp28_round_ss): Ditto.
2635 (_mm_rsqrt28_round_sd): Ditto.
2636 (_mm_rsqrt28_round_ss): Ditto.
2637 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2638 (_mm_rcp14_round_ss): Ditto.
2639 (_mm_rsqrt14_round_sd): Ditto.
2640 (_mm_rsqrt14_round_ss): Ditto.
2641 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2642 the first input operand, get rid of match_dup.
2643 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2645 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2647 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2648 operand as the first input operand, set type attribute.
2649 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2651 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2652 operand as the first input operand, set type attribute.
2654 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2656 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2659 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2662 * config/i386/i386.c (construct_container): Remove TFmode check
2663 for X86_64_INTEGER_CLASS.
2665 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2668 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2669 only when -Wpsabi is enabled.
2671 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2674 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2675 passing arrays in registers are the same as for structs, so remove the
2676 special case for them.
2678 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2680 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2681 destination type, extract only the valid bits if the source type is not
2682 integral and has a different mode.
2684 2014-02-19 Richard Biener <rguenther@suse.de>
2687 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2690 2014-02-19 Richard Biener <rguenther@suse.de>
2693 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2694 (ipa_modify_call_arguments): Emit an argument load explicitely and
2695 preserve virtual SSA form there and for the replacement call.
2696 Do not update SSA form nor free dominance info.
2698 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2700 * ipa.c (function_and_variable_visibility): Also clear WEAK
2701 flag when disolving COMDAT_GROUP.
2703 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2705 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2706 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2708 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2709 do more sanity checks.
2710 (detect_type_change): Return true when giving up early.
2711 (compute_complex_assign_jump_func): Fix type parameter of
2712 ipa_set_ancestor_jf.
2713 (compute_complex_ancestor_jump_func): Likewise.
2714 (update_jump_functions_after_inlining): Fix updating of
2716 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2718 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2720 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2721 inline clones when edge disappears.
2723 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2726 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2727 Split 64-bit moves into 2 patterns. Do not allow the use of
2728 direct move for TDmode in little endian, since the decimal value
2729 has little endian bytes within a word, but the 64-bit pieces are
2730 ordered in a big endian fashion, and normal subreg's of TDmode are
2732 (mov<mode>_64bit_dm): Likewise.
2733 (movtd_64bit_nodm): Likewise.
2735 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2737 PR tree-optimization/60174
2738 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2739 statement of an SSA_NAME that occurs in an abnormal PHI node.
2741 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2744 * final.c (SEEN_BB): Remove.
2745 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2746 (final_scan_insn): Don't force_source_line on second
2747 NOTE_INSN_BASIC_BLOCK.
2749 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2752 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2753 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2754 (type_natural_mode): Warn ABI change when %zmm register is not
2755 available for AVX512F vector value passing.
2757 2014-02-18 Kai Tietz <ktietz@redhat.com>
2760 * config/i386/i386.c (ix86_expand_prologue): Use value in
2761 rax register as displacement when restoring %r10 or %rax.
2762 Fix wrong offset when restoring both registers.
2764 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2766 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2767 assertion with conditional return.
2769 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2770 Uros Bizjak <ubizjak@gmail.com>
2773 * config/i386/driver-i386.c (host_detect_local_cpu): If
2774 YMM state is not saved by the OS, also clear has_f16c. Move
2775 CPUID 0x80000001 handling before YMM state saving checking.
2777 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2779 PR rtl-optimization/58960
2780 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2781 factored out from ...
2782 (sched_init): ... here.
2783 (free_global_sched_pressure_data): New, factored out from ...
2784 (sched_finish): ... here.
2785 * sched-int.h (free_global_sched_pressure_data): Declare.
2786 * sched-rgn.c (nr_regions_initial): New static global.
2787 (haifa_find_rgns): Initialize it.
2788 (schedule_region): Disable sched-pressure for the newly
2791 2014-02-17 Richard Biener <rguenther@suse.de>
2793 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2794 release SSA defs of pattern stmts.
2796 2014-02-17 Richard Biener <rguenther@suse.de>
2798 * tree-inline.c (expand_call_inline): Release the virtual
2799 operand defined by the call we are about to inline.
2801 2014-02-17 Richard Biener <rguenther@suse.de>
2803 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2805 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2806 Ilya Tocar <ilya.tocar@intel.com>
2808 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2809 arguments order in builtin.
2810 (_mm512_permutexvar_epi64): Ditto.
2811 (_mm512_mask_permutexvar_epi64): Ditto
2812 (_mm512_maskz_permutexvar_epi32): Ditto
2813 (_mm512_permutexvar_epi32): Ditto
2814 (_mm512_mask_permutexvar_epi32): Ditto
2816 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2818 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2819 (p8_vmrgow): Likewise.
2821 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2823 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2826 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2829 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2830 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2831 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2832 using direct move instructions on ISA 2.07. Also adjust
2833 instruction length for 64-bit.
2834 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2835 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2837 2014-02-15 Alan Modra <amodra@gmail.com>
2841 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2842 find_replacement on parts of insn rtl that might be reloaded.
2844 2014-02-15 Richard Biener <rguenther@suse.de>
2846 PR tree-optimization/60183
2847 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2848 (tree_ssa_phiprop): Calculate and free post-dominators.
2850 2014-02-14 Jeff Law <law@redhat.com>
2852 PR rtl-optimization/60131
2853 * ree.c (get_extended_src_reg): New function.
2854 (combine_reaching_defs): Use it rather than assuming location of REG.
2855 (find_and_remove_re): Verify first operand of extension is
2856 a REG before adding the insns to the copy list.
2858 2014-02-14 Roland McGrath <mcgrathr@google.com>
2860 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2861 * configure: Regenerated.
2862 * config.in: Regenerated.
2863 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2864 instead of ASM_SHORT.
2866 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2867 Richard Earnshaw <rearnsha@arm.com>
2869 PR rtl-optimization/59535
2870 * lra-constraints.c (process_alt_operands): Encourage alternative
2871 when unassigned pseudo class is superset of the alternative class.
2872 (inherit_reload_reg): Don't inherit when optimizing for code size.
2873 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2874 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2875 modes not less than 4 for Thumb1.
2877 2014-02-14 Kyle McMartin <kyle@redhat.com>
2880 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2882 2014-02-14 Richard Biener <rguenther@suse.de>
2884 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2885 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2886 Do not strip INDIRECT_REFs.
2888 2014-02-14 Richard Biener <rguenther@suse.de>
2891 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2892 DECL_FUNCTION_SPECIFIC_TARGET.
2893 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2894 * tree-streamer-out.c (pack_ts_target_option): Remove.
2895 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2896 (write_ts_function_decl_tree_pointers): Do not stream
2897 DECL_FUNCTION_SPECIFIC_TARGET.
2898 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2899 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2900 (lto_input_ts_function_decl_tree_pointers): Do not stream
2901 DECL_FUNCTION_SPECIFIC_TARGET.
2903 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2905 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2906 (get_initial_def_for_induction, vectorizable_induction): Ignore
2907 debug stmts when looking for exit_phi.
2908 (vectorizable_live_operation): Fix up condition.
2910 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2912 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2913 nreverse() because it changes the content of original tree list.
2915 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2917 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2918 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2920 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2922 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2923 GNU coding standards.
2925 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2928 * dwarf2out.c (gen_subprogram_die): Don't call
2929 add_calling_convention_attribute if subr_die is old_die.
2931 2014-02-13 Sharad Singhai <singhai@google.com>
2933 * doc/optinfo.texi: Fix order of nodes.
2935 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2937 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2938 operands[2], not operands[3].
2940 2014-02-13 Richard Biener <rguenther@suse.de>
2943 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2944 mention the possibility of an in-tree build.
2945 (CLooG): Update recommended version to 0.18.1, mention the
2946 possibility of an in-tree build and clarify that the ISL
2947 bundled with CLooG does not work.
2949 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2952 * expr.c (compress_float_constant): If x is a hard register,
2953 extend into a pseudo and then move to x.
2955 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2957 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2958 caused by bad second argument to warning_at() with -mhotpatch and
2959 nested functions (e.g. with gfortran).
2961 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2963 * opts.c (option_name): Remove "enabled by default" rider.
2965 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2967 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2969 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2970 Uros Bizjak <ubizjak@gmail.com>
2973 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2974 * configure: Regenerated.
2976 2014-02-12 Richard Biener <rguenther@suse.de>
2978 * vec.c (vec_prefix::calculate_allocation): Move as
2979 inline variant to vec.h.
2980 (vec_prefix::calculate_allocation_1): New out-of-line version.
2981 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2982 (vec_prefix::m_has_auto_buf): Rename to ...
2983 (vec_prefix::m_using_auto_storage): ... this.
2984 (vec_prefix::calculate_allocation): Inline the easy cases
2985 and dispatch to calculate_allocation_1 which doesn't need the
2987 (va_heap::reserve): Use gcc_checking_assert.
2988 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2989 m_using_auto_storage.
2990 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2992 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2993 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2994 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2996 2014-02-12 Richard Biener <rguenther@suse.de>
2998 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2999 when we found a dependence.
3001 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3003 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3005 (maybe_fold_stmt): ... into this new function.
3006 * omp-low.c (lower_omp): Update comment.
3008 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3011 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3014 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3016 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3017 identifiers in comments.
3018 (cortexa53_extra_costs): Likewise.
3019 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3020 (cortexa7_extra_costs): Likewise.
3021 (cortexa12_extra_costs): Likewise.
3022 (cortexa15_extra_costs): Likewise.
3023 (v7m_extra_costs): Likewise.
3025 2014-02-12 Richard Biener <rguenther@suse.de>
3028 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3029 of posix_memalign being successful.
3030 (lower_stmt): Restrict lowering of posix_memalign to when
3031 -ftree-bit-ccp is enabled.
3033 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3035 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3037 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3039 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3041 PR rtl-optimization/60116
3042 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3043 other_insn once the combination has been validated.
3045 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3048 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3050 * ipa-devirt.c: Include demangle.h
3051 (odr_violation_reported): New static variable.
3052 (add_type_duplicate): Update odr_violations.
3053 (maybe_record_node): Add completep parameter; update it.
3054 (record_target_from_binfo): Add COMPLETEP parameter;
3055 update it as needed.
3056 (possible_polymorphic_call_targets_1): Likewise.
3057 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3058 rename FINAL to COMPLETE.
3059 (record_targets_from_bases): Sanity check we found the binfo;
3060 fix COMPLETEP updating.
3061 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3062 parameter, fix computing of COMPLETEP.
3063 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3064 at LTO time do demangling.
3065 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3066 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3068 (gimple_get_virt_method_for_binfo): Likewise.
3069 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3070 gimple_get_virt_method_for_vtable): Update prototypes.
3072 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3075 * genautomata.c (add_presence_absence): Fix typo with
3076 {final_}presence_list.
3078 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3081 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3082 for VSX/Altivec vectors that land in GPR registers.
3084 2014-02-11 Richard Henderson <rth@redhat.com>
3085 Jakub Jelinek <jakub@redhat.com>
3088 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3089 around drhs if type conversion to lacc->type is not useless.
3091 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3093 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3095 (cortex-a57.cortex-a53): Likewise.
3096 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3098 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3100 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3103 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3105 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3106 add_options_for_arm_vfp3.
3108 2014-02-11 Jeff Law <law@redhat.com>
3111 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3112 object with an undesirable mode.
3114 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3117 * config/i386/sol2-9.h: New file.
3118 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3119 *-*-solaris2.9*): Use it.
3121 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3123 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3124 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3126 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3128 * config/microblaze/microblaze.c: Extend mcpu version format
3130 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3132 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3134 2014-02-10 Richard Henderson <rth@redhat.com>
3137 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3138 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3139 ms-abi vs -mno-accumulate-outgoing-args.
3140 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3141 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3144 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3147 * cfgexpand.c (expand_asm_operands): Attach source location to
3148 ASM_INPUT rtx objects.
3149 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3151 2014-02-10 Nick Clifton <nickc@redhat.com>
3153 * config/mn10300/mn10300.c (popcount): New function.
