1 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
5 (maybe_fold_stmt): ... into this new function.
6 * omp-low.c (lower_omp): Update comment.
8 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
11 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
14 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
16 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
17 identifiers in comments.
18 (cortexa53_extra_costs): Likewise.
19 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in
21 (cortexa7_extra_costs): Likewise.
22 (cortexa12_extra_costs): Likewise.
23 (cortexa15_extra_costs): Likewise.
24 (v7m_extra_costs): Likewise.
26 2014-02-12 Richard Biener <rguenther@suse.de>
29 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
30 of posix_memalign being successful.
31 (lower_stmt): Restrict lowering of posix_memalign to when
32 -ftree-bit-ccp is enabled.
34 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
36 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
38 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
40 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
42 PR rtl-optimization/60116
43 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
44 other_insn once the combination has been validated.
46 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
49 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
51 * ipa-devirt.c: Include demangle.h
52 (odr_violation_reported): New static variable.
53 (add_type_duplicate): Update odr_violations.
54 (maybe_record_node): Add completep parameter; update it.
55 (record_target_from_binfo): Add COMPLETEP parameter;
57 (possible_polymorphic_call_targets_1): Likewise.
58 (struct polymorphic_call_target_d): Add nonconstruction_targets;
59 rename FINAL to COMPLETE.
60 (record_targets_from_bases): Sanity check we found the binfo;
61 fix COMPLETEP updating.
62 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
63 parameter, fix computing of COMPLETEP.
64 (dump_possible_polymorphic_call_targets): Imrove readability of dump; at
65 LTO time do demangling.
66 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
67 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
69 (gimple_get_virt_method_for_binfo): Likewise.
70 * gimple-fold.h (gimple_get_virt_method_for_binfo,
71 gimple_get_virt_method_for_vtable): Update prototypes.
73 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
76 * genautomata.c (add_presence_absence): Fix typo with
77 {final_}presence_list.
79 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
82 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
83 for VSX/Altivec vectors that land in GPR registers.
85 2014-02-11 Richard Henderson <rth@redhat.com>
86 Jakub Jelinek <jakub@redhat.com>
89 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
90 around drhs if type conversion to lacc->type is not useless.
92 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
94 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
96 (cortex-a57.cortex-a53): Likewise.
97 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
99 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
101 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
104 2014-02-11 Renlin Li <Renlin.Li@arm.com>
106 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
107 add_options_for_arm_vfp3.
109 2014-02-11 Jeff Law <law@redhat.com>
112 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
113 object with an undesirable mode.
115 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
118 * config/i386/sol2-9.h: New file.
119 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
120 *-*-solaris2.9*): Use it.
122 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
124 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
125 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
127 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
129 * config/microblaze/microblaze.c: Extend mcpu version format
131 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
133 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
135 2014-02-10 Richard Henderson <rth@redhat.com>
138 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
139 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
140 ms-abi vs -mno-accumulate-outgoing-args.
141 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
142 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
145 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
148 * cfgexpand.c (expand_asm_operands): Attach source location to
149 ASM_INPUT rtx objects.
150 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
152 2014-02-10 Nick Clifton <nickc@redhat.com>
154 * config/mn10300/mn10300.c (popcount): New function.
155 (mn10300_expand_prologue): Include saved registers in stack usage
158 2014-02-10 Jeff Law <law@redhat.com>
161 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
162 when changing the SET_DEST of a prior insn to avoid an input
165 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
167 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
168 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
169 -mcall-openbsd, or -mcall-linux.
170 (CC1_ENDIAN_BIG_SPEC): Remove.
171 (CC1_ENDIAN_LITTLE_SPEC): Remove.
172 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
173 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
174 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
175 and %cc1_endian_default.
176 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
178 2014-02-10 Richard Biener <rguenther@suse.de>
180 PR tree-optimization/60115
181 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
182 MEM_REF handling. Properly verify that the accesses are not
183 out of the objects bound.
185 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
187 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
190 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
192 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
193 proper constants and fix formatting.
194 (possible_polymorphic_call_targets): Fix formatting.
196 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
197 Ilya Tocar <ilya.tocar@intel.com>
199 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
200 (_mm512_loadu_epi32): Renamed into...
201 (_mm512_loadu_si512): This.
202 (_mm512_storeu_epi32): Renamed into...
203 (_mm512_storeu_si512): This.
204 (_mm512_maskz_ceil_ps): Removed.
205 (_mm512_maskz_ceil_pd): Ditto.
206 (_mm512_maskz_floor_ps): Ditto.
207 (_mm512_maskz_floor_pd): Ditto.
208 (_mm512_floor_round_ps): Ditto.
209 (_mm512_floor_round_pd): Ditto.
210 (_mm512_ceil_round_ps): Ditto.
211 (_mm512_ceil_round_pd): Ditto.
212 (_mm512_mask_floor_round_ps): Ditto.
213 (_mm512_mask_floor_round_pd): Ditto.
214 (_mm512_mask_ceil_round_ps): Ditto.
215 (_mm512_mask_ceil_round_pd): Ditto.
216 (_mm512_maskz_floor_round_ps): Ditto.
217 (_mm512_maskz_floor_round_pd): Ditto.
218 (_mm512_maskz_ceil_round_ps): Ditto.
219 (_mm512_maskz_ceil_round_pd): Ditto.
220 (_mm512_expand_pd): Ditto.
221 (_mm512_expand_ps): Ditto.
222 * config/i386/i386.c (ix86_builtins): Remove
223 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
225 * config/i386/predicates.md (const1256_operand): New.
226 (const_1_to_2_operand): Ditto.
227 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
228 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
229 (*avx512pf_gatherpf<mode>sf): Ditto.
230 (avx512pf_gatherpf<mode>df): Ditto.
231 (*avx512pf_gatherpf<mode>df_mask): Ditto.
232 (*avx512pf_gatherpf<mode>df): Ditto.
233 (avx512pf_scatterpf<mode>sf): Ditto.
234 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
235 (*avx512pf_scatterpf<mode>sf): Ditto.
236 (avx512pf_scatterpf<mode>df): Ditto.
237 (*avx512pf_scatterpf<mode>df_mask): Ditto.
238 (*avx512pf_scatterpf<mode>df): Ditto.
239 (avx512f_expand<mode>): Removed.
240 (<shift_insn><mode>3<mask_name>): Change predicate type.
242 2014-02-08 Jakub Jelinek <jakub@redhat.com>
244 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
245 not at the end of datarefs vector use ordered_remove to avoid
246 reordering datarefs vector.
249 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
250 mark local addressable non-static vars as GOVD_PRIVATE
251 instead of GOVD_LOCAL.
252 * omp-low.c (lower_omp_for): Move gimple_bind_vars
253 and BLOCK_VARS of gimple_bind_block to new_stmt rather
257 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
258 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
259 assume_aligned or alloc_align attributes.
260 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
261 arguments. Handle also assume_aligned and alloc_align attributes.
262 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
263 Handle calls to functions with assume_aligned or alloc_align
265 * doc/extend.texi: Document assume_aligned and alloc_align
268 2014-02-08 Terry Guo <terry.guo@arm.com>
270 * doc/invoke.texi: Document ARM -march=armv7e-m.
272 2014-02-08 Jakub Jelinek <jakub@redhat.com>
274 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
275 flag on __cilkrts_rethrow builtin.
278 * ipa-cp.c (determine_versionability): Fail at -O0
279 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
280 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
283 2014-02-04 Jakub Jelinek <jakub@redhat.com>
286 * tree-inline.c (copy_forbidden): Fail for
287 __attribute__((optimize (0))) functions.
289 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
291 * varpool.c: Include pointer-set.h.
292 (varpool_remove_unreferenced_decls): Variables in other partitions
293 will not be output; be however careful to not lose information
296 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
298 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
299 lookup in the vtable constructor.
301 2014-02-07 Jeff Law <law@redhat.com>
304 * config/m68k/m68k.md (ashldi_extsi): Turn into a
305 define_insn_and_split.
307 * ipa-inline.c (inline_small_functions): Fix typos.
309 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
311 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
312 (s390_can_use_return_insn): Declare.
313 * config/s390/s390.h (EPILOGUE_USES): Define.
314 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
316 (s390_chunkify_start): Handle return JUMP_LABELs.
317 (s390_early_mach): Emit a main_pool instruction on the entry edge.
318 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
319 (s390_can_use_return_insn): New functions.
320 (s390_fix_long_loop_prediction): Handle conditional returns.
321 (TARGET_SET_UP_BY_PROLOGUE): Define.
322 * config/s390/s390.md (ANY_RETURN): New code iterator.
323 (*creturn, *csimple_return, return, simple_return): New patterns.
325 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
327 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
328 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
329 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
330 REG_CFA_RESTORE list when deciding not to restore a register.
332 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
334 * config/s390/s390.c: Include tree-pass.h and context.h.
335 (s390_early_mach): New function, split out from...
336 (s390_emit_prologue): ...here.
337 (pass_data_s390_early_mach): New pass structure.
338 (pass_s390_early_mach): New class.
339 (s390_option_override): Create and register early_mach pass.
342 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
344 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
345 to match for the exit block.
347 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
349 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
350 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
351 Reject misaligned operands.
353 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
355 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
358 2014-02-07 Richard Biener <rguenther@suse.de>
361 * gimple-low.c (lower_builtin_posix_memalign): New function.
362 (lower_stmt): Call it to lower posix_memalign in a way
363 to make alignment info accessible.
365 2014-02-07 Jakub Jelinek <jakub@redhat.com>
368 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
369 __builtin_setjmp_receiver.
371 2014-02-07 Richard Biener <rguenther@suse.de>
374 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
375 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
376 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
377 Handle BUILT_IN_POSIX_MEMALIGN.
378 (find_func_clobbers): Likewise.
379 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
380 (call_may_clobber_ref_p_1): Likewise.
382 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
385 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
387 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
390 * lto-cgraph.c (lto_output_node): Use
391 symtab_get_symbol_partitioning_class.
392 (lto_output_varpool_node): likewise.
393 (symtab_get_symbol_partitioning_class): Move here from
395 * cgraph.h (symbol_partitioning_class): Likewise.
396 (symtab_get_symbol_partitioning_class): Declare.
398 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
400 * ggc.h (ggc_internal_cleared_alloc): New macro.
401 * vec.h (vec_safe_copy): Handle memory stats.
402 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
403 * target-globals.c (save_target_globals): Likewise.
405 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
408 * expr.c (emit_move_resolve_push): Export; be bit more selective
409 on when to clear alias set.
410 * expr.h (emit_move_resolve_push): Declare.
411 * function.h (struct function): Add tail_call_marked.
412 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
413 * config/i386/i386-protos.h (ix86_expand_push): Remove.
414 * config/i386/i386.md (TImode move expander): De not call
416 (FP push expanders): Preserve memory attributes.
417 * config/i386/sse.md (push<mode>1): Remove.
418 * config/i386/i386.c (ix86_expand_vector_move): Handle push
420 (ix86_expand_push): Remove.
