1 2014-04-14 Richard Biener <rguenther@suse.de>
3 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
5 (expand_switch_using_bit_tests_p): Likewise.
6 (process_switch): Compute and pass on speed_p based on the
8 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
9 optimize_bb_for_speed_p.
11 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
13 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
14 * function.h (struct function): Rename has_force_vect_loops into
15 has_force_vectorize_loops.
16 * lto-streamer-in.c (input_cfg): Adjust for renaming.
17 (input_struct_function_base): Likewise.
18 * lto-streamer-out.c (output_cfg): Likewise.
19 (output_struct_function_base): Likewise.
20 * omp-low.c (expand_omp_simd): Likewise.
21 * tree-cfg.c (move_sese_region_to_fn): Likewise.
22 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
23 (version_loop_for_if_conversion): Likewise.
24 (tree_if_conversion): Likewise.
25 (main_tree_if_conversion): Likewise.
26 (gate_tree_if_conversion): Likewise.
27 * tree-inline.c (copy_loops): Likewise.
28 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
29 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
30 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
31 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
32 * tree-vectorizer.c (vectorize_loops): Likewise.
33 * tree-vectorizer.h (unlimited_cost_model): Likewise.
35 2014-04-14 Richard Biener <rguenther@suse.de>
38 * lto-streamer-out.c (wrap_refs): New function.
39 (lto_output): Wrap symbol references in global initializes in
40 type-preserving MEM_REFs.
42 2014-04-14 Christian Bruel <christian.bruel@st.com>
44 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
46 2014-04-14 Christian Bruel <christian.bruel@st.com>
48 * config/sh/sh.md (setmemqi): New expand pattern.
49 * config/sh/sh.h (CLEAR_RATIO): Define.
50 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
51 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
53 2014-04-14 Richard Biener <rguenther@suse.de>
56 * fold-const.c (negate_expr_p): Don't negate directional rounding
58 (fold_negate_expr): Likewise.
60 2014-04-14 Richard Biener <rguenther@suse.de>
62 PR tree-optimization/59817
63 PR tree-optimization/60453
64 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
65 recursion to catch all CHRECs in the scalar evolution and restrict
66 the predicate for the remains appropriately.
68 2014-04-12 Catherine Moore <clm@codesourcery.com>
70 * config/mips/constraints.md: Add new register constraint "kb".
71 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
72 (*movhi_internal): Likewise.
73 (*movqi_internal): Likewise.
74 * config/mips/mips.h (M16_STORE_REGS): New register class.
75 (REG_CLASS_NAMES): Add M16_STORE_REGS.
76 (REG_CLASS_CONTENTS): Likewise.
77 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
79 2014-04-11 Tobias Burnus <burnus@net-b.de>
82 * doc/invoke.texi (-Wformat-signedness): Document it.
83 (Wformat=2): Mention that this enables -Wformat-signedness.
85 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
87 * common/config/epiphany/epiphany-common.c
88 (epiphany_option_optimization_table): Enable section anchors by
89 default at -O1 or higher.
90 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
91 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
92 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
93 carries no extra cost.
94 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
95 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
96 * config/epiphany/predicates.md (memclob_operand): New predicate.
97 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
98 Use memclob_operand predicate and X constraint for operand 3.
100 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
102 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
103 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
106 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
108 PR rtl-optimization/60651
109 * mode-switching.c (optimize_mode_switching): Make sure to emit
110 sets of a lower numbered entity before sets of a higher numbered
111 entity to a mode of the same or lower priority.
112 When creating a seginfo for a basic block that starts with a code
113 label, move the insertion point past the code label.
114 (new_seginfo): Document and enforce requirement that
115 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
116 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
117 * doc/tm.texi: Regenerate.
119 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
122 * config/arc/arc.c (arc_save_restore): Fix assert typo.
124 2013-04-11 Jakub Jelinek <jakub@redhat.com>
126 * BASE-VER: Set to 4.10.0.
128 2014-04-11 Tobias Burnus <burnus@net-b.de>
131 * doc/bugreport.texi (Bugs): Remove nodes pointing to the
133 * doc/gcc.texi (Service): Update description in the @menu
134 * doc/invoke.texi (Option Summary): Remove misplaced and
137 2014-04-11 Steve Ellcey <sellcey@mips.com>
138 Jakub Jelinek <jakub@redhat.com>
141 * expr.c (convert_move): Use emit_store_flag_force instead of
142 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
145 2014-04-11 Richard Biener <rguenther@suse.de>
148 * varasm.c (assemble_alias): Avoid endless error reporting
149 recursion by setting TREE_ASM_WRITTEN.
151 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
153 * config/s390/s390.md: Add a splitter for NOT rtx.
155 2014-04-11 Jakub Jelinek <jakub@redhat.com>
157 PR rtl-optimization/60663
158 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
161 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
162 Jakub Jelinek <jakub@redhat.com>
165 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
168 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
171 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
172 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
173 ameliorating the cases where it can be.
175 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
178 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
180 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
181 (loadsync_<mode>): Change mode.
182 (load_quadpti, store_quadpti): New.
183 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
184 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
185 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
187 2014-04-09 Cong Hou <congh@google.com>
190 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
193 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
195 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
196 instead of vnor to exploit possible fusion opportunity in the
198 (altivec_expand_vec_perm_const_le): Likewise.
200 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
202 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
203 (loadsync_<mode>): Change mode.
204 (load_quadpti, store_quadpti): New.
205 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
206 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
208 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
211 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
212 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
213 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
215 2014-04-08 Richard Biener <rguenther@suse.de>
218 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
219 a 64bit widest int print double-int similar to on HWI64 hosts.
221 2014-04-08 Richard Biener <rguenther@suse.de>
223 PR tree-optimization/60785
224 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
225 default defs properly.
227 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
229 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
232 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
234 * ipa-devirt.c (maybe_record_node): When node is not recorded,
235 set completep to false rather than true.
237 2014-04-07 Douglas B Rupp <rupp@adacore.com>
240 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
241 ARM_TARGET2_DWARF_FORMAT.
243 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
246 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
247 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
250 2014-04-07 Richard Biener <rguenther@suse.de>
252 PR tree-optimization/60766
253 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
254 (may_eliminate_iv): Convert cand_value_at result to desired type.
256 2014-04-07 Jason Merrill <jason@redhat.com>
259 * common.opt (-fno-gnu-unique): Add.
260 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
262 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
264 * haifa-sched.c: Fix outdated function reference and minor
265 grammar errors in introductory comment.
267 2014-04-07 Richard Biener <rguenther@suse.de>
270 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
272 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
274 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
277 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
278 size accounting for thunks.
279 (pa_asm_output_mi_thunk): Use final_start_function() and
280 final_end_function() to output function start and end directives.
282 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
284 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
285 specific ISA/ feature information. Remove short_sp and errata_skip ds.
286 Add avr_device_specific_features enum to have device specific info.
287 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
288 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
289 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
290 updated device specific info.
291 * config/avr/avr-mcus.def: Merge device specific details to
293 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
295 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
296 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
297 assembler if RMW isa supported by current device.
298 * config/avr/genmultilib.awk: Update as device info structure changed.
299 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
301 2014-04-04 Cong Hou <congh@google.com>
303 PR tree-optimization/60656
304 * tree-vect-stmts.c (supportable_widening_operation):
305 Fix a bug that elements in a vector with vect_used_by_reduction
306 property are incorrectly reordered when the operation on it is not
307 consistant with the one in reduction operation.
309 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
311 PR rtl-optimization/60155
312 * gcse.c (record_set_data): New function.
313 (single_set_gcse): New function.
314 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
315 (hoist_code): Likewise.
316 (get_pressure_class_and_nregs): Likewise.
318 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
320 * explow.c (probe_stack_range): Emit a final optimization blockage.
322 2014-04-04 Anthony Green <green@moxielogic.com>
324 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
327 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
330 * lto-cgraph.c (input_overwrite_node): Check that partitioning
331 flags are set only during streaming.
332 * ipa.c (process_references, walk_polymorphic_call_targets,
333 symtab_remove_unreachable_nodes): Drop bodies of always inline
334 after early inlining.
335 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
337 2014-04-04 Jakub Jelinek <jakub@redhat.com>
338 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
341 * dwarf2out.c (const_ok_for_output_1): Reject expressions
344 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
347 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
349 (cortex_a53_fdivd): Likewise.
351 2014-04-04 Martin Jambor <mjambor@suse.cz>
354 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
356 * cgraph.c (clone_of_p): Also return true if thunks match.
357 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
358 cgraph_function_or_thunk_node and an obsolete comment.
359 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
361 (build_function_decl_skip_args): Likewise.
362 (set_new_clone_decl_and_node_flags): New function.
363 (duplicate_thunk_for_node): Likewise.
364 (redirect_edge_duplicating_thunks): Likewise.
365 (cgraph_clone_node): New parameter args_to_skip, pass it to
366 redirect_edge_duplicating_thunks which is called instead of
367 cgraph_redirect_edge_callee.
368 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
369 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
371 2014-04-04 Jeff Law <law@redhat.com>
374 * config/arm/predicates.md (const_int_I_operand): New predicate.
375 (const_int_M_operand): Similarly.
376 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
378 (insv_t2, extv_reg, extzv_t2): Likewise.
379 (load_multiple_with_writeback): Similarly for const_int_I_operand.
380 (pop_multiple_with_writeback_and_return): Likewise.
381 (vfp_pop_multiple_with_writeback): Likewise
383 2014-04-04 Richard Biener <rguenther@suse.de>
386 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
387 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
389 * gimplify.h (gimple_add_tmp_var_fn): Declare.
390 * gimplify.c (gimple_add_tmp_var_fn): New function.
391 * gimple-expr.h (create_tmp_reg_fn): Declare.
392 * gimple-expr.c (create_tmp_reg_fn): New function.
393 * gimple-low.c (record_vars_into): Don't change cfun.
394 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
395 code generation without cfun.
397 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
400 * Makefile.in (install-driver): Fix shell scripting.
402 2014-04-03 Cong Hou <congh@google.com>
404 PR tree-optimization/60505
405 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
406 threshold of number of iterations below which no vectorization will be
408 * tree-vect-loop.c (new_loop_vec_info):
409 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
410 * tree-vect-loop.c (vect_analyze_loop_operations):
411 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
412 * tree-vect-loop.c (vect_transform_loop):
413 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
414 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
415 of iterations of the loop and see if we should build the epilogue.
417 2014-04-03 Richard Biener <rguenther@suse.de>
419 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
421 (streamer_tree_cache_create): Adjust.
422 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
423 to allow optional nodes array.
424 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
425 (streamer_tree_cache_append): Likewise.
426 (streamer_tree_cache_create): Create nodes array optionally
427 as specified by parameter.
428 * lto-streamer-out.c (create_output_block): Avoid maintaining
429 the node array in the writer cache.
430 (DFS_write_tree): Remove assertion.
431 (produce_asm_for_decls): Free the out decl state hash table
433 * lto-streamer-in.c (lto_data_in_create): Adjust for
434 streamer_tree_cache_create prototype change.
436 2014-04-03 Richard Biener <rguenther@suse.de>
438 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
439 set TREE_CHAIN to NULL_TREE.
441 2014-04-03 Richard Biener <rguenther@suse.de>
443 PR tree-optimization/60740
444 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
445 over all GIMPLE_COND operands.
447 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
449 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
450 (Weffc++): Remove Scott's numbering, merge lists and reference
456 * class.c (accessible_nvdtor_p): New.
457 (check_bases): Don't check base destructor here ...
458 (check_bases_and_members): ... check them here. Trigger on
459 Wnon-virtual-dtor flag.
460 (finish_struct_1): Use accessible_nvdtor_p.
463 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
464 * g++.dg/warn/Wnvdtor-2.C: New.
465 * g++.dg/warn/Wnvdtor-3.C: New.
466 * g++.dg/warn/Wnvdtor-4.C: New.
467 * g++.dg/warn/Weff1.C: Delete.
468 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
469 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
471 2014-04-03 Nick Clifton <nickc@redhat.com>
473 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
476 2014-04-03 Martin Jambor <mjambor@suse.cz>
478 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
479 mention gcc_unreachable before failing.
480 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
483 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
486 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
487 code and instead mark the context inconsistent.
488 (possible_polymorphic_call_targets): For inconsistent contexts
489 return empty complete list.
491 2014-04-02 Anthony Green <green@moxielogic.com>
493 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
494 (extendqisi2, extendhisi2): Define.
495 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
496 (WCHAR_TYPE): Change to unsigned int.
498 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
500 PR tree-optimization/60733
501 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
502 insertion point for PHI candidates to be the end of the feeding
503 block for the PHI argument.
505 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
507 PR rtl-optimization/60650
508 * lra-constraints.c (process_alt_operands): Decrease reject for
509 earlyclobber matching.
511 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
513 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
515 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
517 * config/spu/spu.c (pad_bb): Do not crash when the last
518 insn is CODE_FOR_blockage.
520 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
522 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
523 lies outside the target mode.
525 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
528 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
529 software floating point or no floating point registers, do not
530 allow any type in the FPRs. Eliminate a test for SPE SIMD types
531 in GPRs that occurs after we tested for GPRs that would never be
534 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
535 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
536 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
537 specifically allow DDmode, since that does not use the SPE SIMD
540 2014-04-02 Richard Biener <rguenther@suse.de>
543 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
544 MODE_INTs. Properly use negv_optab.
545 (expand_abs): Likewise.
547 2014-04-02 Richard Biener <rguenther@suse.de>
550 * Makefile.in (install-driver): Guard extra installs with special
553 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
555 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
558 2014-04-01 Richard Henderson <rth@redhat.com>
561 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
562 alternative enabled before register allocation.
564 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
566 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
567 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
569 (nios2_large_got_address): Remove unneeded 'sym' parameter.
570 (nios2_got_address): Update nios2_large_got_address call site.
571 (nios2_delegitimize_address): New function.
572 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
573 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
574 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
576 2014-04-01 Martin Husemann <martin@duskware.de>
578 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
581 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
583 PR rtl-optimization/60604
584 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
585 check from register_operand.
586 (register_operand): Redefine in terms of general_operand.
587 (nonmemory_operand): Use register_operand for the non-constant cases.
589 2014-04-01 Richard Biener <rguenther@suse.de>
591 * gimple.h (struct gimple_statement_base): Align subcode to
594 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
596 * doc/invoke.texi (mapp-regs): Clarify.
598 2014-03-31 Ulrich Drepper <drepper@gmail.com>
600 * config/i386/avx512fintrin.h (__v32hi): Define type.
602 (_mm512_set1_epi8): Define.
603 (_mm512_set1_epi16): Define.
604 (_mm512_set4_epi32): Define.
605 (_mm512_set4_epi64): Define.
606 (_mm512_set4_pd): Define.
607 (_mm512_set4_ps): Define.
608 (_mm512_setr4_epi64): Define.
609 (_mm512_setr4_epi32): Define.
610 (_mm512_setr4_pd): Define.
611 (_mm512_setr4_ps): Define.
612 (_mm512_setzero_epi32): Define.
614 2014-03-31 Martin Jambor <mjambor@suse.cz>
617 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
618 callsite_arguments_match_p. Updated all callers. Also check types of
619 corresponding formal parameters and actual arguments.
620 (not_all_callers_have_enough_arguments_p) Renamed to
621 some_callers_have_mismatched_arguments_p.
623 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
625 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
627 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
630 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
633 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
635 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
636 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
638 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
639 Use FMAMODE_NOVF512 mode iterator.
640 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
641 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
642 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
643 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
645 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
646 Use VF_128_256 mode iterator.
647 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
650 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
652 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
653 static chain if needed.
655 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
658 * lra-constraints.c (index_part_to_reg): New.
659 (process_address): Use it.
661 2014-03-27 Jeff Law <law@redhat.com>
662 Jakub Jelinek <jakub@redhat.com>
665 * expr.c (do_tablejump): Use simplify_gen_binary rather than
666 gen_rtx_{PLUS,MULT} to build up the address expression.
668 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
669 creating non-canonical RTL.
671 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
674 * ipa-inline.c (want_inline_small_function_p): Short circuit large
675 functions; reorganize to make cheap checks first.
676 (inline_small_functions): Do not estimate growth when dumping;
678 * ipa-inline.h (inline_summary): Add min_size.
679 (growth_likely_positive): New function.
680 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
681 (set_cond_stmt_execution_predicate): Cleanup.
682 (estimate_edge_size_and_time): Compute min_size.
683 (estimate_calls_size_and_time): Likewise.
684 (estimate_node_size_and_time): Likewise.
685 (inline_update_overall_summary): Update min_size.
686 (do_estimate_edge_time): Likewise.
687 (do_estimate_edge_size): Update.
688 (do_estimate_edge_hints): Update.
689 (growth_likely_positive): New function.
691 2014-03-28 Jakub Jelinek <jakub@redhat.com>
694 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
695 also if addr has VOIDmode.
697 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
699 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
700 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
702 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
703 instructions as well as AdvancedSIMD loads.
705 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
707 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
708 Use crypto_aese type.
709 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
710 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
711 crypto_aese, crypto_aesmc. Move to types.md.
712 * config/arm/types.md (crypto_aes): Split into crypto_aese,
714 * config/arm/iterators.md (crypto_type): Likewise.
716 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
718 * cgraph.c: Include expr.h and tree-dfa.h.
719 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
722 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
725 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
726 regs from checking multi-reg pseudos.
728 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
730 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
732 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
734 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
735 if it would clobber the stack pointer, even temporarily.
737 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
739 * mode-switching.c: Make small adjustments to the top comment.
741 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
743 * config/rs6000/constraints.md (wD constraint): New constraint to
744 match the constant integer to get the top DImode/DFmode out of a
745 vector in a VSX register.
747 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
748 match the constant integer to get the top DImode/DFmode out of a
749 vector in a VSX register.
751 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
754 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
757 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
758 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
760 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
761 Optimize vec_extract of 64-bit values, where the value being
762 extracted is in the top word, where we can use scalar
763 instructions. Add direct move and store support. Combine the big
764 endian/little endian vector select load support into a single insn.
765 (vsx_extract_<mode>_internal1): Likewise.
766 (vsx_extract_<mode>_internal2): Likewise.
767 (vsx_extract_<mode>_load): Likewise.
768 (vsx_extract_<mode>_store): Likewise.
769 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
770 combined into vsx_extract_<mode>_load.
771 (vsx_extract_<mode>_one_le): Likewise.
773 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
774 define the top 64-bit vector element.
776 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
779 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
780 Document vec_vbpermq builtin.
783 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
784 enable use of xxsldwi and xxpermdi builtin functions.
785 (vec_xxpermdi): Likewise.
787 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
788 Document use of vec_xxsldwi and vec_xxpermdi builtins.
790 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
792 PR rtl-optimization/60650
793 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
795 (find_spills_for): New.
796 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
797 Spill all pseudos on the second iteration.
799 2014-03-27 Marek Polacek <polacek@redhat.com>
802 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
805 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
807 * config/s390/s390.c (s390_can_use_return_insn): Check for
808 call-saved FPRs on 31 bit.
810 2014-03-27 Jakub Jelinek <jakub@redhat.com>
813 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
814 if they need regimplification, just drop them instead of
815 calling gimple_regimplify_operands on them.
817 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
820 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
821 (aarch64_frame_pointer_required): Adjust logic.
822 (aarch64_can_eliminate): Adjust logic.
823 (aarch64_override_options_after_change): Adjust logic.
825 2014-03-27 Dehao Chen <dehao@google.com>
827 * ipa-inline.c (early_inliner): Update node's inline info.
829 2014-03-26 Dehao Chen <dehao@google.com>
831 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
832 compiler inserted conditional jumps for NAN float check.
834 2014-03-26 Jakub Jelinek <jakub@redhat.com>
836 * ubsan.h (ubsan_create_data): Change second argument's type
837 to const location_t *.
