1 2014-02-11 Richard Henderson <rth@redhat.com>
2 Jakub Jelinek <jakub@redhat.com>
5 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
6 around drhs if type conversion to lacc->type is not useless.
8 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
12 (cortex-a57.cortex-a53): Likewise.
13 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
15 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
20 2014-02-11 Renlin Li <Renlin.Li@arm.com>
22 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
23 add_options_for_arm_vfp3.
25 2014-02-11 Jeff Law <law@redhat.com>
28 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
29 object with an undesirable mode.
31 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34 * config/i386/sol2-9.h: New file.
35 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
36 *-*-solaris2.9*): Use it.
38 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
40 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
41 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
43 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
45 * config/microblaze/microblaze.c: Extend mcpu version format
47 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
49 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
51 2014-02-10 Richard Henderson <rth@redhat.com>
54 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
55 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
56 ms-abi vs -mno-accumulate-outgoing-args.
57 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
58 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
61 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
64 * cfgexpand.c (expand_asm_operands): Attach source location to
65 ASM_INPUT rtx objects.
66 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
68 2014-02-10 Nick Clifton <nickc@redhat.com>
70 * config/mn10300/mn10300.c (popcount): New function.
71 (mn10300_expand_prologue): Include saved registers in stack usage
74 2014-02-10 Jeff Law <law@redhat.com>
77 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
78 when changing the SET_DEST of a prior insn to avoid an input
81 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
83 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
84 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
85 -mcall-openbsd, or -mcall-linux.
86 (CC1_ENDIAN_BIG_SPEC): Remove.
87 (CC1_ENDIAN_LITTLE_SPEC): Remove.
88 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
89 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
90 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
91 and %cc1_endian_default.
92 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
94 2014-02-10 Richard Biener <rguenther@suse.de>
96 PR tree-optimization/60115
97 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
98 MEM_REF handling. Properly verify that the accesses are not
99 out of the objects bound.
101 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
103 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
106 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
108 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
109 proper constants and fix formatting.
110 (possible_polymorphic_call_targets): Fix formatting.
112 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
113 Ilya Tocar <ilya.tocar@intel.com>
115 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
116 (_mm512_loadu_epi32): Renamed into...
117 (_mm512_loadu_si512): This.
118 (_mm512_storeu_epi32): Renamed into...
119 (_mm512_storeu_si512): This.
120 (_mm512_maskz_ceil_ps): Removed.
121 (_mm512_maskz_ceil_pd): Ditto.
122 (_mm512_maskz_floor_ps): Ditto.
123 (_mm512_maskz_floor_pd): Ditto.
124 (_mm512_floor_round_ps): Ditto.
125 (_mm512_floor_round_pd): Ditto.
126 (_mm512_ceil_round_ps): Ditto.
127 (_mm512_ceil_round_pd): Ditto.
128 (_mm512_mask_floor_round_ps): Ditto.
129 (_mm512_mask_floor_round_pd): Ditto.
130 (_mm512_mask_ceil_round_ps): Ditto.
131 (_mm512_mask_ceil_round_pd): Ditto.
132 (_mm512_maskz_floor_round_ps): Ditto.
133 (_mm512_maskz_floor_round_pd): Ditto.
134 (_mm512_maskz_ceil_round_ps): Ditto.
135 (_mm512_maskz_ceil_round_pd): Ditto.
136 (_mm512_expand_pd): Ditto.
137 (_mm512_expand_ps): Ditto.
138 * config/i386/i386.c (ix86_builtins): Remove
139 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
141 * config/i386/predicates.md (const1256_operand): New.
142 (const_1_to_2_operand): Ditto.
143 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
144 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
145 (*avx512pf_gatherpf<mode>sf): Ditto.
146 (avx512pf_gatherpf<mode>df): Ditto.
147 (*avx512pf_gatherpf<mode>df_mask): Ditto.
148 (*avx512pf_gatherpf<mode>df): Ditto.
149 (avx512pf_scatterpf<mode>sf): Ditto.
150 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
151 (*avx512pf_scatterpf<mode>sf): Ditto.
152 (avx512pf_scatterpf<mode>df): Ditto.
153 (*avx512pf_scatterpf<mode>df_mask): Ditto.
154 (*avx512pf_scatterpf<mode>df): Ditto.
155 (avx512f_expand<mode>): Removed.
156 (<shift_insn><mode>3<mask_name>): Change predicate type.
158 2014-02-08 Jakub Jelinek <jakub@redhat.com>
160 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
161 not at the end of datarefs vector use ordered_remove to avoid
162 reordering datarefs vector.
165 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
166 mark local addressable non-static vars as GOVD_PRIVATE
167 instead of GOVD_LOCAL.
168 * omp-low.c (lower_omp_for): Move gimple_bind_vars
169 and BLOCK_VARS of gimple_bind_block to new_stmt rather
173 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
174 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
175 assume_aligned or alloc_align attributes.
176 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
177 arguments. Handle also assume_aligned and alloc_align attributes.
178 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
179 Handle calls to functions with assume_aligned or alloc_align
181 * doc/extend.texi: Document assume_aligned and alloc_align
184 2014-02-08 Terry Guo <terry.guo@arm.com>
186 * doc/invoke.texi: Document ARM -march=armv7e-m.
188 2014-02-08 Jakub Jelinek <jakub@redhat.com>
190 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
191 flag on __cilkrts_rethrow builtin.
194 * ipa-cp.c (determine_versionability): Fail at -O0
195 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
196 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
199 2014-02-04 Jakub Jelinek <jakub@redhat.com>
202 * tree-inline.c (copy_forbidden): Fail for
203 __attribute__((optimize (0))) functions.
205 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
207 * varpool.c: Include pointer-set.h.
208 (varpool_remove_unreferenced_decls): Variables in other partitions
209 will not be output; be however careful to not lose information
212 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
214 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
215 lookup in the vtable constructor.
217 2014-02-07 Jeff Law <law@redhat.com>
220 * config/m68k/m68k.md (ashldi_extsi): Turn into a
221 define_insn_and_split.
223 * ipa-inline.c (inline_small_functions): Fix typos.
225 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
227 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
228 (s390_can_use_return_insn): Declare.
229 * config/s390/s390.h (EPILOGUE_USES): Define.
230 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
232 (s390_chunkify_start): Handle return JUMP_LABELs.
233 (s390_early_mach): Emit a main_pool instruction on the entry edge.
234 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
235 (s390_can_use_return_insn): New functions.
236 (s390_fix_long_loop_prediction): Handle conditional returns.
237 (TARGET_SET_UP_BY_PROLOGUE): Define.
238 * config/s390/s390.md (ANY_RETURN): New code iterator.
239 (*creturn, *csimple_return, return, simple_return): New patterns.
241 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
243 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
244 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
245 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
246 REG_CFA_RESTORE list when deciding not to restore a register.
248 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
250 * config/s390/s390.c: Include tree-pass.h and context.h.
251 (s390_early_mach): New function, split out from...
252 (s390_emit_prologue): ...here.
253 (pass_data_s390_early_mach): New pass structure.
254 (pass_s390_early_mach): New class.
255 (s390_option_override): Create and register early_mach pass.
258 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
260 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
261 to match for the exit block.
263 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
265 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
266 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
267 Reject misaligned operands.
269 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
271 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
274 2014-02-07 Richard Biener <rguenther@suse.de>
277 * gimple-low.c (lower_builtin_posix_memalign): New function.
278 (lower_stmt): Call it to lower posix_memalign in a way
279 to make alignment info accessible.
281 2014-02-07 Jakub Jelinek <jakub@redhat.com>
284 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
285 __builtin_setjmp_receiver.
287 2014-02-07 Richard Biener <rguenther@suse.de>
290 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
291 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
292 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
293 Handle BUILT_IN_POSIX_MEMALIGN.
294 (find_func_clobbers): Likewise.
295 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
296 (call_may_clobber_ref_p_1): Likewise.
298 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
301 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
303 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
306 * lto-cgraph.c (lto_output_node): Use
307 symtab_get_symbol_partitioning_class.
308 (lto_output_varpool_node): likewise.
309 (symtab_get_symbol_partitioning_class): Move here from
311 * cgraph.h (symbol_partitioning_class): Likewise.
312 (symtab_get_symbol_partitioning_class): Declare.
314 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
316 * ggc.h (ggc_internal_cleared_alloc): New macro.
317 * vec.h (vec_safe_copy): Handle memory stats.
318 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
319 * target-globals.c (save_target_globals): Likewise.
321 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
324 * expr.c (emit_move_resolve_push): Export; be bit more selective
325 on when to clear alias set.
326 * expr.h (emit_move_resolve_push): Declare.
327 * function.h (struct function): Add tail_call_marked.
328 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
329 * config/i386/i386-protos.h (ix86_expand_push): Remove.
330 * config/i386/i386.md (TImode move expander): De not call
332 (FP push expanders): Preserve memory attributes.
333 * config/i386/sse.md (push<mode>1): Remove.
334 * config/i386/i386.c (ix86_expand_vector_move): Handle push
336 (ix86_expand_push): Remove.
337 * config/i386/mmx.md (push<mode>1): Remove.
339 2014-02-06 Jakub Jelinek <jakub@redhat.com>
341 PR rtl-optimization/60030
342 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
343 lopart with paradoxical subreg before shifting it up by hprec.
345 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
347 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
348 Remove extra newline at end of file.
349 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
350 (arm_issue_rate): Handle cortexa57.
351 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
352 (cortex-a57.cortex-a53): Likewise.
354 2014-02-06 Jakub Jelinek <jakub@redhat.com>
357 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
358 don't record in REG_FRAME_RELATED_EXPR registers not set in that
360 (arm_expand_prologue): Adjust all callers.
361 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
362 info, registers also at the lowest numbered registers side. Use
363 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
367 * var-tracking.c (adjust_mems): Before adding a SET
368 to amd->side_effects, adjust it's SET_SRC using
369 simplify_replace_fn_rtx.
371 2014-02-06 Alan Modra <amodra@gmail.com>
374 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
375 change SDmode to DDmode when lra_in_progress.
377 2014-02-06 Jakub Jelinek <jakub@redhat.com>
380 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
381 free_data_ref on the dr first, and before goto again also set dr
382 to the next dr. For simd_lane_access, free old datarefs[i] before
383 overwriting it. For get_vectype_for_scalar_type failure, don't
384 free_data_ref if simd_lane_access.
386 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
389 * tree.h (opts_for_fn): New inline function.
390 (opt_for_fn): Define.
391 * config/i386/i386.c (ix86_function_regparm): Use
392 opt_for_fn (decl, optimize) instead of optimize.
394 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
396 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
397 for SYMBOL_REF in large memory model.
399 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
401 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
403 (cortex-a57): Likewise.
404 (cortex-a57.cortex-a53): Likewise.