421 * config/i386/mmx.md (push<mode>1): Remove.
423 2014-02-06 Jakub Jelinek <jakub@redhat.com>
425 PR rtl-optimization/60030
426 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
427 lopart with paradoxical subreg before shifting it up by hprec.
429 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
431 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
432 Remove extra newline at end of file.
433 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
434 (arm_issue_rate): Handle cortexa57.
435 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
436 (cortex-a57.cortex-a53): Likewise.
438 2014-02-06 Jakub Jelinek <jakub@redhat.com>
441 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
442 don't record in REG_FRAME_RELATED_EXPR registers not set in that
444 (arm_expand_prologue): Adjust all callers.
445 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
446 info, registers also at the lowest numbered registers side. Use
447 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
451 * var-tracking.c (adjust_mems): Before adding a SET
452 to amd->side_effects, adjust it's SET_SRC using
453 simplify_replace_fn_rtx.
455 2014-02-06 Alan Modra <amodra@gmail.com>
458 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
459 change SDmode to DDmode when lra_in_progress.
461 2014-02-06 Jakub Jelinek <jakub@redhat.com>
464 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
465 free_data_ref on the dr first, and before goto again also set dr
466 to the next dr. For simd_lane_access, free old datarefs[i] before
467 overwriting it. For get_vectype_for_scalar_type failure, don't
468 free_data_ref if simd_lane_access.
470 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
473 * tree.h (opts_for_fn): New inline function.
474 (opt_for_fn): Define.
475 * config/i386/i386.c (ix86_function_regparm): Use
476 opt_for_fn (decl, optimize) instead of optimize.
478 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
480 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
481 for SYMBOL_REF in large memory model.
483 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
485 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
487 (cortex-a57): Likewise.
488 (cortex-a57.cortex-a53): Likewise.
490 2014-02-06 Yury Gribov <y.gribov@samsung.com>
491 Kugan Vivekanandarajah <kuganv@linaro.org>
493 * config/arm/arm.c (arm_vector_alignment_reachable): Check
495 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
497 2014-02-06 Richard Biener <rguenther@suse.de>
499 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
500 set_loop_copy and initialize_original_copy_tables.
502 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
504 * config/aarch64/aarch64-simd.md
505 (aarch64_ashr_simddi): Change QI to SI.
507 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
508 Jakub Jelinek <jakub@redhat.com>
511 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
514 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
516 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
517 CODE_FOR_altivec_vpku[hw]um to
518 CODE_FOR_altivec_vpku[hw]um_direct.
519 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
520 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
521 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
522 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
524 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
526 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
527 generation for -maltivec=be.
528 (altivec_vsumsws): Simplify redundant test.
530 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
532 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
533 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
534 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
535 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
537 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
539 (altivec_vpks<VI_char>ss): Likewise.
540 (altivec_vpks<VI_char>us): Likewise.
541 (altivec_vpku<VI_char>us): Likewise.
542 (altivec_vpku<VI_char>um): Likewise.
543 (altivec_vpku<VI_char>um_direct): New (copy of
544 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
546 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
547 target is little endian and -maltivec=be is not specified.
548 (*altivec_vupkhs<VU_char>_direct): New (copy of
549 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
551 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
552 target is little endian and -maltivec=be is not specified.
553 (*altivec_vupkls<VU_char>_direct): New (copy of
554 altivec_vupkls<VU_char> that always emits vupkls*, for internal
556 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
557 little endian and -maltivec=be is not specified.
558 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
559 little endian and -maltivec=be is not specified.
561 2014-02-05 Richard Henderson <rth@redhat.com>
564 * combine-stack-adj.c: Revert r206943.
565 * sched-int.h (struct deps_desc): Add last_args_size.
566 * sched-deps.c (init_deps): Initialize it.
567 (sched_analyze_insn): Add OUTPUT dependencies between insns that
568 contain REG_ARGS_SIZE notes.
570 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
572 * lto-cgraph.c (asm_nodes_output): Make global.
573 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
574 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
576 (driver_handle_option): Handle OPT_fwpa.
578 2014-02-05 Jakub Jelinek <jakub@redhat.com>
581 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
582 a comment typo and formatting issue. If odr_hash hasn't been
583 created, return vNULL and set *completep to false.
586 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
587 bb with no successors.
589 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
592 * doc/invoke.texi (-march): Clarify documentation for ARM.
596 2014-02-05 Richard Biener <rguenther@suse.de>
598 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
599 when not vectorizing because of too many alias checks.
600 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
601 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
603 2014-02-05 Nick Clifton <nickc@redhat.com>
605 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
606 accept extended registers in any mode when compiling for the
609 2014-02-05 Yury Gribov <y.gribov@samsung.com>
611 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
612 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
613 sanitization attributes.
614 (can_inline_edge_p): Likewise.
615 (sanitize_attrs_match_for_inline_p): New function.
617 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
619 * ipa-prop.c (detect_type_change): Shor circuit testing of
620 type changes on THIS pointer.
622 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
625 * config/pa/pa.c (legitimize_tls_address): Return original address
626 if not passed a SYMBOL_REF rtx.
627 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
629 (pa_emit_move_sequence): Simplify TLS source operands.
630 (pa_legitimate_constant_p): Reject all TLS constants.
631 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
632 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
634 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
636 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
637 groups when we know they are controlled by LTO.
638 * varasm.c (default_binds_local_p_1): If object is in other partition,
639 it will be resolved locally.
641 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
643 * config/host-linux.c (linux_gt_pch_use_address): Don't
644 use SSIZE_MAX because it is not always defined.
646 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
649 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
650 threshold for pseudo splitting.
651 (update_ebb_live_info): Process call argument hard registers and
652 hard registers from insn definition too.
653 (max_small_class_regs_num): New constant.
654 (inherit_in_ebb): Update live hard regs through EBBs. Update
655 reloads_num only for small register classes. Don't split for
658 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
661 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
664 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
666 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
668 2014-02-04 Marek Polacek <polacek@redhat.com>
670 * gdbinit.in (pel): Define.
672 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
674 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
677 2014-02-04 Richard Biener <rguenther@suse.de>
680 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
681 in function context local.
682 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
683 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
684 similar to LTO_imported_decl_ref.
686 2014-02-04 Jakub Jelinek <jakub@redhat.com>
688 PR tree-optimization/60002
689 * cgraphclones.c (build_function_decl_skip_args): Clear
692 PR tree-optimization/60023
693 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
694 false to gsi_replace.
695 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
696 has been in some EH region and vec_stmt could throw, add
697 vec_stmt into the same EH region.
698 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
699 has no lhs, ignore it.
700 * internal-fn.c (expand_MASK_LOAD): Likewise.
703 * tree-inline.c (copy_forbidden): Fail for
704 __attribute__((optimize (0))) functions.
707 * omp-low.c (simd_clone_struct_copy): If from->inbranch
708 is set, copy one less argument.
709 (expand_simd_clones): Don't subtract clone_info->inbranch
710 from simd_clone_struct_alloc argument.
712 PR rtl-optimization/57915
713 * recog.c (simplify_while_replacing): If all unary/binary/relational
714 operation arguments are constant, attempt to simplify those.
717 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
718 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
720 2014-02-04 Richard Biener <rguenther@suse.de>
722 PR tree-optimization/60012
723 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
724 TBAA disambiguation to all DDRs.
726 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
729 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
730 (LINK_SPEC): Use it for -shared, -shared-libgcc.
732 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
735 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
737 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
739 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
740 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
742 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
745 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
746 to figure out targets of polymorphic calls with known decl.
747 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
748 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
749 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
750 (get_polymorphic_call_info): ... here.
751 (get_polymorphic_call_info_from_invariant): New function.
753 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
755 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
756 lookup via vtable pointer; check for type consistency
757 and turn inconsitent facts into UNREACHABLE.
758 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
759 * gimple-fold.c (gimple_get_virt_method_for_vtable):
760 Do not ICE on type inconsistent querries; return UNREACHABLE
763 2014-02-03 Richard Henderson <rth@twiddle.net>
766 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
767 already processed this node.
768 (normalize_one_pred_1): Pass along mark_set.
769 (normalize_one_pred): Create and destroy a pointer_set_t.
770 (normalize_one_pred_chain): Likewise.
772 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
774 PR gcov-profile/58602
775 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
777 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
780 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
781 Try to devirtualize by the knowledge of virtual table pointer given by
782 aggregate propagation.
783 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
784 ipa_print_node_jump_functions): Dump also offset that
785 is relevant for polymorphic calls.
786 (determine_known_aggregate_parts): Add arg_type parameter; use it
787 instead of determining the type from pointer type.
788 (ipa_compute_jump_functions_for_edge): Update call of
789 determine_known_aggregate_parts.
790 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
791 (gimple_get_virt_method_for_binfo): ... here; simplify using
792 vtable_pointer_value_to_vtable.
793 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
794 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
795 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
796 (vtable_pointer_value_to_vtable): Break out from ...; handle also
798 (vtable_pointer_value_to_binfo): ... here.
799 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
801 2014-02-03 Teresa Johnson <tejohnson@google.com>
803 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
804 redef of outer loop index variable.
806 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
810 * doc/extend.texi (Function Attributes): Typo.
812 2014-02-03 Cong Hou <congh@google.com>
814 PR tree-optimization/60000
815 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
816 if the vectorized statement is a store. A store statement can only
817 appear at the end of pattern statements.
819 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
821 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
822 (ix86_option_override_internal): Default long double to 64-bit for
823 32-bit Bionic and to 128-bit for 64-bit Bionic.
825 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
826 TARGET_LONG_DOUBLE_128 is true.
827 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
829 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
830 (mlong-double-64): Negate -mlong-double-128.
831 (mlong-double-128): New option.
833 * config/i386/i386-c.c (ix86_target_macros): Define
834 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
836 * doc/invoke.texi: Document -mlong-double-128.
838 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
840 PR rtl-optimization/60024
841 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
843 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
845 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
847 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
849 PR rtl-optimization/57662
850 * sel-sched.c (code_motion_path_driver): Do not mark already not
851 existing blocks in the visiting bitmap.
853 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
855 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
856 on the insn being emitted.
858 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
859 Will Deacon <will.deacon@arm.com>
861 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
863 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
865 * config/arm/arm-tables.opt: Regenerate.
867 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
869 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
870 for vector types other than V16QImode.
871 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
872 define_expand, and call altivec_expand_vec_perm_le when producing
873 code with little endian element order.
874 (*altivec_vperm_<mode>_internal): New insn having previous
875 behavior of altivec_vperm_<mode>.
876 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
877 altivec_expand_vec_perm_le when producing code with little endian
879 (*altivec_vperm_<mode>_uns_internal): New insn having previous
880 behavior of altivec_vperm_<mode>_uns.
882 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
884 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
885 (altivec_vsumsws): Add handling for -maltivec=be with a little
887 (altivec_vsumsws_direct): New.
888 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
891 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
893 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
894 vtable_pointer_value_to_binfo): New functions.