838 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
840 (ubsan_create_data): Change second argument to const location_t *PLOC.
841 Create Loc field whenever PLOC is non-NULL.
842 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
843 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
847 * real.c (real_to_integer2): Change type of low to UHWI.
849 2014-03-26 Tobias Burnus <burnus@net-b.de>
851 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
852 (CILK_SELF_SPECS): New define.
853 (driver_self_specs): Use it.
855 2014-03-26 Richard Biener <rguenther@suse.de>
857 * tree-pretty-print.c (percent_K_format): Implement special
858 case for LTO and its stripped down BLOCK tree.
860 2014-03-26 Jakub Jelinek <jakub@redhat.com>
863 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
865 * tree-vrp.c (simplify_internal_call_using_ranges): If only
866 one range is range_int_cst_p, but not both, at least optimize
867 addition/subtraction of 0 and multiplication by 0 or 1.
868 * gimple-fold.c (gimple_fold_call): Fold
869 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
870 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
871 INTEGER_CSTs, try to fold at least x * 0 and y - y.
873 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
875 PR rtl-optimization/60452
876 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
877 <case REG>: Return 1 for invalid offsets from the frame pointer.
879 2014-03-26 Marek Polacek <polacek@redhat.com>
882 * doc/extend.texi (C Extensions): Mention variable-length arrays in
885 2014-03-26 Marek Polacek <polacek@redhat.com>
888 * doc/extend.texi (Designated Inits): Describe what happens to omitted
891 2014-03-26 Marek Polacek <polacek@redhat.com>
894 * ira-color.c (update_conflict_hard_regno_costs): Perform the
895 multiplication in unsigned type.
897 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
899 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
901 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
903 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
905 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
908 * cif-code.def (UNREACHABLE) New code.
909 * ipa-inline.c (inline_small_functions): Skip edges to
910 __builtlin_unreachable.
911 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
912 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
913 predicate to __bulitin_unreachable.
914 (set_cond_stmt_execution_predicate): Fix issue when
915 invert_tree_comparison returns ERROR_MARK.
916 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
917 propagate to inline clones.
918 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
920 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
921 * cgraphclones.c (cgraph_clone_node): If call destination is already
922 ureachable, do not redirect it back.
923 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
926 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
928 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
929 Do not modify inline clones.
931 2014-03-25 Jakub Jelinek <jakub@redhat.com>
933 * config/i386/i386.md (general_sext_operand): New mode attr.
934 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
935 don't generate (sign_extend (const_int)).
936 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
937 operands[2]. Use We constraint instead of <i> and
938 <general_sext_operand> predicate instead of <general_operand>.
939 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
940 * config/i386/constraints.md (We): New constraint.
941 * config/i386/predicates.md (x86_64_sext_operand,
942 sext_operand): New predicates.
944 2014-03-25 Martin Jambor <mjambor@suse.cz>
947 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
948 inconsistent devirtualizations to __builtin_unreachable.
950 2014-03-25 Marek Polacek <polacek@redhat.com>
953 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
955 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
957 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
958 order of elements for big-endian.
960 2014-03-25 Richard Biener <rguenther@suse.de>
963 * gimplify-me.c (gimple_regimplify_operands): Update the
966 2014-03-25 Martin Jambor <mjambor@suse.cz>
969 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
970 (lto_output_varpool_node): Likewise.
971 (input_overwrite_node): Likewise.
972 (input_varpool_node): Likewise.
974 2014-03-25 Richard Biener <rguenther@suse.de>
976 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
979 2014-03-25 Jakub Jelinek <jakub@redhat.com>
981 * combine.c (simplify_compare_const): Add MODE argument.
982 Handle mode_width 0 as very large mode_width.
983 (try_combine, simplify_comparison): Adjust callers.
985 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
986 type to avoid signed integer overflow.
987 * explow.c (plus_constant): Likewise.
989 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
991 * doc/generic.texi: Correct typos.
993 2014-03-24 Tobias Burnus <burnus@net-b.de>
995 * doc/invoke.texi (-flto): Expand section about
996 using static libraries with LTO.
998 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1000 PR rtl-optimization/60501
1001 * optabs.def (addptr3_optab): New optab.
1002 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1003 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1004 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1006 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1008 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1010 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1012 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1015 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1016 (_mm256_undefined_ps): Define.
1017 (_mm256_undefined_pd): Define.
1018 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1019 (_mm_undefined_pd): Define.
1020 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1021 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1022 (_mm512_undefined_ps): Define.
1023 (_mm512_undefined_pd): Define.
1024 Use _mm*_undefined_*.
1025 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1027 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1029 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1030 (lshr_simd): DI mode added.
1031 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1032 (aarch64_ushr_simddi): Likewise.
1033 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1034 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1035 (vshrd_n_u64): Likewise.
1037 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1039 * Makefile.in (s-macro_list): Depend on cc1.
1041 2014-03-23 Teresa Johnson <tejohnson@google.com>
1043 * ipa-utils.c (ipa_print_order): Use specified dump file.
1045 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1047 PR rtl-optimization/60601
1048 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1050 * gcc.c (eval_spec_function): Initialize save_growing_value.
1052 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1055 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1056 code == MINUS_EXPR, never swap op0 with op1.
1058 * toplev.c (init_local_tick): Avoid signed integer multiplication
1060 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1061 shift by first operand's bitsize.
1063 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1066 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1068 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1069 TARGET_ISA_64BIT_P(x).
1071 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1073 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1074 pattern for vector nor instead of subtract from splat(-1).
1075 (altivec_expand_vec_perm_const_le): Likewise.
1077 2014-03-21 Richard Henderson <rth@twiddle.net>
1080 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1081 related insns after epilogue_completed.
1083 2014-03-21 Martin Jambor <mjambor@suse.cz>
1086 * cgraph.h (symtab_node): New flag body_removed.
1087 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1088 when removing bodies.
1089 * symtab.c (dump_symtab_base): Dump body_removed flag.
1090 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1091 had their bodies removed.
1093 2014-03-21 Martin Jambor <mjambor@suse.cz>
1096 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1099 2014-03-21 Richard Biener <rguenther@suse.de>
1101 PR tree-optimization/60577
1102 * tree-core.h (struct tree_base): Document nothrow_flag use
1103 in VAR_DECL_NONALIASED.
1104 * tree.h (VAR_DECL_NONALIASED): New.
1105 (may_be_aliased): Adjust.
1106 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1108 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1110 * expr.c (expand_expr_real_1): Remove outdated comment.
1112 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1115 * ira.c (adjust_cleared_regs): Call copy_rtx on
1116 *reg_equiv[REGNO (loc)].src_p before passing it to
1117 simplify_replace_fn_rtx.
1120 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1121 into CONST, put pic register as first operand of PLUS. Use
1122 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1124 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1126 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1128 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1130 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1131 around for store forwarding issue in the FPU on the UT699.
1132 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1133 loads and operations if -mfix-ut699 is specified.
1134 (divtf3_hq): Tweak attribute.
1135 (sqrttf2_hq): Likewise.
1137 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1139 * calls.c (store_one_arg): Remove incorrect const qualification on the
1140 type of the temporary.
1141 * cfgexpand.c (expand_return): Likewise.
1142 * expr.c (expand_constructor): Likewise.
1143 (expand_expr_real_1): Likewise.
1145 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1147 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1150 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1153 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1155 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1157 * config/arm/aarch-common-protos.h
1158 (alu_cost_table): Fix spelling of "extend".
1159 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1161 2014-03-19 Richard Biener <rguenther@suse.de>
1164 * tree-core.h (tree_type_common): Re-order pointer members
1165 to reduce recursion depth during GC walks.
1167 2014-03-19 Marek Polacek <polacek@redhat.com>
1170 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1171 before accessing it.
1173 2014-03-19 Richard Biener <rguenther@suse.de>
1176 * lto-streamer-in.c (input_function): In WPA stage do not drop
1179 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1181 PR tree-optimization/60559
1182 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1183 with build_zero_cst assignment.
1185 2014-03-18 Kai Tietz <ktietz@redhat.com>
1187 PR rtl-optimization/56356
1188 * sdbout.c (sdbout_parms): Verify that parms'
1189 incoming argument is valid.
1190 (sdbout_reg_parms): Likewise.
1192 2014-03-18 Richard Henderson <rth@redhat.com>
1195 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1196 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1197 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1199 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1201 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1202 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1203 Italicize plugin event names in description. Explain that
1204 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1205 Remind that no GCC functions should be called after PLUGIN_FINISH.
1206 Explain what pragmas with expansion are.
1208 2014-03-18 Martin Liska <mliska@suse.cz>
1210 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1211 gimple call statement is update.
1212 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1213 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1215 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1218 * ubsan.c (ubsan_instrument_unreachable): Call
1219 initialize_sanitizer_builtins.
1220 (ubsan_pass): Likewise.
1223 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1224 varpool_finalize_decl instead of rest_of_decl_compilation.
1226 2014-03-18 Richard Biener <rguenther@suse.de>
1228 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1229 by using bitmap_and_compl instead of bitmap_and_compl_into.
1230 (df_rd_transfer_function): Likewise.
1232 2014-03-18 Richard Biener <rguenther@suse.de>
1234 * doc/lto.texi (fresolution): Fix typo.
1236 2014-03-18 Richard Biener <rguenther@suse.de>
1238 * doc/invoke.texi (flto): Update for changes in 4.9.
1240 2014-03-18 Richard Biener <rguenther@suse.de>
1242 * doc/loop.texi: Remove section on the removed lambda framework.
1243 Update loop docs with recent changes in preserving loop structure.
1245 2014-03-18 Richard Biener <rguenther@suse.de>
1247 * doc/lto.texi (-fresolution): Document.
1249 2014-03-18 Richard Biener <rguenther@suse.de>
1251 * doc/contrib.texi: Adjust my name.
1253 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1256 * internal-fn.c: Include diagnostic-core.h.
1257 (expand_BUILTIN_EXPECT): New function.
1258 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1259 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1260 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1261 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1263 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1264 Revert 3 argument __builtin_expect code.
1265 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1266 * gimple-fold.c (gimple_fold_call): Likewise.
1267 * tree.h (fold_builtin_expect): New prototype.
1268 * builtins.c (build_builtin_expect_predicate): Add predictor
1269 argument, if non-NULL, create 3 argument __builtin_expect.
1270 (fold_builtin_expect): No longer static. Add ARG2 argument,
1271 pass it through to build_builtin_expect_predicate.
1272 (fold_builtin_2): Adjust caller.
1273 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1274 * internal-fn.def (BUILTIN_EXPECT): New.
1276 2014-03-18 Tobias Burnus <burnus@net-b.de>
1279 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1280 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1281 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1283 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1286 * predict.c (combine_predictions_for_bb): Fix up formatting.
1287 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1288 fill what it points to if non-NULL.
1289 (tree_predict_by_opcode): Adjust caller, use the predictor.
1290 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1292 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1294 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1295 proper constant for the store mode.
1297 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1299 * symtab.c (change_decl_assembler_name): Fix transparent alias
1302 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1304 * config/aarch64/aarch64.c: Correct the comments about the
1305 aarch64 stack layout.
1307 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1309 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1310 check for GF_OMP_FOR_KIND_FOR.
1312 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1314 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1315 ymm and zmm register names.
1317 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1320 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1321 note creation for the 2010-08-31 changes.
1323 2014-03-17 Marek Polacek <polacek@redhat.com>
1326 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1327 as -fno-tree-loop-vectorize.
1328 (expand_omp_simd): Likewise.
1330 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1332 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1333 (eligible_for_call_delay): New prototype.
1334 * config/sparc/sparc.c (tls_call_delay): Rename into...
1335 (eligible_for_call_delay): ...this. Return false if the instruction
1336 cannot be put in the delay slot of a branch.
1337 (eligible_for_restore_insn): Simplify.
1338 (eligible_for_return_delay): Return false if the instruction cannot be
1339 put in the delay slot of a branch and simplify.
1340 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1341 put in the delay slot of a branch.
1342 * config/sparc/sparc.md (fix_ut699): New attribute.
1343 (tls_call_delay): Delete.
1344 (in_call_delay): Reimplement.
1345 (eligible_for_sibcall_delay): Rename into...
1346 (in_sibcall_delay): ...this.
1347 (eligible_for_return_delay): Rename into...
1348 (in_return_delay): ...this.
1349 (in_branch_delay): Reimplement.
1350 (in_uncond_branch_delay): Delete.
1351 (in_annul_branch_delay): Delete.
1353 2014-03-14 Richard Henderson <rth@redhat.com>
1356 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1357 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1358 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1359 (floathi splitters): Remove.
1360 (float<SWI48x>xf2): New pattern.
1361 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1362 code that tried to handle DImode for 32-bit, but which was excluded
1363 by the pattern's condition. Drop allocation of stack temporary.
1364 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1365 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1366 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1367 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1368 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1369 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1370 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1371 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1372 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1373 (*float<SWI48x><X87MODEF>2_i387): Remove.
1374 (all float _with_temp splitters): Remove.
1375 (*float<SWI48x><MODEF>2_i387): New pattern.
1376 (*float<SWI48><MODEF>2_sse): New pattern.
1377 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1378 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1380 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1381 Marek Polacek <polacek@redhat.com>
1384 * common.opt (dump_base_name_prefixed): New Variable.
1385 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1386 if x_dump_base_name_prefixed is already set, set it at the end.
1388 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1390 PR rtl-optimization/60508
1391 * lra-constraints.c (get_reload_reg): Add new parameter
1393 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1394 Pass the new parameter values.
1396 2014-03-14 Richard Biener <rguenther@suse.de>
1398 * common.opt: Revert unintented changes from r205065.
1401 2014-03-14 Richard Biener <rguenther@suse.de>
1404 * cfghooks.c (split_block): Properly adjust all loops the
1405 block was a latch of.
1407 2014-03-14 Martin Jambor <mjambor@suse.cz>
1410 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1413 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1416 * config/avr/avr.c (avr_set_current_function): Pass function name
1417 through default_strip_name_encoding before sanity checking instead
1418 of skipping the first char of the assembler name.
1420 2014-03-13 Richard Henderson <rth@redhat.com>
1423 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1424 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1425 * config/i386/i386-protos.h: Likewise.
1426 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1427 in the expander instead of a splitter.
1428 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1429 any possibility of requiring a memory.
1430 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1431 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1432 (fp branch splitters): Update for ix86_split_fp_branch.
1433 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1434 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1435 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1436 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1437 (*fop_<MODEF>_3_i387): Likewise.
1438 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1439 (splitters for the fop_* register patterns): Remove.
1440 (fscalexf4_i387): Rename from *fscalexf4_i387.
1441 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1443 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1445 PR tree-optimization/59779
1446 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1447 type for bitsize and maxsize instead of HOST_WIDE_INT.
1449 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1451 PR rtl-optimization/57320
1452 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1453 the CFG after thread_prologue_and_epilogue_insns.
1455 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1457 PR rtl-optimization/57189
1458 * lra-constraints.c (process_alt_operands): Disfavor spilling
1461 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1463 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1465 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1467 PR tree-optimization/59025
1469 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1470 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1472 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1475 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1476 calls of avr_out_plus_1.
1478 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1480 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1481 BB's single pred and update the father loop's latch info later.
1483 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1485 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1489 (VEC_base): Likewise.
1490 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1491 registers, we need to swap double words in little endian mode.
1493 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1494 to be a container mode for 128-bit integer operations added in ISA
1495 2.07. Unlike TImode and PTImode, the preferred register set is
1496 the Altivec/VMX registers for the 128-bit operations.
1498 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1500 (rs6000_split_128bit_ok_p): Likewise.
1502 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1503 macros for creating ISA 2.07 normal and overloaded builtin
1504 functions with 3 arguments.
1505 (BU_P8V_OVERLOAD_3): Likewise.
1506 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1507 for use as overloaded functions.
1508 (VPERM_1TI_UNS): Likewise.
1509 (VSEL_1TI): Likewise.
1510 (VSEL_1TI_UNS): Likewise.
1511 (ST_INTERNAL_1ti): Likewise.
1512 (LD_INTERNAL_1ti): Likewise.
1513 (XXSEL_1TI): Likewise.
1514 (XXSEL_1TI_UNS): Likewise.
1515 (VPERM_1TI): Likewise.
1516 (VPERM_1TI_UNS): Likewise.
1517 (XXPERMDI_1TI): Likewise.
1518 (SET_1TI): Likewise.
1519 (LXVD2X_V1TI): Likewise.
1520 (STXVD2X_V1TI): Likewise.
1521 (VEC_INIT_V1TI): Likewise.
1522 (VEC_SET_V1TI): Likewise.
1523 (VEC_EXT_V1TI): Likewise.
1524 (EQV_V1TI): Likewise.
1525 (NAND_V1TI): Likewise.
1526 (ORC_V1TI): Likewise.
1527 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1528 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1530 (VADDUQM): Likewise.
1531 (VSUBCUQ): Likewise.
1532 (VADDEUQM): Likewise.
1533 (VADDECUQ): Likewise.
1534 (VSUBEUQM): Likewise.
1535 (VSUBECUQ): Likewise.
1537 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1538 __int128_t and __uint128_t types.
1539 (__uint128_type): Likewise.
1540 (altivec_categorize_keyword): Add support for vector __int128_t,
1541 vector __uint128_t, vector __int128, and vector unsigned __int128
1542 as a container type for TImode operations that need to be done in
1543 VSX/Altivec registers.
1544 (rs6000_macro_to_expand): Likewise.
1545 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1546 to support 128-bit integer instructions vaddcuq, vadduqm,
1547 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1548 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1550 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1551 for V1TImode, and set up preferences to use VSX/Altivec registers.
1552 Setup VSX reload handlers.
1553 (rs6000_debug_reg_global): Likewise.
1554 (rs6000_init_hard_regno_mode_ok): Likewise.
1555 (rs6000_preferred_simd_mode): Likewise.
1556 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1557 (easy_altivec_constant): Likewise.
1558 (output_vec_const_move): Likewise.
1559 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1561 (rs6000_expand_vector_extract): Likewise.
1562 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1564 (rs6000_const_vec): Add support for V1TImode.
1565 (rs6000_emit_le_vsx_load): Swap double words when loading or
1566 storing TImode/V1TImode.
1567 (rs6000_emit_le_vsx_store): Likewise.
1568 (rs6000_emit_le_vsx_move): Likewise.
1569 (rs6000_emit_move): Add support for V1TImode.
1570 (altivec_expand_ld_builtin): Likewise.
1571 (altivec_expand_st_builtin): Likewise.
1572 (altivec_expand_vec_init_builtin): Likewise.
1573 (altivec_expand_builtin): Likewise.
1574 (rs6000_init_builtins): Add support for V1TImode type. Add
1575 support for ISA 2.07 128-bit integer builtins. Define type names
1576 for the VSX/Altivec vector types.
1577 (altivec_init_builtins): Add support for overloaded vector
1578 functions with V1TImode type.
1579 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1580 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1582 (rs6000_split_128bit_ok_p): Likewise.
1583 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1584 __int128_t and vector __uint128_t.
1586 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1587 and mode attributes.
1594 (VS_scalar): Likewise.
1595 (VS_double): Likewise.
1596 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1598 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1599 we support the ISA 2.07 128-bit integer arithmetic instructions.
1600 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1601 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1602 and TImode types for use with the builtin functions.
1603 (V1TI_type_node): Likewise.
1604 (unsigned_V1TI_type_node): Likewise.
1605 (intTI_type_internal_node): Likewise.
1606 (uintTI_type_internal_node): Likewise.
1608 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1609 128-bit builtin functions.
1610 (UNSPEC_VADDEUQM): Likewise.
1611 (UNSPEC_VADDECUQ): Likewise.
1612 (UNSPEC_VSUBCUQ): Likewise.
1613 (UNSPEC_VSUBEUQM): Likewise.
1614 (UNSPEC_VSUBECUQ): Likewise.
