1 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
4 nreverse() because it changes the content of original tree list.
6 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
8 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
9 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
11 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
13 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
16 2014-02-13 Jakub Jelinek <jakub@redhat.com>
19 * dwarf2out.c (gen_subprogram_die): Don't call
20 add_calling_convention_attribute if subr_die is old_die.
22 2014-02-13 Sharad Singhai <singhai@google.com>
24 * doc/optinfo.texi: Fix order of nodes.
26 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
28 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
29 operands[2], not operands[3].
31 2014-02-13 Richard Biener <rguenther@suse.de>
34 * doc/install.texi (ISL): Update recommended version to 0.12.2,
35 mention the possibility of an in-tree build.
36 (CLooG): Update recommended version to 0.18.1, mention the
37 possibility of an in-tree build and clarify that the ISL
38 bundled with CLooG does not work.
40 2014-02-13 Jakub Jelinek <jakub@redhat.com>
43 * expr.c (compress_float_constant): If x is a hard register,
44 extend into a pseudo and then move to x.
46 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
48 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
49 caused by bad second argument to warning_at() with -mhotpatch and
50 nested functions (e.g. with gfortran).
52 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
54 * opts.c (option_name): Remove "enabled by default" rider.
56 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
58 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
60 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
61 Uros Bizjak <ubizjak@gmail.com>
64 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
65 * configure: Regenerated.
67 2014-02-12 Richard Biener <rguenther@suse.de>
69 * vec.c (vec_prefix::calculate_allocation): Move as
70 inline variant to vec.h.
71 (vec_prefix::calculate_allocation_1): New out-of-line version.
72 * vec.h (vec_prefix::calculate_allocation_1): Declare.
73 (vec_prefix::m_has_auto_buf): Rename to ...
74 (vec_prefix::m_using_auto_storage): ... this.
75 (vec_prefix::calculate_allocation): Inline the easy cases
76 and dispatch to calculate_allocation_1 which doesn't need the
78 (va_heap::reserve): Use gcc_checking_assert.
79 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
81 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
83 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
84 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
85 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
87 2014-02-12 Richard Biener <rguenther@suse.de>
89 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
90 when we found a dependence.
92 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
94 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
96 (maybe_fold_stmt): ... into this new function.
97 * omp-low.c (lower_omp): Update comment.
99 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
102 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
105 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
107 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
108 identifiers in comments.
109 (cortexa53_extra_costs): Likewise.
110 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
111 (cortexa7_extra_costs): Likewise.
112 (cortexa12_extra_costs): Likewise.
113 (cortexa15_extra_costs): Likewise.
114 (v7m_extra_costs): Likewise.
116 2014-02-12 Richard Biener <rguenther@suse.de>
119 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
120 of posix_memalign being successful.
121 (lower_stmt): Restrict lowering of posix_memalign to when
122 -ftree-bit-ccp is enabled.
124 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
126 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
128 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
130 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
132 PR rtl-optimization/60116
133 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
134 other_insn once the combination has been validated.
136 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
139 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
141 * ipa-devirt.c: Include demangle.h
142 (odr_violation_reported): New static variable.
143 (add_type_duplicate): Update odr_violations.
144 (maybe_record_node): Add completep parameter; update it.
145 (record_target_from_binfo): Add COMPLETEP parameter;
147 (possible_polymorphic_call_targets_1): Likewise.
148 (struct polymorphic_call_target_d): Add nonconstruction_targets;
149 rename FINAL to COMPLETE.
150 (record_targets_from_bases): Sanity check we found the binfo;
151 fix COMPLETEP updating.
152 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
153 parameter, fix computing of COMPLETEP.
154 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
155 at LTO time do demangling.
156 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
157 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
159 (gimple_get_virt_method_for_binfo): Likewise.
160 * gimple-fold.h (gimple_get_virt_method_for_binfo,
161 gimple_get_virt_method_for_vtable): Update prototypes.
163 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
166 * genautomata.c (add_presence_absence): Fix typo with
167 {final_}presence_list.
169 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
172 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
173 for VSX/Altivec vectors that land in GPR registers.
175 2014-02-11 Richard Henderson <rth@redhat.com>
176 Jakub Jelinek <jakub@redhat.com>
179 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
180 around drhs if type conversion to lacc->type is not useless.
182 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
184 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
186 (cortex-a57.cortex-a53): Likewise.
187 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
189 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
191 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
194 2014-02-11 Renlin Li <Renlin.Li@arm.com>
196 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
197 add_options_for_arm_vfp3.
199 2014-02-11 Jeff Law <law@redhat.com>
202 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
203 object with an undesirable mode.
205 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
208 * config/i386/sol2-9.h: New file.
209 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
210 *-*-solaris2.9*): Use it.
212 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
214 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
215 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
217 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
219 * config/microblaze/microblaze.c: Extend mcpu version format
221 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
223 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
225 2014-02-10 Richard Henderson <rth@redhat.com>
228 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
229 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
230 ms-abi vs -mno-accumulate-outgoing-args.
231 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
232 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
235 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
238 * cfgexpand.c (expand_asm_operands): Attach source location to
239 ASM_INPUT rtx objects.
240 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
242 2014-02-10 Nick Clifton <nickc@redhat.com>
244 * config/mn10300/mn10300.c (popcount): New function.
245 (mn10300_expand_prologue): Include saved registers in stack usage
248 2014-02-10 Jeff Law <law@redhat.com>
251 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
252 when changing the SET_DEST of a prior insn to avoid an input reload.
254 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
256 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
257 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
258 -mcall-openbsd, or -mcall-linux.
259 (CC1_ENDIAN_BIG_SPEC): Remove.
260 (CC1_ENDIAN_LITTLE_SPEC): Remove.
261 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
262 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
263 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
264 and %cc1_endian_default.
265 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
267 2014-02-10 Richard Biener <rguenther@suse.de>
269 PR tree-optimization/60115
270 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
271 MEM_REF handling. Properly verify that the accesses are not
272 out of the objects bound.
274 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
276 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
279 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
281 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
282 proper constants and fix formatting.
283 (possible_polymorphic_call_targets): Fix formatting.
285 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
286 Ilya Tocar <ilya.tocar@intel.com>
288 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
289 (_mm512_loadu_epi32): Renamed into...
290 (_mm512_loadu_si512): This.
291 (_mm512_storeu_epi32): Renamed into...
292 (_mm512_storeu_si512): This.
293 (_mm512_maskz_ceil_ps): Removed.
294 (_mm512_maskz_ceil_pd): Ditto.
295 (_mm512_maskz_floor_ps): Ditto.
296 (_mm512_maskz_floor_pd): Ditto.
297 (_mm512_floor_round_ps): Ditto.
298 (_mm512_floor_round_pd): Ditto.
299 (_mm512_ceil_round_ps): Ditto.
300 (_mm512_ceil_round_pd): Ditto.
301 (_mm512_mask_floor_round_ps): Ditto.
302 (_mm512_mask_floor_round_pd): Ditto.
303 (_mm512_mask_ceil_round_ps): Ditto.
304 (_mm512_mask_ceil_round_pd): Ditto.
305 (_mm512_maskz_floor_round_ps): Ditto.
306 (_mm512_maskz_floor_round_pd): Ditto.
307 (_mm512_maskz_ceil_round_ps): Ditto.
308 (_mm512_maskz_ceil_round_pd): Ditto.
309 (_mm512_expand_pd): Ditto.
310 (_mm512_expand_ps): Ditto.
311 * config/i386/i386.c (ix86_builtins): Remove
312 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
314 * config/i386/predicates.md (const1256_operand): New.
315 (const_1_to_2_operand): Ditto.
316 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
317 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
318 (*avx512pf_gatherpf<mode>sf): Ditto.
319 (avx512pf_gatherpf<mode>df): Ditto.
320 (*avx512pf_gatherpf<mode>df_mask): Ditto.
321 (*avx512pf_gatherpf<mode>df): Ditto.
322 (avx512pf_scatterpf<mode>sf): Ditto.
323 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
324 (*avx512pf_scatterpf<mode>sf): Ditto.
325 (avx512pf_scatterpf<mode>df): Ditto.
326 (*avx512pf_scatterpf<mode>df_mask): Ditto.
327 (*avx512pf_scatterpf<mode>df): Ditto.
328 (avx512f_expand<mode>): Removed.
329 (<shift_insn><mode>3<mask_name>): Change predicate type.
331 2014-02-08 Jakub Jelinek <jakub@redhat.com>
333 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
334 not at the end of datarefs vector use ordered_remove to avoid
335 reordering datarefs vector.
338 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
339 mark local addressable non-static vars as GOVD_PRIVATE
340 instead of GOVD_LOCAL.
341 * omp-low.c (lower_omp_for): Move gimple_bind_vars
342 and BLOCK_VARS of gimple_bind_block to new_stmt rather
346 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
347 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
348 assume_aligned or alloc_align attributes.
349 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
350 arguments. Handle also assume_aligned and alloc_align attributes.
351 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
352 calls to functions with assume_aligned or alloc_align attributes.
353 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
355 2014-02-08 Terry Guo <terry.guo@arm.com>
357 * doc/invoke.texi: Document ARM -march=armv7e-m.
359 2014-02-08 Jakub Jelinek <jakub@redhat.com>
361 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
362 flag on __cilkrts_rethrow builtin.
365 * ipa-cp.c (determine_versionability): Fail at -O0
366 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
367 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
370 2014-02-04 Jakub Jelinek <jakub@redhat.com>
373 * tree-inline.c (copy_forbidden): Fail for
374 __attribute__((optimize (0))) functions.
376 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
378 * varpool.c: Include pointer-set.h.
379 (varpool_remove_unreferenced_decls): Variables in other partitions
380 will not be output; be however careful to not lose information
383 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
385 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
386 lookup in the vtable constructor.
388 2014-02-07 Jeff Law <law@redhat.com>
391 * config/m68k/m68k.md (ashldi_extsi): Turn into a
392 define_insn_and_split.
394 * ipa-inline.c (inline_small_functions): Fix typos.
396 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
398 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
399 (s390_can_use_return_insn): Declare.
400 * config/s390/s390.h (EPILOGUE_USES): Define.
401 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
403 (s390_chunkify_start): Handle return JUMP_LABELs.
404 (s390_early_mach): Emit a main_pool instruction on the entry edge.
405 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
406 (s390_can_use_return_insn): New functions.
407 (s390_fix_long_loop_prediction): Handle conditional returns.
408 (TARGET_SET_UP_BY_PROLOGUE): Define.
409 * config/s390/s390.md (ANY_RETURN): New code iterator.
410 (*creturn, *csimple_return, return, simple_return): New patterns.
412 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
414 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
415 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
416 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
417 REG_CFA_RESTORE list when deciding not to restore a register.
419 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
421 * config/s390/s390.c: Include tree-pass.h and context.h.
422 (s390_early_mach): New function, split out from...
423 (s390_emit_prologue): ...here.
424 (pass_data_s390_early_mach): New pass structure.
425 (pass_s390_early_mach): New class.
426 (s390_option_override): Create and register early_mach pass.
429 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
431 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
432 to match for the exit block.
434 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
436 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
437 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
438 Reject misaligned operands.
440 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
442 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
444 2014-02-07 Richard Biener <rguenther@suse.de>
447 * gimple-low.c (lower_builtin_posix_memalign): New function.