3154 (mn10300_expand_prologue): Include saved registers in stack usage
3157 2014-02-10 Jeff Law <law@redhat.com>
3160 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3161 when changing the SET_DEST of a prior insn to avoid an input reload.
3163 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3165 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3166 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3167 -mcall-openbsd, or -mcall-linux.
3168 (CC1_ENDIAN_BIG_SPEC): Remove.
3169 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3170 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3171 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3172 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3173 and %cc1_endian_default.
3174 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3176 2014-02-10 Richard Biener <rguenther@suse.de>
3178 PR tree-optimization/60115
3179 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3180 MEM_REF handling. Properly verify that the accesses are not
3181 out of the objects bound.
3183 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3185 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3188 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3190 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3191 proper constants and fix formatting.
3192 (possible_polymorphic_call_targets): Fix formatting.
3194 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3195 Ilya Tocar <ilya.tocar@intel.com>
3197 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3198 (_mm512_loadu_epi32): Renamed into...
3199 (_mm512_loadu_si512): This.
3200 (_mm512_storeu_epi32): Renamed into...
3201 (_mm512_storeu_si512): This.
3202 (_mm512_maskz_ceil_ps): Removed.
3203 (_mm512_maskz_ceil_pd): Ditto.
3204 (_mm512_maskz_floor_ps): Ditto.
3205 (_mm512_maskz_floor_pd): Ditto.
3206 (_mm512_floor_round_ps): Ditto.
3207 (_mm512_floor_round_pd): Ditto.
3208 (_mm512_ceil_round_ps): Ditto.
3209 (_mm512_ceil_round_pd): Ditto.
3210 (_mm512_mask_floor_round_ps): Ditto.
3211 (_mm512_mask_floor_round_pd): Ditto.
3212 (_mm512_mask_ceil_round_ps): Ditto.
3213 (_mm512_mask_ceil_round_pd): Ditto.
3214 (_mm512_maskz_floor_round_ps): Ditto.
3215 (_mm512_maskz_floor_round_pd): Ditto.
3216 (_mm512_maskz_ceil_round_ps): Ditto.
3217 (_mm512_maskz_ceil_round_pd): Ditto.
3218 (_mm512_expand_pd): Ditto.
3219 (_mm512_expand_ps): Ditto.
3220 * config/i386/i386.c (ix86_builtins): Remove
3221 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3222 (bdesc_args): Ditto.
3223 * config/i386/predicates.md (const1256_operand): New.
3224 (const_1_to_2_operand): Ditto.
3225 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3226 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3227 (*avx512pf_gatherpf<mode>sf): Ditto.
3228 (avx512pf_gatherpf<mode>df): Ditto.
3229 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3230 (*avx512pf_gatherpf<mode>df): Ditto.
3231 (avx512pf_scatterpf<mode>sf): Ditto.
3232 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3233 (*avx512pf_scatterpf<mode>sf): Ditto.
3234 (avx512pf_scatterpf<mode>df): Ditto.
3235 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3236 (*avx512pf_scatterpf<mode>df): Ditto.
3237 (avx512f_expand<mode>): Removed.
3238 (<shift_insn><mode>3<mask_name>): Change predicate type.
3240 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3242 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3243 not at the end of datarefs vector use ordered_remove to avoid
3244 reordering datarefs vector.
3247 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3248 mark local addressable non-static vars as GOVD_PRIVATE
3249 instead of GOVD_LOCAL.
3250 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3251 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3255 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3256 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3257 assume_aligned or alloc_align attributes.
3258 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3259 arguments. Handle also assume_aligned and alloc_align attributes.
3260 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3261 calls to functions with assume_aligned or alloc_align attributes.
3262 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3264 2014-02-08 Terry Guo <terry.guo@arm.com>
3266 * doc/invoke.texi: Document ARM -march=armv7e-m.
3268 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3270 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3271 flag on __cilkrts_rethrow builtin.
3274 * ipa-cp.c (determine_versionability): Fail at -O0
3275 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3276 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3279 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3282 * tree-inline.c (copy_forbidden): Fail for
3283 __attribute__((optimize (0))) functions.
3285 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3287 * varpool.c: Include pointer-set.h.
3288 (varpool_remove_unreferenced_decls): Variables in other partitions
3289 will not be output; be however careful to not lose information
3292 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3294 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3295 lookup in the vtable constructor.
3297 2014-02-07 Jeff Law <law@redhat.com>
3300 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3301 define_insn_and_split.
3303 * ipa-inline.c (inline_small_functions): Fix typos.
3305 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3307 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3308 (s390_can_use_return_insn): Declare.
3309 * config/s390/s390.h (EPILOGUE_USES): Define.
3310 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3312 (s390_chunkify_start): Handle return JUMP_LABELs.
3313 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3314 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3315 (s390_can_use_return_insn): New functions.
3316 (s390_fix_long_loop_prediction): Handle conditional returns.
3317 (TARGET_SET_UP_BY_PROLOGUE): Define.
3318 * config/s390/s390.md (ANY_RETURN): New code iterator.
3319 (*creturn, *csimple_return, return, simple_return): New patterns.
3321 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3323 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3324 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3325 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3326 REG_CFA_RESTORE list when deciding not to restore a register.
3328 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3330 * config/s390/s390.c: Include tree-pass.h and context.h.
3331 (s390_early_mach): New function, split out from...
3332 (s390_emit_prologue): ...here.
3333 (pass_data_s390_early_mach): New pass structure.
3334 (pass_s390_early_mach): New class.
3335 (s390_option_override): Create and register early_mach pass.
3336 Move to end of file.
3338 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3340 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3341 to match for the exit block.
3343 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3345 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3346 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3347 Reject misaligned operands.
3349 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3351 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3353 2014-02-07 Richard Biener <rguenther@suse.de>
3356 * gimple-low.c (lower_builtin_posix_memalign): New function.
3357 (lower_stmt): Call it to lower posix_memalign in a way
3358 to make alignment info accessible.
3360 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3363 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3364 __builtin_setjmp_receiver.
3366 2014-02-07 Richard Biener <rguenther@suse.de>
3369 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3370 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3371 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3372 Handle BUILT_IN_POSIX_MEMALIGN.
3373 (find_func_clobbers): Likewise.
3374 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3375 (call_may_clobber_ref_p_1): Likewise.
3377 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3380 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3383 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3386 * lto-cgraph.c (lto_output_node): Use
3387 symtab_get_symbol_partitioning_class.
3388 (lto_output_varpool_node): likewise.
3389 (symtab_get_symbol_partitioning_class): Move here from
3391 * cgraph.h (symbol_partitioning_class): Likewise.
3392 (symtab_get_symbol_partitioning_class): Declare.
3394 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3396 * ggc.h (ggc_internal_cleared_alloc): New macro.
3397 * vec.h (vec_safe_copy): Handle memory stats.
3398 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3399 * target-globals.c (save_target_globals): Likewise.
3401 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3404 * expr.c (emit_move_resolve_push): Export; be bit more selective
3405 on when to clear alias set.
3406 * expr.h (emit_move_resolve_push): Declare.
3407 * function.h (struct function): Add tail_call_marked.
3408 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3409 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3410 * config/i386/i386.md (TImode move expander): De not call
3412 (FP push expanders): Preserve memory attributes.
3413 * config/i386/sse.md (push<mode>1): Remove.
3414 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3415 (ix86_expand_push): Remove.
3416 * config/i386/mmx.md (push<mode>1): Remove.
3418 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3420 PR rtl-optimization/60030
3421 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3422 lopart with paradoxical subreg before shifting it up by hprec.
3424 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3426 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3427 Remove extra newline at end of file.
3428 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3429 (arm_issue_rate): Handle cortexa57.
3430 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3431 (cortex-a57.cortex-a53): Likewise.
3433 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3436 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3437 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3439 (arm_expand_prologue): Adjust all callers.
3440 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3441 info, registers also at the lowest numbered registers side. Use
3442 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3446 * var-tracking.c (adjust_mems): Before adding a SET to
3447 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3449 2014-02-06 Alan Modra <amodra@gmail.com>
3452 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3453 change SDmode to DDmode when lra_in_progress.
3455 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3458 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3459 free_data_ref on the dr first, and before goto again also set dr
3460 to the next dr. For simd_lane_access, free old datarefs[i] before
3461 overwriting it. For get_vectype_for_scalar_type failure, don't
3462 free_data_ref if simd_lane_access.
3464 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3467 * tree.h (opts_for_fn): New inline function.
3468 (opt_for_fn): Define.
3469 * config/i386/i386.c (ix86_function_regparm): Use
3470 opt_for_fn (decl, optimize) instead of optimize.
3472 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3474 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3475 for SYMBOL_REF in large memory model.
3477 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3479 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3481 (cortex-a57): Likewise.
3482 (cortex-a57.cortex-a53): Likewise.
3484 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3485 Kugan Vivekanandarajah <kuganv@linaro.org>
3487 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3489 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3491 2014-02-06 Richard Biener <rguenther@suse.de>
3493 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3494 set_loop_copy and initialize_original_copy_tables.
3496 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3498 * config/aarch64/aarch64-simd.md
3499 (aarch64_ashr_simddi): Change QI to SI.
3501 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3502 Jakub Jelinek <jakub@redhat.com>
3505 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3508 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3510 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3511 CODE_FOR_altivec_vpku[hw]um to
3512 CODE_FOR_altivec_vpku[hw]um_direct.
3513 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3514 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3515 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3516 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3518 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3520 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3521 generation for -maltivec=be.
3522 (altivec_vsumsws): Simplify redundant test.
3524 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3526 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3527 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3528 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3529 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3530 gen_altivec_vpkuwum.
3531 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3533 (altivec_vpks<VI_char>ss): Likewise.
3534 (altivec_vpks<VI_char>us): Likewise.
3535 (altivec_vpku<VI_char>us): Likewise.
3536 (altivec_vpku<VI_char>um): Likewise.
3537 (altivec_vpku<VI_char>um_direct): New (copy of
3538 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3540 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3541 target is little endian and -maltivec=be is not specified.
3542 (*altivec_vupkhs<VU_char>_direct): New (copy of
3543 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3544 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3545 target is little endian and -maltivec=be is not specified.
3546 (*altivec_vupkls<VU_char>_direct): New (copy of
3547 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3548 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3549 little endian and -maltivec=be is not specified.
3550 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3551 little endian and -maltivec=be is not specified.
3553 2014-02-05 Richard Henderson <rth@redhat.com>
3556 * combine-stack-adj.c: Revert r206943.
3557 * sched-int.h (struct deps_desc): Add last_args_size.
3558 * sched-deps.c (init_deps): Initialize it.
3559 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3560 contain REG_ARGS_SIZE notes.
3562 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3564 * lto-cgraph.c (asm_nodes_output): Make global.
3565 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3566 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3567 (driver_handle_option): Handle OPT_fwpa.
3569 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3572 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3573 a comment typo and formatting issue. If odr_hash hasn't been
3574 created, return vNULL and set *completep to false.
3577 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3578 bb with no successors.
3580 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3583 * doc/invoke.texi (-march): Clarify documentation for ARM.
3587 2014-02-05 Richard Biener <rguenther@suse.de>
3589 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3590 when not vectorizing because of too many alias checks.
3591 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3592 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3594 2014-02-05 Nick Clifton <nickc@redhat.com>
3596 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3597 accept extended registers in any mode when compiling for the MN10300.
3599 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3601 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3602 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3603 sanitization attributes.
3604 (can_inline_edge_p): Likewise.
3605 (sanitize_attrs_match_for_inline_p): New function.
3607 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3609 * ipa-prop.c (detect_type_change): Shor circuit testing of
3610 type changes on THIS pointer.
3612 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3615 * config/pa/pa.c (legitimize_tls_address): Return original address
3616 if not passed a SYMBOL_REF rtx.
3617 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3619 (pa_emit_move_sequence): Simplify TLS source operands.