895 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
896 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
898 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
900 * config/nios2/nios2.md (load_got_register): Initialize GOT
901 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
902 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
904 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
906 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
907 preserverd by passthrough, do not propagate the type.
909 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
911 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
912 (mips_atomic_assign_expand_fenv): New function.
913 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
915 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
917 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
918 (__builtin_mips_set_fcsr): Likewise.
919 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
921 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
922 (mips16_expand_set_fcsr): Likewise.
923 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
924 (mips16_set_fcsr_stub): Likewise.
925 (mips16_get_fcsr_one_only_stub): New class.
926 (mips16_set_fcsr_one_only_stub): Likewise.
927 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
928 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
929 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
930 (hard_float): New availability predicate.
931 (mips_builtins): Add get_fcsr and set_fcsr.
932 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
933 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
934 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
935 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
936 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
939 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
941 * config/mips/mips.c (mips_one_only_stub): New class.
942 (mips_need_mips16_rdhwr_p): Replace with...
943 (mips16_rdhwr_stub): ...this new variable.
944 (mips16_stub_call_address): New function.
945 (mips16_rdhwr_one_only_stub): New class.
946 (mips_expand_thread_pointer): Use mips16_stub_call_address.
947 (mips_output_mips16_rdhwr): Delete.
948 (mips_finish_stub): New function.
949 (mips_code_end): Use it to handle rdhwr stubs.
951 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
954 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
955 when calculating size of integer atomic types.
957 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
959 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
961 2014-02-01 Jakub Jelinek <jakub@redhat.com>
963 PR tree-optimization/60003
964 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
965 * profile.c (branch_prob): Use gimple_call_builtin_p
966 to check for BUILT_IN_SETJMP_RECEIVER.
967 * tree-inline.c (copy_bb): Call notice_special_calls.
969 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
972 * lra-constraints.c (process_alt_operands): Update reload_sum only
975 2014-01-31 Richard Henderson <rth@redhat.com>
978 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
979 until after else_eh is processed.
981 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
983 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
984 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
985 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
986 in smmintrin.h, remove them.
987 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
988 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
989 * config/i386/i386.md (ROUND_SAE): Fix value.
990 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
991 (const48_operand): New.
992 * config/i386/subst.md (round), (round_expand): Use
993 const_4_or_8_to_11_operand.
994 (round_saeonly), (round_saeonly_expand): Use const48_operand.
996 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
998 * config/i386/constraints.md (Yk): Swap meaning with k.
999 * config/i386/i386.md (movhi_internal): Change Yk to k.
1000 (movqi_internal): Ditto.
1001 (*k<logic><mode>): Ditto.
1004 (kandn<mode>): Ditto.
1005 (*<code>hi_1): Ditto.
1006 (*<code>qi_1): Ditto.
1007 (kxnor<mode>): Ditto.
1008 (kortestzhi): Ditto.
1009 (kortestchi): Ditto.
1011 (*one_cmplhi2_1): Ditto.
1012 (*one_cmplqi2_1): Ditto.
1013 * config/i386/sse.md (): Change k to Yk.
1014 (avx512f_load<mode>_mask): Ditto.
1015 (avx512f_blendm<mode>): Ditto.
1016 (avx512f_store<mode>_mask): Ditto.
1017 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1018 (avx512f_storedqu<mode>_mask): Ditto.
1019 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1021 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1022 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1023 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1024 (avx512f_maskcmp<mode>3): Ditto.
1025 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1026 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1027 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1028 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1029 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1030 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1031 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1032 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1033 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1034 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1035 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1036 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1037 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1038 (vec_extract_lo_<mode>_maskm): Ditto.
1039 (vec_extract_hi_<mode>_maskm): Ditto.
1040 (avx512f_vternlog<mode>_mask): Ditto.
1041 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1042 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1043 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1044 (avx512f_<code>v8div16qi2_mask): Ditto.
1045 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1046 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1047 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1048 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1049 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1050 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1051 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1052 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1053 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1054 (avx512cd_maskb_vec_dupv8di): Ditto.
1055 (avx512cd_maskw_vec_dupv16si): Ditto.
1056 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1057 (avx512f_vpermi2var<mode>3_mask): Ditto.
1058 (avx512f_vpermi2var<mode>3_mask): Ditto.
1059 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1060 (*avx512f_gathersi<mode>): Ditto.
1061 (*avx512f_gathersi<mode>_2): Ditto.
1062 (*avx512f_gatherdi<mode>): Ditto.
1063 (*avx512f_gatherdi<mode>_2): Ditto.
1064 (*avx512f_scattersi<mode>): Ditto.
1065 (*avx512f_scatterdi<mode>): Ditto.
1066 (avx512f_compress<mode>_mask): Ditto.
1067 (avx512f_compressstore<mode>_mask): Ditto.
1068 (avx512f_expand<mode>_mask): Ditto.
1069 * config/i386/subst.md (mask): Change k to Yk.
1070 (mask_scalar_merge): Ditto.
1073 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1075 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1077 2014-01-31 Richard Biener <rguenther@suse.de>
1080 * builtins.c (fold_builtin_memory_op): Make sure to not
1081 use a floating-point mode or a boolean or enumeral type for
1084 2014-01-30 DJ Delorie <dj@redhat.com>
1086 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1087 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1088 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1089 whenever main() has an epilogue.
1091 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1093 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1094 unused variable "field".
1095 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1096 (vsx_mergeh_<mode>): Likewise.
1097 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1098 (altivec_vmrghh): Likewise.
1099 (altivec_vmrghw): Likewise.
1100 (altivec_vmrglb): Likewise.
1101 (altivec_vmrglh): Likewise.
1102 (altivec_vmrglw): Likewise.
1103 (altivec_vspltb): Add missing uses.
1104 (altivec_vsplth): Likewise.
1105 (altivec_vspltw): Likewise.
1106 (altivec_vspltsf): Likewise.
1108 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1111 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1112 frame related instructions.
1114 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1116 PR rtl-optimization/59959
1117 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1118 any reload of register whose subreg is invalid.
1120 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1122 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1123 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1124 Add missing return type - void.
1126 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1128 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1129 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1130 remove element index adjustment for endian (now handled in vsx.md
1132 (altivec_expand_vec_perm_const): Use
1133 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1134 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1135 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1136 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1137 define_expand and a new define_insn *altivec_vspltb_internal;
1138 adjust for -maltivec=be on a little endian target.
1139 (altivec_vspltb_direct): New.
1140 (altivec_vsplth): Divide into a define_expand and a new
1141 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1142 little endian target.
1143 (altivec_vsplth_direct): New.
1144 (altivec_vspltw): Divide into a define_expand and a new
1145 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1146 little endian target.
1147 (altivec_vspltw_direct): New.
1148 (altivec_vspltsf): Divide into a define_expand and a new
1149 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1150 a little endian target.
1152 2014-01-30 Richard Biener <rguenther@suse.de>
1154 PR tree-optimization/59993
1155 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1156 can propagate form the earlier stmt and avoid the transform
1157 when the intermediate result is needed.
1159 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1161 * README.Portability: Fix typo.
1163 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1165 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1166 comparison_operator with ordered_comparison_operator.
1168 2014-01-30 Nick Clifton <nickc@redhat.com>
1170 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1171 Rename to mn10300_store_multiple_regs.
1172 * config/mn10300/mn10300.c: Likewise.
1173 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1174 store_multiple_regs.
1175 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1176 Call mn10300_store_multiple_regs.
1178 2014-01-30 Nick Clifton <nickc@redhat.com>
1179 DJ Delorie <dj@redhat.com>
1181 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1182 %fp 2 to keep registers after it properly word-aligned.
1183 (rl78_alloc_physical_registers_umul): Handle the case where both
1184 input operands are the same.
1186 2014-01-30 Richard Biener <rguenther@suse.de>
1188 PR tree-optimization/59903
1189 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1192 2014-01-30 Jason Merrill <jason@redhat.com>
1195 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1198 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1200 2014-01-30 Richard Biener <rguenther@suse.de>
1202 PR tree-optimization/59951
1203 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1205 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1208 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1209 SFmode to DFmode case.
1211 2014-01-29 DJ Delorie <dj@redhat.com>
1213 * config/msp430/msp430.opt (-minrt): New.
1214 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1216 (ENDFILE_SPEC): Likewise.
1218 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1220 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1221 (estimate_function_body_sizes): Use it.
1223 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1226 * dwarf2out.c (is_cxx_auto): New.
1227 (is_base_type): Use it.
1228 (gen_type_die_with_usage): Likewise.
1230 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1232 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1233 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1234 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1235 -maltivec=be with LE targets.
1236 (vsx_mergeh_<mode>): Likewise.
1237 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1238 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1239 (altivec_vmrghb): Replace with define_expand and new
1240 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1241 (altivec_vmrghb_direct): New define_insn.
1242 (altivec_vmrghh): Replace with define_expand and new
1243 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1244 (altivec_vmrghh_direct): New define_insn.
1245 (altivec_vmrghw): Replace with define_expand and new
1246 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1247 (altivec_vmrghw_direct): New define_insn.
1248 (*altivec_vmrghsf): Adjust for endianness.
1249 (altivec_vmrglb): Replace with define_expand and new
1250 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1251 (altivec_vmrglb_direct): New define_insn.
1252 (altivec_vmrglh): Replace with define_expand and new
1253 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1254 (altivec_vmrglh_direct): New define_insn.
1255 (altivec_vmrglw): Replace with define_expand and new
1256 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1257 (altivec_vmrglw_direct): New define_insn.
1258 (*altivec_vmrglsf): Adjust for endianness.
1259 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1260 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1261 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1262 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1263 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1264 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1265 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1266 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1268 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1270 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1271 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1274 2014-01-29 Richard Biener <rguenther@suse.de>
1276 PR tree-optimization/58742
1277 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1278 associate_pointerplus_align.
1279 (associate_pointerplus_diff): New function.
1280 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1281 and associate_pointerplus_diff.
1283 2014-01-29 Richard Biener <rguenther@suse.de>
1285 * lto-streamer.h (LTO_major_version): Bump to 3.
1286 (LTO_minor_version): Reset to 0.
1288 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1290 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1291 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1292 (arm_file_start): Generate correct asm header for armv7ve.
1293 * config/arm/bpabi.h: Add multilib support for armv7ve.
1294 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1295 and cortex-a15 to armv7ve.
1296 * config/arm/t-aprofile: Add multilib support for armv7ve.
1297 * doc/invoke.texi: Document -march=armv7ve.
1299 2014-01-29 Richard Biener <rguenther@suse.de>
1301 PR tree-optimization/58742
1302 * tree-ssa-forwprop.c (associate_plusminus): Return true
1303 if we changed sth, defer EH cleanup to ...
1304 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1305 (simplify_mult): New function.
1307 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1310 PR tree-optimization/59920
1311 * tree.c (build_common_builtin_nodes): Remove
1312 __builtin_setjmp_dispatcher initialization.
1313 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1314 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1315 instead of gsi_after_labels + manually skipping debug stmts.
