1 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
4 identifiers in comments.
5 (cortexa53_extra_costs): Likewise.
6 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in
8 (cortexa7_extra_costs): Likewise.
9 (cortexa12_extra_costs): Likewise.
10 (cortexa15_extra_costs): Likewise.
11 (v7m_extra_costs): Likewise.
13 2014-02-12 Richard Biener <rguenther@suse.de>
16 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
17 of posix_memalign being successful.
18 (lower_stmt): Restrict lowering of posix_memalign to when
19 -ftree-bit-ccp is enabled.
21 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
23 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
25 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
27 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
29 PR rtl-optimization/60116
30 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
31 other_insn once the combination has been validated.
33 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
36 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
38 * ipa-devirt.c: Include demangle.h
39 (odr_violation_reported): New static variable.
40 (add_type_duplicate): Update odr_violations.
41 (maybe_record_node): Add completep parameter; update it.
42 (record_target_from_binfo): Add COMPLETEP parameter;
44 (possible_polymorphic_call_targets_1): Likewise.
45 (struct polymorphic_call_target_d): Add nonconstruction_targets;
46 rename FINAL to COMPLETE.
47 (record_targets_from_bases): Sanity check we found the binfo;
48 fix COMPLETEP updating.
49 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
50 parameter, fix computing of COMPLETEP.
51 (dump_possible_polymorphic_call_targets): Imrove readability of dump; at
52 LTO time do demangling.
53 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
54 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
56 (gimple_get_virt_method_for_binfo): Likewise.
57 * gimple-fold.h (gimple_get_virt_method_for_binfo,
58 gimple_get_virt_method_for_vtable): Update prototypes.
60 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
63 * genautomata.c (add_presence_absence): Fix typo with
64 {final_}presence_list.
66 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
69 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
70 for VSX/Altivec vectors that land in GPR registers.
72 2014-02-11 Richard Henderson <rth@redhat.com>
73 Jakub Jelinek <jakub@redhat.com>
76 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
77 around drhs if type conversion to lacc->type is not useless.
79 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
81 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
83 (cortex-a57.cortex-a53): Likewise.
84 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
86 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
88 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
91 2014-02-11 Renlin Li <Renlin.Li@arm.com>
93 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
94 add_options_for_arm_vfp3.
96 2014-02-11 Jeff Law <law@redhat.com>
99 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
100 object with an undesirable mode.
102 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
105 * config/i386/sol2-9.h: New file.
106 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
107 *-*-solaris2.9*): Use it.
109 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
111 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
112 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
114 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
116 * config/microblaze/microblaze.c: Extend mcpu version format
118 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
120 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
122 2014-02-10 Richard Henderson <rth@redhat.com>
125 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
126 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
127 ms-abi vs -mno-accumulate-outgoing-args.
128 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
129 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
132 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
135 * cfgexpand.c (expand_asm_operands): Attach source location to
136 ASM_INPUT rtx objects.
137 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
139 2014-02-10 Nick Clifton <nickc@redhat.com>
141 * config/mn10300/mn10300.c (popcount): New function.
142 (mn10300_expand_prologue): Include saved registers in stack usage
145 2014-02-10 Jeff Law <law@redhat.com>
148 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
149 when changing the SET_DEST of a prior insn to avoid an input
152 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
154 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
155 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
156 -mcall-openbsd, or -mcall-linux.
157 (CC1_ENDIAN_BIG_SPEC): Remove.
158 (CC1_ENDIAN_LITTLE_SPEC): Remove.
159 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
160 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
161 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
162 and %cc1_endian_default.
163 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
165 2014-02-10 Richard Biener <rguenther@suse.de>
167 PR tree-optimization/60115
168 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
169 MEM_REF handling. Properly verify that the accesses are not
170 out of the objects bound.
172 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
174 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
177 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
179 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
180 proper constants and fix formatting.
181 (possible_polymorphic_call_targets): Fix formatting.
183 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
184 Ilya Tocar <ilya.tocar@intel.com>
186 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
187 (_mm512_loadu_epi32): Renamed into...
188 (_mm512_loadu_si512): This.
189 (_mm512_storeu_epi32): Renamed into...
190 (_mm512_storeu_si512): This.
191 (_mm512_maskz_ceil_ps): Removed.
192 (_mm512_maskz_ceil_pd): Ditto.
193 (_mm512_maskz_floor_ps): Ditto.
194 (_mm512_maskz_floor_pd): Ditto.
195 (_mm512_floor_round_ps): Ditto.
196 (_mm512_floor_round_pd): Ditto.
197 (_mm512_ceil_round_ps): Ditto.
198 (_mm512_ceil_round_pd): Ditto.
199 (_mm512_mask_floor_round_ps): Ditto.
200 (_mm512_mask_floor_round_pd): Ditto.
201 (_mm512_mask_ceil_round_ps): Ditto.
202 (_mm512_mask_ceil_round_pd): Ditto.
203 (_mm512_maskz_floor_round_ps): Ditto.
204 (_mm512_maskz_floor_round_pd): Ditto.
205 (_mm512_maskz_ceil_round_ps): Ditto.
206 (_mm512_maskz_ceil_round_pd): Ditto.
207 (_mm512_expand_pd): Ditto.
208 (_mm512_expand_ps): Ditto.
209 * config/i386/i386.c (ix86_builtins): Remove
210 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
212 * config/i386/predicates.md (const1256_operand): New.
213 (const_1_to_2_operand): Ditto.
214 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
215 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
216 (*avx512pf_gatherpf<mode>sf): Ditto.
217 (avx512pf_gatherpf<mode>df): Ditto.
218 (*avx512pf_gatherpf<mode>df_mask): Ditto.
219 (*avx512pf_gatherpf<mode>df): Ditto.
220 (avx512pf_scatterpf<mode>sf): Ditto.
221 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
222 (*avx512pf_scatterpf<mode>sf): Ditto.
223 (avx512pf_scatterpf<mode>df): Ditto.
224 (*avx512pf_scatterpf<mode>df_mask): Ditto.
225 (*avx512pf_scatterpf<mode>df): Ditto.
226 (avx512f_expand<mode>): Removed.
227 (<shift_insn><mode>3<mask_name>): Change predicate type.
229 2014-02-08 Jakub Jelinek <jakub@redhat.com>
231 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
232 not at the end of datarefs vector use ordered_remove to avoid
233 reordering datarefs vector.
236 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
237 mark local addressable non-static vars as GOVD_PRIVATE
238 instead of GOVD_LOCAL.
239 * omp-low.c (lower_omp_for): Move gimple_bind_vars
240 and BLOCK_VARS of gimple_bind_block to new_stmt rather
244 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
245 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
246 assume_aligned or alloc_align attributes.