1615 (VM): Add V1TImode to vector mode iterators.
1617 (VI_unit): Likewise.
1618 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1619 (altivec_vaddcuq): Likewise.
1620 (altivec_vsubuqm): Likewise.
1621 (altivec_vsubcuq): Likewise.
1622 (altivec_vaddeuqm): Likewise.
1623 (altivec_vaddecuq): Likewise.
1624 (altivec_vsubeuqm): Likewise.
1625 (altivec_vsubecuq): Likewise.
1627 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1629 (BOOL_128): Likewise.
1630 (BOOL_REGS_OUTPUT): Likewise.
1631 (BOOL_REGS_OP1): Likewise.
1632 (BOOL_REGS_OP2): Likewise.
1633 (BOOL_REGS_UNARY): Likewise.
1634 (BOOL_REGS_AND_CR0): Likewise.
1636 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1637 128-bit integer builtin support.
1638 (vec_vadduqm): Likewise.
1639 (vec_vaddecuq): Likewise.
1640 (vec_vaddeuqm): Likewise.
1641 (vec_vsubecuq): Likewise.
1642 (vec_vsubeuqm): Likewise.
1643 (vec_vsubcuq): Likewise.
1644 (vec_vsubuqm): Likewise.
1646 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1647 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1648 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1649 128-bit integer add/subtract to ISA 2.07.
1651 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1653 * config/arc/arc.c (arc_predicate_delay_insns):
1654 Fix third argument passed to conditionalize_nonjump.
1656 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1658 * config/aarch64/aarch64-builtins.c
1659 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1660 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1661 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1662 instead of __builtin_lfloor.
1663 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1665 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1667 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1668 (tree_ssa_ifcombine_bb_1): New function.
1669 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1670 is an empty forwarder block to then_bb or vice versa and then_bb
1671 and else_bb are effectively swapped.
1673 2014-03-12 Christian Bruel <christian.bruel@st.com>
1676 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1677 REG_CFA_DEF_CFA note.
1678 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1679 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1681 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1683 PR tree-optimization/60454
1684 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1686 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1688 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1689 Do not define target_cpu_default2 to generic.
1690 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1691 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1692 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1694 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1695 Marc Glisse <marc.glisse@inria.fr>
1697 PR tree-optimization/60502
1698 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1699 instead of build_low_bits_mask.
1701 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1704 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1705 if there are multiple uses, but op doesn't live on E edge.
1706 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1707 clobber stmts before __builtin_unreachable.
1709 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1711 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1712 hard_frame_pointer_rtx.
1713 * cse.c (cse_insn): Remove volatile check.
1714 * cselib.c (cselib_process_insn): Likewise.
1715 * dse.c (scan_insn): Likewise.
1717 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1719 * config/arc/arc.c (conditionalize_nonjump): New function,
1721 (arc_ifcvt): ... this.
1722 (arc_predicate_delay_insns): Use it.
1724 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1726 * config/arc/predicates.md (extend_operand): During/after reload,
1727 allow const_int_operand.
1728 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1729 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1730 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1732 (umulsi3_highpart_i): Likewise.
1734 2014-03-11 Richard Biener <rguenther@suse.de>
1736 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1737 Add asserts to guard possible wrong-code bugs.
1739 2014-03-11 Richard Biener <rguenther@suse.de>
1741 PR tree-optimization/60429
1742 PR tree-optimization/60485
1743 * tree-ssa-structalias.c (set_union_with_increment): Properly
1744 take into account all fields that overlap the shifted vars.
1745 (do_sd_constraint): Likewise.
1746 (do_ds_constraint): Likewise.
1747 (get_constraint_for_ptr_offset): Likewise.
1749 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1751 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1752 (nios2_compute_frame_layout):
1753 Add calculation of cfun->machine->fp_save_offset.
1754 (nios2_expand_prologue): Correct setting of frame pointer register
1756 (nios2_expand_epilogue): Update recovery of stack pointer from
1757 frame pointer accordingly.
1758 (nios2_initial_elimination_offset): Update calculation of offset
1759 for eliminating to HARD_FRAME_POINTER_REGNUM.
1761 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1764 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1765 cgraph_get_create_node on VAR_DECLs.
1767 2014-03-10 Richard Biener <rguenther@suse.de>
1770 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1772 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1774 * config/vms/vms.opt (vms_float_format): New variable.
1776 2014-03-08 Tobias Burnus <burnus@net-b.de>
1778 * doc/invoke.texi (-fcilkplus): Update implementation status.
1780 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1781 Richard Biener <rguenther@suse.de>
1783 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1784 consistently accross all TUs.
1785 (run_gcc): Enable -fshort-double automatically at link at link-time
1786 and disallow override.
1788 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1791 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1792 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1793 if they can't be used.
1795 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1797 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1798 for Solaris 11/x86 ld.
1799 * configure: Regenerate.
1801 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1803 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1804 (LIB_TLS_SPEC): Save as ld_tls_libs.
1805 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1806 (HAVE_AS_IX86_TLSLDM): New test.
1807 * configure, config.in: Regenerate.
1808 * config/i386/i386.c (legitimize_tls_address): Fall back to
1809 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1810 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1811 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1812 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1814 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1816 * common.opt (fira-loop-pressure): Mark as optimization.
1818 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1820 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1821 an OpenMP mappable type.
1823 2014-03-06 Matthias Klose <doko@ubuntu.com>
1825 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1826 MULTILIB_OSDIRNAMES is not defined.
1828 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1829 Meador Inge <meadori@codesourcery.com>
1832 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1833 (arm_legitimize_address): Call legitimize_tls_address for any
1834 arm_tls_referenced_p expression, handle constant addend. Call it
1835 before testing for !TARGET_ARM.
1836 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1838 2014-03-06 Richard Biener <rguenther@suse.de>
1844 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1846 * tree-streamer.c (record_common_node): Assert we don't record
1847 nodes with type double.
1848 (preload_common_node): Skip type double, complex double and double
1849 pointer since it is now frontend dependent due to fshort-double option.
1851 2014-03-06 Richard Biener <rguenther@suse.de>
1853 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1854 or -fno-lto is specified and the linker has full plugin support.
1855 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1856 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1857 * lto-wrapper.c (merge_and_complain): Merge compile-time
1858 optimization levels.
1859 (run_gcc): And pass it through to the link options.
1861 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1865 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1867 * cselib.c (remove_useless_values): Skip to avoid quadratic
1868 behavior if the condition moved from...
1869 (cselib_process_insn): ... here holds.
1871 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1874 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1875 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1878 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1879 (TM_H): Add x86-tune.def.
1881 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1883 * config/aarch64/aarch64.c (generic_tunings):
1884 Use cortexa57_extra_costs.
1886 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1889 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1890 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1891 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1894 2014-03-04 Heiher <r@hev.cc>
1896 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1897 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1899 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1901 * config/i386/predicates.md (const2356_operand): Change to ...
1902 (const2367_operand): ... this.
1903 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1905 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1906 (*avx512pf_scatterpf<mode>sf): Ditto.
1907 (avx512pf_scatterpf<mode>df): Ditto.
1908 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1909 (*avx512pf_scatterpf<mode>df): Ditto.
1910 * config/i386/i386.c (ix86_expand_builtin): Update
1911 incorrect hint operand error message.
1913 2014-03-04 Richard Biener <rguenther@suse.de>
1915 * lto-section-in.c (lto_get_section_data): Fix const cast.
1917 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1919 * tree-streamer.c (record_common_node): Assert we don't record
1920 nodes with type double.
1921 (preload_common_node): Skip type double, complex double and double
1922 pointer since it is now frontend dependent due to fshort-double option.
1924 2014-03-04 Richard Biener <rguenther@suse.de>
1927 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1928 (lto_input_toplevel_asms): Likewise.
1929 * lto-section-in.c (lto_get_section_data): Instead do it here
1932 2014-03-04 Richard Biener <rguenther@suse.de>
1934 PR tree-optimization/60382
1935 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1936 dead PHIs a reduction.
1938 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1940 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1942 (_mm_prefetch): Move out of GCC target("sse") pragma.
1943 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1944 GCC target("prfchw") pragma.
1945 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1947 * config/i386/i386.c (ix86_option_override_internal): Enable
1948 -mprfchw with -mprefetchwt1.
1950 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1952 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1955 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1957 * opts.h (CL_PCH_IGNORE): Define.
1958 * targhooks.c (option_affects_pch_p):
1959 Return false for options that have CL_PCH_IGNORE set.
1960 * opt-functions.awk: Process PchIgnore.
1961 * doc/options.texi: Document PchIgnore.
1963 * config/arc/arc.opt (misize): Add PchIgnore property.
1965 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1967 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1968 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1969 constraint on constants to permit them being loaded into
1970 GENERAL_REGS or BASE_REGS.
1972 2014-03-03 Nick Clifton <nickc@redhat.com>
1974 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1975 anti-cacnonical alternatives.
1976 (negandhi3_real): New pattern.
1977 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1979 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1981 * config/avr/avr-mcus.def: Remove atxmega16x1.
1982 * config/avr/avr-tables.opt: Regenerate.
1983 * config/avr/t-multilib: Regenerate.
1984 * doc/avr-mmcu.texi: Regenerate.
1986 2014-03-03 Tobias Grosser <tobias@grosser.es>
1987 Mircea Namolaru <mircea.namolaru@inria.fr>
1989 PR tree-optimization/58028
1990 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1993 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1995 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1996 not handled by recognizers.
1998 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2001 * function.c (expand_function_end): Don't emit
2002 clobber_return_register sequence if clobber_after is a BARRIER.
2003 * cfgexpand.c (construct_exit_block): Append instructions before
2004 return_label to prev_bb.
2006 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2008 * config/rs6000/constraints.md: Document reserved use of "wc".
2010 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2013 * ipa.c (function_and_variable_visibility): When dissolving comdat
2014 group, also set all symbols to local.
2016 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2021 2013-12-14 Jan Hubicka <jh@suse.cz>
2023 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2025 2014-03-02 Jon Beniston <jon@beniston.com>
2031 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2032 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2033 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2034 (simple_return, *simple_return): New patterns
2035 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2036 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2038 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2040 * dwarf2out.c (gen_subprogram_die): Tidy.
2042 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2045 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2046 (*mov_t_msb_neg_negc): ... this new insn.
2048 2014-02-28 Jason Merrill <jason@redhat.com>
2051 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2054 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2057 * dwarf2out.c (decltype_auto_die): New static.
2058 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2059 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2060 (is_cxx_auto): Likewise.
2062 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2064 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2065 we are not using general regs only.
2067 2014-02-28 Richard Biener <rguenther@suse.de>
2070 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2071 previous fix and only allow to remove trivial pre-headers
2072 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2073 (remove_forwarder_block): Properly update the latch of a loop.
2075 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2078 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2079 (cselib_preserved_hash_table): New.
2080 (preserve_constants_and_equivs): Move preserved vals to it.
2081 (cselib_find_slot): Look it up first.
2082 (cselib_init): Initialize it.
2083 (cselib_finish): Release it.
2084 (dump_cselib_table): Dump it.
2086 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2089 * cselib.c (remove_useless_values): Skip to avoid quadratic
2090 behavior if the condition moved from...
2091 (cselib_process_insn): ... here holds.
2093 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2096 * var-tracking.c (vt_initialize): Apply the same condition to
2097 preserve the CFA base value.
2099 2014-02-28 Joey Ye <joey.ye@arm.com>
2102 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2103 if reload in progress or completed.
2105 2014-02-28 Tobias Burnus <burnus@net-b.de>
2108 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2111 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2113 * doc/tm.texi.in (Condition Code Status): Update documention for
2114 relative locations of cc0-setter and cc0-user.
2116 2014-02-27 Jeff Law <law@redhat.com>
2118 PR rtl-optimization/52714
2119 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2120 into two independent simple sets, if I3 is a jump, ensure the
2121 pattern we place into I3 is a (set (pc) ...).
2123 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2124 Jeff Law <law@redhat.com>
2126 PR rtl-optimization/49847
2127 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2128 are in different blocks.
2129 * doc/tm.texi (Condition Code Status): Update documention for
2130 relative locations of cc0-setter and cc0-user.
2132 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2135 * lra.c (lra_emit_add): Check SUBREG too.
2137 2014-02-27 Andreas Schwab <schwab@suse.de>
2139 * config/m68k/m68k.c (m68k_option_override): Disable
2140 -flive-range-shrinkage for classic m68k.
2141 (m68k_override_options_after_change): Likewise.
2143 2014-02-27 Marek Polacek <polacek@redhat.com>
2146 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2147 -Wmaybe-uninitialized.
2149 2014-02-27 Alan Modra <amodra@gmail.com>
2152 * reload1.c (emit_input_reload_insns): When reload_override_in,
2153 set old to rl->in_reg when rl->in_reg is a subreg.
2155 2014-02-26 Richard Biener <rguenther@suse.de>
2158 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2160 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2162 * common/config/i386/predicates.md (const1256_operand): Remove.
2163 (const2356_operand): New.
2164 (const_1_to_2_operand): Remove.
2165 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2166 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2167 (*avx512pf_gatherpf<mode>sf): Ditto.
2168 (avx512pf_gatherpf<mode>df): Ditto.
2169 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2170 (*avx512pf_gatherpf<mode>df): Ditto.
2171 (avx512pf_scatterpf<mode>sf): Ditto.
2172 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2173 (*avx512pf_scatterpf<mode>sf): Ditto.
2174 (avx512pf_scatterpf<mode>df): Ditto.
2175 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2176 (*avx512pf_scatterpf<mode>df): Ditto.
2177 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2179 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2181 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2182 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2183 (_mm512_mask_testn_epi64_mask): Move to ...
2184 * config/i386/avx512cdintrin.h: Here.
2185 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2186 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2187 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2188 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2189 TARGET_AVX512F from TARGET_AVX512CD.
2191 2014-02-26 Richard Biener <rguenther@suse.de>
2194 * ipa.c (walk_polymorphic_call_targets): Properly guard
2195 call to inline_update_overall_summary.
2197 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2200 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2201 and latches only if requested. Fix latch if it is removed.
2202 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2203 LOOPS_HAVE_PREHEADERS.
2205 2014-02-25 Andrew Pinski <apinski@cavium.com>
2207 * builtins.c (expand_builtin_thread_pointer): Create a new target
2208 when the target is NULL.
2210 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2212 PR rtl-optimization/60317
2213 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2214 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2215 * lra-assigns.c: Include params.h.
2216 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2217 other reload pseudos considerations.
2219 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2221 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2222 to use canonical form for nor<mode>3.
2224 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2227 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2230 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2232 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2233 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2234 (ix86_handle_option): Handle OPT_mprefetchwt1.
2235 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2236 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2238 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2239 OPTION_MASK_ISA_PREFETCHWT1.
2240 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2241 (PTA_PREFETCHWT1): New.
2242 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2243 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2244 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2245 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2246 (*prefetch_avx512pf_<mode>_: Change into ...
2247 (*prefetch_prefetchwt1_<mode>: This.
2248 * config/i386/i386.opt (mprefetchwt1): New.
2249 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2250 (_mm_prefetch): Handle intent to write.
2251 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2253 2014-02-25 Richard Biener <rguenther@suse.de>
2256 * emit-rtl.c (mem_attrs_htab): Remove.
2257 (mem_attrs_htab_hash): Likewise.
2258 (mem_attrs_htab_eq): Likewise.
2259 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2260 (init_emit_once): Do not allocate mem_attrs_htab.
2262 2014-02-25 Richard Biener <rguenther@suse.de>
2265 * lto-opts.c (lto_write_options): Output non-explicit conservative
2266 -fwrapv, -fno-trapv and -fno-strict-overflow.
2267 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2268 (run_gcc): And pass them through.
2270 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2272 * sel-sched.c (calculate_new_fences): New parameter ptime.
2273 Calculate it as a maximum over all fence cycles.
2274 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2275 Print the final schedule timing when sched_verbose.
2277 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2279 PR rtl-optimization/60292
2280 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2281 bit fot the fence instruction.
2283 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2285 * calls.h: Fix typo in comment.
2287 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2289 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2290 adjusting offsetable addresses.
2292 2014-02-24 Guozhi Wei <carrot@google.com>
2294 * sparseset.h (sparseset_pop): Fix the wrong index.
2296 2014-02-24 Walter Lee <walt@tilera.com>
2298 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2299 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2301 * common/config/tilegx/tilegx-common.c
2302 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2303 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2305 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2306 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2307 (tilegx_gimplify_va_arg_expr): Handle big endian.
2308 (tilegx_expand_unaligned_load): Ditto.
2309 (tilegx_expand_unaligned_store): Ditto.
2310 (TARGET_RETURN_IN_MSB): New.
2311 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2312 (TARGET_ENDIAN_DEFAULT): New.
2313 (TARGET_BIG_ENDIAN): Handle big endian.
2314 (BYTES_BIG_ENDIAN): Ditto.
2315 (WORDS_BIG_ENDIAN): Ditto.
2316 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2319 * config/tilegx/tilegx.md (extv): Handle big endian.
2321 (insn_st<n>): Ditto.
2322 (insn_st<n>_add<bitsuffix>): Ditto.
2323 (insn_stnt<n>): Ditto.
2324 (insn_stnt<n>_add<bitsuffix>):Ditto.
2325 (vec_interleave_highv8qi): Handle big endian.
2326 (vec_interleave_highv8qi_be): New.
2327 (vec_interleave_highv8qi_le): New.
2328 (insn_v1int_h): Handle big endian.
2329 (vec_interleave_lowv8qi): Handle big endian.
2330 (vec_interleave_lowv8qi_be): New.
2331 (vec_interleave_lowv8qi_le): New.
2332 (insn_v1int_l): Handle big endian.
2333 (vec_interleave_highv4hi): Handle big endian.
2334 (vec_interleave_highv4hi_be): New.
2335 (vec_interleave_highv4hi_le): New.
2336 (insn_v2int_h): Handle big endian.
2337 (vec_interleave_lowv4hi): Handle big endian.
2338 (vec_interleave_lowv4hi_be): New.
2339 (vec_interleave_lowv4hi_le): New.
2340 (insn_v2int_l): Handle big endian.
2341 (vec_interleave_highv2si): Handle big endian.
2342 (vec_interleave_highv2si_be): New.
2343 (vec_interleave_highv2si_le): New.
2344 (insn_v4int_h): Handle big endian.
2345 (vec_interleave_lowv2si): Handle big endian.
2346 (vec_interleave_lowv2si_be): New.
2347 (vec_interleave_lowv2si_le): New.
2348 (insn_v4int_l): Handle big endian.
2349 * config/tilegx/tilegx.opt (mbig-endian): New option.
2350 (mlittle-endian): New option.
2351 * doc/install.texi: Document tilegxbe-linux.
2352 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2354 2014-02-24 Martin Jambor <mjambor@suse.cz>
2357 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2358 there are no parameter descriptors.
2360 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2362 PR rtl-optimization/60268
2363 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2364 initialization to ...
2365 (sched_rgn_init): ... here.
2366 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2368 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2370 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2373 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2375 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2378 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2380 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2381 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2383 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2385 * config/microblaze/predicates.md: Add cmp_op predicate.
2386 * config/microblaze/microblaze.md: Add branch_compare instruction
2387 which uses cmp_op predicate and emits cmp insn before branch.
2388 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2389 to microblaze_expand_conditional_branch and consolidate logic.
2390 (microblaze_expand_conditional_branch): emit branch_compare
2391 insn instead of handling cmp op separate from branch insn.
2393 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2395 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2398 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2400 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2401 define_insn with define_expand and new define_insn
2402 *altivec_lve<VI_char>x_internal.
2403 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2404 and new define_insn *altivec_stve<VI_char>x_internal.
2405 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2407 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2409 (altivec_expand_stvex_be): New function.
2411 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2413 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2414 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2415 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2416 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2418 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2421 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2422 instead of emit_move_insn.
