1 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
4 * genautomata.c (add_presence_absence): Fix typo with
7 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
10 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
11 for VSX/Altivec vectors that land in GPR registers.
13 2014-02-11 Richard Henderson <rth@redhat.com>
14 Jakub Jelinek <jakub@redhat.com>
17 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
18 around drhs if type conversion to lacc->type is not useless.
20 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
24 (cortex-a57.cortex-a53): Likewise.
25 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
27 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
32 2014-02-11 Renlin Li <Renlin.Li@arm.com>
34 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
35 add_options_for_arm_vfp3.
37 2014-02-11 Jeff Law <law@redhat.com>
40 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
41 object with an undesirable mode.
43 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
46 * config/i386/sol2-9.h: New file.
47 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
48 *-*-solaris2.9*): Use it.
50 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
52 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
53 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
55 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
57 * config/microblaze/microblaze.c: Extend mcpu version format
59 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
61 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
63 2014-02-10 Richard Henderson <rth@redhat.com>
66 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
67 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
68 ms-abi vs -mno-accumulate-outgoing-args.
69 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
70 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
73 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
76 * cfgexpand.c (expand_asm_operands): Attach source location to
77 ASM_INPUT rtx objects.
78 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
80 2014-02-10 Nick Clifton <nickc@redhat.com>
82 * config/mn10300/mn10300.c (popcount): New function.
83 (mn10300_expand_prologue): Include saved registers in stack usage
86 2014-02-10 Jeff Law <law@redhat.com>
89 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
90 when changing the SET_DEST of a prior insn to avoid an input
93 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
95 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
96 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
97 -mcall-openbsd, or -mcall-linux.
98 (CC1_ENDIAN_BIG_SPEC): Remove.
99 (CC1_ENDIAN_LITTLE_SPEC): Remove.
100 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
101 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
102 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
103 and %cc1_endian_default.
104 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
106 2014-02-10 Richard Biener <rguenther@suse.de>
108 PR tree-optimization/60115
109 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
110 MEM_REF handling. Properly verify that the accesses are not
111 out of the objects bound.
113 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
115 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
118 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
120 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
121 proper constants and fix formatting.
122 (possible_polymorphic_call_targets): Fix formatting.
124 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
125 Ilya Tocar <ilya.tocar@intel.com>
127 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
128 (_mm512_loadu_epi32): Renamed into...
129 (_mm512_loadu_si512): This.
130 (_mm512_storeu_epi32): Renamed into...
131 (_mm512_storeu_si512): This.
132 (_mm512_maskz_ceil_ps): Removed.
133 (_mm512_maskz_ceil_pd): Ditto.
134 (_mm512_maskz_floor_ps): Ditto.
135 (_mm512_maskz_floor_pd): Ditto.
136 (_mm512_floor_round_ps): Ditto.
137 (_mm512_floor_round_pd): Ditto.
138 (_mm512_ceil_round_ps): Ditto.
139 (_mm512_ceil_round_pd): Ditto.
140 (_mm512_mask_floor_round_ps): Ditto.
141 (_mm512_mask_floor_round_pd): Ditto.
142 (_mm512_mask_ceil_round_ps): Ditto.
143 (_mm512_mask_ceil_round_pd): Ditto.
144 (_mm512_maskz_floor_round_ps): Ditto.
145 (_mm512_maskz_floor_round_pd): Ditto.
146 (_mm512_maskz_ceil_round_ps): Ditto.
147 (_mm512_maskz_ceil_round_pd): Ditto.
148 (_mm512_expand_pd): Ditto.
149 (_mm512_expand_ps): Ditto.
150 * config/i386/i386.c (ix86_builtins): Remove
151 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
153 * config/i386/predicates.md (const1256_operand): New.
154 (const_1_to_2_operand): Ditto.
155 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
156 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
157 (*avx512pf_gatherpf<mode>sf): Ditto.
158 (avx512pf_gatherpf<mode>df): Ditto.
159 (*avx512pf_gatherpf<mode>df_mask): Ditto.
160 (*avx512pf_gatherpf<mode>df): Ditto.
161 (avx512pf_scatterpf<mode>sf): Ditto.
162 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
163 (*avx512pf_scatterpf<mode>sf): Ditto.
164 (avx512pf_scatterpf<mode>df): Ditto.
165 (*avx512pf_scatterpf<mode>df_mask): Ditto.
166 (*avx512pf_scatterpf<mode>df): Ditto.
167 (avx512f_expand<mode>): Removed.
168 (<shift_insn><mode>3<mask_name>): Change predicate type.
170 2014-02-08 Jakub Jelinek <jakub@redhat.com>
172 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
173 not at the end of datarefs vector use ordered_remove to avoid
174 reordering datarefs vector.
177 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
178 mark local addressable non-static vars as GOVD_PRIVATE
179 instead of GOVD_LOCAL.
180 * omp-low.c (lower_omp_for): Move gimple_bind_vars
181 and BLOCK_VARS of gimple_bind_block to new_stmt rather
185 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
186 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
187 assume_aligned or alloc_align attributes.
188 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
189 arguments. Handle also assume_aligned and alloc_align attributes.
190 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
191 Handle calls to functions with assume_aligned or alloc_align
193 * doc/extend.texi: Document assume_aligned and alloc_align
196 2014-02-08 Terry Guo <terry.guo@arm.com>
198 * doc/invoke.texi: Document ARM -march=armv7e-m.
200 2014-02-08 Jakub Jelinek <jakub@redhat.com>
202 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
203 flag on __cilkrts_rethrow builtin.
206 * ipa-cp.c (determine_versionability): Fail at -O0
207 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
208 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
211 2014-02-04 Jakub Jelinek <jakub@redhat.com>
214 * tree-inline.c (copy_forbidden): Fail for
215 __attribute__((optimize (0))) functions.
217 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
219 * varpool.c: Include pointer-set.h.
220 (varpool_remove_unreferenced_decls): Variables in other partitions
221 will not be output; be however careful to not lose information
224 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
226 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
227 lookup in the vtable constructor.
229 2014-02-07 Jeff Law <law@redhat.com>
232 * config/m68k/m68k.md (ashldi_extsi): Turn into a
233 define_insn_and_split.
235 * ipa-inline.c (inline_small_functions): Fix typos.
237 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
239 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
240 (s390_can_use_return_insn): Declare.
241 * config/s390/s390.h (EPILOGUE_USES): Define.
242 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
244 (s390_chunkify_start): Handle return JUMP_LABELs.
245 (s390_early_mach): Emit a main_pool instruction on the entry edge.
246 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
247 (s390_can_use_return_insn): New functions.
248 (s390_fix_long_loop_prediction): Handle conditional returns.
249 (TARGET_SET_UP_BY_PROLOGUE): Define.
250 * config/s390/s390.md (ANY_RETURN): New code iterator.
251 (*creturn, *csimple_return, return, simple_return): New patterns.
253 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
255 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
256 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
257 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
258 REG_CFA_RESTORE list when deciding not to restore a register.
260 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
262 * config/s390/s390.c: Include tree-pass.h and context.h.
263 (s390_early_mach): New function, split out from...
264 (s390_emit_prologue): ...here.
265 (pass_data_s390_early_mach): New pass structure.
266 (pass_s390_early_mach): New class.
267 (s390_option_override): Create and register early_mach pass.
270 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
272 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
273 to match for the exit block.
275 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
277 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
278 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
279 Reject misaligned operands.
281 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
283 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
286 2014-02-07 Richard Biener <rguenther@suse.de>
289 * gimple-low.c (lower_builtin_posix_memalign): New function.
290 (lower_stmt): Call it to lower posix_memalign in a way
291 to make alignment info accessible.
293 2014-02-07 Jakub Jelinek <jakub@redhat.com>
296 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
297 __builtin_setjmp_receiver.
299 2014-02-07 Richard Biener <rguenther@suse.de>
302 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
303 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
304 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
305 Handle BUILT_IN_POSIX_MEMALIGN.
306 (find_func_clobbers): Likewise.
307 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
308 (call_may_clobber_ref_p_1): Likewise.
310 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
313 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
315 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
318 * lto-cgraph.c (lto_output_node): Use
319 symtab_get_symbol_partitioning_class.
320 (lto_output_varpool_node): likewise.
321 (symtab_get_symbol_partitioning_class): Move here from
323 * cgraph.h (symbol_partitioning_class): Likewise.
324 (symtab_get_symbol_partitioning_class): Declare.
326 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
328 * ggc.h (ggc_internal_cleared_alloc): New macro.
329 * vec.h (vec_safe_copy): Handle memory stats.
330 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
331 * target-globals.c (save_target_globals): Likewise.
333 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
336 * expr.c (emit_move_resolve_push): Export; be bit more selective
337 on when to clear alias set.
338 * expr.h (emit_move_resolve_push): Declare.
339 * function.h (struct function): Add tail_call_marked.
340 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
341 * config/i386/i386-protos.h (ix86_expand_push): Remove.
342 * config/i386/i386.md (TImode move expander): De not call
344 (FP push expanders): Preserve memory attributes.
345 * config/i386/sse.md (push<mode>1): Remove.
346 * config/i386/i386.c (ix86_expand_vector_move): Handle push
348 (ix86_expand_push): Remove.
349 * config/i386/mmx.md (push<mode>1): Remove.
351 2014-02-06 Jakub Jelinek <jakub@redhat.com>
353 PR rtl-optimization/60030
354 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
355 lopart with paradoxical subreg before shifting it up by hprec.
357 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
359 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
360 Remove extra newline at end of file.
361 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
362 (arm_issue_rate): Handle cortexa57.
363 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
364 (cortex-a57.cortex-a53): Likewise.
366 2014-02-06 Jakub Jelinek <jakub@redhat.com>
369 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
370 don't record in REG_FRAME_RELATED_EXPR registers not set in that
372 (arm_expand_prologue): Adjust all callers.
373 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
374 info, registers also at the lowest numbered registers side. Use
375 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
379 * var-tracking.c (adjust_mems): Before adding a SET
380 to amd->side_effects, adjust it's SET_SRC using
381 simplify_replace_fn_rtx.
383 2014-02-06 Alan Modra <amodra@gmail.com>
386 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
387 change SDmode to DDmode when lra_in_progress.
389 2014-02-06 Jakub Jelinek <jakub@redhat.com>
392 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
393 free_data_ref on the dr first, and before goto again also set dr
394 to the next dr. For simd_lane_access, free old datarefs[i] before
395 overwriting it. For get_vectype_for_scalar_type failure, don't
396 free_data_ref if simd_lane_access.
398 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
401 * tree.h (opts_for_fn): New inline function.
402 (opt_for_fn): Define.
403 * config/i386/i386.c (ix86_function_regparm): Use
404 opt_for_fn (decl, optimize) instead of optimize.
406 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
408 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
409 for SYMBOL_REF in large memory model.