247 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
248 arguments. Handle also assume_aligned and alloc_align attributes.
249 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
250 Handle calls to functions with assume_aligned or alloc_align
252 * doc/extend.texi: Document assume_aligned and alloc_align
255 2014-02-08 Terry Guo <terry.guo@arm.com>
257 * doc/invoke.texi: Document ARM -march=armv7e-m.
259 2014-02-08 Jakub Jelinek <jakub@redhat.com>
261 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
262 flag on __cilkrts_rethrow builtin.
265 * ipa-cp.c (determine_versionability): Fail at -O0
266 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
267 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
270 2014-02-04 Jakub Jelinek <jakub@redhat.com>
273 * tree-inline.c (copy_forbidden): Fail for
274 __attribute__((optimize (0))) functions.
276 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
278 * varpool.c: Include pointer-set.h.
279 (varpool_remove_unreferenced_decls): Variables in other partitions
280 will not be output; be however careful to not lose information
283 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
285 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
286 lookup in the vtable constructor.
288 2014-02-07 Jeff Law <law@redhat.com>
291 * config/m68k/m68k.md (ashldi_extsi): Turn into a
292 define_insn_and_split.
294 * ipa-inline.c (inline_small_functions): Fix typos.
296 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
298 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
299 (s390_can_use_return_insn): Declare.
300 * config/s390/s390.h (EPILOGUE_USES): Define.
301 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
303 (s390_chunkify_start): Handle return JUMP_LABELs.
304 (s390_early_mach): Emit a main_pool instruction on the entry edge.
305 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
306 (s390_can_use_return_insn): New functions.
307 (s390_fix_long_loop_prediction): Handle conditional returns.
308 (TARGET_SET_UP_BY_PROLOGUE): Define.
309 * config/s390/s390.md (ANY_RETURN): New code iterator.
310 (*creturn, *csimple_return, return, simple_return): New patterns.
312 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
314 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
315 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
316 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
317 REG_CFA_RESTORE list when deciding not to restore a register.
319 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
321 * config/s390/s390.c: Include tree-pass.h and context.h.
322 (s390_early_mach): New function, split out from...
323 (s390_emit_prologue): ...here.
324 (pass_data_s390_early_mach): New pass structure.
325 (pass_s390_early_mach): New class.
326 (s390_option_override): Create and register early_mach pass.
329 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
331 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
332 to match for the exit block.
334 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
336 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
337 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
338 Reject misaligned operands.
340 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
342 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
345 2014-02-07 Richard Biener <rguenther@suse.de>
348 * gimple-low.c (lower_builtin_posix_memalign): New function.
349 (lower_stmt): Call it to lower posix_memalign in a way
350 to make alignment info accessible.
352 2014-02-07 Jakub Jelinek <jakub@redhat.com>
355 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
356 __builtin_setjmp_receiver.
358 2014-02-07 Richard Biener <rguenther@suse.de>
361 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
362 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
363 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
364 Handle BUILT_IN_POSIX_MEMALIGN.
365 (find_func_clobbers): Likewise.
366 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
367 (call_may_clobber_ref_p_1): Likewise.
369 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
372 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
374 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
377 * lto-cgraph.c (lto_output_node): Use
378 symtab_get_symbol_partitioning_class.
379 (lto_output_varpool_node): likewise.
380 (symtab_get_symbol_partitioning_class): Move here from
382 * cgraph.h (symbol_partitioning_class): Likewise.
383 (symtab_get_symbol_partitioning_class): Declare.
385 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
387 * ggc.h (ggc_internal_cleared_alloc): New macro.
388 * vec.h (vec_safe_copy): Handle memory stats.
389 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
390 * target-globals.c (save_target_globals): Likewise.
392 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
395 * expr.c (emit_move_resolve_push): Export; be bit more selective
396 on when to clear alias set.
397 * expr.h (emit_move_resolve_push): Declare.
398 * function.h (struct function): Add tail_call_marked.
399 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
400 * config/i386/i386-protos.h (ix86_expand_push): Remove.
401 * config/i386/i386.md (TImode move expander): De not call
403 (FP push expanders): Preserve memory attributes.
404 * config/i386/sse.md (push<mode>1): Remove.
405 * config/i386/i386.c (ix86_expand_vector_move): Handle push
407 (ix86_expand_push): Remove.
408 * config/i386/mmx.md (push<mode>1): Remove.
410 2014-02-06 Jakub Jelinek <jakub@redhat.com>
412 PR rtl-optimization/60030
413 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
414 lopart with paradoxical subreg before shifting it up by hprec.
416 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
418 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
419 Remove extra newline at end of file.
420 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
421 (arm_issue_rate): Handle cortexa57.
422 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
423 (cortex-a57.cortex-a53): Likewise.
425 2014-02-06 Jakub Jelinek <jakub@redhat.com>
428 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
429 don't record in REG_FRAME_RELATED_EXPR registers not set in that
431 (arm_expand_prologue): Adjust all callers.
432 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
433 info, registers also at the lowest numbered registers side. Use
434 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
438 * var-tracking.c (adjust_mems): Before adding a SET
439 to amd->side_effects, adjust it's SET_SRC using
440 simplify_replace_fn_rtx.
442 2014-02-06 Alan Modra <amodra@gmail.com>
445 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
446 change SDmode to DDmode when lra_in_progress.
448 2014-02-06 Jakub Jelinek <jakub@redhat.com>
451 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
452 free_data_ref on the dr first, and before goto again also set dr
453 to the next dr. For simd_lane_access, free old datarefs[i] before
454 overwriting it. For get_vectype_for_scalar_type failure, don't
455 free_data_ref if simd_lane_access.
457 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
460 * tree.h (opts_for_fn): New inline function.
461 (opt_for_fn): Define.
462 * config/i386/i386.c (ix86_function_regparm): Use
463 opt_for_fn (decl, optimize) instead of optimize.
465 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
467 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
468 for SYMBOL_REF in large memory model.
470 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
472 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
474 (cortex-a57): Likewise.
475 (cortex-a57.cortex-a53): Likewise.
477 2014-02-06 Yury Gribov <y.gribov@samsung.com>
478 Kugan Vivekanandarajah <kuganv@linaro.org>
480 * config/arm/arm.c (arm_vector_alignment_reachable): Check
482 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
484 2014-02-06 Richard Biener <rguenther@suse.de>
486 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
487 set_loop_copy and initialize_original_copy_tables.
489 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
491 * config/aarch64/aarch64-simd.md
492 (aarch64_ashr_simddi): Change QI to SI.
494 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
495 Jakub Jelinek <jakub@redhat.com>
498 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
501 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
503 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
504 CODE_FOR_altivec_vpku[hw]um to
505 CODE_FOR_altivec_vpku[hw]um_direct.
506 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
507 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
508 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
509 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
511 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
513 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
514 generation for -maltivec=be.
