1 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
5 for VSX/Altivec vectors that land in GPR registers.
7 2014-02-11 Richard Henderson <rth@redhat.com>
8 Jakub Jelinek <jakub@redhat.com>
11 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
12 around drhs if type conversion to lacc->type is not useless.
14 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
18 (cortex-a57.cortex-a53): Likewise.
19 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
21 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
26 2014-02-11 Renlin Li <Renlin.Li@arm.com>
28 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
29 add_options_for_arm_vfp3.
31 2014-02-11 Jeff Law <law@redhat.com>
34 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
35 object with an undesirable mode.
37 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40 * config/i386/sol2-9.h: New file.
41 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
42 *-*-solaris2.9*): Use it.
44 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
46 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
47 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
49 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
51 * config/microblaze/microblaze.c: Extend mcpu version format
53 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
55 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
57 2014-02-10 Richard Henderson <rth@redhat.com>
60 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
61 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
62 ms-abi vs -mno-accumulate-outgoing-args.
63 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
64 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
67 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
70 * cfgexpand.c (expand_asm_operands): Attach source location to
71 ASM_INPUT rtx objects.
72 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
74 2014-02-10 Nick Clifton <nickc@redhat.com>
76 * config/mn10300/mn10300.c (popcount): New function.
77 (mn10300_expand_prologue): Include saved registers in stack usage
80 2014-02-10 Jeff Law <law@redhat.com>
83 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
84 when changing the SET_DEST of a prior insn to avoid an input
87 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
89 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
90 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
91 -mcall-openbsd, or -mcall-linux.
92 (CC1_ENDIAN_BIG_SPEC): Remove.
93 (CC1_ENDIAN_LITTLE_SPEC): Remove.
94 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
95 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
96 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
97 and %cc1_endian_default.
98 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
100 2014-02-10 Richard Biener <rguenther@suse.de>
102 PR tree-optimization/60115
103 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
104 MEM_REF handling. Properly verify that the accesses are not
105 out of the objects bound.
107 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
109 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
112 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
114 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
115 proper constants and fix formatting.
116 (possible_polymorphic_call_targets): Fix formatting.
118 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
119 Ilya Tocar <ilya.tocar@intel.com>
121 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
122 (_mm512_loadu_epi32): Renamed into...
123 (_mm512_loadu_si512): This.
124 (_mm512_storeu_epi32): Renamed into...
125 (_mm512_storeu_si512): This.
126 (_mm512_maskz_ceil_ps): Removed.
127 (_mm512_maskz_ceil_pd): Ditto.
128 (_mm512_maskz_floor_ps): Ditto.
129 (_mm512_maskz_floor_pd): Ditto.
130 (_mm512_floor_round_ps): Ditto.
131 (_mm512_floor_round_pd): Ditto.
132 (_mm512_ceil_round_ps): Ditto.
133 (_mm512_ceil_round_pd): Ditto.
134 (_mm512_mask_floor_round_ps): Ditto.
135 (_mm512_mask_floor_round_pd): Ditto.
136 (_mm512_mask_ceil_round_ps): Ditto.
137 (_mm512_mask_ceil_round_pd): Ditto.
138 (_mm512_maskz_floor_round_ps): Ditto.
139 (_mm512_maskz_floor_round_pd): Ditto.
140 (_mm512_maskz_ceil_round_ps): Ditto.
141 (_mm512_maskz_ceil_round_pd): Ditto.
142 (_mm512_expand_pd): Ditto.
143 (_mm512_expand_ps): Ditto.
144 * config/i386/i386.c (ix86_builtins): Remove
145 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
147 * config/i386/predicates.md (const1256_operand): New.
148 (const_1_to_2_operand): Ditto.
149 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
150 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
151 (*avx512pf_gatherpf<mode>sf): Ditto.
152 (avx512pf_gatherpf<mode>df): Ditto.
153 (*avx512pf_gatherpf<mode>df_mask): Ditto.
154 (*avx512pf_gatherpf<mode>df): Ditto.
155 (avx512pf_scatterpf<mode>sf): Ditto.
156 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
157 (*avx512pf_scatterpf<mode>sf): Ditto.
158 (avx512pf_scatterpf<mode>df): Ditto.
159 (*avx512pf_scatterpf<mode>df_mask): Ditto.
160 (*avx512pf_scatterpf<mode>df): Ditto.
161 (avx512f_expand<mode>): Removed.
162 (<shift_insn><mode>3<mask_name>): Change predicate type.
164 2014-02-08 Jakub Jelinek <jakub@redhat.com>
166 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
167 not at the end of datarefs vector use ordered_remove to avoid
168 reordering datarefs vector.
171 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
172 mark local addressable non-static vars as GOVD_PRIVATE
173 instead of GOVD_LOCAL.
174 * omp-low.c (lower_omp_for): Move gimple_bind_vars
175 and BLOCK_VARS of gimple_bind_block to new_stmt rather
179 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
180 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
181 assume_aligned or alloc_align attributes.
182 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
183 arguments. Handle also assume_aligned and alloc_align attributes.
184 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
185 Handle calls to functions with assume_aligned or alloc_align
187 * doc/extend.texi: Document assume_aligned and alloc_align
190 2014-02-08 Terry Guo <terry.guo@arm.com>
192 * doc/invoke.texi: Document ARM -march=armv7e-m.
194 2014-02-08 Jakub Jelinek <jakub@redhat.com>
196 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
197 flag on __cilkrts_rethrow builtin.
200 * ipa-cp.c (determine_versionability): Fail at -O0
201 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
202 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
205 2014-02-04 Jakub Jelinek <jakub@redhat.com>
208 * tree-inline.c (copy_forbidden): Fail for
209 __attribute__((optimize (0))) functions.
211 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
213 * varpool.c: Include pointer-set.h.
214 (varpool_remove_unreferenced_decls): Variables in other partitions
215 will not be output; be however careful to not lose information
218 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
220 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
221 lookup in the vtable constructor.
223 2014-02-07 Jeff Law <law@redhat.com>
226 * config/m68k/m68k.md (ashldi_extsi): Turn into a
227 define_insn_and_split.
229 * ipa-inline.c (inline_small_functions): Fix typos.
231 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
233 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
234 (s390_can_use_return_insn): Declare.
235 * config/s390/s390.h (EPILOGUE_USES): Define.
236 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
238 (s390_chunkify_start): Handle return JUMP_LABELs.
239 (s390_early_mach): Emit a main_pool instruction on the entry edge.
240 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
241 (s390_can_use_return_insn): New functions.
242 (s390_fix_long_loop_prediction): Handle conditional returns.
243 (TARGET_SET_UP_BY_PROLOGUE): Define.
244 * config/s390/s390.md (ANY_RETURN): New code iterator.
245 (*creturn, *csimple_return, return, simple_return): New patterns.
247 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
249 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
250 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
251 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
252 REG_CFA_RESTORE list when deciding not to restore a register.
254 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
256 * config/s390/s390.c: Include tree-pass.h and context.h.
257 (s390_early_mach): New function, split out from...
258 (s390_emit_prologue): ...here.
259 (pass_data_s390_early_mach): New pass structure.
260 (pass_s390_early_mach): New class.
261 (s390_option_override): Create and register early_mach pass.
264 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
266 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
267 to match for the exit block.
269 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
271 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
272 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
273 Reject misaligned operands.
275 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
277 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
280 2014-02-07 Richard Biener <rguenther@suse.de>
283 * gimple-low.c (lower_builtin_posix_memalign): New function.
284 (lower_stmt): Call it to lower posix_memalign in a way
285 to make alignment info accessible.
287 2014-02-07 Jakub Jelinek <jakub@redhat.com>
290 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
291 __builtin_setjmp_receiver.
293 2014-02-07 Richard Biener <rguenther@suse.de>
296 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
297 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
298 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
299 Handle BUILT_IN_POSIX_MEMALIGN.
300 (find_func_clobbers): Likewise.
301 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
302 (call_may_clobber_ref_p_1): Likewise.
304 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
307 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
309 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
312 * lto-cgraph.c (lto_output_node): Use
313 symtab_get_symbol_partitioning_class.
314 (lto_output_varpool_node): likewise.
315 (symtab_get_symbol_partitioning_class): Move here from
317 * cgraph.h (symbol_partitioning_class): Likewise.
318 (symtab_get_symbol_partitioning_class): Declare.
320 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
322 * ggc.h (ggc_internal_cleared_alloc): New macro.
323 * vec.h (vec_safe_copy): Handle memory stats.
324 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
325 * target-globals.c (save_target_globals): Likewise.
327 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
330 * expr.c (emit_move_resolve_push): Export; be bit more selective
331 on when to clear alias set.
332 * expr.h (emit_move_resolve_push): Declare.
333 * function.h (struct function): Add tail_call_marked.
334 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
335 * config/i386/i386-protos.h (ix86_expand_push): Remove.
336 * config/i386/i386.md (TImode move expander): De not call
338 (FP push expanders): Preserve memory attributes.
339 * config/i386/sse.md (push<mode>1): Remove.
340 * config/i386/i386.c (ix86_expand_vector_move): Handle push
342 (ix86_expand_push): Remove.
343 * config/i386/mmx.md (push<mode>1): Remove.
345 2014-02-06 Jakub Jelinek <jakub@redhat.com>
347 PR rtl-optimization/60030
348 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
349 lopart with paradoxical subreg before shifting it up by hprec.
351 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
353 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
354 Remove extra newline at end of file.
355 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
356 (arm_issue_rate): Handle cortexa57.
357 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
358 (cortex-a57.cortex-a53): Likewise.
360 2014-02-06 Jakub Jelinek <jakub@redhat.com>
363 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
364 don't record in REG_FRAME_RELATED_EXPR registers not set in that
366 (arm_expand_prologue): Adjust all callers.
367 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
368 info, registers also at the lowest numbered registers side. Use
369 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
373 * var-tracking.c (adjust_mems): Before adding a SET
374 to amd->side_effects, adjust it's SET_SRC using
375 simplify_replace_fn_rtx.
377 2014-02-06 Alan Modra <amodra@gmail.com>
380 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
381 change SDmode to DDmode when lra_in_progress.
383 2014-02-06 Jakub Jelinek <jakub@redhat.com>
386 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
387 free_data_ref on the dr first, and before goto again also set dr
388 to the next dr. For simd_lane_access, free old datarefs[i] before
389 overwriting it. For get_vectype_for_scalar_type failure, don't
390 free_data_ref if simd_lane_access.
392 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
395 * tree.h (opts_for_fn): New inline function.
396 (opt_for_fn): Define.
397 * config/i386/i386.c (ix86_function_regparm): Use
398 opt_for_fn (decl, optimize) instead of optimize.
400 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
402 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
403 for SYMBOL_REF in large memory model.
405 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
407 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
409 (cortex-a57): Likewise.