3620 (pa_legitimate_constant_p): Reject all TLS constants.
3621 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3622 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3624 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3626 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3627 groups when we know they are controlled by LTO.
3628 * varasm.c (default_binds_local_p_1): If object is in other partition,
3629 it will be resolved locally.
3631 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3633 * config/host-linux.c (linux_gt_pch_use_address): Don't
3634 use SSIZE_MAX because it is not always defined.
3636 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3639 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3640 threshold for pseudo splitting.
3641 (update_ebb_live_info): Process call argument hard registers and
3642 hard registers from insn definition too.
3643 (max_small_class_regs_num): New constant.
3644 (inherit_in_ebb): Update live hard regs through EBBs. Update
3645 reloads_num only for small register classes. Don't split for
3648 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3651 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3654 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3656 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3659 2014-02-04 Marek Polacek <polacek@redhat.com>
3661 * gdbinit.in (pel): Define.
3663 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3665 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3668 2014-02-04 Richard Biener <rguenther@suse.de>
3671 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3672 in function context local.
3673 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3674 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3675 similar to LTO_imported_decl_ref.
3677 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3679 PR tree-optimization/60002
3680 * cgraphclones.c (build_function_decl_skip_args): Clear
3683 PR tree-optimization/60023
3684 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3685 false to gsi_replace.
3686 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3687 has been in some EH region and vec_stmt could throw, add
3688 vec_stmt into the same EH region.
3689 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3690 has no lhs, ignore it.
3691 * internal-fn.c (expand_MASK_LOAD): Likewise.
3694 * tree-inline.c (copy_forbidden): Fail for
3695 __attribute__((optimize (0))) functions.
3698 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3699 is set, copy one less argument.
3700 (expand_simd_clones): Don't subtract clone_info->inbranch
3701 from simd_clone_struct_alloc argument.
3703 PR rtl-optimization/57915
3704 * recog.c (simplify_while_replacing): If all unary/binary/relational
3705 operation arguments are constant, attempt to simplify those.
3708 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3709 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3711 2014-02-04 Richard Biener <rguenther@suse.de>
3713 PR tree-optimization/60012
3714 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3715 TBAA disambiguation to all DDRs.
3717 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3720 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3721 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3723 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3726 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3728 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3730 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3731 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3733 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3736 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3737 to figure out targets of polymorphic calls with known decl.
3738 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3739 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3740 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3741 (get_polymorphic_call_info): ... here.
3742 (get_polymorphic_call_info_from_invariant): New function.
3744 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3746 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3747 lookup via vtable pointer; check for type consistency
3748 and turn inconsitent facts into UNREACHABLE.
3749 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3750 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3751 type inconsistent querries; return UNREACHABLE instead.
3753 2014-02-03 Richard Henderson <rth@twiddle.net>
3756 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3757 already processed this node.
3758 (normalize_one_pred_1): Pass along mark_set.
3759 (normalize_one_pred): Create and destroy a pointer_set_t.
3760 (normalize_one_pred_chain): Likewise.
3762 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3764 PR gcov-profile/58602
3765 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3767 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3770 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3771 -fno-devirtualize; try to devirtualize by the knowledge of
3772 virtual table pointer given by aggregate propagation.
3773 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3774 (ipa_print_node_jump_functions): Dump also offset that
3775 is relevant for polymorphic calls.
3776 (determine_known_aggregate_parts): Add arg_type parameter; use it
3777 instead of determining the type from pointer type.
3778 (ipa_compute_jump_functions_for_edge): Update call of
3779 determine_known_aggregate_parts.
3780 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3781 (gimple_get_virt_method_for_binfo): ... here; simplify using
3782 vtable_pointer_value_to_vtable.
3783 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3784 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3785 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3786 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3788 (vtable_pointer_value_to_binfo): ... here.
3789 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3791 2014-02-03 Teresa Johnson <tejohnson@google.com>
3793 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3794 redef of outer loop index variable.
3796 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3800 * doc/extend.texi (Function Attributes): Typo.
3802 2014-02-03 Cong Hou <congh@google.com>
3804 PR tree-optimization/60000
3805 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3806 if the vectorized statement is a store. A store statement can only
3807 appear at the end of pattern statements.
3809 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3811 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3812 (ix86_option_override_internal): Default long double to 64-bit for
3813 32-bit Bionic and to 128-bit for 64-bit Bionic.
3815 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3816 TARGET_LONG_DOUBLE_128 is true.
3817 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3819 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3820 (mlong-double-64): Negate -mlong-double-128.
3821 (mlong-double-128): New option.
3823 * config/i386/i386-c.c (ix86_target_macros): Define
3824 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3826 * doc/invoke.texi: Document -mlong-double-128.
3828 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3830 PR rtl-optimization/60024
3831 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3833 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3835 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3837 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3839 PR rtl-optimization/57662
3840 * sel-sched.c (code_motion_path_driver): Do not mark already not
3841 existing blocks in the visiting bitmap.
3843 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3845 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3846 on the insn being emitted.
3848 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3849 Will Deacon <will.deacon@arm.com>
3851 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3853 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3855 * config/arm/arm-tables.opt: Regenerate.
3857 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3859 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3860 for vector types other than V16QImode.
3861 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3862 define_expand, and call altivec_expand_vec_perm_le when producing
3863 code with little endian element order.
3864 (*altivec_vperm_<mode>_internal): New insn having previous
3865 behavior of altivec_vperm_<mode>.
3866 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3867 altivec_expand_vec_perm_le when producing code with little endian
3869 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3870 behavior of altivec_vperm_<mode>_uns.
3872 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3874 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3875 (altivec_vsumsws): Add handling for -maltivec=be with a little
3877 (altivec_vsumsws_direct): New.
3878 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3879 gen_altivec_vsumsws.
3881 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3883 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3884 vtable_pointer_value_to_binfo): New functions.
3885 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3886 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3888 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3890 * config/nios2/nios2.md (load_got_register): Initialize GOT
3891 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3892 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3894 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3896 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3897 preserverd by passthrough, do not propagate the type.
3899 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3901 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3902 (mips_atomic_assign_expand_fenv): New function.
3903 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3905 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3907 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3908 (__builtin_mips_set_fcsr): Likewise.
3909 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3910 MIPS_USI_FTYPE_VOID.
3911 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3912 (mips16_expand_set_fcsr): Likewise.
3913 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3914 (mips16_set_fcsr_stub): Likewise.
3915 (mips16_get_fcsr_one_only_stub): New class.
3916 (mips16_set_fcsr_one_only_stub): Likewise.
3917 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3918 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3919 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3920 (hard_float): New availability predicate.
3921 (mips_builtins): Add get_fcsr and set_fcsr.
3922 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3923 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3924 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3925 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3926 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3929 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3931 * config/mips/mips.c (mips_one_only_stub): New class.
3932 (mips_need_mips16_rdhwr_p): Replace with...
3933 (mips16_rdhwr_stub): ...this new variable.
3934 (mips16_stub_call_address): New function.
3935 (mips16_rdhwr_one_only_stub): New class.
3936 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3937 (mips_output_mips16_rdhwr): Delete.
3938 (mips_finish_stub): New function.
3939 (mips_code_end): Use it to handle rdhwr stubs.
3941 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3944 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3945 when calculating size of integer atomic types.
3947 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3949 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3951 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3953 PR tree-optimization/60003
3954 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3955 * profile.c (branch_prob): Use gimple_call_builtin_p
3956 to check for BUILT_IN_SETJMP_RECEIVER.
3957 * tree-inline.c (copy_bb): Call notice_special_calls.
3959 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3962 * lra-constraints.c (process_alt_operands): Update reload_sum only
3965 2014-01-31 Richard Henderson <rth@redhat.com>
3968 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3969 until after else_eh is processed.
3971 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3973 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3974 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3975 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3976 in smmintrin.h, remove them.
3977 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3978 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3979 * config/i386/i386.md (ROUND_SAE): Fix value.
3980 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3981 (const48_operand): New.
3982 * config/i386/subst.md (round), (round_expand): Use
3983 const_4_or_8_to_11_operand.
3984 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3986 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3988 * config/i386/constraints.md (Yk): Swap meaning with k.
3989 * config/i386/i386.md (movhi_internal): Change Yk to k.
3990 (movqi_internal): Ditto.
3991 (*k<logic><mode>): Ditto.
3994 (kandn<mode>): Ditto.
3995 (*<code>hi_1): Ditto.
3996 (*<code>qi_1): Ditto.
3997 (kxnor<mode>): Ditto.
3998 (kortestzhi): Ditto.
3999 (kortestchi): Ditto.
4001 (*one_cmplhi2_1): Ditto.
4002 (*one_cmplqi2_1): Ditto.
4003 * config/i386/sse.md (): Change k to Yk.
4004 (avx512f_load<mode>_mask): Ditto.
4005 (avx512f_blendm<mode>): Ditto.
4006 (avx512f_store<mode>_mask): Ditto.
4007 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4008 (avx512f_storedqu<mode>_mask): Ditto.
4009 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4011 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4012 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4013 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4014 (avx512f_maskcmp<mode>3): Ditto.
4015 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4016 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4017 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4018 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4019 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4020 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4021 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4022 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4023 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4024 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4025 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4026 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4027 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4028 (vec_extract_lo_<mode>_maskm): Ditto.
4029 (vec_extract_hi_<mode>_maskm): Ditto.
4030 (avx512f_vternlog<mode>_mask): Ditto.
4031 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4032 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4033 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4034 (avx512f_<code>v8div16qi2_mask): Ditto.
4035 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4036 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4037 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4038 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4039 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4040 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4041 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4042 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4043 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4044 (avx512cd_maskb_vec_dupv8di): Ditto.
4045 (avx512cd_maskw_vec_dupv16si): Ditto.
4046 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4047 (avx512f_vpermi2var<mode>3_mask): Ditto.
4048 (avx512f_vpermi2var<mode>3_mask): Ditto.
4049 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4050 (*avx512f_gathersi<mode>): Ditto.
4051 (*avx512f_gathersi<mode>_2): Ditto.
4052 (*avx512f_gatherdi<mode>): Ditto.
4053 (*avx512f_gatherdi<mode>_2): Ditto.
4054 (*avx512f_scattersi<mode>): Ditto.
4055 (*avx512f_scatterdi<mode>): Ditto.
4056 (avx512f_compress<mode>_mask): Ditto.
4057 (avx512f_compressstore<mode>_mask): Ditto.
4058 (avx512f_expand<mode>_mask): Ditto.
4059 * config/i386/subst.md (mask): Change k to Yk.
4060 (mask_scalar_merge): Ditto.
4063 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4065 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4067 2014-01-31 Richard Biener <rguenther@suse.de>
4070 * builtins.c (fold_builtin_memory_op): Make sure to not
4071 use a floating-point mode or a boolean or enumeral type for
4074 2014-01-30 DJ Delorie <dj@redhat.com>
4076 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4077 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4078 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4079 whenever main() has an epilogue.
4081 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4083 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4084 unused variable "field".
4085 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4086 (vsx_mergeh_<mode>): Likewise.
4087 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4088 (altivec_vmrghh): Likewise.
4089 (altivec_vmrghw): Likewise.
4090 (altivec_vmrglb): Likewise.
4091 (altivec_vmrglh): Likewise.
4092 (altivec_vmrglw): Likewise.
4093 (altivec_vspltb): Add missing uses.
4094 (altivec_vsplth): Likewise.
4095 (altivec_vspltw): Likewise.
4096 (altivec_vspltsf): Likewise.
4098 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4101 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4102 frame related instructions.
4104 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4106 PR rtl-optimization/59959
4107 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4108 any reload of register whose subreg is invalid.
4110 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4112 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4113 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4114 Add missing return type - void.
4116 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4118 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4119 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4120 remove element index adjustment for endian (now handled in vsx.md
4122 (altivec_expand_vec_perm_const): Use
4123 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4124 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4125 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4126 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4127 define_expand and a new define_insn *altivec_vspltb_internal;
4128 adjust for -maltivec=be on a little endian target.