406 2014-02-06 Yury Gribov <y.gribov@samsung.com>
407 Kugan Vivekanandarajah <kuganv@linaro.org>
409 * config/arm/arm.c (arm_vector_alignment_reachable): Check
411 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
413 2014-02-06 Richard Biener <rguenther@suse.de>
415 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
416 set_loop_copy and initialize_original_copy_tables.
418 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
420 * config/aarch64/aarch64-simd.md
421 (aarch64_ashr_simddi): Change QI to SI.
423 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
424 Jakub Jelinek <jakub@redhat.com>
427 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
430 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
432 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
433 CODE_FOR_altivec_vpku[hw]um to
434 CODE_FOR_altivec_vpku[hw]um_direct.
435 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
436 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
437 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
438 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
440 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
442 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
443 generation for -maltivec=be.
444 (altivec_vsumsws): Simplify redundant test.
446 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
448 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
449 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
450 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
451 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
453 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
455 (altivec_vpks<VI_char>ss): Likewise.
456 (altivec_vpks<VI_char>us): Likewise.
457 (altivec_vpku<VI_char>us): Likewise.
458 (altivec_vpku<VI_char>um): Likewise.
459 (altivec_vpku<VI_char>um_direct): New (copy of
460 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
462 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
463 target is little endian and -maltivec=be is not specified.
464 (*altivec_vupkhs<VU_char>_direct): New (copy of
465 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
467 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
468 target is little endian and -maltivec=be is not specified.
469 (*altivec_vupkls<VU_char>_direct): New (copy of
470 altivec_vupkls<VU_char> that always emits vupkls*, for internal
472 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
473 little endian and -maltivec=be is not specified.
474 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
475 little endian and -maltivec=be is not specified.
477 2014-02-05 Richard Henderson <rth@redhat.com>
480 * combine-stack-adj.c: Revert r206943.
481 * sched-int.h (struct deps_desc): Add last_args_size.
482 * sched-deps.c (init_deps): Initialize it.
483 (sched_analyze_insn): Add OUTPUT dependencies between insns that
484 contain REG_ARGS_SIZE notes.
486 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
488 * lto-cgraph.c (asm_nodes_output): Make global.
489 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
490 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
492 (driver_handle_option): Handle OPT_fwpa.
494 2014-02-05 Jakub Jelinek <jakub@redhat.com>
497 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
498 a comment typo and formatting issue. If odr_hash hasn't been
499 created, return vNULL and set *completep to false.
502 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
503 bb with no successors.
505 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
508 * doc/invoke.texi (-march): Clarify documentation for ARM.
512 2014-02-05 Richard Biener <rguenther@suse.de>
514 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
515 when not vectorizing because of too many alias checks.
516 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
517 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
519 2014-02-05 Nick Clifton <nickc@redhat.com>
521 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
522 accept extended registers in any mode when compiling for the
525 2014-02-05 Yury Gribov <y.gribov@samsung.com>
527 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
528 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
529 sanitization attributes.
530 (can_inline_edge_p): Likewise.
531 (sanitize_attrs_match_for_inline_p): New function.
533 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
535 * ipa-prop.c (detect_type_change): Shor circuit testing of
536 type changes on THIS pointer.
538 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
541 * config/pa/pa.c (legitimize_tls_address): Return original address
542 if not passed a SYMBOL_REF rtx.
543 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
545 (pa_emit_move_sequence): Simplify TLS source operands.
546 (pa_legitimate_constant_p): Reject all TLS constants.
547 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
548 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
550 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
552 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
553 groups when we know they are controlled by LTO.
554 * varasm.c (default_binds_local_p_1): If object is in other partition,
555 it will be resolved locally.
557 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
559 * config/host-linux.c (linux_gt_pch_use_address): Don't
560 use SSIZE_MAX because it is not always defined.
562 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
565 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
566 threshold for pseudo splitting.
567 (update_ebb_live_info): Process call argument hard registers and
568 hard registers from insn definition too.
569 (max_small_class_regs_num): New constant.
570 (inherit_in_ebb): Update live hard regs through EBBs. Update
571 reloads_num only for small register classes. Don't split for
574 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
577 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
580 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
582 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
584 2014-02-04 Marek Polacek <polacek@redhat.com>
586 * gdbinit.in (pel): Define.
588 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
590 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
593 2014-02-04 Richard Biener <rguenther@suse.de>
596 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
597 in function context local.
598 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
599 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
600 similar to LTO_imported_decl_ref.
602 2014-02-04 Jakub Jelinek <jakub@redhat.com>
604 PR tree-optimization/60002
605 * cgraphclones.c (build_function_decl_skip_args): Clear
608 PR tree-optimization/60023
609 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
610 false to gsi_replace.
611 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
612 has been in some EH region and vec_stmt could throw, add
613 vec_stmt into the same EH region.
614 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
615 has no lhs, ignore it.
616 * internal-fn.c (expand_MASK_LOAD): Likewise.
619 * tree-inline.c (copy_forbidden): Fail for
620 __attribute__((optimize (0))) functions.
623 * omp-low.c (simd_clone_struct_copy): If from->inbranch
624 is set, copy one less argument.
625 (expand_simd_clones): Don't subtract clone_info->inbranch
626 from simd_clone_struct_alloc argument.
628 PR rtl-optimization/57915
629 * recog.c (simplify_while_replacing): If all unary/binary/relational
630 operation arguments are constant, attempt to simplify those.
633 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
634 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
636 2014-02-04 Richard Biener <rguenther@suse.de>
638 PR tree-optimization/60012
639 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
640 TBAA disambiguation to all DDRs.
642 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
645 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
646 (LINK_SPEC): Use it for -shared, -shared-libgcc.
648 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
651 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
653 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
655 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
656 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
658 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
661 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
662 to figure out targets of polymorphic calls with known decl.
663 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
664 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
665 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
666 (get_polymorphic_call_info): ... here.
667 (get_polymorphic_call_info_from_invariant): New function.
669 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
671 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
672 lookup via vtable pointer; check for type consistency
673 and turn inconsitent facts into UNREACHABLE.
674 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
675 * gimple-fold.c (gimple_get_virt_method_for_vtable):
676 Do not ICE on type inconsistent querries; return UNREACHABLE
679 2014-02-03 Richard Henderson <rth@twiddle.net>
682 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
683 already processed this node.
684 (normalize_one_pred_1): Pass along mark_set.
685 (normalize_one_pred): Create and destroy a pointer_set_t.
686 (normalize_one_pred_chain): Likewise.
688 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
690 PR gcov-profile/58602
691 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
693 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
696 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
697 Try to devirtualize by the knowledge of virtual table pointer given by
698 aggregate propagation.
699 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
700 ipa_print_node_jump_functions): Dump also offset that
701 is relevant for polymorphic calls.
702 (determine_known_aggregate_parts): Add arg_type parameter; use it
703 instead of determining the type from pointer type.
704 (ipa_compute_jump_functions_for_edge): Update call of
705 determine_known_aggregate_parts.
706 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
707 (gimple_get_virt_method_for_binfo): ... here; simplify using
708 vtable_pointer_value_to_vtable.
709 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
710 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
711 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
712 (vtable_pointer_value_to_vtable): Break out from ...; handle also
714 (vtable_pointer_value_to_binfo): ... here.
715 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
717 2014-02-03 Teresa Johnson <tejohnson@google.com>
719 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
720 redef of outer loop index variable.
722 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
726 * doc/extend.texi (Function Attributes): Typo.
728 2014-02-03 Cong Hou <congh@google.com>
730 PR tree-optimization/60000
731 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
732 if the vectorized statement is a store. A store statement can only
733 appear at the end of pattern statements.
735 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
737 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
738 (ix86_option_override_internal): Default long double to 64-bit for
739 32-bit Bionic and to 128-bit for 64-bit Bionic.
741 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
742 TARGET_LONG_DOUBLE_128 is true.
743 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
745 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
746 (mlong-double-64): Negate -mlong-double-128.
747 (mlong-double-128): New option.
749 * config/i386/i386-c.c (ix86_target_macros): Define
750 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
752 * doc/invoke.texi: Document -mlong-double-128.
754 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
756 PR rtl-optimization/60024
757 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
759 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
761 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
763 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
765 PR rtl-optimization/57662
766 * sel-sched.c (code_motion_path_driver): Do not mark already not
767 existing blocks in the visiting bitmap.
769 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
771 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
772 on the insn being emitted.
774 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
775 Will Deacon <will.deacon@arm.com>
777 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
779 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
781 * config/arm/arm-tables.opt: Regenerate.
783 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
785 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
786 for vector types other than V16QImode.
787 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
788 define_expand, and call altivec_expand_vec_perm_le when producing
789 code with little endian element order.
790 (*altivec_vperm_<mode>_internal): New insn having previous
791 behavior of altivec_vperm_<mode>.
792 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
793 altivec_expand_vec_perm_le when producing code with little endian
795 (*altivec_vperm_<mode>_uns_internal): New insn having previous
796 behavior of altivec_vperm_<mode>_uns.
798 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
800 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
801 (altivec_vsumsws): Add handling for -maltivec=be with a little
803 (altivec_vsumsws_direct): New.
804 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
807 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
809 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
810 vtable_pointer_value_to_binfo): New functions.
811 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
812 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
814 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
816 * config/nios2/nios2.md (load_got_register): Initialize GOT
817 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
818 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
820 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
822 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
823 preserverd by passthrough, do not propagate the type.
825 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
827 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
828 (mips_atomic_assign_expand_fenv): New function.
829 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
831 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
833 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
834 (__builtin_mips_set_fcsr): Likewise.
835 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
837 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
838 (mips16_expand_set_fcsr): Likewise.
839 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
840 (mips16_set_fcsr_stub): Likewise.
841 (mips16_get_fcsr_one_only_stub): New class.
842 (mips16_set_fcsr_one_only_stub): Likewise.
843 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
844 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
845 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
846 (hard_float): New availability predicate.
847 (mips_builtins): Add get_fcsr and set_fcsr.
848 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
849 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
850 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
851 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
852 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
855 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
857 * config/mips/mips.c (mips_one_only_stub): New class.
858 (mips_need_mips16_rdhwr_p): Replace with...
859 (mips16_rdhwr_stub): ...this new variable.
860 (mips16_stub_call_address): New function.
861 (mips16_rdhwr_one_only_stub): New class.
862 (mips_expand_thread_pointer): Use mips16_stub_call_address.
863 (mips_output_mips16_rdhwr): Delete.
864 (mips_finish_stub): New function.
865 (mips_code_end): Use it to handle rdhwr stubs.
867 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
870 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
871 when calculating size of integer atomic types.
873 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
875 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
877 2014-02-01 Jakub Jelinek <jakub@redhat.com>
879 PR tree-optimization/60003
880 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
881 * profile.c (branch_prob): Use gimple_call_builtin_p
882 to check for BUILT_IN_SETJMP_RECEIVER.