410 (cortex-a57.cortex-a53): Likewise.
412 2014-02-06 Yury Gribov <y.gribov@samsung.com>
413 Kugan Vivekanandarajah <kuganv@linaro.org>
415 * config/arm/arm.c (arm_vector_alignment_reachable): Check
417 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
419 2014-02-06 Richard Biener <rguenther@suse.de>
421 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
422 set_loop_copy and initialize_original_copy_tables.
424 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
426 * config/aarch64/aarch64-simd.md
427 (aarch64_ashr_simddi): Change QI to SI.
429 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
430 Jakub Jelinek <jakub@redhat.com>
433 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
436 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
438 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
439 CODE_FOR_altivec_vpku[hw]um to
440 CODE_FOR_altivec_vpku[hw]um_direct.
441 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
442 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
443 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
444 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
446 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
448 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
449 generation for -maltivec=be.
450 (altivec_vsumsws): Simplify redundant test.
452 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
454 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
455 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
456 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
457 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
459 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
461 (altivec_vpks<VI_char>ss): Likewise.
462 (altivec_vpks<VI_char>us): Likewise.
463 (altivec_vpku<VI_char>us): Likewise.
464 (altivec_vpku<VI_char>um): Likewise.
465 (altivec_vpku<VI_char>um_direct): New (copy of
466 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
468 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
469 target is little endian and -maltivec=be is not specified.
470 (*altivec_vupkhs<VU_char>_direct): New (copy of
471 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
473 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
474 target is little endian and -maltivec=be is not specified.
475 (*altivec_vupkls<VU_char>_direct): New (copy of
476 altivec_vupkls<VU_char> that always emits vupkls*, for internal
478 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
479 little endian and -maltivec=be is not specified.
480 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
481 little endian and -maltivec=be is not specified.
483 2014-02-05 Richard Henderson <rth@redhat.com>
486 * combine-stack-adj.c: Revert r206943.
487 * sched-int.h (struct deps_desc): Add last_args_size.
488 * sched-deps.c (init_deps): Initialize it.
489 (sched_analyze_insn): Add OUTPUT dependencies between insns that
490 contain REG_ARGS_SIZE notes.
492 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
494 * lto-cgraph.c (asm_nodes_output): Make global.
495 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
496 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
498 (driver_handle_option): Handle OPT_fwpa.
500 2014-02-05 Jakub Jelinek <jakub@redhat.com>
503 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
504 a comment typo and formatting issue. If odr_hash hasn't been
505 created, return vNULL and set *completep to false.
508 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
509 bb with no successors.
511 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
514 * doc/invoke.texi (-march): Clarify documentation for ARM.
518 2014-02-05 Richard Biener <rguenther@suse.de>
520 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
521 when not vectorizing because of too many alias checks.
522 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
523 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
525 2014-02-05 Nick Clifton <nickc@redhat.com>
527 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
528 accept extended registers in any mode when compiling for the
531 2014-02-05 Yury Gribov <y.gribov@samsung.com>
533 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
534 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
535 sanitization attributes.
536 (can_inline_edge_p): Likewise.
537 (sanitize_attrs_match_for_inline_p): New function.
539 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
541 * ipa-prop.c (detect_type_change): Shor circuit testing of
542 type changes on THIS pointer.
544 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
547 * config/pa/pa.c (legitimize_tls_address): Return original address
548 if not passed a SYMBOL_REF rtx.
549 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
551 (pa_emit_move_sequence): Simplify TLS source operands.
552 (pa_legitimate_constant_p): Reject all TLS constants.
553 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
554 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
556 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
558 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
559 groups when we know they are controlled by LTO.
560 * varasm.c (default_binds_local_p_1): If object is in other partition,
561 it will be resolved locally.
563 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
565 * config/host-linux.c (linux_gt_pch_use_address): Don't
566 use SSIZE_MAX because it is not always defined.
568 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
571 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
572 threshold for pseudo splitting.
573 (update_ebb_live_info): Process call argument hard registers and
574 hard registers from insn definition too.
575 (max_small_class_regs_num): New constant.
576 (inherit_in_ebb): Update live hard regs through EBBs. Update
577 reloads_num only for small register classes. Don't split for
580 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
583 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
586 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
588 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
590 2014-02-04 Marek Polacek <polacek@redhat.com>
592 * gdbinit.in (pel): Define.
594 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
596 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
599 2014-02-04 Richard Biener <rguenther@suse.de>
602 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
603 in function context local.
604 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
605 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
606 similar to LTO_imported_decl_ref.
608 2014-02-04 Jakub Jelinek <jakub@redhat.com>
610 PR tree-optimization/60002
611 * cgraphclones.c (build_function_decl_skip_args): Clear
614 PR tree-optimization/60023
615 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
616 false to gsi_replace.
617 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
618 has been in some EH region and vec_stmt could throw, add
619 vec_stmt into the same EH region.
620 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
621 has no lhs, ignore it.
622 * internal-fn.c (expand_MASK_LOAD): Likewise.
625 * tree-inline.c (copy_forbidden): Fail for
626 __attribute__((optimize (0))) functions.
629 * omp-low.c (simd_clone_struct_copy): If from->inbranch
630 is set, copy one less argument.
631 (expand_simd_clones): Don't subtract clone_info->inbranch
632 from simd_clone_struct_alloc argument.
634 PR rtl-optimization/57915
635 * recog.c (simplify_while_replacing): If all unary/binary/relational
636 operation arguments are constant, attempt to simplify those.
639 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
640 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
642 2014-02-04 Richard Biener <rguenther@suse.de>
644 PR tree-optimization/60012
645 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
646 TBAA disambiguation to all DDRs.
648 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
651 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
652 (LINK_SPEC): Use it for -shared, -shared-libgcc.
654 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
657 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
659 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
661 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
662 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
664 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
667 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
668 to figure out targets of polymorphic calls with known decl.
669 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
670 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
671 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
672 (get_polymorphic_call_info): ... here.
673 (get_polymorphic_call_info_from_invariant): New function.
675 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
677 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
678 lookup via vtable pointer; check for type consistency
679 and turn inconsitent facts into UNREACHABLE.
680 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
681 * gimple-fold.c (gimple_get_virt_method_for_vtable):
682 Do not ICE on type inconsistent querries; return UNREACHABLE
685 2014-02-03 Richard Henderson <rth@twiddle.net>
688 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
689 already processed this node.
690 (normalize_one_pred_1): Pass along mark_set.
691 (normalize_one_pred): Create and destroy a pointer_set_t.
692 (normalize_one_pred_chain): Likewise.
694 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
696 PR gcov-profile/58602
697 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
699 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
702 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
703 Try to devirtualize by the knowledge of virtual table pointer given by
704 aggregate propagation.
705 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
706 ipa_print_node_jump_functions): Dump also offset that
707 is relevant for polymorphic calls.
708 (determine_known_aggregate_parts): Add arg_type parameter; use it
709 instead of determining the type from pointer type.
710 (ipa_compute_jump_functions_for_edge): Update call of
711 determine_known_aggregate_parts.
712 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
713 (gimple_get_virt_method_for_binfo): ... here; simplify using
714 vtable_pointer_value_to_vtable.
715 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
716 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
717 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
718 (vtable_pointer_value_to_vtable): Break out from ...; handle also
720 (vtable_pointer_value_to_binfo): ... here.
721 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
723 2014-02-03 Teresa Johnson <tejohnson@google.com>
725 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
726 redef of outer loop index variable.
728 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
732 * doc/extend.texi (Function Attributes): Typo.
734 2014-02-03 Cong Hou <congh@google.com>
736 PR tree-optimization/60000
737 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
738 if the vectorized statement is a store. A store statement can only
739 appear at the end of pattern statements.
741 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
743 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
744 (ix86_option_override_internal): Default long double to 64-bit for
745 32-bit Bionic and to 128-bit for 64-bit Bionic.
747 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
748 TARGET_LONG_DOUBLE_128 is true.
749 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
751 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
752 (mlong-double-64): Negate -mlong-double-128.
753 (mlong-double-128): New option.
755 * config/i386/i386-c.c (ix86_target_macros): Define
756 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
758 * doc/invoke.texi: Document -mlong-double-128.
760 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
762 PR rtl-optimization/60024
763 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
765 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
767 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
769 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
771 PR rtl-optimization/57662
772 * sel-sched.c (code_motion_path_driver): Do not mark already not
773 existing blocks in the visiting bitmap.
775 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
777 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
778 on the insn being emitted.
780 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
781 Will Deacon <will.deacon@arm.com>
783 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
785 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
787 * config/arm/arm-tables.opt: Regenerate.
789 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
791 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
792 for vector types other than V16QImode.
793 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
794 define_expand, and call altivec_expand_vec_perm_le when producing
795 code with little endian element order.
796 (*altivec_vperm_<mode>_internal): New insn having previous
797 behavior of altivec_vperm_<mode>.
798 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
799 altivec_expand_vec_perm_le when producing code with little endian
801 (*altivec_vperm_<mode>_uns_internal): New insn having previous
802 behavior of altivec_vperm_<mode>_uns.
804 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
806 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
807 (altivec_vsumsws): Add handling for -maltivec=be with a little
809 (altivec_vsumsws_direct): New.
810 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
813 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
815 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
816 vtable_pointer_value_to_binfo): New functions.
817 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
818 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
820 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
822 * config/nios2/nios2.md (load_got_register): Initialize GOT
823 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
824 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
826 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
828 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
829 preserverd by passthrough, do not propagate the type.
831 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
833 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
834 (mips_atomic_assign_expand_fenv): New function.
835 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
837 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
839 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
840 (__builtin_mips_set_fcsr): Likewise.
841 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
843 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
844 (mips16_expand_set_fcsr): Likewise.
845 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
846 (mips16_set_fcsr_stub): Likewise.
847 (mips16_get_fcsr_one_only_stub): New class.
848 (mips16_set_fcsr_one_only_stub): Likewise.
849 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
850 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
851 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
852 (hard_float): New availability predicate.
853 (mips_builtins): Add get_fcsr and set_fcsr.
854 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
855 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
856 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
857 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
858 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
861 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
863 * config/mips/mips.c (mips_one_only_stub): New class.
864 (mips_need_mips16_rdhwr_p): Replace with...
865 (mips16_rdhwr_stub): ...this new variable.
866 (mips16_stub_call_address): New function.
867 (mips16_rdhwr_one_only_stub): New class.
868 (mips_expand_thread_pointer): Use mips16_stub_call_address.
869 (mips_output_mips16_rdhwr): Delete.
870 (mips_finish_stub): New function.
871 (mips_code_end): Use it to handle rdhwr stubs.
873 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
876 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
877 when calculating size of integer atomic types.