2424 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2426 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2428 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2429 gen_altivec_vsumsws.
2431 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2433 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2434 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2435 (altivec_lvxl_<mode>): New define_expand incorporating
2436 -maltivec=be semantics where needed.
2437 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2438 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2439 semantics where needed.
2440 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2441 (altivec_stvx_<mode>): New define_expand incorporating
2442 -maltivec=be semantics where needed.
2443 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2444 VM2 iterator instead of V4SI.
2445 (altivec_stvxl_<mode>): New define_expand incorporating
2446 -maltivec=be semantics where needed.
2447 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2448 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2449 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2450 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2451 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2452 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2453 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2454 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2455 ALTIVEC_BUILTIN_STVXL.
2456 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2457 (altivec_expand_stvx_be): Likewise.
2458 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2459 (altivec_expand_lvx_be): Likewise.
2460 (altivec_expand_stvx_be): Likewise.
2461 (altivec_expand_builtin): Add cases for
2462 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2463 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2464 (altivec_init_builtins): Add definitions for
2465 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2466 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2468 2014-02-21 Catherine Moore <clm@codesourcery.com>
2470 * doc/invoke.texi (mvirt, mno-virt): Document.
2471 * config/mips/mips.opt (mvirt): New option.
2472 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2474 2014-02-21 Richard Biener <rguenther@suse.de>
2476 PR tree-optimization/60276
2477 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2478 (STMT_VINFO_MIN_NEG_DIST): New macro.
2479 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2480 STMT_VINFO_MIN_NEG_DIST.
2481 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2482 made for negative dependence distances still hold.
2484 2014-02-21 Richard Biener <rguenther@suse.de>
2487 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2488 DECL_INITIAL for globals not in the current function context.
2490 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2492 PR tree-optimization/56490
2493 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2494 * tree-ssa-uninit.c: Include params.h.
2495 (compute_control_dep_chain): Add num_calls argument, return false
2496 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2497 num_calls to recursive call.
2498 (find_predicates): Change dep_chain into normal array,
2499 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2500 variable and adjust compute_control_dep_chain caller.
2501 (find_def_preds): Likewise.
2503 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2505 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2506 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2508 2014-02-21 Nick Clifton <nickc@redhat.com>
2510 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2511 (pushhi1): Likewise.
2512 (popqi1): Add mode to pre_dec.
2515 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2517 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2518 mode for mask of V8SFmode permutation.
2520 2014-02-20 Richard Henderson <rth@redhat.com>
2523 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2524 a new pseudo for OLDVAL.
2526 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2529 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2530 gen_reg_rtx if d->testing_p.
2531 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2532 if d->testing_p and we will certainly return true.
2533 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2536 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2538 * emit-rtl.c (gen_reg_rtx): Assert that
2539 crtl->emit.regno_pointer_align_length is non-zero.
2541 2014-02-20 Richard Henderson <rth@redhat.com>
2544 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2545 on failure the store back into EXPECT.
2547 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2548 Sandra Loosemore <sandra@codesourcery.com>
2550 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2551 * config/nios2/nios2.c (nios2_function_profiler): Add
2552 -fPIC (flag_pic == 2) support.
2553 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2554 (nios2_large_offset_p): New function.
2555 (nios2_unspec_reloc_p): Move up position, update to use
2556 nios2_large_offset_p.
2557 (nios2_unspec_address): Remove function.
2558 (nios2_unspec_offset): New function.
2559 (nios2_large_got_address): New function.
2560 (nios2_got_address): Add large offset support.
2561 (nios2_legitimize_tls_address): Update usage of removed and new
2563 (nios2_symbol_binds_local_p): New function.
2564 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2565 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2566 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2567 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2568 processing for (const (unspec ...)).
2569 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2571 2014-02-20 Richard Biener <rguenther@suse.de>
2573 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2574 doing the substitution.
2575 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2577 2014-02-20 Martin Jambor <mjambor@suse.cz>
2580 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2581 info when checking whether lattices are bottom.
2583 2014-02-20 Richard Biener <rguenther@suse.de>
2586 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2589 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2592 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2593 parameter specifying the scaling.
2594 (inline_call): Update.
2595 (want_inline_recursively): Guard division by zero.
2596 (recursive_inlining): Update.
2597 * ipa-inline.h (clone_inlined_nodes): Update.
2599 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2602 * config/i386/i386.c (classify_argument): Pass structures of size
2603 64 bytes or less in register.
2605 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2606 Kirill Yukhin <kirill.yukhin@intel.com>
2608 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2609 (_mm_rcp28_round_ss): Ditto.
2610 (_mm_rsqrt28_round_sd): Ditto.
2611 (_mm_rsqrt28_round_ss): Ditto.
2612 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2613 (_mm_rcp14_round_ss): Ditto.
2614 (_mm_rsqrt14_round_sd): Ditto.
2615 (_mm_rsqrt14_round_ss): Ditto.
2616 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2617 the first input operand, get rid of match_dup.
2618 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2620 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2622 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2623 operand as the first input operand, set type attribute.
2624 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2626 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2627 operand as the first input operand, set type attribute.
2629 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2631 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2634 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2637 * config/i386/i386.c (construct_container): Remove TFmode check
2638 for X86_64_INTEGER_CLASS.
2640 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2643 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2644 only when -Wpsabi is enabled.
2646 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2649 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2650 passing arrays in registers are the same as for structs, so remove the
2651 special case for them.
2653 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2655 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2656 destination type, extract only the valid bits if the source type is not
2657 integral and has a different mode.
2659 2014-02-19 Richard Biener <rguenther@suse.de>
2662 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2665 2014-02-19 Richard Biener <rguenther@suse.de>
2668 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2669 (ipa_modify_call_arguments): Emit an argument load explicitely and
2670 preserve virtual SSA form there and for the replacement call.
2671 Do not update SSA form nor free dominance info.
2673 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2675 * ipa.c (function_and_variable_visibility): Also clear WEAK
2676 flag when disolving COMDAT_GROUP.
2678 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2680 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2681 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2683 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2684 do more sanity checks.
2685 (detect_type_change): Return true when giving up early.
2686 (compute_complex_assign_jump_func): Fix type parameter of
2687 ipa_set_ancestor_jf.
2688 (compute_complex_ancestor_jump_func): Likewise.
2689 (update_jump_functions_after_inlining): Fix updating of
2691 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2693 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2695 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2696 inline clones when edge disappears.
2698 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2701 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2702 Split 64-bit moves into 2 patterns. Do not allow the use of
2703 direct move for TDmode in little endian, since the decimal value
2704 has little endian bytes within a word, but the 64-bit pieces are
2705 ordered in a big endian fashion, and normal subreg's of TDmode are
2707 (mov<mode>_64bit_dm): Likewise.
2708 (movtd_64bit_nodm): Likewise.
2710 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2712 PR tree-optimization/60174
2713 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2714 statement of an SSA_NAME that occurs in an abnormal PHI node.
2716 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2719 * final.c (SEEN_BB): Remove.
2720 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2721 (final_scan_insn): Don't force_source_line on second
2722 NOTE_INSN_BASIC_BLOCK.
2724 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2727 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2728 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2729 (type_natural_mode): Warn ABI change when %zmm register is not
2730 available for AVX512F vector value passing.
2732 2014-02-18 Kai Tietz <ktietz@redhat.com>
2735 * config/i386/i386.c (ix86_expand_prologue): Use value in
2736 rax register as displacement when restoring %r10 or %rax.
2737 Fix wrong offset when restoring both registers.
2739 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2741 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2742 assertion with conditional return.
2744 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2745 Uros Bizjak <ubizjak@gmail.com>
2748 * config/i386/driver-i386.c (host_detect_local_cpu): If
2749 YMM state is not saved by the OS, also clear has_f16c. Move
2750 CPUID 0x80000001 handling before YMM state saving checking.
2752 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2754 PR rtl-optimization/58960
2755 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2756 factored out from ...
2757 (sched_init): ... here.
2758 (free_global_sched_pressure_data): New, factored out from ...
2759 (sched_finish): ... here.
2760 * sched-int.h (free_global_sched_pressure_data): Declare.
2761 * sched-rgn.c (nr_regions_initial): New static global.
2762 (haifa_find_rgns): Initialize it.
2763 (schedule_region): Disable sched-pressure for the newly
2766 2014-02-17 Richard Biener <rguenther@suse.de>
2768 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2769 release SSA defs of pattern stmts.
2771 2014-02-17 Richard Biener <rguenther@suse.de>
2773 * tree-inline.c (expand_call_inline): Release the virtual
2774 operand defined by the call we are about to inline.
2776 2014-02-17 Richard Biener <rguenther@suse.de>
2778 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2780 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2781 Ilya Tocar <ilya.tocar@intel.com>
2783 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2784 arguments order in builtin.
2785 (_mm512_permutexvar_epi64): Ditto.
2786 (_mm512_mask_permutexvar_epi64): Ditto
2787 (_mm512_maskz_permutexvar_epi32): Ditto
2788 (_mm512_permutexvar_epi32): Ditto
2789 (_mm512_mask_permutexvar_epi32): Ditto
2791 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2793 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2794 (p8_vmrgow): Likewise.
2796 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2798 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2801 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2804 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2805 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2806 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2807 using direct move instructions on ISA 2.07. Also adjust
2808 instruction length for 64-bit.
2809 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2810 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2812 2014-02-15 Alan Modra <amodra@gmail.com>
2816 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2817 find_replacement on parts of insn rtl that might be reloaded.
2819 2014-02-15 Richard Biener <rguenther@suse.de>
2821 PR tree-optimization/60183
2822 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2823 (tree_ssa_phiprop): Calculate and free post-dominators.
2825 2014-02-14 Jeff Law <law@redhat.com>
2827 PR rtl-optimization/60131
2828 * ree.c (get_extended_src_reg): New function.
2829 (combine_reaching_defs): Use it rather than assuming location of REG.
2830 (find_and_remove_re): Verify first operand of extension is
2831 a REG before adding the insns to the copy list.
2833 2014-02-14 Roland McGrath <mcgrathr@google.com>
2835 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2836 * configure: Regenerated.
2837 * config.in: Regenerated.
2838 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2839 instead of ASM_SHORT.
2841 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2842 Richard Earnshaw <rearnsha@arm.com>
2844 PR rtl-optimization/59535
2845 * lra-constraints.c (process_alt_operands): Encourage alternative
2846 when unassigned pseudo class is superset of the alternative class.
2847 (inherit_reload_reg): Don't inherit when optimizing for code size.
2848 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2849 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2850 modes not less than 4 for Thumb1.
2852 2014-02-14 Kyle McMartin <kyle@redhat.com>
2855 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2857 2014-02-14 Richard Biener <rguenther@suse.de>
2859 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2860 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2861 Do not strip INDIRECT_REFs.
2863 2014-02-14 Richard Biener <rguenther@suse.de>
2866 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2867 DECL_FUNCTION_SPECIFIC_TARGET.
2868 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2869 * tree-streamer-out.c (pack_ts_target_option): Remove.
2870 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2871 (write_ts_function_decl_tree_pointers): Do not stream
2872 DECL_FUNCTION_SPECIFIC_TARGET.
2873 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2874 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2875 (lto_input_ts_function_decl_tree_pointers): Do not stream
2876 DECL_FUNCTION_SPECIFIC_TARGET.
2878 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2880 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2881 (get_initial_def_for_induction, vectorizable_induction): Ignore
2882 debug stmts when looking for exit_phi.
2883 (vectorizable_live_operation): Fix up condition.
2885 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2887 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2888 nreverse() because it changes the content of original tree list.
2890 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2892 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2893 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2895 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2897 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2898 GNU coding standards.
2900 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2903 * dwarf2out.c (gen_subprogram_die): Don't call
2904 add_calling_convention_attribute if subr_die is old_die.
2906 2014-02-13 Sharad Singhai <singhai@google.com>
2908 * doc/optinfo.texi: Fix order of nodes.
2910 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2912 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2913 operands[2], not operands[3].
2915 2014-02-13 Richard Biener <rguenther@suse.de>
2918 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2919 mention the possibility of an in-tree build.
2920 (CLooG): Update recommended version to 0.18.1, mention the
2921 possibility of an in-tree build and clarify that the ISL
2922 bundled with CLooG does not work.
2924 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2927 * expr.c (compress_float_constant): If x is a hard register,
2928 extend into a pseudo and then move to x.
2930 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2932 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2933 caused by bad second argument to warning_at() with -mhotpatch and
2934 nested functions (e.g. with gfortran).
2936 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2938 * opts.c (option_name): Remove "enabled by default" rider.
2940 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2942 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2944 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2945 Uros Bizjak <ubizjak@gmail.com>
2948 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2949 * configure: Regenerated.
2951 2014-02-12 Richard Biener <rguenther@suse.de>
2953 * vec.c (vec_prefix::calculate_allocation): Move as
2954 inline variant to vec.h.
2955 (vec_prefix::calculate_allocation_1): New out-of-line version.
2956 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2957 (vec_prefix::m_has_auto_buf): Rename to ...
2958 (vec_prefix::m_using_auto_storage): ... this.
2959 (vec_prefix::calculate_allocation): Inline the easy cases
2960 and dispatch to calculate_allocation_1 which doesn't need the
2962 (va_heap::reserve): Use gcc_checking_assert.
2963 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2964 m_using_auto_storage.
2965 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2967 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2968 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2969 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2971 2014-02-12 Richard Biener <rguenther@suse.de>
2973 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2974 when we found a dependence.
2976 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2978 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2980 (maybe_fold_stmt): ... into this new function.
2981 * omp-low.c (lower_omp): Update comment.
2983 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2986 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2989 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2991 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2992 identifiers in comments.
2993 (cortexa53_extra_costs): Likewise.
2994 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2995 (cortexa7_extra_costs): Likewise.
2996 (cortexa12_extra_costs): Likewise.
2997 (cortexa15_extra_costs): Likewise.
2998 (v7m_extra_costs): Likewise.
3000 2014-02-12 Richard Biener <rguenther@suse.de>
3003 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3004 of posix_memalign being successful.
3005 (lower_stmt): Restrict lowering of posix_memalign to when
3006 -ftree-bit-ccp is enabled.
3008 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3010 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3012 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3014 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3016 PR rtl-optimization/60116
3017 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3018 other_insn once the combination has been validated.
3020 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3023 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3025 * ipa-devirt.c: Include demangle.h
3026 (odr_violation_reported): New static variable.
3027 (add_type_duplicate): Update odr_violations.
3028 (maybe_record_node): Add completep parameter; update it.
3029 (record_target_from_binfo): Add COMPLETEP parameter;
3030 update it as needed.
3031 (possible_polymorphic_call_targets_1): Likewise.
3032 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3033 rename FINAL to COMPLETE.
3034 (record_targets_from_bases): Sanity check we found the binfo;
3035 fix COMPLETEP updating.
3036 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3037 parameter, fix computing of COMPLETEP.
3038 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3039 at LTO time do demangling.
3040 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3041 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3043 (gimple_get_virt_method_for_binfo): Likewise.
3044 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3045 gimple_get_virt_method_for_vtable): Update prototypes.
3047 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3050 * genautomata.c (add_presence_absence): Fix typo with
3051 {final_}presence_list.
3053 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3056 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3057 for VSX/Altivec vectors that land in GPR registers.
3059 2014-02-11 Richard Henderson <rth@redhat.com>
3060 Jakub Jelinek <jakub@redhat.com>
3063 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3064 around drhs if type conversion to lacc->type is not useless.
3066 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3068 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3070 (cortex-a57.cortex-a53): Likewise.
3071 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3073 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3075 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3078 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3080 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3081 add_options_for_arm_vfp3.
3083 2014-02-11 Jeff Law <law@redhat.com>
3086 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3087 object with an undesirable mode.
3089 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3092 * config/i386/sol2-9.h: New file.
3093 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3094 *-*-solaris2.9*): Use it.
3096 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3098 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3099 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3101 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3103 * config/microblaze/microblaze.c: Extend mcpu version format
3105 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3107 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3109 2014-02-10 Richard Henderson <rth@redhat.com>
3112 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3113 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3114 ms-abi vs -mno-accumulate-outgoing-args.
3115 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3116 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3119 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3122 * cfgexpand.c (expand_asm_operands): Attach source location to
3123 ASM_INPUT rtx objects.
3124 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3126 2014-02-10 Nick Clifton <nickc@redhat.com>
3128 * config/mn10300/mn10300.c (popcount): New function.
3129 (mn10300_expand_prologue): Include saved registers in stack usage
3132 2014-02-10 Jeff Law <law@redhat.com>
3135 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3136 when changing the SET_DEST of a prior insn to avoid an input reload.
3138 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3140 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3141 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3142 -mcall-openbsd, or -mcall-linux.
3143 (CC1_ENDIAN_BIG_SPEC): Remove.
3144 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3145 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3146 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3147 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3148 and %cc1_endian_default.
3149 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3151 2014-02-10 Richard Biener <rguenther@suse.de>
3153 PR tree-optimization/60115
3154 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3155 MEM_REF handling. Properly verify that the accesses are not
3156 out of the objects bound.
3158 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3160 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3163 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3165 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3166 proper constants and fix formatting.
3167 (possible_polymorphic_call_targets): Fix formatting.
3169 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3170 Ilya Tocar <ilya.tocar@intel.com>
3172 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3173 (_mm512_loadu_epi32): Renamed into...
3174 (_mm512_loadu_si512): This.
3175 (_mm512_storeu_epi32): Renamed into...
3176 (_mm512_storeu_si512): This.
3177 (_mm512_maskz_ceil_ps): Removed.
3178 (_mm512_maskz_ceil_pd): Ditto.
3179 (_mm512_maskz_floor_ps): Ditto.
3180 (_mm512_maskz_floor_pd): Ditto.
3181 (_mm512_floor_round_ps): Ditto.
3182 (_mm512_floor_round_pd): Ditto.
3183 (_mm512_ceil_round_ps): Ditto.
3184 (_mm512_ceil_round_pd): Ditto.
3185 (_mm512_mask_floor_round_ps): Ditto.
3186 (_mm512_mask_floor_round_pd): Ditto.
3187 (_mm512_mask_ceil_round_ps): Ditto.
3188 (_mm512_mask_ceil_round_pd): Ditto.
3189 (_mm512_maskz_floor_round_ps): Ditto.
3190 (_mm512_maskz_floor_round_pd): Ditto.
3191 (_mm512_maskz_ceil_round_ps): Ditto.
3192 (_mm512_maskz_ceil_round_pd): Ditto.
3193 (_mm512_expand_pd): Ditto.
3194 (_mm512_expand_ps): Ditto.
3195 * config/i386/i386.c (ix86_builtins): Remove
3196 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3197 (bdesc_args): Ditto.
3198 * config/i386/predicates.md (const1256_operand): New.
3199 (const_1_to_2_operand): Ditto.
3200 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3201 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3202 (*avx512pf_gatherpf<mode>sf): Ditto.
3203 (avx512pf_gatherpf<mode>df): Ditto.
3204 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3205 (*avx512pf_gatherpf<mode>df): Ditto.
3206 (avx512pf_scatterpf<mode>sf): Ditto.
3207 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3208 (*avx512pf_scatterpf<mode>sf): Ditto.
3209 (avx512pf_scatterpf<mode>df): Ditto.
3210 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3211 (*avx512pf_scatterpf<mode>df): Ditto.
3212 (avx512f_expand<mode>): Removed.
3213 (<shift_insn><mode>3<mask_name>): Change predicate type.
3215 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3217 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3218 not at the end of datarefs vector use ordered_remove to avoid
3219 reordering datarefs vector.
3222 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3223 mark local addressable non-static vars as GOVD_PRIVATE
3224 instead of GOVD_LOCAL.
3225 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3226 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3230 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3231 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3232 assume_aligned or alloc_align attributes.
3233 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3234 arguments. Handle also assume_aligned and alloc_align attributes.