411 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
413 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
415 (cortex-a57): Likewise.
416 (cortex-a57.cortex-a53): Likewise.
418 2014-02-06 Yury Gribov <y.gribov@samsung.com>
419 Kugan Vivekanandarajah <kuganv@linaro.org>
421 * config/arm/arm.c (arm_vector_alignment_reachable): Check
423 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
425 2014-02-06 Richard Biener <rguenther@suse.de>
427 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
428 set_loop_copy and initialize_original_copy_tables.
430 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
432 * config/aarch64/aarch64-simd.md
433 (aarch64_ashr_simddi): Change QI to SI.
435 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
436 Jakub Jelinek <jakub@redhat.com>
439 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
442 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
444 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
445 CODE_FOR_altivec_vpku[hw]um to
446 CODE_FOR_altivec_vpku[hw]um_direct.
447 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
448 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
449 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
450 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
452 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
454 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
455 generation for -maltivec=be.
456 (altivec_vsumsws): Simplify redundant test.
458 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
460 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
461 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
462 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
463 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
465 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
467 (altivec_vpks<VI_char>ss): Likewise.
468 (altivec_vpks<VI_char>us): Likewise.
469 (altivec_vpku<VI_char>us): Likewise.
470 (altivec_vpku<VI_char>um): Likewise.
471 (altivec_vpku<VI_char>um_direct): New (copy of
472 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
474 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
475 target is little endian and -maltivec=be is not specified.
476 (*altivec_vupkhs<VU_char>_direct): New (copy of
477 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
479 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
480 target is little endian and -maltivec=be is not specified.
481 (*altivec_vupkls<VU_char>_direct): New (copy of
482 altivec_vupkls<VU_char> that always emits vupkls*, for internal
484 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
485 little endian and -maltivec=be is not specified.
486 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
487 little endian and -maltivec=be is not specified.
489 2014-02-05 Richard Henderson <rth@redhat.com>
492 * combine-stack-adj.c: Revert r206943.
493 * sched-int.h (struct deps_desc): Add last_args_size.
494 * sched-deps.c (init_deps): Initialize it.
495 (sched_analyze_insn): Add OUTPUT dependencies between insns that
496 contain REG_ARGS_SIZE notes.
498 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
500 * lto-cgraph.c (asm_nodes_output): Make global.
501 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
502 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
504 (driver_handle_option): Handle OPT_fwpa.
506 2014-02-05 Jakub Jelinek <jakub@redhat.com>
509 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
510 a comment typo and formatting issue. If odr_hash hasn't been
511 created, return vNULL and set *completep to false.
514 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
515 bb with no successors.
517 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
520 * doc/invoke.texi (-march): Clarify documentation for ARM.
524 2014-02-05 Richard Biener <rguenther@suse.de>
526 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
527 when not vectorizing because of too many alias checks.
528 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
529 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
531 2014-02-05 Nick Clifton <nickc@redhat.com>
533 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
534 accept extended registers in any mode when compiling for the
537 2014-02-05 Yury Gribov <y.gribov@samsung.com>
539 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
540 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
541 sanitization attributes.
542 (can_inline_edge_p): Likewise.
543 (sanitize_attrs_match_for_inline_p): New function.
545 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
547 * ipa-prop.c (detect_type_change): Shor circuit testing of
548 type changes on THIS pointer.
550 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
553 * config/pa/pa.c (legitimize_tls_address): Return original address
554 if not passed a SYMBOL_REF rtx.
555 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
557 (pa_emit_move_sequence): Simplify TLS source operands.
558 (pa_legitimate_constant_p): Reject all TLS constants.
559 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
560 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
562 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
564 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
565 groups when we know they are controlled by LTO.
566 * varasm.c (default_binds_local_p_1): If object is in other partition,
567 it will be resolved locally.
569 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
571 * config/host-linux.c (linux_gt_pch_use_address): Don't
572 use SSIZE_MAX because it is not always defined.
574 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
577 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
578 threshold for pseudo splitting.
579 (update_ebb_live_info): Process call argument hard registers and
580 hard registers from insn definition too.
581 (max_small_class_regs_num): New constant.
582 (inherit_in_ebb): Update live hard regs through EBBs. Update
583 reloads_num only for small register classes. Don't split for
586 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
589 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
592 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
594 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
596 2014-02-04 Marek Polacek <polacek@redhat.com>
598 * gdbinit.in (pel): Define.
600 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
602 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
605 2014-02-04 Richard Biener <rguenther@suse.de>
608 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
609 in function context local.
610 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
611 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
612 similar to LTO_imported_decl_ref.
614 2014-02-04 Jakub Jelinek <jakub@redhat.com>
616 PR tree-optimization/60002
617 * cgraphclones.c (build_function_decl_skip_args): Clear
620 PR tree-optimization/60023
621 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
622 false to gsi_replace.
623 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
624 has been in some EH region and vec_stmt could throw, add
625 vec_stmt into the same EH region.
626 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
627 has no lhs, ignore it.
628 * internal-fn.c (expand_MASK_LOAD): Likewise.
631 * tree-inline.c (copy_forbidden): Fail for
632 __attribute__((optimize (0))) functions.
635 * omp-low.c (simd_clone_struct_copy): If from->inbranch
636 is set, copy one less argument.
637 (expand_simd_clones): Don't subtract clone_info->inbranch
638 from simd_clone_struct_alloc argument.
640 PR rtl-optimization/57915
641 * recog.c (simplify_while_replacing): If all unary/binary/relational
642 operation arguments are constant, attempt to simplify those.
645 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
646 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
648 2014-02-04 Richard Biener <rguenther@suse.de>
650 PR tree-optimization/60012
651 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
652 TBAA disambiguation to all DDRs.
654 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
657 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
658 (LINK_SPEC): Use it for -shared, -shared-libgcc.
660 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
663 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
665 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
667 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
668 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
670 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
673 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
674 to figure out targets of polymorphic calls with known decl.
675 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
676 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
677 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
678 (get_polymorphic_call_info): ... here.
679 (get_polymorphic_call_info_from_invariant): New function.
681 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
683 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
684 lookup via vtable pointer; check for type consistency
685 and turn inconsitent facts into UNREACHABLE.
686 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
687 * gimple-fold.c (gimple_get_virt_method_for_vtable):
688 Do not ICE on type inconsistent querries; return UNREACHABLE
691 2014-02-03 Richard Henderson <rth@twiddle.net>
694 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
695 already processed this node.
696 (normalize_one_pred_1): Pass along mark_set.
697 (normalize_one_pred): Create and destroy a pointer_set_t.
698 (normalize_one_pred_chain): Likewise.
700 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
702 PR gcov-profile/58602
703 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
705 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
708 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
709 Try to devirtualize by the knowledge of virtual table pointer given by
710 aggregate propagation.
711 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
712 ipa_print_node_jump_functions): Dump also offset that
713 is relevant for polymorphic calls.
714 (determine_known_aggregate_parts): Add arg_type parameter; use it
715 instead of determining the type from pointer type.
716 (ipa_compute_jump_functions_for_edge): Update call of
717 determine_known_aggregate_parts.
718 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
719 (gimple_get_virt_method_for_binfo): ... here; simplify using
720 vtable_pointer_value_to_vtable.
721 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
722 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
723 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
724 (vtable_pointer_value_to_vtable): Break out from ...; handle also
726 (vtable_pointer_value_to_binfo): ... here.
727 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
729 2014-02-03 Teresa Johnson <tejohnson@google.com>
731 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
732 redef of outer loop index variable.
734 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
738 * doc/extend.texi (Function Attributes): Typo.
740 2014-02-03 Cong Hou <congh@google.com>
742 PR tree-optimization/60000
743 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
744 if the vectorized statement is a store. A store statement can only
745 appear at the end of pattern statements.
747 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
749 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
750 (ix86_option_override_internal): Default long double to 64-bit for
751 32-bit Bionic and to 128-bit for 64-bit Bionic.
753 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
754 TARGET_LONG_DOUBLE_128 is true.
755 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
757 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
758 (mlong-double-64): Negate -mlong-double-128.
759 (mlong-double-128): New option.
761 * config/i386/i386-c.c (ix86_target_macros): Define
762 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
764 * doc/invoke.texi: Document -mlong-double-128.
766 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
768 PR rtl-optimization/60024
769 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
771 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
773 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
775 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
777 PR rtl-optimization/57662
778 * sel-sched.c (code_motion_path_driver): Do not mark already not
779 existing blocks in the visiting bitmap.
781 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
783 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
784 on the insn being emitted.
786 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
787 Will Deacon <will.deacon@arm.com>
789 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
791 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
793 * config/arm/arm-tables.opt: Regenerate.
795 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
797 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
798 for vector types other than V16QImode.
799 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
800 define_expand, and call altivec_expand_vec_perm_le when producing
801 code with little endian element order.
802 (*altivec_vperm_<mode>_internal): New insn having previous
803 behavior of altivec_vperm_<mode>.
804 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
805 altivec_expand_vec_perm_le when producing code with little endian
807 (*altivec_vperm_<mode>_uns_internal): New insn having previous
808 behavior of altivec_vperm_<mode>_uns.
810 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
812 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
813 (altivec_vsumsws): Add handling for -maltivec=be with a little
815 (altivec_vsumsws_direct): New.
816 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
819 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
821 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
822 vtable_pointer_value_to_binfo): New functions.
823 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
824 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
826 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
828 * config/nios2/nios2.md (load_got_register): Initialize GOT
829 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
830 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
832 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
834 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
835 preserverd by passthrough, do not propagate the type.
837 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
839 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
840 (mips_atomic_assign_expand_fenv): New function.
841 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
843 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
845 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
846 (__builtin_mips_set_fcsr): Likewise.
847 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
849 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
850 (mips16_expand_set_fcsr): Likewise.
851 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
852 (mips16_set_fcsr_stub): Likewise.
853 (mips16_get_fcsr_one_only_stub): New class.
854 (mips16_set_fcsr_one_only_stub): Likewise.
855 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
856 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
857 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
858 (hard_float): New availability predicate.
859 (mips_builtins): Add get_fcsr and set_fcsr.
860 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
861 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
862 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
863 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
864 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
867 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
869 * config/mips/mips.c (mips_one_only_stub): New class.
870 (mips_need_mips16_rdhwr_p): Replace with...
871 (mips16_rdhwr_stub): ...this new variable.
872 (mips16_stub_call_address): New function.
873 (mips16_rdhwr_one_only_stub): New class.
874 (mips_expand_thread_pointer): Use mips16_stub_call_address.
875 (mips_output_mips16_rdhwr): Delete.
876 (mips_finish_stub): New function.
877 (mips_code_end): Use it to handle rdhwr stubs.
879 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
882 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
883 when calculating size of integer atomic types.