515 (altivec_vsumsws): Simplify redundant test.
517 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
519 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
520 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
521 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
522 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
524 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
526 (altivec_vpks<VI_char>ss): Likewise.
527 (altivec_vpks<VI_char>us): Likewise.
528 (altivec_vpku<VI_char>us): Likewise.
529 (altivec_vpku<VI_char>um): Likewise.
530 (altivec_vpku<VI_char>um_direct): New (copy of
531 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
533 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
534 target is little endian and -maltivec=be is not specified.
535 (*altivec_vupkhs<VU_char>_direct): New (copy of
536 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
538 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
539 target is little endian and -maltivec=be is not specified.
540 (*altivec_vupkls<VU_char>_direct): New (copy of
541 altivec_vupkls<VU_char> that always emits vupkls*, for internal
543 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
544 little endian and -maltivec=be is not specified.
545 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
546 little endian and -maltivec=be is not specified.
548 2014-02-05 Richard Henderson <rth@redhat.com>
551 * combine-stack-adj.c: Revert r206943.
552 * sched-int.h (struct deps_desc): Add last_args_size.
553 * sched-deps.c (init_deps): Initialize it.
554 (sched_analyze_insn): Add OUTPUT dependencies between insns that
555 contain REG_ARGS_SIZE notes.
557 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
559 * lto-cgraph.c (asm_nodes_output): Make global.
560 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
561 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
563 (driver_handle_option): Handle OPT_fwpa.
565 2014-02-05 Jakub Jelinek <jakub@redhat.com>
568 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
569 a comment typo and formatting issue. If odr_hash hasn't been
570 created, return vNULL and set *completep to false.
573 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
574 bb with no successors.
576 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
579 * doc/invoke.texi (-march): Clarify documentation for ARM.
583 2014-02-05 Richard Biener <rguenther@suse.de>
585 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
586 when not vectorizing because of too many alias checks.
587 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
588 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
590 2014-02-05 Nick Clifton <nickc@redhat.com>
592 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
593 accept extended registers in any mode when compiling for the
596 2014-02-05 Yury Gribov <y.gribov@samsung.com>
598 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
599 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
600 sanitization attributes.
601 (can_inline_edge_p): Likewise.
602 (sanitize_attrs_match_for_inline_p): New function.
604 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
606 * ipa-prop.c (detect_type_change): Shor circuit testing of
607 type changes on THIS pointer.
609 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
612 * config/pa/pa.c (legitimize_tls_address): Return original address
613 if not passed a SYMBOL_REF rtx.
614 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
616 (pa_emit_move_sequence): Simplify TLS source operands.
617 (pa_legitimate_constant_p): Reject all TLS constants.
618 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
619 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
621 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
623 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
624 groups when we know they are controlled by LTO.
625 * varasm.c (default_binds_local_p_1): If object is in other partition,
626 it will be resolved locally.
628 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
630 * config/host-linux.c (linux_gt_pch_use_address): Don't
631 use SSIZE_MAX because it is not always defined.
633 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
636 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
637 threshold for pseudo splitting.
638 (update_ebb_live_info): Process call argument hard registers and
639 hard registers from insn definition too.
640 (max_small_class_regs_num): New constant.
641 (inherit_in_ebb): Update live hard regs through EBBs. Update
642 reloads_num only for small register classes. Don't split for
645 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
648 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
651 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
653 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
655 2014-02-04 Marek Polacek <polacek@redhat.com>
657 * gdbinit.in (pel): Define.
659 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
661 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
664 2014-02-04 Richard Biener <rguenther@suse.de>
667 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
668 in function context local.
669 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
670 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
671 similar to LTO_imported_decl_ref.
673 2014-02-04 Jakub Jelinek <jakub@redhat.com>
675 PR tree-optimization/60002
676 * cgraphclones.c (build_function_decl_skip_args): Clear
679 PR tree-optimization/60023
680 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
681 false to gsi_replace.
682 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
683 has been in some EH region and vec_stmt could throw, add
684 vec_stmt into the same EH region.
685 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
686 has no lhs, ignore it.
687 * internal-fn.c (expand_MASK_LOAD): Likewise.
690 * tree-inline.c (copy_forbidden): Fail for
691 __attribute__((optimize (0))) functions.
694 * omp-low.c (simd_clone_struct_copy): If from->inbranch
695 is set, copy one less argument.
696 (expand_simd_clones): Don't subtract clone_info->inbranch
697 from simd_clone_struct_alloc argument.
699 PR rtl-optimization/57915
700 * recog.c (simplify_while_replacing): If all unary/binary/relational
701 operation arguments are constant, attempt to simplify those.
704 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
705 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
707 2014-02-04 Richard Biener <rguenther@suse.de>
709 PR tree-optimization/60012
710 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
711 TBAA disambiguation to all DDRs.
713 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
716 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
717 (LINK_SPEC): Use it for -shared, -shared-libgcc.
719 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
722 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
724 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
726 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
727 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
729 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
732 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
733 to figure out targets of polymorphic calls with known decl.
734 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
735 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
736 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
737 (get_polymorphic_call_info): ... here.
738 (get_polymorphic_call_info_from_invariant): New function.
740 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
742 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
743 lookup via vtable pointer; check for type consistency
744 and turn inconsitent facts into UNREACHABLE.
745 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
746 * gimple-fold.c (gimple_get_virt_method_for_vtable):
747 Do not ICE on type inconsistent querries; return UNREACHABLE
750 2014-02-03 Richard Henderson <rth@twiddle.net>
753 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
754 already processed this node.
755 (normalize_one_pred_1): Pass along mark_set.
756 (normalize_one_pred): Create and destroy a pointer_set_t.
757 (normalize_one_pred_chain): Likewise.
759 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
761 PR gcov-profile/58602
762 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
764 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
767 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
768 Try to devirtualize by the knowledge of virtual table pointer given by
769 aggregate propagation.
770 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
771 ipa_print_node_jump_functions): Dump also offset that
772 is relevant for polymorphic calls.
773 (determine_known_aggregate_parts): Add arg_type parameter; use it
774 instead of determining the type from pointer type.
775 (ipa_compute_jump_functions_for_edge): Update call of
776 determine_known_aggregate_parts.
777 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
778 (gimple_get_virt_method_for_binfo): ... here; simplify using
779 vtable_pointer_value_to_vtable.
780 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
781 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
782 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
783 (vtable_pointer_value_to_vtable): Break out from ...; handle also
785 (vtable_pointer_value_to_binfo): ... here.
786 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
788 2014-02-03 Teresa Johnson <tejohnson@google.com>
790 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
791 redef of outer loop index variable.
