1 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
5 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
7 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
9 PR rtl-optimization/60116
10 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
11 other_insn once the combination has been validated.
13 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
16 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
18 * ipa-devirt.c: Include demangle.h
19 (odr_violation_reported): New static variable.
20 (add_type_duplicate): Update odr_violations.
21 (maybe_record_node): Add completep parameter; update it.
22 (record_target_from_binfo): Add COMPLETEP parameter;
24 (possible_polymorphic_call_targets_1): Likewise.
25 (struct polymorphic_call_target_d): Add nonconstruction_targets;
26 rename FINAL to COMPLETE.
27 (record_targets_from_bases): Sanity check we found the binfo;
28 fix COMPLETEP updating.
29 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
30 parameter, fix computing of COMPLETEP.
31 (dump_possible_polymorphic_call_targets): Imrove readability of dump; at
32 LTO time do demangling.
33 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
34 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
36 (gimple_get_virt_method_for_binfo): Likewise.
37 * gimple-fold.h (gimple_get_virt_method_for_binfo,
38 gimple_get_virt_method_for_vtable): Update prototypes.
40 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
43 * genautomata.c (add_presence_absence): Fix typo with
44 {final_}presence_list.
46 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
49 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
50 for VSX/Altivec vectors that land in GPR registers.
52 2014-02-11 Richard Henderson <rth@redhat.com>
53 Jakub Jelinek <jakub@redhat.com>
56 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
57 around drhs if type conversion to lacc->type is not useless.
59 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
61 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
63 (cortex-a57.cortex-a53): Likewise.
64 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
66 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
68 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
71 2014-02-11 Renlin Li <Renlin.Li@arm.com>
73 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
74 add_options_for_arm_vfp3.
76 2014-02-11 Jeff Law <law@redhat.com>
79 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
80 object with an undesirable mode.
82 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
85 * config/i386/sol2-9.h: New file.
86 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
87 *-*-solaris2.9*): Use it.
89 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
91 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
92 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
94 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
96 * config/microblaze/microblaze.c: Extend mcpu version format
98 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
100 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
102 2014-02-10 Richard Henderson <rth@redhat.com>
105 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
106 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
107 ms-abi vs -mno-accumulate-outgoing-args.
108 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
109 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
112 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
115 * cfgexpand.c (expand_asm_operands): Attach source location to
116 ASM_INPUT rtx objects.
117 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
119 2014-02-10 Nick Clifton <nickc@redhat.com>
121 * config/mn10300/mn10300.c (popcount): New function.
122 (mn10300_expand_prologue): Include saved registers in stack usage
125 2014-02-10 Jeff Law <law@redhat.com>
128 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
129 when changing the SET_DEST of a prior insn to avoid an input
132 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
134 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
135 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
136 -mcall-openbsd, or -mcall-linux.
137 (CC1_ENDIAN_BIG_SPEC): Remove.
138 (CC1_ENDIAN_LITTLE_SPEC): Remove.
139 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
140 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
141 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
142 and %cc1_endian_default.
143 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
145 2014-02-10 Richard Biener <rguenther@suse.de>
147 PR tree-optimization/60115
148 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
149 MEM_REF handling. Properly verify that the accesses are not
150 out of the objects bound.
152 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
154 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
157 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
159 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
160 proper constants and fix formatting.
161 (possible_polymorphic_call_targets): Fix formatting.
163 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
164 Ilya Tocar <ilya.tocar@intel.com>
166 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
167 (_mm512_loadu_epi32): Renamed into...
168 (_mm512_loadu_si512): This.
169 (_mm512_storeu_epi32): Renamed into...
170 (_mm512_storeu_si512): This.
171 (_mm512_maskz_ceil_ps): Removed.
172 (_mm512_maskz_ceil_pd): Ditto.
173 (_mm512_maskz_floor_ps): Ditto.
174 (_mm512_maskz_floor_pd): Ditto.
175 (_mm512_floor_round_ps): Ditto.
176 (_mm512_floor_round_pd): Ditto.
177 (_mm512_ceil_round_ps): Ditto.
178 (_mm512_ceil_round_pd): Ditto.
179 (_mm512_mask_floor_round_ps): Ditto.
180 (_mm512_mask_floor_round_pd): Ditto.
181 (_mm512_mask_ceil_round_ps): Ditto.
182 (_mm512_mask_ceil_round_pd): Ditto.
183 (_mm512_maskz_floor_round_ps): Ditto.
184 (_mm512_maskz_floor_round_pd): Ditto.
185 (_mm512_maskz_ceil_round_ps): Ditto.
186 (_mm512_maskz_ceil_round_pd): Ditto.
187 (_mm512_expand_pd): Ditto.
188 (_mm512_expand_ps): Ditto.
189 * config/i386/i386.c (ix86_builtins): Remove
190 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
192 * config/i386/predicates.md (const1256_operand): New.
193 (const_1_to_2_operand): Ditto.
194 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
195 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
196 (*avx512pf_gatherpf<mode>sf): Ditto.
197 (avx512pf_gatherpf<mode>df): Ditto.
198 (*avx512pf_gatherpf<mode>df_mask): Ditto.
199 (*avx512pf_gatherpf<mode>df): Ditto.
200 (avx512pf_scatterpf<mode>sf): Ditto.
201 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
202 (*avx512pf_scatterpf<mode>sf): Ditto.
203 (avx512pf_scatterpf<mode>df): Ditto.
204 (*avx512pf_scatterpf<mode>df_mask): Ditto.
205 (*avx512pf_scatterpf<mode>df): Ditto.
206 (avx512f_expand<mode>): Removed.
207 (<shift_insn><mode>3<mask_name>): Change predicate type.
209 2014-02-08 Jakub Jelinek <jakub@redhat.com>
211 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
212 not at the end of datarefs vector use ordered_remove to avoid
213 reordering datarefs vector.
216 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
217 mark local addressable non-static vars as GOVD_PRIVATE
218 instead of GOVD_LOCAL.
219 * omp-low.c (lower_omp_for): Move gimple_bind_vars
220 and BLOCK_VARS of gimple_bind_block to new_stmt rather
224 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
225 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
226 assume_aligned or alloc_align attributes.
227 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
228 arguments. Handle also assume_aligned and alloc_align attributes.
229 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
230 Handle calls to functions with assume_aligned or alloc_align
232 * doc/extend.texi: Document assume_aligned and alloc_align
235 2014-02-08 Terry Guo <terry.guo@arm.com>
237 * doc/invoke.texi: Document ARM -march=armv7e-m.
239 2014-02-08 Jakub Jelinek <jakub@redhat.com>
241 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
242 flag on __cilkrts_rethrow builtin.
245 * ipa-cp.c (determine_versionability): Fail at -O0
246 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
247 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
250 2014-02-04 Jakub Jelinek <jakub@redhat.com>
253 * tree-inline.c (copy_forbidden): Fail for
254 __attribute__((optimize (0))) functions.
256 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
258 * varpool.c: Include pointer-set.h.
259 (varpool_remove_unreferenced_decls): Variables in other partitions
260 will not be output; be however careful to not lose information
263 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
265 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
266 lookup in the vtable constructor.
268 2014-02-07 Jeff Law <law@redhat.com>
271 * config/m68k/m68k.md (ashldi_extsi): Turn into a
272 define_insn_and_split.
274 * ipa-inline.c (inline_small_functions): Fix typos.
276 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
278 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
279 (s390_can_use_return_insn): Declare.
280 * config/s390/s390.h (EPILOGUE_USES): Define.
281 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
283 (s390_chunkify_start): Handle return JUMP_LABELs.
284 (s390_early_mach): Emit a main_pool instruction on the entry edge.
285 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
286 (s390_can_use_return_insn): New functions.
287 (s390_fix_long_loop_prediction): Handle conditional returns.
288 (TARGET_SET_UP_BY_PROLOGUE): Define.
289 * config/s390/s390.md (ANY_RETURN): New code iterator.
290 (*creturn, *csimple_return, return, simple_return): New patterns.
292 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
294 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
295 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
296 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
297 REG_CFA_RESTORE list when deciding not to restore a register.
299 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
301 * config/s390/s390.c: Include tree-pass.h and context.h.
302 (s390_early_mach): New function, split out from...
303 (s390_emit_prologue): ...here.
304 (pass_data_s390_early_mach): New pass structure.
305 (pass_s390_early_mach): New class.
306 (s390_option_override): Create and register early_mach pass.
309 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
311 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
312 to match for the exit block.
314 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
316 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
317 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
318 Reject misaligned operands.
320 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
322 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
325 2014-02-07 Richard Biener <rguenther@suse.de>
328 * gimple-low.c (lower_builtin_posix_memalign): New function.
329 (lower_stmt): Call it to lower posix_memalign in a way
330 to make alignment info accessible.
332 2014-02-07 Jakub Jelinek <jakub@redhat.com>
335 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
336 __builtin_setjmp_receiver.
338 2014-02-07 Richard Biener <rguenther@suse.de>
341 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
342 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
343 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
344 Handle BUILT_IN_POSIX_MEMALIGN.
345 (find_func_clobbers): Likewise.
346 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
347 (call_may_clobber_ref_p_1): Likewise.
349 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
352 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
354 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
357 * lto-cgraph.c (lto_output_node): Use
358 symtab_get_symbol_partitioning_class.
359 (lto_output_varpool_node): likewise.
360 (symtab_get_symbol_partitioning_class): Move here from
362 * cgraph.h (symbol_partitioning_class): Likewise.
363 (symtab_get_symbol_partitioning_class): Declare.
365 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
367 * ggc.h (ggc_internal_cleared_alloc): New macro.
368 * vec.h (vec_safe_copy): Handle memory stats.
369 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
370 * target-globals.c (save_target_globals): Likewise.
372 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
375 * expr.c (emit_move_resolve_push): Export; be bit more selective
376 on when to clear alias set.
377 * expr.h (emit_move_resolve_push): Declare.
378 * function.h (struct function): Add tail_call_marked.
379 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
380 * config/i386/i386-protos.h (ix86_expand_push): Remove.
381 * config/i386/i386.md (TImode move expander): De not call
383 (FP push expanders): Preserve memory attributes.
384 * config/i386/sse.md (push<mode>1): Remove.
385 * config/i386/i386.c (ix86_expand_vector_move): Handle push
387 (ix86_expand_push): Remove.
388 * config/i386/mmx.md (push<mode>1): Remove.
390 2014-02-06 Jakub Jelinek <jakub@redhat.com>
392 PR rtl-optimization/60030
393 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
394 lopart with paradoxical subreg before shifting it up by hprec.
396 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
398 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
399 Remove extra newline at end of file.
400 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
401 (arm_issue_rate): Handle cortexa57.
402 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
403 (cortex-a57.cortex-a53): Likewise.
405 2014-02-06 Jakub Jelinek <jakub@redhat.com>
408 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
409 don't record in REG_FRAME_RELATED_EXPR registers not set in that
411 (arm_expand_prologue): Adjust all callers.