885 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
887 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
889 2014-02-01 Jakub Jelinek <jakub@redhat.com>
891 PR tree-optimization/60003
892 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
893 * profile.c (branch_prob): Use gimple_call_builtin_p
894 to check for BUILT_IN_SETJMP_RECEIVER.
895 * tree-inline.c (copy_bb): Call notice_special_calls.
897 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
900 * lra-constraints.c (process_alt_operands): Update reload_sum only
903 2014-01-31 Richard Henderson <rth@redhat.com>
906 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
907 until after else_eh is processed.
909 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
911 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
912 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
913 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
914 in smmintrin.h, remove them.
915 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
916 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
917 * config/i386/i386.md (ROUND_SAE): Fix value.
918 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
919 (const48_operand): New.
920 * config/i386/subst.md (round), (round_expand): Use
921 const_4_or_8_to_11_operand.
922 (round_saeonly), (round_saeonly_expand): Use const48_operand.
924 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
926 * config/i386/constraints.md (Yk): Swap meaning with k.
927 * config/i386/i386.md (movhi_internal): Change Yk to k.
928 (movqi_internal): Ditto.
929 (*k<logic><mode>): Ditto.
932 (kandn<mode>): Ditto.
933 (*<code>hi_1): Ditto.
934 (*<code>qi_1): Ditto.
935 (kxnor<mode>): Ditto.
939 (*one_cmplhi2_1): Ditto.
940 (*one_cmplqi2_1): Ditto.
941 * config/i386/sse.md (): Change k to Yk.
942 (avx512f_load<mode>_mask): Ditto.
943 (avx512f_blendm<mode>): Ditto.
944 (avx512f_store<mode>_mask): Ditto.
945 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
946 (avx512f_storedqu<mode>_mask): Ditto.
947 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
949 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
950 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
951 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
952 (avx512f_maskcmp<mode>3): Ditto.
953 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
954 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
955 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
956 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
957 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
958 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
959 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
960 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
961 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
962 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
963 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
964 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
965 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
966 (vec_extract_lo_<mode>_maskm): Ditto.
967 (vec_extract_hi_<mode>_maskm): Ditto.
968 (avx512f_vternlog<mode>_mask): Ditto.
969 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
970 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
971 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
972 (avx512f_<code>v8div16qi2_mask): Ditto.
973 (avx512f_<code>v8div16qi2_mask_store): Ditto.
974 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
975 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
976 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
977 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
978 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
979 (*avx512pf_gatherpf<mode>df_mask): Ditto.
980 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
981 (*avx512pf_scatterpf<mode>df_mask): Ditto.
982 (avx512cd_maskb_vec_dupv8di): Ditto.
983 (avx512cd_maskw_vec_dupv16si): Ditto.
984 (avx512f_vpermi2var<mode>3_maskz): Ditto.
985 (avx512f_vpermi2var<mode>3_mask): Ditto.
986 (avx512f_vpermi2var<mode>3_mask): Ditto.
987 (avx512f_vpermt2var<mode>3_maskz): Ditto.
988 (*avx512f_gathersi<mode>): Ditto.
989 (*avx512f_gathersi<mode>_2): Ditto.
990 (*avx512f_gatherdi<mode>): Ditto.
991 (*avx512f_gatherdi<mode>_2): Ditto.
992 (*avx512f_scattersi<mode>): Ditto.
993 (*avx512f_scatterdi<mode>): Ditto.
994 (avx512f_compress<mode>_mask): Ditto.
995 (avx512f_compressstore<mode>_mask): Ditto.
996 (avx512f_expand<mode>_mask): Ditto.
997 * config/i386/subst.md (mask): Change k to Yk.
998 (mask_scalar_merge): Ditto.
1001 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1003 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1005 2014-01-31 Richard Biener <rguenther@suse.de>
1008 * builtins.c (fold_builtin_memory_op): Make sure to not
1009 use a floating-point mode or a boolean or enumeral type for
1012 2014-01-30 DJ Delorie <dj@redhat.com>
1014 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1015 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1016 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1017 whenever main() has an epilogue.
1019 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1021 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1022 unused variable "field".
1023 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1024 (vsx_mergeh_<mode>): Likewise.
1025 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1026 (altivec_vmrghh): Likewise.
1027 (altivec_vmrghw): Likewise.
1028 (altivec_vmrglb): Likewise.
1029 (altivec_vmrglh): Likewise.
1030 (altivec_vmrglw): Likewise.
1031 (altivec_vspltb): Add missing uses.
1032 (altivec_vsplth): Likewise.
1033 (altivec_vspltw): Likewise.
1034 (altivec_vspltsf): Likewise.
1036 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1039 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1040 frame related instructions.
1042 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1044 PR rtl-optimization/59959
1045 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1046 any reload of register whose subreg is invalid.
1048 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1050 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1051 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1052 Add missing return type - void.
1054 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1056 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1057 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1058 remove element index adjustment for endian (now handled in vsx.md
1060 (altivec_expand_vec_perm_const): Use
1061 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1062 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1063 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1064 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1065 define_expand and a new define_insn *altivec_vspltb_internal;
1066 adjust for -maltivec=be on a little endian target.
1067 (altivec_vspltb_direct): New.
1068 (altivec_vsplth): Divide into a define_expand and a new
1069 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1070 little endian target.
1071 (altivec_vsplth_direct): New.
1072 (altivec_vspltw): Divide into a define_expand and a new
1073 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1074 little endian target.
1075 (altivec_vspltw_direct): New.
1076 (altivec_vspltsf): Divide into a define_expand and a new
1077 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1078 a little endian target.
1080 2014-01-30 Richard Biener <rguenther@suse.de>
1082 PR tree-optimization/59993
1083 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1084 can propagate form the earlier stmt and avoid the transform
1085 when the intermediate result is needed.
1087 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1089 * README.Portability: Fix typo.
1091 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1093 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1094 comparison_operator with ordered_comparison_operator.
1096 2014-01-30 Nick Clifton <nickc@redhat.com>
1098 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1099 Rename to mn10300_store_multiple_regs.
1100 * config/mn10300/mn10300.c: Likewise.
1101 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1102 store_multiple_regs.
1103 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1104 Call mn10300_store_multiple_regs.
1106 2014-01-30 Nick Clifton <nickc@redhat.com>
1107 DJ Delorie <dj@redhat.com>
1109 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1110 %fp 2 to keep registers after it properly word-aligned.
1111 (rl78_alloc_physical_registers_umul): Handle the case where both
1112 input operands are the same.
1114 2014-01-30 Richard Biener <rguenther@suse.de>
1116 PR tree-optimization/59903
1117 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1120 2014-01-30 Jason Merrill <jason@redhat.com>
1123 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1126 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1128 2014-01-30 Richard Biener <rguenther@suse.de>
1130 PR tree-optimization/59951
1131 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1133 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1136 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1137 SFmode to DFmode case.
1139 2014-01-29 DJ Delorie <dj@redhat.com>
1141 * config/msp430/msp430.opt (-minrt): New.
1142 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1144 (ENDFILE_SPEC): Likewise.
1146 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1148 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1149 (estimate_function_body_sizes): Use it.
1151 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1154 * dwarf2out.c (is_cxx_auto): New.
1155 (is_base_type): Use it.
1156 (gen_type_die_with_usage): Likewise.
1158 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1160 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1161 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1162 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1163 -maltivec=be with LE targets.
1164 (vsx_mergeh_<mode>): Likewise.
1165 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1166 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1167 (altivec_vmrghb): Replace with define_expand and new
1168 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1169 (altivec_vmrghb_direct): New define_insn.
1170 (altivec_vmrghh): Replace with define_expand and new
1171 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1172 (altivec_vmrghh_direct): New define_insn.
1173 (altivec_vmrghw): Replace with define_expand and new
1174 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1175 (altivec_vmrghw_direct): New define_insn.
1176 (*altivec_vmrghsf): Adjust for endianness.
1177 (altivec_vmrglb): Replace with define_expand and new
1178 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1179 (altivec_vmrglb_direct): New define_insn.
1180 (altivec_vmrglh): Replace with define_expand and new
1181 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1182 (altivec_vmrglh_direct): New define_insn.
1183 (altivec_vmrglw): Replace with define_expand and new
1184 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1185 (altivec_vmrglw_direct): New define_insn.
1186 (*altivec_vmrglsf): Adjust for endianness.
1187 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1188 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1189 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1190 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1191 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1192 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1193 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1194 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1196 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1198 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1199 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1202 2014-01-29 Richard Biener <rguenther@suse.de>
1204 PR tree-optimization/58742
1205 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1206 associate_pointerplus_align.
1207 (associate_pointerplus_diff): New function.
1208 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1209 and associate_pointerplus_diff.
1211 2014-01-29 Richard Biener <rguenther@suse.de>
1213 * lto-streamer.h (LTO_major_version): Bump to 3.
1214 (LTO_minor_version): Reset to 0.
1216 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1218 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1219 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1220 (arm_file_start): Generate correct asm header for armv7ve.
1221 * config/arm/bpabi.h: Add multilib support for armv7ve.
1222 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1223 and cortex-a15 to armv7ve.
1224 * config/arm/t-aprofile: Add multilib support for armv7ve.
1225 * doc/invoke.texi: Document -march=armv7ve.
1227 2014-01-29 Richard Biener <rguenther@suse.de>
1229 PR tree-optimization/58742
1230 * tree-ssa-forwprop.c (associate_plusminus): Return true
1231 if we changed sth, defer EH cleanup to ...
1232 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1233 (simplify_mult): New function.
1235 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1238 PR tree-optimization/59920
1239 * tree.c (build_common_builtin_nodes): Remove
1240 __builtin_setjmp_dispatcher initialization.
1241 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1242 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1243 instead of gsi_after_labels + manually skipping debug stmts.
1244 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1245 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1246 * tree-inline.c (copy_edges_for_bb): Remove
1247 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1248 argument. Ignore computed_goto_p stmts. Don't call
1249 make_abnormal_goto_edges. If a call might need abnormal edges
1250 for non-local gotos, see if it already has an edge to
1251 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1252 with true argument, don't do anything then, otherwise add
1253 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1254 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1256 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1257 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1258 (lower_stmt): Don't set data->calls_builtin_setjmp.
1259 (lower_builtin_setjmp): Adjust comment.
1260 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1261 * tree-cfg.c (found_computed_goto): Remove.
1262 (factor_computed_gotos): Remove.
1263 (make_goto_expr_edges): Return bool, true for computed gotos.
1264 Don't call make_abnormal_goto_edges.
1265 (build_gimple_cfg): Don't set found_computed_goto, don't call
1266 factor_computed_gotos.
1267 (computed_goto_p): No longer static.
1268 (make_blocks): Don't set found_computed_goto.
1269 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1270 (make_edges): If make_goto_expr_edges returns true, push bb
1271 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1272 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1273 vector. Record mapping between bbs and OpenMP regions if there
1274 are any, adjust make_gimple_omp_edges caller. Call
1275 handle_abnormal_edges.