879 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
881 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
883 2014-02-01 Jakub Jelinek <jakub@redhat.com>
885 PR tree-optimization/60003
886 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
887 * profile.c (branch_prob): Use gimple_call_builtin_p
888 to check for BUILT_IN_SETJMP_RECEIVER.
889 * tree-inline.c (copy_bb): Call notice_special_calls.
891 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
894 * lra-constraints.c (process_alt_operands): Update reload_sum only
897 2014-01-31 Richard Henderson <rth@redhat.com>
900 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
901 until after else_eh is processed.
903 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
905 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
906 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
907 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
908 in smmintrin.h, remove them.
909 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
910 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
911 * config/i386/i386.md (ROUND_SAE): Fix value.
912 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
913 (const48_operand): New.
914 * config/i386/subst.md (round), (round_expand): Use
915 const_4_or_8_to_11_operand.
916 (round_saeonly), (round_saeonly_expand): Use const48_operand.
918 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
920 * config/i386/constraints.md (Yk): Swap meaning with k.
921 * config/i386/i386.md (movhi_internal): Change Yk to k.
922 (movqi_internal): Ditto.
923 (*k<logic><mode>): Ditto.
926 (kandn<mode>): Ditto.
927 (*<code>hi_1): Ditto.
928 (*<code>qi_1): Ditto.
929 (kxnor<mode>): Ditto.
933 (*one_cmplhi2_1): Ditto.
934 (*one_cmplqi2_1): Ditto.
935 * config/i386/sse.md (): Change k to Yk.
936 (avx512f_load<mode>_mask): Ditto.
937 (avx512f_blendm<mode>): Ditto.
938 (avx512f_store<mode>_mask): Ditto.
939 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
940 (avx512f_storedqu<mode>_mask): Ditto.
941 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
943 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
944 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
945 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
946 (avx512f_maskcmp<mode>3): Ditto.
947 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
948 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
949 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
950 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
951 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
952 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
953 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
954 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
955 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
956 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
957 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
958 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
959 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
960 (vec_extract_lo_<mode>_maskm): Ditto.
961 (vec_extract_hi_<mode>_maskm): Ditto.
962 (avx512f_vternlog<mode>_mask): Ditto.
963 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
964 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
965 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
966 (avx512f_<code>v8div16qi2_mask): Ditto.
967 (avx512f_<code>v8div16qi2_mask_store): Ditto.
968 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
969 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
970 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
971 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
972 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
973 (*avx512pf_gatherpf<mode>df_mask): Ditto.
974 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
975 (*avx512pf_scatterpf<mode>df_mask): Ditto.
976 (avx512cd_maskb_vec_dupv8di): Ditto.
977 (avx512cd_maskw_vec_dupv16si): Ditto.
978 (avx512f_vpermi2var<mode>3_maskz): Ditto.
979 (avx512f_vpermi2var<mode>3_mask): Ditto.
980 (avx512f_vpermi2var<mode>3_mask): Ditto.
981 (avx512f_vpermt2var<mode>3_maskz): Ditto.
982 (*avx512f_gathersi<mode>): Ditto.
983 (*avx512f_gathersi<mode>_2): Ditto.
984 (*avx512f_gatherdi<mode>): Ditto.
985 (*avx512f_gatherdi<mode>_2): Ditto.
986 (*avx512f_scattersi<mode>): Ditto.
987 (*avx512f_scatterdi<mode>): Ditto.
988 (avx512f_compress<mode>_mask): Ditto.
989 (avx512f_compressstore<mode>_mask): Ditto.
990 (avx512f_expand<mode>_mask): Ditto.
991 * config/i386/subst.md (mask): Change k to Yk.
992 (mask_scalar_merge): Ditto.
995 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
997 * doc/extend.texi (Vector Extensions): Document ?: in C++.
999 2014-01-31 Richard Biener <rguenther@suse.de>
1002 * builtins.c (fold_builtin_memory_op): Make sure to not
1003 use a floating-point mode or a boolean or enumeral type for
1006 2014-01-30 DJ Delorie <dj@redhat.com>
1008 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1009 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1010 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1011 whenever main() has an epilogue.
1013 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1015 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1016 unused variable "field".
1017 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1018 (vsx_mergeh_<mode>): Likewise.
1019 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1020 (altivec_vmrghh): Likewise.
1021 (altivec_vmrghw): Likewise.
1022 (altivec_vmrglb): Likewise.
1023 (altivec_vmrglh): Likewise.
1024 (altivec_vmrglw): Likewise.
1025 (altivec_vspltb): Add missing uses.
1026 (altivec_vsplth): Likewise.
1027 (altivec_vspltw): Likewise.
1028 (altivec_vspltsf): Likewise.
1030 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1033 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1034 frame related instructions.
1036 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1038 PR rtl-optimization/59959
1039 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1040 any reload of register whose subreg is invalid.
1042 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1044 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1045 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1046 Add missing return type - void.
1048 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1050 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1051 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1052 remove element index adjustment for endian (now handled in vsx.md
1054 (altivec_expand_vec_perm_const): Use
1055 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1056 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1057 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1058 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1059 define_expand and a new define_insn *altivec_vspltb_internal;
1060 adjust for -maltivec=be on a little endian target.
1061 (altivec_vspltb_direct): New.
1062 (altivec_vsplth): Divide into a define_expand and a new
1063 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1064 little endian target.
1065 (altivec_vsplth_direct): New.
1066 (altivec_vspltw): Divide into a define_expand and a new
1067 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1068 little endian target.
1069 (altivec_vspltw_direct): New.
1070 (altivec_vspltsf): Divide into a define_expand and a new
1071 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1072 a little endian target.
1074 2014-01-30 Richard Biener <rguenther@suse.de>
1076 PR tree-optimization/59993
1077 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1078 can propagate form the earlier stmt and avoid the transform
1079 when the intermediate result is needed.
1081 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1083 * README.Portability: Fix typo.
1085 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1087 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1088 comparison_operator with ordered_comparison_operator.
1090 2014-01-30 Nick Clifton <nickc@redhat.com>
1092 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1093 Rename to mn10300_store_multiple_regs.
1094 * config/mn10300/mn10300.c: Likewise.
1095 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1096 store_multiple_regs.
1097 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1098 Call mn10300_store_multiple_regs.
1100 2014-01-30 Nick Clifton <nickc@redhat.com>
1101 DJ Delorie <dj@redhat.com>
1103 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1104 %fp 2 to keep registers after it properly word-aligned.
1105 (rl78_alloc_physical_registers_umul): Handle the case where both
1106 input operands are the same.
1108 2014-01-30 Richard Biener <rguenther@suse.de>
1110 PR tree-optimization/59903
1111 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1114 2014-01-30 Jason Merrill <jason@redhat.com>
1117 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1120 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1122 2014-01-30 Richard Biener <rguenther@suse.de>
1124 PR tree-optimization/59951
1125 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1127 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1130 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1131 SFmode to DFmode case.
1133 2014-01-29 DJ Delorie <dj@redhat.com>
1135 * config/msp430/msp430.opt (-minrt): New.
1136 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1138 (ENDFILE_SPEC): Likewise.
1140 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1142 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1143 (estimate_function_body_sizes): Use it.
1145 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1148 * dwarf2out.c (is_cxx_auto): New.
1149 (is_base_type): Use it.
1150 (gen_type_die_with_usage): Likewise.
1152 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1154 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1155 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1156 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1157 -maltivec=be with LE targets.
1158 (vsx_mergeh_<mode>): Likewise.
1159 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1160 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1161 (altivec_vmrghb): Replace with define_expand and new
1162 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1163 (altivec_vmrghb_direct): New define_insn.
1164 (altivec_vmrghh): Replace with define_expand and new
1165 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1166 (altivec_vmrghh_direct): New define_insn.
1167 (altivec_vmrghw): Replace with define_expand and new
1168 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1169 (altivec_vmrghw_direct): New define_insn.
1170 (*altivec_vmrghsf): Adjust for endianness.
1171 (altivec_vmrglb): Replace with define_expand and new
1172 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1173 (altivec_vmrglb_direct): New define_insn.
1174 (altivec_vmrglh): Replace with define_expand and new
1175 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1176 (altivec_vmrglh_direct): New define_insn.
1177 (altivec_vmrglw): Replace with define_expand and new
1178 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1179 (altivec_vmrglw_direct): New define_insn.
1180 (*altivec_vmrglsf): Adjust for endianness.
1181 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1182 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1183 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1184 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1185 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1186 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1187 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1188 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1190 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1192 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1193 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1196 2014-01-29 Richard Biener <rguenther@suse.de>
1198 PR tree-optimization/58742
1199 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1200 associate_pointerplus_align.
1201 (associate_pointerplus_diff): New function.
1202 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1203 and associate_pointerplus_diff.
1205 2014-01-29 Richard Biener <rguenther@suse.de>
1207 * lto-streamer.h (LTO_major_version): Bump to 3.
1208 (LTO_minor_version): Reset to 0.
1210 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1212 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1213 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1214 (arm_file_start): Generate correct asm header for armv7ve.
1215 * config/arm/bpabi.h: Add multilib support for armv7ve.
1216 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1217 and cortex-a15 to armv7ve.
1218 * config/arm/t-aprofile: Add multilib support for armv7ve.
1219 * doc/invoke.texi: Document -march=armv7ve.
1221 2014-01-29 Richard Biener <rguenther@suse.de>
1223 PR tree-optimization/58742
1224 * tree-ssa-forwprop.c (associate_plusminus): Return true
1225 if we changed sth, defer EH cleanup to ...
1226 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1227 (simplify_mult): New function.
1229 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1232 PR tree-optimization/59920
1233 * tree.c (build_common_builtin_nodes): Remove
1234 __builtin_setjmp_dispatcher initialization.
1235 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1236 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1237 instead of gsi_after_labels + manually skipping debug stmts.
1238 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1239 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1240 * tree-inline.c (copy_edges_for_bb): Remove
1241 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1242 argument. Ignore computed_goto_p stmts. Don't call
1243 make_abnormal_goto_edges. If a call might need abnormal edges
1244 for non-local gotos, see if it already has an edge to
1245 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1246 with true argument, don't do anything then, otherwise add
1247 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1248 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1250 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1251 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1252 (lower_stmt): Don't set data->calls_builtin_setjmp.
1253 (lower_builtin_setjmp): Adjust comment.
1254 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1255 * tree-cfg.c (found_computed_goto): Remove.
1256 (factor_computed_gotos): Remove.
1257 (make_goto_expr_edges): Return bool, true for computed gotos.
1258 Don't call make_abnormal_goto_edges.
1259 (build_gimple_cfg): Don't set found_computed_goto, don't call
1260 factor_computed_gotos.
1261 (computed_goto_p): No longer static.
1262 (make_blocks): Don't set found_computed_goto.