4129 (altivec_vspltb_direct): New.
4130 (altivec_vsplth): Divide into a define_expand and a new
4131 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4132 little endian target.
4133 (altivec_vsplth_direct): New.
4134 (altivec_vspltw): Divide into a define_expand and a new
4135 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4136 little endian target.
4137 (altivec_vspltw_direct): New.
4138 (altivec_vspltsf): Divide into a define_expand and a new
4139 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4140 a little endian target.
4142 2014-01-30 Richard Biener <rguenther@suse.de>
4144 PR tree-optimization/59993
4145 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4146 can propagate form the earlier stmt and avoid the transform
4147 when the intermediate result is needed.
4149 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4151 * README.Portability: Fix typo.
4153 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4155 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4156 comparison_operator with ordered_comparison_operator.
4158 2014-01-30 Nick Clifton <nickc@redhat.com>
4160 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4161 Rename to mn10300_store_multiple_regs.
4162 * config/mn10300/mn10300.c: Likewise.
4163 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4164 store_multiple_regs.
4165 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4166 Call mn10300_store_multiple_regs.
4168 2014-01-30 Nick Clifton <nickc@redhat.com>
4169 DJ Delorie <dj@redhat.com>
4171 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4172 %fp 2 to keep registers after it properly word-aligned.
4173 (rl78_alloc_physical_registers_umul): Handle the case where both
4174 input operands are the same.
4176 2014-01-30 Richard Biener <rguenther@suse.de>
4178 PR tree-optimization/59903
4179 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4182 2014-01-30 Jason Merrill <jason@redhat.com>
4185 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4188 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4190 2014-01-30 Richard Biener <rguenther@suse.de>
4192 PR tree-optimization/59951
4193 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4195 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4198 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4199 SFmode to DFmode case.
4201 2014-01-29 DJ Delorie <dj@redhat.com>
4203 * config/msp430/msp430.opt (-minrt): New.
4204 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4206 (ENDFILE_SPEC): Likewise.
4208 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4210 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4211 (estimate_function_body_sizes): Use it.
4213 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4216 * dwarf2out.c (is_cxx_auto): New.
4217 (is_base_type): Use it.
4218 (gen_type_die_with_usage): Likewise.
4220 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4222 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4223 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4224 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4225 -maltivec=be with LE targets.
4226 (vsx_mergeh_<mode>): Likewise.
4227 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4228 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4229 (altivec_vmrghb): Replace with define_expand and new
4230 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4231 (altivec_vmrghb_direct): New define_insn.
4232 (altivec_vmrghh): Replace with define_expand and new
4233 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4234 (altivec_vmrghh_direct): New define_insn.
4235 (altivec_vmrghw): Replace with define_expand and new
4236 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4237 (altivec_vmrghw_direct): New define_insn.
4238 (*altivec_vmrghsf): Adjust for endianness.
4239 (altivec_vmrglb): Replace with define_expand and new
4240 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4241 (altivec_vmrglb_direct): New define_insn.
4242 (altivec_vmrglh): Replace with define_expand and new
4243 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4244 (altivec_vmrglh_direct): New define_insn.
4245 (altivec_vmrglw): Replace with define_expand and new
4246 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4247 (altivec_vmrglw_direct): New define_insn.
4248 (*altivec_vmrglsf): Adjust for endianness.
4249 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4250 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4251 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4252 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4253 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4254 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4255 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4256 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4258 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4260 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4261 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4264 2014-01-29 Richard Biener <rguenther@suse.de>
4266 PR tree-optimization/58742
4267 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4268 associate_pointerplus_align.
4269 (associate_pointerplus_diff): New function.
4270 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4271 and associate_pointerplus_diff.
4273 2014-01-29 Richard Biener <rguenther@suse.de>
4275 * lto-streamer.h (LTO_major_version): Bump to 3.
4276 (LTO_minor_version): Reset to 0.
4278 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4280 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4281 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4282 (arm_file_start): Generate correct asm header for armv7ve.
4283 * config/arm/bpabi.h: Add multilib support for armv7ve.
4284 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4285 and cortex-a15 to armv7ve.
4286 * config/arm/t-aprofile: Add multilib support for armv7ve.
4287 * doc/invoke.texi: Document -march=armv7ve.
4289 2014-01-29 Richard Biener <rguenther@suse.de>
4291 PR tree-optimization/58742
4292 * tree-ssa-forwprop.c (associate_plusminus): Return true
4293 if we changed sth, defer EH cleanup to ...
4294 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4295 (simplify_mult): New function.
4297 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4300 PR tree-optimization/59920
4301 * tree.c (build_common_builtin_nodes): Remove
4302 __builtin_setjmp_dispatcher initialization.
4303 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4304 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4305 instead of gsi_after_labels + manually skipping debug stmts.
4306 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4307 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4308 * tree-inline.c (copy_edges_for_bb): Remove
4309 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4310 argument. Ignore computed_goto_p stmts. Don't call
4311 make_abnormal_goto_edges. If a call might need abnormal edges
4312 for non-local gotos, see if it already has an edge to
4313 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4314 with true argument, don't do anything then, otherwise add
4315 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4316 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4318 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4319 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4320 (lower_stmt): Don't set data->calls_builtin_setjmp.
4321 (lower_builtin_setjmp): Adjust comment.
4322 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4323 * tree-cfg.c (found_computed_goto): Remove.
4324 (factor_computed_gotos): Remove.
4325 (make_goto_expr_edges): Return bool, true for computed gotos.
4326 Don't call make_abnormal_goto_edges.
4327 (build_gimple_cfg): Don't set found_computed_goto, don't call
4328 factor_computed_gotos.
4329 (computed_goto_p): No longer static.
4330 (make_blocks): Don't set found_computed_goto.
4331 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4332 (make_edges): If make_goto_expr_edges returns true, push bb
4333 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4334 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4335 vector. Record mapping between bbs and OpenMP regions if there
4336 are any, adjust make_gimple_omp_edges caller. Call
4337 handle_abnormal_edges.
4338 (make_abnormal_goto_edges): Remove.
4339 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4340 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4341 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4342 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4343 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4344 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4345 filling *region also set *region_idx to (*region)->entry->index.
4348 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4349 For REGs set ORIGINAL_REGNO.
4351 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4353 * doc/md.texi: Mention that a target shouldn't implement
4354 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4357 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4359 PR tree-optimization/59594
4360 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4361 a copy of the datarefs vector rather than the vector itself.
4363 2014-01-28 Jason Merrill <jason@redhat.com>
4366 * dwarf2out.c (auto_die): New static.
4367 (gen_type_die_with_usage): Handle C++1y 'auto'.
4368 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4371 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4374 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4375 (SPEC_X32): Likewise.
4376 (SPEC_64): Likewise.
4377 * config/i386/i386.c (ix86_option_override_internal): Turn off
4378 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4380 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4381 * config/i386/i386.h (TARGET_16BIT): New macro.
4382 (TARGET_16BIT_P): Likewise.
4383 * config/i386/i386.opt: Add m16.
4384 * doc/invoke.texi: Document -m16.
4386 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4388 PR preprocessor/59935
4389 * input.c (location_get_source_line): Bail out on when line number
4390 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4392 2014-01-28 Richard Biener <rguenther@suse.de>
4394 PR tree-optimization/58742
4395 * tree-ssa-forwprop.c (associate_plusminus): Handle
4396 pointer subtraction of the form (T)(P + A) - (T)P.
4398 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4400 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4403 2014-01-28 Richard Biener <rguenther@suse.de>
4406 2014-01-28 Richard Biener <rguenther@suse.de>
4408 PR rtl-optimization/45364
4409 PR rtl-optimization/59890
4410 * var-tracking.c (local_get_addr_clear_given_value): Handle
4411 already cleared slot.
4412 (val_reset): Handle not allocated local_get_addr_cache.
4413 (vt_find_locations): Use post-order on the inverted CFG.
4415 2014-01-28 Richard Biener <rguenther@suse.de>
4417 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4419 2014-01-28 Richard Biener <rguenther@suse.de>
4421 PR rtl-optimization/45364
4422 PR rtl-optimization/59890
4423 * var-tracking.c (local_get_addr_clear_given_value): Handle
4424 already cleared slot.
4425 (val_reset): Handle not allocated local_get_addr_cache.
4426 (vt_find_locations): Use post-order on the inverted CFG.
4428 2014-01-28 Alan Modra <amodra@gmail.com>
4430 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4431 * configure.ac <recursive call for build != host>: Define
4432 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4433 and LD_FOR_BUILD too.
4434 * configure: Regenerate.
4436 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4438 * config/i386/i386.c (get_builtin_code_for_version): Separate
4439 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4440 Broadwell from Haswell.
4442 2014-01-27 Steve Ellcey <sellcey@mips.com>
4444 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4445 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4446 * config/mips/mips.c (mips_option_override): Change setting
4448 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4449 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4450 Change from Mask to Var.
4452 2014-01-27 Jeff Law <law@redhat.com>
4454 * ipa-inline.c (inline_small_functions): Fix typo.
4456 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4458 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4459 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4460 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4461 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4462 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4463 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4464 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4465 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4466 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4467 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4468 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4469 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4470 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4471 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4472 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4473 (_mm512_storeu_epi64): Ditto.
4474 (_mm512_cmpge_epi32_mask): Ditto.
4475 (_mm512_cmpge_epu32_mask): Ditto.
4476 (_mm512_cmpge_epi64_mask): Ditto.
4477 (_mm512_cmpge_epu64_mask): Ditto.
4478 (_mm512_cmple_epi32_mask): Ditto.
4479 (_mm512_cmple_epu32_mask): Ditto.
4480 (_mm512_cmple_epi64_mask): Ditto.
4481 (_mm512_cmple_epu64_mask): Ditto.
4482 (_mm512_cmplt_epi32_mask): Ditto.
4483 (_mm512_cmplt_epu32_mask): Ditto.
4484 (_mm512_cmplt_epi64_mask): Ditto.
4485 (_mm512_cmplt_epu64_mask): Ditto.
4486 (_mm512_cmpneq_epi32_mask): Ditto.
4487 (_mm512_cmpneq_epu32_mask): Ditto.
4488 (_mm512_cmpneq_epi64_mask): Ditto.
4489 (_mm512_cmpneq_epu64_mask): Ditto.
4490 (_mm512_expand_pd): Ditto.
4491 (_mm512_expand_ps): Ditto.
4492 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4493 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4494 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4495 * config/i386/i386.c (ix86_builtins): Add
4496 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4497 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4498 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4499 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4500 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4501 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4502 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4503 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4504 IX86_BUILTIN_PMOVUSQW512_MEM.
4505 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4506 __builtin_ia32_pmovsqd512mem_mask,
4507 __builtin_ia32_pmovqd512mem_mask,
4508 __builtin_ia32_pmovusqw512mem_mask,
4509 __builtin_ia32_pmovsqw512mem_mask,
4510 __builtin_ia32_pmovqw512mem_mask,
4511 __builtin_ia32_pmovusdw512mem_mask,
4512 __builtin_ia32_pmovsdw512mem_mask,
4513 __builtin_ia32_pmovdw512mem_mask,
4514 __builtin_ia32_pmovqb512mem_mask,
4515 __builtin_ia32_pmovusqb512mem_mask,
4516 __builtin_ia32_pmovsqb512mem_mask,
4517 __builtin_ia32_pmovusdb512mem_mask,
4518 __builtin_ia32_pmovsdb512mem_mask,
4519 __builtin_ia32_pmovdb512mem_mask.
4520 (bdesc_args): Add __builtin_ia32_expanddf512,
4521 __builtin_ia32_expandsf512.
4522 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4523 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4524 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4525 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4526 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4527 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4528 (avx512f_<code>v8div16qi2_mask_store): This.
4529 (avx512f_expand<mode>): New.
4531 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4533 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4535 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4536 (_mm512_prefetch_i32scatter_pd): Ditto.