883 * tree-inline.c (copy_bb): Call notice_special_calls.
885 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
888 * lra-constraints.c (process_alt_operands): Update reload_sum only
891 2014-01-31 Richard Henderson <rth@redhat.com>
894 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
895 until after else_eh is processed.
897 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
899 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
900 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
901 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
902 in smmintrin.h, remove them.
903 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
904 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
905 * config/i386/i386.md (ROUND_SAE): Fix value.
906 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
907 (const48_operand): New.
908 * config/i386/subst.md (round), (round_expand): Use
909 const_4_or_8_to_11_operand.
910 (round_saeonly), (round_saeonly_expand): Use const48_operand.
912 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
914 * config/i386/constraints.md (Yk): Swap meaning with k.
915 * config/i386/i386.md (movhi_internal): Change Yk to k.
916 (movqi_internal): Ditto.
917 (*k<logic><mode>): Ditto.
920 (kandn<mode>): Ditto.
921 (*<code>hi_1): Ditto.
922 (*<code>qi_1): Ditto.
923 (kxnor<mode>): Ditto.
927 (*one_cmplhi2_1): Ditto.
928 (*one_cmplqi2_1): Ditto.
929 * config/i386/sse.md (): Change k to Yk.
930 (avx512f_load<mode>_mask): Ditto.
931 (avx512f_blendm<mode>): Ditto.
932 (avx512f_store<mode>_mask): Ditto.
933 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
934 (avx512f_storedqu<mode>_mask): Ditto.
935 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
937 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
938 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
939 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
940 (avx512f_maskcmp<mode>3): Ditto.
941 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
942 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
943 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
944 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
945 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
946 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
947 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
948 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
949 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
950 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
951 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
952 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
953 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
954 (vec_extract_lo_<mode>_maskm): Ditto.
955 (vec_extract_hi_<mode>_maskm): Ditto.
956 (avx512f_vternlog<mode>_mask): Ditto.
957 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
958 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
959 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
960 (avx512f_<code>v8div16qi2_mask): Ditto.
961 (avx512f_<code>v8div16qi2_mask_store): Ditto.
962 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
963 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
964 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
965 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
966 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
967 (*avx512pf_gatherpf<mode>df_mask): Ditto.
968 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
969 (*avx512pf_scatterpf<mode>df_mask): Ditto.
970 (avx512cd_maskb_vec_dupv8di): Ditto.
971 (avx512cd_maskw_vec_dupv16si): Ditto.
972 (avx512f_vpermi2var<mode>3_maskz): Ditto.
973 (avx512f_vpermi2var<mode>3_mask): Ditto.
974 (avx512f_vpermi2var<mode>3_mask): Ditto.
975 (avx512f_vpermt2var<mode>3_maskz): Ditto.
976 (*avx512f_gathersi<mode>): Ditto.
977 (*avx512f_gathersi<mode>_2): Ditto.
978 (*avx512f_gatherdi<mode>): Ditto.
979 (*avx512f_gatherdi<mode>_2): Ditto.
980 (*avx512f_scattersi<mode>): Ditto.
981 (*avx512f_scatterdi<mode>): Ditto.
982 (avx512f_compress<mode>_mask): Ditto.
983 (avx512f_compressstore<mode>_mask): Ditto.
984 (avx512f_expand<mode>_mask): Ditto.
985 * config/i386/subst.md (mask): Change k to Yk.
986 (mask_scalar_merge): Ditto.
989 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
991 * doc/extend.texi (Vector Extensions): Document ?: in C++.
993 2014-01-31 Richard Biener <rguenther@suse.de>
996 * builtins.c (fold_builtin_memory_op): Make sure to not
997 use a floating-point mode or a boolean or enumeral type for
1000 2014-01-30 DJ Delorie <dj@redhat.com>
1002 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1003 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1004 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1005 whenever main() has an epilogue.
1007 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1009 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1010 unused variable "field".
1011 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1012 (vsx_mergeh_<mode>): Likewise.
1013 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1014 (altivec_vmrghh): Likewise.
1015 (altivec_vmrghw): Likewise.
1016 (altivec_vmrglb): Likewise.
1017 (altivec_vmrglh): Likewise.
1018 (altivec_vmrglw): Likewise.
1019 (altivec_vspltb): Add missing uses.
1020 (altivec_vsplth): Likewise.
1021 (altivec_vspltw): Likewise.
1022 (altivec_vspltsf): Likewise.
1024 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1027 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1028 frame related instructions.
1030 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1032 PR rtl-optimization/59959
1033 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1034 any reload of register whose subreg is invalid.
1036 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1038 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1039 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1040 Add missing return type - void.
1042 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1044 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1045 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1046 remove element index adjustment for endian (now handled in vsx.md
1048 (altivec_expand_vec_perm_const): Use
1049 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1050 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1051 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1052 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1053 define_expand and a new define_insn *altivec_vspltb_internal;
1054 adjust for -maltivec=be on a little endian target.
1055 (altivec_vspltb_direct): New.
1056 (altivec_vsplth): Divide into a define_expand and a new
1057 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1058 little endian target.
1059 (altivec_vsplth_direct): New.
1060 (altivec_vspltw): Divide into a define_expand and a new
1061 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1062 little endian target.
1063 (altivec_vspltw_direct): New.
1064 (altivec_vspltsf): Divide into a define_expand and a new
1065 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1066 a little endian target.
1068 2014-01-30 Richard Biener <rguenther@suse.de>
1070 PR tree-optimization/59993
1071 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1072 can propagate form the earlier stmt and avoid the transform
1073 when the intermediate result is needed.
1075 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1077 * README.Portability: Fix typo.
1079 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1081 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1082 comparison_operator with ordered_comparison_operator.
1084 2014-01-30 Nick Clifton <nickc@redhat.com>
1086 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1087 Rename to mn10300_store_multiple_regs.
1088 * config/mn10300/mn10300.c: Likewise.
1089 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1090 store_multiple_regs.
1091 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1092 Call mn10300_store_multiple_regs.
1094 2014-01-30 Nick Clifton <nickc@redhat.com>
1095 DJ Delorie <dj@redhat.com>
1097 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1098 %fp 2 to keep registers after it properly word-aligned.
1099 (rl78_alloc_physical_registers_umul): Handle the case where both
1100 input operands are the same.
1102 2014-01-30 Richard Biener <rguenther@suse.de>
1104 PR tree-optimization/59903
1105 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1108 2014-01-30 Jason Merrill <jason@redhat.com>
1111 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1114 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1116 2014-01-30 Richard Biener <rguenther@suse.de>
1118 PR tree-optimization/59951
1119 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1121 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1124 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1125 SFmode to DFmode case.
1127 2014-01-29 DJ Delorie <dj@redhat.com>
1129 * config/msp430/msp430.opt (-minrt): New.
1130 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1132 (ENDFILE_SPEC): Likewise.
1134 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1136 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1137 (estimate_function_body_sizes): Use it.
1139 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1142 * dwarf2out.c (is_cxx_auto): New.
1143 (is_base_type): Use it.
1144 (gen_type_die_with_usage): Likewise.
1146 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1148 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1149 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1150 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1151 -maltivec=be with LE targets.
1152 (vsx_mergeh_<mode>): Likewise.
1153 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1154 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1155 (altivec_vmrghb): Replace with define_expand and new
1156 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1157 (altivec_vmrghb_direct): New define_insn.
1158 (altivec_vmrghh): Replace with define_expand and new
1159 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1160 (altivec_vmrghh_direct): New define_insn.
1161 (altivec_vmrghw): Replace with define_expand and new
1162 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1163 (altivec_vmrghw_direct): New define_insn.
1164 (*altivec_vmrghsf): Adjust for endianness.
1165 (altivec_vmrglb): Replace with define_expand and new
1166 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1167 (altivec_vmrglb_direct): New define_insn.
1168 (altivec_vmrglh): Replace with define_expand and new
1169 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1170 (altivec_vmrglh_direct): New define_insn.
1171 (altivec_vmrglw): Replace with define_expand and new
1172 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1173 (altivec_vmrglw_direct): New define_insn.
1174 (*altivec_vmrglsf): Adjust for endianness.
1175 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1176 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1177 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1178 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1179 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1180 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1181 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1182 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1184 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1186 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1187 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1190 2014-01-29 Richard Biener <rguenther@suse.de>
1192 PR tree-optimization/58742
1193 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1194 associate_pointerplus_align.
1195 (associate_pointerplus_diff): New function.
1196 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1197 and associate_pointerplus_diff.
1199 2014-01-29 Richard Biener <rguenther@suse.de>
1201 * lto-streamer.h (LTO_major_version): Bump to 3.
1202 (LTO_minor_version): Reset to 0.
1204 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1206 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1207 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1208 (arm_file_start): Generate correct asm header for armv7ve.
1209 * config/arm/bpabi.h: Add multilib support for armv7ve.
1210 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1211 and cortex-a15 to armv7ve.
1212 * config/arm/t-aprofile: Add multilib support for armv7ve.
1213 * doc/invoke.texi: Document -march=armv7ve.
1215 2014-01-29 Richard Biener <rguenther@suse.de>
1217 PR tree-optimization/58742
1218 * tree-ssa-forwprop.c (associate_plusminus): Return true
1219 if we changed sth, defer EH cleanup to ...
1220 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1221 (simplify_mult): New function.
1223 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1226 PR tree-optimization/59920
1227 * tree.c (build_common_builtin_nodes): Remove
1228 __builtin_setjmp_dispatcher initialization.
1229 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1230 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1231 instead of gsi_after_labels + manually skipping debug stmts.
1232 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1233 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1234 * tree-inline.c (copy_edges_for_bb): Remove
1235 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1236 argument. Ignore computed_goto_p stmts. Don't call
1237 make_abnormal_goto_edges. If a call might need abnormal edges
1238 for non-local gotos, see if it already has an edge to
1239 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1240 with true argument, don't do anything then, otherwise add
1241 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1242 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1244 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1245 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1246 (lower_stmt): Don't set data->calls_builtin_setjmp.
1247 (lower_builtin_setjmp): Adjust comment.
1248 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1249 * tree-cfg.c (found_computed_goto): Remove.
1250 (factor_computed_gotos): Remove.
1251 (make_goto_expr_edges): Return bool, true for computed gotos.
1252 Don't call make_abnormal_goto_edges.
1253 (build_gimple_cfg): Don't set found_computed_goto, don't call
1254 factor_computed_gotos.
1255 (computed_goto_p): No longer static.
1256 (make_blocks): Don't set found_computed_goto.
1257 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1258 (make_edges): If make_goto_expr_edges returns true, push bb
1259 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1260 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1261 vector. Record mapping between bbs and OpenMP regions if there
1262 are any, adjust make_gimple_omp_edges caller. Call
1263 handle_abnormal_edges.