1316 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1317 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1318 * tree-inline.c (copy_edges_for_bb): Remove
1319 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1320 argument. Ignore computed_goto_p stmts. Don't call
1321 make_abnormal_goto_edges. If a call might need abnormal edges
1322 for non-local gotos, see if it already has an edge to
1323 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1324 with true argument, don't do anything then, otherwise add
1325 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1326 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1328 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1329 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1330 (lower_stmt): Don't set data->calls_builtin_setjmp.
1331 (lower_builtin_setjmp): Adjust comment.
1332 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1333 * tree-cfg.c (found_computed_goto): Remove.
1334 (factor_computed_gotos): Remove.
1335 (make_goto_expr_edges): Return bool, true for computed gotos.
1336 Don't call make_abnormal_goto_edges.
1337 (build_gimple_cfg): Don't set found_computed_goto, don't call
1338 factor_computed_gotos.
1339 (computed_goto_p): No longer static.
1340 (make_blocks): Don't set found_computed_goto.
1341 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1342 (make_edges): If make_goto_expr_edges returns true, push bb
1343 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1344 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1345 vector. Record mapping between bbs and OpenMP regions if there
1346 are any, adjust make_gimple_omp_edges caller. Call
1347 handle_abnormal_edges.
1348 (make_abnormal_goto_edges): Remove.
1349 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1350 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1351 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1352 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1353 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1354 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1355 filling *region also set *region_idx to (*region)->entry->index.
1358 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1359 For REGs set ORIGINAL_REGNO.
1361 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1363 * doc/md.texi: Mention that a target shouldn't implement
1364 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1367 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1369 PR tree-optimization/59594
1370 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1371 a copy of the datarefs vector rather than the vector itself.
1373 2014-01-28 Jason Merrill <jason@redhat.com>
1376 * dwarf2out.c (auto_die): New static.
1377 (gen_type_die_with_usage): Handle C++1y 'auto'.
1378 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1381 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1384 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1385 (SPEC_X32): Likewise.
1386 (SPEC_64): Likewise.
1387 * config/i386/i386.c (ix86_option_override_internal): Turn off
1388 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1390 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1391 * config/i386/i386.h (TARGET_16BIT): New macro.
1392 (TARGET_16BIT_P): Likewise.
1393 * config/i386/i386.opt: Add m16.
1394 * doc/invoke.texi: Document -m16.
1396 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1398 PR preprocessor/59935
1399 * input.c (location_get_source_line): Bail out on when line number
1400 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1402 2014-01-28 Richard Biener <rguenther@suse.de>
1404 PR tree-optimization/58742
1405 * tree-ssa-forwprop.c (associate_plusminus): Handle
1406 pointer subtraction of the form (T)(P + A) - (T)P.
1408 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1410 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1413 2014-01-28 Richard Biener <rguenther@suse.de>
1416 2014-01-28 Richard Biener <rguenther@suse.de>
1418 PR rtl-optimization/45364
1419 PR rtl-optimization/59890
1420 * var-tracking.c (local_get_addr_clear_given_value): Handle
1421 already cleared slot.
1422 (val_reset): Handle not allocated local_get_addr_cache.
1423 (vt_find_locations): Use post-order on the inverted CFG.
1425 2014-01-28 Richard Biener <rguenther@suse.de>
1427 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1429 2014-01-28 Richard Biener <rguenther@suse.de>
1431 PR rtl-optimization/45364
1432 PR rtl-optimization/59890
1433 * var-tracking.c (local_get_addr_clear_given_value): Handle
1434 already cleared slot.
1435 (val_reset): Handle not allocated local_get_addr_cache.
1436 (vt_find_locations): Use post-order on the inverted CFG.
1438 2014-01-28 Alan Modra <amodra@gmail.com>
1440 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1441 * configure.ac <recursive call for build != host>: Define
1442 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1443 and LD_FOR_BUILD too.
1444 * configure: Regenerate.
1446 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1448 * config/i386/i386.c (get_builtin_code_for_version): Separate
1449 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1450 Broadwell from Haswell.
1452 2014-01-27 Steve Ellcey <sellcey@mips.com>
1454 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1455 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1456 * config/mips/mips.c (mips_option_override): Change setting
1458 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1459 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1460 Change from Mask to Var.
1462 2014-01-27 Jeff Law <law@redhat.com>
1464 * ipa-inline.c (inline_small_functions): Fix typo.
1466 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1468 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1469 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1470 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1471 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1472 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1473 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1474 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1475 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1476 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1477 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1478 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1479 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1480 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1481 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1482 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1483 (_mm512_storeu_epi64): Ditto.
1484 (_mm512_cmpge_epi32_mask): Ditto.
1485 (_mm512_cmpge_epu32_mask): Ditto.
1486 (_mm512_cmpge_epi64_mask): Ditto.
1487 (_mm512_cmpge_epu64_mask): Ditto.
1488 (_mm512_cmple_epi32_mask): Ditto.
1489 (_mm512_cmple_epu32_mask): Ditto.
1490 (_mm512_cmple_epi64_mask): Ditto.
1491 (_mm512_cmple_epu64_mask): Ditto.
1492 (_mm512_cmplt_epi32_mask): Ditto.
1493 (_mm512_cmplt_epu32_mask): Ditto.
1494 (_mm512_cmplt_epi64_mask): Ditto.
1495 (_mm512_cmplt_epu64_mask): Ditto.
1496 (_mm512_cmpneq_epi32_mask): Ditto.
1497 (_mm512_cmpneq_epu32_mask): Ditto.
1498 (_mm512_cmpneq_epi64_mask): Ditto.
1499 (_mm512_cmpneq_epu64_mask): Ditto.
1500 (_mm512_expand_pd): Ditto.
1501 (_mm512_expand_ps): Ditto.
1502 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1503 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1504 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1505 * config/i386/i386.c (ix86_builtins): Add
1506 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1507 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1508 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1509 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1510 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1511 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1512 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1513 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1514 IX86_BUILTIN_PMOVUSQW512_MEM.
1515 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1516 __builtin_ia32_pmovsqd512mem_mask,
1517 __builtin_ia32_pmovqd512mem_mask,
1518 __builtin_ia32_pmovusqw512mem_mask,
1519 __builtin_ia32_pmovsqw512mem_mask,
1520 __builtin_ia32_pmovqw512mem_mask,
1521 __builtin_ia32_pmovusdw512mem_mask,
1522 __builtin_ia32_pmovsdw512mem_mask,
1523 __builtin_ia32_pmovdw512mem_mask,
1524 __builtin_ia32_pmovqb512mem_mask,
1525 __builtin_ia32_pmovusqb512mem_mask,
1526 __builtin_ia32_pmovsqb512mem_mask,
1527 __builtin_ia32_pmovusdb512mem_mask,
1528 __builtin_ia32_pmovsdb512mem_mask,
1529 __builtin_ia32_pmovdb512mem_mask.
1530 (bdesc_args): Add __builtin_ia32_expanddf512,
1531 __builtin_ia32_expandsf512.
1532 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1533 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1534 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1535 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1536 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1537 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1538 (avx512f_<code>v8div16qi2_mask_store): This.
1539 (avx512f_expand<mode>): New.
1541 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1543 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1545 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1546 (_mm512_prefetch_i32scatter_pd): Ditto.
1547 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1548 (_mm512_prefetch_i64scatter_pd): Ditto.
1549 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1550 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1551 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1552 (_mm512_prefetch_i32scatter_ps): Ditto.
1553 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1554 (_mm512_prefetch_i64scatter_ps): Ditto.
1555 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1556 * config/i386/i386-builtin-types.def: Define
1557 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1558 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1559 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1560 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1561 IX86_BUILTIN_SCATTERPFQPD.
1562 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1563 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1564 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1565 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1566 __builtin_ia32_scatterpfqps.
1567 (ix86_expand_builtin): Expand new built-ins.
1568 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1569 fix memory access data type.
1570 (*avx512pf_gatherpf<mode>_mask): Ditto.
1571 (*avx512pf_gatherpf<mode>): Ditto.
1572 (avx512pf_scatterpf<mode>): Ditto.
1573 (*avx512pf_scatterpf<mode>_mask): Ditto.
1574 (*avx512pf_scatterpf<mode>): Ditto.
1575 (GATHER_SCATTER_SF_MEM_MODE): New.
1576 (avx512pf_gatherpf<mode>df): Ditto.
1577 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1578 (*avx512pf_scatterpf<mode>df): Ditto.
1580 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1583 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1584 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1587 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1589 * common/config/arm/arm-common.c
1590 (arm_rewrite_mcpu): Handle multiple names.
1592 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1594 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1596 * gimple-builder.h (create_gimple_tmp): Delete.
1598 2014-01-27 Christian Bruel <christian.bruel@st.com>
1600 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1603 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1605 * config/pa/pa.md (call): Generate indirect long calls to non-local
1606 functions when outputing 32-bit code.
1607 (call_value): Likewise except for special call to buggy powf function.
1609 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1610 portable runtime and PIC indirect calls.
1611 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1612 and PIC call sequences. Use ldo instead of blr to set return register
1613 in PIC call sequence.
1615 2014-01-25 Walter Lee <walt@tilera.com>
1617 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1618 avoid clobbering a live register.
1620 2014-01-25 Walter Lee <walt@tilera.com>
1622 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1623 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1624 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1625 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1627 2014-01-25 Walter Lee <walt@tilera.com>
1629 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1630 arguments on even registers.
1631 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1633 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1634 (BIGGEST_ALIGNMENT): Ditto.
1635 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1637 2014-01-25 Walter Lee <walt@tilera.com>
1639 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1640 insns before bundling.
1641 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1643 2014-01-25 Walter Lee <walt@tilera.com>
1645 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1646 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1647 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1649 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1651 * config/mips/constraints.md (kl): Delete.
1652 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1653 define expands, using...
1654 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1655 instructions for MIPS16.
1656 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1657 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1659 2014-01-25 Walter Lee <walt@tilera.com>
1661 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1665 2014-01-25 Walter Lee <walt@tilera.com>
1667 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1668 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1670 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1672 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1675 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1677 * print-rtl.c (in_call_function_usage): New var.
1678 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1679 EXPR_LIST mode as mode and not as reg note name.
1682 * cfgloopmanip.c (copy_loop_info): If
1683 loop->warned_aggressive_loop_optimizations, make sure
1684 the flag is set in target loop too.
1686 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1688 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1690 * builtins.def: Likewise.
1691 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1692 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1693 * ira.c (ira_setup_eliminable_regset): Likewise.
1694 * omp-low.c (gate_expand_omp): Likewise.
1695 (execute_lower_omp): Likewise.
1696 (diagnose_sb_0): Likewise.
1697 (gate_diagnose_omp_blocks): Likewise.
1698 (simd_clone_clauses_extract): Likewise.
1701 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1703 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1704 correction for little endian...
1705 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1708 2014-01-24 Jeff Law <law@redhat.com>
1710 PR tree-optimization/59919
1711 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1712 for non-returning calls.
1714 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1716 * common/config/aarch64/aarch64-common.c
1717 (aarch64_rewrite_mcpu): Handle multiple names.