1276 (make_abnormal_goto_edges): Remove.
1277 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1278 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1279 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1280 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1281 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1282 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1283 filling *region also set *region_idx to (*region)->entry->index.
1286 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1287 For REGs set ORIGINAL_REGNO.
1289 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1291 * doc/md.texi: Mention that a target shouldn't implement
1292 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1295 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1297 PR tree-optimization/59594
1298 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1299 a copy of the datarefs vector rather than the vector itself.
1301 2014-01-28 Jason Merrill <jason@redhat.com>
1304 * dwarf2out.c (auto_die): New static.
1305 (gen_type_die_with_usage): Handle C++1y 'auto'.
1306 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1309 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1312 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1313 (SPEC_X32): Likewise.
1314 (SPEC_64): Likewise.
1315 * config/i386/i386.c (ix86_option_override_internal): Turn off
1316 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1318 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1319 * config/i386/i386.h (TARGET_16BIT): New macro.
1320 (TARGET_16BIT_P): Likewise.
1321 * config/i386/i386.opt: Add m16.
1322 * doc/invoke.texi: Document -m16.
1324 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1326 PR preprocessor/59935
1327 * input.c (location_get_source_line): Bail out on when line number
1328 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1330 2014-01-28 Richard Biener <rguenther@suse.de>
1332 PR tree-optimization/58742
1333 * tree-ssa-forwprop.c (associate_plusminus): Handle
1334 pointer subtraction of the form (T)(P + A) - (T)P.
1336 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1338 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1341 2014-01-28 Richard Biener <rguenther@suse.de>
1344 2014-01-28 Richard Biener <rguenther@suse.de>
1346 PR rtl-optimization/45364
1347 PR rtl-optimization/59890
1348 * var-tracking.c (local_get_addr_clear_given_value): Handle
1349 already cleared slot.
1350 (val_reset): Handle not allocated local_get_addr_cache.
1351 (vt_find_locations): Use post-order on the inverted CFG.
1353 2014-01-28 Richard Biener <rguenther@suse.de>
1355 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1357 2014-01-28 Richard Biener <rguenther@suse.de>
1359 PR rtl-optimization/45364
1360 PR rtl-optimization/59890
1361 * var-tracking.c (local_get_addr_clear_given_value): Handle
1362 already cleared slot.
1363 (val_reset): Handle not allocated local_get_addr_cache.
1364 (vt_find_locations): Use post-order on the inverted CFG.
1366 2014-01-28 Alan Modra <amodra@gmail.com>
1368 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1369 * configure.ac <recursive call for build != host>: Define
1370 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1371 and LD_FOR_BUILD too.
1372 * configure: Regenerate.
1374 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1376 * config/i386/i386.c (get_builtin_code_for_version): Separate
1377 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1378 Broadwell from Haswell.
1380 2014-01-27 Steve Ellcey <sellcey@mips.com>
1382 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1383 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1384 * config/mips/mips.c (mips_option_override): Change setting
1386 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1387 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1388 Change from Mask to Var.
1390 2014-01-27 Jeff Law <law@redhat.com>
1392 * ipa-inline.c (inline_small_functions): Fix typo.
1394 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1396 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1397 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1398 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1399 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1400 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1401 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1402 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1403 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1404 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1405 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1406 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1407 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1408 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1409 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1410 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1411 (_mm512_storeu_epi64): Ditto.
1412 (_mm512_cmpge_epi32_mask): Ditto.
1413 (_mm512_cmpge_epu32_mask): Ditto.
1414 (_mm512_cmpge_epi64_mask): Ditto.
1415 (_mm512_cmpge_epu64_mask): Ditto.
1416 (_mm512_cmple_epi32_mask): Ditto.
1417 (_mm512_cmple_epu32_mask): Ditto.
1418 (_mm512_cmple_epi64_mask): Ditto.
1419 (_mm512_cmple_epu64_mask): Ditto.
1420 (_mm512_cmplt_epi32_mask): Ditto.
1421 (_mm512_cmplt_epu32_mask): Ditto.
1422 (_mm512_cmplt_epi64_mask): Ditto.
1423 (_mm512_cmplt_epu64_mask): Ditto.
1424 (_mm512_cmpneq_epi32_mask): Ditto.
1425 (_mm512_cmpneq_epu32_mask): Ditto.
1426 (_mm512_cmpneq_epi64_mask): Ditto.
1427 (_mm512_cmpneq_epu64_mask): Ditto.
1428 (_mm512_expand_pd): Ditto.
1429 (_mm512_expand_ps): Ditto.
1430 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1431 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1432 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1433 * config/i386/i386.c (ix86_builtins): Add
1434 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1435 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1436 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1437 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1438 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1439 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1440 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1441 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1442 IX86_BUILTIN_PMOVUSQW512_MEM.
1443 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1444 __builtin_ia32_pmovsqd512mem_mask,
1445 __builtin_ia32_pmovqd512mem_mask,
1446 __builtin_ia32_pmovusqw512mem_mask,
1447 __builtin_ia32_pmovsqw512mem_mask,
1448 __builtin_ia32_pmovqw512mem_mask,
1449 __builtin_ia32_pmovusdw512mem_mask,
1450 __builtin_ia32_pmovsdw512mem_mask,
1451 __builtin_ia32_pmovdw512mem_mask,
1452 __builtin_ia32_pmovqb512mem_mask,
1453 __builtin_ia32_pmovusqb512mem_mask,
1454 __builtin_ia32_pmovsqb512mem_mask,
1455 __builtin_ia32_pmovusdb512mem_mask,
1456 __builtin_ia32_pmovsdb512mem_mask,
1457 __builtin_ia32_pmovdb512mem_mask.
1458 (bdesc_args): Add __builtin_ia32_expanddf512,
1459 __builtin_ia32_expandsf512.
1460 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1461 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1462 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1463 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1464 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1465 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1466 (avx512f_<code>v8div16qi2_mask_store): This.
1467 (avx512f_expand<mode>): New.
1469 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1471 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1473 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1474 (_mm512_prefetch_i32scatter_pd): Ditto.
1475 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1476 (_mm512_prefetch_i64scatter_pd): Ditto.
1477 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1478 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1479 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1480 (_mm512_prefetch_i32scatter_ps): Ditto.
1481 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1482 (_mm512_prefetch_i64scatter_ps): Ditto.
1483 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1484 * config/i386/i386-builtin-types.def: Define
1485 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1486 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1487 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1488 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1489 IX86_BUILTIN_SCATTERPFQPD.
1490 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1491 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1492 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1493 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1494 __builtin_ia32_scatterpfqps.
1495 (ix86_expand_builtin): Expand new built-ins.
1496 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1497 fix memory access data type.
1498 (*avx512pf_gatherpf<mode>_mask): Ditto.
1499 (*avx512pf_gatherpf<mode>): Ditto.
1500 (avx512pf_scatterpf<mode>): Ditto.
1501 (*avx512pf_scatterpf<mode>_mask): Ditto.
1502 (*avx512pf_scatterpf<mode>): Ditto.
1503 (GATHER_SCATTER_SF_MEM_MODE): New.
1504 (avx512pf_gatherpf<mode>df): Ditto.
1505 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1506 (*avx512pf_scatterpf<mode>df): Ditto.
1508 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1511 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1512 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1515 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1517 * common/config/arm/arm-common.c
1518 (arm_rewrite_mcpu): Handle multiple names.
1520 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1522 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1524 * gimple-builder.h (create_gimple_tmp): Delete.
1526 2014-01-27 Christian Bruel <christian.bruel@st.com>
1528 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1531 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1533 * config/pa/pa.md (call): Generate indirect long calls to non-local
1534 functions when outputing 32-bit code.
1535 (call_value): Likewise except for special call to buggy powf function.
1537 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1538 portable runtime and PIC indirect calls.
1539 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1540 and PIC call sequences. Use ldo instead of blr to set return register
1541 in PIC call sequence.
1543 2014-01-25 Walter Lee <walt@tilera.com>
1545 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1546 avoid clobbering a live register.
1548 2014-01-25 Walter Lee <walt@tilera.com>
1550 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1551 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1552 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1553 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1555 2014-01-25 Walter Lee <walt@tilera.com>
1557 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1558 arguments on even registers.
1559 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1561 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1562 (BIGGEST_ALIGNMENT): Ditto.
1563 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1565 2014-01-25 Walter Lee <walt@tilera.com>
1567 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1568 insns before bundling.
1569 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1571 2014-01-25 Walter Lee <walt@tilera.com>
1573 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1574 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1575 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1577 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1579 * config/mips/constraints.md (kl): Delete.
1580 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1581 define expands, using...
1582 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1583 instructions for MIPS16.
1584 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1585 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1587 2014-01-25 Walter Lee <walt@tilera.com>
1589 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1593 2014-01-25 Walter Lee <walt@tilera.com>
1595 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1596 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1598 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1600 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1603 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1605 * print-rtl.c (in_call_function_usage): New var.
1606 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1607 EXPR_LIST mode as mode and not as reg note name.
1610 * cfgloopmanip.c (copy_loop_info): If
1611 loop->warned_aggressive_loop_optimizations, make sure
1612 the flag is set in target loop too.
1614 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1616 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1618 * builtins.def: Likewise.
1619 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1620 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1621 * ira.c (ira_setup_eliminable_regset): Likewise.
1622 * omp-low.c (gate_expand_omp): Likewise.
1623 (execute_lower_omp): Likewise.
1624 (diagnose_sb_0): Likewise.
1625 (gate_diagnose_omp_blocks): Likewise.
1626 (simd_clone_clauses_extract): Likewise.
1629 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1631 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1632 correction for little endian...
1633 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1636 2014-01-24 Jeff Law <law@redhat.com>
1638 PR tree-optimization/59919
1639 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1640 for non-returning calls.
1642 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1644 * common/config/aarch64/aarch64-common.c
1645 (aarch64_rewrite_mcpu): Handle multiple names.
1646 * config/aarch64/aarch64.h
1647 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1649 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1651 * input.c (add_file_to_cache_tab): Handle the case where fopen
1654 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1657 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1658 from push operand if code of push isn't PRE_DEC.
1660 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1663 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1664 -mquad-memory-atomic. Update -mquad-memory documentation to say
1665 it is only used for non-atomic loads/stores.
1667 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1668 -mquad-memory or -mquad-memory-atomic switches.
1670 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1671 -mquad-memory-atomic to ISA 2.07 support.
1673 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1674 to separate support of normal quad word memory operations (ldq, stq)
1675 from the atomic quad word memory operations.
1677 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1678 support to separate non-atomic quad word operations from atomic
1679 quad word operations. Disable non-atomic quad word operations in
1680 little endian mode so that we don't have to swap words after the
1681 load and before the store.