1264 (make_abnormal_goto_edges): Remove.
1265 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1266 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1267 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1268 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1269 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1270 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1271 filling *region also set *region_idx to (*region)->entry->index.
1274 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1275 For REGs set ORIGINAL_REGNO.
1277 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1279 * doc/md.texi: Mention that a target shouldn't implement
1280 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1283 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1285 PR tree-optimization/59594
1286 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1287 a copy of the datarefs vector rather than the vector itself.
1289 2014-01-28 Jason Merrill <jason@redhat.com>
1292 * dwarf2out.c (auto_die): New static.
1293 (gen_type_die_with_usage): Handle C++1y 'auto'.
1294 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1297 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1300 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1301 (SPEC_X32): Likewise.
1302 (SPEC_64): Likewise.
1303 * config/i386/i386.c (ix86_option_override_internal): Turn off
1304 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1306 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1307 * config/i386/i386.h (TARGET_16BIT): New macro.
1308 (TARGET_16BIT_P): Likewise.
1309 * config/i386/i386.opt: Add m16.
1310 * doc/invoke.texi: Document -m16.
1312 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1314 PR preprocessor/59935
1315 * input.c (location_get_source_line): Bail out on when line number
1316 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1318 2014-01-28 Richard Biener <rguenther@suse.de>
1320 PR tree-optimization/58742
1321 * tree-ssa-forwprop.c (associate_plusminus): Handle
1322 pointer subtraction of the form (T)(P + A) - (T)P.
1324 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1326 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1329 2014-01-28 Richard Biener <rguenther@suse.de>
1332 2014-01-28 Richard Biener <rguenther@suse.de>
1334 PR rtl-optimization/45364
1335 PR rtl-optimization/59890
1336 * var-tracking.c (local_get_addr_clear_given_value): Handle
1337 already cleared slot.
1338 (val_reset): Handle not allocated local_get_addr_cache.
1339 (vt_find_locations): Use post-order on the inverted CFG.
1341 2014-01-28 Richard Biener <rguenther@suse.de>
1343 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1345 2014-01-28 Richard Biener <rguenther@suse.de>
1347 PR rtl-optimization/45364
1348 PR rtl-optimization/59890
1349 * var-tracking.c (local_get_addr_clear_given_value): Handle
1350 already cleared slot.
1351 (val_reset): Handle not allocated local_get_addr_cache.
1352 (vt_find_locations): Use post-order on the inverted CFG.
1354 2014-01-28 Alan Modra <amodra@gmail.com>
1356 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1357 * configure.ac <recursive call for build != host>: Define
1358 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1359 and LD_FOR_BUILD too.
1360 * configure: Regenerate.
1362 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1364 * config/i386/i386.c (get_builtin_code_for_version): Separate
1365 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1366 Broadwell from Haswell.
1368 2014-01-27 Steve Ellcey <sellcey@mips.com>
1370 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1371 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1372 * config/mips/mips.c (mips_option_override): Change setting
1374 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1375 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1376 Change from Mask to Var.
1378 2014-01-27 Jeff Law <law@redhat.com>
1380 * ipa-inline.c (inline_small_functions): Fix typo.
1382 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1384 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1385 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1386 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1387 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1388 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1389 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1390 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1391 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1392 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1393 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1394 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1395 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1396 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1397 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1398 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1399 (_mm512_storeu_epi64): Ditto.
1400 (_mm512_cmpge_epi32_mask): Ditto.
1401 (_mm512_cmpge_epu32_mask): Ditto.
1402 (_mm512_cmpge_epi64_mask): Ditto.
1403 (_mm512_cmpge_epu64_mask): Ditto.
1404 (_mm512_cmple_epi32_mask): Ditto.
1405 (_mm512_cmple_epu32_mask): Ditto.
1406 (_mm512_cmple_epi64_mask): Ditto.
1407 (_mm512_cmple_epu64_mask): Ditto.
1408 (_mm512_cmplt_epi32_mask): Ditto.
1409 (_mm512_cmplt_epu32_mask): Ditto.
1410 (_mm512_cmplt_epi64_mask): Ditto.
1411 (_mm512_cmplt_epu64_mask): Ditto.
1412 (_mm512_cmpneq_epi32_mask): Ditto.
1413 (_mm512_cmpneq_epu32_mask): Ditto.
1414 (_mm512_cmpneq_epi64_mask): Ditto.
1415 (_mm512_cmpneq_epu64_mask): Ditto.
1416 (_mm512_expand_pd): Ditto.
1417 (_mm512_expand_ps): Ditto.
1418 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1419 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1420 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1421 * config/i386/i386.c (ix86_builtins): Add
1422 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1423 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1424 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1425 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1426 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1427 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1428 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1429 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1430 IX86_BUILTIN_PMOVUSQW512_MEM.
1431 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1432 __builtin_ia32_pmovsqd512mem_mask,
1433 __builtin_ia32_pmovqd512mem_mask,
1434 __builtin_ia32_pmovusqw512mem_mask,
1435 __builtin_ia32_pmovsqw512mem_mask,
1436 __builtin_ia32_pmovqw512mem_mask,
1437 __builtin_ia32_pmovusdw512mem_mask,
1438 __builtin_ia32_pmovsdw512mem_mask,
1439 __builtin_ia32_pmovdw512mem_mask,
1440 __builtin_ia32_pmovqb512mem_mask,
1441 __builtin_ia32_pmovusqb512mem_mask,
1442 __builtin_ia32_pmovsqb512mem_mask,
1443 __builtin_ia32_pmovusdb512mem_mask,
1444 __builtin_ia32_pmovsdb512mem_mask,
1445 __builtin_ia32_pmovdb512mem_mask.
1446 (bdesc_args): Add __builtin_ia32_expanddf512,
1447 __builtin_ia32_expandsf512.
1448 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1449 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1450 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1451 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1452 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1453 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1454 (avx512f_<code>v8div16qi2_mask_store): This.
1455 (avx512f_expand<mode>): New.
1457 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1459 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1461 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1462 (_mm512_prefetch_i32scatter_pd): Ditto.
1463 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1464 (_mm512_prefetch_i64scatter_pd): Ditto.
1465 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1466 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1467 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1468 (_mm512_prefetch_i32scatter_ps): Ditto.
1469 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1470 (_mm512_prefetch_i64scatter_ps): Ditto.
1471 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1472 * config/i386/i386-builtin-types.def: Define
1473 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1474 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1475 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1476 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1477 IX86_BUILTIN_SCATTERPFQPD.
1478 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1479 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1480 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1481 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1482 __builtin_ia32_scatterpfqps.
1483 (ix86_expand_builtin): Expand new built-ins.
1484 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1485 fix memory access data type.
1486 (*avx512pf_gatherpf<mode>_mask): Ditto.
1487 (*avx512pf_gatherpf<mode>): Ditto.
1488 (avx512pf_scatterpf<mode>): Ditto.
1489 (*avx512pf_scatterpf<mode>_mask): Ditto.
1490 (*avx512pf_scatterpf<mode>): Ditto.
1491 (GATHER_SCATTER_SF_MEM_MODE): New.
1492 (avx512pf_gatherpf<mode>df): Ditto.
1493 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1494 (*avx512pf_scatterpf<mode>df): Ditto.
1496 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1499 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1500 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1503 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1505 * common/config/arm/arm-common.c
1506 (arm_rewrite_mcpu): Handle multiple names.
1508 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1510 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1512 * gimple-builder.h (create_gimple_tmp): Delete.
1514 2014-01-27 Christian Bruel <christian.bruel@st.com>
1516 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1519 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1521 * config/pa/pa.md (call): Generate indirect long calls to non-local
1522 functions when outputing 32-bit code.
1523 (call_value): Likewise except for special call to buggy powf function.
1525 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1526 portable runtime and PIC indirect calls.
1527 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1528 and PIC call sequences. Use ldo instead of blr to set return register
1529 in PIC call sequence.
1531 2014-01-25 Walter Lee <walt@tilera.com>
1533 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1534 avoid clobbering a live register.
1536 2014-01-25 Walter Lee <walt@tilera.com>
1538 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1539 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1540 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1541 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1543 2014-01-25 Walter Lee <walt@tilera.com>
1545 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1546 arguments on even registers.
1547 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1549 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1550 (BIGGEST_ALIGNMENT): Ditto.
1551 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1553 2014-01-25 Walter Lee <walt@tilera.com>
1555 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1556 insns before bundling.
1557 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1559 2014-01-25 Walter Lee <walt@tilera.com>
1561 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1562 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1563 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1565 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1567 * config/mips/constraints.md (kl): Delete.
1568 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1569 define expands, using...
1570 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1571 instructions for MIPS16.
1572 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1573 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1575 2014-01-25 Walter Lee <walt@tilera.com>
1577 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1581 2014-01-25 Walter Lee <walt@tilera.com>
1583 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1584 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1586 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1588 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1591 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1593 * print-rtl.c (in_call_function_usage): New var.
1594 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1595 EXPR_LIST mode as mode and not as reg note name.
1598 * cfgloopmanip.c (copy_loop_info): If
1599 loop->warned_aggressive_loop_optimizations, make sure
1600 the flag is set in target loop too.
1602 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1604 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1606 * builtins.def: Likewise.
1607 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1608 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1609 * ira.c (ira_setup_eliminable_regset): Likewise.
1610 * omp-low.c (gate_expand_omp): Likewise.
1611 (execute_lower_omp): Likewise.
1612 (diagnose_sb_0): Likewise.
1613 (gate_diagnose_omp_blocks): Likewise.
1614 (simd_clone_clauses_extract): Likewise.
1617 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1619 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1620 correction for little endian...
1621 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1624 2014-01-24 Jeff Law <law@redhat.com>
1626 PR tree-optimization/59919
1627 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1628 for non-returning calls.
1630 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1632 * common/config/aarch64/aarch64-common.c
1633 (aarch64_rewrite_mcpu): Handle multiple names.
1634 * config/aarch64/aarch64.h
1635 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1637 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1639 * input.c (add_file_to_cache_tab): Handle the case where fopen
1642 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1645 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1646 from push operand if code of push isn't PRE_DEC.
1648 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1651 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1652 -mquad-memory-atomic. Update -mquad-memory documentation to say
1653 it is only used for non-atomic loads/stores.
1655 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1656 -mquad-memory or -mquad-memory-atomic switches.
1658 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1659 -mquad-memory-atomic to ISA 2.07 support.
1661 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1662 to separate support of normal quad word memory operations (ldq, stq)
1663 from the atomic quad word memory operations.
1665 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1666 support to separate non-atomic quad word operations from atomic
1667 quad word operations. Disable non-atomic quad word operations in
1668 little endian mode so that we don't have to swap words after the
1669 load and before the store.