4537 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4538 (_mm512_prefetch_i64scatter_pd): Ditto.
4539 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4540 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4541 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4542 (_mm512_prefetch_i32scatter_ps): Ditto.
4543 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4544 (_mm512_prefetch_i64scatter_ps): Ditto.
4545 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4546 * config/i386/i386-builtin-types.def: Define
4547 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4548 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4549 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4550 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4551 IX86_BUILTIN_SCATTERPFQPD.
4552 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4553 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4554 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4555 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4556 __builtin_ia32_scatterpfqps.
4557 (ix86_expand_builtin): Expand new built-ins.
4558 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4559 fix memory access data type.
4560 (*avx512pf_gatherpf<mode>_mask): Ditto.
4561 (*avx512pf_gatherpf<mode>): Ditto.
4562 (avx512pf_scatterpf<mode>): Ditto.
4563 (*avx512pf_scatterpf<mode>_mask): Ditto.
4564 (*avx512pf_scatterpf<mode>): Ditto.
4565 (GATHER_SCATTER_SF_MEM_MODE): New.
4566 (avx512pf_gatherpf<mode>df): Ditto.
4567 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4568 (*avx512pf_scatterpf<mode>df): Ditto.
4570 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4573 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4574 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4577 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4579 * common/config/arm/arm-common.c
4580 (arm_rewrite_mcpu): Handle multiple names.
4582 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4584 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4586 * gimple-builder.h (create_gimple_tmp): Delete.
4588 2014-01-27 Christian Bruel <christian.bruel@st.com>
4590 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4593 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4595 * config/pa/pa.md (call): Generate indirect long calls to non-local
4596 functions when outputing 32-bit code.
4597 (call_value): Likewise except for special call to buggy powf function.
4599 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4600 portable runtime and PIC indirect calls.
4601 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4602 and PIC call sequences. Use ldo instead of blr to set return register
4603 in PIC call sequence.
4605 2014-01-25 Walter Lee <walt@tilera.com>
4607 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4608 avoid clobbering a live register.
4610 2014-01-25 Walter Lee <walt@tilera.com>
4612 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4613 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4614 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4615 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4617 2014-01-25 Walter Lee <walt@tilera.com>
4619 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4620 arguments on even registers.
4621 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4623 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4624 (BIGGEST_ALIGNMENT): Ditto.
4625 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4627 2014-01-25 Walter Lee <walt@tilera.com>
4629 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4630 insns before bundling.
4631 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4633 2014-01-25 Walter Lee <walt@tilera.com>
4635 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4636 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4637 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4639 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4641 * config/mips/constraints.md (kl): Delete.
4642 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4643 define expands, using...
4644 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4645 instructions for MIPS16.
4646 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4647 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4649 2014-01-25 Walter Lee <walt@tilera.com>
4651 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4655 2014-01-25 Walter Lee <walt@tilera.com>
4657 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4658 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4660 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4662 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4665 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4667 * print-rtl.c (in_call_function_usage): New var.
4668 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4669 EXPR_LIST mode as mode and not as reg note name.
4672 * cfgloopmanip.c (copy_loop_info): If
4673 loop->warned_aggressive_loop_optimizations, make sure
4674 the flag is set in target loop too.
4676 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4678 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4680 * builtins.def: Likewise.
4681 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4682 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4683 * ira.c (ira_setup_eliminable_regset): Likewise.
4684 * omp-low.c (gate_expand_omp): Likewise.
4685 (execute_lower_omp): Likewise.
4686 (diagnose_sb_0): Likewise.
4687 (gate_diagnose_omp_blocks): Likewise.
4688 (simd_clone_clauses_extract): Likewise.
4691 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4693 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4694 correction for little endian...
4695 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4698 2014-01-24 Jeff Law <law@redhat.com>
4700 PR tree-optimization/59919
4701 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4702 for non-returning calls.
4704 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4706 * common/config/aarch64/aarch64-common.c
4707 (aarch64_rewrite_mcpu): Handle multiple names.
4708 * config/aarch64/aarch64.h
4709 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4711 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4713 * input.c (add_file_to_cache_tab): Handle the case where fopen
4716 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4719 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4720 from push operand if code of push isn't PRE_DEC.
4722 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4725 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4726 -mquad-memory-atomic. Update -mquad-memory documentation to say
4727 it is only used for non-atomic loads/stores.
4729 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4730 -mquad-memory or -mquad-memory-atomic switches.
4732 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4733 -mquad-memory-atomic to ISA 2.07 support.
4735 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4736 to separate support of normal quad word memory operations (ldq, stq)
4737 from the atomic quad word memory operations.
4739 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4740 support to separate non-atomic quad word operations from atomic
4741 quad word operations. Disable non-atomic quad word operations in
4742 little endian mode so that we don't have to swap words after the
4743 load and before the store.
4744 (quad_load_store_p): Add comment about atomic quad word support.
4745 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4746 options printed with -mdebug=reg.
4748 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4749 -mquad-memory-atomic as the test for whether we have quad word
4750 atomic instructions.
4751 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4752 or -mp8-vector are used, allow byte/half-word atomic operations.
4754 * config/rs6000/sync.md (load_lockedti): Insure that the address
4755 is a proper indexed or indirect address for the lqarx instruction.
4756 On little endian systems, swap the hi/lo registers after the lqarx
4758 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4759 insure the address is valid for the lqarx instruction.
4760 (store_conditionalti): Insure that the address is a proper indexed
4761 or indirect address for the stqcrx. instruction. On little endian
4762 systems, swap the hi/lo registers before doing the stqcrx.
4764 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4765 insure the address is valid for the stqcrx. instruction.
4767 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4768 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4769 type of quad memory support is available.
4771 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4774 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4775 there is a danger of looping.
4777 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4780 force flag_ira_loop_pressure if set via command line.
4782 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4784 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4785 (ashr_simd): New builtin handling DI mode.
4786 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4787 (aarch64_sshr_simddi): New match pattern.
4788 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4789 (vshrd_n_s64): Likewise.
4790 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4792 2014-01-23 Nick Clifton <nickc@redhat.com>
4794 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4795 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4796 favour of mcu specific scripts.
4797 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4800 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4801 Alex Velenko <Alex.Velenko@arm.com>
4803 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4804 (vaddv_s16): Likewise.
4805 (vaddv_s32): Likewise.
4806 (vaddv_u8): Likewise.
4807 (vaddv_u16): Likewise.
4808 (vaddv_u32): Likewise.
4809 (vaddvq_s8): Likewise.
4810 (vaddvq_s16): Likewise.
4811 (vaddvq_s32): Likewise.
4812 (vaddvq_s64): Likewise.
4813 (vaddvq_u8): Likewise.
4814 (vaddvq_u16): Likewise.
4815 (vaddvq_u32): Likewise.
4816 (vaddvq_u64): Likewise.
4817 (vaddv_f32): Likewise.
4818 (vaddvq_f32): Likewise.
4819 (vaddvq_f64): Likewise.
4820 (vmaxv_f32): Likewise.
4821 (vmaxv_s8): Likewise.
4822 (vmaxv_s16): Likewise.
4823 (vmaxv_s32): Likewise.
4824 (vmaxv_u8): Likewise.
4825 (vmaxv_u16): Likewise.
4826 (vmaxv_u32): Likewise.
4827 (vmaxvq_f32): Likewise.
4828 (vmaxvq_f64): Likewise.
4829 (vmaxvq_s8): Likewise.
4830 (vmaxvq_s16): Likewise.
4831 (vmaxvq_s32): Likewise.
4832 (vmaxvq_u8): Likewise.
4833 (vmaxvq_u16): Likewise.
4834 (vmaxvq_u32): Likewise.
4835 (vmaxnmv_f32): Likewise.
4836 (vmaxnmvq_f32): Likewise.
4837 (vmaxnmvq_f64): Likewise.
4838 (vminv_f32): Likewise.
4839 (vminv_s8): Likewise.
4840 (vminv_s16): Likewise.
4841 (vminv_s32): Likewise.
4842 (vminv_u8): Likewise.
4843 (vminv_u16): Likewise.
4844 (vminv_u32): Likewise.
4845 (vminvq_f32): Likewise.
4846 (vminvq_f64): Likewise.
4847 (vminvq_s8): Likewise.
4848 (vminvq_s16): Likewise.
4849 (vminvq_s32): Likewise.
4850 (vminvq_u8): Likewise.
4851 (vminvq_u16): Likewise.
4852 (vminvq_u32): Likewise.
4853 (vminnmv_f32): Likewise.
4854 (vminnmvq_f32): Likewise.
4855 (vminnmvq_f64): Likewise.
4857 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4859 * config/aarch64/aarch64-simd.md
4860 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4861 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4862 (*aarch64_mul3_elt<mode>): Likewise.
4863 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4864 (*aarch64_mul3_elt_to_64v2df): Likewise.
4865 (*aarch64_mla_elt<mode>): Likewise.
4866 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4867 (*aarch64_mls_elt<mode>): Likewise.
4868 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4869 (*aarch64_fma4_elt<mode>): Likewise.
4870 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4871 (*aarch64_fma4_elt_to_64v2df): Likewise.
4872 (*aarch64_fnma4_elt<mode>): Likewise.
4873 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4874 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4875 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4876 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4877 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4878 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4879 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4880 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4881 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4883 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4885 * config/aarch64/aarch64-simd.md
4886 (aarch64_be_checked_get_lane<mode>): New define_expand.
4887 * config/aarch64/aarch64-simd-builtins.def
4888 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4889 New builtin definition.
4890 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4891 Use new safe be builtin.
4893 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4895 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4897 (aarch64_be_st1<mode>): Likewise.
4898 (aarch_ld1<VALL:mode>): Define_expand modified.
4899 (aarch_st1<VALL:mode>): Likewise.
4900 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4901 (UNSPEC_ST1): Likewise.
4903 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4905 * config/microblaze/microblaze.md: Add trap insn and attribute
4907 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4909 PR preprocessor/58580
4910 * input.h (location_get_source_line): Take an additional line_size
4912 (void diagnostics_file_cache_fini): Declare new function.
4913 * input.c (struct fcache): New type.
4914 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4915 New static constants.
4916 (diagnostic_file_cache_init, total_lines_num)
4917 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4918 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4919 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4920 (get_next_line, read_next_line, goto_next_line, read_line_num):
4921 New static function definitions.
4922 (diagnostic_file_cache_fini): New function.
4923 (location_get_source_line): Take an additional output line_len
4924 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4926 * diagnostic.c (diagnostic_finish): Call
4927 diagnostic_file_cache_fini.
4928 (adjust_line): Take an additional input parameter for the length
4929 of the line, rather than calculating it with strlen.
4930 (diagnostic_show_locus): Adjust the use of
4931 location_get_source_line and adjust_line with respect to their new
4932 signature. While displaying a line now, do not stop at the first
4933 null byte. Rather, display the zero byte as a space and keep
4934 going until we reach the size of the line.
4935 * Makefile.in: Add vec.o to OBJS-libcommon
4937 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4938 Ilya Tocar <ilya.tocar@intel.com>
4940 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4941 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4942 (__builtin_ia32_kmov16): Ditto.
4943 * config/i386/i386.md (UNSPEC_KMOV): New.
4946 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4948 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4949 (_mm512_storeu_si512): Ditto.
4951 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4954 * rtl.h (get_referenced_operands): Declare.
4955 * recog.c (get_referenced_operands): New function.
4956 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4957 operands have been referenced when recording LO_SUM references.
4959 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4961 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4963 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4965 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4966 Enable for generic and recent AMD targets.
4968 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4970 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4971 ARG_SIZE note when adjustment was eliminated.
4973 2014-01-22 Jeff Law <law@redhat.com>
4975 PR tree-optimization/59597
4976 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4977 in file. Accept new argument REGISTERING and use it to modify
4978 dump output appropriately.
4979 (register_jump_thread): Corresponding changes.
4980 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4981 thread paths involving joiner blocks. Add code to dump cancelled
4982 jump threading paths.