793 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
797 * doc/extend.texi (Function Attributes): Typo.
799 2014-02-03 Cong Hou <congh@google.com>
801 PR tree-optimization/60000
802 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
803 if the vectorized statement is a store. A store statement can only
804 appear at the end of pattern statements.
806 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
808 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
809 (ix86_option_override_internal): Default long double to 64-bit for
810 32-bit Bionic and to 128-bit for 64-bit Bionic.
812 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
813 TARGET_LONG_DOUBLE_128 is true.
814 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
816 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
817 (mlong-double-64): Negate -mlong-double-128.
818 (mlong-double-128): New option.
820 * config/i386/i386-c.c (ix86_target_macros): Define
821 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
823 * doc/invoke.texi: Document -mlong-double-128.
825 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
827 PR rtl-optimization/60024
828 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
830 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
832 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
834 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
836 PR rtl-optimization/57662
837 * sel-sched.c (code_motion_path_driver): Do not mark already not
838 existing blocks in the visiting bitmap.
840 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
842 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
843 on the insn being emitted.
845 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
846 Will Deacon <will.deacon@arm.com>
848 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
850 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
852 * config/arm/arm-tables.opt: Regenerate.
854 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
856 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
857 for vector types other than V16QImode.
858 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
859 define_expand, and call altivec_expand_vec_perm_le when producing
860 code with little endian element order.
861 (*altivec_vperm_<mode>_internal): New insn having previous
862 behavior of altivec_vperm_<mode>.
863 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
864 altivec_expand_vec_perm_le when producing code with little endian
866 (*altivec_vperm_<mode>_uns_internal): New insn having previous
867 behavior of altivec_vperm_<mode>_uns.
869 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
871 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
872 (altivec_vsumsws): Add handling for -maltivec=be with a little
874 (altivec_vsumsws_direct): New.
875 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
878 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
880 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
881 vtable_pointer_value_to_binfo): New functions.
882 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
883 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
885 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
887 * config/nios2/nios2.md (load_got_register): Initialize GOT
888 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
889 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
891 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
893 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
894 preserverd by passthrough, do not propagate the type.
896 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
898 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
899 (mips_atomic_assign_expand_fenv): New function.
900 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
902 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
904 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
905 (__builtin_mips_set_fcsr): Likewise.
906 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
908 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
909 (mips16_expand_set_fcsr): Likewise.
910 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
911 (mips16_set_fcsr_stub): Likewise.
912 (mips16_get_fcsr_one_only_stub): New class.
913 (mips16_set_fcsr_one_only_stub): Likewise.
914 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
915 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
916 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
917 (hard_float): New availability predicate.
918 (mips_builtins): Add get_fcsr and set_fcsr.
919 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
920 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
921 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
922 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
923 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
926 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
928 * config/mips/mips.c (mips_one_only_stub): New class.
929 (mips_need_mips16_rdhwr_p): Replace with...
930 (mips16_rdhwr_stub): ...this new variable.
931 (mips16_stub_call_address): New function.
932 (mips16_rdhwr_one_only_stub): New class.
933 (mips_expand_thread_pointer): Use mips16_stub_call_address.
934 (mips_output_mips16_rdhwr): Delete.
935 (mips_finish_stub): New function.
936 (mips_code_end): Use it to handle rdhwr stubs.
938 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
941 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
942 when calculating size of integer atomic types.
944 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
946 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
948 2014-02-01 Jakub Jelinek <jakub@redhat.com>
950 PR tree-optimization/60003
951 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
952 * profile.c (branch_prob): Use gimple_call_builtin_p
953 to check for BUILT_IN_SETJMP_RECEIVER.
954 * tree-inline.c (copy_bb): Call notice_special_calls.
956 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
959 * lra-constraints.c (process_alt_operands): Update reload_sum only
962 2014-01-31 Richard Henderson <rth@redhat.com>
965 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
966 until after else_eh is processed.
968 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
970 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
971 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
972 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
973 in smmintrin.h, remove them.
974 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
975 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
976 * config/i386/i386.md (ROUND_SAE): Fix value.
977 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
978 (const48_operand): New.
979 * config/i386/subst.md (round), (round_expand): Use
980 const_4_or_8_to_11_operand.
981 (round_saeonly), (round_saeonly_expand): Use const48_operand.
983 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
985 * config/i386/constraints.md (Yk): Swap meaning with k.
986 * config/i386/i386.md (movhi_internal): Change Yk to k.
987 (movqi_internal): Ditto.
988 (*k<logic><mode>): Ditto.
991 (kandn<mode>): Ditto.
992 (*<code>hi_1): Ditto.
993 (*<code>qi_1): Ditto.
994 (kxnor<mode>): Ditto.
998 (*one_cmplhi2_1): Ditto.
999 (*one_cmplqi2_1): Ditto.
1000 * config/i386/sse.md (): Change k to Yk.
1001 (avx512f_load<mode>_mask): Ditto.
1002 (avx512f_blendm<mode>): Ditto.
1003 (avx512f_store<mode>_mask): Ditto.
1004 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1005 (avx512f_storedqu<mode>_mask): Ditto.
1006 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1008 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1009 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1010 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1011 (avx512f_maskcmp<mode>3): Ditto.
1012 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1013 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1014 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1015 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1016 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1017 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1018 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1019 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1020 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1021 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1022 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1023 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1024 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1025 (vec_extract_lo_<mode>_maskm): Ditto.
1026 (vec_extract_hi_<mode>_maskm): Ditto.
1027 (avx512f_vternlog<mode>_mask): Ditto.
1028 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1029 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1030 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1031 (avx512f_<code>v8div16qi2_mask): Ditto.
1032 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1033 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1034 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1035 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1036 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1037 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1038 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1039 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1040 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1041 (avx512cd_maskb_vec_dupv8di): Ditto.
1042 (avx512cd_maskw_vec_dupv16si): Ditto.
1043 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1044 (avx512f_vpermi2var<mode>3_mask): Ditto.
1045 (avx512f_vpermi2var<mode>3_mask): Ditto.
1046 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1047 (*avx512f_gathersi<mode>): Ditto.
1048 (*avx512f_gathersi<mode>_2): Ditto.
1049 (*avx512f_gatherdi<mode>): Ditto.
1050 (*avx512f_gatherdi<mode>_2): Ditto.
1051 (*avx512f_scattersi<mode>): Ditto.
1052 (*avx512f_scatterdi<mode>): Ditto.
1053 (avx512f_compress<mode>_mask): Ditto.
1054 (avx512f_compressstore<mode>_mask): Ditto.
1055 (avx512f_expand<mode>_mask): Ditto.
1056 * config/i386/subst.md (mask): Change k to Yk.