1670 (quad_load_store_p): Add comment about atomic quad word support.
1671 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1672 options printed with -mdebug=reg.
1674 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1675 -mquad-memory-atomic as the test for whether we have quad word
1676 atomic instructions.
1677 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1678 or -mp8-vector are used, allow byte/half-word atomic operations.
1680 * config/rs6000/sync.md (load_lockedti): Insure that the address
1681 is a proper indexed or indirect address for the lqarx instruction.
1682 On little endian systems, swap the hi/lo registers after the lqarx
1684 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1685 insure the address is valid for the lqarx instruction.
1686 (store_conditionalti): Insure that the address is a proper indexed
1687 or indirect address for the stqcrx. instruction. On little endian
1688 systems, swap the hi/lo registers before doing the stqcrx.
1690 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1691 insure the address is valid for the stqcrx. instruction.
1693 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1694 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1695 type of quad memory support is available.
1697 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1700 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1701 there is a danger of looping.
1703 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1706 force flag_ira_loop_pressure if set via command line.
1708 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1710 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1711 (ashr_simd): New builtin handling DI mode.
1712 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1713 (aarch64_sshr_simddi): New match pattern.
1714 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1715 (vshrd_n_s64): Likewise.
1716 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1718 2014-01-23 Nick Clifton <nickc@redhat.com>
1720 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1721 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1722 favour of mcu specific scripts.
1723 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1726 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1727 Alex Velenko <Alex.Velenko@arm.com>
1729 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1730 (vaddv_s16): Likewise.
1731 (vaddv_s32): Likewise.
1732 (vaddv_u8): Likewise.
1733 (vaddv_u16): Likewise.
1734 (vaddv_u32): Likewise.
1735 (vaddvq_s8): Likewise.
1736 (vaddvq_s16): Likewise.
1737 (vaddvq_s32): Likewise.
1738 (vaddvq_s64): Likewise.
1739 (vaddvq_u8): Likewise.
1740 (vaddvq_u16): Likewise.
1741 (vaddvq_u32): Likewise.
1742 (vaddvq_u64): Likewise.
1743 (vaddv_f32): Likewise.
1744 (vaddvq_f32): Likewise.
1745 (vaddvq_f64): Likewise.
1746 (vmaxv_f32): Likewise.
1747 (vmaxv_s8): Likewise.
1748 (vmaxv_s16): Likewise.
1749 (vmaxv_s32): Likewise.
1750 (vmaxv_u8): Likewise.
1751 (vmaxv_u16): Likewise.
1752 (vmaxv_u32): Likewise.
1753 (vmaxvq_f32): Likewise.
1754 (vmaxvq_f64): Likewise.
1755 (vmaxvq_s8): Likewise.
1756 (vmaxvq_s16): Likewise.
1757 (vmaxvq_s32): Likewise.
1758 (vmaxvq_u8): Likewise.
1759 (vmaxvq_u16): Likewise.
1760 (vmaxvq_u32): Likewise.
1761 (vmaxnmv_f32): Likewise.
1762 (vmaxnmvq_f32): Likewise.
1763 (vmaxnmvq_f64): Likewise.
1764 (vminv_f32): Likewise.
1765 (vminv_s8): Likewise.
1766 (vminv_s16): Likewise.
1767 (vminv_s32): Likewise.
1768 (vminv_u8): Likewise.
1769 (vminv_u16): Likewise.
1770 (vminv_u32): Likewise.
1771 (vminvq_f32): Likewise.
1772 (vminvq_f64): Likewise.
1773 (vminvq_s8): Likewise.
1774 (vminvq_s16): Likewise.
1775 (vminvq_s32): Likewise.
1776 (vminvq_u8): Likewise.
1777 (vminvq_u16): Likewise.
1778 (vminvq_u32): Likewise.
1779 (vminnmv_f32): Likewise.
1780 (vminnmvq_f32): Likewise.
1781 (vminnmvq_f64): Likewise.
1783 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1785 * config/aarch64/aarch64-simd.md
1786 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1787 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1788 (*aarch64_mul3_elt<mode>): Likewise.
1789 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1790 (*aarch64_mul3_elt_to_64v2df): Likewise.
1791 (*aarch64_mla_elt<mode>): Likewise.
1792 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1793 (*aarch64_mls_elt<mode>): Likewise.
1794 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1795 (*aarch64_fma4_elt<mode>): Likewise.
1796 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1797 (*aarch64_fma4_elt_to_64v2df): Likewise.
1798 (*aarch64_fnma4_elt<mode>): Likewise.
1799 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1800 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1801 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1802 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1803 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1804 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1805 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1806 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1807 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1809 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1811 * config/aarch64/aarch64-simd.md
1812 (aarch64_be_checked_get_lane<mode>): New define_expand.
1813 * config/aarch64/aarch64-simd-builtins.def
1814 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1815 New builtin definition.
1816 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1817 Use new safe be builtin.
1819 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1821 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1823 (aarch64_be_st1<mode>): Likewise.
1824 (aarch_ld1<VALL:mode>): Define_expand modified.
1825 (aarch_st1<VALL:mode>): Likewise.
1826 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1827 (UNSPEC_ST1): Likewise.
1829 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1831 * config/microblaze/microblaze.md: Add trap insn and attribute
1833 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1835 PR preprocessor/58580
1836 * input.h (location_get_source_line): Take an additional line_size
1838 (void diagnostics_file_cache_fini): Declare new function.
1839 * input.c (struct fcache): New type.
1840 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1841 New static constants.
1842 (diagnostic_file_cache_init, total_lines_num)
1843 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1844 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1845 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1846 (get_next_line, read_next_line, goto_next_line, read_line_num):
1847 New static function definitions.
1848 (diagnostic_file_cache_fini): New function.
1849 (location_get_source_line): Take an additional output line_len
1850 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1852 * diagnostic.c (diagnostic_finish): Call
1853 diagnostic_file_cache_fini.
1854 (adjust_line): Take an additional input parameter for the length
1855 of the line, rather than calculating it with strlen.
1856 (diagnostic_show_locus): Adjust the use of
1857 location_get_source_line and adjust_line with respect to their new
1858 signature. While displaying a line now, do not stop at the first
1859 null byte. Rather, display the zero byte as a space and keep
1860 going until we reach the size of the line.
1861 * Makefile.in: Add vec.o to OBJS-libcommon
1863 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1864 Ilya Tocar <ilya.tocar@intel.com>
1866 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1867 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1868 (__builtin_ia32_kmov16): Ditto.
1869 * config/i386/i386.md (UNSPEC_KMOV): New.
1872 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1874 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1875 (_mm512_storeu_si512): Ditto.
1877 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1880 * rtl.h (get_referenced_operands): Declare.
1881 * recog.c (get_referenced_operands): New function.
1882 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1883 operands have been referenced when recording LO_SUM references.
1885 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1887 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1889 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1891 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1892 Enable for generic and recent AMD targets.
1894 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1896 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1897 ARG_SIZE note when adjustment was eliminated.
1899 2014-01-22 Jeff Law <law@redhat.com>
1901 PR tree-optimization/59597
1902 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1903 in file. Accept new argument REGISTERING and use it to modify
1904 dump output appropriately.
1905 (register_jump_thread): Corresponding changes.
1906 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1907 thread paths involving joiner blocks. Add code to dump cancelled
1908 jump threading paths.
1910 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1912 PR rtl-optimization/59477
1913 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1914 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1916 2014-01-22 Tom Tromey <tromey@redhat.com>
1918 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1920 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1922 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1924 PR rtl-optimization/59896
1925 * lra-constraints.c (process_alt_operands): Check unused note for
1926 matched operands of insn with no output reloads.
1928 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1930 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1931 (mips_move_from_gpr_cost): Likewise.
1933 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1935 PR rtl-optimization/59858
1936 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1937 ira_class_hard_regs_num.
1938 (process_alt_operands): Increase reject for dying matched operand.
1940 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1943 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1944 smaller than size, perform several stores or loads and stores
1945 at dst + count - size to store or copy all of size bytes, rather
1946 than just last modesize bytes.
1948 2014-01-20 DJ Delorie <dj@redhat.com>
1950 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1951 that CLOBBERs are REGs before propogating their values.
1953 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1956 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1957 (cgraph_inline_failed_type): New function.
1958 * cgraph.h (DEFCIFCODE): Add type.
1959 (cgraph_inline_failed_type_t): New enum.
1960 (cgraph_inline_failed_type): New prototype.
1961 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1962 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1963 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1964 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1965 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1966 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1967 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1968 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1969 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1970 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1971 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1972 OPTIMIZATION_MISMATCH.
1973 * tree-inline.c (expand_call_inline): Emit errors during
1974 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1976 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1979 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1980 mode attribute in insn output.
1982 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1984 * output.h (output_constant): Delete.
1985 * varasm.c (output_constant): Make private.
1987 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1989 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1991 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1994 * tree.h (fold_builtin_strcat): New prototype.
1995 * builtins.c (fold_builtin_strcat): No longer static. Add len
1996 argument, if non-NULL, don't call c_strlen. Optimize
1997 directly into __builtin_memcpy instead of __builtin_strcpy.
1998 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1999 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2001 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2003 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2004 for SImode_address_operand operands, having only a REG argument.
2006 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2008 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2009 loader name using mbig-endian.
2010 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2012 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2014 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2018 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2020 * config/aarch64/aarch64-protos.h
2021 (aarch64_cannot_change_mode_class_ptr): Declare.
2022 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2023 aarch64_cannot_change_mode_class_ptr): New.
2024 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2025 backend hook aarch64_cannot_change_mode_class.
2027 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2029 * common/config/aarch64/aarch64-common.c
2030 (aarch64_handle_option): Don't handle any option order logic here.
2031 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2032 selected_cpu, warn on architecture version mismatch.
2033 (aarch64_override_options): Fix parsing order for option strings.
2035 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2036 Iain Sandoe <iain@codesourcery.com>
2039 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2040 warning. Amend comment to reflect current functionality.
2042 2014-01-20 Richard Biener <rguenther@suse.de>
2045 * builtins.c (fold_builtin_strcat): Remove case better handled
2046 by tree-ssa-strlen.c.
2048 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2050 * config/aarch64/aarch64.opt
2051 (mcpu, march, mtune): Make case-insensitive.
2053 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2056 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2057 if operands[1] is a REG or ZERO_EXTEND of a REG.
2059 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2061 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2063 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2065 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2066 long non-pic millicode calls.
2068 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2070 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2072 2014-01-19 Kito Cheng <kito@0xlab.org>
2074 * builtins.c (expand_movstr): Check movstr expand done or fail.
2076 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2077 H.J. Lu <hongjiu.lu@intel.com>
2080 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2081 to DImode for zero-extended addresses.
2083 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2085 PR rtl-optimization/57763
2086 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2087 on the new indirect jump_insn and increment LABEL_NUSES (label).