3235 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3236 calls to functions with assume_aligned or alloc_align attributes.
3237 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3239 2014-02-08 Terry Guo <terry.guo@arm.com>
3241 * doc/invoke.texi: Document ARM -march=armv7e-m.
3243 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3245 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3246 flag on __cilkrts_rethrow builtin.
3249 * ipa-cp.c (determine_versionability): Fail at -O0
3250 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3251 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3254 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3257 * tree-inline.c (copy_forbidden): Fail for
3258 __attribute__((optimize (0))) functions.
3260 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3262 * varpool.c: Include pointer-set.h.
3263 (varpool_remove_unreferenced_decls): Variables in other partitions
3264 will not be output; be however careful to not lose information
3267 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3269 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3270 lookup in the vtable constructor.
3272 2014-02-07 Jeff Law <law@redhat.com>
3275 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3276 define_insn_and_split.
3278 * ipa-inline.c (inline_small_functions): Fix typos.
3280 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3282 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3283 (s390_can_use_return_insn): Declare.
3284 * config/s390/s390.h (EPILOGUE_USES): Define.
3285 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3287 (s390_chunkify_start): Handle return JUMP_LABELs.
3288 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3289 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3290 (s390_can_use_return_insn): New functions.
3291 (s390_fix_long_loop_prediction): Handle conditional returns.
3292 (TARGET_SET_UP_BY_PROLOGUE): Define.
3293 * config/s390/s390.md (ANY_RETURN): New code iterator.
3294 (*creturn, *csimple_return, return, simple_return): New patterns.
3296 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3298 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3299 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3300 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3301 REG_CFA_RESTORE list when deciding not to restore a register.
3303 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3305 * config/s390/s390.c: Include tree-pass.h and context.h.
3306 (s390_early_mach): New function, split out from...
3307 (s390_emit_prologue): ...here.
3308 (pass_data_s390_early_mach): New pass structure.
3309 (pass_s390_early_mach): New class.
3310 (s390_option_override): Create and register early_mach pass.
3311 Move to end of file.
3313 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3315 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3316 to match for the exit block.
3318 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3320 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3321 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3322 Reject misaligned operands.
3324 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3326 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3328 2014-02-07 Richard Biener <rguenther@suse.de>
3331 * gimple-low.c (lower_builtin_posix_memalign): New function.
3332 (lower_stmt): Call it to lower posix_memalign in a way
3333 to make alignment info accessible.
3335 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3338 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3339 __builtin_setjmp_receiver.
3341 2014-02-07 Richard Biener <rguenther@suse.de>
3344 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3345 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3346 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3347 Handle BUILT_IN_POSIX_MEMALIGN.
3348 (find_func_clobbers): Likewise.
3349 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3350 (call_may_clobber_ref_p_1): Likewise.
3352 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3355 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3358 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3361 * lto-cgraph.c (lto_output_node): Use
3362 symtab_get_symbol_partitioning_class.
3363 (lto_output_varpool_node): likewise.
3364 (symtab_get_symbol_partitioning_class): Move here from
3366 * cgraph.h (symbol_partitioning_class): Likewise.
3367 (symtab_get_symbol_partitioning_class): Declare.
3369 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3371 * ggc.h (ggc_internal_cleared_alloc): New macro.
3372 * vec.h (vec_safe_copy): Handle memory stats.
3373 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3374 * target-globals.c (save_target_globals): Likewise.
3376 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3379 * expr.c (emit_move_resolve_push): Export; be bit more selective
3380 on when to clear alias set.
3381 * expr.h (emit_move_resolve_push): Declare.
3382 * function.h (struct function): Add tail_call_marked.
3383 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3384 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3385 * config/i386/i386.md (TImode move expander): De not call
3387 (FP push expanders): Preserve memory attributes.
3388 * config/i386/sse.md (push<mode>1): Remove.
3389 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3390 (ix86_expand_push): Remove.
3391 * config/i386/mmx.md (push<mode>1): Remove.
3393 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3395 PR rtl-optimization/60030
3396 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3397 lopart with paradoxical subreg before shifting it up by hprec.
3399 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3401 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3402 Remove extra newline at end of file.
3403 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3404 (arm_issue_rate): Handle cortexa57.
3405 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3406 (cortex-a57.cortex-a53): Likewise.
3408 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3411 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3412 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3414 (arm_expand_prologue): Adjust all callers.
3415 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3416 info, registers also at the lowest numbered registers side. Use
3417 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3421 * var-tracking.c (adjust_mems): Before adding a SET to
3422 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3424 2014-02-06 Alan Modra <amodra@gmail.com>
3427 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3428 change SDmode to DDmode when lra_in_progress.
3430 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3433 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3434 free_data_ref on the dr first, and before goto again also set dr
3435 to the next dr. For simd_lane_access, free old datarefs[i] before
3436 overwriting it. For get_vectype_for_scalar_type failure, don't
3437 free_data_ref if simd_lane_access.
3439 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3442 * tree.h (opts_for_fn): New inline function.
3443 (opt_for_fn): Define.
3444 * config/i386/i386.c (ix86_function_regparm): Use
3445 opt_for_fn (decl, optimize) instead of optimize.
3447 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3449 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3450 for SYMBOL_REF in large memory model.
3452 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3454 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3456 (cortex-a57): Likewise.
3457 (cortex-a57.cortex-a53): Likewise.
3459 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3460 Kugan Vivekanandarajah <kuganv@linaro.org>
3462 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3464 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3466 2014-02-06 Richard Biener <rguenther@suse.de>
3468 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3469 set_loop_copy and initialize_original_copy_tables.
3471 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3473 * config/aarch64/aarch64-simd.md
3474 (aarch64_ashr_simddi): Change QI to SI.
3476 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3477 Jakub Jelinek <jakub@redhat.com>
3480 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3483 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3485 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3486 CODE_FOR_altivec_vpku[hw]um to
3487 CODE_FOR_altivec_vpku[hw]um_direct.
3488 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3489 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3490 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3491 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3493 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3495 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3496 generation for -maltivec=be.
3497 (altivec_vsumsws): Simplify redundant test.
3499 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3501 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3502 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3503 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3504 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3505 gen_altivec_vpkuwum.
3506 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3508 (altivec_vpks<VI_char>ss): Likewise.
3509 (altivec_vpks<VI_char>us): Likewise.
3510 (altivec_vpku<VI_char>us): Likewise.
3511 (altivec_vpku<VI_char>um): Likewise.
3512 (altivec_vpku<VI_char>um_direct): New (copy of
3513 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3515 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3516 target is little endian and -maltivec=be is not specified.
3517 (*altivec_vupkhs<VU_char>_direct): New (copy of
3518 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3519 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3520 target is little endian and -maltivec=be is not specified.
3521 (*altivec_vupkls<VU_char>_direct): New (copy of
3522 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3523 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3524 little endian and -maltivec=be is not specified.
3525 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3526 little endian and -maltivec=be is not specified.
3528 2014-02-05 Richard Henderson <rth@redhat.com>
3531 * combine-stack-adj.c: Revert r206943.
3532 * sched-int.h (struct deps_desc): Add last_args_size.
3533 * sched-deps.c (init_deps): Initialize it.
3534 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3535 contain REG_ARGS_SIZE notes.
3537 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3539 * lto-cgraph.c (asm_nodes_output): Make global.
3540 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3541 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3542 (driver_handle_option): Handle OPT_fwpa.
3544 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3547 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3548 a comment typo and formatting issue. If odr_hash hasn't been
3549 created, return vNULL and set *completep to false.
3552 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3553 bb with no successors.
3555 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3558 * doc/invoke.texi (-march): Clarify documentation for ARM.
3562 2014-02-05 Richard Biener <rguenther@suse.de>
3564 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3565 when not vectorizing because of too many alias checks.
3566 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3567 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3569 2014-02-05 Nick Clifton <nickc@redhat.com>
3571 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3572 accept extended registers in any mode when compiling for the MN10300.
3574 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3576 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3577 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3578 sanitization attributes.
3579 (can_inline_edge_p): Likewise.
3580 (sanitize_attrs_match_for_inline_p): New function.
3582 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3584 * ipa-prop.c (detect_type_change): Shor circuit testing of
3585 type changes on THIS pointer.
3587 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3590 * config/pa/pa.c (legitimize_tls_address): Return original address
3591 if not passed a SYMBOL_REF rtx.
3592 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3594 (pa_emit_move_sequence): Simplify TLS source operands.
3595 (pa_legitimate_constant_p): Reject all TLS constants.
3596 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3597 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3599 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3601 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3602 groups when we know they are controlled by LTO.
3603 * varasm.c (default_binds_local_p_1): If object is in other partition,
3604 it will be resolved locally.
3606 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3608 * config/host-linux.c (linux_gt_pch_use_address): Don't
3609 use SSIZE_MAX because it is not always defined.
3611 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3614 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3615 threshold for pseudo splitting.
3616 (update_ebb_live_info): Process call argument hard registers and
3617 hard registers from insn definition too.
3618 (max_small_class_regs_num): New constant.
3619 (inherit_in_ebb): Update live hard regs through EBBs. Update
3620 reloads_num only for small register classes. Don't split for
3623 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3626 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3629 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3631 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3634 2014-02-04 Marek Polacek <polacek@redhat.com>
3636 * gdbinit.in (pel): Define.
3638 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3640 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3643 2014-02-04 Richard Biener <rguenther@suse.de>
3646 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3647 in function context local.
3648 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3649 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3650 similar to LTO_imported_decl_ref.
3652 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3654 PR tree-optimization/60002
3655 * cgraphclones.c (build_function_decl_skip_args): Clear
3658 PR tree-optimization/60023
3659 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3660 false to gsi_replace.
3661 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3662 has been in some EH region and vec_stmt could throw, add
3663 vec_stmt into the same EH region.
3664 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3665 has no lhs, ignore it.
3666 * internal-fn.c (expand_MASK_LOAD): Likewise.
3669 * tree-inline.c (copy_forbidden): Fail for
3670 __attribute__((optimize (0))) functions.
3673 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3674 is set, copy one less argument.
3675 (expand_simd_clones): Don't subtract clone_info->inbranch
3676 from simd_clone_struct_alloc argument.
3678 PR rtl-optimization/57915
3679 * recog.c (simplify_while_replacing): If all unary/binary/relational
3680 operation arguments are constant, attempt to simplify those.
3683 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3684 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3686 2014-02-04 Richard Biener <rguenther@suse.de>
3688 PR tree-optimization/60012
3689 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3690 TBAA disambiguation to all DDRs.
3692 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3695 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3696 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3698 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3701 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3703 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3705 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3706 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3708 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3711 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3712 to figure out targets of polymorphic calls with known decl.
3713 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3714 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3715 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3716 (get_polymorphic_call_info): ... here.
3717 (get_polymorphic_call_info_from_invariant): New function.
3719 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3721 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3722 lookup via vtable pointer; check for type consistency
3723 and turn inconsitent facts into UNREACHABLE.
3724 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3725 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3726 type inconsistent querries; return UNREACHABLE instead.
3728 2014-02-03 Richard Henderson <rth@twiddle.net>
3731 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3732 already processed this node.
3733 (normalize_one_pred_1): Pass along mark_set.
3734 (normalize_one_pred): Create and destroy a pointer_set_t.
3735 (normalize_one_pred_chain): Likewise.
3737 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3739 PR gcov-profile/58602
3740 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3742 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3745 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3746 -fno-devirtualize; try to devirtualize by the knowledge of
3747 virtual table pointer given by aggregate propagation.
3748 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3749 (ipa_print_node_jump_functions): Dump also offset that
3750 is relevant for polymorphic calls.
3751 (determine_known_aggregate_parts): Add arg_type parameter; use it
3752 instead of determining the type from pointer type.
3753 (ipa_compute_jump_functions_for_edge): Update call of
3754 determine_known_aggregate_parts.
3755 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3756 (gimple_get_virt_method_for_binfo): ... here; simplify using
3757 vtable_pointer_value_to_vtable.
3758 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3759 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3760 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3761 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3763 (vtable_pointer_value_to_binfo): ... here.
3764 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3766 2014-02-03 Teresa Johnson <tejohnson@google.com>
3768 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3769 redef of outer loop index variable.
3771 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3775 * doc/extend.texi (Function Attributes): Typo.
3777 2014-02-03 Cong Hou <congh@google.com>
3779 PR tree-optimization/60000
3780 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3781 if the vectorized statement is a store. A store statement can only
3782 appear at the end of pattern statements.
3784 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3786 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3787 (ix86_option_override_internal): Default long double to 64-bit for
3788 32-bit Bionic and to 128-bit for 64-bit Bionic.
3790 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3791 TARGET_LONG_DOUBLE_128 is true.
3792 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3794 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3795 (mlong-double-64): Negate -mlong-double-128.
3796 (mlong-double-128): New option.
3798 * config/i386/i386-c.c (ix86_target_macros): Define
3799 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3801 * doc/invoke.texi: Document -mlong-double-128.
3803 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3805 PR rtl-optimization/60024
3806 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3808 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3810 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3812 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3814 PR rtl-optimization/57662
3815 * sel-sched.c (code_motion_path_driver): Do not mark already not
3816 existing blocks in the visiting bitmap.
3818 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3820 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3821 on the insn being emitted.
3823 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3824 Will Deacon <will.deacon@arm.com>
3826 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3828 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3830 * config/arm/arm-tables.opt: Regenerate.
3832 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3834 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3835 for vector types other than V16QImode.
3836 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3837 define_expand, and call altivec_expand_vec_perm_le when producing
3838 code with little endian element order.
3839 (*altivec_vperm_<mode>_internal): New insn having previous
3840 behavior of altivec_vperm_<mode>.
3841 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3842 altivec_expand_vec_perm_le when producing code with little endian
3844 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3845 behavior of altivec_vperm_<mode>_uns.
3847 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3849 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3850 (altivec_vsumsws): Add handling for -maltivec=be with a little
3852 (altivec_vsumsws_direct): New.
3853 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3854 gen_altivec_vsumsws.
3856 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3858 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3859 vtable_pointer_value_to_binfo): New functions.
3860 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3861 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3863 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3865 * config/nios2/nios2.md (load_got_register): Initialize GOT
3866 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3867 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3869 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3871 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3872 preserverd by passthrough, do not propagate the type.
3874 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3876 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3877 (mips_atomic_assign_expand_fenv): New function.
3878 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3880 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3882 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3883 (__builtin_mips_set_fcsr): Likewise.
3884 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3885 MIPS_USI_FTYPE_VOID.
3886 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3887 (mips16_expand_set_fcsr): Likewise.
3888 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3889 (mips16_set_fcsr_stub): Likewise.
3890 (mips16_get_fcsr_one_only_stub): New class.
3891 (mips16_set_fcsr_one_only_stub): Likewise.
3892 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3893 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3894 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3895 (hard_float): New availability predicate.
3896 (mips_builtins): Add get_fcsr and set_fcsr.
3897 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3898 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3899 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3900 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3901 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3904 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3906 * config/mips/mips.c (mips_one_only_stub): New class.
3907 (mips_need_mips16_rdhwr_p): Replace with...
3908 (mips16_rdhwr_stub): ...this new variable.
3909 (mips16_stub_call_address): New function.
3910 (mips16_rdhwr_one_only_stub): New class.
3911 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3912 (mips_output_mips16_rdhwr): Delete.
3913 (mips_finish_stub): New function.
3914 (mips_code_end): Use it to handle rdhwr stubs.
3916 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3919 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3920 when calculating size of integer atomic types.
3922 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3924 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3926 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3928 PR tree-optimization/60003
3929 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3930 * profile.c (branch_prob): Use gimple_call_builtin_p
3931 to check for BUILT_IN_SETJMP_RECEIVER.
3932 * tree-inline.c (copy_bb): Call notice_special_calls.
3934 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3937 * lra-constraints.c (process_alt_operands): Update reload_sum only
3940 2014-01-31 Richard Henderson <rth@redhat.com>
3943 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3944 until after else_eh is processed.
3946 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3948 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3949 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3950 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3951 in smmintrin.h, remove them.
3952 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3953 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3954 * config/i386/i386.md (ROUND_SAE): Fix value.
3955 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3956 (const48_operand): New.
3957 * config/i386/subst.md (round), (round_expand): Use
3958 const_4_or_8_to_11_operand.
3959 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3961 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3963 * config/i386/constraints.md (Yk): Swap meaning with k.
3964 * config/i386/i386.md (movhi_internal): Change Yk to k.
3965 (movqi_internal): Ditto.
3966 (*k<logic><mode>): Ditto.
3969 (kandn<mode>): Ditto.
3970 (*<code>hi_1): Ditto.
3971 (*<code>qi_1): Ditto.
3972 (kxnor<mode>): Ditto.
3973 (kortestzhi): Ditto.
3974 (kortestchi): Ditto.
3976 (*one_cmplhi2_1): Ditto.
3977 (*one_cmplqi2_1): Ditto.
3978 * config/i386/sse.md (): Change k to Yk.
3979 (avx512f_load<mode>_mask): Ditto.
3980 (avx512f_blendm<mode>): Ditto.
3981 (avx512f_store<mode>_mask): Ditto.
3982 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3983 (avx512f_storedqu<mode>_mask): Ditto.
3984 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3986 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3987 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3988 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3989 (avx512f_maskcmp<mode>3): Ditto.
3990 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3991 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3992 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3993 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3994 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3995 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3996 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3997 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3998 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3999 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4000 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4001 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4002 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4003 (vec_extract_lo_<mode>_maskm): Ditto.
4004 (vec_extract_hi_<mode>_maskm): Ditto.
4005 (avx512f_vternlog<mode>_mask): Ditto.
4006 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4007 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4008 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4009 (avx512f_<code>v8div16qi2_mask): Ditto.
4010 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4011 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4012 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4013 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4014 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4015 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4016 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4017 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4018 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4019 (avx512cd_maskb_vec_dupv8di): Ditto.
4020 (avx512cd_maskw_vec_dupv16si): Ditto.
4021 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4022 (avx512f_vpermi2var<mode>3_mask): Ditto.
4023 (avx512f_vpermi2var<mode>3_mask): Ditto.
4024 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4025 (*avx512f_gathersi<mode>): Ditto.
4026 (*avx512f_gathersi<mode>_2): Ditto.
4027 (*avx512f_gatherdi<mode>): Ditto.
4028 (*avx512f_gatherdi<mode>_2): Ditto.
4029 (*avx512f_scattersi<mode>): Ditto.
4030 (*avx512f_scatterdi<mode>): Ditto.
4031 (avx512f_compress<mode>_mask): Ditto.
4032 (avx512f_compressstore<mode>_mask): Ditto.
4033 (avx512f_expand<mode>_mask): Ditto.
4034 * config/i386/subst.md (mask): Change k to Yk.
4035 (mask_scalar_merge): Ditto.
4038 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4040 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4042 2014-01-31 Richard Biener <rguenther@suse.de>
4045 * builtins.c (fold_builtin_memory_op): Make sure to not
4046 use a floating-point mode or a boolean or enumeral type for
4049 2014-01-30 DJ Delorie <dj@redhat.com>
4051 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4052 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4053 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4054 whenever main() has an epilogue.
4056 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4058 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4059 unused variable "field".
4060 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4061 (vsx_mergeh_<mode>): Likewise.
4062 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4063 (altivec_vmrghh): Likewise.
4064 (altivec_vmrghw): Likewise.
4065 (altivec_vmrglb): Likewise.
4066 (altivec_vmrglh): Likewise.
4067 (altivec_vmrglw): Likewise.
4068 (altivec_vspltb): Add missing uses.
4069 (altivec_vsplth): Likewise.
4070 (altivec_vspltw): Likewise.
4071 (altivec_vspltsf): Likewise.
4073 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4076 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4077 frame related instructions.
4079 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4081 PR rtl-optimization/59959
4082 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4083 any reload of register whose subreg is invalid.