1682 (quad_load_store_p): Add comment about atomic quad word support.
1683 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1684 options printed with -mdebug=reg.
1686 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1687 -mquad-memory-atomic as the test for whether we have quad word
1688 atomic instructions.
1689 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1690 or -mp8-vector are used, allow byte/half-word atomic operations.
1692 * config/rs6000/sync.md (load_lockedti): Insure that the address
1693 is a proper indexed or indirect address for the lqarx instruction.
1694 On little endian systems, swap the hi/lo registers after the lqarx
1696 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1697 insure the address is valid for the lqarx instruction.
1698 (store_conditionalti): Insure that the address is a proper indexed
1699 or indirect address for the stqcrx. instruction. On little endian
1700 systems, swap the hi/lo registers before doing the stqcrx.
1702 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1703 insure the address is valid for the stqcrx. instruction.
1705 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1706 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1707 type of quad memory support is available.
1709 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1712 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1713 there is a danger of looping.
1715 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1717 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1718 force flag_ira_loop_pressure if set via command line.
1720 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1722 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1723 (ashr_simd): New builtin handling DI mode.
1724 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1725 (aarch64_sshr_simddi): New match pattern.
1726 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1727 (vshrd_n_s64): Likewise.
1728 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1730 2014-01-23 Nick Clifton <nickc@redhat.com>
1732 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1733 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1734 favour of mcu specific scripts.
1735 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1738 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1739 Alex Velenko <Alex.Velenko@arm.com>
1741 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1742 (vaddv_s16): Likewise.
1743 (vaddv_s32): Likewise.
1744 (vaddv_u8): Likewise.
1745 (vaddv_u16): Likewise.
1746 (vaddv_u32): Likewise.
1747 (vaddvq_s8): Likewise.
1748 (vaddvq_s16): Likewise.
1749 (vaddvq_s32): Likewise.
1750 (vaddvq_s64): Likewise.
1751 (vaddvq_u8): Likewise.
1752 (vaddvq_u16): Likewise.
1753 (vaddvq_u32): Likewise.
1754 (vaddvq_u64): Likewise.
1755 (vaddv_f32): Likewise.
1756 (vaddvq_f32): Likewise.
1757 (vaddvq_f64): Likewise.
1758 (vmaxv_f32): Likewise.
1759 (vmaxv_s8): Likewise.
1760 (vmaxv_s16): Likewise.
1761 (vmaxv_s32): Likewise.
1762 (vmaxv_u8): Likewise.
1763 (vmaxv_u16): Likewise.
1764 (vmaxv_u32): Likewise.
1765 (vmaxvq_f32): Likewise.
1766 (vmaxvq_f64): Likewise.
1767 (vmaxvq_s8): Likewise.
1768 (vmaxvq_s16): Likewise.
1769 (vmaxvq_s32): Likewise.
1770 (vmaxvq_u8): Likewise.
1771 (vmaxvq_u16): Likewise.
1772 (vmaxvq_u32): Likewise.
1773 (vmaxnmv_f32): Likewise.
1774 (vmaxnmvq_f32): Likewise.
1775 (vmaxnmvq_f64): Likewise.
1776 (vminv_f32): Likewise.
1777 (vminv_s8): Likewise.
1778 (vminv_s16): Likewise.
1779 (vminv_s32): Likewise.
1780 (vminv_u8): Likewise.
1781 (vminv_u16): Likewise.
1782 (vminv_u32): Likewise.
1783 (vminvq_f32): Likewise.
1784 (vminvq_f64): Likewise.
1785 (vminvq_s8): Likewise.
1786 (vminvq_s16): Likewise.
1787 (vminvq_s32): Likewise.
1788 (vminvq_u8): Likewise.
1789 (vminvq_u16): Likewise.
1790 (vminvq_u32): Likewise.
1791 (vminnmv_f32): Likewise.
1792 (vminnmvq_f32): Likewise.
1793 (vminnmvq_f64): Likewise.
1795 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1797 * config/aarch64/aarch64-simd.md
1798 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1799 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1800 (*aarch64_mul3_elt<mode>): Likewise.
1801 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1802 (*aarch64_mul3_elt_to_64v2df): Likewise.
1803 (*aarch64_mla_elt<mode>): Likewise.
1804 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1805 (*aarch64_mls_elt<mode>): Likewise.
1806 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1807 (*aarch64_fma4_elt<mode>): Likewise.
1808 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1809 (*aarch64_fma4_elt_to_64v2df): Likewise.
1810 (*aarch64_fnma4_elt<mode>): Likewise.
1811 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1812 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1813 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1814 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1815 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1816 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1817 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1818 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1819 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1821 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1823 * config/aarch64/aarch64-simd.md
1824 (aarch64_be_checked_get_lane<mode>): New define_expand.
1825 * config/aarch64/aarch64-simd-builtins.def
1826 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1827 New builtin definition.
1828 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1829 Use new safe be builtin.
1831 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1833 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1835 (aarch64_be_st1<mode>): Likewise.
1836 (aarch_ld1<VALL:mode>): Define_expand modified.
1837 (aarch_st1<VALL:mode>): Likewise.
1838 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1839 (UNSPEC_ST1): Likewise.
1841 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1843 * config/microblaze/microblaze.md: Add trap insn and attribute
1845 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1847 PR preprocessor/58580
1848 * input.h (location_get_source_line): Take an additional line_size
1850 (void diagnostics_file_cache_fini): Declare new function.
1851 * input.c (struct fcache): New type.
1852 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1853 New static constants.
1854 (diagnostic_file_cache_init, total_lines_num)
1855 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1856 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1857 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1858 (get_next_line, read_next_line, goto_next_line, read_line_num):
1859 New static function definitions.
1860 (diagnostic_file_cache_fini): New function.
1861 (location_get_source_line): Take an additional output line_len
1862 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1864 * diagnostic.c (diagnostic_finish): Call
1865 diagnostic_file_cache_fini.
1866 (adjust_line): Take an additional input parameter for the length
1867 of the line, rather than calculating it with strlen.
1868 (diagnostic_show_locus): Adjust the use of
1869 location_get_source_line and adjust_line with respect to their new
1870 signature. While displaying a line now, do not stop at the first
1871 null byte. Rather, display the zero byte as a space and keep
1872 going until we reach the size of the line.
1873 * Makefile.in: Add vec.o to OBJS-libcommon
1875 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1876 Ilya Tocar <ilya.tocar@intel.com>
1878 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1879 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1880 (__builtin_ia32_kmov16): Ditto.
1881 * config/i386/i386.md (UNSPEC_KMOV): New.
1884 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1886 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1887 (_mm512_storeu_si512): Ditto.
1889 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1892 * rtl.h (get_referenced_operands): Declare.
1893 * recog.c (get_referenced_operands): New function.
1894 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1895 operands have been referenced when recording LO_SUM references.
1897 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1899 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1901 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1903 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1904 Enable for generic and recent AMD targets.
1906 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1908 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1909 ARG_SIZE note when adjustment was eliminated.
1911 2014-01-22 Jeff Law <law@redhat.com>
1913 PR tree-optimization/59597
1914 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1915 in file. Accept new argument REGISTERING and use it to modify
1916 dump output appropriately.
1917 (register_jump_thread): Corresponding changes.
1918 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1919 thread paths involving joiner blocks. Add code to dump cancelled
1920 jump threading paths.
1922 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1924 PR rtl-optimization/59477
1925 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1926 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1928 2014-01-22 Tom Tromey <tromey@redhat.com>
1930 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1932 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1934 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1936 PR rtl-optimization/59896
1937 * lra-constraints.c (process_alt_operands): Check unused note for
1938 matched operands of insn with no output reloads.
1940 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1942 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1943 (mips_move_from_gpr_cost): Likewise.
1945 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1947 PR rtl-optimization/59858
1948 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1949 ira_class_hard_regs_num.
1950 (process_alt_operands): Increase reject for dying matched operand.
1952 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1955 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1956 smaller than size, perform several stores or loads and stores
1957 at dst + count - size to store or copy all of size bytes, rather
1958 than just last modesize bytes.
1960 2014-01-20 DJ Delorie <dj@redhat.com>
1962 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1963 that CLOBBERs are REGs before propogating their values.
1965 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1968 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1969 (cgraph_inline_failed_type): New function.
1970 * cgraph.h (DEFCIFCODE): Add type.
1971 (cgraph_inline_failed_type_t): New enum.
1972 (cgraph_inline_failed_type): New prototype.
1973 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1974 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1975 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1976 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1977 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1978 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1979 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1980 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1981 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1982 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1983 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1984 OPTIMIZATION_MISMATCH.
1985 * tree-inline.c (expand_call_inline): Emit errors during
1986 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1988 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1991 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1992 mode attribute in insn output.
1994 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1996 * output.h (output_constant): Delete.
1997 * varasm.c (output_constant): Make private.
1999 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2001 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2003 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2006 * tree.h (fold_builtin_strcat): New prototype.
2007 * builtins.c (fold_builtin_strcat): No longer static. Add len
2008 argument, if non-NULL, don't call c_strlen. Optimize
2009 directly into __builtin_memcpy instead of __builtin_strcpy.
2010 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2011 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2013 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2015 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2016 for SImode_address_operand operands, having only a REG argument.
2018 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2020 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2021 loader name using mbig-endian.
2022 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2024 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2026 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2030 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2032 * config/aarch64/aarch64-protos.h
2033 (aarch64_cannot_change_mode_class_ptr): Declare.
2034 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2035 aarch64_cannot_change_mode_class_ptr): New.
2036 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2037 backend hook aarch64_cannot_change_mode_class.
2039 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2041 * common/config/aarch64/aarch64-common.c
2042 (aarch64_handle_option): Don't handle any option order logic here.
2043 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2044 selected_cpu, warn on architecture version mismatch.
2045 (aarch64_override_options): Fix parsing order for option strings.
2047 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2048 Iain Sandoe <iain@codesourcery.com>
2051 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2052 warning. Amend comment to reflect current functionality.
2054 2014-01-20 Richard Biener <rguenther@suse.de>
2057 * builtins.c (fold_builtin_strcat): Remove case better handled
2058 by tree-ssa-strlen.c.
2060 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2062 * config/aarch64/aarch64.opt
2063 (mcpu, march, mtune): Make case-insensitive.
2065 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2068 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2069 if operands[1] is a REG or ZERO_EXTEND of a REG.
2071 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2073 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2075 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2077 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2078 long non-pic millicode calls.
2080 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2082 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2084 2014-01-19 Kito Cheng <kito@0xlab.org>
2086 * builtins.c (expand_movstr): Check movstr expand done or fail.
2088 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2089 H.J. Lu <hongjiu.lu@intel.com>
2092 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2093 to DImode for zero-extended addresses.