1718 * config/aarch64/aarch64.h
1719 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1721 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1723 * input.c (add_file_to_cache_tab): Handle the case where fopen
1726 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1729 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1730 from push operand if code of push isn't PRE_DEC.
1732 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1735 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1736 -mquad-memory-atomic. Update -mquad-memory documentation to say
1737 it is only used for non-atomic loads/stores.
1739 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1740 -mquad-memory or -mquad-memory-atomic switches.
1742 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1743 -mquad-memory-atomic to ISA 2.07 support.
1745 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1746 to separate support of normal quad word memory operations (ldq, stq)
1747 from the atomic quad word memory operations.
1749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1750 support to separate non-atomic quad word operations from atomic
1751 quad word operations. Disable non-atomic quad word operations in
1752 little endian mode so that we don't have to swap words after the
1753 load and before the store.
1754 (quad_load_store_p): Add comment about atomic quad word support.
1755 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1756 options printed with -mdebug=reg.
1758 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1759 -mquad-memory-atomic as the test for whether we have quad word
1760 atomic instructions.
1761 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1762 or -mp8-vector are used, allow byte/half-word atomic operations.
1764 * config/rs6000/sync.md (load_lockedti): Insure that the address
1765 is a proper indexed or indirect address for the lqarx instruction.
1766 On little endian systems, swap the hi/lo registers after the lqarx
1768 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1769 insure the address is valid for the lqarx instruction.
1770 (store_conditionalti): Insure that the address is a proper indexed
1771 or indirect address for the stqcrx. instruction. On little endian
1772 systems, swap the hi/lo registers before doing the stqcrx.
1774 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1775 insure the address is valid for the stqcrx. instruction.
1777 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1778 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1779 type of quad memory support is available.
1781 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1784 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1785 there is a danger of looping.
1787 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1789 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1790 force flag_ira_loop_pressure if set via command line.
1792 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1794 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1795 (ashr_simd): New builtin handling DI mode.
1796 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1797 (aarch64_sshr_simddi): New match pattern.
1798 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1799 (vshrd_n_s64): Likewise.
1800 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1802 2014-01-23 Nick Clifton <nickc@redhat.com>
1804 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1805 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1806 favour of mcu specific scripts.
1807 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1810 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1811 Alex Velenko <Alex.Velenko@arm.com>
1813 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1814 (vaddv_s16): Likewise.
1815 (vaddv_s32): Likewise.
1816 (vaddv_u8): Likewise.
1817 (vaddv_u16): Likewise.
1818 (vaddv_u32): Likewise.
1819 (vaddvq_s8): Likewise.
1820 (vaddvq_s16): Likewise.
1821 (vaddvq_s32): Likewise.
1822 (vaddvq_s64): Likewise.
1823 (vaddvq_u8): Likewise.
1824 (vaddvq_u16): Likewise.
1825 (vaddvq_u32): Likewise.
1826 (vaddvq_u64): Likewise.
1827 (vaddv_f32): Likewise.
1828 (vaddvq_f32): Likewise.
1829 (vaddvq_f64): Likewise.
1830 (vmaxv_f32): Likewise.
1831 (vmaxv_s8): Likewise.
1832 (vmaxv_s16): Likewise.
1833 (vmaxv_s32): Likewise.
1834 (vmaxv_u8): Likewise.
1835 (vmaxv_u16): Likewise.
1836 (vmaxv_u32): Likewise.
1837 (vmaxvq_f32): Likewise.
1838 (vmaxvq_f64): Likewise.
1839 (vmaxvq_s8): Likewise.
1840 (vmaxvq_s16): Likewise.
1841 (vmaxvq_s32): Likewise.
1842 (vmaxvq_u8): Likewise.
1843 (vmaxvq_u16): Likewise.
1844 (vmaxvq_u32): Likewise.
1845 (vmaxnmv_f32): Likewise.
1846 (vmaxnmvq_f32): Likewise.
1847 (vmaxnmvq_f64): Likewise.
1848 (vminv_f32): Likewise.
1849 (vminv_s8): Likewise.
1850 (vminv_s16): Likewise.
1851 (vminv_s32): Likewise.
1852 (vminv_u8): Likewise.
1853 (vminv_u16): Likewise.
1854 (vminv_u32): Likewise.
1855 (vminvq_f32): Likewise.
1856 (vminvq_f64): Likewise.
1857 (vminvq_s8): Likewise.
1858 (vminvq_s16): Likewise.
1859 (vminvq_s32): Likewise.
1860 (vminvq_u8): Likewise.
1861 (vminvq_u16): Likewise.
1862 (vminvq_u32): Likewise.
1863 (vminnmv_f32): Likewise.
1864 (vminnmvq_f32): Likewise.
1865 (vminnmvq_f64): Likewise.
1867 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1869 * config/aarch64/aarch64-simd.md
1870 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1871 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1872 (*aarch64_mul3_elt<mode>): Likewise.
1873 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1874 (*aarch64_mul3_elt_to_64v2df): Likewise.
1875 (*aarch64_mla_elt<mode>): Likewise.
1876 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1877 (*aarch64_mls_elt<mode>): Likewise.
1878 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1879 (*aarch64_fma4_elt<mode>): Likewise.
1880 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1881 (*aarch64_fma4_elt_to_64v2df): Likewise.
1882 (*aarch64_fnma4_elt<mode>): Likewise.
1883 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1884 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1885 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1886 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1887 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1888 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1889 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1890 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1891 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1893 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1895 * config/aarch64/aarch64-simd.md
1896 (aarch64_be_checked_get_lane<mode>): New define_expand.
1897 * config/aarch64/aarch64-simd-builtins.def
1898 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1899 New builtin definition.
1900 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1901 Use new safe be builtin.
1903 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1905 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1907 (aarch64_be_st1<mode>): Likewise.
1908 (aarch_ld1<VALL:mode>): Define_expand modified.
1909 (aarch_st1<VALL:mode>): Likewise.
1910 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1911 (UNSPEC_ST1): Likewise.
1913 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1915 * config/microblaze/microblaze.md: Add trap insn and attribute
1917 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1919 PR preprocessor/58580
1920 * input.h (location_get_source_line): Take an additional line_size
1922 (void diagnostics_file_cache_fini): Declare new function.
1923 * input.c (struct fcache): New type.
1924 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1925 New static constants.
1926 (diagnostic_file_cache_init, total_lines_num)
1927 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1928 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1929 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1930 (get_next_line, read_next_line, goto_next_line, read_line_num):
1931 New static function definitions.
1932 (diagnostic_file_cache_fini): New function.
1933 (location_get_source_line): Take an additional output line_len
1934 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1936 * diagnostic.c (diagnostic_finish): Call
1937 diagnostic_file_cache_fini.
1938 (adjust_line): Take an additional input parameter for the length
1939 of the line, rather than calculating it with strlen.
1940 (diagnostic_show_locus): Adjust the use of
1941 location_get_source_line and adjust_line with respect to their new
1942 signature. While displaying a line now, do not stop at the first
1943 null byte. Rather, display the zero byte as a space and keep
1944 going until we reach the size of the line.
1945 * Makefile.in: Add vec.o to OBJS-libcommon
1947 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1948 Ilya Tocar <ilya.tocar@intel.com>
1950 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1951 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1952 (__builtin_ia32_kmov16): Ditto.
1953 * config/i386/i386.md (UNSPEC_KMOV): New.
1956 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1958 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1959 (_mm512_storeu_si512): Ditto.
1961 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1964 * rtl.h (get_referenced_operands): Declare.
1965 * recog.c (get_referenced_operands): New function.
1966 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1967 operands have been referenced when recording LO_SUM references.
1969 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1971 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1973 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1975 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1976 Enable for generic and recent AMD targets.
1978 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1980 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1981 ARG_SIZE note when adjustment was eliminated.
1983 2014-01-22 Jeff Law <law@redhat.com>
1985 PR tree-optimization/59597
1986 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1987 in file. Accept new argument REGISTERING and use it to modify
1988 dump output appropriately.
1989 (register_jump_thread): Corresponding changes.
1990 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1991 thread paths involving joiner blocks. Add code to dump cancelled
1992 jump threading paths.
1994 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1996 PR rtl-optimization/59477
1997 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1998 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2000 2014-01-22 Tom Tromey <tromey@redhat.com>
2002 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2004 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2006 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2008 PR rtl-optimization/59896
2009 * lra-constraints.c (process_alt_operands): Check unused note for
2010 matched operands of insn with no output reloads.
2012 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2014 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2015 (mips_move_from_gpr_cost): Likewise.
2017 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2019 PR rtl-optimization/59858
2020 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2021 ira_class_hard_regs_num.
2022 (process_alt_operands): Increase reject for dying matched operand.
2024 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2027 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2028 smaller than size, perform several stores or loads and stores
2029 at dst + count - size to store or copy all of size bytes, rather
2030 than just last modesize bytes.
2032 2014-01-20 DJ Delorie <dj@redhat.com>
2034 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2035 that CLOBBERs are REGs before propogating their values.
2037 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2040 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2041 (cgraph_inline_failed_type): New function.
2042 * cgraph.h (DEFCIFCODE): Add type.
2043 (cgraph_inline_failed_type_t): New enum.
2044 (cgraph_inline_failed_type): New prototype.
2045 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2046 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2047 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2048 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2049 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2050 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2051 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2052 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2053 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2054 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2055 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2056 OPTIMIZATION_MISMATCH.
2057 * tree-inline.c (expand_call_inline): Emit errors during
2058 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2060 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2063 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2064 mode attribute in insn output.
2066 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2068 * output.h (output_constant): Delete.
2069 * varasm.c (output_constant): Make private.
2071 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2073 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2075 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2078 * tree.h (fold_builtin_strcat): New prototype.
2079 * builtins.c (fold_builtin_strcat): No longer static. Add len
2080 argument, if non-NULL, don't call c_strlen. Optimize
2081 directly into __builtin_memcpy instead of __builtin_strcpy.
2082 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2083 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2085 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2087 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2088 for SImode_address_operand operands, having only a REG argument.
2090 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2092 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2093 loader name using mbig-endian.
2094 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2096 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2098 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2102 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2104 * config/aarch64/aarch64-protos.h
2105 (aarch64_cannot_change_mode_class_ptr): Declare.
2106 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2107 aarch64_cannot_change_mode_class_ptr): New.
2108 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2109 backend hook aarch64_cannot_change_mode_class.
2111 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2113 * common/config/aarch64/aarch64-common.c
2114 (aarch64_handle_option): Don't handle any option order logic here.
2115 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2116 selected_cpu, warn on architecture version mismatch.
2117 (aarch64_override_options): Fix parsing order for option strings.
2119 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2120 Iain Sandoe <iain@codesourcery.com>
2123 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2124 warning. Amend comment to reflect current functionality.
2126 2014-01-20 Richard Biener <rguenther@suse.de>
2129 * builtins.c (fold_builtin_strcat): Remove case better handled
2130 by tree-ssa-strlen.c.