412 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
413 info, registers also at the lowest numbered registers side. Use
414 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
418 * var-tracking.c (adjust_mems): Before adding a SET
419 to amd->side_effects, adjust it's SET_SRC using
420 simplify_replace_fn_rtx.
422 2014-02-06 Alan Modra <amodra@gmail.com>
425 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
426 change SDmode to DDmode when lra_in_progress.
428 2014-02-06 Jakub Jelinek <jakub@redhat.com>
431 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
432 free_data_ref on the dr first, and before goto again also set dr
433 to the next dr. For simd_lane_access, free old datarefs[i] before
434 overwriting it. For get_vectype_for_scalar_type failure, don't
435 free_data_ref if simd_lane_access.
437 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
440 * tree.h (opts_for_fn): New inline function.
441 (opt_for_fn): Define.
442 * config/i386/i386.c (ix86_function_regparm): Use
443 opt_for_fn (decl, optimize) instead of optimize.
445 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
447 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
448 for SYMBOL_REF in large memory model.
450 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
452 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
454 (cortex-a57): Likewise.
455 (cortex-a57.cortex-a53): Likewise.
457 2014-02-06 Yury Gribov <y.gribov@samsung.com>
458 Kugan Vivekanandarajah <kuganv@linaro.org>
460 * config/arm/arm.c (arm_vector_alignment_reachable): Check
462 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
464 2014-02-06 Richard Biener <rguenther@suse.de>
466 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
467 set_loop_copy and initialize_original_copy_tables.
469 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
471 * config/aarch64/aarch64-simd.md
472 (aarch64_ashr_simddi): Change QI to SI.
474 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
475 Jakub Jelinek <jakub@redhat.com>
478 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
481 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
483 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
484 CODE_FOR_altivec_vpku[hw]um to
485 CODE_FOR_altivec_vpku[hw]um_direct.
486 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
487 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
488 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
489 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
491 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
493 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
494 generation for -maltivec=be.
495 (altivec_vsumsws): Simplify redundant test.
497 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
499 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
500 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
501 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
502 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
504 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
506 (altivec_vpks<VI_char>ss): Likewise.
507 (altivec_vpks<VI_char>us): Likewise.
508 (altivec_vpku<VI_char>us): Likewise.
509 (altivec_vpku<VI_char>um): Likewise.
510 (altivec_vpku<VI_char>um_direct): New (copy of
511 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
513 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
514 target is little endian and -maltivec=be is not specified.
515 (*altivec_vupkhs<VU_char>_direct): New (copy of
516 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
518 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
519 target is little endian and -maltivec=be is not specified.
520 (*altivec_vupkls<VU_char>_direct): New (copy of
521 altivec_vupkls<VU_char> that always emits vupkls*, for internal
523 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
524 little endian and -maltivec=be is not specified.
525 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
526 little endian and -maltivec=be is not specified.
528 2014-02-05 Richard Henderson <rth@redhat.com>
531 * combine-stack-adj.c: Revert r206943.
532 * sched-int.h (struct deps_desc): Add last_args_size.
533 * sched-deps.c (init_deps): Initialize it.
534 (sched_analyze_insn): Add OUTPUT dependencies between insns that
535 contain REG_ARGS_SIZE notes.
537 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
539 * lto-cgraph.c (asm_nodes_output): Make global.
540 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
541 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
543 (driver_handle_option): Handle OPT_fwpa.
545 2014-02-05 Jakub Jelinek <jakub@redhat.com>
548 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
549 a comment typo and formatting issue. If odr_hash hasn't been
550 created, return vNULL and set *completep to false.
553 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
554 bb with no successors.
556 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
559 * doc/invoke.texi (-march): Clarify documentation for ARM.
563 2014-02-05 Richard Biener <rguenther@suse.de>
565 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
566 when not vectorizing because of too many alias checks.
567 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
568 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
570 2014-02-05 Nick Clifton <nickc@redhat.com>
572 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
573 accept extended registers in any mode when compiling for the
576 2014-02-05 Yury Gribov <y.gribov@samsung.com>
578 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
579 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
580 sanitization attributes.
581 (can_inline_edge_p): Likewise.
582 (sanitize_attrs_match_for_inline_p): New function.
584 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
586 * ipa-prop.c (detect_type_change): Shor circuit testing of
587 type changes on THIS pointer.
589 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
592 * config/pa/pa.c (legitimize_tls_address): Return original address
593 if not passed a SYMBOL_REF rtx.
594 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
596 (pa_emit_move_sequence): Simplify TLS source operands.
597 (pa_legitimate_constant_p): Reject all TLS constants.
598 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
599 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
601 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
603 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
604 groups when we know they are controlled by LTO.
605 * varasm.c (default_binds_local_p_1): If object is in other partition,
606 it will be resolved locally.
608 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
610 * config/host-linux.c (linux_gt_pch_use_address): Don't
611 use SSIZE_MAX because it is not always defined.
613 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
616 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
617 threshold for pseudo splitting.
618 (update_ebb_live_info): Process call argument hard registers and
619 hard registers from insn definition too.
620 (max_small_class_regs_num): New constant.
621 (inherit_in_ebb): Update live hard regs through EBBs. Update
622 reloads_num only for small register classes. Don't split for
625 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
628 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
631 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
633 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
635 2014-02-04 Marek Polacek <polacek@redhat.com>
637 * gdbinit.in (pel): Define.
639 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
641 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
644 2014-02-04 Richard Biener <rguenther@suse.de>
647 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
648 in function context local.
649 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
650 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
651 similar to LTO_imported_decl_ref.
653 2014-02-04 Jakub Jelinek <jakub@redhat.com>
655 PR tree-optimization/60002
656 * cgraphclones.c (build_function_decl_skip_args): Clear
659 PR tree-optimization/60023
660 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
661 false to gsi_replace.
662 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
663 has been in some EH region and vec_stmt could throw, add
664 vec_stmt into the same EH region.
665 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
666 has no lhs, ignore it.
667 * internal-fn.c (expand_MASK_LOAD): Likewise.
670 * tree-inline.c (copy_forbidden): Fail for
671 __attribute__((optimize (0))) functions.
674 * omp-low.c (simd_clone_struct_copy): If from->inbranch
675 is set, copy one less argument.
676 (expand_simd_clones): Don't subtract clone_info->inbranch
677 from simd_clone_struct_alloc argument.
679 PR rtl-optimization/57915
680 * recog.c (simplify_while_replacing): If all unary/binary/relational
681 operation arguments are constant, attempt to simplify those.
684 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
685 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
687 2014-02-04 Richard Biener <rguenther@suse.de>
689 PR tree-optimization/60012
690 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
691 TBAA disambiguation to all DDRs.
693 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
696 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
697 (LINK_SPEC): Use it for -shared, -shared-libgcc.
699 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
702 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
704 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
706 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
707 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
709 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
712 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
713 to figure out targets of polymorphic calls with known decl.
714 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
715 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
716 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
717 (get_polymorphic_call_info): ... here.
718 (get_polymorphic_call_info_from_invariant): New function.
720 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
722 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
723 lookup via vtable pointer; check for type consistency
724 and turn inconsitent facts into UNREACHABLE.
725 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
726 * gimple-fold.c (gimple_get_virt_method_for_vtable):
727 Do not ICE on type inconsistent querries; return UNREACHABLE
730 2014-02-03 Richard Henderson <rth@twiddle.net>
733 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
734 already processed this node.
735 (normalize_one_pred_1): Pass along mark_set.
736 (normalize_one_pred): Create and destroy a pointer_set_t.
737 (normalize_one_pred_chain): Likewise.
739 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
741 PR gcov-profile/58602
742 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
744 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
747 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
748 Try to devirtualize by the knowledge of virtual table pointer given by
749 aggregate propagation.
750 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
751 ipa_print_node_jump_functions): Dump also offset that
752 is relevant for polymorphic calls.
753 (determine_known_aggregate_parts): Add arg_type parameter; use it
754 instead of determining the type from pointer type.
755 (ipa_compute_jump_functions_for_edge): Update call of
756 determine_known_aggregate_parts.
757 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
758 (gimple_get_virt_method_for_binfo): ... here; simplify using
759 vtable_pointer_value_to_vtable.
760 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
761 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
762 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
763 (vtable_pointer_value_to_vtable): Break out from ...; handle also
765 (vtable_pointer_value_to_binfo): ... here.
766 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
768 2014-02-03 Teresa Johnson <tejohnson@google.com>
770 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
771 redef of outer loop index variable.
773 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
777 * doc/extend.texi (Function Attributes): Typo.
779 2014-02-03 Cong Hou <congh@google.com>
781 PR tree-optimization/60000
782 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
783 if the vectorized statement is a store. A store statement can only
784 appear at the end of pattern statements.
786 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
788 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
789 (ix86_option_override_internal): Default long double to 64-bit for
790 32-bit Bionic and to 128-bit for 64-bit Bionic.
792 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
793 TARGET_LONG_DOUBLE_128 is true.
794 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
796 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
797 (mlong-double-64): Negate -mlong-double-128.
798 (mlong-double-128): New option.
800 * config/i386/i386-c.c (ix86_target_macros): Define
801 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
803 * doc/invoke.texi: Document -mlong-double-128.
805 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
807 PR rtl-optimization/60024
808 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
810 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
812 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
814 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
816 PR rtl-optimization/57662
817 * sel-sched.c (code_motion_path_driver): Do not mark already not
818 existing blocks in the visiting bitmap.
820 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
822 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
823 on the insn being emitted.
825 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
826 Will Deacon <will.deacon@arm.com>
828 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
830 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
832 * config/arm/arm-tables.opt: Regenerate.
834 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
836 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
837 for vector types other than V16QImode.
838 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
839 define_expand, and call altivec_expand_vec_perm_le when producing
840 code with little endian element order.
841 (*altivec_vperm_<mode>_internal): New insn having previous
842 behavior of altivec_vperm_<mode>.
843 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
844 altivec_expand_vec_perm_le when producing code with little endian
846 (*altivec_vperm_<mode>_uns_internal): New insn having previous
847 behavior of altivec_vperm_<mode>_uns.
849 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
851 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
852 (altivec_vsumsws): Add handling for -maltivec=be with a little
854 (altivec_vsumsws_direct): New.
855 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
858 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
860 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
861 vtable_pointer_value_to_binfo): New functions.
862 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
863 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
865 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
867 * config/nios2/nios2.md (load_got_register): Initialize GOT
868 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
869 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
871 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
873 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
874 preserverd by passthrough, do not propagate the type.
876 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
878 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
879 (mips_atomic_assign_expand_fenv): New function.
880 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
882 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
884 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
885 (__builtin_mips_set_fcsr): Likewise.
886 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
888 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
889 (mips16_expand_set_fcsr): Likewise.
890 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
891 (mips16_set_fcsr_stub): Likewise.
892 (mips16_get_fcsr_one_only_stub): New class.