448 (lower_stmt): Call it to lower posix_memalign in a way
449 to make alignment info accessible.
451 2014-02-07 Jakub Jelinek <jakub@redhat.com>
454 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
455 __builtin_setjmp_receiver.
457 2014-02-07 Richard Biener <rguenther@suse.de>
460 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
461 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
462 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
463 Handle BUILT_IN_POSIX_MEMALIGN.
464 (find_func_clobbers): Likewise.
465 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
466 (call_may_clobber_ref_p_1): Likewise.
468 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
471 * ipa-devirt.c (record_target_from_binfo): Remove overactive
474 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
477 * lto-cgraph.c (lto_output_node): Use
478 symtab_get_symbol_partitioning_class.
479 (lto_output_varpool_node): likewise.
480 (symtab_get_symbol_partitioning_class): Move here from
482 * cgraph.h (symbol_partitioning_class): Likewise.
483 (symtab_get_symbol_partitioning_class): Declare.
485 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
487 * ggc.h (ggc_internal_cleared_alloc): New macro.
488 * vec.h (vec_safe_copy): Handle memory stats.
489 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
490 * target-globals.c (save_target_globals): Likewise.
492 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
495 * expr.c (emit_move_resolve_push): Export; be bit more selective
496 on when to clear alias set.
497 * expr.h (emit_move_resolve_push): Declare.
498 * function.h (struct function): Add tail_call_marked.
499 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
500 * config/i386/i386-protos.h (ix86_expand_push): Remove.
501 * config/i386/i386.md (TImode move expander): De not call
503 (FP push expanders): Preserve memory attributes.
504 * config/i386/sse.md (push<mode>1): Remove.
505 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
506 (ix86_expand_push): Remove.
507 * config/i386/mmx.md (push<mode>1): Remove.
509 2014-02-06 Jakub Jelinek <jakub@redhat.com>
511 PR rtl-optimization/60030
512 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
513 lopart with paradoxical subreg before shifting it up by hprec.
515 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
517 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
518 Remove extra newline at end of file.
519 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
520 (arm_issue_rate): Handle cortexa57.
521 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
522 (cortex-a57.cortex-a53): Likewise.
524 2014-02-06 Jakub Jelinek <jakub@redhat.com>
527 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
528 don't record in REG_FRAME_RELATED_EXPR registers not set in that
530 (arm_expand_prologue): Adjust all callers.
531 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
532 info, registers also at the lowest numbered registers side. Use
533 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
537 * var-tracking.c (adjust_mems): Before adding a SET to
538 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
540 2014-02-06 Alan Modra <amodra@gmail.com>
543 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
544 change SDmode to DDmode when lra_in_progress.
546 2014-02-06 Jakub Jelinek <jakub@redhat.com>
549 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
550 free_data_ref on the dr first, and before goto again also set dr
551 to the next dr. For simd_lane_access, free old datarefs[i] before
552 overwriting it. For get_vectype_for_scalar_type failure, don't
553 free_data_ref if simd_lane_access.
555 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
558 * tree.h (opts_for_fn): New inline function.
559 (opt_for_fn): Define.
560 * config/i386/i386.c (ix86_function_regparm): Use
561 opt_for_fn (decl, optimize) instead of optimize.
563 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
565 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
566 for SYMBOL_REF in large memory model.
568 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
570 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
572 (cortex-a57): Likewise.
573 (cortex-a57.cortex-a53): Likewise.
575 2014-02-06 Yury Gribov <y.gribov@samsung.com>
576 Kugan Vivekanandarajah <kuganv@linaro.org>
578 * config/arm/arm.c (arm_vector_alignment_reachable): Check
580 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
582 2014-02-06 Richard Biener <rguenther@suse.de>
584 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
585 set_loop_copy and initialize_original_copy_tables.
587 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
589 * config/aarch64/aarch64-simd.md
590 (aarch64_ashr_simddi): Change QI to SI.
592 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
593 Jakub Jelinek <jakub@redhat.com>
596 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
599 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
601 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
602 CODE_FOR_altivec_vpku[hw]um to
603 CODE_FOR_altivec_vpku[hw]um_direct.
604 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
605 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
606 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
607 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
609 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
611 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
612 generation for -maltivec=be.
613 (altivec_vsumsws): Simplify redundant test.
615 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
617 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
618 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
619 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
620 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
622 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
624 (altivec_vpks<VI_char>ss): Likewise.
625 (altivec_vpks<VI_char>us): Likewise.
626 (altivec_vpku<VI_char>us): Likewise.
627 (altivec_vpku<VI_char>um): Likewise.
628 (altivec_vpku<VI_char>um_direct): New (copy of
629 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
631 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
632 target is little endian and -maltivec=be is not specified.
633 (*altivec_vupkhs<VU_char>_direct): New (copy of
634 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
635 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
636 target is little endian and -maltivec=be is not specified.
637 (*altivec_vupkls<VU_char>_direct): New (copy of
638 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
639 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
640 little endian and -maltivec=be is not specified.
641 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
642 little endian and -maltivec=be is not specified.
644 2014-02-05 Richard Henderson <rth@redhat.com>
647 * combine-stack-adj.c: Revert r206943.
648 * sched-int.h (struct deps_desc): Add last_args_size.
649 * sched-deps.c (init_deps): Initialize it.
650 (sched_analyze_insn): Add OUTPUT dependencies between insns that
651 contain REG_ARGS_SIZE notes.
653 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
655 * lto-cgraph.c (asm_nodes_output): Make global.
656 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
657 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
658 (driver_handle_option): Handle OPT_fwpa.
660 2014-02-05 Jakub Jelinek <jakub@redhat.com>
663 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
664 a comment typo and formatting issue. If odr_hash hasn't been
665 created, return vNULL and set *completep to false.
668 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
669 bb with no successors.
671 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
674 * doc/invoke.texi (-march): Clarify documentation for ARM.
678 2014-02-05 Richard Biener <rguenther@suse.de>
680 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
681 when not vectorizing because of too many alias checks.
682 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
683 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
685 2014-02-05 Nick Clifton <nickc@redhat.com>
687 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
688 accept extended registers in any mode when compiling for the MN10300.
690 2014-02-05 Yury Gribov <y.gribov@samsung.com>
692 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
693 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
694 sanitization attributes.
695 (can_inline_edge_p): Likewise.
696 (sanitize_attrs_match_for_inline_p): New function.
698 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
700 * ipa-prop.c (detect_type_change): Shor circuit testing of
701 type changes on THIS pointer.
703 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
706 * config/pa/pa.c (legitimize_tls_address): Return original address
707 if not passed a SYMBOL_REF rtx.
708 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
710 (pa_emit_move_sequence): Simplify TLS source operands.
711 (pa_legitimate_constant_p): Reject all TLS constants.
712 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
713 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
715 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
717 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
718 groups when we know they are controlled by LTO.
719 * varasm.c (default_binds_local_p_1): If object is in other partition,
720 it will be resolved locally.
722 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
724 * config/host-linux.c (linux_gt_pch_use_address): Don't
725 use SSIZE_MAX because it is not always defined.
727 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
730 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
731 threshold for pseudo splitting.
732 (update_ebb_live_info): Process call argument hard registers and
733 hard registers from insn definition too.
734 (max_small_class_regs_num): New constant.
735 (inherit_in_ebb): Update live hard regs through EBBs. Update
736 reloads_num only for small register classes. Don't split for
739 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
742 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
745 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
747 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
750 2014-02-04 Marek Polacek <polacek@redhat.com>
752 * gdbinit.in (pel): Define.
754 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
756 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
759 2014-02-04 Richard Biener <rguenther@suse.de>
762 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
763 in function context local.
764 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
765 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
766 similar to LTO_imported_decl_ref.
768 2014-02-04 Jakub Jelinek <jakub@redhat.com>
770 PR tree-optimization/60002
771 * cgraphclones.c (build_function_decl_skip_args): Clear
774 PR tree-optimization/60023
775 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
776 false to gsi_replace.
777 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
778 has been in some EH region and vec_stmt could throw, add
779 vec_stmt into the same EH region.
780 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
781 has no lhs, ignore it.
782 * internal-fn.c (expand_MASK_LOAD): Likewise.
785 * tree-inline.c (copy_forbidden): Fail for
786 __attribute__((optimize (0))) functions.
789 * omp-low.c (simd_clone_struct_copy): If from->inbranch
790 is set, copy one less argument.
791 (expand_simd_clones): Don't subtract clone_info->inbranch
792 from simd_clone_struct_alloc argument.
794 PR rtl-optimization/57915
795 * recog.c (simplify_while_replacing): If all unary/binary/relational
796 operation arguments are constant, attempt to simplify those.
799 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
800 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
802 2014-02-04 Richard Biener <rguenther@suse.de>
804 PR tree-optimization/60012
805 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
806 TBAA disambiguation to all DDRs.
808 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
811 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
812 (LINK_SPEC): Use it for -shared, -shared-libgcc.
814 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
817 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
819 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
821 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
822 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
824 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
827 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
828 to figure out targets of polymorphic calls with known decl.
829 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
830 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
831 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
832 (get_polymorphic_call_info): ... here.
833 (get_polymorphic_call_info_from_invariant): New function.
835 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
837 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
838 lookup via vtable pointer; check for type consistency
839 and turn inconsitent facts into UNREACHABLE.
840 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
841 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
842 type inconsistent querries; return UNREACHABLE instead.
844 2014-02-03 Richard Henderson <rth@twiddle.net>
847 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
848 already processed this node.
849 (normalize_one_pred_1): Pass along mark_set.
850 (normalize_one_pred): Create and destroy a pointer_set_t.
851 (normalize_one_pred_chain): Likewise.
853 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
855 PR gcov-profile/58602
856 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
858 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
861 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
862 -fno-devirtualize; try to devirtualize by the knowledge of
863 virtual table pointer given by aggregate propagation.
864 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
865 (ipa_print_node_jump_functions): Dump also offset that
866 is relevant for polymorphic calls.
867 (determine_known_aggregate_parts): Add arg_type parameter; use it
868 instead of determining the type from pointer type.
869 (ipa_compute_jump_functions_for_edge): Update call of
870 determine_known_aggregate_parts.
871 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
872 (gimple_get_virt_method_for_binfo): ... here; simplify using
873 vtable_pointer_value_to_vtable.
874 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
875 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
876 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
877 (vtable_pointer_value_to_vtable): Break out from ...; handle also
879 (vtable_pointer_value_to_binfo): ... here.
880 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
882 2014-02-03 Teresa Johnson <tejohnson@google.com>
884 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
885 redef of outer loop index variable.
887 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
891 * doc/extend.texi (Function Attributes): Typo.
893 2014-02-03 Cong Hou <congh@google.com>
895 PR tree-optimization/60000
896 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
897 if the vectorized statement is a store. A store statement can only
898 appear at the end of pattern statements.
900 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
902 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
903 (ix86_option_override_internal): Default long double to 64-bit for
904 32-bit Bionic and to 128-bit for 64-bit Bionic.
906 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
907 TARGET_LONG_DOUBLE_128 is true.
908 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
910 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
911 (mlong-double-64): Negate -mlong-double-128.