2132 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2134 * config/aarch64/aarch64.opt
2135 (mcpu, march, mtune): Make case-insensitive.
2137 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2140 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2141 if operands[1] is a REG or ZERO_EXTEND of a REG.
2143 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2145 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2147 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2149 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2150 long non-pic millicode calls.
2152 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2154 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2156 2014-01-19 Kito Cheng <kito@0xlab.org>
2158 * builtins.c (expand_movstr): Check movstr expand done or fail.
2160 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2161 H.J. Lu <hongjiu.lu@intel.com>
2164 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2165 to DImode for zero-extended addresses.
2167 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2169 PR rtl-optimization/57763
2170 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2171 on the new indirect jump_insn and increment LABEL_NUSES (label).
2173 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2177 * config.gcc (x86_archs): New variable.
2178 (x86_64_archs): Likewise.
2179 (x86_cpus): Likewise.
2180 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2181 --with-arch/--with-cpu= options.
2182 Support --with-arch=/--with-cpu={nehalem,westmere,
2183 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2185 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2187 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2188 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2190 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2192 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2194 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2197 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2198 clear cpp_get_options (parse_in)->warn_unused_macros for
2199 ix86_target_macros_internal with cpp_define.
2201 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2203 * jump.c (delete_related_insns): Keep (use (insn))s.
2204 * reorg.c (redundant_insn): Check for barriers too.
2206 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2208 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2210 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2212 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2213 call to $$dyncall when TARGET_LONG_CALLS is true.
2215 2014-01-17 Jeff Law <law@redhat.com>
2217 * ree.c (combine_set_extension): Temporarily disable test for
2218 changing number of hard registers.
2220 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2223 * ipa-inline-analysis.c (inline_free_summary):
2224 Do not free summary of aliases.
2226 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2229 * gimplify.c (gimplify_expr): Use create_tmp_var
2230 instead of create_tmp_var_raw. If cond doesn't have
2231 integral type, don't add the IFN_ANNOTATE builtin at all.
2233 2014-01-17 Martin Jambor <mjambor@suse.cz>
2236 * ipa-cp.c (prev_edge_clone): New variable.
2237 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2238 Also resize prev_edge_clone vector.
2239 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2240 (ipcp_edge_removal_hook): New function.
2241 (ipcp_driver): Register ipcp_edge_removal_hook.
2243 2014-01-17 Andrew Pinski <apinski@cavium.com>
2244 Steve Ellcey <sellcey@mips.com>
2247 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2250 2014-01-17 Jeff Law <law@redhat.com>
2253 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2254 so that pass_ccp runs first.
2256 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2258 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2259 (ix86_adjust_cost): Use !TARGET_XXX.
2260 (do_reorder_for_imul): Likewise.
2261 (swap_top_of_ready_list): Likewise.
2262 (ix86_sched_reorder): Likewise.
2264 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2266 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2267 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2268 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2269 (intel_memset): New. Duplicate slm_memset.
2270 (intel_cost): New. Duplicate slm_cost.
2271 (m_INTEL): New macro.
2272 (processor_target_table): Add "intel".
2273 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2274 with PROCESSOR_INTEL for "intel".
2275 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2276 PROCESSOR_SILVERMONT.
2277 (ix86_issue_rate): Likewise.
2278 (ix86_adjust_cost): Likewise.
2279 (ia32_multipass_dfa_lookahead): Likewise.
2280 (swap_top_of_ready_list): Likewise.
2281 (ix86_sched_reorder): Likewise.
2282 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2283 instead of TARGET_OPT_AGU.
2284 * config/i386/i386.h (TARGET_INTEL): New.
2285 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2286 (processor_type): Add PROCESSOR_INTEL.
2287 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2288 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2290 2014-01-17 Marek Polacek <polacek@redhat.com>
2293 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2296 2014-01-17 Richard Biener <rguenther@suse.de>
2298 PR tree-optimization/46590
2299 * opts.c (default_options_table): Add entries for
2300 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2301 all enabled at -O1 but not for -Og.
2302 * common.opt (fbranch-count-reg): Remove Init(1).
2303 (fmove-loop-invariants): Likewise.
2304 (ftree-pta): Likewise.
2306 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2308 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2309 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2310 decls to at least the GCC 4.8 used alignments.
2313 * tree-nested.c (convert_nonlocal_reference_stmt,
2314 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2315 of GIMPLE_BIND stmts, adjust associated decls.
2317 2014-01-17 Richard Biener <rguenther@suse.de>
2319 PR tree-optimization/46590
2320 * vec.h (vec<>::bseach): New member function implementing
2321 binary search according to C89 bsearch.
2322 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2323 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2324 bitmap pointer again. Make accesses_in_loop a flat array.
2325 (mem_ref_obstack): New global.
2326 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2327 (mark_ref_stored): Likewise.
2328 (ref_indep_loop_p_2): Likewise.
2329 (set_ref_stored_in_loop): New helper function.
2330 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2331 (memref_free): Adjust.
2332 (record_mem_ref_loc): Simplify.
2333 (gather_mem_refs_stmt): Adjust.
2334 (sort_locs_in_loop_postorder_cmp): New function.
2335 (analyze_memory_references): Sort accesses_in_loop after
2336 loop postorder number.
2337 (find_ref_loc_in_loop_cmp): New function.
2338 (for_all_locs_in_loop): Find relevant cluster of locs in
2339 accesses_in_loop and iterate without recursion.
2340 (execute_sm): Avoid uninit warning.
2341 (struct ref_always_accessed): Simplify.
2342 (ref_always_accessed::operator ()): Likewise.
2343 (ref_always_accessed_p): Likewise.
2344 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2345 loop postorder numbers here.
2346 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2349 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2352 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2353 on decls for which assemble_alias has been called.
2355 2014-01-17 Nick Clifton <nickc@redhat.com>
2357 * config/msp430/msp430.opt: (mcpu): New option.
2358 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2359 (msp430_option_override): Parse target_cpu. If the MCU name
2360 matches a generic string, clear target_mcu.
2361 (msp430_attr): Allow numeric interrupt values up to 63.
2362 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2363 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2365 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2367 * config/msp430/msp430.md (popm): Use %J rather than %I.
2368 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2369 (addhi_cy_i): Use immediate_operand for operand 2.
2370 * doc/invoke.texi: Document -mcpu option.
2372 2014-01-17 Richard Biener <rguenther@suse.de>
2374 PR rtl-optimization/38518
2375 * df.h (df_analyze_loop): Declare.
2376 * df-core.c: Include cfgloop.h.
2377 (df_analyze_1): Split out main part of df_analyze.
2378 (df_analyze): Adjust.
2379 (loop_inverted_post_order_compute): New function.
2380 (loop_post_order_compute): Likewise.
2381 (df_analyze_loop): New function avoiding whole-function
2383 * loop-invariant.c (find_defs): Use df_analyze_loop.
2384 (find_invariants): Adjust.
2385 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2387 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2389 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2390 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2392 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2394 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2395 traversal when removing references.
2397 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2400 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2402 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2405 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2406 pushing a reload for an autoinc when we had previously reloaded an
2407 inner part of the address.
2409 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2411 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2413 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2414 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2415 when not giving up or versioning for alias only because of
2417 (vect_analyze_data_ref_dependences): Set to true.
2418 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2420 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2421 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2425 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2428 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2429 operand 0 predicate for gathers, use a new pseudo as subtarget.
2431 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2434 * lra-constraints.c (process_alt_operands): Add printing debug info.
2435 Check absence of input/output reloads for matched operands too.
2437 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2439 PR rtl-optimization/59835
2440 * ira.c (ira_init_register_move_cost): Increase cost for
2443 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2445 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2447 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2450 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2451 non-register objects. Use gen_(high/low)part more consistently.
2454 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2457 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2458 endian support, remove tests for WORDS_BIG_ENDIAN.
2459 (p8_mfvsrd_3_<mode>): Likewise.
2460 (reload_gpr_from_vsx<mode>): Likewise.
2461 (reload_gpr_from_vsxsf): Likewise.
2462 (p8_mfvsrd_4_disf): Likewise.
2464 2014-01-16 Richard Biener <rguenther@suse.de>
2466 PR rtl-optimization/46590
2467 * lcm.c (compute_antinout_edge): Use postorder iteration.
2468 (compute_laterin): Use inverted postorder iteration.
2470 2014-01-16 Nick Clifton <nickc@redhat.com>
2473 * varasm.c (output_constant): Return the number of bytes actually
2475 (output_constructor_array_range): Update the field size with the
2476 number of bytes emitted by output_constant.
2477 (output_constructor_regular_field): Likewise. Also do not
2478 complain if the total number of bytes emitted is now greater
2479 than the expected fieldpos.
2480 * output.h (output_constant): Update prototype and descriptive comment.
2482 2014-01-16 Marek Polacek <polacek@redhat.com>
2485 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2486 it is error_mark_node.
2488 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2490 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2491 VALID_AVX256_REG_OR_OI_MODE.
2493 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2495 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2496 current procedure should be profiled.
2498 2014-01-15 Andrew Pinski <apinski@cavium.com>
2500 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2501 of moving from/to the STACK_REG register class.
2503 2014-01-15 Richard Henderson <rth@redhat.com>
2506 * reginfo.c (global_regs_decl): Globalize.
2507 * rtl.h (global_regs_decl): Declare.
2508 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2509 reserved via global_regs.
2511 2014-01-15 Teresa Johnson <tejohnson@google.com>
2513 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2515 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2517 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2518 and vmulosh rather than call gen_vec_widen_smult_*.
2519 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2520 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2521 (vec_widen_smult_even_v16qi): Likewise.
2522 (vec_widen_umult_even_v8hi): Likewise.
2523 (vec_widen_smult_even_v8hi): Likewise.
2524 (vec_widen_umult_odd_v16qi): Likewise.
2525 (vec_widen_smult_odd_v16qi): Likewise.
2526 (vec_widen_umult_odd_v8hi): Likewise.
2527 (vec_widen_smult_odd_v8hi): Likewise.
2528 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2529 vmuloub rather than call gen_vec_widen_umult_*.
2530 (vec_widen_umult_lo_v16qi): Likewise.
2531 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2532 vmulosb rather than call gen_vec_widen_smult_*.
2533 (vec_widen_smult_lo_v16qi): Likewise.
2534 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2535 rather than call gen_vec_widen_umult_*.
2536 (vec_widen_umult_lo_v8hi): Likewise.
2537 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2538 rather than call gen_vec_widen_smult_*.
2539 (vec_widen_smult_lo_v8hi): Likewise.
2541 2014-01-15 Jeff Law <law@redhat.com>
2543 PR tree-optimization/59747
2544 * ree.c (find_and_remove_re): Properly handle case where a second
2545 eliminated extension requires widening a copy created for elimination
2546 of a prior extension.
2547 (combine_set_extension): Ensure that the number of hard regs needed
2548 for a destination register does not change when we widen it.
2550 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2552 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2553 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2554 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2555 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2556 (avr-*-rtems*): Likewise.
2557 (bfin*-rtems*): Likewise.