893 (mips16_set_fcsr_one_only_stub): Likewise.
894 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
895 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
896 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
897 (hard_float): New availability predicate.
898 (mips_builtins): Add get_fcsr and set_fcsr.
899 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
900 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
901 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
902 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
903 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
906 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
908 * config/mips/mips.c (mips_one_only_stub): New class.
909 (mips_need_mips16_rdhwr_p): Replace with...
910 (mips16_rdhwr_stub): ...this new variable.
911 (mips16_stub_call_address): New function.
912 (mips16_rdhwr_one_only_stub): New class.
913 (mips_expand_thread_pointer): Use mips16_stub_call_address.
914 (mips_output_mips16_rdhwr): Delete.
915 (mips_finish_stub): New function.
916 (mips_code_end): Use it to handle rdhwr stubs.
918 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
921 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
922 when calculating size of integer atomic types.
924 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
926 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
928 2014-02-01 Jakub Jelinek <jakub@redhat.com>
930 PR tree-optimization/60003
931 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
932 * profile.c (branch_prob): Use gimple_call_builtin_p
933 to check for BUILT_IN_SETJMP_RECEIVER.
934 * tree-inline.c (copy_bb): Call notice_special_calls.
936 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
939 * lra-constraints.c (process_alt_operands): Update reload_sum only
942 2014-01-31 Richard Henderson <rth@redhat.com>
945 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
946 until after else_eh is processed.
948 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
950 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
951 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
952 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
953 in smmintrin.h, remove them.
954 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
955 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
956 * config/i386/i386.md (ROUND_SAE): Fix value.
957 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
958 (const48_operand): New.
959 * config/i386/subst.md (round), (round_expand): Use
960 const_4_or_8_to_11_operand.
961 (round_saeonly), (round_saeonly_expand): Use const48_operand.
963 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
965 * config/i386/constraints.md (Yk): Swap meaning with k.
966 * config/i386/i386.md (movhi_internal): Change Yk to k.
967 (movqi_internal): Ditto.
968 (*k<logic><mode>): Ditto.
971 (kandn<mode>): Ditto.
972 (*<code>hi_1): Ditto.
973 (*<code>qi_1): Ditto.
974 (kxnor<mode>): Ditto.
978 (*one_cmplhi2_1): Ditto.
979 (*one_cmplqi2_1): Ditto.
980 * config/i386/sse.md (): Change k to Yk.
981 (avx512f_load<mode>_mask): Ditto.
982 (avx512f_blendm<mode>): Ditto.
983 (avx512f_store<mode>_mask): Ditto.
984 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
985 (avx512f_storedqu<mode>_mask): Ditto.
986 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
988 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
989 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
990 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
991 (avx512f_maskcmp<mode>3): Ditto.
992 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
993 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
994 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
995 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
996 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
997 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
998 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
999 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1000 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1001 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1002 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1003 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1004 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1005 (vec_extract_lo_<mode>_maskm): Ditto.
1006 (vec_extract_hi_<mode>_maskm): Ditto.
1007 (avx512f_vternlog<mode>_mask): Ditto.
1008 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1009 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1010 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1011 (avx512f_<code>v8div16qi2_mask): Ditto.
1012 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1013 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1014 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1015 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1016 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1017 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1018 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1019 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1020 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1021 (avx512cd_maskb_vec_dupv8di): Ditto.
1022 (avx512cd_maskw_vec_dupv16si): Ditto.
1023 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1024 (avx512f_vpermi2var<mode>3_mask): Ditto.
1025 (avx512f_vpermi2var<mode>3_mask): Ditto.
1026 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1027 (*avx512f_gathersi<mode>): Ditto.
1028 (*avx512f_gathersi<mode>_2): Ditto.
1029 (*avx512f_gatherdi<mode>): Ditto.
1030 (*avx512f_gatherdi<mode>_2): Ditto.
1031 (*avx512f_scattersi<mode>): Ditto.
1032 (*avx512f_scatterdi<mode>): Ditto.
1033 (avx512f_compress<mode>_mask): Ditto.
1034 (avx512f_compressstore<mode>_mask): Ditto.
1035 (avx512f_expand<mode>_mask): Ditto.
1036 * config/i386/subst.md (mask): Change k to Yk.
1037 (mask_scalar_merge): Ditto.
1040 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1042 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1044 2014-01-31 Richard Biener <rguenther@suse.de>
1047 * builtins.c (fold_builtin_memory_op): Make sure to not
1048 use a floating-point mode or a boolean or enumeral type for
1051 2014-01-30 DJ Delorie <dj@redhat.com>
1053 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1054 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1055 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1056 whenever main() has an epilogue.
1058 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1060 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1061 unused variable "field".
1062 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1063 (vsx_mergeh_<mode>): Likewise.
1064 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1065 (altivec_vmrghh): Likewise.
1066 (altivec_vmrghw): Likewise.
1067 (altivec_vmrglb): Likewise.
1068 (altivec_vmrglh): Likewise.
1069 (altivec_vmrglw): Likewise.
1070 (altivec_vspltb): Add missing uses.
1071 (altivec_vsplth): Likewise.
1072 (altivec_vspltw): Likewise.
1073 (altivec_vspltsf): Likewise.
1075 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1078 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1079 frame related instructions.
1081 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1083 PR rtl-optimization/59959
1084 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1085 any reload of register whose subreg is invalid.
1087 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1089 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1090 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1091 Add missing return type - void.
1093 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1095 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1096 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1097 remove element index adjustment for endian (now handled in vsx.md
1099 (altivec_expand_vec_perm_const): Use
1100 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1101 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1102 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1103 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1104 define_expand and a new define_insn *altivec_vspltb_internal;
1105 adjust for -maltivec=be on a little endian target.
1106 (altivec_vspltb_direct): New.
1107 (altivec_vsplth): Divide into a define_expand and a new
1108 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1109 little endian target.
1110 (altivec_vsplth_direct): New.
1111 (altivec_vspltw): Divide into a define_expand and a new
1112 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1113 little endian target.
1114 (altivec_vspltw_direct): New.
1115 (altivec_vspltsf): Divide into a define_expand and a new
1116 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1117 a little endian target.
1119 2014-01-30 Richard Biener <rguenther@suse.de>
1121 PR tree-optimization/59993
1122 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1123 can propagate form the earlier stmt and avoid the transform
1124 when the intermediate result is needed.
1126 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1128 * README.Portability: Fix typo.
1130 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1132 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1133 comparison_operator with ordered_comparison_operator.
1135 2014-01-30 Nick Clifton <nickc@redhat.com>
1137 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1138 Rename to mn10300_store_multiple_regs.
1139 * config/mn10300/mn10300.c: Likewise.
1140 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1141 store_multiple_regs.
1142 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1143 Call mn10300_store_multiple_regs.
1145 2014-01-30 Nick Clifton <nickc@redhat.com>
1146 DJ Delorie <dj@redhat.com>
1148 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1149 %fp 2 to keep registers after it properly word-aligned.
1150 (rl78_alloc_physical_registers_umul): Handle the case where both
1151 input operands are the same.
1153 2014-01-30 Richard Biener <rguenther@suse.de>
1155 PR tree-optimization/59903
1156 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1159 2014-01-30 Jason Merrill <jason@redhat.com>
1162 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1165 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1167 2014-01-30 Richard Biener <rguenther@suse.de>
1169 PR tree-optimization/59951
1170 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1172 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1175 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1176 SFmode to DFmode case.
1178 2014-01-29 DJ Delorie <dj@redhat.com>
1180 * config/msp430/msp430.opt (-minrt): New.
1181 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1183 (ENDFILE_SPEC): Likewise.
1185 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1187 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1188 (estimate_function_body_sizes): Use it.
1190 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1193 * dwarf2out.c (is_cxx_auto): New.
1194 (is_base_type): Use it.
1195 (gen_type_die_with_usage): Likewise.
1197 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1199 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1200 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1201 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1202 -maltivec=be with LE targets.
1203 (vsx_mergeh_<mode>): Likewise.
1204 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1205 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1206 (altivec_vmrghb): Replace with define_expand and new
1207 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1208 (altivec_vmrghb_direct): New define_insn.
1209 (altivec_vmrghh): Replace with define_expand and new
1210 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1211 (altivec_vmrghh_direct): New define_insn.
1212 (altivec_vmrghw): Replace with define_expand and new
1213 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1214 (altivec_vmrghw_direct): New define_insn.
1215 (*altivec_vmrghsf): Adjust for endianness.
1216 (altivec_vmrglb): Replace with define_expand and new
1217 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1218 (altivec_vmrglb_direct): New define_insn.
1219 (altivec_vmrglh): Replace with define_expand and new
1220 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1221 (altivec_vmrglh_direct): New define_insn.
1222 (altivec_vmrglw): Replace with define_expand and new
1223 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1224 (altivec_vmrglw_direct): New define_insn.
1225 (*altivec_vmrglsf): Adjust for endianness.
1226 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1227 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1228 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1229 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1230 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1231 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1232 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1233 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1235 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1237 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1238 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1241 2014-01-29 Richard Biener <rguenther@suse.de>
1243 PR tree-optimization/58742
1244 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1245 associate_pointerplus_align.
1246 (associate_pointerplus_diff): New function.
1247 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1248 and associate_pointerplus_diff.
1250 2014-01-29 Richard Biener <rguenther@suse.de>
1252 * lto-streamer.h (LTO_major_version): Bump to 3.
1253 (LTO_minor_version): Reset to 0.
1255 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1257 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1258 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1259 (arm_file_start): Generate correct asm header for armv7ve.
1260 * config/arm/bpabi.h: Add multilib support for armv7ve.
1261 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1262 and cortex-a15 to armv7ve.
1263 * config/arm/t-aprofile: Add multilib support for armv7ve.
1264 * doc/invoke.texi: Document -march=armv7ve.
1266 2014-01-29 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/58742
1269 * tree-ssa-forwprop.c (associate_plusminus): Return true
1270 if we changed sth, defer EH cleanup to ...
1271 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1272 (simplify_mult): New function.
1274 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1277 PR tree-optimization/59920
1278 * tree.c (build_common_builtin_nodes): Remove
1279 __builtin_setjmp_dispatcher initialization.
1280 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1281 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1282 instead of gsi_after_labels + manually skipping debug stmts.
1283 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1284 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1285 * tree-inline.c (copy_edges_for_bb): Remove
1286 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1287 argument. Ignore computed_goto_p stmts. Don't call
1288 make_abnormal_goto_edges. If a call might need abnormal edges
1289 for non-local gotos, see if it already has an edge to
1290 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1291 with true argument, don't do anything then, otherwise add
1292 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1293 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1295 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1296 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1297 (lower_stmt): Don't set data->calls_builtin_setjmp.
1298 (lower_builtin_setjmp): Adjust comment.
1299 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1300 * tree-cfg.c (found_computed_goto): Remove.
1301 (factor_computed_gotos): Remove.