912 (mlong-double-128): New option.
914 * config/i386/i386-c.c (ix86_target_macros): Define
915 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
917 * doc/invoke.texi: Document -mlong-double-128.
919 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
921 PR rtl-optimization/60024
922 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
924 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
926 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
928 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
930 PR rtl-optimization/57662
931 * sel-sched.c (code_motion_path_driver): Do not mark already not
932 existing blocks in the visiting bitmap.
934 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
936 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
937 on the insn being emitted.
939 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
940 Will Deacon <will.deacon@arm.com>
942 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
944 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
946 * config/arm/arm-tables.opt: Regenerate.
948 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
950 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
951 for vector types other than V16QImode.
952 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
953 define_expand, and call altivec_expand_vec_perm_le when producing
954 code with little endian element order.
955 (*altivec_vperm_<mode>_internal): New insn having previous
956 behavior of altivec_vperm_<mode>.
957 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
958 altivec_expand_vec_perm_le when producing code with little endian
960 (*altivec_vperm_<mode>_uns_internal): New insn having previous
961 behavior of altivec_vperm_<mode>_uns.
963 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
965 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
966 (altivec_vsumsws): Add handling for -maltivec=be with a little
968 (altivec_vsumsws_direct): New.
969 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
972 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
974 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
975 vtable_pointer_value_to_binfo): New functions.
976 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
977 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
979 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
981 * config/nios2/nios2.md (load_got_register): Initialize GOT
982 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
983 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
985 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
987 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
988 preserverd by passthrough, do not propagate the type.
990 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
992 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
993 (mips_atomic_assign_expand_fenv): New function.
994 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
996 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
998 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
999 (__builtin_mips_set_fcsr): Likewise.
1000 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1001 MIPS_USI_FTYPE_VOID.
1002 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1003 (mips16_expand_set_fcsr): Likewise.
1004 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1005 (mips16_set_fcsr_stub): Likewise.
1006 (mips16_get_fcsr_one_only_stub): New class.
1007 (mips16_set_fcsr_one_only_stub): Likewise.
1008 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1009 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1010 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1011 (hard_float): New availability predicate.
1012 (mips_builtins): Add get_fcsr and set_fcsr.
1013 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1014 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1015 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1016 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1017 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1020 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1022 * config/mips/mips.c (mips_one_only_stub): New class.
1023 (mips_need_mips16_rdhwr_p): Replace with...
1024 (mips16_rdhwr_stub): ...this new variable.
1025 (mips16_stub_call_address): New function.
1026 (mips16_rdhwr_one_only_stub): New class.
1027 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1028 (mips_output_mips16_rdhwr): Delete.
1029 (mips_finish_stub): New function.
1030 (mips_code_end): Use it to handle rdhwr stubs.
1032 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1035 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1036 when calculating size of integer atomic types.
1038 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1040 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1042 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1044 PR tree-optimization/60003
1045 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1046 * profile.c (branch_prob): Use gimple_call_builtin_p
1047 to check for BUILT_IN_SETJMP_RECEIVER.
1048 * tree-inline.c (copy_bb): Call notice_special_calls.
1050 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1053 * lra-constraints.c (process_alt_operands): Update reload_sum only
1056 2014-01-31 Richard Henderson <rth@redhat.com>
1059 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1060 until after else_eh is processed.
1062 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1064 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1065 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1066 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1067 in smmintrin.h, remove them.
1068 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1069 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1070 * config/i386/i386.md (ROUND_SAE): Fix value.
1071 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1072 (const48_operand): New.
1073 * config/i386/subst.md (round), (round_expand): Use
1074 const_4_or_8_to_11_operand.
1075 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1077 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1079 * config/i386/constraints.md (Yk): Swap meaning with k.
1080 * config/i386/i386.md (movhi_internal): Change Yk to k.
1081 (movqi_internal): Ditto.
1082 (*k<logic><mode>): Ditto.
1085 (kandn<mode>): Ditto.
1086 (*<code>hi_1): Ditto.
1087 (*<code>qi_1): Ditto.
1088 (kxnor<mode>): Ditto.
1089 (kortestzhi): Ditto.
1090 (kortestchi): Ditto.
1092 (*one_cmplhi2_1): Ditto.
1093 (*one_cmplqi2_1): Ditto.
1094 * config/i386/sse.md (): Change k to Yk.
1095 (avx512f_load<mode>_mask): Ditto.
1096 (avx512f_blendm<mode>): Ditto.
1097 (avx512f_store<mode>_mask): Ditto.
1098 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1099 (avx512f_storedqu<mode>_mask): Ditto.
1100 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1102 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1103 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1104 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1105 (avx512f_maskcmp<mode>3): Ditto.
1106 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1107 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1108 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1109 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1110 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1111 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1112 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1113 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1114 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1115 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1116 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1117 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1118 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1119 (vec_extract_lo_<mode>_maskm): Ditto.
1120 (vec_extract_hi_<mode>_maskm): Ditto.
1121 (avx512f_vternlog<mode>_mask): Ditto.
1122 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1123 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1124 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1125 (avx512f_<code>v8div16qi2_mask): Ditto.
1126 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1127 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1128 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1129 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1130 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1131 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1132 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1133 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1134 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1135 (avx512cd_maskb_vec_dupv8di): Ditto.
1136 (avx512cd_maskw_vec_dupv16si): Ditto.
1137 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1138 (avx512f_vpermi2var<mode>3_mask): Ditto.
1139 (avx512f_vpermi2var<mode>3_mask): Ditto.
1140 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1141 (*avx512f_gathersi<mode>): Ditto.
1142 (*avx512f_gathersi<mode>_2): Ditto.
1143 (*avx512f_gatherdi<mode>): Ditto.
1144 (*avx512f_gatherdi<mode>_2): Ditto.
1145 (*avx512f_scattersi<mode>): Ditto.
1146 (*avx512f_scatterdi<mode>): Ditto.
1147 (avx512f_compress<mode>_mask): Ditto.
1148 (avx512f_compressstore<mode>_mask): Ditto.
1149 (avx512f_expand<mode>_mask): Ditto.
1150 * config/i386/subst.md (mask): Change k to Yk.
1151 (mask_scalar_merge): Ditto.
1154 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1156 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1158 2014-01-31 Richard Biener <rguenther@suse.de>
1161 * builtins.c (fold_builtin_memory_op): Make sure to not
1162 use a floating-point mode or a boolean or enumeral type for
1165 2014-01-30 DJ Delorie <dj@redhat.com>
1167 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1168 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1169 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1170 whenever main() has an epilogue.
1172 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1174 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1175 unused variable "field".
1176 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1177 (vsx_mergeh_<mode>): Likewise.
1178 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1179 (altivec_vmrghh): Likewise.
1180 (altivec_vmrghw): Likewise.
1181 (altivec_vmrglb): Likewise.
1182 (altivec_vmrglh): Likewise.
1183 (altivec_vmrglw): Likewise.
1184 (altivec_vspltb): Add missing uses.
1185 (altivec_vsplth): Likewise.
1186 (altivec_vspltw): Likewise.
1187 (altivec_vspltsf): Likewise.
1189 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1192 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1193 frame related instructions.
1195 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1197 PR rtl-optimization/59959
1198 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1199 any reload of register whose subreg is invalid.
1201 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1203 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1204 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1205 Add missing return type - void.
1207 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1209 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1210 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1211 remove element index adjustment for endian (now handled in vsx.md
1213 (altivec_expand_vec_perm_const): Use
1214 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1215 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1216 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1217 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1218 define_expand and a new define_insn *altivec_vspltb_internal;
1219 adjust for -maltivec=be on a little endian target.
1220 (altivec_vspltb_direct): New.
1221 (altivec_vsplth): Divide into a define_expand and a new
1222 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1223 little endian target.
1224 (altivec_vsplth_direct): New.
1225 (altivec_vspltw): Divide into a define_expand and a new
1226 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1227 little endian target.
1228 (altivec_vspltw_direct): New.
1229 (altivec_vspltsf): Divide into a define_expand and a new
1230 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1231 a little endian target.
1233 2014-01-30 Richard Biener <rguenther@suse.de>
1235 PR tree-optimization/59993
1236 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1237 can propagate form the earlier stmt and avoid the transform
1238 when the intermediate result is needed.
1240 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1242 * README.Portability: Fix typo.
1244 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1246 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1247 comparison_operator with ordered_comparison_operator.
1249 2014-01-30 Nick Clifton <nickc@redhat.com>
1251 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1252 Rename to mn10300_store_multiple_regs.
1253 * config/mn10300/mn10300.c: Likewise.
1254 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1255 store_multiple_regs.
1256 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1257 Call mn10300_store_multiple_regs.
1259 2014-01-30 Nick Clifton <nickc@redhat.com>
1260 DJ Delorie <dj@redhat.com>
1262 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1263 %fp 2 to keep registers after it properly word-aligned.
1264 (rl78_alloc_physical_registers_umul): Handle the case where both
1265 input operands are the same.
1267 2014-01-30 Richard Biener <rguenther@suse.de>
1269 PR tree-optimization/59903
1270 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1273 2014-01-30 Jason Merrill <jason@redhat.com>
1276 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1279 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1281 2014-01-30 Richard Biener <rguenther@suse.de>
1283 PR tree-optimization/59951
1284 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1286 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1289 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1290 SFmode to DFmode case.
1292 2014-01-29 DJ Delorie <dj@redhat.com>
1294 * config/msp430/msp430.opt (-minrt): New.
1295 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1297 (ENDFILE_SPEC): Likewise.
1299 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1301 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1302 (estimate_function_body_sizes): Use it.
1304 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1307 * dwarf2out.c (is_cxx_auto): New.
1308 (is_base_type): Use it.
1309 (gen_type_die_with_usage): Likewise.
1311 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1313 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1314 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1315 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1316 -maltivec=be with LE targets.
1317 (vsx_mergeh_<mode>): Likewise.
1318 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1319 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1320 (altivec_vmrghb): Replace with define_expand and new
1321 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1322 (altivec_vmrghb_direct): New define_insn.
1323 (altivec_vmrghh): Replace with define_expand and new
1324 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1325 (altivec_vmrghh_direct): New define_insn.
1326 (altivec_vmrghw): Replace with define_expand and new
1327 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1328 (altivec_vmrghw_direct): New define_insn.
1329 (*altivec_vmrghsf): Adjust for endianness.
1330 (altivec_vmrglb): Replace with define_expand and new
1331 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1332 (altivec_vmrglb_direct): New define_insn.
1333 (altivec_vmrglh): Replace with define_expand and new
1334 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1335 (altivec_vmrglh_direct): New define_insn.
1336 (altivec_vmrglw): Replace with define_expand and new
1337 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1338 (altivec_vmrglw_direct): New define_insn.
1339 (*altivec_vmrglsf): Adjust for endianness.
1340 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1341 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1342 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1343 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1344 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1345 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1346 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1347 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1349 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1351 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1352 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1355 2014-01-29 Richard Biener <rguenther@suse.de>
1357 PR tree-optimization/58742
1358 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1359 associate_pointerplus_align.
1360 (associate_pointerplus_diff): New function.
1361 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1362 and associate_pointerplus_diff.