4984 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4986 PR rtl-optimization/59477
4987 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4988 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4990 2014-01-22 Tom Tromey <tromey@redhat.com>
4992 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4994 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4996 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4998 PR rtl-optimization/59896
4999 * lra-constraints.c (process_alt_operands): Check unused note for
5000 matched operands of insn with no output reloads.
5002 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5004 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5005 (mips_move_from_gpr_cost): Likewise.
5007 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5009 PR rtl-optimization/59858
5010 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5011 ira_class_hard_regs_num.
5012 (process_alt_operands): Increase reject for dying matched operand.
5014 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5017 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5018 smaller than size, perform several stores or loads and stores
5019 at dst + count - size to store or copy all of size bytes, rather
5020 than just last modesize bytes.
5022 2014-01-20 DJ Delorie <dj@redhat.com>
5024 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5025 that CLOBBERs are REGs before propogating their values.
5027 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5030 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5031 (cgraph_inline_failed_type): New function.
5032 * cgraph.h (DEFCIFCODE): Add type.
5033 (cgraph_inline_failed_type_t): New enum.
5034 (cgraph_inline_failed_type): New prototype.
5035 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5036 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5037 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5038 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5039 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5040 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5041 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5042 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5043 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5044 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5045 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5046 OPTIMIZATION_MISMATCH.
5047 * tree-inline.c (expand_call_inline): Emit errors during
5048 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5050 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5053 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5054 mode attribute in insn output.
5056 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5058 * output.h (output_constant): Delete.
5059 * varasm.c (output_constant): Make private.
5061 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5063 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5065 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5068 * tree.h (fold_builtin_strcat): New prototype.
5069 * builtins.c (fold_builtin_strcat): No longer static. Add len
5070 argument, if non-NULL, don't call c_strlen. Optimize
5071 directly into __builtin_memcpy instead of __builtin_strcpy.
5072 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5073 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5075 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5077 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5078 for SImode_address_operand operands, having only a REG argument.
5080 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5082 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5083 loader name using mbig-endian.
5084 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5086 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5088 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5092 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5094 * config/aarch64/aarch64-protos.h
5095 (aarch64_cannot_change_mode_class_ptr): Declare.
5096 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5097 aarch64_cannot_change_mode_class_ptr): New.
5098 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5099 backend hook aarch64_cannot_change_mode_class.
5101 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5103 * common/config/aarch64/aarch64-common.c
5104 (aarch64_handle_option): Don't handle any option order logic here.
5105 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5106 selected_cpu, warn on architecture version mismatch.
5107 (aarch64_override_options): Fix parsing order for option strings.
5109 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5110 Iain Sandoe <iain@codesourcery.com>
5113 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5114 warning. Amend comment to reflect current functionality.
5116 2014-01-20 Richard Biener <rguenther@suse.de>
5119 * builtins.c (fold_builtin_strcat): Remove case better handled
5120 by tree-ssa-strlen.c.
5122 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5124 * config/aarch64/aarch64.opt
5125 (mcpu, march, mtune): Make case-insensitive.
5127 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5130 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5131 if operands[1] is a REG or ZERO_EXTEND of a REG.
5133 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5135 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5137 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5139 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5140 long non-pic millicode calls.
5142 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5144 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5146 2014-01-19 Kito Cheng <kito@0xlab.org>
5148 * builtins.c (expand_movstr): Check movstr expand done or fail.
5150 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5151 H.J. Lu <hongjiu.lu@intel.com>
5154 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5155 to DImode for zero-extended addresses.
5157 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5159 PR rtl-optimization/57763
5160 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5161 on the new indirect jump_insn and increment LABEL_NUSES (label).
5163 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5167 * config.gcc (x86_archs): New variable.
5168 (x86_64_archs): Likewise.
5169 (x86_cpus): Likewise.
5170 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5171 --with-arch/--with-cpu= options.
5172 Support --with-arch=/--with-cpu={nehalem,westmere,
5173 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5175 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5177 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5178 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5180 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5182 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5184 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5187 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5188 clear cpp_get_options (parse_in)->warn_unused_macros for
5189 ix86_target_macros_internal with cpp_define.
5191 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5193 * jump.c (delete_related_insns): Keep (use (insn))s.
5194 * reorg.c (redundant_insn): Check for barriers too.
5196 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5198 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5200 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5202 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5203 call to $$dyncall when TARGET_LONG_CALLS is true.
5205 2014-01-17 Jeff Law <law@redhat.com>
5207 * ree.c (combine_set_extension): Temporarily disable test for
5208 changing number of hard registers.
5210 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5213 * ipa-inline-analysis.c (inline_free_summary):
5214 Do not free summary of aliases.
5216 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5219 * gimplify.c (gimplify_expr): Use create_tmp_var
5220 instead of create_tmp_var_raw. If cond doesn't have
5221 integral type, don't add the IFN_ANNOTATE builtin at all.
5223 2014-01-17 Martin Jambor <mjambor@suse.cz>
5226 * ipa-cp.c (prev_edge_clone): New variable.
5227 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5228 Also resize prev_edge_clone vector.
5229 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5230 (ipcp_edge_removal_hook): New function.
5231 (ipcp_driver): Register ipcp_edge_removal_hook.
5233 2014-01-17 Andrew Pinski <apinski@cavium.com>
5234 Steve Ellcey <sellcey@mips.com>
5237 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5240 2014-01-17 Jeff Law <law@redhat.com>
5243 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5244 so that pass_ccp runs first.
5246 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5248 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5249 (ix86_adjust_cost): Use !TARGET_XXX.
5250 (do_reorder_for_imul): Likewise.
5251 (swap_top_of_ready_list): Likewise.
5252 (ix86_sched_reorder): Likewise.
5254 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5256 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5257 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5258 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5259 (intel_memset): New. Duplicate slm_memset.
5260 (intel_cost): New. Duplicate slm_cost.
5261 (m_INTEL): New macro.
5262 (processor_target_table): Add "intel".
5263 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5264 with PROCESSOR_INTEL for "intel".
5265 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5266 PROCESSOR_SILVERMONT.
5267 (ix86_issue_rate): Likewise.
5268 (ix86_adjust_cost): Likewise.
5269 (ia32_multipass_dfa_lookahead): Likewise.
5270 (swap_top_of_ready_list): Likewise.
5271 (ix86_sched_reorder): Likewise.
5272 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5273 instead of TARGET_OPT_AGU.
5274 * config/i386/i386.h (TARGET_INTEL): New.
5275 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5276 (processor_type): Add PROCESSOR_INTEL.
5277 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5278 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5280 2014-01-17 Marek Polacek <polacek@redhat.com>
5283 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5286 2014-01-17 Richard Biener <rguenther@suse.de>
5288 PR tree-optimization/46590
5289 * opts.c (default_options_table): Add entries for
5290 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5291 all enabled at -O1 but not for -Og.
5292 * common.opt (fbranch-count-reg): Remove Init(1).
5293 (fmove-loop-invariants): Likewise.
5294 (ftree-pta): Likewise.
5296 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5298 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5299 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5300 decls to at least the GCC 4.8 used alignments.
5303 * tree-nested.c (convert_nonlocal_reference_stmt,
5304 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5305 of GIMPLE_BIND stmts, adjust associated decls.
5307 2014-01-17 Richard Biener <rguenther@suse.de>
5309 PR tree-optimization/46590
5310 * vec.h (vec<>::bseach): New member function implementing
5311 binary search according to C89 bsearch.
5312 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5313 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5314 bitmap pointer again. Make accesses_in_loop a flat array.
5315 (mem_ref_obstack): New global.
5316 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5317 (mark_ref_stored): Likewise.
5318 (ref_indep_loop_p_2): Likewise.
5319 (set_ref_stored_in_loop): New helper function.
5320 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5321 (memref_free): Adjust.
5322 (record_mem_ref_loc): Simplify.
5323 (gather_mem_refs_stmt): Adjust.
5324 (sort_locs_in_loop_postorder_cmp): New function.
5325 (analyze_memory_references): Sort accesses_in_loop after
5326 loop postorder number.
5327 (find_ref_loc_in_loop_cmp): New function.
5328 (for_all_locs_in_loop): Find relevant cluster of locs in
5329 accesses_in_loop and iterate without recursion.
5330 (execute_sm): Avoid uninit warning.
5331 (struct ref_always_accessed): Simplify.
5332 (ref_always_accessed::operator ()): Likewise.
5333 (ref_always_accessed_p): Likewise.
5334 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5335 loop postorder numbers here.
5336 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5339 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5342 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5343 on decls for which assemble_alias has been called.
5345 2014-01-17 Nick Clifton <nickc@redhat.com>
5347 * config/msp430/msp430.opt: (mcpu): New option.
5348 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5349 (msp430_option_override): Parse target_cpu. If the MCU name
5350 matches a generic string, clear target_mcu.
5351 (msp430_attr): Allow numeric interrupt values up to 63.
5352 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5353 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5355 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5357 * config/msp430/msp430.md (popm): Use %J rather than %I.
5358 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5359 (addhi_cy_i): Use immediate_operand for operand 2.
5360 * doc/invoke.texi: Document -mcpu option.
5362 2014-01-17 Richard Biener <rguenther@suse.de>
5364 PR rtl-optimization/38518
5365 * df.h (df_analyze_loop): Declare.
5366 * df-core.c: Include cfgloop.h.
5367 (df_analyze_1): Split out main part of df_analyze.
5368 (df_analyze): Adjust.
5369 (loop_inverted_post_order_compute): New function.
5370 (loop_post_order_compute): Likewise.
5371 (df_analyze_loop): New function avoiding whole-function
5373 * loop-invariant.c (find_defs): Use df_analyze_loop.
5374 (find_invariants): Adjust.
5375 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5377 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5379 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5380 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5382 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5384 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5385 traversal when removing references.
5387 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5390 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5392 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5395 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5396 pushing a reload for an autoinc when we had previously reloaded an
5397 inner part of the address.
5399 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5401 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5403 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5404 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5405 when not giving up or versioning for alias only because of
5407 (vect_analyze_data_ref_dependences): Set to true.
5408 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5410 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5411 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5415 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5418 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5419 operand 0 predicate for gathers, use a new pseudo as subtarget.
5421 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5424 * lra-constraints.c (process_alt_operands): Add printing debug info.
5425 Check absence of input/output reloads for matched operands too.
5427 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5429 PR rtl-optimization/59835
5430 * ira.c (ira_init_register_move_cost): Increase cost for
5433 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5435 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5437 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5440 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5441 non-register objects. Use gen_(high/low)part more consistently.
5444 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5447 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5448 endian support, remove tests for WORDS_BIG_ENDIAN.
5449 (p8_mfvsrd_3_<mode>): Likewise.
5450 (reload_gpr_from_vsx<mode>): Likewise.
5451 (reload_gpr_from_vsxsf): Likewise.
5452 (p8_mfvsrd_4_disf): Likewise.
5454 2014-01-16 Richard Biener <rguenther@suse.de>
5456 PR rtl-optimization/46590
5457 * lcm.c (compute_antinout_edge): Use postorder iteration.
5458 (compute_laterin): Use inverted postorder iteration.
5460 2014-01-16 Nick Clifton <nickc@redhat.com>
5463 * varasm.c (output_constant): Return the number of bytes actually
5465 (output_constructor_array_range): Update the field size with the
5466 number of bytes emitted by output_constant.
5467 (output_constructor_regular_field): Likewise. Also do not
5468 complain if the total number of bytes emitted is now greater
5469 than the expected fieldpos.
5470 * output.h (output_constant): Update prototype and descriptive comment.
5472 2014-01-16 Marek Polacek <polacek@redhat.com>
5475 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5476 it is error_mark_node.
5478 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5480 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5481 VALID_AVX256_REG_OR_OI_MODE.
5483 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5485 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5486 current procedure should be profiled.
5488 2014-01-15 Andrew Pinski <apinski@cavium.com>
5490 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5491 of moving from/to the STACK_REG register class.