2558 (moxie-*-rtems*): Likewise.
2559 (h8300-*-rtems*): Likewise.
2560 (i[34567]86-*-rtems*): Likewise.
2561 (lm32-*-rtems*): Likewise.
2562 (m32r-*-rtems*): Likewise.
2563 (m68k-*-rtems*): Likewise.
2564 (microblaze*-*-rtems*): Likewise.
2565 (mips*-*-rtems*): Likewise.
2566 (powerpc-*-rtems*): Likewise.
2567 (sh-*-rtems*): Likewise.
2568 (sparc-*-rtems*): Likewise.
2569 (sparc64-*-rtems*): Likewise.
2570 (v850-*-rtems*): Likewise.
2571 (m32c-*-rtems*): Likewise.
2573 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2575 PR rtl-optimization/59511
2576 * ira.c (ira_init_register_move_cost): Use memory costs for some
2577 cases of register move cost calculations.
2578 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2579 instead of BB frequency.
2580 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2581 * lra-assigns.c (find_hard_regno_for): Ditto.
2583 2014-01-15 Richard Biener <rguenther@suse.de>
2585 PR tree-optimization/59822
2586 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2587 (vectorizable_load): Use it to hoist defs of uses of invariant
2588 loads out of the loop.
2590 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2591 Kugan Vivekanandarajah <kuganv@linaro.org>
2594 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2597 2014-01-15 Richard Biener <rguenther@suse.de>
2599 PR rtl-optimization/59802
2600 * lcm.c (compute_available): Use inverted postorder to seed
2601 the initial worklist.
2603 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2606 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2607 ADDR_REGS for invalid symrefs in non-PIC code.
2609 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2612 * builtins.c (determine_block_size): Initialize *probable_max_size
2613 even if len_rtx is CONST_INT.
2615 2014-01-14 Andrew Pinski <apinski@cavium.com>
2617 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2618 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2619 (cortexa53_tunings): Likewise.
2620 (aarch64_sched_issue_rate): New function.
2621 (TARGET_SCHED_ISSUE_RATE): Define.
2623 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2625 * ira-costs.c (find_costs_and_classes): Add missed
2626 ira_init_register_move_cost_if_necessary.
2628 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2631 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2633 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2636 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2637 to indicate if type is used for function return value. Warn ABI
2638 change if the vector mode isn't available for function return value.
2639 (ix86_function_arg_advance): Pass false to type_natural_mode.
2640 (ix86_function_arg): Likewise.
2641 (ix86_gimplify_va_arg): Likewise.
2642 (function_arg_32): Don't warn ABI change.
2643 (ix86_function_value): Pass true to type_natural_mode.
2644 (ix86_return_in_memory): Likewise.
2645 (ix86_struct_value_rtx): Removed.
2646 (TARGET_STRUCT_VALUE_RTX): Likewise.
2648 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2650 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2651 converting a conditional jump into a conditional return.
2653 2014-01-14 Richard Biener <rguenther@suse.de>
2655 PR tree-optimization/58921
2656 PR tree-optimization/59006
2657 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2658 hoisting invariant stmts.
2659 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2660 invariant loads on the preheader edge if possible.
2662 2014-01-14 Joey Ye <joey.ye@arm.com>
2664 * doc/plugin.texi (Building GCC plugins): Update to C++.
2666 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2668 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2669 (_mm_rcp28_round_ss): Ditto.
2670 (_mm_rsqrt28_round_sd): Ditto.
2671 (_mm_rsqrt28_round_ss): Ditto.
2672 (_mm_rcp28_sd): Ditto.
2673 (_mm_rcp28_ss): Ditto.
2674 (_mm_rsqrt28_sd): Ditto.
2675 (_mm_rsqrt28_ss): Ditto.
2676 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2677 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2678 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2679 (IX86_BUILTIN_RCP28SD): Ditto.
2680 (IX86_BUILTIN_RCP28SS): Ditto.
2681 (IX86_BUILTIN_RSQRT28SD): Ditto.
2682 (IX86_BUILTIN_RSQRT28SS): Ditto.
2683 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2684 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2685 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2686 (ix86_expand_special_args_builtin): Expand new FTYPE.
2687 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2688 (srcp14<mode>): Make insn unary.
2689 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2690 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2691 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2692 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2693 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2694 Fix rounding: make it SAE only.
2695 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2697 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2699 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2700 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2701 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2702 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2703 (round_saeonly_mask_scalar_operand4): Ditto.
2704 (round_saeonly_mask_scalar_op3): Ditto.
2705 (round_saeonly_mask_scalar_op4): Ditto.
2707 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2709 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2710 Implement -maltivec=be for vec_insert and vec_extract.
2712 2014-01-10 DJ Delorie <dj@redhat.com>
2714 * config/msp430/msp430.md (call_internal): Don't allow memory
2715 references with SP as the base register.
2716 (call_value_internal): Likewise.
2717 * config/msp430/constraints.md (Yc): New. For memory references
2718 that don't use SP as a base register.
2720 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2721 "an integer without a # prefix"
2722 * config/msp430/msp430.md (epilogue_helper): Use it.
2724 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2727 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2728 AVX512F gather builtins.
2729 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2730 on gather decls with INTEGER_TYPE masktype.
2731 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2732 directly into the builtin rather than hoisting it before loop.
2734 PR tree-optimization/59387
2735 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2736 (scev_const_prop): If folded_casts and type has undefined overflow,
2737 use force_gimple_operand instead of force_gimple_operand_gsi and
2738 for each added stmt if it is assign with
2739 arith_code_with_undefined_signed_overflow, call
2740 rewrite_to_defined_overflow.
2741 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2742 gimple-fold.h instead.
2743 (arith_code_with_undefined_signed_overflow,
2744 rewrite_to_defined_overflow): Moved to ...
2745 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2746 rewrite_to_defined_overflow): ... here. No longer static.
2747 Include gimplify-me.h.
2748 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2749 rewrite_to_defined_overflow): New prototypes.
2751 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2753 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2755 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2757 * builtins.c (get_object_alignment_2): Minor tweak.
2758 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2760 2014-01-13 Christian Bruel <christian.bruel@st.com>
2762 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2763 optimized non constant lengths.
2765 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2768 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2769 load as __atomic_load_N if possible.
2771 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2773 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2775 (rs6000_expand_builtin): Adjust call.
2777 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2780 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2781 * config/rs6000/rs6000.c: Include target-globals.h.
2782 (rs6000_set_current_function): Instead of doing target_reinit
2783 unconditionally, use save_target_globals_default_opts and
2784 restore_target_globals.
2786 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2788 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2789 (rs6000_expand_builtin): Handle mffs and mtfsf.
2790 (rs6000_init_builtins): Define mffs and mtfsf.
2791 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2792 (rs6000_mffs): New pattern.
2793 (rs6000_mtfsf): New pattern.
2795 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2797 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2798 Start narrowing with START. Apply candidate-use pair
2799 and check overall cost in narrowing.
2800 (iv_ca_prune): Pass new argument.
2802 2014-01-10 Jeff Law <law@redhat.com>
2805 * ree.c (combine_reaching_defs): Ensure the defining statement
2806 occurs before the extension when optimizing extensions with
2807 different source and destination hard registers.
2809 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2812 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2813 vtables into the type inheritance graph.
2815 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2817 PR rtl-optimization/59754
2818 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2819 modes in the REGNO != REGNO case.
2821 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2823 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2825 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2827 PR tree-optimization/59745
2828 * tree-predcom.c (tree_predictive_commoning_loop): Call
2829 free_affine_expand_cache if giving up because components is NULL.
2831 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2832 GC in payload of target_globals struct instead of allocating them on
2833 the heap and the larger structs separately using GC.
2834 * target-globals.h (struct target_globals): Make regs, hard_regs,
2835 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2836 of GTY((skip)) and change type to void *.
2837 (reset_target_globals): Cast loads from those fields to corresponding
2840 2014-01-10 Steve Ellcey <sellcey@mips.com>
2843 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2844 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2845 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2847 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2850 * aarch64-modes.def (CC_Zmode): New flags mode.
2851 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2852 represents an equality.
2853 (aarch64_get_condition_code): Handle CC_Zmode.
2854 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2856 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2858 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2859 extraction in good case.
2861 2014-01-10 Richard Biener <rguenther@suse.de>
2863 PR tree-optimization/59374
2864 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2865 checking after SLP discovery. Mark stmts not participating
2866 in any SLP instance properly.
2868 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2870 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2871 when handling a SET rtx.
2873 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2875 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2876 (cortex-a57): Likewise.
2877 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2879 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2881 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2882 non-iwmmxt builtins.
2884 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2888 * ipa-devirt.c record_target_from_binfo): Take as argument
2889 stack of binfos and lookup matching one for virtual inheritance.
2890 (possible_polymorphic_call_targets_1): Update.
2892 2014-01-10 Huacai Chen <chenhc@lemote.com>
2894 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2895 kernel strings for Loongson-2E/2F/3A.
2897 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2900 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2901 is_gimple_call before calling gimple_call_internal_p.
2903 2014-01-09 Steve Ellcey <sellcey@mips.com>
2905 * Makefile.in (TREE_FLOW_H): Remove.
2906 (TREE_SSA_H): Add file names from tree-flow.h.
2907 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2908 * tree.h: Remove tree-flow.h reference.
2909 * hash-table.h: Remove tree-flow.h reference.
2910 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2911 reference with tree-ssa-loop.h.
2913 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2915 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2916 default element-order behavior for -maltivec.
2917 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2918 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2919 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2920 when targeting big endian, at least for now.
2921 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2923 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2926 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2927 var satisfies use_register_for_decl, just take into account type
2928 alignment, rather than decl alignment.
2930 PR tree-optimization/59622
2931 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2932 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2933 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2934 Don't devirtualize for inplace at all. For targets.length () == 1,
2935 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2937 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2939 * config/i386/i386.md (cpu): Remove the unused btver1.
2941 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2943 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2945 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2948 * tree-core.h (struct target_globals): New forward declaration.
2949 (struct tree_target_option): Add globals field.
2950 * tree.h (TREE_TARGET_GLOBALS): Define.
2951 (prepare_target_option_nodes_for_pch): New prototype.
2952 * target-globals.h (struct target_globals): Define even if
2954 * tree.c (prepare_target_option_node_for_pch,
2955 prepare_target_option_nodes_for_pch): New functions.
2956 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2957 * config/i386/i386.c: Include target-globals.h.
2958 (ix86_set_current_function): Instead of doing target_reinit
2959 unconditionally, use save_target_globals_default_opts and
2960 restore_target_globals.
2962 2014-01-09 Richard Biener <rguenther@suse.de>
2964 PR tree-optimization/59715
2965 * tree-cfg.h (split_critical_edges): Declare.
2966 * tree-cfg.c (split_critical_edges): Export.
2967 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2969 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2971 * cfgexpand.c (expand_stack_vars): Optionally disable
2972 asan stack protection.
2973 (expand_used_vars): Likewise.
2974 (partition_stack_vars): Likewise.