1364 2014-01-29 Richard Biener <rguenther@suse.de>
1366 * lto-streamer.h (LTO_major_version): Bump to 3.
1367 (LTO_minor_version): Reset to 0.
1369 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1371 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1372 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1373 (arm_file_start): Generate correct asm header for armv7ve.
1374 * config/arm/bpabi.h: Add multilib support for armv7ve.
1375 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1376 and cortex-a15 to armv7ve.
1377 * config/arm/t-aprofile: Add multilib support for armv7ve.
1378 * doc/invoke.texi: Document -march=armv7ve.
1380 2014-01-29 Richard Biener <rguenther@suse.de>
1382 PR tree-optimization/58742
1383 * tree-ssa-forwprop.c (associate_plusminus): Return true
1384 if we changed sth, defer EH cleanup to ...
1385 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1386 (simplify_mult): New function.
1388 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1391 PR tree-optimization/59920
1392 * tree.c (build_common_builtin_nodes): Remove
1393 __builtin_setjmp_dispatcher initialization.
1394 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1395 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1396 instead of gsi_after_labels + manually skipping debug stmts.
1397 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1398 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1399 * tree-inline.c (copy_edges_for_bb): Remove
1400 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1401 argument. Ignore computed_goto_p stmts. Don't call
1402 make_abnormal_goto_edges. If a call might need abnormal edges
1403 for non-local gotos, see if it already has an edge to
1404 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1405 with true argument, don't do anything then, otherwise add
1406 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1407 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1409 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1410 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1411 (lower_stmt): Don't set data->calls_builtin_setjmp.
1412 (lower_builtin_setjmp): Adjust comment.
1413 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1414 * tree-cfg.c (found_computed_goto): Remove.
1415 (factor_computed_gotos): Remove.
1416 (make_goto_expr_edges): Return bool, true for computed gotos.
1417 Don't call make_abnormal_goto_edges.
1418 (build_gimple_cfg): Don't set found_computed_goto, don't call
1419 factor_computed_gotos.
1420 (computed_goto_p): No longer static.
1421 (make_blocks): Don't set found_computed_goto.
1422 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1423 (make_edges): If make_goto_expr_edges returns true, push bb
1424 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1425 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1426 vector. Record mapping between bbs and OpenMP regions if there
1427 are any, adjust make_gimple_omp_edges caller. Call
1428 handle_abnormal_edges.
1429 (make_abnormal_goto_edges): Remove.
1430 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1431 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1432 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1433 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1434 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1435 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1436 filling *region also set *region_idx to (*region)->entry->index.
1439 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1440 For REGs set ORIGINAL_REGNO.
1442 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1444 * doc/md.texi: Mention that a target shouldn't implement
1445 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1448 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1450 PR tree-optimization/59594
1451 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1452 a copy of the datarefs vector rather than the vector itself.
1454 2014-01-28 Jason Merrill <jason@redhat.com>
1457 * dwarf2out.c (auto_die): New static.
1458 (gen_type_die_with_usage): Handle C++1y 'auto'.
1459 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1462 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1465 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1466 (SPEC_X32): Likewise.
1467 (SPEC_64): Likewise.
1468 * config/i386/i386.c (ix86_option_override_internal): Turn off
1469 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1471 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1472 * config/i386/i386.h (TARGET_16BIT): New macro.
1473 (TARGET_16BIT_P): Likewise.
1474 * config/i386/i386.opt: Add m16.
1475 * doc/invoke.texi: Document -m16.
1477 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1479 PR preprocessor/59935
1480 * input.c (location_get_source_line): Bail out on when line number
1481 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1483 2014-01-28 Richard Biener <rguenther@suse.de>
1485 PR tree-optimization/58742
1486 * tree-ssa-forwprop.c (associate_plusminus): Handle
1487 pointer subtraction of the form (T)(P + A) - (T)P.
1489 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1491 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1494 2014-01-28 Richard Biener <rguenther@suse.de>
1497 2014-01-28 Richard Biener <rguenther@suse.de>
1499 PR rtl-optimization/45364
1500 PR rtl-optimization/59890
1501 * var-tracking.c (local_get_addr_clear_given_value): Handle
1502 already cleared slot.
1503 (val_reset): Handle not allocated local_get_addr_cache.
1504 (vt_find_locations): Use post-order on the inverted CFG.
1506 2014-01-28 Richard Biener <rguenther@suse.de>
1508 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1510 2014-01-28 Richard Biener <rguenther@suse.de>
1512 PR rtl-optimization/45364
1513 PR rtl-optimization/59890
1514 * var-tracking.c (local_get_addr_clear_given_value): Handle
1515 already cleared slot.
1516 (val_reset): Handle not allocated local_get_addr_cache.
1517 (vt_find_locations): Use post-order on the inverted CFG.
1519 2014-01-28 Alan Modra <amodra@gmail.com>
1521 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1522 * configure.ac <recursive call for build != host>: Define
1523 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1524 and LD_FOR_BUILD too.
1525 * configure: Regenerate.
1527 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1529 * config/i386/i386.c (get_builtin_code_for_version): Separate
1530 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1531 Broadwell from Haswell.
1533 2014-01-27 Steve Ellcey <sellcey@mips.com>
1535 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1536 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1537 * config/mips/mips.c (mips_option_override): Change setting
1539 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1540 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1541 Change from Mask to Var.
1543 2014-01-27 Jeff Law <law@redhat.com>
1545 * ipa-inline.c (inline_small_functions): Fix typo.
1547 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1549 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1550 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1551 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1552 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1553 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1554 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1555 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1556 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1557 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1558 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1559 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1560 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1561 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1562 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1563 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1564 (_mm512_storeu_epi64): Ditto.
1565 (_mm512_cmpge_epi32_mask): Ditto.
1566 (_mm512_cmpge_epu32_mask): Ditto.
1567 (_mm512_cmpge_epi64_mask): Ditto.
1568 (_mm512_cmpge_epu64_mask): Ditto.
1569 (_mm512_cmple_epi32_mask): Ditto.
1570 (_mm512_cmple_epu32_mask): Ditto.
1571 (_mm512_cmple_epi64_mask): Ditto.
1572 (_mm512_cmple_epu64_mask): Ditto.
1573 (_mm512_cmplt_epi32_mask): Ditto.
1574 (_mm512_cmplt_epu32_mask): Ditto.
1575 (_mm512_cmplt_epi64_mask): Ditto.
1576 (_mm512_cmplt_epu64_mask): Ditto.
1577 (_mm512_cmpneq_epi32_mask): Ditto.
1578 (_mm512_cmpneq_epu32_mask): Ditto.
1579 (_mm512_cmpneq_epi64_mask): Ditto.
1580 (_mm512_cmpneq_epu64_mask): Ditto.
1581 (_mm512_expand_pd): Ditto.
1582 (_mm512_expand_ps): Ditto.
1583 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1584 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1585 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1586 * config/i386/i386.c (ix86_builtins): Add
1587 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1588 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1589 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1590 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1591 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1592 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1593 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1594 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1595 IX86_BUILTIN_PMOVUSQW512_MEM.
1596 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1597 __builtin_ia32_pmovsqd512mem_mask,
1598 __builtin_ia32_pmovqd512mem_mask,
1599 __builtin_ia32_pmovusqw512mem_mask,
1600 __builtin_ia32_pmovsqw512mem_mask,
1601 __builtin_ia32_pmovqw512mem_mask,
1602 __builtin_ia32_pmovusdw512mem_mask,
1603 __builtin_ia32_pmovsdw512mem_mask,
1604 __builtin_ia32_pmovdw512mem_mask,
1605 __builtin_ia32_pmovqb512mem_mask,
1606 __builtin_ia32_pmovusqb512mem_mask,
1607 __builtin_ia32_pmovsqb512mem_mask,
1608 __builtin_ia32_pmovusdb512mem_mask,
1609 __builtin_ia32_pmovsdb512mem_mask,
1610 __builtin_ia32_pmovdb512mem_mask.
1611 (bdesc_args): Add __builtin_ia32_expanddf512,
1612 __builtin_ia32_expandsf512.
1613 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1614 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1615 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1616 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1617 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1618 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1619 (avx512f_<code>v8div16qi2_mask_store): This.
1620 (avx512f_expand<mode>): New.
1622 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1624 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1626 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1627 (_mm512_prefetch_i32scatter_pd): Ditto.
1628 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1629 (_mm512_prefetch_i64scatter_pd): Ditto.
1630 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1631 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1632 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1633 (_mm512_prefetch_i32scatter_ps): Ditto.
1634 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1635 (_mm512_prefetch_i64scatter_ps): Ditto.
1636 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1637 * config/i386/i386-builtin-types.def: Define
1638 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1639 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1640 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1641 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1642 IX86_BUILTIN_SCATTERPFQPD.
1643 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1644 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1645 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1646 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1647 __builtin_ia32_scatterpfqps.
1648 (ix86_expand_builtin): Expand new built-ins.
1649 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1650 fix memory access data type.
1651 (*avx512pf_gatherpf<mode>_mask): Ditto.
1652 (*avx512pf_gatherpf<mode>): Ditto.
1653 (avx512pf_scatterpf<mode>): Ditto.
1654 (*avx512pf_scatterpf<mode>_mask): Ditto.
1655 (*avx512pf_scatterpf<mode>): Ditto.
1656 (GATHER_SCATTER_SF_MEM_MODE): New.
1657 (avx512pf_gatherpf<mode>df): Ditto.
1658 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1659 (*avx512pf_scatterpf<mode>df): Ditto.
1661 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1664 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1665 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1668 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1670 * common/config/arm/arm-common.c
1671 (arm_rewrite_mcpu): Handle multiple names.
1673 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1675 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1677 * gimple-builder.h (create_gimple_tmp): Delete.
1679 2014-01-27 Christian Bruel <christian.bruel@st.com>
1681 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1684 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1686 * config/pa/pa.md (call): Generate indirect long calls to non-local
1687 functions when outputing 32-bit code.
1688 (call_value): Likewise except for special call to buggy powf function.
1690 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1691 portable runtime and PIC indirect calls.
1692 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1693 and PIC call sequences. Use ldo instead of blr to set return register
1694 in PIC call sequence.
1696 2014-01-25 Walter Lee <walt@tilera.com>
1698 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1699 avoid clobbering a live register.
1701 2014-01-25 Walter Lee <walt@tilera.com>
1703 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1704 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1705 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1706 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1708 2014-01-25 Walter Lee <walt@tilera.com>
1710 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1711 arguments on even registers.
1712 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1714 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1715 (BIGGEST_ALIGNMENT): Ditto.
1716 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1718 2014-01-25 Walter Lee <walt@tilera.com>
1720 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1721 insns before bundling.
1722 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1724 2014-01-25 Walter Lee <walt@tilera.com>
1726 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1727 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1728 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1730 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1732 * config/mips/constraints.md (kl): Delete.
1733 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1734 define expands, using...
1735 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1736 instructions for MIPS16.
1737 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1738 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1740 2014-01-25 Walter Lee <walt@tilera.com>
1742 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1746 2014-01-25 Walter Lee <walt@tilera.com>
1748 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1749 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1751 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1753 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1756 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1758 * print-rtl.c (in_call_function_usage): New var.
1759 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1760 EXPR_LIST mode as mode and not as reg note name.