2089 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2093 * config.gcc (x86_archs): New variable.
2094 (x86_64_archs): Likewise.
2095 (x86_cpus): Likewise.
2096 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2097 --with-arch/--with-cpu= options.
2098 Support --with-arch=/--with-cpu={nehalem,westmere,
2099 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2101 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2103 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2104 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2106 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2108 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2110 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2113 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2114 clear cpp_get_options (parse_in)->warn_unused_macros for
2115 ix86_target_macros_internal with cpp_define.
2117 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2119 * jump.c (delete_related_insns): Keep (use (insn))s.
2120 * reorg.c (redundant_insn): Check for barriers too.
2122 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2124 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2126 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2128 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2129 call to $$dyncall when TARGET_LONG_CALLS is true.
2131 2014-01-17 Jeff Law <law@redhat.com>
2133 * ree.c (combine_set_extension): Temporarily disable test for
2134 changing number of hard registers.
2136 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2139 * ipa-inline-analysis.c (inline_free_summary):
2140 Do not free summary of aliases.
2142 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2145 * gimplify.c (gimplify_expr): Use create_tmp_var
2146 instead of create_tmp_var_raw. If cond doesn't have
2147 integral type, don't add the IFN_ANNOTATE builtin at all.
2149 2014-01-17 Martin Jambor <mjambor@suse.cz>
2152 * ipa-cp.c (prev_edge_clone): New variable.
2153 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2154 Also resize prev_edge_clone vector.
2155 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2156 (ipcp_edge_removal_hook): New function.
2157 (ipcp_driver): Register ipcp_edge_removal_hook.
2159 2014-01-17 Andrew Pinski <apinski@cavium.com>
2160 Steve Ellcey <sellcey@mips.com>
2163 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2166 2014-01-17 Jeff Law <law@redhat.com>
2169 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2170 so that pass_ccp runs first.
2172 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2174 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2175 (ix86_adjust_cost): Use !TARGET_XXX.
2176 (do_reorder_for_imul): Likewise.
2177 (swap_top_of_ready_list): Likewise.
2178 (ix86_sched_reorder): Likewise.
2180 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2182 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2183 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2184 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2185 (intel_memset): New. Duplicate slm_memset.
2186 (intel_cost): New. Duplicate slm_cost.
2187 (m_INTEL): New macro.
2188 (processor_target_table): Add "intel".
2189 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2190 with PROCESSOR_INTEL for "intel".
2191 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2192 PROCESSOR_SILVERMONT.
2193 (ix86_issue_rate): Likewise.
2194 (ix86_adjust_cost): Likewise.
2195 (ia32_multipass_dfa_lookahead): Likewise.
2196 (swap_top_of_ready_list): Likewise.
2197 (ix86_sched_reorder): Likewise.
2198 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2199 instead of TARGET_OPT_AGU.
2200 * config/i386/i386.h (TARGET_INTEL): New.
2201 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2202 (processor_type): Add PROCESSOR_INTEL.
2203 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2204 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2206 2014-01-17 Marek Polacek <polacek@redhat.com>
2209 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2212 2014-01-17 Richard Biener <rguenther@suse.de>
2214 PR tree-optimization/46590
2215 * opts.c (default_options_table): Add entries for
2216 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2217 all enabled at -O1 but not for -Og.
2218 * common.opt (fbranch-count-reg): Remove Init(1).
2219 (fmove-loop-invariants): Likewise.
2220 (ftree-pta): Likewise.
2222 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2224 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2225 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2226 decls to at least the GCC 4.8 used alignments.
2229 * tree-nested.c (convert_nonlocal_reference_stmt,
2230 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2231 of GIMPLE_BIND stmts, adjust associated decls.
2233 2014-01-17 Richard Biener <rguenther@suse.de>
2235 PR tree-optimization/46590
2236 * vec.h (vec<>::bseach): New member function implementing
2237 binary search according to C89 bsearch.
2238 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2239 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2240 bitmap pointer again. Make accesses_in_loop a flat array.
2241 (mem_ref_obstack): New global.
2242 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2243 (mark_ref_stored): Likewise.
2244 (ref_indep_loop_p_2): Likewise.
2245 (set_ref_stored_in_loop): New helper function.
2246 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2247 (memref_free): Adjust.
2248 (record_mem_ref_loc): Simplify.
2249 (gather_mem_refs_stmt): Adjust.
2250 (sort_locs_in_loop_postorder_cmp): New function.
2251 (analyze_memory_references): Sort accesses_in_loop after
2252 loop postorder number.
2253 (find_ref_loc_in_loop_cmp): New function.
2254 (for_all_locs_in_loop): Find relevant cluster of locs in
2255 accesses_in_loop and iterate without recursion.
2256 (execute_sm): Avoid uninit warning.
2257 (struct ref_always_accessed): Simplify.
2258 (ref_always_accessed::operator ()): Likewise.
2259 (ref_always_accessed_p): Likewise.
2260 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2261 loop postorder numbers here.
2262 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2265 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2268 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2269 on decls for which assemble_alias has been called.
2271 2014-01-17 Nick Clifton <nickc@redhat.com>
2273 * config/msp430/msp430.opt: (mcpu): New option.
2274 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2275 (msp430_option_override): Parse target_cpu. If the MCU name
2276 matches a generic string, clear target_mcu.
2277 (msp430_attr): Allow numeric interrupt values up to 63.
2278 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2279 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2281 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2283 * config/msp430/msp430.md (popm): Use %J rather than %I.
2284 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2285 (addhi_cy_i): Use immediate_operand for operand 2.
2286 * doc/invoke.texi: Document -mcpu option.
2288 2014-01-17 Richard Biener <rguenther@suse.de>
2290 PR rtl-optimization/38518
2291 * df.h (df_analyze_loop): Declare.
2292 * df-core.c: Include cfgloop.h.
2293 (df_analyze_1): Split out main part of df_analyze.
2294 (df_analyze): Adjust.
2295 (loop_inverted_post_order_compute): New function.
2296 (loop_post_order_compute): Likewise.
2297 (df_analyze_loop): New function avoiding whole-function
2299 * loop-invariant.c (find_defs): Use df_analyze_loop.
2300 (find_invariants): Adjust.
2301 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2303 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2305 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2306 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2308 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2310 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2311 traversal when removing references.
2313 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2316 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2318 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2321 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2322 pushing a reload for an autoinc when we had previously reloaded an
2323 inner part of the address.
2325 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2327 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2329 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2330 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2331 when not giving up or versioning for alias only because of
2333 (vect_analyze_data_ref_dependences): Set to true.
2334 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2336 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2337 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2341 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2344 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2345 operand 0 predicate for gathers, use a new pseudo as subtarget.
2347 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2350 * lra-constraints.c (process_alt_operands): Add printing debug info.
2351 Check absence of input/output reloads for matched operands too.
2353 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2355 PR rtl-optimization/59835
2356 * ira.c (ira_init_register_move_cost): Increase cost for
2359 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2361 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2363 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2366 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2367 non-register objects. Use gen_(high/low)part more consistently.
2370 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2373 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2374 endian support, remove tests for WORDS_BIG_ENDIAN.
2375 (p8_mfvsrd_3_<mode>): Likewise.
2376 (reload_gpr_from_vsx<mode>): Likewise.
2377 (reload_gpr_from_vsxsf): Likewise.
2378 (p8_mfvsrd_4_disf): Likewise.
2380 2014-01-16 Richard Biener <rguenther@suse.de>
2382 PR rtl-optimization/46590
2383 * lcm.c (compute_antinout_edge): Use postorder iteration.
2384 (compute_laterin): Use inverted postorder iteration.
2386 2014-01-16 Nick Clifton <nickc@redhat.com>
2389 * varasm.c (output_constant): Return the number of bytes actually
2391 (output_constructor_array_range): Update the field size with the
2392 number of bytes emitted by output_constant.
2393 (output_constructor_regular_field): Likewise. Also do not
2394 complain if the total number of bytes emitted is now greater
2395 than the expected fieldpos.
2396 * output.h (output_constant): Update prototype and descriptive comment.
2398 2014-01-16 Marek Polacek <polacek@redhat.com>
2401 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2402 it is error_mark_node.
2404 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2406 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2407 VALID_AVX256_REG_OR_OI_MODE.
2409 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2411 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2412 current procedure should be profiled.
2414 2014-01-15 Andrew Pinski <apinski@cavium.com>
2416 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2417 of moving from/to the STACK_REG register class.
2419 2014-01-15 Richard Henderson <rth@redhat.com>
2422 * reginfo.c (global_regs_decl): Globalize.
2423 * rtl.h (global_regs_decl): Declare.
2424 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2425 reserved via global_regs.
2427 2014-01-15 Teresa Johnson <tejohnson@google.com>
2429 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2431 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2433 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2434 and vmulosh rather than call gen_vec_widen_smult_*.
2435 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2436 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2437 (vec_widen_smult_even_v16qi): Likewise.
2438 (vec_widen_umult_even_v8hi): Likewise.
2439 (vec_widen_smult_even_v8hi): Likewise.
2440 (vec_widen_umult_odd_v16qi): Likewise.
2441 (vec_widen_smult_odd_v16qi): Likewise.
2442 (vec_widen_umult_odd_v8hi): Likewise.
2443 (vec_widen_smult_odd_v8hi): Likewise.
2444 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2445 vmuloub rather than call gen_vec_widen_umult_*.
2446 (vec_widen_umult_lo_v16qi): Likewise.
2447 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2448 vmulosb rather than call gen_vec_widen_smult_*.
2449 (vec_widen_smult_lo_v16qi): Likewise.
2450 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2451 rather than call gen_vec_widen_umult_*.
2452 (vec_widen_umult_lo_v8hi): Likewise.
2453 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2454 rather than call gen_vec_widen_smult_*.
2455 (vec_widen_smult_lo_v8hi): Likewise.
2457 2014-01-15 Jeff Law <law@redhat.com>
2459 PR tree-optimization/59747
2460 * ree.c (find_and_remove_re): Properly handle case where a second
2461 eliminated extension requires widening a copy created for elimination
2462 of a prior extension.
2463 (combine_set_extension): Ensure that the number of hard regs needed
2464 for a destination register does not change when we widen it.
2466 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2468 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2469 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2470 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2471 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2472 (avr-*-rtems*): Likewise.
2473 (bfin*-rtems*): Likewise.
2474 (moxie-*-rtems*): Likewise.
2475 (h8300-*-rtems*): Likewise.
2476 (i[34567]86-*-rtems*): Likewise.
2477 (lm32-*-rtems*): Likewise.
2478 (m32r-*-rtems*): Likewise.
2479 (m68k-*-rtems*): Likewise.
2480 (microblaze*-*-rtems*): Likewise.
2481 (mips*-*-rtems*): Likewise.