1302 (make_goto_expr_edges): Return bool, true for computed gotos.
1303 Don't call make_abnormal_goto_edges.
1304 (build_gimple_cfg): Don't set found_computed_goto, don't call
1305 factor_computed_gotos.
1306 (computed_goto_p): No longer static.
1307 (make_blocks): Don't set found_computed_goto.
1308 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1309 (make_edges): If make_goto_expr_edges returns true, push bb
1310 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1311 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1312 vector. Record mapping between bbs and OpenMP regions if there
1313 are any, adjust make_gimple_omp_edges caller. Call
1314 handle_abnormal_edges.
1315 (make_abnormal_goto_edges): Remove.
1316 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1317 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1318 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1319 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1320 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1321 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1322 filling *region also set *region_idx to (*region)->entry->index.
1325 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1326 For REGs set ORIGINAL_REGNO.
1328 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1330 * doc/md.texi: Mention that a target shouldn't implement
1331 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1334 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1336 PR tree-optimization/59594
1337 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1338 a copy of the datarefs vector rather than the vector itself.
1340 2014-01-28 Jason Merrill <jason@redhat.com>
1343 * dwarf2out.c (auto_die): New static.
1344 (gen_type_die_with_usage): Handle C++1y 'auto'.
1345 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1348 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1351 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1352 (SPEC_X32): Likewise.
1353 (SPEC_64): Likewise.
1354 * config/i386/i386.c (ix86_option_override_internal): Turn off
1355 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1357 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1358 * config/i386/i386.h (TARGET_16BIT): New macro.
1359 (TARGET_16BIT_P): Likewise.
1360 * config/i386/i386.opt: Add m16.
1361 * doc/invoke.texi: Document -m16.
1363 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1365 PR preprocessor/59935
1366 * input.c (location_get_source_line): Bail out on when line number
1367 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1369 2014-01-28 Richard Biener <rguenther@suse.de>
1371 PR tree-optimization/58742
1372 * tree-ssa-forwprop.c (associate_plusminus): Handle
1373 pointer subtraction of the form (T)(P + A) - (T)P.
1375 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1377 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1380 2014-01-28 Richard Biener <rguenther@suse.de>
1383 2014-01-28 Richard Biener <rguenther@suse.de>
1385 PR rtl-optimization/45364
1386 PR rtl-optimization/59890
1387 * var-tracking.c (local_get_addr_clear_given_value): Handle
1388 already cleared slot.
1389 (val_reset): Handle not allocated local_get_addr_cache.
1390 (vt_find_locations): Use post-order on the inverted CFG.
1392 2014-01-28 Richard Biener <rguenther@suse.de>
1394 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1396 2014-01-28 Richard Biener <rguenther@suse.de>
1398 PR rtl-optimization/45364
1399 PR rtl-optimization/59890
1400 * var-tracking.c (local_get_addr_clear_given_value): Handle
1401 already cleared slot.
1402 (val_reset): Handle not allocated local_get_addr_cache.
1403 (vt_find_locations): Use post-order on the inverted CFG.
1405 2014-01-28 Alan Modra <amodra@gmail.com>
1407 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1408 * configure.ac <recursive call for build != host>: Define
1409 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1410 and LD_FOR_BUILD too.
1411 * configure: Regenerate.
1413 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1415 * config/i386/i386.c (get_builtin_code_for_version): Separate
1416 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1417 Broadwell from Haswell.
1419 2014-01-27 Steve Ellcey <sellcey@mips.com>
1421 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1422 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1423 * config/mips/mips.c (mips_option_override): Change setting
1425 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1426 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1427 Change from Mask to Var.
1429 2014-01-27 Jeff Law <law@redhat.com>
1431 * ipa-inline.c (inline_small_functions): Fix typo.
1433 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1435 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1436 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1437 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1438 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1439 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1440 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1441 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1442 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1443 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1444 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1445 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1446 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1447 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1448 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1449 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1450 (_mm512_storeu_epi64): Ditto.
1451 (_mm512_cmpge_epi32_mask): Ditto.
1452 (_mm512_cmpge_epu32_mask): Ditto.
1453 (_mm512_cmpge_epi64_mask): Ditto.
1454 (_mm512_cmpge_epu64_mask): Ditto.
1455 (_mm512_cmple_epi32_mask): Ditto.
1456 (_mm512_cmple_epu32_mask): Ditto.
1457 (_mm512_cmple_epi64_mask): Ditto.
1458 (_mm512_cmple_epu64_mask): Ditto.
1459 (_mm512_cmplt_epi32_mask): Ditto.
1460 (_mm512_cmplt_epu32_mask): Ditto.
1461 (_mm512_cmplt_epi64_mask): Ditto.
1462 (_mm512_cmplt_epu64_mask): Ditto.
1463 (_mm512_cmpneq_epi32_mask): Ditto.
1464 (_mm512_cmpneq_epu32_mask): Ditto.
1465 (_mm512_cmpneq_epi64_mask): Ditto.
1466 (_mm512_cmpneq_epu64_mask): Ditto.
1467 (_mm512_expand_pd): Ditto.
1468 (_mm512_expand_ps): Ditto.
1469 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1470 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1471 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1472 * config/i386/i386.c (ix86_builtins): Add
1473 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1474 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1475 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1476 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1477 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1478 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1479 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1480 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1481 IX86_BUILTIN_PMOVUSQW512_MEM.
1482 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1483 __builtin_ia32_pmovsqd512mem_mask,
1484 __builtin_ia32_pmovqd512mem_mask,
1485 __builtin_ia32_pmovusqw512mem_mask,
1486 __builtin_ia32_pmovsqw512mem_mask,
1487 __builtin_ia32_pmovqw512mem_mask,
1488 __builtin_ia32_pmovusdw512mem_mask,
1489 __builtin_ia32_pmovsdw512mem_mask,
1490 __builtin_ia32_pmovdw512mem_mask,
1491 __builtin_ia32_pmovqb512mem_mask,
1492 __builtin_ia32_pmovusqb512mem_mask,
1493 __builtin_ia32_pmovsqb512mem_mask,
1494 __builtin_ia32_pmovusdb512mem_mask,
1495 __builtin_ia32_pmovsdb512mem_mask,
1496 __builtin_ia32_pmovdb512mem_mask.
1497 (bdesc_args): Add __builtin_ia32_expanddf512,
1498 __builtin_ia32_expandsf512.
1499 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1500 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1501 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1502 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1503 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1504 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1505 (avx512f_<code>v8div16qi2_mask_store): This.
1506 (avx512f_expand<mode>): New.
1508 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1510 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1512 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1513 (_mm512_prefetch_i32scatter_pd): Ditto.
1514 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1515 (_mm512_prefetch_i64scatter_pd): Ditto.
1516 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1517 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1518 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1519 (_mm512_prefetch_i32scatter_ps): Ditto.
1520 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1521 (_mm512_prefetch_i64scatter_ps): Ditto.
1522 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1523 * config/i386/i386-builtin-types.def: Define
1524 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1525 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1526 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1527 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1528 IX86_BUILTIN_SCATTERPFQPD.
1529 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1530 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1531 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1532 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1533 __builtin_ia32_scatterpfqps.
1534 (ix86_expand_builtin): Expand new built-ins.
1535 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1536 fix memory access data type.
1537 (*avx512pf_gatherpf<mode>_mask): Ditto.
1538 (*avx512pf_gatherpf<mode>): Ditto.
1539 (avx512pf_scatterpf<mode>): Ditto.
1540 (*avx512pf_scatterpf<mode>_mask): Ditto.
1541 (*avx512pf_scatterpf<mode>): Ditto.
1542 (GATHER_SCATTER_SF_MEM_MODE): New.
1543 (avx512pf_gatherpf<mode>df): Ditto.
1544 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1545 (*avx512pf_scatterpf<mode>df): Ditto.
1547 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1550 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1551 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1554 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1556 * common/config/arm/arm-common.c
1557 (arm_rewrite_mcpu): Handle multiple names.
1559 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1561 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1563 * gimple-builder.h (create_gimple_tmp): Delete.
1565 2014-01-27 Christian Bruel <christian.bruel@st.com>
1567 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1570 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1572 * config/pa/pa.md (call): Generate indirect long calls to non-local
1573 functions when outputing 32-bit code.
1574 (call_value): Likewise except for special call to buggy powf function.
1576 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1577 portable runtime and PIC indirect calls.
1578 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1579 and PIC call sequences. Use ldo instead of blr to set return register
1580 in PIC call sequence.
1582 2014-01-25 Walter Lee <walt@tilera.com>
1584 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1585 avoid clobbering a live register.
1587 2014-01-25 Walter Lee <walt@tilera.com>
1589 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1590 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1591 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1592 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1594 2014-01-25 Walter Lee <walt@tilera.com>
1596 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1597 arguments on even registers.
1598 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1600 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1601 (BIGGEST_ALIGNMENT): Ditto.
1602 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1604 2014-01-25 Walter Lee <walt@tilera.com>
1606 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1607 insns before bundling.
1608 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1610 2014-01-25 Walter Lee <walt@tilera.com>
1612 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1613 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1614 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1616 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1618 * config/mips/constraints.md (kl): Delete.
1619 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1620 define expands, using...
1621 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1622 instructions for MIPS16.
1623 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1624 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1626 2014-01-25 Walter Lee <walt@tilera.com>
1628 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1632 2014-01-25 Walter Lee <walt@tilera.com>
1634 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1635 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1637 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1639 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1642 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1644 * print-rtl.c (in_call_function_usage): New var.
1645 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1646 EXPR_LIST mode as mode and not as reg note name.
1649 * cfgloopmanip.c (copy_loop_info): If
1650 loop->warned_aggressive_loop_optimizations, make sure
1651 the flag is set in target loop too.
1653 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1655 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1657 * builtins.def: Likewise.
1658 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1659 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1660 * ira.c (ira_setup_eliminable_regset): Likewise.
1661 * omp-low.c (gate_expand_omp): Likewise.
1662 (execute_lower_omp): Likewise.
1663 (diagnose_sb_0): Likewise.
1664 (gate_diagnose_omp_blocks): Likewise.
1665 (simd_clone_clauses_extract): Likewise.
1668 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1670 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1671 correction for little endian...
1672 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1675 2014-01-24 Jeff Law <law@redhat.com>
1677 PR tree-optimization/59919
1678 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1679 for non-returning calls.
1681 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1683 * common/config/aarch64/aarch64-common.c
1684 (aarch64_rewrite_mcpu): Handle multiple names.
1685 * config/aarch64/aarch64.h
1686 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1688 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1690 * input.c (add_file_to_cache_tab): Handle the case where fopen
1693 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1696 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1697 from push operand if code of push isn't PRE_DEC.
1699 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1702 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1703 -mquad-memory-atomic. Update -mquad-memory documentation to say
1704 it is only used for non-atomic loads/stores.
1706 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1707 -mquad-memory or -mquad-memory-atomic switches.