1263 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1264 (make_edges): If make_goto_expr_edges returns true, push bb
1265 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1266 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1267 vector. Record mapping between bbs and OpenMP regions if there
1268 are any, adjust make_gimple_omp_edges caller. Call
1269 handle_abnormal_edges.
1270 (make_abnormal_goto_edges): Remove.
1271 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1272 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1273 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1274 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1275 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1276 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1277 filling *region also set *region_idx to (*region)->entry->index.
1280 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1281 For REGs set ORIGINAL_REGNO.
1283 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1285 * doc/md.texi: Mention that a target shouldn't implement
1286 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1289 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1291 PR tree-optimization/59594
1292 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1293 a copy of the datarefs vector rather than the vector itself.
1295 2014-01-28 Jason Merrill <jason@redhat.com>
1298 * dwarf2out.c (auto_die): New static.
1299 (gen_type_die_with_usage): Handle C++1y 'auto'.
1300 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1303 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1306 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1307 (SPEC_X32): Likewise.
1308 (SPEC_64): Likewise.
1309 * config/i386/i386.c (ix86_option_override_internal): Turn off
1310 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1312 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1313 * config/i386/i386.h (TARGET_16BIT): New macro.
1314 (TARGET_16BIT_P): Likewise.
1315 * config/i386/i386.opt: Add m16.
1316 * doc/invoke.texi: Document -m16.
1318 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1320 PR preprocessor/59935
1321 * input.c (location_get_source_line): Bail out on when line number
1322 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1324 2014-01-28 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/58742
1327 * tree-ssa-forwprop.c (associate_plusminus): Handle
1328 pointer subtraction of the form (T)(P + A) - (T)P.
1330 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1332 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1335 2014-01-28 Richard Biener <rguenther@suse.de>
1338 2014-01-28 Richard Biener <rguenther@suse.de>
1340 PR rtl-optimization/45364
1341 PR rtl-optimization/59890
1342 * var-tracking.c (local_get_addr_clear_given_value): Handle
1343 already cleared slot.
1344 (val_reset): Handle not allocated local_get_addr_cache.
1345 (vt_find_locations): Use post-order on the inverted CFG.
1347 2014-01-28 Richard Biener <rguenther@suse.de>
1349 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1351 2014-01-28 Richard Biener <rguenther@suse.de>
1353 PR rtl-optimization/45364
1354 PR rtl-optimization/59890
1355 * var-tracking.c (local_get_addr_clear_given_value): Handle
1356 already cleared slot.
1357 (val_reset): Handle not allocated local_get_addr_cache.
1358 (vt_find_locations): Use post-order on the inverted CFG.
1360 2014-01-28 Alan Modra <amodra@gmail.com>
1362 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1363 * configure.ac <recursive call for build != host>: Define
1364 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1365 and LD_FOR_BUILD too.
1366 * configure: Regenerate.
1368 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1370 * config/i386/i386.c (get_builtin_code_for_version): Separate
1371 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1372 Broadwell from Haswell.
1374 2014-01-27 Steve Ellcey <sellcey@mips.com>
1376 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1377 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1378 * config/mips/mips.c (mips_option_override): Change setting
1380 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1381 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1382 Change from Mask to Var.
1384 2014-01-27 Jeff Law <law@redhat.com>
1386 * ipa-inline.c (inline_small_functions): Fix typo.
1388 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1390 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1391 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1392 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1393 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1394 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1395 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1396 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1397 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1398 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1399 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1400 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1401 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1402 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1403 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1404 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1405 (_mm512_storeu_epi64): Ditto.
1406 (_mm512_cmpge_epi32_mask): Ditto.
1407 (_mm512_cmpge_epu32_mask): Ditto.
1408 (_mm512_cmpge_epi64_mask): Ditto.
1409 (_mm512_cmpge_epu64_mask): Ditto.
1410 (_mm512_cmple_epi32_mask): Ditto.
1411 (_mm512_cmple_epu32_mask): Ditto.
1412 (_mm512_cmple_epi64_mask): Ditto.
1413 (_mm512_cmple_epu64_mask): Ditto.
1414 (_mm512_cmplt_epi32_mask): Ditto.
1415 (_mm512_cmplt_epu32_mask): Ditto.
1416 (_mm512_cmplt_epi64_mask): Ditto.
1417 (_mm512_cmplt_epu64_mask): Ditto.
1418 (_mm512_cmpneq_epi32_mask): Ditto.
1419 (_mm512_cmpneq_epu32_mask): Ditto.
1420 (_mm512_cmpneq_epi64_mask): Ditto.
1421 (_mm512_cmpneq_epu64_mask): Ditto.
1422 (_mm512_expand_pd): Ditto.
1423 (_mm512_expand_ps): Ditto.
1424 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1425 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1426 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1427 * config/i386/i386.c (ix86_builtins): Add
1428 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1429 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1430 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1431 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1432 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1433 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1434 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1435 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1436 IX86_BUILTIN_PMOVUSQW512_MEM.
1437 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1438 __builtin_ia32_pmovsqd512mem_mask,
1439 __builtin_ia32_pmovqd512mem_mask,
1440 __builtin_ia32_pmovusqw512mem_mask,
1441 __builtin_ia32_pmovsqw512mem_mask,
1442 __builtin_ia32_pmovqw512mem_mask,
1443 __builtin_ia32_pmovusdw512mem_mask,
1444 __builtin_ia32_pmovsdw512mem_mask,
1445 __builtin_ia32_pmovdw512mem_mask,
1446 __builtin_ia32_pmovqb512mem_mask,
1447 __builtin_ia32_pmovusqb512mem_mask,
1448 __builtin_ia32_pmovsqb512mem_mask,
1449 __builtin_ia32_pmovusdb512mem_mask,
1450 __builtin_ia32_pmovsdb512mem_mask,
1451 __builtin_ia32_pmovdb512mem_mask.
1452 (bdesc_args): Add __builtin_ia32_expanddf512,
1453 __builtin_ia32_expandsf512.
1454 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1455 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1456 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1457 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1458 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1459 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1460 (avx512f_<code>v8div16qi2_mask_store): This.
1461 (avx512f_expand<mode>): New.
1463 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1465 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1467 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1468 (_mm512_prefetch_i32scatter_pd): Ditto.
1469 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1470 (_mm512_prefetch_i64scatter_pd): Ditto.
1471 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1472 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1473 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1474 (_mm512_prefetch_i32scatter_ps): Ditto.
1475 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1476 (_mm512_prefetch_i64scatter_ps): Ditto.
1477 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1478 * config/i386/i386-builtin-types.def: Define
1479 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1480 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1481 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1482 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1483 IX86_BUILTIN_SCATTERPFQPD.
1484 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1485 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1486 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1487 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1488 __builtin_ia32_scatterpfqps.
1489 (ix86_expand_builtin): Expand new built-ins.
1490 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1491 fix memory access data type.
1492 (*avx512pf_gatherpf<mode>_mask): Ditto.
1493 (*avx512pf_gatherpf<mode>): Ditto.
1494 (avx512pf_scatterpf<mode>): Ditto.
1495 (*avx512pf_scatterpf<mode>_mask): Ditto.
1496 (*avx512pf_scatterpf<mode>): Ditto.
1497 (GATHER_SCATTER_SF_MEM_MODE): New.
1498 (avx512pf_gatherpf<mode>df): Ditto.
1499 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1500 (*avx512pf_scatterpf<mode>df): Ditto.
1502 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1505 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1506 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1509 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1511 * common/config/arm/arm-common.c
1512 (arm_rewrite_mcpu): Handle multiple names.
1514 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1516 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1518 * gimple-builder.h (create_gimple_tmp): Delete.
1520 2014-01-27 Christian Bruel <christian.bruel@st.com>
1522 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1525 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1527 * config/pa/pa.md (call): Generate indirect long calls to non-local
1528 functions when outputing 32-bit code.
1529 (call_value): Likewise except for special call to buggy powf function.
1531 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1532 portable runtime and PIC indirect calls.
1533 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1534 and PIC call sequences. Use ldo instead of blr to set return register
1535 in PIC call sequence.
1537 2014-01-25 Walter Lee <walt@tilera.com>
1539 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1540 avoid clobbering a live register.
1542 2014-01-25 Walter Lee <walt@tilera.com>
1544 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1545 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1546 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1547 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1549 2014-01-25 Walter Lee <walt@tilera.com>
1551 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1552 arguments on even registers.
1553 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1555 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1556 (BIGGEST_ALIGNMENT): Ditto.
1557 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1559 2014-01-25 Walter Lee <walt@tilera.com>
1561 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1562 insns before bundling.
1563 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1565 2014-01-25 Walter Lee <walt@tilera.com>
1567 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1568 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1569 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1571 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1573 * config/mips/constraints.md (kl): Delete.
1574 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1575 define expands, using...
1576 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1577 instructions for MIPS16.
1578 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1579 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1581 2014-01-25 Walter Lee <walt@tilera.com>
1583 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1587 2014-01-25 Walter Lee <walt@tilera.com>
1589 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1590 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1592 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1594 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1597 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1599 * print-rtl.c (in_call_function_usage): New var.
1600 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1601 EXPR_LIST mode as mode and not as reg note name.
1604 * cfgloopmanip.c (copy_loop_info): If
1605 loop->warned_aggressive_loop_optimizations, make sure
1606 the flag is set in target loop too.
1608 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1610 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1612 * builtins.def: Likewise.
1613 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1614 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1615 * ira.c (ira_setup_eliminable_regset): Likewise.
1616 * omp-low.c (gate_expand_omp): Likewise.
1617 (execute_lower_omp): Likewise.
1618 (diagnose_sb_0): Likewise.
1619 (gate_diagnose_omp_blocks): Likewise.
1620 (simd_clone_clauses_extract): Likewise.
1623 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1625 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1626 correction for little endian...
1627 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1630 2014-01-24 Jeff Law <law@redhat.com>
1632 PR tree-optimization/59919
1633 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1634 for non-returning calls.
1636 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1638 * common/config/aarch64/aarch64-common.c
1639 (aarch64_rewrite_mcpu): Handle multiple names.
1640 * config/aarch64/aarch64.h
1641 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1643 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1645 * input.c (add_file_to_cache_tab): Handle the case where fopen
1648 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1651 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1652 from push operand if code of push isn't PRE_DEC.
1654 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1657 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1658 -mquad-memory-atomic. Update -mquad-memory documentation to say
1659 it is only used for non-atomic loads/stores.
1661 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1662 -mquad-memory or -mquad-memory-atomic switches.
1664 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1665 -mquad-memory-atomic to ISA 2.07 support.
1667 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1668 to separate support of normal quad word memory operations (ldq, stq)
1669 from the atomic quad word memory operations.
1671 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1672 support to separate non-atomic quad word operations from atomic
1673 quad word operations. Disable non-atomic quad word operations in
1674 little endian mode so that we don't have to swap words after the
1675 load and before the store.