1763 * cfgloopmanip.c (copy_loop_info): If
1764 loop->warned_aggressive_loop_optimizations, make sure
1765 the flag is set in target loop too.
1767 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1769 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1771 * builtins.def: Likewise.
1772 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1773 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1774 * ira.c (ira_setup_eliminable_regset): Likewise.
1775 * omp-low.c (gate_expand_omp): Likewise.
1776 (execute_lower_omp): Likewise.
1777 (diagnose_sb_0): Likewise.
1778 (gate_diagnose_omp_blocks): Likewise.
1779 (simd_clone_clauses_extract): Likewise.
1782 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1784 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1785 correction for little endian...
1786 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1789 2014-01-24 Jeff Law <law@redhat.com>
1791 PR tree-optimization/59919
1792 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1793 for non-returning calls.
1795 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1797 * common/config/aarch64/aarch64-common.c
1798 (aarch64_rewrite_mcpu): Handle multiple names.
1799 * config/aarch64/aarch64.h
1800 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1802 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1804 * input.c (add_file_to_cache_tab): Handle the case where fopen
1807 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1810 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1811 from push operand if code of push isn't PRE_DEC.
1813 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1816 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1817 -mquad-memory-atomic. Update -mquad-memory documentation to say
1818 it is only used for non-atomic loads/stores.
1820 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1821 -mquad-memory or -mquad-memory-atomic switches.
1823 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1824 -mquad-memory-atomic to ISA 2.07 support.
1826 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1827 to separate support of normal quad word memory operations (ldq, stq)
1828 from the atomic quad word memory operations.
1830 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1831 support to separate non-atomic quad word operations from atomic
1832 quad word operations. Disable non-atomic quad word operations in
1833 little endian mode so that we don't have to swap words after the
1834 load and before the store.
1835 (quad_load_store_p): Add comment about atomic quad word support.
1836 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1837 options printed with -mdebug=reg.
1839 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1840 -mquad-memory-atomic as the test for whether we have quad word
1841 atomic instructions.
1842 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1843 or -mp8-vector are used, allow byte/half-word atomic operations.
1845 * config/rs6000/sync.md (load_lockedti): Insure that the address
1846 is a proper indexed or indirect address for the lqarx instruction.
1847 On little endian systems, swap the hi/lo registers after the lqarx
1849 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1850 insure the address is valid for the lqarx instruction.
1851 (store_conditionalti): Insure that the address is a proper indexed
1852 or indirect address for the stqcrx. instruction. On little endian
1853 systems, swap the hi/lo registers before doing the stqcrx.
1855 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1856 insure the address is valid for the stqcrx. instruction.
1858 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1859 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1860 type of quad memory support is available.
1862 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1865 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1866 there is a danger of looping.
1868 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1870 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1871 force flag_ira_loop_pressure if set via command line.
1873 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1875 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1876 (ashr_simd): New builtin handling DI mode.
1877 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1878 (aarch64_sshr_simddi): New match pattern.
1879 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1880 (vshrd_n_s64): Likewise.
1881 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1883 2014-01-23 Nick Clifton <nickc@redhat.com>
1885 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1886 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1887 favour of mcu specific scripts.
1888 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1891 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1892 Alex Velenko <Alex.Velenko@arm.com>
1894 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1895 (vaddv_s16): Likewise.
1896 (vaddv_s32): Likewise.
1897 (vaddv_u8): Likewise.
1898 (vaddv_u16): Likewise.
1899 (vaddv_u32): Likewise.
1900 (vaddvq_s8): Likewise.
1901 (vaddvq_s16): Likewise.
1902 (vaddvq_s32): Likewise.
1903 (vaddvq_s64): Likewise.
1904 (vaddvq_u8): Likewise.
1905 (vaddvq_u16): Likewise.
1906 (vaddvq_u32): Likewise.
1907 (vaddvq_u64): Likewise.
1908 (vaddv_f32): Likewise.
1909 (vaddvq_f32): Likewise.
1910 (vaddvq_f64): Likewise.
1911 (vmaxv_f32): Likewise.
1912 (vmaxv_s8): Likewise.
1913 (vmaxv_s16): Likewise.
1914 (vmaxv_s32): Likewise.
1915 (vmaxv_u8): Likewise.
1916 (vmaxv_u16): Likewise.
1917 (vmaxv_u32): Likewise.
1918 (vmaxvq_f32): Likewise.
1919 (vmaxvq_f64): Likewise.
1920 (vmaxvq_s8): Likewise.
1921 (vmaxvq_s16): Likewise.
1922 (vmaxvq_s32): Likewise.
1923 (vmaxvq_u8): Likewise.
1924 (vmaxvq_u16): Likewise.
1925 (vmaxvq_u32): Likewise.
1926 (vmaxnmv_f32): Likewise.
1927 (vmaxnmvq_f32): Likewise.
1928 (vmaxnmvq_f64): Likewise.
1929 (vminv_f32): Likewise.
1930 (vminv_s8): Likewise.
1931 (vminv_s16): Likewise.
1932 (vminv_s32): Likewise.
1933 (vminv_u8): Likewise.
1934 (vminv_u16): Likewise.
1935 (vminv_u32): Likewise.
1936 (vminvq_f32): Likewise.
1937 (vminvq_f64): Likewise.
1938 (vminvq_s8): Likewise.
1939 (vminvq_s16): Likewise.
1940 (vminvq_s32): Likewise.
1941 (vminvq_u8): Likewise.
1942 (vminvq_u16): Likewise.
1943 (vminvq_u32): Likewise.
1944 (vminnmv_f32): Likewise.
1945 (vminnmvq_f32): Likewise.
1946 (vminnmvq_f64): Likewise.
1948 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1950 * config/aarch64/aarch64-simd.md
1951 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1952 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1953 (*aarch64_mul3_elt<mode>): Likewise.
1954 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1955 (*aarch64_mul3_elt_to_64v2df): Likewise.
1956 (*aarch64_mla_elt<mode>): Likewise.
1957 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1958 (*aarch64_mls_elt<mode>): Likewise.
1959 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1960 (*aarch64_fma4_elt<mode>): Likewise.
1961 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1962 (*aarch64_fma4_elt_to_64v2df): Likewise.
1963 (*aarch64_fnma4_elt<mode>): Likewise.
1964 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1965 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1966 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1967 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1968 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1969 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1970 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1971 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1972 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1974 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1976 * config/aarch64/aarch64-simd.md
1977 (aarch64_be_checked_get_lane<mode>): New define_expand.
1978 * config/aarch64/aarch64-simd-builtins.def
1979 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1980 New builtin definition.
1981 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1982 Use new safe be builtin.
1984 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1986 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1988 (aarch64_be_st1<mode>): Likewise.
1989 (aarch_ld1<VALL:mode>): Define_expand modified.
1990 (aarch_st1<VALL:mode>): Likewise.
1991 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1992 (UNSPEC_ST1): Likewise.
1994 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1996 * config/microblaze/microblaze.md: Add trap insn and attribute
1998 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2000 PR preprocessor/58580
2001 * input.h (location_get_source_line): Take an additional line_size
2003 (void diagnostics_file_cache_fini): Declare new function.
2004 * input.c (struct fcache): New type.
2005 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2006 New static constants.
2007 (diagnostic_file_cache_init, total_lines_num)
2008 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2009 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2010 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2011 (get_next_line, read_next_line, goto_next_line, read_line_num):
2012 New static function definitions.
2013 (diagnostic_file_cache_fini): New function.
2014 (location_get_source_line): Take an additional output line_len
2015 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2017 * diagnostic.c (diagnostic_finish): Call
2018 diagnostic_file_cache_fini.
2019 (adjust_line): Take an additional input parameter for the length
2020 of the line, rather than calculating it with strlen.
2021 (diagnostic_show_locus): Adjust the use of
2022 location_get_source_line and adjust_line with respect to their new
2023 signature. While displaying a line now, do not stop at the first
2024 null byte. Rather, display the zero byte as a space and keep
2025 going until we reach the size of the line.
2026 * Makefile.in: Add vec.o to OBJS-libcommon
2028 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2029 Ilya Tocar <ilya.tocar@intel.com>
2031 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2032 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2033 (__builtin_ia32_kmov16): Ditto.
2034 * config/i386/i386.md (UNSPEC_KMOV): New.
2037 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2039 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2040 (_mm512_storeu_si512): Ditto.
2042 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2045 * rtl.h (get_referenced_operands): Declare.
2046 * recog.c (get_referenced_operands): New function.
2047 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2048 operands have been referenced when recording LO_SUM references.
2050 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2052 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2054 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2056 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2057 Enable for generic and recent AMD targets.
2059 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2061 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2062 ARG_SIZE note when adjustment was eliminated.
2064 2014-01-22 Jeff Law <law@redhat.com>
2066 PR tree-optimization/59597
2067 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2068 in file. Accept new argument REGISTERING and use it to modify
2069 dump output appropriately.
2070 (register_jump_thread): Corresponding changes.
2071 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2072 thread paths involving joiner blocks. Add code to dump cancelled
2073 jump threading paths.
2075 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2077 PR rtl-optimization/59477
2078 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2079 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2081 2014-01-22 Tom Tromey <tromey@redhat.com>
2083 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2085 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2087 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2089 PR rtl-optimization/59896
2090 * lra-constraints.c (process_alt_operands): Check unused note for
2091 matched operands of insn with no output reloads.
2093 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2095 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2096 (mips_move_from_gpr_cost): Likewise.
2098 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2100 PR rtl-optimization/59858
2101 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2102 ira_class_hard_regs_num.
2103 (process_alt_operands): Increase reject for dying matched operand.
2105 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2108 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2109 smaller than size, perform several stores or loads and stores
2110 at dst + count - size to store or copy all of size bytes, rather
2111 than just last modesize bytes.
2113 2014-01-20 DJ Delorie <dj@redhat.com>
2115 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2116 that CLOBBERs are REGs before propogating their values.
2118 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2121 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2122 (cgraph_inline_failed_type): New function.
2123 * cgraph.h (DEFCIFCODE): Add type.
2124 (cgraph_inline_failed_type_t): New enum.
2125 (cgraph_inline_failed_type): New prototype.
2126 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2127 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2128 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2129 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2130 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2131 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2132 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2133 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2134 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2135 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2136 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2137 OPTIMIZATION_MISMATCH.
2138 * tree-inline.c (expand_call_inline): Emit errors during
2139 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2141 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2144 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2145 mode attribute in insn output.
2147 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2149 * output.h (output_constant): Delete.
2150 * varasm.c (output_constant): Make private.
2152 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2154 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2156 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2159 * tree.h (fold_builtin_strcat): New prototype.
2160 * builtins.c (fold_builtin_strcat): No longer static. Add len
2161 argument, if non-NULL, don't call c_strlen. Optimize
2162 directly into __builtin_memcpy instead of __builtin_strcpy.
2163 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2164 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2166 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2168 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2169 for SImode_address_operand operands, having only a REG argument.
2171 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2173 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2174 loader name using mbig-endian.
2175 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2177 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2179 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2183 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2185 * config/aarch64/aarch64-protos.h
2186 (aarch64_cannot_change_mode_class_ptr): Declare.
2187 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2188 aarch64_cannot_change_mode_class_ptr): New.