4085 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4087 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4088 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4089 Add missing return type - void.
4091 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4093 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4094 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4095 remove element index adjustment for endian (now handled in vsx.md
4097 (altivec_expand_vec_perm_const): Use
4098 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4099 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4100 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4101 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4102 define_expand and a new define_insn *altivec_vspltb_internal;
4103 adjust for -maltivec=be on a little endian target.
4104 (altivec_vspltb_direct): New.
4105 (altivec_vsplth): Divide into a define_expand and a new
4106 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4107 little endian target.
4108 (altivec_vsplth_direct): New.
4109 (altivec_vspltw): Divide into a define_expand and a new
4110 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4111 little endian target.
4112 (altivec_vspltw_direct): New.
4113 (altivec_vspltsf): Divide into a define_expand and a new
4114 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4115 a little endian target.
4117 2014-01-30 Richard Biener <rguenther@suse.de>
4119 PR tree-optimization/59993
4120 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4121 can propagate form the earlier stmt and avoid the transform
4122 when the intermediate result is needed.
4124 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4126 * README.Portability: Fix typo.
4128 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4130 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4131 comparison_operator with ordered_comparison_operator.
4133 2014-01-30 Nick Clifton <nickc@redhat.com>
4135 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4136 Rename to mn10300_store_multiple_regs.
4137 * config/mn10300/mn10300.c: Likewise.
4138 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4139 store_multiple_regs.
4140 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4141 Call mn10300_store_multiple_regs.
4143 2014-01-30 Nick Clifton <nickc@redhat.com>
4144 DJ Delorie <dj@redhat.com>
4146 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4147 %fp 2 to keep registers after it properly word-aligned.
4148 (rl78_alloc_physical_registers_umul): Handle the case where both
4149 input operands are the same.
4151 2014-01-30 Richard Biener <rguenther@suse.de>
4153 PR tree-optimization/59903
4154 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4157 2014-01-30 Jason Merrill <jason@redhat.com>
4160 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4163 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4165 2014-01-30 Richard Biener <rguenther@suse.de>
4167 PR tree-optimization/59951
4168 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4170 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4173 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4174 SFmode to DFmode case.
4176 2014-01-29 DJ Delorie <dj@redhat.com>
4178 * config/msp430/msp430.opt (-minrt): New.
4179 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4181 (ENDFILE_SPEC): Likewise.
4183 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4185 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4186 (estimate_function_body_sizes): Use it.
4188 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4191 * dwarf2out.c (is_cxx_auto): New.
4192 (is_base_type): Use it.
4193 (gen_type_die_with_usage): Likewise.
4195 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4197 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4198 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4199 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4200 -maltivec=be with LE targets.
4201 (vsx_mergeh_<mode>): Likewise.
4202 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4203 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4204 (altivec_vmrghb): Replace with define_expand and new
4205 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4206 (altivec_vmrghb_direct): New define_insn.
4207 (altivec_vmrghh): Replace with define_expand and new
4208 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4209 (altivec_vmrghh_direct): New define_insn.
4210 (altivec_vmrghw): Replace with define_expand and new
4211 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4212 (altivec_vmrghw_direct): New define_insn.
4213 (*altivec_vmrghsf): Adjust for endianness.
4214 (altivec_vmrglb): Replace with define_expand and new
4215 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4216 (altivec_vmrglb_direct): New define_insn.
4217 (altivec_vmrglh): Replace with define_expand and new
4218 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4219 (altivec_vmrglh_direct): New define_insn.
4220 (altivec_vmrglw): Replace with define_expand and new
4221 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4222 (altivec_vmrglw_direct): New define_insn.
4223 (*altivec_vmrglsf): Adjust for endianness.
4224 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4225 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4226 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4227 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4228 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4229 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4230 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4231 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4233 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4235 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4236 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4239 2014-01-29 Richard Biener <rguenther@suse.de>
4241 PR tree-optimization/58742
4242 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4243 associate_pointerplus_align.
4244 (associate_pointerplus_diff): New function.
4245 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4246 and associate_pointerplus_diff.
4248 2014-01-29 Richard Biener <rguenther@suse.de>
4250 * lto-streamer.h (LTO_major_version): Bump to 3.
4251 (LTO_minor_version): Reset to 0.
4253 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4255 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4256 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4257 (arm_file_start): Generate correct asm header for armv7ve.
4258 * config/arm/bpabi.h: Add multilib support for armv7ve.
4259 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4260 and cortex-a15 to armv7ve.
4261 * config/arm/t-aprofile: Add multilib support for armv7ve.
4262 * doc/invoke.texi: Document -march=armv7ve.
4264 2014-01-29 Richard Biener <rguenther@suse.de>
4266 PR tree-optimization/58742
4267 * tree-ssa-forwprop.c (associate_plusminus): Return true
4268 if we changed sth, defer EH cleanup to ...
4269 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4270 (simplify_mult): New function.
4272 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4275 PR tree-optimization/59920
4276 * tree.c (build_common_builtin_nodes): Remove
4277 __builtin_setjmp_dispatcher initialization.
4278 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4279 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4280 instead of gsi_after_labels + manually skipping debug stmts.
4281 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4282 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4283 * tree-inline.c (copy_edges_for_bb): Remove
4284 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4285 argument. Ignore computed_goto_p stmts. Don't call
4286 make_abnormal_goto_edges. If a call might need abnormal edges
4287 for non-local gotos, see if it already has an edge to
4288 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4289 with true argument, don't do anything then, otherwise add
4290 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4291 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4293 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4294 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4295 (lower_stmt): Don't set data->calls_builtin_setjmp.
4296 (lower_builtin_setjmp): Adjust comment.
4297 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4298 * tree-cfg.c (found_computed_goto): Remove.
4299 (factor_computed_gotos): Remove.
4300 (make_goto_expr_edges): Return bool, true for computed gotos.
4301 Don't call make_abnormal_goto_edges.
4302 (build_gimple_cfg): Don't set found_computed_goto, don't call
4303 factor_computed_gotos.
4304 (computed_goto_p): No longer static.
4305 (make_blocks): Don't set found_computed_goto.
4306 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4307 (make_edges): If make_goto_expr_edges returns true, push bb
4308 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4309 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4310 vector. Record mapping between bbs and OpenMP regions if there
4311 are any, adjust make_gimple_omp_edges caller. Call
4312 handle_abnormal_edges.
4313 (make_abnormal_goto_edges): Remove.
4314 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4315 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4316 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4317 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4318 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4319 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4320 filling *region also set *region_idx to (*region)->entry->index.
4323 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4324 For REGs set ORIGINAL_REGNO.
4326 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4328 * doc/md.texi: Mention that a target shouldn't implement
4329 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4332 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4334 PR tree-optimization/59594
4335 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4336 a copy of the datarefs vector rather than the vector itself.
4338 2014-01-28 Jason Merrill <jason@redhat.com>
4341 * dwarf2out.c (auto_die): New static.
4342 (gen_type_die_with_usage): Handle C++1y 'auto'.
4343 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4346 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4349 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4350 (SPEC_X32): Likewise.
4351 (SPEC_64): Likewise.
4352 * config/i386/i386.c (ix86_option_override_internal): Turn off
4353 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4355 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4356 * config/i386/i386.h (TARGET_16BIT): New macro.
4357 (TARGET_16BIT_P): Likewise.
4358 * config/i386/i386.opt: Add m16.
4359 * doc/invoke.texi: Document -m16.
4361 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4363 PR preprocessor/59935
4364 * input.c (location_get_source_line): Bail out on when line number
4365 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4367 2014-01-28 Richard Biener <rguenther@suse.de>
4369 PR tree-optimization/58742
4370 * tree-ssa-forwprop.c (associate_plusminus): Handle
4371 pointer subtraction of the form (T)(P + A) - (T)P.
4373 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4375 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4378 2014-01-28 Richard Biener <rguenther@suse.de>
4381 2014-01-28 Richard Biener <rguenther@suse.de>
4383 PR rtl-optimization/45364
4384 PR rtl-optimization/59890
4385 * var-tracking.c (local_get_addr_clear_given_value): Handle
4386 already cleared slot.
4387 (val_reset): Handle not allocated local_get_addr_cache.
4388 (vt_find_locations): Use post-order on the inverted CFG.
4390 2014-01-28 Richard Biener <rguenther@suse.de>
4392 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4394 2014-01-28 Richard Biener <rguenther@suse.de>
4396 PR rtl-optimization/45364
4397 PR rtl-optimization/59890
4398 * var-tracking.c (local_get_addr_clear_given_value): Handle
4399 already cleared slot.
4400 (val_reset): Handle not allocated local_get_addr_cache.
4401 (vt_find_locations): Use post-order on the inverted CFG.
4403 2014-01-28 Alan Modra <amodra@gmail.com>
4405 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4406 * configure.ac <recursive call for build != host>: Define
4407 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4408 and LD_FOR_BUILD too.
4409 * configure: Regenerate.
4411 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4413 * config/i386/i386.c (get_builtin_code_for_version): Separate
4414 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4415 Broadwell from Haswell.
4417 2014-01-27 Steve Ellcey <sellcey@mips.com>
4419 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4420 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4421 * config/mips/mips.c (mips_option_override): Change setting
4423 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4424 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4425 Change from Mask to Var.
4427 2014-01-27 Jeff Law <law@redhat.com>
4429 * ipa-inline.c (inline_small_functions): Fix typo.
4431 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4433 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4434 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4435 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4436 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4437 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4438 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4439 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4440 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4441 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4442 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4443 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4444 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4445 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4446 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4447 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4448 (_mm512_storeu_epi64): Ditto.
4449 (_mm512_cmpge_epi32_mask): Ditto.
4450 (_mm512_cmpge_epu32_mask): Ditto.
4451 (_mm512_cmpge_epi64_mask): Ditto.
4452 (_mm512_cmpge_epu64_mask): Ditto.
4453 (_mm512_cmple_epi32_mask): Ditto.
4454 (_mm512_cmple_epu32_mask): Ditto.
4455 (_mm512_cmple_epi64_mask): Ditto.
4456 (_mm512_cmple_epu64_mask): Ditto.
4457 (_mm512_cmplt_epi32_mask): Ditto.
4458 (_mm512_cmplt_epu32_mask): Ditto.
4459 (_mm512_cmplt_epi64_mask): Ditto.
4460 (_mm512_cmplt_epu64_mask): Ditto.
4461 (_mm512_cmpneq_epi32_mask): Ditto.
4462 (_mm512_cmpneq_epu32_mask): Ditto.
4463 (_mm512_cmpneq_epi64_mask): Ditto.
4464 (_mm512_cmpneq_epu64_mask): Ditto.
4465 (_mm512_expand_pd): Ditto.
4466 (_mm512_expand_ps): Ditto.
4467 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4468 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4469 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4470 * config/i386/i386.c (ix86_builtins): Add
4471 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4472 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4473 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4474 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4475 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4476 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4477 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4478 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4479 IX86_BUILTIN_PMOVUSQW512_MEM.
4480 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4481 __builtin_ia32_pmovsqd512mem_mask,
4482 __builtin_ia32_pmovqd512mem_mask,
4483 __builtin_ia32_pmovusqw512mem_mask,
4484 __builtin_ia32_pmovsqw512mem_mask,
4485 __builtin_ia32_pmovqw512mem_mask,
4486 __builtin_ia32_pmovusdw512mem_mask,
4487 __builtin_ia32_pmovsdw512mem_mask,
4488 __builtin_ia32_pmovdw512mem_mask,
4489 __builtin_ia32_pmovqb512mem_mask,
4490 __builtin_ia32_pmovusqb512mem_mask,
4491 __builtin_ia32_pmovsqb512mem_mask,
4492 __builtin_ia32_pmovusdb512mem_mask,
4493 __builtin_ia32_pmovsdb512mem_mask,
4494 __builtin_ia32_pmovdb512mem_mask.
4495 (bdesc_args): Add __builtin_ia32_expanddf512,
4496 __builtin_ia32_expandsf512.
4497 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4498 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4499 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4500 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4501 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4502 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4503 (avx512f_<code>v8div16qi2_mask_store): This.
4504 (avx512f_expand<mode>): New.
4506 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4508 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4510 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4511 (_mm512_prefetch_i32scatter_pd): Ditto.
4512 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4513 (_mm512_prefetch_i64scatter_pd): Ditto.
4514 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4515 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4516 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4517 (_mm512_prefetch_i32scatter_ps): Ditto.
4518 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4519 (_mm512_prefetch_i64scatter_ps): Ditto.
4520 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4521 * config/i386/i386-builtin-types.def: Define
4522 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4523 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4524 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4525 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4526 IX86_BUILTIN_SCATTERPFQPD.
4527 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4528 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4529 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4530 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4531 __builtin_ia32_scatterpfqps.
4532 (ix86_expand_builtin): Expand new built-ins.
4533 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4534 fix memory access data type.
4535 (*avx512pf_gatherpf<mode>_mask): Ditto.
4536 (*avx512pf_gatherpf<mode>): Ditto.
4537 (avx512pf_scatterpf<mode>): Ditto.
4538 (*avx512pf_scatterpf<mode>_mask): Ditto.
4539 (*avx512pf_scatterpf<mode>): Ditto.
4540 (GATHER_SCATTER_SF_MEM_MODE): New.
4541 (avx512pf_gatherpf<mode>df): Ditto.
4542 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4543 (*avx512pf_scatterpf<mode>df): Ditto.
4545 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4548 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4549 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4552 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4554 * common/config/arm/arm-common.c
4555 (arm_rewrite_mcpu): Handle multiple names.
4557 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4559 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4561 * gimple-builder.h (create_gimple_tmp): Delete.
4563 2014-01-27 Christian Bruel <christian.bruel@st.com>
4565 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4568 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4570 * config/pa/pa.md (call): Generate indirect long calls to non-local
4571 functions when outputing 32-bit code.
4572 (call_value): Likewise except for special call to buggy powf function.
4574 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4575 portable runtime and PIC indirect calls.
4576 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4577 and PIC call sequences. Use ldo instead of blr to set return register
4578 in PIC call sequence.
4580 2014-01-25 Walter Lee <walt@tilera.com>
4582 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4583 avoid clobbering a live register.
4585 2014-01-25 Walter Lee <walt@tilera.com>
4587 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4588 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4589 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4590 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4592 2014-01-25 Walter Lee <walt@tilera.com>
4594 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4595 arguments on even registers.
4596 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4598 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4599 (BIGGEST_ALIGNMENT): Ditto.
4600 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4602 2014-01-25 Walter Lee <walt@tilera.com>
4604 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4605 insns before bundling.
4606 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4608 2014-01-25 Walter Lee <walt@tilera.com>
4610 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4611 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4612 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4614 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4616 * config/mips/constraints.md (kl): Delete.
4617 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4618 define expands, using...
4619 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4620 instructions for MIPS16.
4621 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4622 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4624 2014-01-25 Walter Lee <walt@tilera.com>
4626 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4630 2014-01-25 Walter Lee <walt@tilera.com>
4632 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4633 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4635 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4637 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4640 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4642 * print-rtl.c (in_call_function_usage): New var.
4643 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4644 EXPR_LIST mode as mode and not as reg note name.
4647 * cfgloopmanip.c (copy_loop_info): If
4648 loop->warned_aggressive_loop_optimizations, make sure
4649 the flag is set in target loop too.
4651 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4653 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4655 * builtins.def: Likewise.
4656 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4657 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4658 * ira.c (ira_setup_eliminable_regset): Likewise.
4659 * omp-low.c (gate_expand_omp): Likewise.
4660 (execute_lower_omp): Likewise.
4661 (diagnose_sb_0): Likewise.
4662 (gate_diagnose_omp_blocks): Likewise.
4663 (simd_clone_clauses_extract): Likewise.
4666 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4668 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4669 correction for little endian...
4670 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4673 2014-01-24 Jeff Law <law@redhat.com>
4675 PR tree-optimization/59919
4676 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4677 for non-returning calls.
4679 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4681 * common/config/aarch64/aarch64-common.c
4682 (aarch64_rewrite_mcpu): Handle multiple names.
4683 * config/aarch64/aarch64.h
4684 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4686 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4688 * input.c (add_file_to_cache_tab): Handle the case where fopen
4691 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4694 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4695 from push operand if code of push isn't PRE_DEC.
4697 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4700 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4701 -mquad-memory-atomic. Update -mquad-memory documentation to say
4702 it is only used for non-atomic loads/stores.
4704 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4705 -mquad-memory or -mquad-memory-atomic switches.
4707 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4708 -mquad-memory-atomic to ISA 2.07 support.
4710 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4711 to separate support of normal quad word memory operations (ldq, stq)
4712 from the atomic quad word memory operations.
4714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4715 support to separate non-atomic quad word operations from atomic
4716 quad word operations. Disable non-atomic quad word operations in
4717 little endian mode so that we don't have to swap words after the
4718 load and before the store.
4719 (quad_load_store_p): Add comment about atomic quad word support.
4720 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4721 options printed with -mdebug=reg.
4723 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4724 -mquad-memory-atomic as the test for whether we have quad word
4725 atomic instructions.
4726 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4727 or -mp8-vector are used, allow byte/half-word atomic operations.
4729 * config/rs6000/sync.md (load_lockedti): Insure that the address
4730 is a proper indexed or indirect address for the lqarx instruction.
4731 On little endian systems, swap the hi/lo registers after the lqarx
4733 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4734 insure the address is valid for the lqarx instruction.
4735 (store_conditionalti): Insure that the address is a proper indexed
4736 or indirect address for the stqcrx. instruction. On little endian
4737 systems, swap the hi/lo registers before doing the stqcrx.
4739 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4740 insure the address is valid for the stqcrx. instruction.
4742 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4743 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4744 type of quad memory support is available.
4746 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4749 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4750 there is a danger of looping.
4752 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4754 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4755 force flag_ira_loop_pressure if set via command line.
4757 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4759 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4760 (ashr_simd): New builtin handling DI mode.
4761 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4762 (aarch64_sshr_simddi): New match pattern.
4763 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4764 (vshrd_n_s64): Likewise.
4765 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4767 2014-01-23 Nick Clifton <nickc@redhat.com>
4769 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4770 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4771 favour of mcu specific scripts.
4772 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4775 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4776 Alex Velenko <Alex.Velenko@arm.com>
4778 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4779 (vaddv_s16): Likewise.
4780 (vaddv_s32): Likewise.
4781 (vaddv_u8): Likewise.
4782 (vaddv_u16): Likewise.
4783 (vaddv_u32): Likewise.
4784 (vaddvq_s8): Likewise.
4785 (vaddvq_s16): Likewise.
4786 (vaddvq_s32): Likewise.
4787 (vaddvq_s64): Likewise.
4788 (vaddvq_u8): Likewise.
4789 (vaddvq_u16): Likewise.
4790 (vaddvq_u32): Likewise.
4791 (vaddvq_u64): Likewise.
4792 (vaddv_f32): Likewise.
4793 (vaddvq_f32): Likewise.
4794 (vaddvq_f64): Likewise.
4795 (vmaxv_f32): Likewise.
4796 (vmaxv_s8): Likewise.
4797 (vmaxv_s16): Likewise.
4798 (vmaxv_s32): Likewise.
4799 (vmaxv_u8): Likewise.
4800 (vmaxv_u16): Likewise.
4801 (vmaxv_u32): Likewise.
4802 (vmaxvq_f32): Likewise.
4803 (vmaxvq_f64): Likewise.
4804 (vmaxvq_s8): Likewise.
4805 (vmaxvq_s16): Likewise.
4806 (vmaxvq_s32): Likewise.
4807 (vmaxvq_u8): Likewise.
4808 (vmaxvq_u16): Likewise.
4809 (vmaxvq_u32): Likewise.
4810 (vmaxnmv_f32): Likewise.
4811 (vmaxnmvq_f32): Likewise.
4812 (vmaxnmvq_f64): Likewise.