1676 (quad_load_store_p): Add comment about atomic quad word support.
1677 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1678 options printed with -mdebug=reg.
1680 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1681 -mquad-memory-atomic as the test for whether we have quad word
1682 atomic instructions.
1683 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1684 or -mp8-vector are used, allow byte/half-word atomic operations.
1686 * config/rs6000/sync.md (load_lockedti): Insure that the address
1687 is a proper indexed or indirect address for the lqarx instruction.
1688 On little endian systems, swap the hi/lo registers after the lqarx
1690 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1691 insure the address is valid for the lqarx instruction.
1692 (store_conditionalti): Insure that the address is a proper indexed
1693 or indirect address for the stqcrx. instruction. On little endian
1694 systems, swap the hi/lo registers before doing the stqcrx.
1696 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1697 insure the address is valid for the stqcrx. instruction.
1699 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1700 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1701 type of quad memory support is available.
1703 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1706 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1707 there is a danger of looping.
1709 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1711 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1712 force flag_ira_loop_pressure if set via command line.
1714 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1716 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1717 (ashr_simd): New builtin handling DI mode.
1718 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1719 (aarch64_sshr_simddi): New match pattern.
1720 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1721 (vshrd_n_s64): Likewise.
1722 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1724 2014-01-23 Nick Clifton <nickc@redhat.com>
1726 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1727 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1728 favour of mcu specific scripts.
1729 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1732 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1733 Alex Velenko <Alex.Velenko@arm.com>
1735 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1736 (vaddv_s16): Likewise.
1737 (vaddv_s32): Likewise.
1738 (vaddv_u8): Likewise.
1739 (vaddv_u16): Likewise.
1740 (vaddv_u32): Likewise.
1741 (vaddvq_s8): Likewise.
1742 (vaddvq_s16): Likewise.
1743 (vaddvq_s32): Likewise.
1744 (vaddvq_s64): Likewise.
1745 (vaddvq_u8): Likewise.
1746 (vaddvq_u16): Likewise.
1747 (vaddvq_u32): Likewise.
1748 (vaddvq_u64): Likewise.
1749 (vaddv_f32): Likewise.
1750 (vaddvq_f32): Likewise.
1751 (vaddvq_f64): Likewise.
1752 (vmaxv_f32): Likewise.
1753 (vmaxv_s8): Likewise.
1754 (vmaxv_s16): Likewise.
1755 (vmaxv_s32): Likewise.
1756 (vmaxv_u8): Likewise.
1757 (vmaxv_u16): Likewise.
1758 (vmaxv_u32): Likewise.
1759 (vmaxvq_f32): Likewise.
1760 (vmaxvq_f64): Likewise.
1761 (vmaxvq_s8): Likewise.
1762 (vmaxvq_s16): Likewise.
1763 (vmaxvq_s32): Likewise.
1764 (vmaxvq_u8): Likewise.
1765 (vmaxvq_u16): Likewise.
1766 (vmaxvq_u32): Likewise.
1767 (vmaxnmv_f32): Likewise.
1768 (vmaxnmvq_f32): Likewise.
1769 (vmaxnmvq_f64): Likewise.
1770 (vminv_f32): Likewise.
1771 (vminv_s8): Likewise.
1772 (vminv_s16): Likewise.
1773 (vminv_s32): Likewise.
1774 (vminv_u8): Likewise.
1775 (vminv_u16): Likewise.
1776 (vminv_u32): Likewise.
1777 (vminvq_f32): Likewise.
1778 (vminvq_f64): Likewise.
1779 (vminvq_s8): Likewise.
1780 (vminvq_s16): Likewise.
1781 (vminvq_s32): Likewise.
1782 (vminvq_u8): Likewise.
1783 (vminvq_u16): Likewise.
1784 (vminvq_u32): Likewise.
1785 (vminnmv_f32): Likewise.
1786 (vminnmvq_f32): Likewise.
1787 (vminnmvq_f64): Likewise.
1789 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1791 * config/aarch64/aarch64-simd.md
1792 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1793 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1794 (*aarch64_mul3_elt<mode>): Likewise.
1795 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1796 (*aarch64_mul3_elt_to_64v2df): Likewise.
1797 (*aarch64_mla_elt<mode>): Likewise.
1798 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1799 (*aarch64_mls_elt<mode>): Likewise.
1800 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1801 (*aarch64_fma4_elt<mode>): Likewise.
1802 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1803 (*aarch64_fma4_elt_to_64v2df): Likewise.
1804 (*aarch64_fnma4_elt<mode>): Likewise.
1805 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1806 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1807 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1808 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1809 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1810 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1811 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1812 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1813 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1815 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1817 * config/aarch64/aarch64-simd.md
1818 (aarch64_be_checked_get_lane<mode>): New define_expand.
1819 * config/aarch64/aarch64-simd-builtins.def
1820 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1821 New builtin definition.
1822 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1823 Use new safe be builtin.
1825 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1827 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1829 (aarch64_be_st1<mode>): Likewise.
1830 (aarch_ld1<VALL:mode>): Define_expand modified.
1831 (aarch_st1<VALL:mode>): Likewise.
1832 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1833 (UNSPEC_ST1): Likewise.
1835 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1837 * config/microblaze/microblaze.md: Add trap insn and attribute
1839 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1841 PR preprocessor/58580
1842 * input.h (location_get_source_line): Take an additional line_size
1844 (void diagnostics_file_cache_fini): Declare new function.
1845 * input.c (struct fcache): New type.
1846 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1847 New static constants.
1848 (diagnostic_file_cache_init, total_lines_num)
1849 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1850 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1851 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1852 (get_next_line, read_next_line, goto_next_line, read_line_num):
1853 New static function definitions.
1854 (diagnostic_file_cache_fini): New function.
1855 (location_get_source_line): Take an additional output line_len
1856 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1858 * diagnostic.c (diagnostic_finish): Call
1859 diagnostic_file_cache_fini.
1860 (adjust_line): Take an additional input parameter for the length
1861 of the line, rather than calculating it with strlen.
1862 (diagnostic_show_locus): Adjust the use of
1863 location_get_source_line and adjust_line with respect to their new
1864 signature. While displaying a line now, do not stop at the first
1865 null byte. Rather, display the zero byte as a space and keep
1866 going until we reach the size of the line.
1867 * Makefile.in: Add vec.o to OBJS-libcommon
1869 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1870 Ilya Tocar <ilya.tocar@intel.com>
1872 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1873 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1874 (__builtin_ia32_kmov16): Ditto.
1875 * config/i386/i386.md (UNSPEC_KMOV): New.
1878 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1880 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1881 (_mm512_storeu_si512): Ditto.
1883 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1886 * rtl.h (get_referenced_operands): Declare.
1887 * recog.c (get_referenced_operands): New function.
1888 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1889 operands have been referenced when recording LO_SUM references.
1891 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1893 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1895 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1897 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1898 Enable for generic and recent AMD targets.
1900 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1902 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1903 ARG_SIZE note when adjustment was eliminated.
1905 2014-01-22 Jeff Law <law@redhat.com>
1907 PR tree-optimization/59597
1908 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1909 in file. Accept new argument REGISTERING and use it to modify
1910 dump output appropriately.
1911 (register_jump_thread): Corresponding changes.
1912 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1913 thread paths involving joiner blocks. Add code to dump cancelled
1914 jump threading paths.
1916 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1918 PR rtl-optimization/59477
1919 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1920 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1922 2014-01-22 Tom Tromey <tromey@redhat.com>
1924 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1926 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1928 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1930 PR rtl-optimization/59896
1931 * lra-constraints.c (process_alt_operands): Check unused note for
1932 matched operands of insn with no output reloads.
1934 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1936 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1937 (mips_move_from_gpr_cost): Likewise.
1939 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1941 PR rtl-optimization/59858
1942 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1943 ira_class_hard_regs_num.
1944 (process_alt_operands): Increase reject for dying matched operand.
1946 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1949 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1950 smaller than size, perform several stores or loads and stores
1951 at dst + count - size to store or copy all of size bytes, rather
1952 than just last modesize bytes.
1954 2014-01-20 DJ Delorie <dj@redhat.com>
1956 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1957 that CLOBBERs are REGs before propogating their values.
1959 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1962 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1963 (cgraph_inline_failed_type): New function.
1964 * cgraph.h (DEFCIFCODE): Add type.
1965 (cgraph_inline_failed_type_t): New enum.
1966 (cgraph_inline_failed_type): New prototype.
1967 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1968 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1969 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1970 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1971 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1972 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1973 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1974 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1975 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1976 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1977 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1978 OPTIMIZATION_MISMATCH.
1979 * tree-inline.c (expand_call_inline): Emit errors during
1980 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1982 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1985 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1986 mode attribute in insn output.
1988 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1990 * output.h (output_constant): Delete.
1991 * varasm.c (output_constant): Make private.
1993 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1995 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1997 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2000 * tree.h (fold_builtin_strcat): New prototype.
2001 * builtins.c (fold_builtin_strcat): No longer static. Add len
2002 argument, if non-NULL, don't call c_strlen. Optimize
2003 directly into __builtin_memcpy instead of __builtin_strcpy.
2004 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2005 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2007 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2009 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2010 for SImode_address_operand operands, having only a REG argument.
2012 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2014 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2015 loader name using mbig-endian.
2016 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2018 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2020 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2024 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2026 * config/aarch64/aarch64-protos.h
2027 (aarch64_cannot_change_mode_class_ptr): Declare.
2028 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2029 aarch64_cannot_change_mode_class_ptr): New.
2030 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2031 backend hook aarch64_cannot_change_mode_class.
2033 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2035 * common/config/aarch64/aarch64-common.c
2036 (aarch64_handle_option): Don't handle any option order logic here.
2037 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2038 selected_cpu, warn on architecture version mismatch.
2039 (aarch64_override_options): Fix parsing order for option strings.
2041 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2042 Iain Sandoe <iain@codesourcery.com>
2045 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2046 warning. Amend comment to reflect current functionality.
2048 2014-01-20 Richard Biener <rguenther@suse.de>
2051 * builtins.c (fold_builtin_strcat): Remove case better handled
2052 by tree-ssa-strlen.c.
2054 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2056 * config/aarch64/aarch64.opt
2057 (mcpu, march, mtune): Make case-insensitive.
2059 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2062 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2063 if operands[1] is a REG or ZERO_EXTEND of a REG.
2065 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2067 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2069 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2071 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2072 long non-pic millicode calls.
2074 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2076 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2078 2014-01-19 Kito Cheng <kito@0xlab.org>
2080 * builtins.c (expand_movstr): Check movstr expand done or fail.
2082 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2083 H.J. Lu <hongjiu.lu@intel.com>
2086 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2087 to DImode for zero-extended addresses.
2089 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2091 PR rtl-optimization/57763
2092 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2093 on the new indirect jump_insn and increment LABEL_NUSES (label).