2189 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2190 backend hook aarch64_cannot_change_mode_class.
2192 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2194 * common/config/aarch64/aarch64-common.c
2195 (aarch64_handle_option): Don't handle any option order logic here.
2196 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2197 selected_cpu, warn on architecture version mismatch.
2198 (aarch64_override_options): Fix parsing order for option strings.
2200 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2201 Iain Sandoe <iain@codesourcery.com>
2204 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2205 warning. Amend comment to reflect current functionality.
2207 2014-01-20 Richard Biener <rguenther@suse.de>
2210 * builtins.c (fold_builtin_strcat): Remove case better handled
2211 by tree-ssa-strlen.c.
2213 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2215 * config/aarch64/aarch64.opt
2216 (mcpu, march, mtune): Make case-insensitive.
2218 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2221 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2222 if operands[1] is a REG or ZERO_EXTEND of a REG.
2224 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2226 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2228 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2230 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2231 long non-pic millicode calls.
2233 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2235 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2237 2014-01-19 Kito Cheng <kito@0xlab.org>
2239 * builtins.c (expand_movstr): Check movstr expand done or fail.
2241 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2242 H.J. Lu <hongjiu.lu@intel.com>
2245 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2246 to DImode for zero-extended addresses.
2248 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2250 PR rtl-optimization/57763
2251 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2252 on the new indirect jump_insn and increment LABEL_NUSES (label).
2254 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2258 * config.gcc (x86_archs): New variable.
2259 (x86_64_archs): Likewise.
2260 (x86_cpus): Likewise.
2261 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2262 --with-arch/--with-cpu= options.
2263 Support --with-arch=/--with-cpu={nehalem,westmere,
2264 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2266 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2268 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2269 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2271 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2273 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2275 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2278 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2279 clear cpp_get_options (parse_in)->warn_unused_macros for
2280 ix86_target_macros_internal with cpp_define.
2282 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2284 * jump.c (delete_related_insns): Keep (use (insn))s.
2285 * reorg.c (redundant_insn): Check for barriers too.
2287 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2289 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2291 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2293 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2294 call to $$dyncall when TARGET_LONG_CALLS is true.
2296 2014-01-17 Jeff Law <law@redhat.com>
2298 * ree.c (combine_set_extension): Temporarily disable test for
2299 changing number of hard registers.
2301 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2304 * ipa-inline-analysis.c (inline_free_summary):
2305 Do not free summary of aliases.
2307 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2310 * gimplify.c (gimplify_expr): Use create_tmp_var
2311 instead of create_tmp_var_raw. If cond doesn't have
2312 integral type, don't add the IFN_ANNOTATE builtin at all.
2314 2014-01-17 Martin Jambor <mjambor@suse.cz>
2317 * ipa-cp.c (prev_edge_clone): New variable.
2318 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2319 Also resize prev_edge_clone vector.
2320 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2321 (ipcp_edge_removal_hook): New function.
2322 (ipcp_driver): Register ipcp_edge_removal_hook.
2324 2014-01-17 Andrew Pinski <apinski@cavium.com>
2325 Steve Ellcey <sellcey@mips.com>
2328 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2331 2014-01-17 Jeff Law <law@redhat.com>
2334 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2335 so that pass_ccp runs first.
2337 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2339 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2340 (ix86_adjust_cost): Use !TARGET_XXX.
2341 (do_reorder_for_imul): Likewise.
2342 (swap_top_of_ready_list): Likewise.
2343 (ix86_sched_reorder): Likewise.
2345 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2347 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2348 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2349 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2350 (intel_memset): New. Duplicate slm_memset.
2351 (intel_cost): New. Duplicate slm_cost.
2352 (m_INTEL): New macro.
2353 (processor_target_table): Add "intel".
2354 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2355 with PROCESSOR_INTEL for "intel".
2356 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2357 PROCESSOR_SILVERMONT.
2358 (ix86_issue_rate): Likewise.
2359 (ix86_adjust_cost): Likewise.
2360 (ia32_multipass_dfa_lookahead): Likewise.
2361 (swap_top_of_ready_list): Likewise.
2362 (ix86_sched_reorder): Likewise.
2363 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2364 instead of TARGET_OPT_AGU.
2365 * config/i386/i386.h (TARGET_INTEL): New.
2366 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2367 (processor_type): Add PROCESSOR_INTEL.
2368 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2369 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2371 2014-01-17 Marek Polacek <polacek@redhat.com>
2374 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2377 2014-01-17 Richard Biener <rguenther@suse.de>
2379 PR tree-optimization/46590
2380 * opts.c (default_options_table): Add entries for
2381 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2382 all enabled at -O1 but not for -Og.
2383 * common.opt (fbranch-count-reg): Remove Init(1).
2384 (fmove-loop-invariants): Likewise.
2385 (ftree-pta): Likewise.
2387 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2389 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2390 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2391 decls to at least the GCC 4.8 used alignments.
2394 * tree-nested.c (convert_nonlocal_reference_stmt,
2395 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2396 of GIMPLE_BIND stmts, adjust associated decls.
2398 2014-01-17 Richard Biener <rguenther@suse.de>
2400 PR tree-optimization/46590
2401 * vec.h (vec<>::bseach): New member function implementing
2402 binary search according to C89 bsearch.
2403 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2404 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2405 bitmap pointer again. Make accesses_in_loop a flat array.
2406 (mem_ref_obstack): New global.
2407 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2408 (mark_ref_stored): Likewise.
2409 (ref_indep_loop_p_2): Likewise.
2410 (set_ref_stored_in_loop): New helper function.
2411 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2412 (memref_free): Adjust.
2413 (record_mem_ref_loc): Simplify.
2414 (gather_mem_refs_stmt): Adjust.
2415 (sort_locs_in_loop_postorder_cmp): New function.
2416 (analyze_memory_references): Sort accesses_in_loop after
2417 loop postorder number.
2418 (find_ref_loc_in_loop_cmp): New function.
2419 (for_all_locs_in_loop): Find relevant cluster of locs in
2420 accesses_in_loop and iterate without recursion.
2421 (execute_sm): Avoid uninit warning.
2422 (struct ref_always_accessed): Simplify.
2423 (ref_always_accessed::operator ()): Likewise.
2424 (ref_always_accessed_p): Likewise.
2425 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2426 loop postorder numbers here.
2427 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2430 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2433 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2434 on decls for which assemble_alias has been called.
2436 2014-01-17 Nick Clifton <nickc@redhat.com>
2438 * config/msp430/msp430.opt: (mcpu): New option.
2439 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2440 (msp430_option_override): Parse target_cpu. If the MCU name
2441 matches a generic string, clear target_mcu.
2442 (msp430_attr): Allow numeric interrupt values up to 63.
2443 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2444 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2446 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2448 * config/msp430/msp430.md (popm): Use %J rather than %I.
2449 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2450 (addhi_cy_i): Use immediate_operand for operand 2.
2451 * doc/invoke.texi: Document -mcpu option.
2453 2014-01-17 Richard Biener <rguenther@suse.de>
2455 PR rtl-optimization/38518
2456 * df.h (df_analyze_loop): Declare.
2457 * df-core.c: Include cfgloop.h.
2458 (df_analyze_1): Split out main part of df_analyze.
2459 (df_analyze): Adjust.
2460 (loop_inverted_post_order_compute): New function.
2461 (loop_post_order_compute): Likewise.
2462 (df_analyze_loop): New function avoiding whole-function
2464 * loop-invariant.c (find_defs): Use df_analyze_loop.
2465 (find_invariants): Adjust.
2466 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2468 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2470 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2471 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2473 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2475 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2476 traversal when removing references.
2478 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2481 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2483 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2486 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2487 pushing a reload for an autoinc when we had previously reloaded an
2488 inner part of the address.
2490 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2492 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2494 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2495 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2496 when not giving up or versioning for alias only because of
2498 (vect_analyze_data_ref_dependences): Set to true.
2499 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2501 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2502 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2506 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2509 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2510 operand 0 predicate for gathers, use a new pseudo as subtarget.
2512 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2515 * lra-constraints.c (process_alt_operands): Add printing debug info.
2516 Check absence of input/output reloads for matched operands too.
2518 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2520 PR rtl-optimization/59835
2521 * ira.c (ira_init_register_move_cost): Increase cost for
2524 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2526 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2528 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2531 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2532 non-register objects. Use gen_(high/low)part more consistently.
2535 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2538 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2539 endian support, remove tests for WORDS_BIG_ENDIAN.
2540 (p8_mfvsrd_3_<mode>): Likewise.
2541 (reload_gpr_from_vsx<mode>): Likewise.
2542 (reload_gpr_from_vsxsf): Likewise.
2543 (p8_mfvsrd_4_disf): Likewise.
2545 2014-01-16 Richard Biener <rguenther@suse.de>
2547 PR rtl-optimization/46590
2548 * lcm.c (compute_antinout_edge): Use postorder iteration.
2549 (compute_laterin): Use inverted postorder iteration.
2551 2014-01-16 Nick Clifton <nickc@redhat.com>
2554 * varasm.c (output_constant): Return the number of bytes actually
2556 (output_constructor_array_range): Update the field size with the
2557 number of bytes emitted by output_constant.
2558 (output_constructor_regular_field): Likewise. Also do not
2559 complain if the total number of bytes emitted is now greater
2560 than the expected fieldpos.
2561 * output.h (output_constant): Update prototype and descriptive comment.
2563 2014-01-16 Marek Polacek <polacek@redhat.com>
2566 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2567 it is error_mark_node.
2569 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2571 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2572 VALID_AVX256_REG_OR_OI_MODE.
2574 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2576 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2577 current procedure should be profiled.
2579 2014-01-15 Andrew Pinski <apinski@cavium.com>
2581 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2582 of moving from/to the STACK_REG register class.
2584 2014-01-15 Richard Henderson <rth@redhat.com>
2587 * reginfo.c (global_regs_decl): Globalize.
2588 * rtl.h (global_regs_decl): Declare.
2589 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2590 reserved via global_regs.
2592 2014-01-15 Teresa Johnson <tejohnson@google.com>
2594 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2596 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2598 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2599 and vmulosh rather than call gen_vec_widen_smult_*.
2600 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2601 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2602 (vec_widen_smult_even_v16qi): Likewise.
2603 (vec_widen_umult_even_v8hi): Likewise.
2604 (vec_widen_smult_even_v8hi): Likewise.
2605 (vec_widen_umult_odd_v16qi): Likewise.
2606 (vec_widen_smult_odd_v16qi): Likewise.
2607 (vec_widen_umult_odd_v8hi): Likewise.
2608 (vec_widen_smult_odd_v8hi): Likewise.
2609 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2610 vmuloub rather than call gen_vec_widen_umult_*.
2611 (vec_widen_umult_lo_v16qi): Likewise.
2612 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2613 vmulosb rather than call gen_vec_widen_smult_*.
2614 (vec_widen_smult_lo_v16qi): Likewise.
2615 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2616 rather than call gen_vec_widen_umult_*.
2617 (vec_widen_umult_lo_v8hi): Likewise.
2618 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2619 rather than call gen_vec_widen_smult_*.
2620 (vec_widen_smult_lo_v8hi): Likewise.