1709 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1710 -mquad-memory-atomic to ISA 2.07 support.
1712 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1713 to separate support of normal quad word memory operations (ldq, stq)
1714 from the atomic quad word memory operations.
1716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1717 support to separate non-atomic quad word operations from atomic
1718 quad word operations. Disable non-atomic quad word operations in
1719 little endian mode so that we don't have to swap words after the
1720 load and before the store.
1721 (quad_load_store_p): Add comment about atomic quad word support.
1722 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1723 options printed with -mdebug=reg.
1725 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1726 -mquad-memory-atomic as the test for whether we have quad word
1727 atomic instructions.
1728 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1729 or -mp8-vector are used, allow byte/half-word atomic operations.
1731 * config/rs6000/sync.md (load_lockedti): Insure that the address
1732 is a proper indexed or indirect address for the lqarx instruction.
1733 On little endian systems, swap the hi/lo registers after the lqarx
1735 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1736 insure the address is valid for the lqarx instruction.
1737 (store_conditionalti): Insure that the address is a proper indexed
1738 or indirect address for the stqcrx. instruction. On little endian
1739 systems, swap the hi/lo registers before doing the stqcrx.
1741 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1742 insure the address is valid for the stqcrx. instruction.
1744 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1745 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1746 type of quad memory support is available.
1748 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1751 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1752 there is a danger of looping.
1754 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1756 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1757 force flag_ira_loop_pressure if set via command line.
1759 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1761 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1762 (ashr_simd): New builtin handling DI mode.
1763 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1764 (aarch64_sshr_simddi): New match pattern.
1765 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1766 (vshrd_n_s64): Likewise.
1767 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1769 2014-01-23 Nick Clifton <nickc@redhat.com>
1771 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1772 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1773 favour of mcu specific scripts.
1774 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1777 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1778 Alex Velenko <Alex.Velenko@arm.com>
1780 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1781 (vaddv_s16): Likewise.
1782 (vaddv_s32): Likewise.
1783 (vaddv_u8): Likewise.
1784 (vaddv_u16): Likewise.
1785 (vaddv_u32): Likewise.
1786 (vaddvq_s8): Likewise.
1787 (vaddvq_s16): Likewise.
1788 (vaddvq_s32): Likewise.
1789 (vaddvq_s64): Likewise.
1790 (vaddvq_u8): Likewise.
1791 (vaddvq_u16): Likewise.
1792 (vaddvq_u32): Likewise.
1793 (vaddvq_u64): Likewise.
1794 (vaddv_f32): Likewise.
1795 (vaddvq_f32): Likewise.
1796 (vaddvq_f64): Likewise.
1797 (vmaxv_f32): Likewise.
1798 (vmaxv_s8): Likewise.
1799 (vmaxv_s16): Likewise.
1800 (vmaxv_s32): Likewise.
1801 (vmaxv_u8): Likewise.
1802 (vmaxv_u16): Likewise.
1803 (vmaxv_u32): Likewise.
1804 (vmaxvq_f32): Likewise.
1805 (vmaxvq_f64): Likewise.
1806 (vmaxvq_s8): Likewise.
1807 (vmaxvq_s16): Likewise.
1808 (vmaxvq_s32): Likewise.
1809 (vmaxvq_u8): Likewise.
1810 (vmaxvq_u16): Likewise.
1811 (vmaxvq_u32): Likewise.
1812 (vmaxnmv_f32): Likewise.
1813 (vmaxnmvq_f32): Likewise.
1814 (vmaxnmvq_f64): Likewise.
1815 (vminv_f32): Likewise.
1816 (vminv_s8): Likewise.
1817 (vminv_s16): Likewise.
1818 (vminv_s32): Likewise.
1819 (vminv_u8): Likewise.
1820 (vminv_u16): Likewise.
1821 (vminv_u32): Likewise.
1822 (vminvq_f32): Likewise.
1823 (vminvq_f64): Likewise.
1824 (vminvq_s8): Likewise.
1825 (vminvq_s16): Likewise.
1826 (vminvq_s32): Likewise.
1827 (vminvq_u8): Likewise.
1828 (vminvq_u16): Likewise.
1829 (vminvq_u32): Likewise.
1830 (vminnmv_f32): Likewise.
1831 (vminnmvq_f32): Likewise.
1832 (vminnmvq_f64): Likewise.
1834 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1836 * config/aarch64/aarch64-simd.md
1837 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1838 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1839 (*aarch64_mul3_elt<mode>): Likewise.
1840 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1841 (*aarch64_mul3_elt_to_64v2df): Likewise.
1842 (*aarch64_mla_elt<mode>): Likewise.
1843 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1844 (*aarch64_mls_elt<mode>): Likewise.
1845 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1846 (*aarch64_fma4_elt<mode>): Likewise.
1847 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1848 (*aarch64_fma4_elt_to_64v2df): Likewise.
1849 (*aarch64_fnma4_elt<mode>): Likewise.
1850 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1851 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1852 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1853 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1854 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1855 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1856 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1857 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1858 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1860 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1862 * config/aarch64/aarch64-simd.md
1863 (aarch64_be_checked_get_lane<mode>): New define_expand.
1864 * config/aarch64/aarch64-simd-builtins.def
1865 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1866 New builtin definition.
1867 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1868 Use new safe be builtin.
1870 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1872 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1874 (aarch64_be_st1<mode>): Likewise.
1875 (aarch_ld1<VALL:mode>): Define_expand modified.
1876 (aarch_st1<VALL:mode>): Likewise.
1877 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1878 (UNSPEC_ST1): Likewise.
1880 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1882 * config/microblaze/microblaze.md: Add trap insn and attribute
1884 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1886 PR preprocessor/58580
1887 * input.h (location_get_source_line): Take an additional line_size
1889 (void diagnostics_file_cache_fini): Declare new function.
1890 * input.c (struct fcache): New type.
1891 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1892 New static constants.
1893 (diagnostic_file_cache_init, total_lines_num)
1894 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1895 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1896 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1897 (get_next_line, read_next_line, goto_next_line, read_line_num):
1898 New static function definitions.
1899 (diagnostic_file_cache_fini): New function.
1900 (location_get_source_line): Take an additional output line_len
1901 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1903 * diagnostic.c (diagnostic_finish): Call
1904 diagnostic_file_cache_fini.
1905 (adjust_line): Take an additional input parameter for the length
1906 of the line, rather than calculating it with strlen.
1907 (diagnostic_show_locus): Adjust the use of
1908 location_get_source_line and adjust_line with respect to their new
1909 signature. While displaying a line now, do not stop at the first
1910 null byte. Rather, display the zero byte as a space and keep
1911 going until we reach the size of the line.
1912 * Makefile.in: Add vec.o to OBJS-libcommon
1914 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1915 Ilya Tocar <ilya.tocar@intel.com>
1917 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1918 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1919 (__builtin_ia32_kmov16): Ditto.
1920 * config/i386/i386.md (UNSPEC_KMOV): New.
1923 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1925 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1926 (_mm512_storeu_si512): Ditto.
1928 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1931 * rtl.h (get_referenced_operands): Declare.
1932 * recog.c (get_referenced_operands): New function.
1933 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1934 operands have been referenced when recording LO_SUM references.
1936 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1938 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1940 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1942 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1943 Enable for generic and recent AMD targets.
1945 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1947 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1948 ARG_SIZE note when adjustment was eliminated.
1950 2014-01-22 Jeff Law <law@redhat.com>
1952 PR tree-optimization/59597
1953 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1954 in file. Accept new argument REGISTERING and use it to modify
1955 dump output appropriately.
1956 (register_jump_thread): Corresponding changes.
1957 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1958 thread paths involving joiner blocks. Add code to dump cancelled
1959 jump threading paths.
1961 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1963 PR rtl-optimization/59477
1964 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1965 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1967 2014-01-22 Tom Tromey <tromey@redhat.com>
1969 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1971 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1973 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1975 PR rtl-optimization/59896
1976 * lra-constraints.c (process_alt_operands): Check unused note for
1977 matched operands of insn with no output reloads.
1979 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1981 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1982 (mips_move_from_gpr_cost): Likewise.
1984 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1986 PR rtl-optimization/59858
1987 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1988 ira_class_hard_regs_num.
1989 (process_alt_operands): Increase reject for dying matched operand.
1991 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1994 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1995 smaller than size, perform several stores or loads and stores
1996 at dst + count - size to store or copy all of size bytes, rather
1997 than just last modesize bytes.
1999 2014-01-20 DJ Delorie <dj@redhat.com>
2001 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2002 that CLOBBERs are REGs before propogating their values.
2004 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2007 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2008 (cgraph_inline_failed_type): New function.
2009 * cgraph.h (DEFCIFCODE): Add type.
2010 (cgraph_inline_failed_type_t): New enum.
2011 (cgraph_inline_failed_type): New prototype.
2012 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2013 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2014 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2015 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2016 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2017 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2018 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2019 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2020 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2021 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2022 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2023 OPTIMIZATION_MISMATCH.
2024 * tree-inline.c (expand_call_inline): Emit errors during
2025 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2027 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2030 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2031 mode attribute in insn output.
2033 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2035 * output.h (output_constant): Delete.
2036 * varasm.c (output_constant): Make private.
2038 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2040 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2042 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2045 * tree.h (fold_builtin_strcat): New prototype.
2046 * builtins.c (fold_builtin_strcat): No longer static. Add len
2047 argument, if non-NULL, don't call c_strlen. Optimize
2048 directly into __builtin_memcpy instead of __builtin_strcpy.
2049 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2050 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2052 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2054 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2055 for SImode_address_operand operands, having only a REG argument.
2057 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2059 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2060 loader name using mbig-endian.
2061 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2063 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2065 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2069 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2071 * config/aarch64/aarch64-protos.h
2072 (aarch64_cannot_change_mode_class_ptr): Declare.
2073 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2074 aarch64_cannot_change_mode_class_ptr): New.
2075 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2076 backend hook aarch64_cannot_change_mode_class.
2078 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2080 * common/config/aarch64/aarch64-common.c
2081 (aarch64_handle_option): Don't handle any option order logic here.
2082 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2083 selected_cpu, warn on architecture version mismatch.
2084 (aarch64_override_options): Fix parsing order for option strings.
2086 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2087 Iain Sandoe <iain@codesourcery.com>
2090 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2091 warning. Amend comment to reflect current functionality.
2093 2014-01-20 Richard Biener <rguenther@suse.de>
2096 * builtins.c (fold_builtin_strcat): Remove case better handled
2097 by tree-ssa-strlen.c.
2099 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2101 * config/aarch64/aarch64.opt
2102 (mcpu, march, mtune): Make case-insensitive.
2104 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2107 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2108 if operands[1] is a REG or ZERO_EXTEND of a REG.
2110 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2112 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2114 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2116 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2117 long non-pic millicode calls.