5493 2014-01-15 Richard Henderson <rth@redhat.com>
5496 * reginfo.c (global_regs_decl): Globalize.
5497 * rtl.h (global_regs_decl): Declare.
5498 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5499 reserved via global_regs.
5501 2014-01-15 Teresa Johnson <tejohnson@google.com>
5503 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5505 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5507 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5508 and vmulosh rather than call gen_vec_widen_smult_*.
5509 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5510 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5511 (vec_widen_smult_even_v16qi): Likewise.
5512 (vec_widen_umult_even_v8hi): Likewise.
5513 (vec_widen_smult_even_v8hi): Likewise.
5514 (vec_widen_umult_odd_v16qi): Likewise.
5515 (vec_widen_smult_odd_v16qi): Likewise.
5516 (vec_widen_umult_odd_v8hi): Likewise.
5517 (vec_widen_smult_odd_v8hi): Likewise.
5518 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5519 vmuloub rather than call gen_vec_widen_umult_*.
5520 (vec_widen_umult_lo_v16qi): Likewise.
5521 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5522 vmulosb rather than call gen_vec_widen_smult_*.
5523 (vec_widen_smult_lo_v16qi): Likewise.
5524 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5525 rather than call gen_vec_widen_umult_*.
5526 (vec_widen_umult_lo_v8hi): Likewise.
5527 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5528 rather than call gen_vec_widen_smult_*.
5529 (vec_widen_smult_lo_v8hi): Likewise.
5531 2014-01-15 Jeff Law <law@redhat.com>
5533 PR tree-optimization/59747
5534 * ree.c (find_and_remove_re): Properly handle case where a second
5535 eliminated extension requires widening a copy created for elimination
5536 of a prior extension.
5537 (combine_set_extension): Ensure that the number of hard regs needed
5538 for a destination register does not change when we widen it.
5540 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5542 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5543 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5544 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5545 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5546 (avr-*-rtems*): Likewise.
5547 (bfin*-rtems*): Likewise.
5548 (moxie-*-rtems*): Likewise.
5549 (h8300-*-rtems*): Likewise.
5550 (i[34567]86-*-rtems*): Likewise.
5551 (lm32-*-rtems*): Likewise.
5552 (m32r-*-rtems*): Likewise.
5553 (m68k-*-rtems*): Likewise.
5554 (microblaze*-*-rtems*): Likewise.
5555 (mips*-*-rtems*): Likewise.
5556 (powerpc-*-rtems*): Likewise.
5557 (sh-*-rtems*): Likewise.
5558 (sparc-*-rtems*): Likewise.
5559 (sparc64-*-rtems*): Likewise.
5560 (v850-*-rtems*): Likewise.
5561 (m32c-*-rtems*): Likewise.
5563 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5565 PR rtl-optimization/59511
5566 * ira.c (ira_init_register_move_cost): Use memory costs for some
5567 cases of register move cost calculations.
5568 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5569 instead of BB frequency.
5570 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5571 * lra-assigns.c (find_hard_regno_for): Ditto.
5573 2014-01-15 Richard Biener <rguenther@suse.de>
5575 PR tree-optimization/59822
5576 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5577 (vectorizable_load): Use it to hoist defs of uses of invariant
5578 loads out of the loop.
5580 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5581 Kugan Vivekanandarajah <kuganv@linaro.org>
5584 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5587 2014-01-15 Richard Biener <rguenther@suse.de>
5589 PR rtl-optimization/59802
5590 * lcm.c (compute_available): Use inverted postorder to seed
5591 the initial worklist.
5593 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5596 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5597 ADDR_REGS for invalid symrefs in non-PIC code.
5599 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5602 * builtins.c (determine_block_size): Initialize *probable_max_size
5603 even if len_rtx is CONST_INT.
5605 2014-01-14 Andrew Pinski <apinski@cavium.com>
5607 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5608 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5609 (cortexa53_tunings): Likewise.
5610 (aarch64_sched_issue_rate): New function.
5611 (TARGET_SCHED_ISSUE_RATE): Define.
5613 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5615 * ira-costs.c (find_costs_and_classes): Add missed
5616 ira_init_register_move_cost_if_necessary.
5618 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5621 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5623 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5626 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5627 to indicate if type is used for function return value. Warn ABI
5628 change if the vector mode isn't available for function return value.
5629 (ix86_function_arg_advance): Pass false to type_natural_mode.
5630 (ix86_function_arg): Likewise.
5631 (ix86_gimplify_va_arg): Likewise.
5632 (function_arg_32): Don't warn ABI change.
5633 (ix86_function_value): Pass true to type_natural_mode.
5634 (ix86_return_in_memory): Likewise.
5635 (ix86_struct_value_rtx): Removed.
5636 (TARGET_STRUCT_VALUE_RTX): Likewise.
5638 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5640 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5641 converting a conditional jump into a conditional return.
5643 2014-01-14 Richard Biener <rguenther@suse.de>
5645 PR tree-optimization/58921
5646 PR tree-optimization/59006
5647 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5648 hoisting invariant stmts.
5649 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5650 invariant loads on the preheader edge if possible.
5652 2014-01-14 Joey Ye <joey.ye@arm.com>
5654 * doc/plugin.texi (Building GCC plugins): Update to C++.
5656 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5658 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5659 (_mm_rcp28_round_ss): Ditto.
5660 (_mm_rsqrt28_round_sd): Ditto.
5661 (_mm_rsqrt28_round_ss): Ditto.
5662 (_mm_rcp28_sd): Ditto.
5663 (_mm_rcp28_ss): Ditto.
5664 (_mm_rsqrt28_sd): Ditto.
5665 (_mm_rsqrt28_ss): Ditto.
5666 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5667 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5668 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5669 (IX86_BUILTIN_RCP28SD): Ditto.
5670 (IX86_BUILTIN_RCP28SS): Ditto.
5671 (IX86_BUILTIN_RSQRT28SD): Ditto.
5672 (IX86_BUILTIN_RSQRT28SS): Ditto.
5673 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5674 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5675 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5676 (ix86_expand_special_args_builtin): Expand new FTYPE.
5677 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5678 (srcp14<mode>): Make insn unary.
5679 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5680 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5681 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5682 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5683 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5684 Fix rounding: make it SAE only.
5685 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5687 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5689 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5690 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5691 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5692 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5693 (round_saeonly_mask_scalar_operand4): Ditto.
5694 (round_saeonly_mask_scalar_op3): Ditto.
5695 (round_saeonly_mask_scalar_op4): Ditto.
5697 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5699 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5700 Implement -maltivec=be for vec_insert and vec_extract.
5702 2014-01-10 DJ Delorie <dj@redhat.com>
5704 * config/msp430/msp430.md (call_internal): Don't allow memory
5705 references with SP as the base register.
5706 (call_value_internal): Likewise.
5707 * config/msp430/constraints.md (Yc): New. For memory references
5708 that don't use SP as a base register.
5710 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5711 "an integer without a # prefix"
5712 * config/msp430/msp430.md (epilogue_helper): Use it.
5714 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5717 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5718 AVX512F gather builtins.
5719 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5720 on gather decls with INTEGER_TYPE masktype.
5721 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5722 directly into the builtin rather than hoisting it before loop.
5724 PR tree-optimization/59387
5725 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5726 (scev_const_prop): If folded_casts and type has undefined overflow,
5727 use force_gimple_operand instead of force_gimple_operand_gsi and
5728 for each added stmt if it is assign with
5729 arith_code_with_undefined_signed_overflow, call
5730 rewrite_to_defined_overflow.
5731 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5732 gimple-fold.h instead.
5733 (arith_code_with_undefined_signed_overflow,
5734 rewrite_to_defined_overflow): Moved to ...
5735 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5736 rewrite_to_defined_overflow): ... here. No longer static.
5737 Include gimplify-me.h.
5738 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5739 rewrite_to_defined_overflow): New prototypes.
5741 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5743 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5745 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5747 * builtins.c (get_object_alignment_2): Minor tweak.
5748 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5750 2014-01-13 Christian Bruel <christian.bruel@st.com>
5752 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5753 optimized non constant lengths.
5755 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5758 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5759 load as __atomic_load_N if possible.
5761 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5763 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5765 (rs6000_expand_builtin): Adjust call.
5767 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5770 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5771 * config/rs6000/rs6000.c: Include target-globals.h.
5772 (rs6000_set_current_function): Instead of doing target_reinit
5773 unconditionally, use save_target_globals_default_opts and
5774 restore_target_globals.
5776 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5778 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5779 (rs6000_expand_builtin): Handle mffs and mtfsf.
5780 (rs6000_init_builtins): Define mffs and mtfsf.
5781 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5782 (rs6000_mffs): New pattern.
5783 (rs6000_mtfsf): New pattern.
5785 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5787 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5788 Start narrowing with START. Apply candidate-use pair
5789 and check overall cost in narrowing.
5790 (iv_ca_prune): Pass new argument.
5792 2014-01-10 Jeff Law <law@redhat.com>
5795 * ree.c (combine_reaching_defs): Ensure the defining statement
5796 occurs before the extension when optimizing extensions with
5797 different source and destination hard registers.
5799 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5802 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5803 vtables into the type inheritance graph.
5805 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5807 PR rtl-optimization/59754
5808 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5809 modes in the REGNO != REGNO case.
5811 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5813 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5815 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5817 PR tree-optimization/59745
5818 * tree-predcom.c (tree_predictive_commoning_loop): Call
5819 free_affine_expand_cache if giving up because components is NULL.
5821 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5822 GC in payload of target_globals struct instead of allocating them on
5823 the heap and the larger structs separately using GC.
5824 * target-globals.h (struct target_globals): Make regs, hard_regs,
5825 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5826 of GTY((skip)) and change type to void *.
5827 (reset_target_globals): Cast loads from those fields to corresponding
5830 2014-01-10 Steve Ellcey <sellcey@mips.com>
5833 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5834 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5835 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5837 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5840 * aarch64-modes.def (CC_Zmode): New flags mode.
5841 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5842 represents an equality.
5843 (aarch64_get_condition_code): Handle CC_Zmode.
5844 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5846 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5848 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5849 extraction in good case.
5851 2014-01-10 Richard Biener <rguenther@suse.de>
5853 PR tree-optimization/59374
5854 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5855 checking after SLP discovery. Mark stmts not participating
5856 in any SLP instance properly.
5858 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5860 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5861 when handling a SET rtx.
5863 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5865 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5866 (cortex-a57): Likewise.
5867 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5869 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5871 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5872 non-iwmmxt builtins.
5874 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5878 * ipa-devirt.c record_target_from_binfo): Take as argument
5879 stack of binfos and lookup matching one for virtual inheritance.
5880 (possible_polymorphic_call_targets_1): Update.
5882 2014-01-10 Huacai Chen <chenhc@lemote.com>
5884 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5885 kernel strings for Loongson-2E/2F/3A.
5887 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5890 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5891 is_gimple_call before calling gimple_call_internal_p.
5893 2014-01-09 Steve Ellcey <sellcey@mips.com>
5895 * Makefile.in (TREE_FLOW_H): Remove.
5896 (TREE_SSA_H): Add file names from tree-flow.h.
5897 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5898 * tree.h: Remove tree-flow.h reference.
5899 * hash-table.h: Remove tree-flow.h reference.
5900 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5901 reference with tree-ssa-loop.h.
5903 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5905 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5906 default element-order behavior for -maltivec.
5907 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5909 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5910 when targeting big endian, at least for now.
5911 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5913 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5916 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5917 var satisfies use_register_for_decl, just take into account type
5918 alignment, rather than decl alignment.
5920 PR tree-optimization/59622
5921 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5922 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5923 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5924 Don't devirtualize for inplace at all. For targets.length () == 1,
5925 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5927 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5929 * config/i386/i386.md (cpu): Remove the unused btver1.
5931 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5933 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5935 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5938 * tree-core.h (struct target_globals): New forward declaration.
5939 (struct tree_target_option): Add globals field.