2482 (powerpc-*-rtems*): Likewise.
2483 (sh-*-rtems*): Likewise.
2484 (sparc-*-rtems*): Likewise.
2485 (sparc64-*-rtems*): Likewise.
2486 (v850-*-rtems*): Likewise.
2487 (m32c-*-rtems*): Likewise.
2489 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2491 PR rtl-optimization/59511
2492 * ira.c (ira_init_register_move_cost): Use memory costs for some
2493 cases of register move cost calculations.
2494 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2495 instead of BB frequency.
2496 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2497 * lra-assigns.c (find_hard_regno_for): Ditto.
2499 2014-01-15 Richard Biener <rguenther@suse.de>
2501 PR tree-optimization/59822
2502 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2503 (vectorizable_load): Use it to hoist defs of uses of invariant
2504 loads out of the loop.
2506 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2507 Kugan Vivekanandarajah <kuganv@linaro.org>
2510 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2513 2014-01-15 Richard Biener <rguenther@suse.de>
2515 PR rtl-optimization/59802
2516 * lcm.c (compute_available): Use inverted postorder to seed
2517 the initial worklist.
2519 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2522 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2523 ADDR_REGS for invalid symrefs in non-PIC code.
2525 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2528 * builtins.c (determine_block_size): Initialize *probable_max_size
2529 even if len_rtx is CONST_INT.
2531 2014-01-14 Andrew Pinski <apinski@cavium.com>
2533 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2534 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2535 (cortexa53_tunings): Likewise.
2536 (aarch64_sched_issue_rate): New function.
2537 (TARGET_SCHED_ISSUE_RATE): Define.
2539 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2541 * ira-costs.c (find_costs_and_classes): Add missed
2542 ira_init_register_move_cost_if_necessary.
2544 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2547 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2549 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2552 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2553 to indicate if type is used for function return value. Warn ABI
2554 change if the vector mode isn't available for function return value.
2555 (ix86_function_arg_advance): Pass false to type_natural_mode.
2556 (ix86_function_arg): Likewise.
2557 (ix86_gimplify_va_arg): Likewise.
2558 (function_arg_32): Don't warn ABI change.
2559 (ix86_function_value): Pass true to type_natural_mode.
2560 (ix86_return_in_memory): Likewise.
2561 (ix86_struct_value_rtx): Removed.
2562 (TARGET_STRUCT_VALUE_RTX): Likewise.
2564 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2566 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2567 converting a conditional jump into a conditional return.
2569 2014-01-14 Richard Biener <rguenther@suse.de>
2571 PR tree-optimization/58921
2572 PR tree-optimization/59006
2573 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2574 hoisting invariant stmts.
2575 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2576 invariant loads on the preheader edge if possible.
2578 2014-01-14 Joey Ye <joey.ye@arm.com>
2580 * doc/plugin.texi (Building GCC plugins): Update to C++.
2582 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2584 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2585 (_mm_rcp28_round_ss): Ditto.
2586 (_mm_rsqrt28_round_sd): Ditto.
2587 (_mm_rsqrt28_round_ss): Ditto.
2588 (_mm_rcp28_sd): Ditto.
2589 (_mm_rcp28_ss): Ditto.
2590 (_mm_rsqrt28_sd): Ditto.
2591 (_mm_rsqrt28_ss): Ditto.
2592 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2593 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2594 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2595 (IX86_BUILTIN_RCP28SD): Ditto.
2596 (IX86_BUILTIN_RCP28SS): Ditto.
2597 (IX86_BUILTIN_RSQRT28SD): Ditto.
2598 (IX86_BUILTIN_RSQRT28SS): Ditto.
2599 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2600 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2601 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2602 (ix86_expand_special_args_builtin): Expand new FTYPE.
2603 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2604 (srcp14<mode>): Make insn unary.
2605 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2606 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2607 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2608 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2609 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2610 Fix rounding: make it SAE only.
2611 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2613 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2615 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2616 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2617 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2618 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2619 (round_saeonly_mask_scalar_operand4): Ditto.
2620 (round_saeonly_mask_scalar_op3): Ditto.
2621 (round_saeonly_mask_scalar_op4): Ditto.
2623 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2625 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2626 Implement -maltivec=be for vec_insert and vec_extract.
2628 2014-01-10 DJ Delorie <dj@redhat.com>
2630 * config/msp430/msp430.md (call_internal): Don't allow memory
2631 references with SP as the base register.
2632 (call_value_internal): Likewise.
2633 * config/msp430/constraints.md (Yc): New. For memory references
2634 that don't use SP as a base register.
2636 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2637 "an integer without a # prefix"
2638 * config/msp430/msp430.md (epilogue_helper): Use it.
2640 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2643 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2644 AVX512F gather builtins.
2645 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2646 on gather decls with INTEGER_TYPE masktype.
2647 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2648 directly into the builtin rather than hoisting it before loop.
2650 PR tree-optimization/59387
2651 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2652 (scev_const_prop): If folded_casts and type has undefined overflow,
2653 use force_gimple_operand instead of force_gimple_operand_gsi and
2654 for each added stmt if it is assign with
2655 arith_code_with_undefined_signed_overflow, call
2656 rewrite_to_defined_overflow.
2657 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2658 gimple-fold.h instead.
2659 (arith_code_with_undefined_signed_overflow,
2660 rewrite_to_defined_overflow): Moved to ...
2661 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2662 rewrite_to_defined_overflow): ... here. No longer static.
2663 Include gimplify-me.h.
2664 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2665 rewrite_to_defined_overflow): New prototypes.
2667 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2669 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2671 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2673 * builtins.c (get_object_alignment_2): Minor tweak.
2674 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2676 2014-01-13 Christian Bruel <christian.bruel@st.com>
2678 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2679 optimized non constant lengths.
2681 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2684 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2685 load as __atomic_load_N if possible.
2687 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2689 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2691 (rs6000_expand_builtin): Adjust call.
2693 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2696 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2697 * config/rs6000/rs6000.c: Include target-globals.h.
2698 (rs6000_set_current_function): Instead of doing target_reinit
2699 unconditionally, use save_target_globals_default_opts and
2700 restore_target_globals.
2702 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2704 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2705 (rs6000_expand_builtin): Handle mffs and mtfsf.
2706 (rs6000_init_builtins): Define mffs and mtfsf.
2707 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2708 (rs6000_mffs): New pattern.
2709 (rs6000_mtfsf): New pattern.
2711 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2713 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2714 Start narrowing with START. Apply candidate-use pair
2715 and check overall cost in narrowing.
2716 (iv_ca_prune): Pass new argument.
2718 2014-01-10 Jeff Law <law@redhat.com>
2721 * ree.c (combine_reaching_defs): Ensure the defining statement
2722 occurs before the extension when optimizing extensions with
2723 different source and destination hard registers.
2725 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2728 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2729 vtables into the type inheritance graph.
2731 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2733 PR rtl-optimization/59754
2734 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2735 modes in the REGNO != REGNO case.
2737 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2739 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2741 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2743 PR tree-optimization/59745
2744 * tree-predcom.c (tree_predictive_commoning_loop): Call
2745 free_affine_expand_cache if giving up because components is NULL.
2747 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2748 GC in payload of target_globals struct instead of allocating them on
2749 the heap and the larger structs separately using GC.
2750 * target-globals.h (struct target_globals): Make regs, hard_regs,
2751 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2752 of GTY((skip)) and change type to void *.
2753 (reset_target_globals): Cast loads from those fields to corresponding
2756 2014-01-10 Steve Ellcey <sellcey@mips.com>
2759 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2760 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2761 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2763 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2766 * aarch64-modes.def (CC_Zmode): New flags mode.
2767 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2768 represents an equality.
2769 (aarch64_get_condition_code): Handle CC_Zmode.
2770 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2772 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2774 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2775 extraction in good case.
2777 2014-01-10 Richard Biener <rguenther@suse.de>
2779 PR tree-optimization/59374
2780 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2781 checking after SLP discovery. Mark stmts not participating
2782 in any SLP instance properly.
2784 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2786 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2787 when handling a SET rtx.
2789 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2791 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2792 (cortex-a57): Likewise.
2793 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2795 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2797 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2798 non-iwmmxt builtins.
2800 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2804 * ipa-devirt.c record_target_from_binfo): Take as argument
2805 stack of binfos and lookup matching one for virtual inheritance.
2806 (possible_polymorphic_call_targets_1): Update.
2808 2014-01-10 Huacai Chen <chenhc@lemote.com>
2810 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2811 kernel strings for Loongson-2E/2F/3A.
2813 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2816 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2817 is_gimple_call before calling gimple_call_internal_p.
2819 2014-01-09 Steve Ellcey <sellcey@mips.com>
2821 * Makefile.in (TREE_FLOW_H): Remove.
2822 (TREE_SSA_H): Add file names from tree-flow.h.
2823 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2824 * tree.h: Remove tree-flow.h reference.
2825 * hash-table.h: Remove tree-flow.h reference.
2826 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2827 reference with tree-ssa-loop.h.
2829 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2831 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2832 default element-order behavior for -maltivec.
2833 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2835 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2836 when targeting big endian, at least for now.
2837 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2839 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2842 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2843 var satisfies use_register_for_decl, just take into account type
2844 alignment, rather than decl alignment.
2846 PR tree-optimization/59622
2847 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2848 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2849 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2850 Don't devirtualize for inplace at all. For targets.length () == 1,
2851 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2853 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2855 * config/i386/i386.md (cpu): Remove the unused btver1.
2857 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2859 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2861 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2864 * tree-core.h (struct target_globals): New forward declaration.
2865 (struct tree_target_option): Add globals field.
2866 * tree.h (TREE_TARGET_GLOBALS): Define.
2867 (prepare_target_option_nodes_for_pch): New prototype.
2868 * target-globals.h (struct target_globals): Define even if
2870 * tree.c (prepare_target_option_node_for_pch,
2871 prepare_target_option_nodes_for_pch): New functions.
2872 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2873 * config/i386/i386.c: Include target-globals.h.
2874 (ix86_set_current_function): Instead of doing target_reinit
2875 unconditionally, use save_target_globals_default_opts and
2876 restore_target_globals.
2878 2014-01-09 Richard Biener <rguenther@suse.de>
2880 PR tree-optimization/59715
2881 * tree-cfg.h (split_critical_edges): Declare.
2882 * tree-cfg.c (split_critical_edges): Export.
2883 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2885 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2887 * cfgexpand.c (expand_stack_vars): Optionally disable
2888 asan stack protection.
2889 (expand_used_vars): Likewise.
2890 (partition_stack_vars): Likewise.
2891 * asan.c (asan_emit_stack_protection): Optionally disable
2892 after return stack usage.
2893 (instrument_derefs): Optionally disable memory access instrumentation.
2894 (instrument_builtin_call): Likewise.