2095 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2099 * config.gcc (x86_archs): New variable.
2100 (x86_64_archs): Likewise.
2101 (x86_cpus): Likewise.
2102 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2103 --with-arch/--with-cpu= options.
2104 Support --with-arch=/--with-cpu={nehalem,westmere,
2105 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2107 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2109 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2110 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2112 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2114 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2116 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2119 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2120 clear cpp_get_options (parse_in)->warn_unused_macros for
2121 ix86_target_macros_internal with cpp_define.
2123 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2125 * jump.c (delete_related_insns): Keep (use (insn))s.
2126 * reorg.c (redundant_insn): Check for barriers too.
2128 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2130 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2132 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2134 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2135 call to $$dyncall when TARGET_LONG_CALLS is true.
2137 2014-01-17 Jeff Law <law@redhat.com>
2139 * ree.c (combine_set_extension): Temporarily disable test for
2140 changing number of hard registers.
2142 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2145 * ipa-inline-analysis.c (inline_free_summary):
2146 Do not free summary of aliases.
2148 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2151 * gimplify.c (gimplify_expr): Use create_tmp_var
2152 instead of create_tmp_var_raw. If cond doesn't have
2153 integral type, don't add the IFN_ANNOTATE builtin at all.
2155 2014-01-17 Martin Jambor <mjambor@suse.cz>
2158 * ipa-cp.c (prev_edge_clone): New variable.
2159 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2160 Also resize prev_edge_clone vector.
2161 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2162 (ipcp_edge_removal_hook): New function.
2163 (ipcp_driver): Register ipcp_edge_removal_hook.
2165 2014-01-17 Andrew Pinski <apinski@cavium.com>
2166 Steve Ellcey <sellcey@mips.com>
2169 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2172 2014-01-17 Jeff Law <law@redhat.com>
2175 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2176 so that pass_ccp runs first.
2178 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2180 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2181 (ix86_adjust_cost): Use !TARGET_XXX.
2182 (do_reorder_for_imul): Likewise.
2183 (swap_top_of_ready_list): Likewise.
2184 (ix86_sched_reorder): Likewise.
2186 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2188 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2189 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2190 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2191 (intel_memset): New. Duplicate slm_memset.
2192 (intel_cost): New. Duplicate slm_cost.
2193 (m_INTEL): New macro.
2194 (processor_target_table): Add "intel".
2195 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2196 with PROCESSOR_INTEL for "intel".
2197 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2198 PROCESSOR_SILVERMONT.
2199 (ix86_issue_rate): Likewise.
2200 (ix86_adjust_cost): Likewise.
2201 (ia32_multipass_dfa_lookahead): Likewise.
2202 (swap_top_of_ready_list): Likewise.
2203 (ix86_sched_reorder): Likewise.
2204 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2205 instead of TARGET_OPT_AGU.
2206 * config/i386/i386.h (TARGET_INTEL): New.
2207 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2208 (processor_type): Add PROCESSOR_INTEL.
2209 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2210 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2212 2014-01-17 Marek Polacek <polacek@redhat.com>
2215 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2218 2014-01-17 Richard Biener <rguenther@suse.de>
2220 PR tree-optimization/46590
2221 * opts.c (default_options_table): Add entries for
2222 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2223 all enabled at -O1 but not for -Og.
2224 * common.opt (fbranch-count-reg): Remove Init(1).
2225 (fmove-loop-invariants): Likewise.
2226 (ftree-pta): Likewise.
2228 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2230 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2231 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2232 decls to at least the GCC 4.8 used alignments.
2235 * tree-nested.c (convert_nonlocal_reference_stmt,
2236 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2237 of GIMPLE_BIND stmts, adjust associated decls.
2239 2014-01-17 Richard Biener <rguenther@suse.de>
2241 PR tree-optimization/46590
2242 * vec.h (vec<>::bseach): New member function implementing
2243 binary search according to C89 bsearch.
2244 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2245 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2246 bitmap pointer again. Make accesses_in_loop a flat array.
2247 (mem_ref_obstack): New global.
2248 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2249 (mark_ref_stored): Likewise.
2250 (ref_indep_loop_p_2): Likewise.
2251 (set_ref_stored_in_loop): New helper function.
2252 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2253 (memref_free): Adjust.
2254 (record_mem_ref_loc): Simplify.
2255 (gather_mem_refs_stmt): Adjust.
2256 (sort_locs_in_loop_postorder_cmp): New function.
2257 (analyze_memory_references): Sort accesses_in_loop after
2258 loop postorder number.
2259 (find_ref_loc_in_loop_cmp): New function.
2260 (for_all_locs_in_loop): Find relevant cluster of locs in
2261 accesses_in_loop and iterate without recursion.
2262 (execute_sm): Avoid uninit warning.
2263 (struct ref_always_accessed): Simplify.
2264 (ref_always_accessed::operator ()): Likewise.
2265 (ref_always_accessed_p): Likewise.
2266 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2267 loop postorder numbers here.
2268 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2271 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2274 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2275 on decls for which assemble_alias has been called.
2277 2014-01-17 Nick Clifton <nickc@redhat.com>
2279 * config/msp430/msp430.opt: (mcpu): New option.
2280 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2281 (msp430_option_override): Parse target_cpu. If the MCU name
2282 matches a generic string, clear target_mcu.
2283 (msp430_attr): Allow numeric interrupt values up to 63.
2284 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2285 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2287 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2289 * config/msp430/msp430.md (popm): Use %J rather than %I.
2290 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2291 (addhi_cy_i): Use immediate_operand for operand 2.
2292 * doc/invoke.texi: Document -mcpu option.
2294 2014-01-17 Richard Biener <rguenther@suse.de>
2296 PR rtl-optimization/38518
2297 * df.h (df_analyze_loop): Declare.
2298 * df-core.c: Include cfgloop.h.
2299 (df_analyze_1): Split out main part of df_analyze.
2300 (df_analyze): Adjust.
2301 (loop_inverted_post_order_compute): New function.
2302 (loop_post_order_compute): Likewise.
2303 (df_analyze_loop): New function avoiding whole-function
2305 * loop-invariant.c (find_defs): Use df_analyze_loop.
2306 (find_invariants): Adjust.
2307 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2309 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2311 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2312 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2314 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2316 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2317 traversal when removing references.
2319 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2322 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2324 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2327 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2328 pushing a reload for an autoinc when we had previously reloaded an
2329 inner part of the address.
2331 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2333 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2335 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2336 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2337 when not giving up or versioning for alias only because of
2339 (vect_analyze_data_ref_dependences): Set to true.
2340 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2342 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2343 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2347 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2350 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2351 operand 0 predicate for gathers, use a new pseudo as subtarget.
2353 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2356 * lra-constraints.c (process_alt_operands): Add printing debug info.
2357 Check absence of input/output reloads for matched operands too.
2359 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2361 PR rtl-optimization/59835
2362 * ira.c (ira_init_register_move_cost): Increase cost for
2365 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2367 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2369 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2372 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2373 non-register objects. Use gen_(high/low)part more consistently.
2376 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2379 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2380 endian support, remove tests for WORDS_BIG_ENDIAN.
2381 (p8_mfvsrd_3_<mode>): Likewise.
2382 (reload_gpr_from_vsx<mode>): Likewise.
2383 (reload_gpr_from_vsxsf): Likewise.
2384 (p8_mfvsrd_4_disf): Likewise.
2386 2014-01-16 Richard Biener <rguenther@suse.de>
2388 PR rtl-optimization/46590
2389 * lcm.c (compute_antinout_edge): Use postorder iteration.
2390 (compute_laterin): Use inverted postorder iteration.
2392 2014-01-16 Nick Clifton <nickc@redhat.com>
2395 * varasm.c (output_constant): Return the number of bytes actually
2397 (output_constructor_array_range): Update the field size with the
2398 number of bytes emitted by output_constant.
2399 (output_constructor_regular_field): Likewise. Also do not
2400 complain if the total number of bytes emitted is now greater
2401 than the expected fieldpos.
2402 * output.h (output_constant): Update prototype and descriptive comment.
2404 2014-01-16 Marek Polacek <polacek@redhat.com>
2407 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2408 it is error_mark_node.
2410 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2412 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2413 VALID_AVX256_REG_OR_OI_MODE.
2415 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2417 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2418 current procedure should be profiled.
2420 2014-01-15 Andrew Pinski <apinski@cavium.com>
2422 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2423 of moving from/to the STACK_REG register class.
2425 2014-01-15 Richard Henderson <rth@redhat.com>
2428 * reginfo.c (global_regs_decl): Globalize.
2429 * rtl.h (global_regs_decl): Declare.
2430 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2431 reserved via global_regs.
2433 2014-01-15 Teresa Johnson <tejohnson@google.com>
2435 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2437 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2439 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2440 and vmulosh rather than call gen_vec_widen_smult_*.
2441 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2442 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2443 (vec_widen_smult_even_v16qi): Likewise.
2444 (vec_widen_umult_even_v8hi): Likewise.
2445 (vec_widen_smult_even_v8hi): Likewise.
2446 (vec_widen_umult_odd_v16qi): Likewise.
2447 (vec_widen_smult_odd_v16qi): Likewise.
2448 (vec_widen_umult_odd_v8hi): Likewise.
2449 (vec_widen_smult_odd_v8hi): Likewise.
2450 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2451 vmuloub rather than call gen_vec_widen_umult_*.
2452 (vec_widen_umult_lo_v16qi): Likewise.
2453 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2454 vmulosb rather than call gen_vec_widen_smult_*.
2455 (vec_widen_smult_lo_v16qi): Likewise.
2456 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2457 rather than call gen_vec_widen_umult_*.
2458 (vec_widen_umult_lo_v8hi): Likewise.
2459 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2460 rather than call gen_vec_widen_smult_*.
2461 (vec_widen_smult_lo_v8hi): Likewise.
2463 2014-01-15 Jeff Law <law@redhat.com>
2465 PR tree-optimization/59747
2466 * ree.c (find_and_remove_re): Properly handle case where a second
2467 eliminated extension requires widening a copy created for elimination
2468 of a prior extension.
2469 (combine_set_extension): Ensure that the number of hard regs needed
2470 for a destination register does not change when we widen it.
2472 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2474 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2475 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2476 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2477 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2478 (avr-*-rtems*): Likewise.
2479 (bfin*-rtems*): Likewise.
2480 (moxie-*-rtems*): Likewise.
2481 (h8300-*-rtems*): Likewise.
2482 (i[34567]86-*-rtems*): Likewise.
2483 (lm32-*-rtems*): Likewise.
2484 (m32r-*-rtems*): Likewise.
2485 (m68k-*-rtems*): Likewise.
2486 (microblaze*-*-rtems*): Likewise.
2487 (mips*-*-rtems*): Likewise.