5940 * tree.h (TREE_TARGET_GLOBALS): Define.
5941 (prepare_target_option_nodes_for_pch): New prototype.
5942 * target-globals.h (struct target_globals): Define even if
5944 * tree.c (prepare_target_option_node_for_pch,
5945 prepare_target_option_nodes_for_pch): New functions.
5946 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5947 * config/i386/i386.c: Include target-globals.h.
5948 (ix86_set_current_function): Instead of doing target_reinit
5949 unconditionally, use save_target_globals_default_opts and
5950 restore_target_globals.
5952 2014-01-09 Richard Biener <rguenther@suse.de>
5954 PR tree-optimization/59715
5955 * tree-cfg.h (split_critical_edges): Declare.
5956 * tree-cfg.c (split_critical_edges): Export.
5957 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5959 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5961 * cfgexpand.c (expand_stack_vars): Optionally disable
5962 asan stack protection.
5963 (expand_used_vars): Likewise.
5964 (partition_stack_vars): Likewise.
5965 * asan.c (asan_emit_stack_protection): Optionally disable
5966 after return stack usage.
5967 (instrument_derefs): Optionally disable memory access instrumentation.
5968 (instrument_builtin_call): Likewise.
5969 (instrument_strlen_call): Likewise.
5970 (asan_protect_global): Optionally disable global variables protection.
5971 * doc/invoke.texi: Added doc for new options.
5972 * params.def: Added new options.
5973 * params.h: Likewise.
5975 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5977 PR rtl-optimization/59724
5978 * ifcvt.c (cond_exec_process_if_block): Don't call
5979 flow_find_head_matching_sequence with 0 longest_match.
5980 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5981 non-active insns if !stop_after.
5982 (try_head_merge_bb): Revert 2014-01-07 changes.
5984 2014-01-08 Jeff Law <law@redhat.com>
5986 * ree.c (get_sub_rtx): New function, extracted from...
5987 (merge_def_and_ext): Here.
5988 (combine_reaching_defs): Use get_sub_rtx.
5990 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5992 * cgraph.h (varpool_variable_node): Do not choke on null node.
5994 2014-01-08 Catherine Moore <clm@codesourcery.com>
5996 * config/mips/mips.md (simple_return): Attempt to use JRC
5998 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6000 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6002 PR rtl-optimization/59137
6003 * reorg.c (steal_delay_list_from_target): Call update_block for
6005 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6007 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6009 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6010 two duplicate entries.
6012 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6015 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6017 * config/mips/mips.c (mips_truncated_op_cost): New function.
6018 (mips_rtx_costs): Adjust test for BADDU.
6019 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6021 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6023 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6024 (*baddu_si): ...this new pattern.
6026 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6029 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6031 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6034 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6036 (expand_expr, expand_normal): Adjust.
6037 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6038 inner_reference_p. Use inner_reference_p to expand inner references.
6039 (store_expr): Adjust.
6040 * cfgexpand.c (expand_call_stmt): Adjust.
6042 2014-01-08 Rong Xu <xur@google.com>
6044 * gcov-io.c (gcov_var): Move from gcov-io.h.
6045 (gcov_position): Ditto.
6046 (gcov_is_error): Ditto.
6047 (gcov_rewrite): Ditto.
6048 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6049 only part to libgcc/libgcov.h.
6051 2014-01-08 Marek Polacek <polacek@redhat.com>
6054 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6056 2014-01-08 Marek Polacek <polacek@redhat.com>
6059 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6061 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6063 PR rtl-optimization/59649
6064 * stor-layout.c (get_mode_bounds): For BImode return
6065 0 and STORE_FLAG_VALUE.
6067 2014-01-08 Richard Biener <rguenther@suse.de>
6070 * gimple.h (is_gimple_builtin_call): Remove.
6071 (gimple_builtin_call_types_compatible_p): New.
6072 (gimple_call_builtin_p): New overload.
6073 * gimple.c (is_gimple_builtin_call): Remove.
6074 (validate_call): Rename to ...
6075 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6077 (validate_type): New static function.
6078 (gimple_call_builtin_p): New overload and adjust.
6079 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6080 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6081 (gimple_fold_stmt_to_constant_1): Likewise.
6082 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6084 2014-01-08 Richard Biener <rguenther@suse.de>
6087 * gimplify.c (gimplify_expr): Gimplify register-register type
6088 VIEW_CONVERT_EXPRs to separate stmts.
6090 2014-01-07 Jeff Law <law@redhat.com>
6093 * ree.c (combine_set_extension): Handle case where source
6094 and destination registers in an extension insn are different.
6095 (combine_reaching_defs): Allow source and destination registers
6096 in extension to be different under limited circumstances.
6097 (add_removable_extension): Remove restriction that the
6098 source and destination registers in the extension are the same.
6099 (find_and_remove_re): Emit a copy from the extension's
6100 destination to its source after the defining insn if
6101 the source and destination registers are different.
6104 * ifcvt.c (merge_if_block): If we are merging a block with more than
6105 one successor with a block with no successors, remove any BARRIER
6106 after the second block.
6108 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6110 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6112 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6115 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6116 for 14-bit register offsets when INT14_OK_STRICT is false.
6118 2014-01-07 Roland Stigge <stigge@antcom.de>
6119 Michael Meissner <meissner@linux.vnet.ibm.com>
6122 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6123 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6125 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6127 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6130 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6132 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6133 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6134 rtx is const0_rtx or not.
6136 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6139 * target-globals.c (save_target_globals): Remove this_fn_optab
6141 * toplev.c: Include optabs.h.
6142 (target_reinit): Temporarily restore the global options if another
6143 set of options are in force.
6145 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6147 PR rtl-optimization/58668
6148 * cfgcleanup.c (flow_find_cross_jump): Don't count
6149 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6150 to determine what is counted.
6151 (flow_find_head_matching_sequence): Use active_insn_p to determine
6153 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6155 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6156 determine what is counted.
6158 PR tree-optimization/59643
6159 * tree-predcom.c (split_data_refs_to_components): If one dr is
6160 read and one write, determine_offset fails and the write isn't
6161 in the bad component, just put the read into the bad component.
6163 2014-01-07 Mike Stump <mikestump@comcast.net>
6164 Jakub Jelinek <jakub@redhat.com>
6167 * tree-core.h (struct tree_optimization_option): Change optabs
6168 type from unsigned char * to void *.
6169 * optabs.c (init_tree_optimization_optabs): Adjust
6170 TREE_OPTIMIZATION_OPTABS initialization.
6172 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6175 * config/i386/i386.h (struct machine_function): Add
6176 no_drap_save_restore field.
6177 * config/i386/i386.c (ix86_save_reg): Use
6178 !cfun->machine->no_drap_save_restore instead of
6179 crtl->stack_realign_needed.
6180 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6181 this function clears frame_pointer_needed. Set
6182 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6183 and DRAP reg is needed.
6185 2014-01-06 Marek Polacek <polacek@redhat.com>
6188 * doc/implement-c.texi: Mention that other integer types are
6189 permitted as bit-field types in strictly conforming mode.
6191 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6193 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6196 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6198 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6200 2014-01-06 Martin Jambor <mjambor@suse.cz>
6203 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6205 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6207 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6211 * var-tracking.c (add_stores): Preserve the value of the source even if
6212 we don't record the store.
6214 2014-01-06 Terry Guo <terry.guo@arm.com>
6216 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6218 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6221 * config/darwin.c (darwin_function_section): Adjust return values to
6222 correspond to optimisation changes made in r206070.
6224 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6226 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6227 from prefetch_block tune setting.
6228 (nocona_cost): Correct size of prefetch block to 64.
6230 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6232 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6233 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6234 used to save the static chain register in the computation of the offset
6235 from which the FP registers need to be restored.
6237 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6239 PR tree-optimization/59519
6240 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6241 ICE if get_current_def (current_new_name) is already non-NULL, as long
6242 as it is a phi result of some other phi in *new_exit_bb that has
6245 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6246 or vmovdqu* for misaligned_operand.
6247 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6248 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6249 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6250 aligned_mem for AVX512F masked aligned load and store builtins and for
6253 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6255 PR tree-optimization/59651
6256 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6257 Address range for negative step should be added by TYPE_SIZE_UNIT.
6259 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6261 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6262 overlapping registers also for registers other than the stack pointer.
6264 2014-01-03 Marek Polacek <polacek@redhat.com>
6267 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6270 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6273 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6276 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6277 (push splitter): Use <P:MODE_SIZE> instead of
6278 GET_MODE_SIZE (<P:MODE>mode).
6279 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6280 (mov -1, reg peephole2): Likewise.
6281 * config/i386/sse.md (*mov<mode>_internal,
6282 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6283 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6284 *<code><mode>3, *andnot<mode>3<mask_name>,
6285 <mask_codefor><code><mode>3<mask_name>): Likewise.
6286 * config/i386/subst.md (mask_mode512bit_condition,
6287 sd_mask_mode512bit_condition): Likewise.
6289 2014-01-02 Xinliang David Li <davidxl@google.com>
6291 PR tree-optimization/59303
6292 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6293 (dump_predicates): Better output format.
6294 (pred_equal_p): New function.
6295 (is_neq_relop_p): Ditto.
6296 (is_neq_zero_form_p): Ditto.
6297 (pred_expr_equal_p): Ditto.
6298 (pred_neg_p): Ditto.
6299 (simplify_pred): Ditto.
6300 (simplify_preds_2): Ditto.
6301 (simplify_preds_3): Ditto.
6302 (simplify_preds_4): Ditto.
6303 (simplify_preds): Ditto.
6305 (push_to_worklist): Ditto.
6306 (get_pred_info_from_cmp): Ditto.
6307 (is_degenerated_phi): Ditto.
6308 (normalize_one_pred_1): Ditto.
6309 (normalize_one_pred): Ditto.
6310 (normalize_one_pred_chain): Ditto.
6311 (normalize_preds): Ditto.
6312 (normalize_cond_1): Remove function.
6313 (normalize_cond): Ditto.
6314 (is_gcond_subset_of): Ditto.
6315 (is_subset_of_any): Ditto.
6316 (is_or_set_subset_of): Ditto.
6317 (is_and_set_subset_of): Ditto.
6318 (is_norm_cond_subset_of): Ditto.
6319 (pred_chain_length_cmp): Ditto.
6320 (convert_control_dep_chain_into_preds): Type change.
6321 (find_predicates): Ditto.
6322 (find_def_preds): Ditto.
6323 (destroy_predicates_vecs): Ditto.
6324 (find_matching_predicates_in_rest_chains): Ditto.
6325 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6326 (is_pred_expr_subset): Ditto.
6327 (is_pred_chain_subset_of): Ditto.
6328 (is_included_in): Ditto.
6329 (is_superset_of): Ditto.
6331 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6333 Update copyright years.
6335 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6337 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6338 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6339 config/arc/arc.md, config/arc/arc.opt,
6340 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6341 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6342 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6343 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6344 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6345 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6346 vtable-verify.c, vtable-verify.h: Use the standard form for the
6349 2014-01-02 Tobias Burnus <burnus@net-b.de>
6351 * gcc.c (process_command): Update copyright notice dates.
6352 * gcov-dump.c: Ditto.
6354 * doc/cpp.texi: Bump @copying's copyright year.
6355 * doc/cppinternals.texi: Ditto.
6356 * doc/gcc.texi: Ditto.
6357 * doc/gccint.texi: Ditto.
6358 * doc/gcov.texi: Ditto.
6359 * doc/install.texi: Ditto.
6360 * doc/invoke.texi: Ditto.
6362 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6364 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6366 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6368 * config/i386/sse.md (*mov<mode>_internal): Guard
6369 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6371 PR rtl-optimization/59647
6372 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6373 new_rtx into UNSIGNED_FLOAT rtxes.
6375 Copyright (C) 2014 Free Software Foundation, Inc.
6377 Copying and distribution of this file, with or without modification,
6378 are permitted in any medium without royalty provided the copyright
6379 notice and this notice are preserved.