1057 (mask_scalar_merge): Ditto.
1060 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1062 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1064 2014-01-31 Richard Biener <rguenther@suse.de>
1067 * builtins.c (fold_builtin_memory_op): Make sure to not
1068 use a floating-point mode or a boolean or enumeral type for
1071 2014-01-30 DJ Delorie <dj@redhat.com>
1073 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1074 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1075 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1076 whenever main() has an epilogue.
1078 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1080 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1081 unused variable "field".
1082 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1083 (vsx_mergeh_<mode>): Likewise.
1084 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1085 (altivec_vmrghh): Likewise.
1086 (altivec_vmrghw): Likewise.
1087 (altivec_vmrglb): Likewise.
1088 (altivec_vmrglh): Likewise.
1089 (altivec_vmrglw): Likewise.
1090 (altivec_vspltb): Add missing uses.
1091 (altivec_vsplth): Likewise.
1092 (altivec_vspltw): Likewise.
1093 (altivec_vspltsf): Likewise.
1095 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1098 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1099 frame related instructions.
1101 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1103 PR rtl-optimization/59959
1104 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1105 any reload of register whose subreg is invalid.
1107 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1109 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1110 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1111 Add missing return type - void.
1113 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1115 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1116 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1117 remove element index adjustment for endian (now handled in vsx.md
1119 (altivec_expand_vec_perm_const): Use
1120 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1121 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1122 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1123 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1124 define_expand and a new define_insn *altivec_vspltb_internal;
1125 adjust for -maltivec=be on a little endian target.
1126 (altivec_vspltb_direct): New.
1127 (altivec_vsplth): Divide into a define_expand and a new
1128 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1129 little endian target.
1130 (altivec_vsplth_direct): New.
1131 (altivec_vspltw): Divide into a define_expand and a new
1132 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1133 little endian target.
1134 (altivec_vspltw_direct): New.
1135 (altivec_vspltsf): Divide into a define_expand and a new
1136 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1137 a little endian target.
1139 2014-01-30 Richard Biener <rguenther@suse.de>
1141 PR tree-optimization/59993
1142 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1143 can propagate form the earlier stmt and avoid the transform
1144 when the intermediate result is needed.
1146 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1148 * README.Portability: Fix typo.
1150 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1152 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1153 comparison_operator with ordered_comparison_operator.
1155 2014-01-30 Nick Clifton <nickc@redhat.com>
1157 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1158 Rename to mn10300_store_multiple_regs.
1159 * config/mn10300/mn10300.c: Likewise.
1160 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1161 store_multiple_regs.
1162 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1163 Call mn10300_store_multiple_regs.
1165 2014-01-30 Nick Clifton <nickc@redhat.com>
1166 DJ Delorie <dj@redhat.com>
1168 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1169 %fp 2 to keep registers after it properly word-aligned.
1170 (rl78_alloc_physical_registers_umul): Handle the case where both
1171 input operands are the same.
1173 2014-01-30 Richard Biener <rguenther@suse.de>
1175 PR tree-optimization/59903
1176 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1179 2014-01-30 Jason Merrill <jason@redhat.com>
1182 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1185 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1187 2014-01-30 Richard Biener <rguenther@suse.de>
1189 PR tree-optimization/59951
1190 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1192 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1195 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1196 SFmode to DFmode case.
1198 2014-01-29 DJ Delorie <dj@redhat.com>
1200 * config/msp430/msp430.opt (-minrt): New.
1201 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1203 (ENDFILE_SPEC): Likewise.
1205 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1207 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1208 (estimate_function_body_sizes): Use it.
1210 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1213 * dwarf2out.c (is_cxx_auto): New.
1214 (is_base_type): Use it.
1215 (gen_type_die_with_usage): Likewise.
1217 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1219 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1220 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1221 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1222 -maltivec=be with LE targets.
1223 (vsx_mergeh_<mode>): Likewise.
1224 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1225 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1226 (altivec_vmrghb): Replace with define_expand and new
1227 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1228 (altivec_vmrghb_direct): New define_insn.
1229 (altivec_vmrghh): Replace with define_expand and new
1230 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1231 (altivec_vmrghh_direct): New define_insn.
1232 (altivec_vmrghw): Replace with define_expand and new
1233 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1234 (altivec_vmrghw_direct): New define_insn.
1235 (*altivec_vmrghsf): Adjust for endianness.
1236 (altivec_vmrglb): Replace with define_expand and new
1237 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1238 (altivec_vmrglb_direct): New define_insn.
1239 (altivec_vmrglh): Replace with define_expand and new
1240 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1241 (altivec_vmrglh_direct): New define_insn.
1242 (altivec_vmrglw): Replace with define_expand and new
1243 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1244 (altivec_vmrglw_direct): New define_insn.
1245 (*altivec_vmrglsf): Adjust for endianness.
1246 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1247 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1248 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1249 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1250 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1251 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1252 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1253 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1255 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1257 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1258 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1261 2014-01-29 Richard Biener <rguenther@suse.de>
1263 PR tree-optimization/58742
1264 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1265 associate_pointerplus_align.
1266 (associate_pointerplus_diff): New function.
1267 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1268 and associate_pointerplus_diff.
1270 2014-01-29 Richard Biener <rguenther@suse.de>
1272 * lto-streamer.h (LTO_major_version): Bump to 3.
1273 (LTO_minor_version): Reset to 0.
1275 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1277 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1278 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1279 (arm_file_start): Generate correct asm header for armv7ve.
1280 * config/arm/bpabi.h: Add multilib support for armv7ve.
1281 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1282 and cortex-a15 to armv7ve.
1283 * config/arm/t-aprofile: Add multilib support for armv7ve.
1284 * doc/invoke.texi: Document -march=armv7ve.
1286 2014-01-29 Richard Biener <rguenther@suse.de>
1288 PR tree-optimization/58742
1289 * tree-ssa-forwprop.c (associate_plusminus): Return true
1290 if we changed sth, defer EH cleanup to ...
1291 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1292 (simplify_mult): New function.
1294 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1297 PR tree-optimization/59920
1298 * tree.c (build_common_builtin_nodes): Remove
1299 __builtin_setjmp_dispatcher initialization.
1300 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1301 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1302 instead of gsi_after_labels + manually skipping debug stmts.
1303 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1304 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1305 * tree-inline.c (copy_edges_for_bb): Remove
1306 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1307 argument. Ignore computed_goto_p stmts. Don't call
1308 make_abnormal_goto_edges. If a call might need abnormal edges
1309 for non-local gotos, see if it already has an edge to
1310 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1311 with true argument, don't do anything then, otherwise add
1312 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1313 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1315 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1316 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1317 (lower_stmt): Don't set data->calls_builtin_setjmp.