2975 * asan.c (asan_emit_stack_protection): Optionally disable
2976 after return stack usage.
2977 (instrument_derefs): Optionally disable memory access instrumentation.
2978 (instrument_builtin_call): Likewise.
2979 (instrument_strlen_call): Likewise.
2980 (asan_protect_global): Optionally disable global variables protection.
2981 * doc/invoke.texi: Added doc for new options.
2982 * params.def: Added new options.
2983 * params.h: Likewise.
2985 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2987 PR rtl-optimization/59724
2988 * ifcvt.c (cond_exec_process_if_block): Don't call
2989 flow_find_head_matching_sequence with 0 longest_match.
2990 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2991 non-active insns if !stop_after.
2992 (try_head_merge_bb): Revert 2014-01-07 changes.
2994 2014-01-08 Jeff Law <law@redhat.com>
2996 * ree.c (get_sub_rtx): New function, extracted from...
2997 (merge_def_and_ext): Here.
2998 (combine_reaching_defs): Use get_sub_rtx.
3000 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3002 * cgraph.h (varpool_variable_node): Do not choke on null node.
3004 2014-01-08 Catherine Moore <clm@codesourcery.com>
3006 * config/mips/mips.md (simple_return): Attempt to use JRC
3008 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3010 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3012 PR rtl-optimization/59137
3013 * reorg.c (steal_delay_list_from_target): Call update_block for
3015 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3017 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3019 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3020 two duplicate entries.
3022 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3025 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3027 * config/mips/mips.c (mips_truncated_op_cost): New function.
3028 (mips_rtx_costs): Adjust test for BADDU.
3029 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3031 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3033 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3034 (*baddu_si): ...this new pattern.
3036 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3039 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3041 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3044 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3046 (expand_expr, expand_normal): Adjust.
3047 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3048 inner_reference_p. Use inner_reference_p to expand inner references.
3049 (store_expr): Adjust.
3050 * cfgexpand.c (expand_call_stmt): Adjust.
3052 2014-01-08 Rong Xu <xur@google.com>
3054 * gcov-io.c (gcov_var): Move from gcov-io.h.
3055 (gcov_position): Ditto.
3056 (gcov_is_error): Ditto.
3057 (gcov_rewrite): Ditto.
3058 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3059 only part to libgcc/libgcov.h.
3061 2014-01-08 Marek Polacek <polacek@redhat.com>
3064 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3066 2014-01-08 Marek Polacek <polacek@redhat.com>
3069 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3071 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3073 PR rtl-optimization/59649
3074 * stor-layout.c (get_mode_bounds): For BImode return
3075 0 and STORE_FLAG_VALUE.
3077 2014-01-08 Richard Biener <rguenther@suse.de>
3080 * gimple.h (is_gimple_builtin_call): Remove.
3081 (gimple_builtin_call_types_compatible_p): New.
3082 (gimple_call_builtin_p): New overload.
3083 * gimple.c (is_gimple_builtin_call): Remove.
3084 (validate_call): Rename to ...
3085 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3087 (validate_type): New static function.
3088 (gimple_call_builtin_p): New overload and adjust.
3089 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3090 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3091 (gimple_fold_stmt_to_constant_1): Likewise.
3092 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3094 2014-01-08 Richard Biener <rguenther@suse.de>
3097 * gimplify.c (gimplify_expr): Gimplify register-register type
3098 VIEW_CONVERT_EXPRs to separate stmts.
3100 2014-01-07 Jeff Law <law@redhat.com>
3103 * ree.c (combine_set_extension): Handle case where source
3104 and destination registers in an extension insn are different.
3105 (combine_reaching_defs): Allow source and destination registers
3106 in extension to be different under limited circumstances.
3107 (add_removable_extension): Remove restriction that the
3108 source and destination registers in the extension are the same.
3109 (find_and_remove_re): Emit a copy from the extension's
3110 destination to its source after the defining insn if
3111 the source and destination registers are different.
3114 * ifcvt.c (merge_if_block): If we are merging a block with more than
3115 one successor with a block with no successors, remove any BARRIER
3116 after the second block.
3118 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3120 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3122 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3125 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3126 for 14-bit register offsets when INT14_OK_STRICT is false.
3128 2014-01-07 Roland Stigge <stigge@antcom.de>
3129 Michael Meissner <meissner@linux.vnet.ibm.com>
3132 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3133 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3135 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3137 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3140 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3142 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3143 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3144 rtx is const0_rtx or not.
3146 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3149 * target-globals.c (save_target_globals): Remove this_fn_optab
3151 * toplev.c: Include optabs.h.
3152 (target_reinit): Temporarily restore the global options if another
3153 set of options are in force.
3155 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3157 PR rtl-optimization/58668
3158 * cfgcleanup.c (flow_find_cross_jump): Don't count
3159 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3160 to determine what is counted.
3161 (flow_find_head_matching_sequence): Use active_insn_p to determine
3163 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3165 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3166 determine what is counted.
3168 PR tree-optimization/59643
3169 * tree-predcom.c (split_data_refs_to_components): If one dr is
3170 read and one write, determine_offset fails and the write isn't
3171 in the bad component, just put the read into the bad component.
3173 2014-01-07 Mike Stump <mikestump@comcast.net>
3174 Jakub Jelinek <jakub@redhat.com>
3177 * tree-core.h (struct tree_optimization_option): Change optabs
3178 type from unsigned char * to void *.
3179 * optabs.c (init_tree_optimization_optabs): Adjust
3180 TREE_OPTIMIZATION_OPTABS initialization.
3182 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3185 * config/i386/i386.h (struct machine_function): Add
3186 no_drap_save_restore field.
3187 * config/i386/i386.c (ix86_save_reg): Use
3188 !cfun->machine->no_drap_save_restore instead of
3189 crtl->stack_realign_needed.
3190 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3191 this function clears frame_pointer_needed. Set
3192 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3193 and DRAP reg is needed.
3195 2014-01-06 Marek Polacek <polacek@redhat.com>
3198 * doc/implement-c.texi: Mention that other integer types are
3199 permitted as bit-field types in strictly conforming mode.
3201 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3203 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3206 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3208 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3210 2014-01-06 Martin Jambor <mjambor@suse.cz>
3213 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3215 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3217 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3221 * var-tracking.c (add_stores): Preserve the value of the source even if
3222 we don't record the store.
3224 2014-01-06 Terry Guo <terry.guo@arm.com>
3226 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3228 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3231 * config/darwin.c (darwin_function_section): Adjust return values to
3232 correspond to optimisation changes made in r206070.
3234 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3236 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3237 from prefetch_block tune setting.
3238 (nocona_cost): Correct size of prefetch block to 64.
3240 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3242 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3243 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3244 used to save the static chain register in the computation of the offset
3245 from which the FP registers need to be restored.
3247 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3249 PR tree-optimization/59519
3250 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3251 ICE if get_current_def (current_new_name) is already non-NULL, as long
3252 as it is a phi result of some other phi in *new_exit_bb that has
3255 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3256 or vmovdqu* for misaligned_operand.
3257 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3258 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3259 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3260 aligned_mem for AVX512F masked aligned load and store builtins and for
3263 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3265 PR tree-optimization/59651
3266 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3267 Address range for negative step should be added by TYPE_SIZE_UNIT.
3269 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3271 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3272 overlapping registers also for registers other than the stack pointer.
3274 2014-01-03 Marek Polacek <polacek@redhat.com>
3277 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3280 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3283 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3286 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3287 (push splitter): Use <P:MODE_SIZE> instead of
3288 GET_MODE_SIZE (<P:MODE>mode).
3289 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3290 (mov -1, reg peephole2): Likewise.
3291 * config/i386/sse.md (*mov<mode>_internal,
3292 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3293 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3294 *<code><mode>3, *andnot<mode>3<mask_name>,
3295 <mask_codefor><code><mode>3<mask_name>): Likewise.
3296 * config/i386/subst.md (mask_mode512bit_condition,
3297 sd_mask_mode512bit_condition): Likewise.
3299 2014-01-02 Xinliang David Li <davidxl@google.com>
3301 PR tree-optimization/59303
3302 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3303 (dump_predicates): Better output format.
3304 (pred_equal_p): New function.
3305 (is_neq_relop_p): Ditto.
3306 (is_neq_zero_form_p): Ditto.
3307 (pred_expr_equal_p): Ditto.
3308 (pred_neg_p): Ditto.
3309 (simplify_pred): Ditto.
3310 (simplify_preds_2): Ditto.
3311 (simplify_preds_3): Ditto.
3312 (simplify_preds_4): Ditto.
3313 (simplify_preds): Ditto.
3315 (push_to_worklist): Ditto.
3316 (get_pred_info_from_cmp): Ditto.
3317 (is_degenerated_phi): Ditto.
3318 (normalize_one_pred_1): Ditto.
3319 (normalize_one_pred): Ditto.
3320 (normalize_one_pred_chain): Ditto.
3321 (normalize_preds): Ditto.
3322 (normalize_cond_1): Remove function.
3323 (normalize_cond): Ditto.
3324 (is_gcond_subset_of): Ditto.
3325 (is_subset_of_any): Ditto.
3326 (is_or_set_subset_of): Ditto.
3327 (is_and_set_subset_of): Ditto.
3328 (is_norm_cond_subset_of): Ditto.
3329 (pred_chain_length_cmp): Ditto.
3330 (convert_control_dep_chain_into_preds): Type change.
3331 (find_predicates): Ditto.
3332 (find_def_preds): Ditto.
3333 (destroy_predicates_vecs): Ditto.
3334 (find_matching_predicates_in_rest_chains): Ditto.
3335 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3336 (is_pred_expr_subset): Ditto.
3337 (is_pred_chain_subset_of): Ditto.
3338 (is_included_in): Ditto.
3339 (is_superset_of): Ditto.
3341 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3343 Update copyright years.
3345 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3347 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3348 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3349 config/arc/arc.md, config/arc/arc.opt,
3350 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3351 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3352 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3353 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3354 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3355 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3356 vtable-verify.c, vtable-verify.h: Use the standard form for the
3359 2014-01-02 Tobias Burnus <burnus@net-b.de>
3361 * gcc.c (process_command): Update copyright notice dates.
3362 * gcov-dump.c: Ditto.
3364 * doc/cpp.texi: Bump @copying's copyright year.
3365 * doc/cppinternals.texi: Ditto.
3366 * doc/gcc.texi: Ditto.
3367 * doc/gccint.texi: Ditto.
3368 * doc/gcov.texi: Ditto.
3369 * doc/install.texi: Ditto.
3370 * doc/invoke.texi: Ditto.
3372 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3374 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3376 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3378 * config/i386/sse.md (*mov<mode>_internal): Guard
3379 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3381 PR rtl-optimization/59647
3382 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3383 new_rtx into UNSIGNED_FLOAT rtxes.
3385 Copyright (C) 2014 Free Software Foundation, Inc.
3387 Copying and distribution of this file, with or without modification,
3388 are permitted in any medium without royalty provided the copyright
3389 notice and this notice are preserved.