2095 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2097 PR rtl-optimization/57763
2098 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2099 on the new indirect jump_insn and increment LABEL_NUSES (label).
2101 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2105 * config.gcc (x86_archs): New variable.
2106 (x86_64_archs): Likewise.
2107 (x86_cpus): Likewise.
2108 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2109 --with-arch/--with-cpu= options.
2110 Support --with-arch=/--with-cpu={nehalem,westmere,
2111 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2113 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2115 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2116 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2118 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2120 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2122 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2125 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2126 clear cpp_get_options (parse_in)->warn_unused_macros for
2127 ix86_target_macros_internal with cpp_define.
2129 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2131 * jump.c (delete_related_insns): Keep (use (insn))s.
2132 * reorg.c (redundant_insn): Check for barriers too.
2134 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2136 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2138 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2140 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2141 call to $$dyncall when TARGET_LONG_CALLS is true.
2143 2014-01-17 Jeff Law <law@redhat.com>
2145 * ree.c (combine_set_extension): Temporarily disable test for
2146 changing number of hard registers.
2148 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2151 * ipa-inline-analysis.c (inline_free_summary):
2152 Do not free summary of aliases.
2154 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2157 * gimplify.c (gimplify_expr): Use create_tmp_var
2158 instead of create_tmp_var_raw. If cond doesn't have
2159 integral type, don't add the IFN_ANNOTATE builtin at all.
2161 2014-01-17 Martin Jambor <mjambor@suse.cz>
2164 * ipa-cp.c (prev_edge_clone): New variable.
2165 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2166 Also resize prev_edge_clone vector.
2167 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2168 (ipcp_edge_removal_hook): New function.
2169 (ipcp_driver): Register ipcp_edge_removal_hook.
2171 2014-01-17 Andrew Pinski <apinski@cavium.com>
2172 Steve Ellcey <sellcey@mips.com>
2175 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2178 2014-01-17 Jeff Law <law@redhat.com>
2181 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2182 so that pass_ccp runs first.
2184 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2186 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2187 (ix86_adjust_cost): Use !TARGET_XXX.
2188 (do_reorder_for_imul): Likewise.
2189 (swap_top_of_ready_list): Likewise.
2190 (ix86_sched_reorder): Likewise.
2192 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2194 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2195 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2196 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2197 (intel_memset): New. Duplicate slm_memset.
2198 (intel_cost): New. Duplicate slm_cost.
2199 (m_INTEL): New macro.
2200 (processor_target_table): Add "intel".
2201 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2202 with PROCESSOR_INTEL for "intel".
2203 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2204 PROCESSOR_SILVERMONT.
2205 (ix86_issue_rate): Likewise.
2206 (ix86_adjust_cost): Likewise.
2207 (ia32_multipass_dfa_lookahead): Likewise.
2208 (swap_top_of_ready_list): Likewise.
2209 (ix86_sched_reorder): Likewise.
2210 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2211 instead of TARGET_OPT_AGU.
2212 * config/i386/i386.h (TARGET_INTEL): New.
2213 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2214 (processor_type): Add PROCESSOR_INTEL.
2215 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2216 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2218 2014-01-17 Marek Polacek <polacek@redhat.com>
2221 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2224 2014-01-17 Richard Biener <rguenther@suse.de>
2226 PR tree-optimization/46590
2227 * opts.c (default_options_table): Add entries for
2228 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2229 all enabled at -O1 but not for -Og.
2230 * common.opt (fbranch-count-reg): Remove Init(1).
2231 (fmove-loop-invariants): Likewise.
2232 (ftree-pta): Likewise.
2234 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2236 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2237 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2238 decls to at least the GCC 4.8 used alignments.
2241 * tree-nested.c (convert_nonlocal_reference_stmt,
2242 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2243 of GIMPLE_BIND stmts, adjust associated decls.
2245 2014-01-17 Richard Biener <rguenther@suse.de>
2247 PR tree-optimization/46590
2248 * vec.h (vec<>::bseach): New member function implementing
2249 binary search according to C89 bsearch.
2250 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2251 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2252 bitmap pointer again. Make accesses_in_loop a flat array.
2253 (mem_ref_obstack): New global.
2254 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2255 (mark_ref_stored): Likewise.
2256 (ref_indep_loop_p_2): Likewise.
2257 (set_ref_stored_in_loop): New helper function.
2258 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2259 (memref_free): Adjust.
2260 (record_mem_ref_loc): Simplify.
2261 (gather_mem_refs_stmt): Adjust.
2262 (sort_locs_in_loop_postorder_cmp): New function.
2263 (analyze_memory_references): Sort accesses_in_loop after
2264 loop postorder number.
2265 (find_ref_loc_in_loop_cmp): New function.
2266 (for_all_locs_in_loop): Find relevant cluster of locs in
2267 accesses_in_loop and iterate without recursion.
2268 (execute_sm): Avoid uninit warning.
2269 (struct ref_always_accessed): Simplify.
2270 (ref_always_accessed::operator ()): Likewise.
2271 (ref_always_accessed_p): Likewise.
2272 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2273 loop postorder numbers here.
2274 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2277 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2280 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2281 on decls for which assemble_alias has been called.
2283 2014-01-17 Nick Clifton <nickc@redhat.com>
2285 * config/msp430/msp430.opt: (mcpu): New option.
2286 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2287 (msp430_option_override): Parse target_cpu. If the MCU name
2288 matches a generic string, clear target_mcu.
2289 (msp430_attr): Allow numeric interrupt values up to 63.
2290 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2291 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2293 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2295 * config/msp430/msp430.md (popm): Use %J rather than %I.
2296 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2297 (addhi_cy_i): Use immediate_operand for operand 2.
2298 * doc/invoke.texi: Document -mcpu option.
2300 2014-01-17 Richard Biener <rguenther@suse.de>
2302 PR rtl-optimization/38518
2303 * df.h (df_analyze_loop): Declare.
2304 * df-core.c: Include cfgloop.h.
2305 (df_analyze_1): Split out main part of df_analyze.
2306 (df_analyze): Adjust.
2307 (loop_inverted_post_order_compute): New function.
2308 (loop_post_order_compute): Likewise.
2309 (df_analyze_loop): New function avoiding whole-function
2311 * loop-invariant.c (find_defs): Use df_analyze_loop.
2312 (find_invariants): Adjust.
2313 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2315 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2317 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2318 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2320 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2322 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2323 traversal when removing references.
2325 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2328 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2330 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2333 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2334 pushing a reload for an autoinc when we had previously reloaded an
2335 inner part of the address.
2337 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2339 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2341 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2342 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2343 when not giving up or versioning for alias only because of
2345 (vect_analyze_data_ref_dependences): Set to true.
2346 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2348 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2349 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2353 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2356 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2357 operand 0 predicate for gathers, use a new pseudo as subtarget.
2359 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2362 * lra-constraints.c (process_alt_operands): Add printing debug info.
2363 Check absence of input/output reloads for matched operands too.
2365 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2367 PR rtl-optimization/59835
2368 * ira.c (ira_init_register_move_cost): Increase cost for
2371 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2373 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2375 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2378 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2379 non-register objects. Use gen_(high/low)part more consistently.
2382 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2385 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2386 endian support, remove tests for WORDS_BIG_ENDIAN.
2387 (p8_mfvsrd_3_<mode>): Likewise.
2388 (reload_gpr_from_vsx<mode>): Likewise.
2389 (reload_gpr_from_vsxsf): Likewise.
2390 (p8_mfvsrd_4_disf): Likewise.
2392 2014-01-16 Richard Biener <rguenther@suse.de>
2394 PR rtl-optimization/46590
2395 * lcm.c (compute_antinout_edge): Use postorder iteration.
2396 (compute_laterin): Use inverted postorder iteration.
2398 2014-01-16 Nick Clifton <nickc@redhat.com>
2401 * varasm.c (output_constant): Return the number of bytes actually
2403 (output_constructor_array_range): Update the field size with the
2404 number of bytes emitted by output_constant.
2405 (output_constructor_regular_field): Likewise. Also do not
2406 complain if the total number of bytes emitted is now greater
2407 than the expected fieldpos.
2408 * output.h (output_constant): Update prototype and descriptive comment.
2410 2014-01-16 Marek Polacek <polacek@redhat.com>
2413 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2414 it is error_mark_node.
2416 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2418 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2419 VALID_AVX256_REG_OR_OI_MODE.
2421 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2423 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2424 current procedure should be profiled.
2426 2014-01-15 Andrew Pinski <apinski@cavium.com>
2428 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2429 of moving from/to the STACK_REG register class.
2431 2014-01-15 Richard Henderson <rth@redhat.com>
2434 * reginfo.c (global_regs_decl): Globalize.
2435 * rtl.h (global_regs_decl): Declare.
2436 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2437 reserved via global_regs.
2439 2014-01-15 Teresa Johnson <tejohnson@google.com>
2441 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2443 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2445 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2446 and vmulosh rather than call gen_vec_widen_smult_*.
2447 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2448 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2449 (vec_widen_smult_even_v16qi): Likewise.
2450 (vec_widen_umult_even_v8hi): Likewise.
2451 (vec_widen_smult_even_v8hi): Likewise.
2452 (vec_widen_umult_odd_v16qi): Likewise.
2453 (vec_widen_smult_odd_v16qi): Likewise.
2454 (vec_widen_umult_odd_v8hi): Likewise.
2455 (vec_widen_smult_odd_v8hi): Likewise.
2456 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2457 vmuloub rather than call gen_vec_widen_umult_*.
2458 (vec_widen_umult_lo_v16qi): Likewise.
2459 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2460 vmulosb rather than call gen_vec_widen_smult_*.
2461 (vec_widen_smult_lo_v16qi): Likewise.
2462 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2463 rather than call gen_vec_widen_umult_*.
2464 (vec_widen_umult_lo_v8hi): Likewise.
2465 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2466 rather than call gen_vec_widen_smult_*.
2467 (vec_widen_smult_lo_v8hi): Likewise.
2469 2014-01-15 Jeff Law <law@redhat.com>
2471 PR tree-optimization/59747
2472 * ree.c (find_and_remove_re): Properly handle case where a second
2473 eliminated extension requires widening a copy created for elimination
2474 of a prior extension.
2475 (combine_set_extension): Ensure that the number of hard regs needed
2476 for a destination register does not change when we widen it.
2478 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2480 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2481 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2482 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2483 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2484 (avr-*-rtems*): Likewise.
2485 (bfin*-rtems*): Likewise.
2486 (moxie-*-rtems*): Likewise.
2487 (h8300-*-rtems*): Likewise.
2488 (i[34567]86-*-rtems*): Likewise.
2489 (lm32-*-rtems*): Likewise.
2490 (m32r-*-rtems*): Likewise.
2491 (m68k-*-rtems*): Likewise.
2492 (microblaze*-*-rtems*): Likewise.
2493 (mips*-*-rtems*): Likewise.