4813 (vminv_f32): Likewise.
4814 (vminv_s8): Likewise.
4815 (vminv_s16): Likewise.
4816 (vminv_s32): Likewise.
4817 (vminv_u8): Likewise.
4818 (vminv_u16): Likewise.
4819 (vminv_u32): Likewise.
4820 (vminvq_f32): Likewise.
4821 (vminvq_f64): Likewise.
4822 (vminvq_s8): Likewise.
4823 (vminvq_s16): Likewise.
4824 (vminvq_s32): Likewise.
4825 (vminvq_u8): Likewise.
4826 (vminvq_u16): Likewise.
4827 (vminvq_u32): Likewise.
4828 (vminnmv_f32): Likewise.
4829 (vminnmvq_f32): Likewise.
4830 (vminnmvq_f64): Likewise.
4832 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4834 * config/aarch64/aarch64-simd.md
4835 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4836 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4837 (*aarch64_mul3_elt<mode>): Likewise.
4838 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4839 (*aarch64_mul3_elt_to_64v2df): Likewise.
4840 (*aarch64_mla_elt<mode>): Likewise.
4841 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4842 (*aarch64_mls_elt<mode>): Likewise.
4843 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4844 (*aarch64_fma4_elt<mode>): Likewise.
4845 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4846 (*aarch64_fma4_elt_to_64v2df): Likewise.
4847 (*aarch64_fnma4_elt<mode>): Likewise.
4848 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4849 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4850 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4851 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4852 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4853 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4854 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4855 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4856 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4858 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4860 * config/aarch64/aarch64-simd.md
4861 (aarch64_be_checked_get_lane<mode>): New define_expand.
4862 * config/aarch64/aarch64-simd-builtins.def
4863 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4864 New builtin definition.
4865 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4866 Use new safe be builtin.
4868 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4870 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4872 (aarch64_be_st1<mode>): Likewise.
4873 (aarch_ld1<VALL:mode>): Define_expand modified.
4874 (aarch_st1<VALL:mode>): Likewise.
4875 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4876 (UNSPEC_ST1): Likewise.
4878 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4880 * config/microblaze/microblaze.md: Add trap insn and attribute
4882 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4884 PR preprocessor/58580
4885 * input.h (location_get_source_line): Take an additional line_size
4887 (void diagnostics_file_cache_fini): Declare new function.
4888 * input.c (struct fcache): New type.
4889 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4890 New static constants.
4891 (diagnostic_file_cache_init, total_lines_num)
4892 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4893 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4894 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4895 (get_next_line, read_next_line, goto_next_line, read_line_num):
4896 New static function definitions.
4897 (diagnostic_file_cache_fini): New function.
4898 (location_get_source_line): Take an additional output line_len
4899 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4901 * diagnostic.c (diagnostic_finish): Call
4902 diagnostic_file_cache_fini.
4903 (adjust_line): Take an additional input parameter for the length
4904 of the line, rather than calculating it with strlen.
4905 (diagnostic_show_locus): Adjust the use of
4906 location_get_source_line and adjust_line with respect to their new
4907 signature. While displaying a line now, do not stop at the first
4908 null byte. Rather, display the zero byte as a space and keep
4909 going until we reach the size of the line.
4910 * Makefile.in: Add vec.o to OBJS-libcommon
4912 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4913 Ilya Tocar <ilya.tocar@intel.com>
4915 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4916 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4917 (__builtin_ia32_kmov16): Ditto.
4918 * config/i386/i386.md (UNSPEC_KMOV): New.
4921 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4923 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4924 (_mm512_storeu_si512): Ditto.
4926 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4929 * rtl.h (get_referenced_operands): Declare.
4930 * recog.c (get_referenced_operands): New function.
4931 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4932 operands have been referenced when recording LO_SUM references.
4934 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4936 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4938 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4940 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4941 Enable for generic and recent AMD targets.
4943 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4945 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4946 ARG_SIZE note when adjustment was eliminated.
4948 2014-01-22 Jeff Law <law@redhat.com>
4950 PR tree-optimization/59597
4951 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4952 in file. Accept new argument REGISTERING and use it to modify
4953 dump output appropriately.
4954 (register_jump_thread): Corresponding changes.
4955 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4956 thread paths involving joiner blocks. Add code to dump cancelled
4957 jump threading paths.
4959 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4961 PR rtl-optimization/59477
4962 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4963 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4965 2014-01-22 Tom Tromey <tromey@redhat.com>
4967 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4969 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4971 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4973 PR rtl-optimization/59896
4974 * lra-constraints.c (process_alt_operands): Check unused note for
4975 matched operands of insn with no output reloads.
4977 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4979 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4980 (mips_move_from_gpr_cost): Likewise.
4982 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4984 PR rtl-optimization/59858
4985 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4986 ira_class_hard_regs_num.
4987 (process_alt_operands): Increase reject for dying matched operand.
4989 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4992 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4993 smaller than size, perform several stores or loads and stores
4994 at dst + count - size to store or copy all of size bytes, rather
4995 than just last modesize bytes.
4997 2014-01-20 DJ Delorie <dj@redhat.com>
4999 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5000 that CLOBBERs are REGs before propogating their values.
5002 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5005 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5006 (cgraph_inline_failed_type): New function.
5007 * cgraph.h (DEFCIFCODE): Add type.
5008 (cgraph_inline_failed_type_t): New enum.
5009 (cgraph_inline_failed_type): New prototype.
5010 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5011 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5012 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5013 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5014 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5015 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5016 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5017 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5018 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5019 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5020 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5021 OPTIMIZATION_MISMATCH.
5022 * tree-inline.c (expand_call_inline): Emit errors during
5023 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5025 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5028 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5029 mode attribute in insn output.
5031 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5033 * output.h (output_constant): Delete.
5034 * varasm.c (output_constant): Make private.
5036 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5038 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5040 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5043 * tree.h (fold_builtin_strcat): New prototype.
5044 * builtins.c (fold_builtin_strcat): No longer static. Add len
5045 argument, if non-NULL, don't call c_strlen. Optimize
5046 directly into __builtin_memcpy instead of __builtin_strcpy.
5047 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5048 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5050 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5052 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5053 for SImode_address_operand operands, having only a REG argument.
5055 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5057 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5058 loader name using mbig-endian.
5059 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5061 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5063 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5067 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5069 * config/aarch64/aarch64-protos.h
5070 (aarch64_cannot_change_mode_class_ptr): Declare.
5071 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5072 aarch64_cannot_change_mode_class_ptr): New.
5073 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5074 backend hook aarch64_cannot_change_mode_class.
5076 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5078 * common/config/aarch64/aarch64-common.c
5079 (aarch64_handle_option): Don't handle any option order logic here.
5080 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5081 selected_cpu, warn on architecture version mismatch.
5082 (aarch64_override_options): Fix parsing order for option strings.
5084 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5085 Iain Sandoe <iain@codesourcery.com>
5088 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5089 warning. Amend comment to reflect current functionality.
5091 2014-01-20 Richard Biener <rguenther@suse.de>
5094 * builtins.c (fold_builtin_strcat): Remove case better handled
5095 by tree-ssa-strlen.c.
5097 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5099 * config/aarch64/aarch64.opt
5100 (mcpu, march, mtune): Make case-insensitive.
5102 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5105 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5106 if operands[1] is a REG or ZERO_EXTEND of a REG.
5108 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5110 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5112 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5114 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5115 long non-pic millicode calls.
5117 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5119 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5121 2014-01-19 Kito Cheng <kito@0xlab.org>
5123 * builtins.c (expand_movstr): Check movstr expand done or fail.
5125 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5126 H.J. Lu <hongjiu.lu@intel.com>
5129 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5130 to DImode for zero-extended addresses.
5132 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5134 PR rtl-optimization/57763
5135 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5136 on the new indirect jump_insn and increment LABEL_NUSES (label).
5138 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5142 * config.gcc (x86_archs): New variable.
5143 (x86_64_archs): Likewise.
5144 (x86_cpus): Likewise.
5145 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5146 --with-arch/--with-cpu= options.
5147 Support --with-arch=/--with-cpu={nehalem,westmere,
5148 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5150 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5152 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5153 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5155 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5157 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5159 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5162 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5163 clear cpp_get_options (parse_in)->warn_unused_macros for
5164 ix86_target_macros_internal with cpp_define.
5166 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5168 * jump.c (delete_related_insns): Keep (use (insn))s.
5169 * reorg.c (redundant_insn): Check for barriers too.
5171 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5173 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5175 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5177 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5178 call to $$dyncall when TARGET_LONG_CALLS is true.
5180 2014-01-17 Jeff Law <law@redhat.com>
5182 * ree.c (combine_set_extension): Temporarily disable test for
5183 changing number of hard registers.
5185 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5188 * ipa-inline-analysis.c (inline_free_summary):
5189 Do not free summary of aliases.
5191 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5194 * gimplify.c (gimplify_expr): Use create_tmp_var
5195 instead of create_tmp_var_raw. If cond doesn't have
5196 integral type, don't add the IFN_ANNOTATE builtin at all.
5198 2014-01-17 Martin Jambor <mjambor@suse.cz>
5201 * ipa-cp.c (prev_edge_clone): New variable.
5202 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5203 Also resize prev_edge_clone vector.
5204 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5205 (ipcp_edge_removal_hook): New function.
5206 (ipcp_driver): Register ipcp_edge_removal_hook.
5208 2014-01-17 Andrew Pinski <apinski@cavium.com>
5209 Steve Ellcey <sellcey@mips.com>
5212 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5215 2014-01-17 Jeff Law <law@redhat.com>
5218 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5219 so that pass_ccp runs first.
5221 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5223 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5224 (ix86_adjust_cost): Use !TARGET_XXX.
5225 (do_reorder_for_imul): Likewise.
5226 (swap_top_of_ready_list): Likewise.
5227 (ix86_sched_reorder): Likewise.
5229 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5231 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5232 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5233 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5234 (intel_memset): New. Duplicate slm_memset.
5235 (intel_cost): New. Duplicate slm_cost.
5236 (m_INTEL): New macro.
5237 (processor_target_table): Add "intel".
5238 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5239 with PROCESSOR_INTEL for "intel".
5240 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5241 PROCESSOR_SILVERMONT.
5242 (ix86_issue_rate): Likewise.
5243 (ix86_adjust_cost): Likewise.
5244 (ia32_multipass_dfa_lookahead): Likewise.
5245 (swap_top_of_ready_list): Likewise.
5246 (ix86_sched_reorder): Likewise.
5247 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5248 instead of TARGET_OPT_AGU.
5249 * config/i386/i386.h (TARGET_INTEL): New.
5250 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5251 (processor_type): Add PROCESSOR_INTEL.
5252 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5253 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5255 2014-01-17 Marek Polacek <polacek@redhat.com>
5258 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5261 2014-01-17 Richard Biener <rguenther@suse.de>
5263 PR tree-optimization/46590
5264 * opts.c (default_options_table): Add entries for
5265 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5266 all enabled at -O1 but not for -Og.
5267 * common.opt (fbranch-count-reg): Remove Init(1).
5268 (fmove-loop-invariants): Likewise.
5269 (ftree-pta): Likewise.
5271 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5273 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5274 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5275 decls to at least the GCC 4.8 used alignments.
5278 * tree-nested.c (convert_nonlocal_reference_stmt,
5279 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5280 of GIMPLE_BIND stmts, adjust associated decls.
5282 2014-01-17 Richard Biener <rguenther@suse.de>
5284 PR tree-optimization/46590
5285 * vec.h (vec<>::bseach): New member function implementing
5286 binary search according to C89 bsearch.
5287 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5288 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5289 bitmap pointer again. Make accesses_in_loop a flat array.
5290 (mem_ref_obstack): New global.
5291 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5292 (mark_ref_stored): Likewise.
5293 (ref_indep_loop_p_2): Likewise.
5294 (set_ref_stored_in_loop): New helper function.
5295 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5296 (memref_free): Adjust.
5297 (record_mem_ref_loc): Simplify.
5298 (gather_mem_refs_stmt): Adjust.
5299 (sort_locs_in_loop_postorder_cmp): New function.
5300 (analyze_memory_references): Sort accesses_in_loop after
5301 loop postorder number.
5302 (find_ref_loc_in_loop_cmp): New function.
5303 (for_all_locs_in_loop): Find relevant cluster of locs in
5304 accesses_in_loop and iterate without recursion.
5305 (execute_sm): Avoid uninit warning.
5306 (struct ref_always_accessed): Simplify.
5307 (ref_always_accessed::operator ()): Likewise.
5308 (ref_always_accessed_p): Likewise.
5309 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5310 loop postorder numbers here.
5311 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5314 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5317 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5318 on decls for which assemble_alias has been called.
5320 2014-01-17 Nick Clifton <nickc@redhat.com>
5322 * config/msp430/msp430.opt: (mcpu): New option.
5323 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5324 (msp430_option_override): Parse target_cpu. If the MCU name
5325 matches a generic string, clear target_mcu.
5326 (msp430_attr): Allow numeric interrupt values up to 63.
5327 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5328 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5330 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5332 * config/msp430/msp430.md (popm): Use %J rather than %I.
5333 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5334 (addhi_cy_i): Use immediate_operand for operand 2.
5335 * doc/invoke.texi: Document -mcpu option.
5337 2014-01-17 Richard Biener <rguenther@suse.de>
5339 PR rtl-optimization/38518
5340 * df.h (df_analyze_loop): Declare.
5341 * df-core.c: Include cfgloop.h.
5342 (df_analyze_1): Split out main part of df_analyze.
5343 (df_analyze): Adjust.
5344 (loop_inverted_post_order_compute): New function.
5345 (loop_post_order_compute): Likewise.
5346 (df_analyze_loop): New function avoiding whole-function
5348 * loop-invariant.c (find_defs): Use df_analyze_loop.
5349 (find_invariants): Adjust.
5350 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5352 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5354 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5355 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5357 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5359 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5360 traversal when removing references.
5362 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5365 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5367 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5370 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5371 pushing a reload for an autoinc when we had previously reloaded an
5372 inner part of the address.
5374 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5376 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5378 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5379 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5380 when not giving up or versioning for alias only because of
5382 (vect_analyze_data_ref_dependences): Set to true.
5383 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5385 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5386 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5390 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5393 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5394 operand 0 predicate for gathers, use a new pseudo as subtarget.
5396 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5399 * lra-constraints.c (process_alt_operands): Add printing debug info.
5400 Check absence of input/output reloads for matched operands too.
5402 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5404 PR rtl-optimization/59835
5405 * ira.c (ira_init_register_move_cost): Increase cost for
5408 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5410 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5412 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5415 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5416 non-register objects. Use gen_(high/low)part more consistently.
5419 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5422 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5423 endian support, remove tests for WORDS_BIG_ENDIAN.
5424 (p8_mfvsrd_3_<mode>): Likewise.
5425 (reload_gpr_from_vsx<mode>): Likewise.
5426 (reload_gpr_from_vsxsf): Likewise.
5427 (p8_mfvsrd_4_disf): Likewise.
5429 2014-01-16 Richard Biener <rguenther@suse.de>
5431 PR rtl-optimization/46590
5432 * lcm.c (compute_antinout_edge): Use postorder iteration.
5433 (compute_laterin): Use inverted postorder iteration.
5435 2014-01-16 Nick Clifton <nickc@redhat.com>
5438 * varasm.c (output_constant): Return the number of bytes actually
5440 (output_constructor_array_range): Update the field size with the
5441 number of bytes emitted by output_constant.
5442 (output_constructor_regular_field): Likewise. Also do not
5443 complain if the total number of bytes emitted is now greater
5444 than the expected fieldpos.
5445 * output.h (output_constant): Update prototype and descriptive comment.
5447 2014-01-16 Marek Polacek <polacek@redhat.com>
5450 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5451 it is error_mark_node.
5453 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5455 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5456 VALID_AVX256_REG_OR_OI_MODE.
5458 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5460 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5461 current procedure should be profiled.
5463 2014-01-15 Andrew Pinski <apinski@cavium.com>
5465 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5466 of moving from/to the STACK_REG register class.
5468 2014-01-15 Richard Henderson <rth@redhat.com>
5471 * reginfo.c (global_regs_decl): Globalize.
5472 * rtl.h (global_regs_decl): Declare.
5473 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5474 reserved via global_regs.
5476 2014-01-15 Teresa Johnson <tejohnson@google.com>
5478 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5480 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5482 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5483 and vmulosh rather than call gen_vec_widen_smult_*.
5484 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5485 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5486 (vec_widen_smult_even_v16qi): Likewise.
5487 (vec_widen_umult_even_v8hi): Likewise.
5488 (vec_widen_smult_even_v8hi): Likewise.
5489 (vec_widen_umult_odd_v16qi): Likewise.
5490 (vec_widen_smult_odd_v16qi): Likewise.
5491 (vec_widen_umult_odd_v8hi): Likewise.
5492 (vec_widen_smult_odd_v8hi): Likewise.
5493 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5494 vmuloub rather than call gen_vec_widen_umult_*.
5495 (vec_widen_umult_lo_v16qi): Likewise.
5496 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5497 vmulosb rather than call gen_vec_widen_smult_*.
5498 (vec_widen_smult_lo_v16qi): Likewise.
5499 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5500 rather than call gen_vec_widen_umult_*.
5501 (vec_widen_umult_lo_v8hi): Likewise.
5502 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5503 rather than call gen_vec_widen_smult_*.
5504 (vec_widen_smult_lo_v8hi): Likewise.
5506 2014-01-15 Jeff Law <law@redhat.com>
5508 PR tree-optimization/59747
5509 * ree.c (find_and_remove_re): Properly handle case where a second
5510 eliminated extension requires widening a copy created for elimination
5511 of a prior extension.
5512 (combine_set_extension): Ensure that the number of hard regs needed
5513 for a destination register does not change when we widen it.
5515 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5517 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5518 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5519 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5520 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5521 (avr-*-rtems*): Likewise.
5522 (bfin*-rtems*): Likewise.
5523 (moxie-*-rtems*): Likewise.
5524 (h8300-*-rtems*): Likewise.
5525 (i[34567]86-*-rtems*): Likewise.
5526 (lm32-*-rtems*): Likewise.
5527 (m32r-*-rtems*): Likewise.
5528 (m68k-*-rtems*): Likewise.
5529 (microblaze*-*-rtems*): Likewise.
5530 (mips*-*-rtems*): Likewise.
5531 (powerpc-*-rtems*): Likewise.
5532 (sh-*-rtems*): Likewise.
5533 (sparc-*-rtems*): Likewise.
5534 (sparc64-*-rtems*): Likewise.
5535 (v850-*-rtems*): Likewise.
5536 (m32c-*-rtems*): Likewise.
5538 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5540 PR rtl-optimization/59511
5541 * ira.c (ira_init_register_move_cost): Use memory costs for some
5542 cases of register move cost calculations.
5543 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5544 instead of BB frequency.
5545 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5546 * lra-assigns.c (find_hard_regno_for): Ditto.
5548 2014-01-15 Richard Biener <rguenther@suse.de>
5550 PR tree-optimization/59822
5551 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5552 (vectorizable_load): Use it to hoist defs of uses of invariant
5553 loads out of the loop.
5555 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5556 Kugan Vivekanandarajah <kuganv@linaro.org>
5559 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5562 2014-01-15 Richard Biener <rguenther@suse.de>
5564 PR rtl-optimization/59802
5565 * lcm.c (compute_available): Use inverted postorder to seed
5566 the initial worklist.
5568 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5571 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5572 ADDR_REGS for invalid symrefs in non-PIC code.
5574 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5577 * builtins.c (determine_block_size): Initialize *probable_max_size
5578 even if len_rtx is CONST_INT.
5580 2014-01-14 Andrew Pinski <apinski@cavium.com>
5582 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5583 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5584 (cortexa53_tunings): Likewise.
5585 (aarch64_sched_issue_rate): New function.