2488 (powerpc-*-rtems*): Likewise.
2489 (sh-*-rtems*): Likewise.
2490 (sparc-*-rtems*): Likewise.
2491 (sparc64-*-rtems*): Likewise.
2492 (v850-*-rtems*): Likewise.
2493 (m32c-*-rtems*): Likewise.
2495 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2497 PR rtl-optimization/59511
2498 * ira.c (ira_init_register_move_cost): Use memory costs for some
2499 cases of register move cost calculations.
2500 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2501 instead of BB frequency.
2502 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2503 * lra-assigns.c (find_hard_regno_for): Ditto.
2505 2014-01-15 Richard Biener <rguenther@suse.de>
2507 PR tree-optimization/59822
2508 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2509 (vectorizable_load): Use it to hoist defs of uses of invariant
2510 loads out of the loop.
2512 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2513 Kugan Vivekanandarajah <kuganv@linaro.org>
2516 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2519 2014-01-15 Richard Biener <rguenther@suse.de>
2521 PR rtl-optimization/59802
2522 * lcm.c (compute_available): Use inverted postorder to seed
2523 the initial worklist.
2525 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2528 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2529 ADDR_REGS for invalid symrefs in non-PIC code.
2531 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2534 * builtins.c (determine_block_size): Initialize *probable_max_size
2535 even if len_rtx is CONST_INT.
2537 2014-01-14 Andrew Pinski <apinski@cavium.com>
2539 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2540 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2541 (cortexa53_tunings): Likewise.
2542 (aarch64_sched_issue_rate): New function.
2543 (TARGET_SCHED_ISSUE_RATE): Define.
2545 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2547 * ira-costs.c (find_costs_and_classes): Add missed
2548 ira_init_register_move_cost_if_necessary.
2550 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2553 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2555 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2558 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2559 to indicate if type is used for function return value. Warn ABI
2560 change if the vector mode isn't available for function return value.
2561 (ix86_function_arg_advance): Pass false to type_natural_mode.
2562 (ix86_function_arg): Likewise.
2563 (ix86_gimplify_va_arg): Likewise.
2564 (function_arg_32): Don't warn ABI change.
2565 (ix86_function_value): Pass true to type_natural_mode.
2566 (ix86_return_in_memory): Likewise.
2567 (ix86_struct_value_rtx): Removed.
2568 (TARGET_STRUCT_VALUE_RTX): Likewise.
2570 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2572 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2573 converting a conditional jump into a conditional return.
2575 2014-01-14 Richard Biener <rguenther@suse.de>
2577 PR tree-optimization/58921
2578 PR tree-optimization/59006
2579 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2580 hoisting invariant stmts.
2581 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2582 invariant loads on the preheader edge if possible.
2584 2014-01-14 Joey Ye <joey.ye@arm.com>
2586 * doc/plugin.texi (Building GCC plugins): Update to C++.
2588 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2590 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2591 (_mm_rcp28_round_ss): Ditto.
2592 (_mm_rsqrt28_round_sd): Ditto.
2593 (_mm_rsqrt28_round_ss): Ditto.
2594 (_mm_rcp28_sd): Ditto.
2595 (_mm_rcp28_ss): Ditto.
2596 (_mm_rsqrt28_sd): Ditto.
2597 (_mm_rsqrt28_ss): Ditto.
2598 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2599 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2600 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2601 (IX86_BUILTIN_RCP28SD): Ditto.
2602 (IX86_BUILTIN_RCP28SS): Ditto.
2603 (IX86_BUILTIN_RSQRT28SD): Ditto.
2604 (IX86_BUILTIN_RSQRT28SS): Ditto.
2605 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2606 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2607 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2608 (ix86_expand_special_args_builtin): Expand new FTYPE.
2609 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2610 (srcp14<mode>): Make insn unary.
2611 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2612 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2613 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2614 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2615 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2616 Fix rounding: make it SAE only.
2617 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2619 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2621 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2622 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2623 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2624 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2625 (round_saeonly_mask_scalar_operand4): Ditto.
2626 (round_saeonly_mask_scalar_op3): Ditto.
2627 (round_saeonly_mask_scalar_op4): Ditto.
2629 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2631 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2632 Implement -maltivec=be for vec_insert and vec_extract.
2634 2014-01-10 DJ Delorie <dj@redhat.com>
2636 * config/msp430/msp430.md (call_internal): Don't allow memory
2637 references with SP as the base register.
2638 (call_value_internal): Likewise.
2639 * config/msp430/constraints.md (Yc): New. For memory references
2640 that don't use SP as a base register.
2642 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2643 "an integer without a # prefix"
2644 * config/msp430/msp430.md (epilogue_helper): Use it.
2646 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2649 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2650 AVX512F gather builtins.
2651 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2652 on gather decls with INTEGER_TYPE masktype.
2653 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2654 directly into the builtin rather than hoisting it before loop.
2656 PR tree-optimization/59387
2657 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2658 (scev_const_prop): If folded_casts and type has undefined overflow,
2659 use force_gimple_operand instead of force_gimple_operand_gsi and
2660 for each added stmt if it is assign with
2661 arith_code_with_undefined_signed_overflow, call
2662 rewrite_to_defined_overflow.
2663 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2664 gimple-fold.h instead.
2665 (arith_code_with_undefined_signed_overflow,
2666 rewrite_to_defined_overflow): Moved to ...
2667 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2668 rewrite_to_defined_overflow): ... here. No longer static.
2669 Include gimplify-me.h.
2670 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2671 rewrite_to_defined_overflow): New prototypes.
2673 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2675 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2677 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2679 * builtins.c (get_object_alignment_2): Minor tweak.
2680 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2682 2014-01-13 Christian Bruel <christian.bruel@st.com>
2684 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2685 optimized non constant lengths.
2687 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2690 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2691 load as __atomic_load_N if possible.
2693 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2695 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2697 (rs6000_expand_builtin): Adjust call.
2699 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2702 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2703 * config/rs6000/rs6000.c: Include target-globals.h.
2704 (rs6000_set_current_function): Instead of doing target_reinit
2705 unconditionally, use save_target_globals_default_opts and
2706 restore_target_globals.
2708 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2710 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2711 (rs6000_expand_builtin): Handle mffs and mtfsf.
2712 (rs6000_init_builtins): Define mffs and mtfsf.
2713 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2714 (rs6000_mffs): New pattern.
2715 (rs6000_mtfsf): New pattern.
2717 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2719 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2720 Start narrowing with START. Apply candidate-use pair
2721 and check overall cost in narrowing.
2722 (iv_ca_prune): Pass new argument.
2724 2014-01-10 Jeff Law <law@redhat.com>
2727 * ree.c (combine_reaching_defs): Ensure the defining statement
2728 occurs before the extension when optimizing extensions with
2729 different source and destination hard registers.
2731 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2734 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2735 vtables into the type inheritance graph.
2737 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2739 PR rtl-optimization/59754
2740 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2741 modes in the REGNO != REGNO case.
2743 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2745 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2747 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2749 PR tree-optimization/59745
2750 * tree-predcom.c (tree_predictive_commoning_loop): Call
2751 free_affine_expand_cache if giving up because components is NULL.
2753 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2754 GC in payload of target_globals struct instead of allocating them on
2755 the heap and the larger structs separately using GC.
2756 * target-globals.h (struct target_globals): Make regs, hard_regs,
2757 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2758 of GTY((skip)) and change type to void *.
2759 (reset_target_globals): Cast loads from those fields to corresponding
2762 2014-01-10 Steve Ellcey <sellcey@mips.com>
2765 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2766 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2767 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2769 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2772 * aarch64-modes.def (CC_Zmode): New flags mode.
2773 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2774 represents an equality.
2775 (aarch64_get_condition_code): Handle CC_Zmode.
2776 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2778 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2780 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2781 extraction in good case.
2783 2014-01-10 Richard Biener <rguenther@suse.de>
2785 PR tree-optimization/59374
2786 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2787 checking after SLP discovery. Mark stmts not participating
2788 in any SLP instance properly.
2790 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2792 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2793 when handling a SET rtx.
2795 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2797 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2798 (cortex-a57): Likewise.
2799 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2801 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2803 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2804 non-iwmmxt builtins.
2806 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2810 * ipa-devirt.c record_target_from_binfo): Take as argument
2811 stack of binfos and lookup matching one for virtual inheritance.
2812 (possible_polymorphic_call_targets_1): Update.
2814 2014-01-10 Huacai Chen <chenhc@lemote.com>
2816 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2817 kernel strings for Loongson-2E/2F/3A.
2819 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2822 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2823 is_gimple_call before calling gimple_call_internal_p.
2825 2014-01-09 Steve Ellcey <sellcey@mips.com>
2827 * Makefile.in (TREE_FLOW_H): Remove.
2828 (TREE_SSA_H): Add file names from tree-flow.h.
2829 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2830 * tree.h: Remove tree-flow.h reference.
2831 * hash-table.h: Remove tree-flow.h reference.
2832 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2833 reference with tree-ssa-loop.h.
2835 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2837 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2838 default element-order behavior for -maltivec.
2839 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2841 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2842 when targeting big endian, at least for now.
2843 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2845 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2848 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2849 var satisfies use_register_for_decl, just take into account type
2850 alignment, rather than decl alignment.
2852 PR tree-optimization/59622
2853 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2854 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2855 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2856 Don't devirtualize for inplace at all. For targets.length () == 1,
2857 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2859 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2861 * config/i386/i386.md (cpu): Remove the unused btver1.
2863 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2865 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2867 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2870 * tree-core.h (struct target_globals): New forward declaration.
2871 (struct tree_target_option): Add globals field.
2872 * tree.h (TREE_TARGET_GLOBALS): Define.
2873 (prepare_target_option_nodes_for_pch): New prototype.
2874 * target-globals.h (struct target_globals): Define even if
2876 * tree.c (prepare_target_option_node_for_pch,
2877 prepare_target_option_nodes_for_pch): New functions.
2878 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2879 * config/i386/i386.c: Include target-globals.h.
2880 (ix86_set_current_function): Instead of doing target_reinit
2881 unconditionally, use save_target_globals_default_opts and
2882 restore_target_globals.
2884 2014-01-09 Richard Biener <rguenther@suse.de>
2886 PR tree-optimization/59715
2887 * tree-cfg.h (split_critical_edges): Declare.
2888 * tree-cfg.c (split_critical_edges): Export.
2889 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2891 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2893 * cfgexpand.c (expand_stack_vars): Optionally disable
2894 asan stack protection.
2895 (expand_used_vars): Likewise.
2896 (partition_stack_vars): Likewise.
2897 * asan.c (asan_emit_stack_protection): Optionally disable
2898 after return stack usage.
2899 (instrument_derefs): Optionally disable memory access instrumentation.
2900 (instrument_builtin_call): Likewise.