2119 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2121 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2123 2014-01-19 Kito Cheng <kito@0xlab.org>
2125 * builtins.c (expand_movstr): Check movstr expand done or fail.
2127 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2128 H.J. Lu <hongjiu.lu@intel.com>
2131 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2132 to DImode for zero-extended addresses.
2134 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2136 PR rtl-optimization/57763
2137 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2138 on the new indirect jump_insn and increment LABEL_NUSES (label).
2140 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2144 * config.gcc (x86_archs): New variable.
2145 (x86_64_archs): Likewise.
2146 (x86_cpus): Likewise.
2147 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2148 --with-arch/--with-cpu= options.
2149 Support --with-arch=/--with-cpu={nehalem,westmere,
2150 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2152 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2154 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2155 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2157 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2159 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2161 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2164 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2165 clear cpp_get_options (parse_in)->warn_unused_macros for
2166 ix86_target_macros_internal with cpp_define.
2168 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2170 * jump.c (delete_related_insns): Keep (use (insn))s.
2171 * reorg.c (redundant_insn): Check for barriers too.
2173 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2175 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2177 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2179 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2180 call to $$dyncall when TARGET_LONG_CALLS is true.
2182 2014-01-17 Jeff Law <law@redhat.com>
2184 * ree.c (combine_set_extension): Temporarily disable test for
2185 changing number of hard registers.
2187 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2190 * ipa-inline-analysis.c (inline_free_summary):
2191 Do not free summary of aliases.
2193 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2196 * gimplify.c (gimplify_expr): Use create_tmp_var
2197 instead of create_tmp_var_raw. If cond doesn't have
2198 integral type, don't add the IFN_ANNOTATE builtin at all.
2200 2014-01-17 Martin Jambor <mjambor@suse.cz>
2203 * ipa-cp.c (prev_edge_clone): New variable.
2204 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2205 Also resize prev_edge_clone vector.
2206 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2207 (ipcp_edge_removal_hook): New function.
2208 (ipcp_driver): Register ipcp_edge_removal_hook.
2210 2014-01-17 Andrew Pinski <apinski@cavium.com>
2211 Steve Ellcey <sellcey@mips.com>
2214 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2217 2014-01-17 Jeff Law <law@redhat.com>
2220 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2221 so that pass_ccp runs first.
2223 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2225 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2226 (ix86_adjust_cost): Use !TARGET_XXX.
2227 (do_reorder_for_imul): Likewise.
2228 (swap_top_of_ready_list): Likewise.
2229 (ix86_sched_reorder): Likewise.
2231 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2233 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2234 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2235 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2236 (intel_memset): New. Duplicate slm_memset.
2237 (intel_cost): New. Duplicate slm_cost.
2238 (m_INTEL): New macro.
2239 (processor_target_table): Add "intel".
2240 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2241 with PROCESSOR_INTEL for "intel".
2242 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2243 PROCESSOR_SILVERMONT.
2244 (ix86_issue_rate): Likewise.
2245 (ix86_adjust_cost): Likewise.
2246 (ia32_multipass_dfa_lookahead): Likewise.
2247 (swap_top_of_ready_list): Likewise.
2248 (ix86_sched_reorder): Likewise.
2249 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2250 instead of TARGET_OPT_AGU.
2251 * config/i386/i386.h (TARGET_INTEL): New.
2252 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2253 (processor_type): Add PROCESSOR_INTEL.
2254 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2255 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2257 2014-01-17 Marek Polacek <polacek@redhat.com>
2260 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2263 2014-01-17 Richard Biener <rguenther@suse.de>
2265 PR tree-optimization/46590
2266 * opts.c (default_options_table): Add entries for
2267 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2268 all enabled at -O1 but not for -Og.
2269 * common.opt (fbranch-count-reg): Remove Init(1).
2270 (fmove-loop-invariants): Likewise.
2271 (ftree-pta): Likewise.
2273 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2275 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2276 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2277 decls to at least the GCC 4.8 used alignments.
2280 * tree-nested.c (convert_nonlocal_reference_stmt,
2281 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2282 of GIMPLE_BIND stmts, adjust associated decls.
2284 2014-01-17 Richard Biener <rguenther@suse.de>
2286 PR tree-optimization/46590
2287 * vec.h (vec<>::bseach): New member function implementing
2288 binary search according to C89 bsearch.
2289 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2290 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2291 bitmap pointer again. Make accesses_in_loop a flat array.
2292 (mem_ref_obstack): New global.
2293 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2294 (mark_ref_stored): Likewise.
2295 (ref_indep_loop_p_2): Likewise.
2296 (set_ref_stored_in_loop): New helper function.
2297 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2298 (memref_free): Adjust.
2299 (record_mem_ref_loc): Simplify.
2300 (gather_mem_refs_stmt): Adjust.
2301 (sort_locs_in_loop_postorder_cmp): New function.
2302 (analyze_memory_references): Sort accesses_in_loop after
2303 loop postorder number.
2304 (find_ref_loc_in_loop_cmp): New function.
2305 (for_all_locs_in_loop): Find relevant cluster of locs in
2306 accesses_in_loop and iterate without recursion.
2307 (execute_sm): Avoid uninit warning.
2308 (struct ref_always_accessed): Simplify.
2309 (ref_always_accessed::operator ()): Likewise.
2310 (ref_always_accessed_p): Likewise.
2311 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2312 loop postorder numbers here.
2313 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2316 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2319 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2320 on decls for which assemble_alias has been called.
2322 2014-01-17 Nick Clifton <nickc@redhat.com>
2324 * config/msp430/msp430.opt: (mcpu): New option.
2325 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2326 (msp430_option_override): Parse target_cpu. If the MCU name
2327 matches a generic string, clear target_mcu.
2328 (msp430_attr): Allow numeric interrupt values up to 63.
2329 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2330 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2332 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2334 * config/msp430/msp430.md (popm): Use %J rather than %I.
2335 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2336 (addhi_cy_i): Use immediate_operand for operand 2.
2337 * doc/invoke.texi: Document -mcpu option.
2339 2014-01-17 Richard Biener <rguenther@suse.de>
2341 PR rtl-optimization/38518
2342 * df.h (df_analyze_loop): Declare.
2343 * df-core.c: Include cfgloop.h.
2344 (df_analyze_1): Split out main part of df_analyze.
2345 (df_analyze): Adjust.
2346 (loop_inverted_post_order_compute): New function.
2347 (loop_post_order_compute): Likewise.
2348 (df_analyze_loop): New function avoiding whole-function
2350 * loop-invariant.c (find_defs): Use df_analyze_loop.
2351 (find_invariants): Adjust.
2352 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2354 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2356 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2357 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2359 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2361 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2362 traversal when removing references.
2364 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2367 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2369 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2372 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2373 pushing a reload for an autoinc when we had previously reloaded an
2374 inner part of the address.
2376 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2378 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2380 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2381 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2382 when not giving up or versioning for alias only because of
2384 (vect_analyze_data_ref_dependences): Set to true.
2385 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2387 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2388 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2392 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2395 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2396 operand 0 predicate for gathers, use a new pseudo as subtarget.
2398 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2401 * lra-constraints.c (process_alt_operands): Add printing debug info.
2402 Check absence of input/output reloads for matched operands too.
2404 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2406 PR rtl-optimization/59835
2407 * ira.c (ira_init_register_move_cost): Increase cost for
2410 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2412 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2414 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2417 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2418 non-register objects. Use gen_(high/low)part more consistently.
2421 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2424 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2425 endian support, remove tests for WORDS_BIG_ENDIAN.
2426 (p8_mfvsrd_3_<mode>): Likewise.
2427 (reload_gpr_from_vsx<mode>): Likewise.
2428 (reload_gpr_from_vsxsf): Likewise.
2429 (p8_mfvsrd_4_disf): Likewise.
2431 2014-01-16 Richard Biener <rguenther@suse.de>
2433 PR rtl-optimization/46590
2434 * lcm.c (compute_antinout_edge): Use postorder iteration.
2435 (compute_laterin): Use inverted postorder iteration.
2437 2014-01-16 Nick Clifton <nickc@redhat.com>
2440 * varasm.c (output_constant): Return the number of bytes actually
2442 (output_constructor_array_range): Update the field size with the
2443 number of bytes emitted by output_constant.
2444 (output_constructor_regular_field): Likewise. Also do not
2445 complain if the total number of bytes emitted is now greater
2446 than the expected fieldpos.
2447 * output.h (output_constant): Update prototype and descriptive comment.
2449 2014-01-16 Marek Polacek <polacek@redhat.com>
2452 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2453 it is error_mark_node.
2455 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2457 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2458 VALID_AVX256_REG_OR_OI_MODE.
2460 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2462 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2463 current procedure should be profiled.
2465 2014-01-15 Andrew Pinski <apinski@cavium.com>
2467 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2468 of moving from/to the STACK_REG register class.
2470 2014-01-15 Richard Henderson <rth@redhat.com>
2473 * reginfo.c (global_regs_decl): Globalize.
2474 * rtl.h (global_regs_decl): Declare.
2475 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2476 reserved via global_regs.
2478 2014-01-15 Teresa Johnson <tejohnson@google.com>
2480 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2482 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2484 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2485 and vmulosh rather than call gen_vec_widen_smult_*.
2486 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2487 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2488 (vec_widen_smult_even_v16qi): Likewise.
2489 (vec_widen_umult_even_v8hi): Likewise.
2490 (vec_widen_smult_even_v8hi): Likewise.
2491 (vec_widen_umult_odd_v16qi): Likewise.
2492 (vec_widen_smult_odd_v16qi): Likewise.
2493 (vec_widen_umult_odd_v8hi): Likewise.
2494 (vec_widen_smult_odd_v8hi): Likewise.
2495 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2496 vmuloub rather than call gen_vec_widen_umult_*.
2497 (vec_widen_umult_lo_v16qi): Likewise.
2498 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2499 vmulosb rather than call gen_vec_widen_smult_*.
2500 (vec_widen_smult_lo_v16qi): Likewise.
2501 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2502 rather than call gen_vec_widen_umult_*.
2503 (vec_widen_umult_lo_v8hi): Likewise.
2504 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2505 rather than call gen_vec_widen_smult_*.
2506 (vec_widen_smult_lo_v8hi): Likewise.
2508 2014-01-15 Jeff Law <law@redhat.com>
2510 PR tree-optimization/59747
2511 * ree.c (find_and_remove_re): Properly handle case where a second
2512 eliminated extension requires widening a copy created for elimination
2513 of a prior extension.
2514 (combine_set_extension): Ensure that the number of hard regs needed
2515 for a destination register does not change when we widen it.
2517 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2519 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2520 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2521 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2522 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2523 (avr-*-rtems*): Likewise.
2524 (bfin*-rtems*): Likewise.
2525 (moxie-*-rtems*): Likewise.
2526 (h8300-*-rtems*): Likewise.
2527 (i[34567]86-*-rtems*): Likewise.
2528 (lm32-*-rtems*): Likewise.