1318 (lower_builtin_setjmp): Adjust comment.
1319 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1320 * tree-cfg.c (found_computed_goto): Remove.
1321 (factor_computed_gotos): Remove.
1322 (make_goto_expr_edges): Return bool, true for computed gotos.
1323 Don't call make_abnormal_goto_edges.
1324 (build_gimple_cfg): Don't set found_computed_goto, don't call
1325 factor_computed_gotos.
1326 (computed_goto_p): No longer static.
1327 (make_blocks): Don't set found_computed_goto.
1328 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1329 (make_edges): If make_goto_expr_edges returns true, push bb
1330 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1331 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1332 vector. Record mapping between bbs and OpenMP regions if there
1333 are any, adjust make_gimple_omp_edges caller. Call
1334 handle_abnormal_edges.
1335 (make_abnormal_goto_edges): Remove.
1336 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1337 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1338 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1339 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1340 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1341 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1342 filling *region also set *region_idx to (*region)->entry->index.
1345 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1346 For REGs set ORIGINAL_REGNO.
1348 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1350 * doc/md.texi: Mention that a target shouldn't implement
1351 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1354 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1356 PR tree-optimization/59594
1357 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1358 a copy of the datarefs vector rather than the vector itself.
1360 2014-01-28 Jason Merrill <jason@redhat.com>
1363 * dwarf2out.c (auto_die): New static.
1364 (gen_type_die_with_usage): Handle C++1y 'auto'.
1365 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1368 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1371 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1372 (SPEC_X32): Likewise.
1373 (SPEC_64): Likewise.
1374 * config/i386/i386.c (ix86_option_override_internal): Turn off
1375 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1377 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1378 * config/i386/i386.h (TARGET_16BIT): New macro.
1379 (TARGET_16BIT_P): Likewise.
1380 * config/i386/i386.opt: Add m16.
1381 * doc/invoke.texi: Document -m16.
1383 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1385 PR preprocessor/59935
1386 * input.c (location_get_source_line): Bail out on when line number
1387 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1389 2014-01-28 Richard Biener <rguenther@suse.de>
1391 PR tree-optimization/58742
1392 * tree-ssa-forwprop.c (associate_plusminus): Handle
1393 pointer subtraction of the form (T)(P + A) - (T)P.
1395 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1397 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1400 2014-01-28 Richard Biener <rguenther@suse.de>
1403 2014-01-28 Richard Biener <rguenther@suse.de>
1405 PR rtl-optimization/45364
1406 PR rtl-optimization/59890
1407 * var-tracking.c (local_get_addr_clear_given_value): Handle
1408 already cleared slot.
1409 (val_reset): Handle not allocated local_get_addr_cache.
1410 (vt_find_locations): Use post-order on the inverted CFG.
1412 2014-01-28 Richard Biener <rguenther@suse.de>
1414 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1416 2014-01-28 Richard Biener <rguenther@suse.de>
1418 PR rtl-optimization/45364
1419 PR rtl-optimization/59890
1420 * var-tracking.c (local_get_addr_clear_given_value): Handle
1421 already cleared slot.
1422 (val_reset): Handle not allocated local_get_addr_cache.
1423 (vt_find_locations): Use post-order on the inverted CFG.
1425 2014-01-28 Alan Modra <amodra@gmail.com>
1427 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1428 * configure.ac <recursive call for build != host>: Define
1429 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1430 and LD_FOR_BUILD too.
1431 * configure: Regenerate.
1433 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1435 * config/i386/i386.c (get_builtin_code_for_version): Separate
1436 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1437 Broadwell from Haswell.
1439 2014-01-27 Steve Ellcey <sellcey@mips.com>
1441 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1442 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1443 * config/mips/mips.c (mips_option_override): Change setting
1445 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1446 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1447 Change from Mask to Var.
1449 2014-01-27 Jeff Law <law@redhat.com>
1451 * ipa-inline.c (inline_small_functions): Fix typo.
1453 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1455 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1456 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1457 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1458 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1459 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1460 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1461 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1462 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1463 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1464 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1465 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1466 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1467 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1468 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1469 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1470 (_mm512_storeu_epi64): Ditto.
1471 (_mm512_cmpge_epi32_mask): Ditto.
1472 (_mm512_cmpge_epu32_mask): Ditto.
1473 (_mm512_cmpge_epi64_mask): Ditto.
1474 (_mm512_cmpge_epu64_mask): Ditto.
1475 (_mm512_cmple_epi32_mask): Ditto.
1476 (_mm512_cmple_epu32_mask): Ditto.
1477 (_mm512_cmple_epi64_mask): Ditto.
1478 (_mm512_cmple_epu64_mask): Ditto.
1479 (_mm512_cmplt_epi32_mask): Ditto.
1480 (_mm512_cmplt_epu32_mask): Ditto.
1481 (_mm512_cmplt_epi64_mask): Ditto.
1482 (_mm512_cmplt_epu64_mask): Ditto.
1483 (_mm512_cmpneq_epi32_mask): Ditto.
1484 (_mm512_cmpneq_epu32_mask): Ditto.
1485 (_mm512_cmpneq_epi64_mask): Ditto.
1486 (_mm512_cmpneq_epu64_mask): Ditto.
1487 (_mm512_expand_pd): Ditto.
1488 (_mm512_expand_ps): Ditto.
1489 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1490 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1491 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1492 * config/i386/i386.c (ix86_builtins): Add
1493 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1494 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1495 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1496 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1497 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1498 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1499 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1500 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1501 IX86_BUILTIN_PMOVUSQW512_MEM.
1502 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1503 __builtin_ia32_pmovsqd512mem_mask,
1504 __builtin_ia32_pmovqd512mem_mask,
1505 __builtin_ia32_pmovusqw512mem_mask,
1506 __builtin_ia32_pmovsqw512mem_mask,
1507 __builtin_ia32_pmovqw512mem_mask,
1508 __builtin_ia32_pmovusdw512mem_mask,
1509 __builtin_ia32_pmovsdw512mem_mask,
1510 __builtin_ia32_pmovdw512mem_mask,
1511 __builtin_ia32_pmovqb512mem_mask,
1512 __builtin_ia32_pmovusqb512mem_mask,
1513 __builtin_ia32_pmovsqb512mem_mask,
1514 __builtin_ia32_pmovusdb512mem_mask,
1515 __builtin_ia32_pmovsdb512mem_mask,
1516 __builtin_ia32_pmovdb512mem_mask.
1517 (bdesc_args): Add __builtin_ia32_expanddf512,
1518 __builtin_ia32_expandsf512.
1519 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1520 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1521 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1522 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1523 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1524 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1525 (avx512f_<code>v8div16qi2_mask_store): This.
1526 (avx512f_expand<mode>): New.