2494 (powerpc-*-rtems*): Likewise.
2495 (sh-*-rtems*): Likewise.
2496 (sparc-*-rtems*): Likewise.
2497 (sparc64-*-rtems*): Likewise.
2498 (v850-*-rtems*): Likewise.
2499 (m32c-*-rtems*): Likewise.
2501 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2503 PR rtl-optimization/59511
2504 * ira.c (ira_init_register_move_cost): Use memory costs for some
2505 cases of register move cost calculations.
2506 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2507 instead of BB frequency.
2508 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2509 * lra-assigns.c (find_hard_regno_for): Ditto.
2511 2014-01-15 Richard Biener <rguenther@suse.de>
2513 PR tree-optimization/59822
2514 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2515 (vectorizable_load): Use it to hoist defs of uses of invariant
2516 loads out of the loop.
2518 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2519 Kugan Vivekanandarajah <kuganv@linaro.org>
2522 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2525 2014-01-15 Richard Biener <rguenther@suse.de>
2527 PR rtl-optimization/59802
2528 * lcm.c (compute_available): Use inverted postorder to seed
2529 the initial worklist.
2531 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2534 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2535 ADDR_REGS for invalid symrefs in non-PIC code.
2537 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2540 * builtins.c (determine_block_size): Initialize *probable_max_size
2541 even if len_rtx is CONST_INT.
2543 2014-01-14 Andrew Pinski <apinski@cavium.com>
2545 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2546 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2547 (cortexa53_tunings): Likewise.
2548 (aarch64_sched_issue_rate): New function.
2549 (TARGET_SCHED_ISSUE_RATE): Define.
2551 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2553 * ira-costs.c (find_costs_and_classes): Add missed
2554 ira_init_register_move_cost_if_necessary.
2556 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2559 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2561 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2564 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2565 to indicate if type is used for function return value. Warn ABI
2566 change if the vector mode isn't available for function return value.
2567 (ix86_function_arg_advance): Pass false to type_natural_mode.
2568 (ix86_function_arg): Likewise.
2569 (ix86_gimplify_va_arg): Likewise.
2570 (function_arg_32): Don't warn ABI change.
2571 (ix86_function_value): Pass true to type_natural_mode.
2572 (ix86_return_in_memory): Likewise.
2573 (ix86_struct_value_rtx): Removed.
2574 (TARGET_STRUCT_VALUE_RTX): Likewise.
2576 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2578 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2579 converting a conditional jump into a conditional return.
2581 2014-01-14 Richard Biener <rguenther@suse.de>
2583 PR tree-optimization/58921
2584 PR tree-optimization/59006
2585 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2586 hoisting invariant stmts.
2587 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2588 invariant loads on the preheader edge if possible.
2590 2014-01-14 Joey Ye <joey.ye@arm.com>
2592 * doc/plugin.texi (Building GCC plugins): Update to C++.
2594 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2596 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2597 (_mm_rcp28_round_ss): Ditto.
2598 (_mm_rsqrt28_round_sd): Ditto.
2599 (_mm_rsqrt28_round_ss): Ditto.
2600 (_mm_rcp28_sd): Ditto.
2601 (_mm_rcp28_ss): Ditto.
2602 (_mm_rsqrt28_sd): Ditto.
2603 (_mm_rsqrt28_ss): Ditto.
2604 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2605 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2606 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2607 (IX86_BUILTIN_RCP28SD): Ditto.
2608 (IX86_BUILTIN_RCP28SS): Ditto.
2609 (IX86_BUILTIN_RSQRT28SD): Ditto.
2610 (IX86_BUILTIN_RSQRT28SS): Ditto.
2611 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2612 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2613 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2614 (ix86_expand_special_args_builtin): Expand new FTYPE.
2615 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2616 (srcp14<mode>): Make insn unary.
2617 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2618 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2619 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2620 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2621 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2622 Fix rounding: make it SAE only.
2623 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2625 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2627 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2628 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2629 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2630 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2631 (round_saeonly_mask_scalar_operand4): Ditto.
2632 (round_saeonly_mask_scalar_op3): Ditto.
2633 (round_saeonly_mask_scalar_op4): Ditto.
2635 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2637 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2638 Implement -maltivec=be for vec_insert and vec_extract.
2640 2014-01-10 DJ Delorie <dj@redhat.com>
2642 * config/msp430/msp430.md (call_internal): Don't allow memory
2643 references with SP as the base register.
2644 (call_value_internal): Likewise.
2645 * config/msp430/constraints.md (Yc): New. For memory references
2646 that don't use SP as a base register.
2648 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2649 "an integer without a # prefix"
2650 * config/msp430/msp430.md (epilogue_helper): Use it.
2652 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2655 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2656 AVX512F gather builtins.
2657 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2658 on gather decls with INTEGER_TYPE masktype.
2659 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2660 directly into the builtin rather than hoisting it before loop.
2662 PR tree-optimization/59387
2663 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2664 (scev_const_prop): If folded_casts and type has undefined overflow,
2665 use force_gimple_operand instead of force_gimple_operand_gsi and
2666 for each added stmt if it is assign with
2667 arith_code_with_undefined_signed_overflow, call
2668 rewrite_to_defined_overflow.
2669 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2670 gimple-fold.h instead.
2671 (arith_code_with_undefined_signed_overflow,
2672 rewrite_to_defined_overflow): Moved to ...
2673 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2674 rewrite_to_defined_overflow): ... here. No longer static.
2675 Include gimplify-me.h.
2676 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2677 rewrite_to_defined_overflow): New prototypes.
2679 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2681 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2683 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2685 * builtins.c (get_object_alignment_2): Minor tweak.
2686 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2688 2014-01-13 Christian Bruel <christian.bruel@st.com>
2690 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2691 optimized non constant lengths.
2693 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2696 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2697 load as __atomic_load_N if possible.
2699 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2701 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2703 (rs6000_expand_builtin): Adjust call.
2705 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2708 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2709 * config/rs6000/rs6000.c: Include target-globals.h.
2710 (rs6000_set_current_function): Instead of doing target_reinit
2711 unconditionally, use save_target_globals_default_opts and
2712 restore_target_globals.
2714 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2716 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2717 (rs6000_expand_builtin): Handle mffs and mtfsf.
2718 (rs6000_init_builtins): Define mffs and mtfsf.
2719 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2720 (rs6000_mffs): New pattern.
2721 (rs6000_mtfsf): New pattern.
2723 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2725 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2726 Start narrowing with START. Apply candidate-use pair
2727 and check overall cost in narrowing.
2728 (iv_ca_prune): Pass new argument.
2730 2014-01-10 Jeff Law <law@redhat.com>
2733 * ree.c (combine_reaching_defs): Ensure the defining statement
2734 occurs before the extension when optimizing extensions with
2735 different source and destination hard registers.
2737 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2740 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2741 vtables into the type inheritance graph.
2743 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2745 PR rtl-optimization/59754
2746 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2747 modes in the REGNO != REGNO case.
2749 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2751 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2753 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2755 PR tree-optimization/59745
2756 * tree-predcom.c (tree_predictive_commoning_loop): Call
2757 free_affine_expand_cache if giving up because components is NULL.
2759 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2760 GC in payload of target_globals struct instead of allocating them on
2761 the heap and the larger structs separately using GC.
2762 * target-globals.h (struct target_globals): Make regs, hard_regs,
2763 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2764 of GTY((skip)) and change type to void *.
2765 (reset_target_globals): Cast loads from those fields to corresponding
2768 2014-01-10 Steve Ellcey <sellcey@mips.com>
2771 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2772 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2773 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2775 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2778 * aarch64-modes.def (CC_Zmode): New flags mode.
2779 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2780 represents an equality.
2781 (aarch64_get_condition_code): Handle CC_Zmode.
2782 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2784 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2786 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2787 extraction in good case.
2789 2014-01-10 Richard Biener <rguenther@suse.de>
2791 PR tree-optimization/59374
2792 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2793 checking after SLP discovery. Mark stmts not participating
2794 in any SLP instance properly.
2796 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2798 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2799 when handling a SET rtx.
2801 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2803 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2804 (cortex-a57): Likewise.
2805 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2807 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2809 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2810 non-iwmmxt builtins.
2812 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2816 * ipa-devirt.c record_target_from_binfo): Take as argument
2817 stack of binfos and lookup matching one for virtual inheritance.
2818 (possible_polymorphic_call_targets_1): Update.
2820 2014-01-10 Huacai Chen <chenhc@lemote.com>
2822 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2823 kernel strings for Loongson-2E/2F/3A.
2825 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2828 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2829 is_gimple_call before calling gimple_call_internal_p.
2831 2014-01-09 Steve Ellcey <sellcey@mips.com>
2833 * Makefile.in (TREE_FLOW_H): Remove.
2834 (TREE_SSA_H): Add file names from tree-flow.h.
2835 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2836 * tree.h: Remove tree-flow.h reference.
2837 * hash-table.h: Remove tree-flow.h reference.
2838 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2839 reference with tree-ssa-loop.h.
2841 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2843 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2844 default element-order behavior for -maltivec.
2845 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2847 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2848 when targeting big endian, at least for now.
2849 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2851 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2854 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2855 var satisfies use_register_for_decl, just take into account type
2856 alignment, rather than decl alignment.
2858 PR tree-optimization/59622
2859 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2860 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2861 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2862 Don't devirtualize for inplace at all. For targets.length () == 1,
2863 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2865 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2867 * config/i386/i386.md (cpu): Remove the unused btver1.
2869 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2871 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2873 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2876 * tree-core.h (struct target_globals): New forward declaration.
2877 (struct tree_target_option): Add globals field.
2878 * tree.h (TREE_TARGET_GLOBALS): Define.
2879 (prepare_target_option_nodes_for_pch): New prototype.
2880 * target-globals.h (struct target_globals): Define even if
2882 * tree.c (prepare_target_option_node_for_pch,
2883 prepare_target_option_nodes_for_pch): New functions.
2884 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2885 * config/i386/i386.c: Include target-globals.h.
2886 (ix86_set_current_function): Instead of doing target_reinit
2887 unconditionally, use save_target_globals_default_opts and
2888 restore_target_globals.
2890 2014-01-09 Richard Biener <rguenther@suse.de>
2892 PR tree-optimization/59715
2893 * tree-cfg.h (split_critical_edges): Declare.
2894 * tree-cfg.c (split_critical_edges): Export.
2895 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2897 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2899 * cfgexpand.c (expand_stack_vars): Optionally disable
2900 asan stack protection.
2901 (expand_used_vars): Likewise.
2902 (partition_stack_vars): Likewise.
2903 * asan.c (asan_emit_stack_protection): Optionally disable
2904 after return stack usage.
2905 (instrument_derefs): Optionally disable memory access instrumentation.
2906 (instrument_builtin_call): Likewise.