5586 (TARGET_SCHED_ISSUE_RATE): Define.
5588 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5590 * ira-costs.c (find_costs_and_classes): Add missed
5591 ira_init_register_move_cost_if_necessary.
5593 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5596 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5598 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5601 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5602 to indicate if type is used for function return value. Warn ABI
5603 change if the vector mode isn't available for function return value.
5604 (ix86_function_arg_advance): Pass false to type_natural_mode.
5605 (ix86_function_arg): Likewise.
5606 (ix86_gimplify_va_arg): Likewise.
5607 (function_arg_32): Don't warn ABI change.
5608 (ix86_function_value): Pass true to type_natural_mode.
5609 (ix86_return_in_memory): Likewise.
5610 (ix86_struct_value_rtx): Removed.
5611 (TARGET_STRUCT_VALUE_RTX): Likewise.
5613 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5615 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5616 converting a conditional jump into a conditional return.
5618 2014-01-14 Richard Biener <rguenther@suse.de>
5620 PR tree-optimization/58921
5621 PR tree-optimization/59006
5622 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5623 hoisting invariant stmts.
5624 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5625 invariant loads on the preheader edge if possible.
5627 2014-01-14 Joey Ye <joey.ye@arm.com>
5629 * doc/plugin.texi (Building GCC plugins): Update to C++.
5631 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5633 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5634 (_mm_rcp28_round_ss): Ditto.
5635 (_mm_rsqrt28_round_sd): Ditto.
5636 (_mm_rsqrt28_round_ss): Ditto.
5637 (_mm_rcp28_sd): Ditto.
5638 (_mm_rcp28_ss): Ditto.
5639 (_mm_rsqrt28_sd): Ditto.
5640 (_mm_rsqrt28_ss): Ditto.
5641 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5642 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5643 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5644 (IX86_BUILTIN_RCP28SD): Ditto.
5645 (IX86_BUILTIN_RCP28SS): Ditto.
5646 (IX86_BUILTIN_RSQRT28SD): Ditto.
5647 (IX86_BUILTIN_RSQRT28SS): Ditto.
5648 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5649 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5650 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5651 (ix86_expand_special_args_builtin): Expand new FTYPE.
5652 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5653 (srcp14<mode>): Make insn unary.
5654 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5655 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5656 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5657 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5658 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5659 Fix rounding: make it SAE only.
5660 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5662 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5664 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5665 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5666 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5667 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5668 (round_saeonly_mask_scalar_operand4): Ditto.
5669 (round_saeonly_mask_scalar_op3): Ditto.
5670 (round_saeonly_mask_scalar_op4): Ditto.
5672 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5674 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5675 Implement -maltivec=be for vec_insert and vec_extract.
5677 2014-01-10 DJ Delorie <dj@redhat.com>
5679 * config/msp430/msp430.md (call_internal): Don't allow memory
5680 references with SP as the base register.
5681 (call_value_internal): Likewise.
5682 * config/msp430/constraints.md (Yc): New. For memory references
5683 that don't use SP as a base register.
5685 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5686 "an integer without a # prefix"
5687 * config/msp430/msp430.md (epilogue_helper): Use it.
5689 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5692 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5693 AVX512F gather builtins.
5694 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5695 on gather decls with INTEGER_TYPE masktype.
5696 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5697 directly into the builtin rather than hoisting it before loop.
5699 PR tree-optimization/59387
5700 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5701 (scev_const_prop): If folded_casts and type has undefined overflow,
5702 use force_gimple_operand instead of force_gimple_operand_gsi and
5703 for each added stmt if it is assign with
5704 arith_code_with_undefined_signed_overflow, call
5705 rewrite_to_defined_overflow.
5706 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5707 gimple-fold.h instead.
5708 (arith_code_with_undefined_signed_overflow,
5709 rewrite_to_defined_overflow): Moved to ...
5710 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5711 rewrite_to_defined_overflow): ... here. No longer static.
5712 Include gimplify-me.h.
5713 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5714 rewrite_to_defined_overflow): New prototypes.
5716 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5718 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5720 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5722 * builtins.c (get_object_alignment_2): Minor tweak.
5723 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5725 2014-01-13 Christian Bruel <christian.bruel@st.com>
5727 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5728 optimized non constant lengths.
5730 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5733 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5734 load as __atomic_load_N if possible.
5736 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5738 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5740 (rs6000_expand_builtin): Adjust call.
5742 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5745 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5746 * config/rs6000/rs6000.c: Include target-globals.h.
5747 (rs6000_set_current_function): Instead of doing target_reinit
5748 unconditionally, use save_target_globals_default_opts and
5749 restore_target_globals.
5751 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5753 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5754 (rs6000_expand_builtin): Handle mffs and mtfsf.
5755 (rs6000_init_builtins): Define mffs and mtfsf.
5756 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5757 (rs6000_mffs): New pattern.
5758 (rs6000_mtfsf): New pattern.
5760 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5762 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5763 Start narrowing with START. Apply candidate-use pair
5764 and check overall cost in narrowing.
5765 (iv_ca_prune): Pass new argument.
5767 2014-01-10 Jeff Law <law@redhat.com>
5770 * ree.c (combine_reaching_defs): Ensure the defining statement
5771 occurs before the extension when optimizing extensions with
5772 different source and destination hard registers.
5774 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5777 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5778 vtables into the type inheritance graph.
5780 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5782 PR rtl-optimization/59754
5783 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5784 modes in the REGNO != REGNO case.
5786 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5788 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5790 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5792 PR tree-optimization/59745
5793 * tree-predcom.c (tree_predictive_commoning_loop): Call
5794 free_affine_expand_cache if giving up because components is NULL.
5796 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5797 GC in payload of target_globals struct instead of allocating them on
5798 the heap and the larger structs separately using GC.
5799 * target-globals.h (struct target_globals): Make regs, hard_regs,
5800 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5801 of GTY((skip)) and change type to void *.
5802 (reset_target_globals): Cast loads from those fields to corresponding
5805 2014-01-10 Steve Ellcey <sellcey@mips.com>
5808 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5809 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5810 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5812 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5815 * aarch64-modes.def (CC_Zmode): New flags mode.
5816 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5817 represents an equality.
5818 (aarch64_get_condition_code): Handle CC_Zmode.
5819 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5821 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5823 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5824 extraction in good case.
5826 2014-01-10 Richard Biener <rguenther@suse.de>
5828 PR tree-optimization/59374
5829 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5830 checking after SLP discovery. Mark stmts not participating
5831 in any SLP instance properly.
5833 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5835 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5836 when handling a SET rtx.
5838 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5840 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5841 (cortex-a57): Likewise.
5842 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5844 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5846 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5847 non-iwmmxt builtins.
5849 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5853 * ipa-devirt.c record_target_from_binfo): Take as argument
5854 stack of binfos and lookup matching one for virtual inheritance.
5855 (possible_polymorphic_call_targets_1): Update.
5857 2014-01-10 Huacai Chen <chenhc@lemote.com>
5859 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5860 kernel strings for Loongson-2E/2F/3A.
5862 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5865 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5866 is_gimple_call before calling gimple_call_internal_p.
5868 2014-01-09 Steve Ellcey <sellcey@mips.com>
5870 * Makefile.in (TREE_FLOW_H): Remove.
5871 (TREE_SSA_H): Add file names from tree-flow.h.
5872 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5873 * tree.h: Remove tree-flow.h reference.
5874 * hash-table.h: Remove tree-flow.h reference.
5875 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5876 reference with tree-ssa-loop.h.
5878 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5880 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5881 default element-order behavior for -maltivec.
5882 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5884 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5885 when targeting big endian, at least for now.
5886 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5888 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5891 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5892 var satisfies use_register_for_decl, just take into account type
5893 alignment, rather than decl alignment.
5895 PR tree-optimization/59622
5896 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5897 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5898 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5899 Don't devirtualize for inplace at all. For targets.length () == 1,
5900 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5902 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5904 * config/i386/i386.md (cpu): Remove the unused btver1.
5906 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5908 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5910 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5913 * tree-core.h (struct target_globals): New forward declaration.
5914 (struct tree_target_option): Add globals field.
5915 * tree.h (TREE_TARGET_GLOBALS): Define.
5916 (prepare_target_option_nodes_for_pch): New prototype.
5917 * target-globals.h (struct target_globals): Define even if
5919 * tree.c (prepare_target_option_node_for_pch,
5920 prepare_target_option_nodes_for_pch): New functions.
5921 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5922 * config/i386/i386.c: Include target-globals.h.
5923 (ix86_set_current_function): Instead of doing target_reinit
5924 unconditionally, use save_target_globals_default_opts and
5925 restore_target_globals.
5927 2014-01-09 Richard Biener <rguenther@suse.de>
5929 PR tree-optimization/59715
5930 * tree-cfg.h (split_critical_edges): Declare.
5931 * tree-cfg.c (split_critical_edges): Export.
5932 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5934 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5936 * cfgexpand.c (expand_stack_vars): Optionally disable
5937 asan stack protection.
5938 (expand_used_vars): Likewise.
5939 (partition_stack_vars): Likewise.
5940 * asan.c (asan_emit_stack_protection): Optionally disable
5941 after return stack usage.
5942 (instrument_derefs): Optionally disable memory access instrumentation.
5943 (instrument_builtin_call): Likewise.
5944 (instrument_strlen_call): Likewise.
5945 (asan_protect_global): Optionally disable global variables protection.
5946 * doc/invoke.texi: Added doc for new options.
5947 * params.def: Added new options.
5948 * params.h: Likewise.
5950 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5952 PR rtl-optimization/59724
5953 * ifcvt.c (cond_exec_process_if_block): Don't call
5954 flow_find_head_matching_sequence with 0 longest_match.
5955 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5956 non-active insns if !stop_after.
5957 (try_head_merge_bb): Revert 2014-01-07 changes.
5959 2014-01-08 Jeff Law <law@redhat.com>
5961 * ree.c (get_sub_rtx): New function, extracted from...
5962 (merge_def_and_ext): Here.
5963 (combine_reaching_defs): Use get_sub_rtx.
5965 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5967 * cgraph.h (varpool_variable_node): Do not choke on null node.
5969 2014-01-08 Catherine Moore <clm@codesourcery.com>
5971 * config/mips/mips.md (simple_return): Attempt to use JRC
5973 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5975 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5977 PR rtl-optimization/59137
5978 * reorg.c (steal_delay_list_from_target): Call update_block for
5980 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5982 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5984 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5985 two duplicate entries.
5987 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5990 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5992 * config/mips/mips.c (mips_truncated_op_cost): New function.
5993 (mips_rtx_costs): Adjust test for BADDU.
5994 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5996 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5998 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5999 (*baddu_si): ...this new pattern.
6001 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6004 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6006 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6009 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6011 (expand_expr, expand_normal): Adjust.
6012 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6013 inner_reference_p. Use inner_reference_p to expand inner references.
6014 (store_expr): Adjust.
6015 * cfgexpand.c (expand_call_stmt): Adjust.
6017 2014-01-08 Rong Xu <xur@google.com>
6019 * gcov-io.c (gcov_var): Move from gcov-io.h.
6020 (gcov_position): Ditto.
6021 (gcov_is_error): Ditto.
6022 (gcov_rewrite): Ditto.
6023 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6024 only part to libgcc/libgcov.h.
6026 2014-01-08 Marek Polacek <polacek@redhat.com>
6029 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6031 2014-01-08 Marek Polacek <polacek@redhat.com>
6034 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6036 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6038 PR rtl-optimization/59649
6039 * stor-layout.c (get_mode_bounds): For BImode return
6040 0 and STORE_FLAG_VALUE.
6042 2014-01-08 Richard Biener <rguenther@suse.de>
6045 * gimple.h (is_gimple_builtin_call): Remove.
6046 (gimple_builtin_call_types_compatible_p): New.
6047 (gimple_call_builtin_p): New overload.
6048 * gimple.c (is_gimple_builtin_call): Remove.
6049 (validate_call): Rename to ...
6050 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6052 (validate_type): New static function.
6053 (gimple_call_builtin_p): New overload and adjust.
6054 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6055 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6056 (gimple_fold_stmt_to_constant_1): Likewise.
6057 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6059 2014-01-08 Richard Biener <rguenther@suse.de>
6062 * gimplify.c (gimplify_expr): Gimplify register-register type
6063 VIEW_CONVERT_EXPRs to separate stmts.
6065 2014-01-07 Jeff Law <law@redhat.com>
6068 * ree.c (combine_set_extension): Handle case where source
6069 and destination registers in an extension insn are different.
6070 (combine_reaching_defs): Allow source and destination registers
6071 in extension to be different under limited circumstances.
6072 (add_removable_extension): Remove restriction that the
6073 source and destination registers in the extension are the same.
6074 (find_and_remove_re): Emit a copy from the extension's
6075 destination to its source after the defining insn if
6076 the source and destination registers are different.
6079 * ifcvt.c (merge_if_block): If we are merging a block with more than
6080 one successor with a block with no successors, remove any BARRIER
6081 after the second block.
6083 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6085 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6087 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6090 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6091 for 14-bit register offsets when INT14_OK_STRICT is false.
6093 2014-01-07 Roland Stigge <stigge@antcom.de>
6094 Michael Meissner <meissner@linux.vnet.ibm.com>
6097 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6098 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6100 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6102 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6105 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6107 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6108 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6109 rtx is const0_rtx or not.
6111 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6114 * target-globals.c (save_target_globals): Remove this_fn_optab
6116 * toplev.c: Include optabs.h.
6117 (target_reinit): Temporarily restore the global options if another
6118 set of options are in force.
6120 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6122 PR rtl-optimization/58668
6123 * cfgcleanup.c (flow_find_cross_jump): Don't count
6124 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6125 to determine what is counted.
6126 (flow_find_head_matching_sequence): Use active_insn_p to determine
6128 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6130 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6131 determine what is counted.
6133 PR tree-optimization/59643
6134 * tree-predcom.c (split_data_refs_to_components): If one dr is
6135 read and one write, determine_offset fails and the write isn't
6136 in the bad component, just put the read into the bad component.
6138 2014-01-07 Mike Stump <mikestump@comcast.net>
6139 Jakub Jelinek <jakub@redhat.com>
6142 * tree-core.h (struct tree_optimization_option): Change optabs
6143 type from unsigned char * to void *.
6144 * optabs.c (init_tree_optimization_optabs): Adjust
6145 TREE_OPTIMIZATION_OPTABS initialization.
6147 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6150 * config/i386/i386.h (struct machine_function): Add
6151 no_drap_save_restore field.
6152 * config/i386/i386.c (ix86_save_reg): Use
6153 !cfun->machine->no_drap_save_restore instead of
6154 crtl->stack_realign_needed.
6155 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6156 this function clears frame_pointer_needed. Set
6157 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6158 and DRAP reg is needed.
6160 2014-01-06 Marek Polacek <polacek@redhat.com>
6163 * doc/implement-c.texi: Mention that other integer types are
6164 permitted as bit-field types in strictly conforming mode.
6166 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6168 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6171 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6173 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6175 2014-01-06 Martin Jambor <mjambor@suse.cz>
6178 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6180 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6182 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6186 * var-tracking.c (add_stores): Preserve the value of the source even if
6187 we don't record the store.
6189 2014-01-06 Terry Guo <terry.guo@arm.com>
6191 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6193 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6196 * config/darwin.c (darwin_function_section): Adjust return values to
6197 correspond to optimisation changes made in r206070.
6199 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6201 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6202 from prefetch_block tune setting.
6203 (nocona_cost): Correct size of prefetch block to 64.
6205 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6207 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6208 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6209 used to save the static chain register in the computation of the offset
6210 from which the FP registers need to be restored.
6212 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6214 PR tree-optimization/59519
6215 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6216 ICE if get_current_def (current_new_name) is already non-NULL, as long
6217 as it is a phi result of some other phi in *new_exit_bb that has
6220 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6221 or vmovdqu* for misaligned_operand.
6222 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6223 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6224 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6225 aligned_mem for AVX512F masked aligned load and store builtins and for
6228 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6230 PR tree-optimization/59651
6231 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6232 Address range for negative step should be added by TYPE_SIZE_UNIT.
6234 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6236 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6237 overlapping registers also for registers other than the stack pointer.
6239 2014-01-03 Marek Polacek <polacek@redhat.com>
6242 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6245 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6248 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6251 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6252 (push splitter): Use <P:MODE_SIZE> instead of
6253 GET_MODE_SIZE (<P:MODE>mode).
6254 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6255 (mov -1, reg peephole2): Likewise.
6256 * config/i386/sse.md (*mov<mode>_internal,
6257 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6258 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6259 *<code><mode>3, *andnot<mode>3<mask_name>,
6260 <mask_codefor><code><mode>3<mask_name>): Likewise.
6261 * config/i386/subst.md (mask_mode512bit_condition,
6262 sd_mask_mode512bit_condition): Likewise.
6264 2014-01-02 Xinliang David Li <davidxl@google.com>
6266 PR tree-optimization/59303
6267 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6268 (dump_predicates): Better output format.
6269 (pred_equal_p): New function.
6270 (is_neq_relop_p): Ditto.
6271 (is_neq_zero_form_p): Ditto.
6272 (pred_expr_equal_p): Ditto.
6273 (pred_neg_p): Ditto.
6274 (simplify_pred): Ditto.
6275 (simplify_preds_2): Ditto.
6276 (simplify_preds_3): Ditto.
6277 (simplify_preds_4): Ditto.
6278 (simplify_preds): Ditto.
6280 (push_to_worklist): Ditto.
6281 (get_pred_info_from_cmp): Ditto.
6282 (is_degenerated_phi): Ditto.
6283 (normalize_one_pred_1): Ditto.
6284 (normalize_one_pred): Ditto.
6285 (normalize_one_pred_chain): Ditto.
6286 (normalize_preds): Ditto.
6287 (normalize_cond_1): Remove function.
6288 (normalize_cond): Ditto.
6289 (is_gcond_subset_of): Ditto.
6290 (is_subset_of_any): Ditto.
6291 (is_or_set_subset_of): Ditto.
6292 (is_and_set_subset_of): Ditto.
6293 (is_norm_cond_subset_of): Ditto.
6294 (pred_chain_length_cmp): Ditto.
6295 (convert_control_dep_chain_into_preds): Type change.
6296 (find_predicates): Ditto.
6297 (find_def_preds): Ditto.
6298 (destroy_predicates_vecs): Ditto.
6299 (find_matching_predicates_in_rest_chains): Ditto.
6300 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6301 (is_pred_expr_subset): Ditto.
6302 (is_pred_chain_subset_of): Ditto.
6303 (is_included_in): Ditto.
6304 (is_superset_of): Ditto.
6306 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6308 Update copyright years.
6310 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6312 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6313 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6314 config/arc/arc.md, config/arc/arc.opt,
6315 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6316 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6317 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6318 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6319 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6320 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6321 vtable-verify.c, vtable-verify.h: Use the standard form for the
6324 2014-01-02 Tobias Burnus <burnus@net-b.de>
6326 * gcc.c (process_command): Update copyright notice dates.
6327 * gcov-dump.c: Ditto.
6329 * doc/cpp.texi: Bump @copying's copyright year.
6330 * doc/cppinternals.texi: Ditto.
6331 * doc/gcc.texi: Ditto.
6332 * doc/gccint.texi: Ditto.
6333 * doc/gcov.texi: Ditto.
6334 * doc/install.texi: Ditto.
6335 * doc/invoke.texi: Ditto.
6337 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6339 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6341 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6343 * config/i386/sse.md (*mov<mode>_internal): Guard
6344 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6346 PR rtl-optimization/59647
6347 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6348 new_rtx into UNSIGNED_FLOAT rtxes.
6350 Copyright (C) 2014 Free Software Foundation, Inc.
6352 Copying and distribution of this file, with or without modification,
6353 are permitted in any medium without royalty provided the copyright
6354 notice and this notice are preserved.