2901 (instrument_strlen_call): Likewise.
2902 (asan_protect_global): Optionally disable global variables protection.
2903 * doc/invoke.texi: Added doc for new options.
2904 * params.def: Added new options.
2905 * params.h: Likewise.
2907 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2909 PR rtl-optimization/59724
2910 * ifcvt.c (cond_exec_process_if_block): Don't call
2911 flow_find_head_matching_sequence with 0 longest_match.
2912 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2913 non-active insns if !stop_after.
2914 (try_head_merge_bb): Revert 2014-01-07 changes.
2916 2014-01-08 Jeff Law <law@redhat.com>
2918 * ree.c (get_sub_rtx): New function, extracted from...
2919 (merge_def_and_ext): Here.
2920 (combine_reaching_defs): Use get_sub_rtx.
2922 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2924 * cgraph.h (varpool_variable_node): Do not choke on null node.
2926 2014-01-08 Catherine Moore <clm@codesourcery.com>
2928 * config/mips/mips.md (simple_return): Attempt to use JRC
2930 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2932 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2934 PR rtl-optimization/59137
2935 * reorg.c (steal_delay_list_from_target): Call update_block for
2937 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2939 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2941 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2942 two duplicate entries.
2944 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2947 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2949 * config/mips/mips.c (mips_truncated_op_cost): New function.
2950 (mips_rtx_costs): Adjust test for BADDU.
2951 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2953 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2955 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2956 (*baddu_si): ...this new pattern.
2958 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2961 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2963 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2966 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2968 (expand_expr, expand_normal): Adjust.
2969 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2970 inner_reference_p. Use inner_reference_p to expand inner references.
2971 (store_expr): Adjust.
2972 * cfgexpand.c (expand_call_stmt): Adjust.
2974 2014-01-08 Rong Xu <xur@google.com>
2976 * gcov-io.c (gcov_var): Move from gcov-io.h.
2977 (gcov_position): Ditto.
2978 (gcov_is_error): Ditto.
2979 (gcov_rewrite): Ditto.
2980 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2981 only part to libgcc/libgcov.h.
2983 2014-01-08 Marek Polacek <polacek@redhat.com>
2986 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2988 2014-01-08 Marek Polacek <polacek@redhat.com>
2991 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2993 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2995 PR rtl-optimization/59649
2996 * stor-layout.c (get_mode_bounds): For BImode return
2997 0 and STORE_FLAG_VALUE.
2999 2014-01-08 Richard Biener <rguenther@suse.de>
3002 * gimple.h (is_gimple_builtin_call): Remove.
3003 (gimple_builtin_call_types_compatible_p): New.
3004 (gimple_call_builtin_p): New overload.
3005 * gimple.c (is_gimple_builtin_call): Remove.
3006 (validate_call): Rename to ...
3007 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3009 (validate_type): New static function.
3010 (gimple_call_builtin_p): New overload and adjust.
3011 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3012 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3013 (gimple_fold_stmt_to_constant_1): Likewise.
3014 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3016 2014-01-08 Richard Biener <rguenther@suse.de>
3019 * gimplify.c (gimplify_expr): Gimplify register-register type
3020 VIEW_CONVERT_EXPRs to separate stmts.
3022 2014-01-07 Jeff Law <law@redhat.com>
3025 * ree.c (combine_set_extension): Handle case where source
3026 and destination registers in an extension insn are different.
3027 (combine_reaching_defs): Allow source and destination registers
3028 in extension to be different under limited circumstances.
3029 (add_removable_extension): Remove restriction that the
3030 source and destination registers in the extension are the same.
3031 (find_and_remove_re): Emit a copy from the extension's
3032 destination to its source after the defining insn if
3033 the source and destination registers are different.
3036 * ifcvt.c (merge_if_block): If we are merging a block with more than
3037 one successor with a block with no successors, remove any BARRIER
3038 after the second block.
3040 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3042 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3044 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3047 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3048 for 14-bit register offsets when INT14_OK_STRICT is false.
3050 2014-01-07 Roland Stigge <stigge@antcom.de>
3051 Michael Meissner <meissner@linux.vnet.ibm.com>
3054 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3055 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3057 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3059 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3062 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3064 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3065 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3066 rtx is const0_rtx or not.
3068 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3071 * target-globals.c (save_target_globals): Remove this_fn_optab
3073 * toplev.c: Include optabs.h.
3074 (target_reinit): Temporarily restore the global options if another
3075 set of options are in force.
3077 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3079 PR rtl-optimization/58668
3080 * cfgcleanup.c (flow_find_cross_jump): Don't count
3081 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3082 to determine what is counted.
3083 (flow_find_head_matching_sequence): Use active_insn_p to determine
3085 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3087 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3088 determine what is counted.
3090 PR tree-optimization/59643
3091 * tree-predcom.c (split_data_refs_to_components): If one dr is
3092 read and one write, determine_offset fails and the write isn't
3093 in the bad component, just put the read into the bad component.
3095 2014-01-07 Mike Stump <mikestump@comcast.net>
3096 Jakub Jelinek <jakub@redhat.com>
3099 * tree-core.h (struct tree_optimization_option): Change optabs
3100 type from unsigned char * to void *.
3101 * optabs.c (init_tree_optimization_optabs): Adjust
3102 TREE_OPTIMIZATION_OPTABS initialization.
3104 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3107 * config/i386/i386.h (struct machine_function): Add
3108 no_drap_save_restore field.
3109 * config/i386/i386.c (ix86_save_reg): Use
3110 !cfun->machine->no_drap_save_restore instead of
3111 crtl->stack_realign_needed.
3112 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3113 this function clears frame_pointer_needed. Set
3114 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3115 and DRAP reg is needed.
3117 2014-01-06 Marek Polacek <polacek@redhat.com>
3120 * doc/implement-c.texi: Mention that other integer types are
3121 permitted as bit-field types in strictly conforming mode.
3123 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3125 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3128 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3130 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3132 2014-01-06 Martin Jambor <mjambor@suse.cz>
3135 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3137 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3139 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3143 * var-tracking.c (add_stores): Preserve the value of the source even if
3144 we don't record the store.
3146 2014-01-06 Terry Guo <terry.guo@arm.com>
3148 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3150 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3153 * config/darwin.c (darwin_function_section): Adjust return values to
3154 correspond to optimisation changes made in r206070.
3156 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3158 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3159 from prefetch_block tune setting.
3160 (nocona_cost): Correct size of prefetch block to 64.
3162 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3164 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3165 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3166 used to save the static chain register in the computation of the offset
3167 from which the FP registers need to be restored.
3169 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3171 PR tree-optimization/59519
3172 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3173 ICE if get_current_def (current_new_name) is already non-NULL, as long
3174 as it is a phi result of some other phi in *new_exit_bb that has
3177 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3178 or vmovdqu* for misaligned_operand.
3179 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3180 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3181 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3182 aligned_mem for AVX512F masked aligned load and store builtins and for
3185 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3187 PR tree-optimization/59651
3188 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3189 Address range for negative step should be added by TYPE_SIZE_UNIT.
3191 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3193 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3194 overlapping registers also for registers other than the stack pointer.
3196 2014-01-03 Marek Polacek <polacek@redhat.com>
3199 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3202 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3205 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3208 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3209 (push splitter): Use <P:MODE_SIZE> instead of
3210 GET_MODE_SIZE (<P:MODE>mode).
3211 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3212 (mov -1, reg peephole2): Likewise.
3213 * config/i386/sse.md (*mov<mode>_internal,
3214 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3215 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3216 *<code><mode>3, *andnot<mode>3<mask_name>,
3217 <mask_codefor><code><mode>3<mask_name>): Likewise.
3218 * config/i386/subst.md (mask_mode512bit_condition,
3219 sd_mask_mode512bit_condition): Likewise.
3221 2014-01-02 Xinliang David Li <davidxl@google.com>
3223 PR tree-optimization/59303
3224 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3225 (dump_predicates): Better output format.
3226 (pred_equal_p): New function.
3227 (is_neq_relop_p): Ditto.
3228 (is_neq_zero_form_p): Ditto.
3229 (pred_expr_equal_p): Ditto.
3230 (pred_neg_p): Ditto.
3231 (simplify_pred): Ditto.
3232 (simplify_preds_2): Ditto.
3233 (simplify_preds_3): Ditto.
3234 (simplify_preds_4): Ditto.
3235 (simplify_preds): Ditto.
3237 (push_to_worklist): Ditto.
3238 (get_pred_info_from_cmp): Ditto.
3239 (is_degenerated_phi): Ditto.
3240 (normalize_one_pred_1): Ditto.
3241 (normalize_one_pred): Ditto.
3242 (normalize_one_pred_chain): Ditto.
3243 (normalize_preds): Ditto.
3244 (normalize_cond_1): Remove function.
3245 (normalize_cond): Ditto.
3246 (is_gcond_subset_of): Ditto.
3247 (is_subset_of_any): Ditto.
3248 (is_or_set_subset_of): Ditto.
3249 (is_and_set_subset_of): Ditto.
3250 (is_norm_cond_subset_of): Ditto.
3251 (pred_chain_length_cmp): Ditto.
3252 (convert_control_dep_chain_into_preds): Type change.
3253 (find_predicates): Ditto.
3254 (find_def_preds): Ditto.
3255 (destroy_predicates_vecs): Ditto.
3256 (find_matching_predicates_in_rest_chains): Ditto.
3257 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3258 (is_pred_expr_subset): Ditto.
3259 (is_pred_chain_subset_of): Ditto.
3260 (is_included_in): Ditto.
3261 (is_superset_of): Ditto.
3263 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3265 Update copyright years.
3267 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3269 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3270 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3271 config/arc/arc.md, config/arc/arc.opt,
3272 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3273 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3274 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3275 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3276 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3277 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3278 vtable-verify.c, vtable-verify.h: Use the standard form for the
3281 2014-01-02 Tobias Burnus <burnus@net-b.de>
3283 * gcc.c (process_command): Update copyright notice dates.
3284 * gcov-dump.c: Ditto.
3286 * doc/cpp.texi: Bump @copying's copyright year.
3287 * doc/cppinternals.texi: Ditto.
3288 * doc/gcc.texi: Ditto.
3289 * doc/gccint.texi: Ditto.
3290 * doc/gcov.texi: Ditto.
3291 * doc/install.texi: Ditto.
3292 * doc/invoke.texi: Ditto.
3294 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3296 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3298 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3300 * config/i386/sse.md (*mov<mode>_internal): Guard
3301 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3303 PR rtl-optimization/59647
3304 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3305 new_rtx into UNSIGNED_FLOAT rtxes.
3307 Copyright (C) 2014 Free Software Foundation, Inc.
3309 Copying and distribution of this file, with or without modification,
3310 are permitted in any medium without royalty provided the copyright
3311 notice and this notice are preserved.