1528 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1530 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1532 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1533 (_mm512_prefetch_i32scatter_pd): Ditto.
1534 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1535 (_mm512_prefetch_i64scatter_pd): Ditto.
1536 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1537 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1538 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1539 (_mm512_prefetch_i32scatter_ps): Ditto.
1540 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1541 (_mm512_prefetch_i64scatter_ps): Ditto.
1542 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1543 * config/i386/i386-builtin-types.def: Define
1544 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1545 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1546 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1547 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1548 IX86_BUILTIN_SCATTERPFQPD.
1549 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1550 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1551 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1552 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1553 __builtin_ia32_scatterpfqps.
1554 (ix86_expand_builtin): Expand new built-ins.
1555 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1556 fix memory access data type.
1557 (*avx512pf_gatherpf<mode>_mask): Ditto.
1558 (*avx512pf_gatherpf<mode>): Ditto.
1559 (avx512pf_scatterpf<mode>): Ditto.
1560 (*avx512pf_scatterpf<mode>_mask): Ditto.
1561 (*avx512pf_scatterpf<mode>): Ditto.
1562 (GATHER_SCATTER_SF_MEM_MODE): New.
1563 (avx512pf_gatherpf<mode>df): Ditto.
1564 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1565 (*avx512pf_scatterpf<mode>df): Ditto.
1567 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1570 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1571 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1574 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1576 * common/config/arm/arm-common.c
1577 (arm_rewrite_mcpu): Handle multiple names.
1579 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1581 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1583 * gimple-builder.h (create_gimple_tmp): Delete.
1585 2014-01-27 Christian Bruel <christian.bruel@st.com>
1587 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1590 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1592 * config/pa/pa.md (call): Generate indirect long calls to non-local
1593 functions when outputing 32-bit code.
1594 (call_value): Likewise except for special call to buggy powf function.
1596 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1597 portable runtime and PIC indirect calls.
1598 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1599 and PIC call sequences. Use ldo instead of blr to set return register
1600 in PIC call sequence.
1602 2014-01-25 Walter Lee <walt@tilera.com>
1604 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1605 avoid clobbering a live register.
1607 2014-01-25 Walter Lee <walt@tilera.com>
1609 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1610 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1611 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1612 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1614 2014-01-25 Walter Lee <walt@tilera.com>
1616 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1617 arguments on even registers.
1618 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1620 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1621 (BIGGEST_ALIGNMENT): Ditto.
1622 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1624 2014-01-25 Walter Lee <walt@tilera.com>
1626 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1627 insns before bundling.
1628 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1630 2014-01-25 Walter Lee <walt@tilera.com>
1632 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1633 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1634 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1636 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1638 * config/mips/constraints.md (kl): Delete.
1639 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1640 define expands, using...
1641 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1642 instructions for MIPS16.
1643 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1644 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1646 2014-01-25 Walter Lee <walt@tilera.com>
1648 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1652 2014-01-25 Walter Lee <walt@tilera.com>
1654 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1655 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1657 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1659 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1662 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1664 * print-rtl.c (in_call_function_usage): New var.
1665 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1666 EXPR_LIST mode as mode and not as reg note name.
1669 * cfgloopmanip.c (copy_loop_info): If
1670 loop->warned_aggressive_loop_optimizations, make sure
1671 the flag is set in target loop too.
1673 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1675 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1677 * builtins.def: Likewise.
1678 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1679 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1680 * ira.c (ira_setup_eliminable_regset): Likewise.
1681 * omp-low.c (gate_expand_omp): Likewise.
1682 (execute_lower_omp): Likewise.
1683 (diagnose_sb_0): Likewise.
1684 (gate_diagnose_omp_blocks): Likewise.
1685 (simd_clone_clauses_extract): Likewise.
1688 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1690 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1691 correction for little endian...
1692 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1695 2014-01-24 Jeff Law <law@redhat.com>
1697 PR tree-optimization/59919
1698 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1699 for non-returning calls.
1701 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1703 * common/config/aarch64/aarch64-common.c
1704 (aarch64_rewrite_mcpu): Handle multiple names.
1705 * config/aarch64/aarch64.h
1706 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1708 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1710 * input.c (add_file_to_cache_tab): Handle the case where fopen
1713 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1716 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1717 from push operand if code of push isn't PRE_DEC.
1719 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1722 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1723 -mquad-memory-atomic. Update -mquad-memory documentation to say
1724 it is only used for non-atomic loads/stores.
1726 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1727 -mquad-memory or -mquad-memory-atomic switches.
1729 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1730 -mquad-memory-atomic to ISA 2.07 support.
1732 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1733 to separate support of normal quad word memory operations (ldq, stq)
1734 from the atomic quad word memory operations.
1736 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1737 support to separate non-atomic quad word operations from atomic
1738 quad word operations. Disable non-atomic quad word operations in
1739 little endian mode so that we don't have to swap words after the
1740 load and before the store.
1741 (quad_load_store_p): Add comment about atomic quad word support.
1742 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1743 options printed with -mdebug=reg.
1745 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1746 -mquad-memory-atomic as the test for whether we have quad word
1747 atomic instructions.
1748 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1749 or -mp8-vector are used, allow byte/half-word atomic operations.
1751 * config/rs6000/sync.md (load_lockedti): Insure that the address
1752 is a proper indexed or indirect address for the lqarx instruction.
1753 On little endian systems, swap the hi/lo registers after the lqarx
1755 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1756 insure the address is valid for the lqarx instruction.
1757 (store_conditionalti): Insure that the address is a proper indexed
1758 or indirect address for the stqcrx. instruction. On little endian
1759 systems, swap the hi/lo registers before doing the stqcrx.
1761 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1762 insure the address is valid for the stqcrx. instruction.
1764 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1765 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1766 type of quad memory support is available.
1768 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1771 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1772 there is a danger of looping.
1774 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1776 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1777 force flag_ira_loop_pressure if set via command line.
1779 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1781 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1782 (ashr_simd): New builtin handling DI mode.
1783 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1784 (aarch64_sshr_simddi): New match pattern.
1785 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1786 (vshrd_n_s64): Likewise.
1787 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1789 2014-01-23 Nick Clifton <nickc@redhat.com>
1791 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1792 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1793 favour of mcu specific scripts.
1794 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1797 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1798 Alex Velenko <Alex.Velenko@arm.com>
1800 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1801 (vaddv_s16): Likewise.
1802 (vaddv_s32): Likewise.
1803 (vaddv_u8): Likewise.
1804 (vaddv_u16): Likewise.
1805 (vaddv_u32): Likewise.
1806 (vaddvq_s8): Likewise.
1807 (vaddvq_s16): Likewise.
1808 (vaddvq_s32): Likewise.