2895 (instrument_strlen_call): Likewise.
2896 (asan_protect_global): Optionally disable global variables protection.
2897 * doc/invoke.texi: Added doc for new options.
2898 * params.def: Added new options.
2899 * params.h: Likewise.
2901 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2903 PR rtl-optimization/59724
2904 * ifcvt.c (cond_exec_process_if_block): Don't call
2905 flow_find_head_matching_sequence with 0 longest_match.
2906 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2907 non-active insns if !stop_after.
2908 (try_head_merge_bb): Revert 2014-01-07 changes.
2910 2014-01-08 Jeff Law <law@redhat.com>
2912 * ree.c (get_sub_rtx): New function, extracted from...
2913 (merge_def_and_ext): Here.
2914 (combine_reaching_defs): Use get_sub_rtx.
2916 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2918 * cgraph.h (varpool_variable_node): Do not choke on null node.
2920 2014-01-08 Catherine Moore <clm@codesourcery.com>
2922 * config/mips/mips.md (simple_return): Attempt to use JRC
2924 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2926 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2928 PR rtl-optimization/59137
2929 * reorg.c (steal_delay_list_from_target): Call update_block for
2931 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2933 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2935 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2936 two duplicate entries.
2938 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2941 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2943 * config/mips/mips.c (mips_truncated_op_cost): New function.
2944 (mips_rtx_costs): Adjust test for BADDU.
2945 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2947 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2949 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2950 (*baddu_si): ...this new pattern.
2952 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2955 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2957 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2960 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2962 (expand_expr, expand_normal): Adjust.
2963 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2964 inner_reference_p. Use inner_reference_p to expand inner references.
2965 (store_expr): Adjust.
2966 * cfgexpand.c (expand_call_stmt): Adjust.
2968 2014-01-08 Rong Xu <xur@google.com>
2970 * gcov-io.c (gcov_var): Move from gcov-io.h.
2971 (gcov_position): Ditto.
2972 (gcov_is_error): Ditto.
2973 (gcov_rewrite): Ditto.
2974 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2975 only part to libgcc/libgcov.h.
2977 2014-01-08 Marek Polacek <polacek@redhat.com>
2980 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2982 2014-01-08 Marek Polacek <polacek@redhat.com>
2985 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2987 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2989 PR rtl-optimization/59649
2990 * stor-layout.c (get_mode_bounds): For BImode return
2991 0 and STORE_FLAG_VALUE.
2993 2014-01-08 Richard Biener <rguenther@suse.de>
2996 * gimple.h (is_gimple_builtin_call): Remove.
2997 (gimple_builtin_call_types_compatible_p): New.
2998 (gimple_call_builtin_p): New overload.
2999 * gimple.c (is_gimple_builtin_call): Remove.
3000 (validate_call): Rename to ...
3001 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3003 (validate_type): New static function.
3004 (gimple_call_builtin_p): New overload and adjust.
3005 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3006 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3007 (gimple_fold_stmt_to_constant_1): Likewise.
3008 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3010 2014-01-08 Richard Biener <rguenther@suse.de>
3013 * gimplify.c (gimplify_expr): Gimplify register-register type
3014 VIEW_CONVERT_EXPRs to separate stmts.
3016 2014-01-07 Jeff Law <law@redhat.com>
3019 * ree.c (combine_set_extension): Handle case where source
3020 and destination registers in an extension insn are different.
3021 (combine_reaching_defs): Allow source and destination registers
3022 in extension to be different under limited circumstances.
3023 (add_removable_extension): Remove restriction that the
3024 source and destination registers in the extension are the same.
3025 (find_and_remove_re): Emit a copy from the extension's
3026 destination to its source after the defining insn if
3027 the source and destination registers are different.
3030 * ifcvt.c (merge_if_block): If we are merging a block with more than
3031 one successor with a block with no successors, remove any BARRIER
3032 after the second block.
3034 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3036 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3038 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3041 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3042 for 14-bit register offsets when INT14_OK_STRICT is false.
3044 2014-01-07 Roland Stigge <stigge@antcom.de>
3045 Michael Meissner <meissner@linux.vnet.ibm.com>
3048 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3049 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3051 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3053 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3056 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3058 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3059 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3060 rtx is const0_rtx or not.
3062 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3065 * target-globals.c (save_target_globals): Remove this_fn_optab
3067 * toplev.c: Include optabs.h.
3068 (target_reinit): Temporarily restore the global options if another
3069 set of options are in force.
3071 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3073 PR rtl-optimization/58668
3074 * cfgcleanup.c (flow_find_cross_jump): Don't count
3075 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3076 to determine what is counted.
3077 (flow_find_head_matching_sequence): Use active_insn_p to determine
3079 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3081 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3082 determine what is counted.
3084 PR tree-optimization/59643
3085 * tree-predcom.c (split_data_refs_to_components): If one dr is
3086 read and one write, determine_offset fails and the write isn't
3087 in the bad component, just put the read into the bad component.
3089 2014-01-07 Mike Stump <mikestump@comcast.net>
3090 Jakub Jelinek <jakub@redhat.com>
3093 * tree-core.h (struct tree_optimization_option): Change optabs
3094 type from unsigned char * to void *.
3095 * optabs.c (init_tree_optimization_optabs): Adjust
3096 TREE_OPTIMIZATION_OPTABS initialization.
3098 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3101 * config/i386/i386.h (struct machine_function): Add
3102 no_drap_save_restore field.
3103 * config/i386/i386.c (ix86_save_reg): Use
3104 !cfun->machine->no_drap_save_restore instead of
3105 crtl->stack_realign_needed.
3106 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3107 this function clears frame_pointer_needed. Set
3108 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3109 and DRAP reg is needed.
3111 2014-01-06 Marek Polacek <polacek@redhat.com>
3114 * doc/implement-c.texi: Mention that other integer types are
3115 permitted as bit-field types in strictly conforming mode.
3117 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3119 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3122 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3124 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3126 2014-01-06 Martin Jambor <mjambor@suse.cz>
3129 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3131 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3133 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3137 * var-tracking.c (add_stores): Preserve the value of the source even if
3138 we don't record the store.
3140 2014-01-06 Terry Guo <terry.guo@arm.com>
3142 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3144 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3147 * config/darwin.c (darwin_function_section): Adjust return values to
3148 correspond to optimisation changes made in r206070.
3150 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3152 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3153 from prefetch_block tune setting.
3154 (nocona_cost): Correct size of prefetch block to 64.
3156 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3158 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3159 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3160 used to save the static chain register in the computation of the offset
3161 from which the FP registers need to be restored.
3163 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3165 PR tree-optimization/59519
3166 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3167 ICE if get_current_def (current_new_name) is already non-NULL, as long
3168 as it is a phi result of some other phi in *new_exit_bb that has
3171 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3172 or vmovdqu* for misaligned_operand.
3173 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3174 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3175 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3176 aligned_mem for AVX512F masked aligned load and store builtins and for
3179 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3181 PR tree-optimization/59651
3182 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3183 Address range for negative step should be added by TYPE_SIZE_UNIT.
3185 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3187 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3188 overlapping registers also for registers other than the stack pointer.
3190 2014-01-03 Marek Polacek <polacek@redhat.com>
3193 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3196 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3199 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3202 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3203 (push splitter): Use <P:MODE_SIZE> instead of
3204 GET_MODE_SIZE (<P:MODE>mode).
3205 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3206 (mov -1, reg peephole2): Likewise.
3207 * config/i386/sse.md (*mov<mode>_internal,
3208 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3209 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3210 *<code><mode>3, *andnot<mode>3<mask_name>,
3211 <mask_codefor><code><mode>3<mask_name>): Likewise.
3212 * config/i386/subst.md (mask_mode512bit_condition,
3213 sd_mask_mode512bit_condition): Likewise.
3215 2014-01-02 Xinliang David Li <davidxl@google.com>
3217 PR tree-optimization/59303
3218 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3219 (dump_predicates): Better output format.
3220 (pred_equal_p): New function.
3221 (is_neq_relop_p): Ditto.
3222 (is_neq_zero_form_p): Ditto.
3223 (pred_expr_equal_p): Ditto.
3224 (pred_neg_p): Ditto.
3225 (simplify_pred): Ditto.
3226 (simplify_preds_2): Ditto.
3227 (simplify_preds_3): Ditto.
3228 (simplify_preds_4): Ditto.
3229 (simplify_preds): Ditto.
3231 (push_to_worklist): Ditto.
3232 (get_pred_info_from_cmp): Ditto.
3233 (is_degenerated_phi): Ditto.
3234 (normalize_one_pred_1): Ditto.
3235 (normalize_one_pred): Ditto.
3236 (normalize_one_pred_chain): Ditto.
3237 (normalize_preds): Ditto.
3238 (normalize_cond_1): Remove function.
3239 (normalize_cond): Ditto.
3240 (is_gcond_subset_of): Ditto.
3241 (is_subset_of_any): Ditto.
3242 (is_or_set_subset_of): Ditto.
3243 (is_and_set_subset_of): Ditto.
3244 (is_norm_cond_subset_of): Ditto.
3245 (pred_chain_length_cmp): Ditto.
3246 (convert_control_dep_chain_into_preds): Type change.
3247 (find_predicates): Ditto.
3248 (find_def_preds): Ditto.
3249 (destroy_predicates_vecs): Ditto.
3250 (find_matching_predicates_in_rest_chains): Ditto.
3251 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3252 (is_pred_expr_subset): Ditto.
3253 (is_pred_chain_subset_of): Ditto.
3254 (is_included_in): Ditto.
3255 (is_superset_of): Ditto.
3257 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3259 Update copyright years.
3261 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3263 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3264 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3265 config/arc/arc.md, config/arc/arc.opt,
3266 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3267 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3268 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3269 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3270 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3271 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3272 vtable-verify.c, vtable-verify.h: Use the standard form for the
3275 2014-01-02 Tobias Burnus <burnus@net-b.de>
3277 * gcc.c (process_command): Update copyright notice dates.
3278 * gcov-dump.c: Ditto.
3280 * doc/cpp.texi: Bump @copying's copyright year.
3281 * doc/cppinternals.texi: Ditto.
3282 * doc/gcc.texi: Ditto.
3283 * doc/gccint.texi: Ditto.
3284 * doc/gcov.texi: Ditto.
3285 * doc/install.texi: Ditto.
3286 * doc/invoke.texi: Ditto.
3288 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3290 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3292 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3294 * config/i386/sse.md (*mov<mode>_internal): Guard
3295 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3297 PR rtl-optimization/59647
3298 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3299 new_rtx into UNSIGNED_FLOAT rtxes.
3301 Copyright (C) 2014 Free Software Foundation, Inc.
3303 Copying and distribution of this file, with or without modification,
3304 are permitted in any medium without royalty provided the copyright
3305 notice and this notice are preserved.