2622 2014-01-15 Jeff Law <law@redhat.com>
2624 PR tree-optimization/59747
2625 * ree.c (find_and_remove_re): Properly handle case where a second
2626 eliminated extension requires widening a copy created for elimination
2627 of a prior extension.
2628 (combine_set_extension): Ensure that the number of hard regs needed
2629 for a destination register does not change when we widen it.
2631 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2633 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2634 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2635 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2636 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2637 (avr-*-rtems*): Likewise.
2638 (bfin*-rtems*): Likewise.
2639 (moxie-*-rtems*): Likewise.
2640 (h8300-*-rtems*): Likewise.
2641 (i[34567]86-*-rtems*): Likewise.
2642 (lm32-*-rtems*): Likewise.
2643 (m32r-*-rtems*): Likewise.
2644 (m68k-*-rtems*): Likewise.
2645 (microblaze*-*-rtems*): Likewise.
2646 (mips*-*-rtems*): Likewise.
2647 (powerpc-*-rtems*): Likewise.
2648 (sh-*-rtems*): Likewise.
2649 (sparc-*-rtems*): Likewise.
2650 (sparc64-*-rtems*): Likewise.
2651 (v850-*-rtems*): Likewise.
2652 (m32c-*-rtems*): Likewise.
2654 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2656 PR rtl-optimization/59511
2657 * ira.c (ira_init_register_move_cost): Use memory costs for some
2658 cases of register move cost calculations.
2659 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2660 instead of BB frequency.
2661 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2662 * lra-assigns.c (find_hard_regno_for): Ditto.
2664 2014-01-15 Richard Biener <rguenther@suse.de>
2666 PR tree-optimization/59822
2667 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2668 (vectorizable_load): Use it to hoist defs of uses of invariant
2669 loads out of the loop.
2671 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2672 Kugan Vivekanandarajah <kuganv@linaro.org>
2675 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2678 2014-01-15 Richard Biener <rguenther@suse.de>
2680 PR rtl-optimization/59802
2681 * lcm.c (compute_available): Use inverted postorder to seed
2682 the initial worklist.
2684 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2687 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2688 ADDR_REGS for invalid symrefs in non-PIC code.
2690 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2693 * builtins.c (determine_block_size): Initialize *probable_max_size
2694 even if len_rtx is CONST_INT.
2696 2014-01-14 Andrew Pinski <apinski@cavium.com>
2698 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2699 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2700 (cortexa53_tunings): Likewise.
2701 (aarch64_sched_issue_rate): New function.
2702 (TARGET_SCHED_ISSUE_RATE): Define.
2704 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2706 * ira-costs.c (find_costs_and_classes): Add missed
2707 ira_init_register_move_cost_if_necessary.
2709 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2712 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2714 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2717 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2718 to indicate if type is used for function return value. Warn ABI
2719 change if the vector mode isn't available for function return value.
2720 (ix86_function_arg_advance): Pass false to type_natural_mode.
2721 (ix86_function_arg): Likewise.
2722 (ix86_gimplify_va_arg): Likewise.
2723 (function_arg_32): Don't warn ABI change.
2724 (ix86_function_value): Pass true to type_natural_mode.
2725 (ix86_return_in_memory): Likewise.
2726 (ix86_struct_value_rtx): Removed.
2727 (TARGET_STRUCT_VALUE_RTX): Likewise.
2729 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2731 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2732 converting a conditional jump into a conditional return.
2734 2014-01-14 Richard Biener <rguenther@suse.de>
2736 PR tree-optimization/58921
2737 PR tree-optimization/59006
2738 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2739 hoisting invariant stmts.
2740 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2741 invariant loads on the preheader edge if possible.
2743 2014-01-14 Joey Ye <joey.ye@arm.com>
2745 * doc/plugin.texi (Building GCC plugins): Update to C++.
2747 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2749 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2750 (_mm_rcp28_round_ss): Ditto.
2751 (_mm_rsqrt28_round_sd): Ditto.
2752 (_mm_rsqrt28_round_ss): Ditto.
2753 (_mm_rcp28_sd): Ditto.
2754 (_mm_rcp28_ss): Ditto.
2755 (_mm_rsqrt28_sd): Ditto.
2756 (_mm_rsqrt28_ss): Ditto.
2757 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2758 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2759 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2760 (IX86_BUILTIN_RCP28SD): Ditto.
2761 (IX86_BUILTIN_RCP28SS): Ditto.
2762 (IX86_BUILTIN_RSQRT28SD): Ditto.
2763 (IX86_BUILTIN_RSQRT28SS): Ditto.
2764 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2765 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2766 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2767 (ix86_expand_special_args_builtin): Expand new FTYPE.
2768 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2769 (srcp14<mode>): Make insn unary.
2770 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2771 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2772 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2773 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2774 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2775 Fix rounding: make it SAE only.
2776 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2778 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2780 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2781 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2782 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2783 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2784 (round_saeonly_mask_scalar_operand4): Ditto.
2785 (round_saeonly_mask_scalar_op3): Ditto.
2786 (round_saeonly_mask_scalar_op4): Ditto.
2788 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2790 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2791 Implement -maltivec=be for vec_insert and vec_extract.
2793 2014-01-10 DJ Delorie <dj@redhat.com>
2795 * config/msp430/msp430.md (call_internal): Don't allow memory
2796 references with SP as the base register.
2797 (call_value_internal): Likewise.
2798 * config/msp430/constraints.md (Yc): New. For memory references
2799 that don't use SP as a base register.
2801 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2802 "an integer without a # prefix"
2803 * config/msp430/msp430.md (epilogue_helper): Use it.
2805 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2808 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2809 AVX512F gather builtins.
2810 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2811 on gather decls with INTEGER_TYPE masktype.
2812 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2813 directly into the builtin rather than hoisting it before loop.
2815 PR tree-optimization/59387
2816 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2817 (scev_const_prop): If folded_casts and type has undefined overflow,
2818 use force_gimple_operand instead of force_gimple_operand_gsi and
2819 for each added stmt if it is assign with
2820 arith_code_with_undefined_signed_overflow, call
2821 rewrite_to_defined_overflow.
2822 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2823 gimple-fold.h instead.
2824 (arith_code_with_undefined_signed_overflow,
2825 rewrite_to_defined_overflow): Moved to ...
2826 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2827 rewrite_to_defined_overflow): ... here. No longer static.
2828 Include gimplify-me.h.
2829 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2830 rewrite_to_defined_overflow): New prototypes.
2832 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2834 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2836 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2838 * builtins.c (get_object_alignment_2): Minor tweak.
2839 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2841 2014-01-13 Christian Bruel <christian.bruel@st.com>
2843 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2844 optimized non constant lengths.
2846 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2849 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2850 load as __atomic_load_N if possible.
2852 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2854 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2856 (rs6000_expand_builtin): Adjust call.
2858 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2861 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2862 * config/rs6000/rs6000.c: Include target-globals.h.
2863 (rs6000_set_current_function): Instead of doing target_reinit
2864 unconditionally, use save_target_globals_default_opts and
2865 restore_target_globals.
2867 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2869 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2870 (rs6000_expand_builtin): Handle mffs and mtfsf.
2871 (rs6000_init_builtins): Define mffs and mtfsf.
2872 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2873 (rs6000_mffs): New pattern.
2874 (rs6000_mtfsf): New pattern.
2876 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2878 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2879 Start narrowing with START. Apply candidate-use pair
2880 and check overall cost in narrowing.
2881 (iv_ca_prune): Pass new argument.
2883 2014-01-10 Jeff Law <law@redhat.com>
2886 * ree.c (combine_reaching_defs): Ensure the defining statement
2887 occurs before the extension when optimizing extensions with
2888 different source and destination hard registers.
2890 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2893 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2894 vtables into the type inheritance graph.
2896 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2898 PR rtl-optimization/59754
2899 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2900 modes in the REGNO != REGNO case.
2902 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2904 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2906 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2908 PR tree-optimization/59745
2909 * tree-predcom.c (tree_predictive_commoning_loop): Call
2910 free_affine_expand_cache if giving up because components is NULL.
2912 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2913 GC in payload of target_globals struct instead of allocating them on
2914 the heap and the larger structs separately using GC.
2915 * target-globals.h (struct target_globals): Make regs, hard_regs,
2916 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2917 of GTY((skip)) and change type to void *.
2918 (reset_target_globals): Cast loads from those fields to corresponding
2921 2014-01-10 Steve Ellcey <sellcey@mips.com>
2924 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2925 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2926 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2928 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2931 * aarch64-modes.def (CC_Zmode): New flags mode.
2932 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2933 represents an equality.
2934 (aarch64_get_condition_code): Handle CC_Zmode.
2935 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2937 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2939 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2940 extraction in good case.
2942 2014-01-10 Richard Biener <rguenther@suse.de>
2944 PR tree-optimization/59374
2945 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2946 checking after SLP discovery. Mark stmts not participating
2947 in any SLP instance properly.
2949 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2951 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2952 when handling a SET rtx.
2954 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2956 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2957 (cortex-a57): Likewise.
2958 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2960 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2962 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2963 non-iwmmxt builtins.
2965 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2969 * ipa-devirt.c record_target_from_binfo): Take as argument
2970 stack of binfos and lookup matching one for virtual inheritance.
2971 (possible_polymorphic_call_targets_1): Update.
2973 2014-01-10 Huacai Chen <chenhc@lemote.com>
2975 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2976 kernel strings for Loongson-2E/2F/3A.
2978 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2981 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2982 is_gimple_call before calling gimple_call_internal_p.
2984 2014-01-09 Steve Ellcey <sellcey@mips.com>
2986 * Makefile.in (TREE_FLOW_H): Remove.
2987 (TREE_SSA_H): Add file names from tree-flow.h.
2988 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2989 * tree.h: Remove tree-flow.h reference.
2990 * hash-table.h: Remove tree-flow.h reference.
2991 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2992 reference with tree-ssa-loop.h.
2994 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2996 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2997 default element-order behavior for -maltivec.
2998 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2999 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3000 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3001 when targeting big endian, at least for now.
3002 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3004 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3007 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3008 var satisfies use_register_for_decl, just take into account type
3009 alignment, rather than decl alignment.
3011 PR tree-optimization/59622
3012 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3013 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3014 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3015 Don't devirtualize for inplace at all. For targets.length () == 1,
3016 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3018 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3020 * config/i386/i386.md (cpu): Remove the unused btver1.
3022 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3024 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3026 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3029 * tree-core.h (struct target_globals): New forward declaration.
3030 (struct tree_target_option): Add globals field.
3031 * tree.h (TREE_TARGET_GLOBALS): Define.
3032 (prepare_target_option_nodes_for_pch): New prototype.
3033 * target-globals.h (struct target_globals): Define even if
3035 * tree.c (prepare_target_option_node_for_pch,
3036 prepare_target_option_nodes_for_pch): New functions.
3037 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3038 * config/i386/i386.c: Include target-globals.h.
3039 (ix86_set_current_function): Instead of doing target_reinit
3040 unconditionally, use save_target_globals_default_opts and
3041 restore_target_globals.
3043 2014-01-09 Richard Biener <rguenther@suse.de>
3045 PR tree-optimization/59715
3046 * tree-cfg.h (split_critical_edges): Declare.
3047 * tree-cfg.c (split_critical_edges): Export.