2907 (instrument_strlen_call): Likewise.
2908 (asan_protect_global): Optionally disable global variables protection.
2909 * doc/invoke.texi: Added doc for new options.
2910 * params.def: Added new options.
2911 * params.h: Likewise.
2913 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2915 PR rtl-optimization/59724
2916 * ifcvt.c (cond_exec_process_if_block): Don't call
2917 flow_find_head_matching_sequence with 0 longest_match.
2918 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2919 non-active insns if !stop_after.
2920 (try_head_merge_bb): Revert 2014-01-07 changes.
2922 2014-01-08 Jeff Law <law@redhat.com>
2924 * ree.c (get_sub_rtx): New function, extracted from...
2925 (merge_def_and_ext): Here.
2926 (combine_reaching_defs): Use get_sub_rtx.
2928 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2930 * cgraph.h (varpool_variable_node): Do not choke on null node.
2932 2014-01-08 Catherine Moore <clm@codesourcery.com>
2934 * config/mips/mips.md (simple_return): Attempt to use JRC
2936 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2938 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2940 PR rtl-optimization/59137
2941 * reorg.c (steal_delay_list_from_target): Call update_block for
2943 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2945 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2947 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2948 two duplicate entries.
2950 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2953 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2955 * config/mips/mips.c (mips_truncated_op_cost): New function.
2956 (mips_rtx_costs): Adjust test for BADDU.
2957 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2959 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2961 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2962 (*baddu_si): ...this new pattern.
2964 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2967 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2969 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2972 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2974 (expand_expr, expand_normal): Adjust.
2975 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2976 inner_reference_p. Use inner_reference_p to expand inner references.
2977 (store_expr): Adjust.
2978 * cfgexpand.c (expand_call_stmt): Adjust.
2980 2014-01-08 Rong Xu <xur@google.com>
2982 * gcov-io.c (gcov_var): Move from gcov-io.h.
2983 (gcov_position): Ditto.
2984 (gcov_is_error): Ditto.
2985 (gcov_rewrite): Ditto.
2986 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2987 only part to libgcc/libgcov.h.
2989 2014-01-08 Marek Polacek <polacek@redhat.com>
2992 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2994 2014-01-08 Marek Polacek <polacek@redhat.com>
2997 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2999 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3001 PR rtl-optimization/59649
3002 * stor-layout.c (get_mode_bounds): For BImode return
3003 0 and STORE_FLAG_VALUE.
3005 2014-01-08 Richard Biener <rguenther@suse.de>
3008 * gimple.h (is_gimple_builtin_call): Remove.
3009 (gimple_builtin_call_types_compatible_p): New.
3010 (gimple_call_builtin_p): New overload.
3011 * gimple.c (is_gimple_builtin_call): Remove.
3012 (validate_call): Rename to ...
3013 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3015 (validate_type): New static function.
3016 (gimple_call_builtin_p): New overload and adjust.
3017 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3018 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3019 (gimple_fold_stmt_to_constant_1): Likewise.
3020 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3022 2014-01-08 Richard Biener <rguenther@suse.de>
3025 * gimplify.c (gimplify_expr): Gimplify register-register type
3026 VIEW_CONVERT_EXPRs to separate stmts.
3028 2014-01-07 Jeff Law <law@redhat.com>
3031 * ree.c (combine_set_extension): Handle case where source
3032 and destination registers in an extension insn are different.
3033 (combine_reaching_defs): Allow source and destination registers
3034 in extension to be different under limited circumstances.
3035 (add_removable_extension): Remove restriction that the
3036 source and destination registers in the extension are the same.
3037 (find_and_remove_re): Emit a copy from the extension's
3038 destination to its source after the defining insn if
3039 the source and destination registers are different.
3042 * ifcvt.c (merge_if_block): If we are merging a block with more than
3043 one successor with a block with no successors, remove any BARRIER
3044 after the second block.
3046 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3048 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3050 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3053 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3054 for 14-bit register offsets when INT14_OK_STRICT is false.
3056 2014-01-07 Roland Stigge <stigge@antcom.de>
3057 Michael Meissner <meissner@linux.vnet.ibm.com>
3060 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3061 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3063 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3065 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3068 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3070 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3071 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3072 rtx is const0_rtx or not.
3074 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3077 * target-globals.c (save_target_globals): Remove this_fn_optab
3079 * toplev.c: Include optabs.h.
3080 (target_reinit): Temporarily restore the global options if another
3081 set of options are in force.
3083 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3085 PR rtl-optimization/58668
3086 * cfgcleanup.c (flow_find_cross_jump): Don't count
3087 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3088 to determine what is counted.
3089 (flow_find_head_matching_sequence): Use active_insn_p to determine
3091 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3093 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3094 determine what is counted.
3096 PR tree-optimization/59643
3097 * tree-predcom.c (split_data_refs_to_components): If one dr is
3098 read and one write, determine_offset fails and the write isn't
3099 in the bad component, just put the read into the bad component.
3101 2014-01-07 Mike Stump <mikestump@comcast.net>
3102 Jakub Jelinek <jakub@redhat.com>
3105 * tree-core.h (struct tree_optimization_option): Change optabs
3106 type from unsigned char * to void *.
3107 * optabs.c (init_tree_optimization_optabs): Adjust
3108 TREE_OPTIMIZATION_OPTABS initialization.
3110 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3113 * config/i386/i386.h (struct machine_function): Add
3114 no_drap_save_restore field.
3115 * config/i386/i386.c (ix86_save_reg): Use
3116 !cfun->machine->no_drap_save_restore instead of
3117 crtl->stack_realign_needed.
3118 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3119 this function clears frame_pointer_needed. Set
3120 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3121 and DRAP reg is needed.
3123 2014-01-06 Marek Polacek <polacek@redhat.com>
3126 * doc/implement-c.texi: Mention that other integer types are
3127 permitted as bit-field types in strictly conforming mode.
3129 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3131 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3134 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3136 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3138 2014-01-06 Martin Jambor <mjambor@suse.cz>
3141 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3143 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3145 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3149 * var-tracking.c (add_stores): Preserve the value of the source even if
3150 we don't record the store.
3152 2014-01-06 Terry Guo <terry.guo@arm.com>
3154 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3156 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3159 * config/darwin.c (darwin_function_section): Adjust return values to
3160 correspond to optimisation changes made in r206070.
3162 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3164 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3165 from prefetch_block tune setting.
3166 (nocona_cost): Correct size of prefetch block to 64.
3168 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3170 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3171 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3172 used to save the static chain register in the computation of the offset
3173 from which the FP registers need to be restored.
3175 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3177 PR tree-optimization/59519
3178 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3179 ICE if get_current_def (current_new_name) is already non-NULL, as long
3180 as it is a phi result of some other phi in *new_exit_bb that has
3183 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3184 or vmovdqu* for misaligned_operand.
3185 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3186 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3187 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3188 aligned_mem for AVX512F masked aligned load and store builtins and for
3191 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3193 PR tree-optimization/59651
3194 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3195 Address range for negative step should be added by TYPE_SIZE_UNIT.
3197 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3199 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3200 overlapping registers also for registers other than the stack pointer.
3202 2014-01-03 Marek Polacek <polacek@redhat.com>
3205 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3208 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3211 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3214 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3215 (push splitter): Use <P:MODE_SIZE> instead of
3216 GET_MODE_SIZE (<P:MODE>mode).
3217 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3218 (mov -1, reg peephole2): Likewise.
3219 * config/i386/sse.md (*mov<mode>_internal,
3220 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3221 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3222 *<code><mode>3, *andnot<mode>3<mask_name>,
3223 <mask_codefor><code><mode>3<mask_name>): Likewise.
3224 * config/i386/subst.md (mask_mode512bit_condition,
3225 sd_mask_mode512bit_condition): Likewise.
3227 2014-01-02 Xinliang David Li <davidxl@google.com>
3229 PR tree-optimization/59303
3230 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3231 (dump_predicates): Better output format.
3232 (pred_equal_p): New function.
3233 (is_neq_relop_p): Ditto.
3234 (is_neq_zero_form_p): Ditto.
3235 (pred_expr_equal_p): Ditto.
3236 (pred_neg_p): Ditto.
3237 (simplify_pred): Ditto.
3238 (simplify_preds_2): Ditto.
3239 (simplify_preds_3): Ditto.
3240 (simplify_preds_4): Ditto.
3241 (simplify_preds): Ditto.
3243 (push_to_worklist): Ditto.
3244 (get_pred_info_from_cmp): Ditto.
3245 (is_degenerated_phi): Ditto.
3246 (normalize_one_pred_1): Ditto.
3247 (normalize_one_pred): Ditto.
3248 (normalize_one_pred_chain): Ditto.
3249 (normalize_preds): Ditto.
3250 (normalize_cond_1): Remove function.
3251 (normalize_cond): Ditto.
3252 (is_gcond_subset_of): Ditto.
3253 (is_subset_of_any): Ditto.
3254 (is_or_set_subset_of): Ditto.
3255 (is_and_set_subset_of): Ditto.
3256 (is_norm_cond_subset_of): Ditto.
3257 (pred_chain_length_cmp): Ditto.
3258 (convert_control_dep_chain_into_preds): Type change.
3259 (find_predicates): Ditto.
3260 (find_def_preds): Ditto.
3261 (destroy_predicates_vecs): Ditto.
3262 (find_matching_predicates_in_rest_chains): Ditto.
3263 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3264 (is_pred_expr_subset): Ditto.
3265 (is_pred_chain_subset_of): Ditto.
3266 (is_included_in): Ditto.
3267 (is_superset_of): Ditto.
3269 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3271 Update copyright years.
3273 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3275 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3276 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3277 config/arc/arc.md, config/arc/arc.opt,
3278 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3279 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3280 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3281 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3282 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3283 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3284 vtable-verify.c, vtable-verify.h: Use the standard form for the
3287 2014-01-02 Tobias Burnus <burnus@net-b.de>
3289 * gcc.c (process_command): Update copyright notice dates.
3290 * gcov-dump.c: Ditto.
3292 * doc/cpp.texi: Bump @copying's copyright year.
3293 * doc/cppinternals.texi: Ditto.
3294 * doc/gcc.texi: Ditto.
3295 * doc/gccint.texi: Ditto.
3296 * doc/gcov.texi: Ditto.
3297 * doc/install.texi: Ditto.
3298 * doc/invoke.texi: Ditto.
3300 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3302 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3304 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3306 * config/i386/sse.md (*mov<mode>_internal): Guard
3307 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3309 PR rtl-optimization/59647
3310 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3311 new_rtx into UNSIGNED_FLOAT rtxes.
3313 Copyright (C) 2014 Free Software Foundation, Inc.
3315 Copying and distribution of this file, with or without modification,
3316 are permitted in any medium without royalty provided the copyright
3317 notice and this notice are preserved.