2529 (m32r-*-rtems*): Likewise.
2530 (m68k-*-rtems*): Likewise.
2531 (microblaze*-*-rtems*): Likewise.
2532 (mips*-*-rtems*): Likewise.
2533 (powerpc-*-rtems*): Likewise.
2534 (sh-*-rtems*): Likewise.
2535 (sparc-*-rtems*): Likewise.
2536 (sparc64-*-rtems*): Likewise.
2537 (v850-*-rtems*): Likewise.
2538 (m32c-*-rtems*): Likewise.
2540 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2542 PR rtl-optimization/59511
2543 * ira.c (ira_init_register_move_cost): Use memory costs for some
2544 cases of register move cost calculations.
2545 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2546 instead of BB frequency.
2547 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2548 * lra-assigns.c (find_hard_regno_for): Ditto.
2550 2014-01-15 Richard Biener <rguenther@suse.de>
2552 PR tree-optimization/59822
2553 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2554 (vectorizable_load): Use it to hoist defs of uses of invariant
2555 loads out of the loop.
2557 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2558 Kugan Vivekanandarajah <kuganv@linaro.org>
2561 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2564 2014-01-15 Richard Biener <rguenther@suse.de>
2566 PR rtl-optimization/59802
2567 * lcm.c (compute_available): Use inverted postorder to seed
2568 the initial worklist.
2570 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2573 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2574 ADDR_REGS for invalid symrefs in non-PIC code.
2576 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2579 * builtins.c (determine_block_size): Initialize *probable_max_size
2580 even if len_rtx is CONST_INT.
2582 2014-01-14 Andrew Pinski <apinski@cavium.com>
2584 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2585 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2586 (cortexa53_tunings): Likewise.
2587 (aarch64_sched_issue_rate): New function.
2588 (TARGET_SCHED_ISSUE_RATE): Define.
2590 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2592 * ira-costs.c (find_costs_and_classes): Add missed
2593 ira_init_register_move_cost_if_necessary.
2595 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2598 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2600 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2603 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2604 to indicate if type is used for function return value. Warn ABI
2605 change if the vector mode isn't available for function return value.
2606 (ix86_function_arg_advance): Pass false to type_natural_mode.
2607 (ix86_function_arg): Likewise.
2608 (ix86_gimplify_va_arg): Likewise.
2609 (function_arg_32): Don't warn ABI change.
2610 (ix86_function_value): Pass true to type_natural_mode.
2611 (ix86_return_in_memory): Likewise.
2612 (ix86_struct_value_rtx): Removed.
2613 (TARGET_STRUCT_VALUE_RTX): Likewise.
2615 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2617 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2618 converting a conditional jump into a conditional return.
2620 2014-01-14 Richard Biener <rguenther@suse.de>
2622 PR tree-optimization/58921
2623 PR tree-optimization/59006
2624 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2625 hoisting invariant stmts.
2626 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2627 invariant loads on the preheader edge if possible.
2629 2014-01-14 Joey Ye <joey.ye@arm.com>
2631 * doc/plugin.texi (Building GCC plugins): Update to C++.
2633 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2635 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2636 (_mm_rcp28_round_ss): Ditto.
2637 (_mm_rsqrt28_round_sd): Ditto.
2638 (_mm_rsqrt28_round_ss): Ditto.
2639 (_mm_rcp28_sd): Ditto.
2640 (_mm_rcp28_ss): Ditto.
2641 (_mm_rsqrt28_sd): Ditto.
2642 (_mm_rsqrt28_ss): Ditto.
2643 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2644 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2645 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2646 (IX86_BUILTIN_RCP28SD): Ditto.
2647 (IX86_BUILTIN_RCP28SS): Ditto.
2648 (IX86_BUILTIN_RSQRT28SD): Ditto.
2649 (IX86_BUILTIN_RSQRT28SS): Ditto.
2650 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2651 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2652 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2653 (ix86_expand_special_args_builtin): Expand new FTYPE.
2654 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2655 (srcp14<mode>): Make insn unary.
2656 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2657 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2658 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2659 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2660 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2661 Fix rounding: make it SAE only.
2662 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2664 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2666 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2667 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2668 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2669 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2670 (round_saeonly_mask_scalar_operand4): Ditto.
2671 (round_saeonly_mask_scalar_op3): Ditto.
2672 (round_saeonly_mask_scalar_op4): Ditto.
2674 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2676 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2677 Implement -maltivec=be for vec_insert and vec_extract.
2679 2014-01-10 DJ Delorie <dj@redhat.com>
2681 * config/msp430/msp430.md (call_internal): Don't allow memory
2682 references with SP as the base register.
2683 (call_value_internal): Likewise.
2684 * config/msp430/constraints.md (Yc): New. For memory references
2685 that don't use SP as a base register.
2687 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2688 "an integer without a # prefix"
2689 * config/msp430/msp430.md (epilogue_helper): Use it.
2691 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2694 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2695 AVX512F gather builtins.
2696 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2697 on gather decls with INTEGER_TYPE masktype.
2698 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2699 directly into the builtin rather than hoisting it before loop.
2701 PR tree-optimization/59387
2702 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2703 (scev_const_prop): If folded_casts and type has undefined overflow,
2704 use force_gimple_operand instead of force_gimple_operand_gsi and
2705 for each added stmt if it is assign with
2706 arith_code_with_undefined_signed_overflow, call
2707 rewrite_to_defined_overflow.
2708 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2709 gimple-fold.h instead.
2710 (arith_code_with_undefined_signed_overflow,
2711 rewrite_to_defined_overflow): Moved to ...
2712 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2713 rewrite_to_defined_overflow): ... here. No longer static.
2714 Include gimplify-me.h.
2715 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2716 rewrite_to_defined_overflow): New prototypes.
2718 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2720 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2722 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2724 * builtins.c (get_object_alignment_2): Minor tweak.
2725 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2727 2014-01-13 Christian Bruel <christian.bruel@st.com>
2729 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2730 optimized non constant lengths.
2732 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2735 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2736 load as __atomic_load_N if possible.
2738 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2740 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2742 (rs6000_expand_builtin): Adjust call.
2744 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2747 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2748 * config/rs6000/rs6000.c: Include target-globals.h.
2749 (rs6000_set_current_function): Instead of doing target_reinit
2750 unconditionally, use save_target_globals_default_opts and
2751 restore_target_globals.
2753 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2755 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2756 (rs6000_expand_builtin): Handle mffs and mtfsf.
2757 (rs6000_init_builtins): Define mffs and mtfsf.
2758 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2759 (rs6000_mffs): New pattern.
2760 (rs6000_mtfsf): New pattern.
2762 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2764 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2765 Start narrowing with START. Apply candidate-use pair
2766 and check overall cost in narrowing.
2767 (iv_ca_prune): Pass new argument.
2769 2014-01-10 Jeff Law <law@redhat.com>
2772 * ree.c (combine_reaching_defs): Ensure the defining statement
2773 occurs before the extension when optimizing extensions with
2774 different source and destination hard registers.
2776 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2779 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2780 vtables into the type inheritance graph.
2782 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2784 PR rtl-optimization/59754
2785 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2786 modes in the REGNO != REGNO case.
2788 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2790 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2792 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2794 PR tree-optimization/59745
2795 * tree-predcom.c (tree_predictive_commoning_loop): Call
2796 free_affine_expand_cache if giving up because components is NULL.
2798 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2799 GC in payload of target_globals struct instead of allocating them on
2800 the heap and the larger structs separately using GC.
2801 * target-globals.h (struct target_globals): Make regs, hard_regs,
2802 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2803 of GTY((skip)) and change type to void *.
2804 (reset_target_globals): Cast loads from those fields to corresponding
2807 2014-01-10 Steve Ellcey <sellcey@mips.com>
2810 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2811 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2812 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2814 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2817 * aarch64-modes.def (CC_Zmode): New flags mode.
2818 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2819 represents an equality.
2820 (aarch64_get_condition_code): Handle CC_Zmode.
2821 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2823 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2825 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2826 extraction in good case.
2828 2014-01-10 Richard Biener <rguenther@suse.de>
2830 PR tree-optimization/59374
2831 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2832 checking after SLP discovery. Mark stmts not participating
2833 in any SLP instance properly.
2835 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2837 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2838 when handling a SET rtx.
2840 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2842 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2843 (cortex-a57): Likewise.
2844 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2846 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2848 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2849 non-iwmmxt builtins.
2851 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2855 * ipa-devirt.c record_target_from_binfo): Take as argument
2856 stack of binfos and lookup matching one for virtual inheritance.
2857 (possible_polymorphic_call_targets_1): Update.
2859 2014-01-10 Huacai Chen <chenhc@lemote.com>
2861 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2862 kernel strings for Loongson-2E/2F/3A.
2864 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2867 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2868 is_gimple_call before calling gimple_call_internal_p.
2870 2014-01-09 Steve Ellcey <sellcey@mips.com>
2872 * Makefile.in (TREE_FLOW_H): Remove.
2873 (TREE_SSA_H): Add file names from tree-flow.h.
2874 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2875 * tree.h: Remove tree-flow.h reference.
2876 * hash-table.h: Remove tree-flow.h reference.
2877 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2878 reference with tree-ssa-loop.h.
2880 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2882 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2883 default element-order behavior for -maltivec.
2884 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2885 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2886 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2887 when targeting big endian, at least for now.
2888 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2890 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2893 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2894 var satisfies use_register_for_decl, just take into account type
2895 alignment, rather than decl alignment.
2897 PR tree-optimization/59622
2898 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2899 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2900 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2901 Don't devirtualize for inplace at all. For targets.length () == 1,
2902 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2904 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2906 * config/i386/i386.md (cpu): Remove the unused btver1.
2908 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2910 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2912 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2915 * tree-core.h (struct target_globals): New forward declaration.
2916 (struct tree_target_option): Add globals field.
2917 * tree.h (TREE_TARGET_GLOBALS): Define.
2918 (prepare_target_option_nodes_for_pch): New prototype.
2919 * target-globals.h (struct target_globals): Define even if
2921 * tree.c (prepare_target_option_node_for_pch,
2922 prepare_target_option_nodes_for_pch): New functions.
2923 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2924 * config/i386/i386.c: Include target-globals.h.
2925 (ix86_set_current_function): Instead of doing target_reinit
2926 unconditionally, use save_target_globals_default_opts and
2927 restore_target_globals.
2929 2014-01-09 Richard Biener <rguenther@suse.de>
2931 PR tree-optimization/59715
2932 * tree-cfg.h (split_critical_edges): Declare.
2933 * tree-cfg.c (split_critical_edges): Export.
2934 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2936 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2938 * cfgexpand.c (expand_stack_vars): Optionally disable
2939 asan stack protection.
2940 (expand_used_vars): Likewise.
2941 (partition_stack_vars): Likewise.
2942 * asan.c (asan_emit_stack_protection): Optionally disable
2943 after return stack usage.