3048 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3050 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3052 * cfgexpand.c (expand_stack_vars): Optionally disable
3053 asan stack protection.
3054 (expand_used_vars): Likewise.
3055 (partition_stack_vars): Likewise.
3056 * asan.c (asan_emit_stack_protection): Optionally disable
3057 after return stack usage.
3058 (instrument_derefs): Optionally disable memory access instrumentation.
3059 (instrument_builtin_call): Likewise.
3060 (instrument_strlen_call): Likewise.
3061 (asan_protect_global): Optionally disable global variables protection.
3062 * doc/invoke.texi: Added doc for new options.
3063 * params.def: Added new options.
3064 * params.h: Likewise.
3066 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3068 PR rtl-optimization/59724
3069 * ifcvt.c (cond_exec_process_if_block): Don't call
3070 flow_find_head_matching_sequence with 0 longest_match.
3071 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3072 non-active insns if !stop_after.
3073 (try_head_merge_bb): Revert 2014-01-07 changes.
3075 2014-01-08 Jeff Law <law@redhat.com>
3077 * ree.c (get_sub_rtx): New function, extracted from...
3078 (merge_def_and_ext): Here.
3079 (combine_reaching_defs): Use get_sub_rtx.
3081 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3083 * cgraph.h (varpool_variable_node): Do not choke on null node.
3085 2014-01-08 Catherine Moore <clm@codesourcery.com>
3087 * config/mips/mips.md (simple_return): Attempt to use JRC
3089 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3091 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3093 PR rtl-optimization/59137
3094 * reorg.c (steal_delay_list_from_target): Call update_block for
3096 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3098 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3100 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3101 two duplicate entries.
3103 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3106 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3108 * config/mips/mips.c (mips_truncated_op_cost): New function.
3109 (mips_rtx_costs): Adjust test for BADDU.
3110 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3112 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3114 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3115 (*baddu_si): ...this new pattern.
3117 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3120 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3122 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3125 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3127 (expand_expr, expand_normal): Adjust.
3128 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3129 inner_reference_p. Use inner_reference_p to expand inner references.
3130 (store_expr): Adjust.
3131 * cfgexpand.c (expand_call_stmt): Adjust.
3133 2014-01-08 Rong Xu <xur@google.com>
3135 * gcov-io.c (gcov_var): Move from gcov-io.h.
3136 (gcov_position): Ditto.
3137 (gcov_is_error): Ditto.
3138 (gcov_rewrite): Ditto.
3139 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3140 only part to libgcc/libgcov.h.
3142 2014-01-08 Marek Polacek <polacek@redhat.com>
3145 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3147 2014-01-08 Marek Polacek <polacek@redhat.com>
3150 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3152 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3154 PR rtl-optimization/59649
3155 * stor-layout.c (get_mode_bounds): For BImode return
3156 0 and STORE_FLAG_VALUE.
3158 2014-01-08 Richard Biener <rguenther@suse.de>
3161 * gimple.h (is_gimple_builtin_call): Remove.
3162 (gimple_builtin_call_types_compatible_p): New.
3163 (gimple_call_builtin_p): New overload.
3164 * gimple.c (is_gimple_builtin_call): Remove.
3165 (validate_call): Rename to ...
3166 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3168 (validate_type): New static function.
3169 (gimple_call_builtin_p): New overload and adjust.
3170 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3171 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3172 (gimple_fold_stmt_to_constant_1): Likewise.
3173 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3175 2014-01-08 Richard Biener <rguenther@suse.de>
3178 * gimplify.c (gimplify_expr): Gimplify register-register type
3179 VIEW_CONVERT_EXPRs to separate stmts.
3181 2014-01-07 Jeff Law <law@redhat.com>
3184 * ree.c (combine_set_extension): Handle case where source
3185 and destination registers in an extension insn are different.
3186 (combine_reaching_defs): Allow source and destination registers
3187 in extension to be different under limited circumstances.
3188 (add_removable_extension): Remove restriction that the
3189 source and destination registers in the extension are the same.
3190 (find_and_remove_re): Emit a copy from the extension's
3191 destination to its source after the defining insn if
3192 the source and destination registers are different.
3195 * ifcvt.c (merge_if_block): If we are merging a block with more than
3196 one successor with a block with no successors, remove any BARRIER
3197 after the second block.
3199 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3201 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3203 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3206 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3207 for 14-bit register offsets when INT14_OK_STRICT is false.
3209 2014-01-07 Roland Stigge <stigge@antcom.de>
3210 Michael Meissner <meissner@linux.vnet.ibm.com>
3213 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3214 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3216 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3218 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3221 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3223 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3224 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3225 rtx is const0_rtx or not.
3227 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3230 * target-globals.c (save_target_globals): Remove this_fn_optab
3232 * toplev.c: Include optabs.h.
3233 (target_reinit): Temporarily restore the global options if another
3234 set of options are in force.
3236 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3238 PR rtl-optimization/58668
3239 * cfgcleanup.c (flow_find_cross_jump): Don't count
3240 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3241 to determine what is counted.
3242 (flow_find_head_matching_sequence): Use active_insn_p to determine
3244 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3246 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3247 determine what is counted.
3249 PR tree-optimization/59643
3250 * tree-predcom.c (split_data_refs_to_components): If one dr is
3251 read and one write, determine_offset fails and the write isn't
3252 in the bad component, just put the read into the bad component.
3254 2014-01-07 Mike Stump <mikestump@comcast.net>
3255 Jakub Jelinek <jakub@redhat.com>
3258 * tree-core.h (struct tree_optimization_option): Change optabs
3259 type from unsigned char * to void *.
3260 * optabs.c (init_tree_optimization_optabs): Adjust
3261 TREE_OPTIMIZATION_OPTABS initialization.
3263 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3266 * config/i386/i386.h (struct machine_function): Add
3267 no_drap_save_restore field.
3268 * config/i386/i386.c (ix86_save_reg): Use
3269 !cfun->machine->no_drap_save_restore instead of
3270 crtl->stack_realign_needed.
3271 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3272 this function clears frame_pointer_needed. Set
3273 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3274 and DRAP reg is needed.
3276 2014-01-06 Marek Polacek <polacek@redhat.com>
3279 * doc/implement-c.texi: Mention that other integer types are
3280 permitted as bit-field types in strictly conforming mode.
3282 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3284 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3287 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3289 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3291 2014-01-06 Martin Jambor <mjambor@suse.cz>
3294 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3296 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3298 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3302 * var-tracking.c (add_stores): Preserve the value of the source even if
3303 we don't record the store.
3305 2014-01-06 Terry Guo <terry.guo@arm.com>
3307 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3309 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3312 * config/darwin.c (darwin_function_section): Adjust return values to
3313 correspond to optimisation changes made in r206070.
3315 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3317 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3318 from prefetch_block tune setting.
3319 (nocona_cost): Correct size of prefetch block to 64.
3321 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3323 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3324 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3325 used to save the static chain register in the computation of the offset
3326 from which the FP registers need to be restored.
3328 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3330 PR tree-optimization/59519
3331 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3332 ICE if get_current_def (current_new_name) is already non-NULL, as long
3333 as it is a phi result of some other phi in *new_exit_bb that has
3336 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3337 or vmovdqu* for misaligned_operand.
3338 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3339 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3340 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3341 aligned_mem for AVX512F masked aligned load and store builtins and for
3344 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3346 PR tree-optimization/59651
3347 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3348 Address range for negative step should be added by TYPE_SIZE_UNIT.
3350 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3352 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3353 overlapping registers also for registers other than the stack pointer.
3355 2014-01-03 Marek Polacek <polacek@redhat.com>
3358 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3361 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3364 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3367 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3368 (push splitter): Use <P:MODE_SIZE> instead of
3369 GET_MODE_SIZE (<P:MODE>mode).
3370 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3371 (mov -1, reg peephole2): Likewise.
3372 * config/i386/sse.md (*mov<mode>_internal,
3373 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3374 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3375 *<code><mode>3, *andnot<mode>3<mask_name>,
3376 <mask_codefor><code><mode>3<mask_name>): Likewise.
3377 * config/i386/subst.md (mask_mode512bit_condition,
3378 sd_mask_mode512bit_condition): Likewise.
3380 2014-01-02 Xinliang David Li <davidxl@google.com>
3382 PR tree-optimization/59303
3383 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3384 (dump_predicates): Better output format.
3385 (pred_equal_p): New function.
3386 (is_neq_relop_p): Ditto.
3387 (is_neq_zero_form_p): Ditto.
3388 (pred_expr_equal_p): Ditto.
3389 (pred_neg_p): Ditto.
3390 (simplify_pred): Ditto.
3391 (simplify_preds_2): Ditto.
3392 (simplify_preds_3): Ditto.
3393 (simplify_preds_4): Ditto.
3394 (simplify_preds): Ditto.
3396 (push_to_worklist): Ditto.
3397 (get_pred_info_from_cmp): Ditto.
3398 (is_degenerated_phi): Ditto.
3399 (normalize_one_pred_1): Ditto.
3400 (normalize_one_pred): Ditto.
3401 (normalize_one_pred_chain): Ditto.
3402 (normalize_preds): Ditto.
3403 (normalize_cond_1): Remove function.
3404 (normalize_cond): Ditto.
3405 (is_gcond_subset_of): Ditto.
3406 (is_subset_of_any): Ditto.
3407 (is_or_set_subset_of): Ditto.
3408 (is_and_set_subset_of): Ditto.
3409 (is_norm_cond_subset_of): Ditto.
3410 (pred_chain_length_cmp): Ditto.
3411 (convert_control_dep_chain_into_preds): Type change.
3412 (find_predicates): Ditto.
3413 (find_def_preds): Ditto.
3414 (destroy_predicates_vecs): Ditto.
3415 (find_matching_predicates_in_rest_chains): Ditto.
3416 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3417 (is_pred_expr_subset): Ditto.
3418 (is_pred_chain_subset_of): Ditto.
3419 (is_included_in): Ditto.
3420 (is_superset_of): Ditto.
3422 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3424 Update copyright years.
3426 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3428 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3429 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3430 config/arc/arc.md, config/arc/arc.opt,
3431 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3432 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3433 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3434 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3435 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3436 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3437 vtable-verify.c, vtable-verify.h: Use the standard form for the
3440 2014-01-02 Tobias Burnus <burnus@net-b.de>
3442 * gcc.c (process_command): Update copyright notice dates.
3443 * gcov-dump.c: Ditto.
3445 * doc/cpp.texi: Bump @copying's copyright year.
3446 * doc/cppinternals.texi: Ditto.
3447 * doc/gcc.texi: Ditto.
3448 * doc/gccint.texi: Ditto.
3449 * doc/gcov.texi: Ditto.
3450 * doc/install.texi: Ditto.
3451 * doc/invoke.texi: Ditto.
3453 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3455 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3457 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3459 * config/i386/sse.md (*mov<mode>_internal): Guard
3460 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3462 PR rtl-optimization/59647
3463 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3464 new_rtx into UNSIGNED_FLOAT rtxes.
3466 Copyright (C) 2014 Free Software Foundation, Inc.
3468 Copying and distribution of this file, with or without modification,
3469 are permitted in any medium without royalty provided the copyright
3470 notice and this notice are preserved.