2944 (instrument_derefs): Optionally disable memory access instrumentation.
2945 (instrument_builtin_call): Likewise.
2946 (instrument_strlen_call): Likewise.
2947 (asan_protect_global): Optionally disable global variables protection.
2948 * doc/invoke.texi: Added doc for new options.
2949 * params.def: Added new options.
2950 * params.h: Likewise.
2952 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2954 PR rtl-optimization/59724
2955 * ifcvt.c (cond_exec_process_if_block): Don't call
2956 flow_find_head_matching_sequence with 0 longest_match.
2957 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2958 non-active insns if !stop_after.
2959 (try_head_merge_bb): Revert 2014-01-07 changes.
2961 2014-01-08 Jeff Law <law@redhat.com>
2963 * ree.c (get_sub_rtx): New function, extracted from...
2964 (merge_def_and_ext): Here.
2965 (combine_reaching_defs): Use get_sub_rtx.
2967 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2969 * cgraph.h (varpool_variable_node): Do not choke on null node.
2971 2014-01-08 Catherine Moore <clm@codesourcery.com>
2973 * config/mips/mips.md (simple_return): Attempt to use JRC
2975 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2977 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2979 PR rtl-optimization/59137
2980 * reorg.c (steal_delay_list_from_target): Call update_block for
2982 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2984 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2986 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2987 two duplicate entries.
2989 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2992 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2994 * config/mips/mips.c (mips_truncated_op_cost): New function.
2995 (mips_rtx_costs): Adjust test for BADDU.
2996 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2998 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3000 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3001 (*baddu_si): ...this new pattern.
3003 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3006 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3008 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3011 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3013 (expand_expr, expand_normal): Adjust.
3014 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3015 inner_reference_p. Use inner_reference_p to expand inner references.
3016 (store_expr): Adjust.
3017 * cfgexpand.c (expand_call_stmt): Adjust.
3019 2014-01-08 Rong Xu <xur@google.com>
3021 * gcov-io.c (gcov_var): Move from gcov-io.h.
3022 (gcov_position): Ditto.
3023 (gcov_is_error): Ditto.
3024 (gcov_rewrite): Ditto.
3025 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3026 only part to libgcc/libgcov.h.
3028 2014-01-08 Marek Polacek <polacek@redhat.com>
3031 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3033 2014-01-08 Marek Polacek <polacek@redhat.com>
3036 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3038 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3040 PR rtl-optimization/59649
3041 * stor-layout.c (get_mode_bounds): For BImode return
3042 0 and STORE_FLAG_VALUE.
3044 2014-01-08 Richard Biener <rguenther@suse.de>
3047 * gimple.h (is_gimple_builtin_call): Remove.
3048 (gimple_builtin_call_types_compatible_p): New.
3049 (gimple_call_builtin_p): New overload.
3050 * gimple.c (is_gimple_builtin_call): Remove.
3051 (validate_call): Rename to ...
3052 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3054 (validate_type): New static function.
3055 (gimple_call_builtin_p): New overload and adjust.
3056 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3057 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3058 (gimple_fold_stmt_to_constant_1): Likewise.
3059 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3061 2014-01-08 Richard Biener <rguenther@suse.de>
3064 * gimplify.c (gimplify_expr): Gimplify register-register type
3065 VIEW_CONVERT_EXPRs to separate stmts.
3067 2014-01-07 Jeff Law <law@redhat.com>
3070 * ree.c (combine_set_extension): Handle case where source
3071 and destination registers in an extension insn are different.
3072 (combine_reaching_defs): Allow source and destination registers
3073 in extension to be different under limited circumstances.
3074 (add_removable_extension): Remove restriction that the
3075 source and destination registers in the extension are the same.
3076 (find_and_remove_re): Emit a copy from the extension's
3077 destination to its source after the defining insn if
3078 the source and destination registers are different.
3081 * ifcvt.c (merge_if_block): If we are merging a block with more than
3082 one successor with a block with no successors, remove any BARRIER
3083 after the second block.
3085 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3087 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3089 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3092 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3093 for 14-bit register offsets when INT14_OK_STRICT is false.
3095 2014-01-07 Roland Stigge <stigge@antcom.de>
3096 Michael Meissner <meissner@linux.vnet.ibm.com>
3099 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3100 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3102 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3104 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3107 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3109 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3110 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3111 rtx is const0_rtx or not.
3113 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3116 * target-globals.c (save_target_globals): Remove this_fn_optab
3118 * toplev.c: Include optabs.h.
3119 (target_reinit): Temporarily restore the global options if another
3120 set of options are in force.
3122 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3124 PR rtl-optimization/58668
3125 * cfgcleanup.c (flow_find_cross_jump): Don't count
3126 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3127 to determine what is counted.
3128 (flow_find_head_matching_sequence): Use active_insn_p to determine
3130 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3132 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3133 determine what is counted.
3135 PR tree-optimization/59643
3136 * tree-predcom.c (split_data_refs_to_components): If one dr is
3137 read and one write, determine_offset fails and the write isn't
3138 in the bad component, just put the read into the bad component.
3140 2014-01-07 Mike Stump <mikestump@comcast.net>
3141 Jakub Jelinek <jakub@redhat.com>
3144 * tree-core.h (struct tree_optimization_option): Change optabs
3145 type from unsigned char * to void *.
3146 * optabs.c (init_tree_optimization_optabs): Adjust
3147 TREE_OPTIMIZATION_OPTABS initialization.
3149 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3152 * config/i386/i386.h (struct machine_function): Add
3153 no_drap_save_restore field.
3154 * config/i386/i386.c (ix86_save_reg): Use
3155 !cfun->machine->no_drap_save_restore instead of
3156 crtl->stack_realign_needed.
3157 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3158 this function clears frame_pointer_needed. Set
3159 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3160 and DRAP reg is needed.
3162 2014-01-06 Marek Polacek <polacek@redhat.com>
3165 * doc/implement-c.texi: Mention that other integer types are
3166 permitted as bit-field types in strictly conforming mode.
3168 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3170 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3173 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3175 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3177 2014-01-06 Martin Jambor <mjambor@suse.cz>
3180 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3182 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3184 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3188 * var-tracking.c (add_stores): Preserve the value of the source even if
3189 we don't record the store.
3191 2014-01-06 Terry Guo <terry.guo@arm.com>
3193 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3195 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3198 * config/darwin.c (darwin_function_section): Adjust return values to
3199 correspond to optimisation changes made in r206070.
3201 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3203 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3204 from prefetch_block tune setting.
3205 (nocona_cost): Correct size of prefetch block to 64.
3207 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3209 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3210 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3211 used to save the static chain register in the computation of the offset
3212 from which the FP registers need to be restored.
3214 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3216 PR tree-optimization/59519
3217 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3218 ICE if get_current_def (current_new_name) is already non-NULL, as long
3219 as it is a phi result of some other phi in *new_exit_bb that has
3222 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3223 or vmovdqu* for misaligned_operand.
3224 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3225 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3226 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3227 aligned_mem for AVX512F masked aligned load and store builtins and for
3230 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3232 PR tree-optimization/59651
3233 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3234 Address range for negative step should be added by TYPE_SIZE_UNIT.
3236 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3238 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3239 overlapping registers also for registers other than the stack pointer.
3241 2014-01-03 Marek Polacek <polacek@redhat.com>
3244 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3247 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3250 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3253 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3254 (push splitter): Use <P:MODE_SIZE> instead of
3255 GET_MODE_SIZE (<P:MODE>mode).
3256 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3257 (mov -1, reg peephole2): Likewise.
3258 * config/i386/sse.md (*mov<mode>_internal,
3259 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3260 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3261 *<code><mode>3, *andnot<mode>3<mask_name>,
3262 <mask_codefor><code><mode>3<mask_name>): Likewise.
3263 * config/i386/subst.md (mask_mode512bit_condition,
3264 sd_mask_mode512bit_condition): Likewise.
3266 2014-01-02 Xinliang David Li <davidxl@google.com>
3268 PR tree-optimization/59303
3269 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3270 (dump_predicates): Better output format.
3271 (pred_equal_p): New function.
3272 (is_neq_relop_p): Ditto.
3273 (is_neq_zero_form_p): Ditto.
3274 (pred_expr_equal_p): Ditto.
3275 (pred_neg_p): Ditto.
3276 (simplify_pred): Ditto.
3277 (simplify_preds_2): Ditto.
3278 (simplify_preds_3): Ditto.
3279 (simplify_preds_4): Ditto.
3280 (simplify_preds): Ditto.
3282 (push_to_worklist): Ditto.
3283 (get_pred_info_from_cmp): Ditto.
3284 (is_degenerated_phi): Ditto.
3285 (normalize_one_pred_1): Ditto.
3286 (normalize_one_pred): Ditto.
3287 (normalize_one_pred_chain): Ditto.
3288 (normalize_preds): Ditto.
3289 (normalize_cond_1): Remove function.
3290 (normalize_cond): Ditto.
3291 (is_gcond_subset_of): Ditto.
3292 (is_subset_of_any): Ditto.
3293 (is_or_set_subset_of): Ditto.
3294 (is_and_set_subset_of): Ditto.
3295 (is_norm_cond_subset_of): Ditto.
3296 (pred_chain_length_cmp): Ditto.
3297 (convert_control_dep_chain_into_preds): Type change.
3298 (find_predicates): Ditto.
3299 (find_def_preds): Ditto.
3300 (destroy_predicates_vecs): Ditto.
3301 (find_matching_predicates_in_rest_chains): Ditto.
3302 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3303 (is_pred_expr_subset): Ditto.
3304 (is_pred_chain_subset_of): Ditto.
3305 (is_included_in): Ditto.
3306 (is_superset_of): Ditto.
3308 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3310 Update copyright years.
3312 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3314 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3315 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3316 config/arc/arc.md, config/arc/arc.opt,
3317 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3318 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3319 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3320 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3321 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3322 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3323 vtable-verify.c, vtable-verify.h: Use the standard form for the
3326 2014-01-02 Tobias Burnus <burnus@net-b.de>
3328 * gcc.c (process_command): Update copyright notice dates.
3329 * gcov-dump.c: Ditto.
3331 * doc/cpp.texi: Bump @copying's copyright year.
3332 * doc/cppinternals.texi: Ditto.
3333 * doc/gcc.texi: Ditto.
3334 * doc/gccint.texi: Ditto.
3335 * doc/gcov.texi: Ditto.
3336 * doc/install.texi: Ditto.
3337 * doc/invoke.texi: Ditto.
3339 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3341 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3343 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3345 * config/i386/sse.md (*mov<mode>_internal): Guard
3346 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3348 PR rtl-optimization/59647
3349 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3350 new_rtx into UNSIGNED_FLOAT rtxes.
3352 Copyright (C) 2014 Free Software Foundation, Inc.
3354 Copying and distribution of this file, with or without modification,
3355 are permitted in any medium without royalty provided the copyright
3356 notice and this notice are preserved.