1809 (vaddvq_s64): Likewise.
1810 (vaddvq_u8): Likewise.
1811 (vaddvq_u16): Likewise.
1812 (vaddvq_u32): Likewise.
1813 (vaddvq_u64): Likewise.
1814 (vaddv_f32): Likewise.
1815 (vaddvq_f32): Likewise.
1816 (vaddvq_f64): Likewise.
1817 (vmaxv_f32): Likewise.
1818 (vmaxv_s8): Likewise.
1819 (vmaxv_s16): Likewise.
1820 (vmaxv_s32): Likewise.
1821 (vmaxv_u8): Likewise.
1822 (vmaxv_u16): Likewise.
1823 (vmaxv_u32): Likewise.
1824 (vmaxvq_f32): Likewise.
1825 (vmaxvq_f64): Likewise.
1826 (vmaxvq_s8): Likewise.
1827 (vmaxvq_s16): Likewise.
1828 (vmaxvq_s32): Likewise.
1829 (vmaxvq_u8): Likewise.
1830 (vmaxvq_u16): Likewise.
1831 (vmaxvq_u32): Likewise.
1832 (vmaxnmv_f32): Likewise.
1833 (vmaxnmvq_f32): Likewise.
1834 (vmaxnmvq_f64): Likewise.
1835 (vminv_f32): Likewise.
1836 (vminv_s8): Likewise.
1837 (vminv_s16): Likewise.
1838 (vminv_s32): Likewise.
1839 (vminv_u8): Likewise.
1840 (vminv_u16): Likewise.
1841 (vminv_u32): Likewise.
1842 (vminvq_f32): Likewise.
1843 (vminvq_f64): Likewise.
1844 (vminvq_s8): Likewise.
1845 (vminvq_s16): Likewise.
1846 (vminvq_s32): Likewise.
1847 (vminvq_u8): Likewise.
1848 (vminvq_u16): Likewise.
1849 (vminvq_u32): Likewise.
1850 (vminnmv_f32): Likewise.
1851 (vminnmvq_f32): Likewise.
1852 (vminnmvq_f64): Likewise.
1854 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1856 * config/aarch64/aarch64-simd.md
1857 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1858 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1859 (*aarch64_mul3_elt<mode>): Likewise.
1860 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1861 (*aarch64_mul3_elt_to_64v2df): Likewise.
1862 (*aarch64_mla_elt<mode>): Likewise.
1863 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1864 (*aarch64_mls_elt<mode>): Likewise.
1865 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1866 (*aarch64_fma4_elt<mode>): Likewise.
1867 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1868 (*aarch64_fma4_elt_to_64v2df): Likewise.
1869 (*aarch64_fnma4_elt<mode>): Likewise.
1870 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1871 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1872 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1873 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1874 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1875 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1876 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1877 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1878 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1880 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1882 * config/aarch64/aarch64-simd.md
1883 (aarch64_be_checked_get_lane<mode>): New define_expand.
1884 * config/aarch64/aarch64-simd-builtins.def
1885 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1886 New builtin definition.
1887 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1888 Use new safe be builtin.
1890 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1892 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1894 (aarch64_be_st1<mode>): Likewise.
1895 (aarch_ld1<VALL:mode>): Define_expand modified.
1896 (aarch_st1<VALL:mode>): Likewise.
1897 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1898 (UNSPEC_ST1): Likewise.
1900 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1902 * config/microblaze/microblaze.md: Add trap insn and attribute
1904 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1906 PR preprocessor/58580
1907 * input.h (location_get_source_line): Take an additional line_size
1909 (void diagnostics_file_cache_fini): Declare new function.
1910 * input.c (struct fcache): New type.
1911 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1912 New static constants.
1913 (diagnostic_file_cache_init, total_lines_num)
1914 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1915 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1916 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1917 (get_next_line, read_next_line, goto_next_line, read_line_num):
1918 New static function definitions.
1919 (diagnostic_file_cache_fini): New function.
1920 (location_get_source_line): Take an additional output line_len
1921 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1923 * diagnostic.c (diagnostic_finish): Call
1924 diagnostic_file_cache_fini.
1925 (adjust_line): Take an additional input parameter for the length
1926 of the line, rather than calculating it with strlen.
1927 (diagnostic_show_locus): Adjust the use of
1928 location_get_source_line and adjust_line with respect to their new
1929 signature. While displaying a line now, do not stop at the first
1930 null byte. Rather, display the zero byte as a space and keep
1931 going until we reach the size of the line.
1932 * Makefile.in: Add vec.o to OBJS-libcommon
1934 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1935 Ilya Tocar <ilya.tocar@intel.com>
1937 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1938 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1939 (__builtin_ia32_kmov16): Ditto.
1940 * config/i386/i386.md (UNSPEC_KMOV): New.
1943 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1945 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1946 (_mm512_storeu_si512): Ditto.
1948 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1951 * rtl.h (get_referenced_operands): Declare.
1952 * recog.c (get_referenced_operands): New function.
1953 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1954 operands have been referenced when recording LO_SUM references.
1956 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1958 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1960 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1962 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1963 Enable for generic and recent AMD targets.
1965 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1967 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1968 ARG_SIZE note when adjustment was eliminated.
1970 2014-01-22 Jeff Law <law@redhat.com>
1972 PR tree-optimization/59597
1973 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1974 in file. Accept new argument REGISTERING and use it to modify
1975 dump output appropriately.
1976 (register_jump_thread): Corresponding changes.
1977 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1978 thread paths involving joiner blocks. Add code to dump cancelled
1979 jump threading paths.
1981 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1983 PR rtl-optimization/59477
1984 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1985 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1987 2014-01-22 Tom Tromey <tromey@redhat.com>
1989 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1991 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1993 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1995 PR rtl-optimization/59896
1996 * lra-constraints.c (process_alt_operands): Check unused note for
1997 matched operands of insn with no output reloads.
1999 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2001 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2002 (mips_move_from_gpr_cost): Likewise.
2004 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2006 PR rtl-optimization/59858
2007 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2008 ira_class_hard_regs_num.
2009 (process_alt_operands): Increase reject for dying matched operand.
2011 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2014 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2015 smaller than size, perform several stores or loads and stores
2016 at dst + count - size to store or copy all of size bytes, rather
2017 than just last modesize bytes.
2019 2014-01-20 DJ Delorie <dj@redhat.com>
2021 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2022 that CLOBBERs are REGs before propogating their values.
2024 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2027 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2028 (cgraph_inline_failed_type): New function.
2029 * cgraph.h (DEFCIFCODE): Add type.
2030 (cgraph_inline_failed_type_t): New enum.
2031 (cgraph_inline_failed_type): New prototype.
2032 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2033 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2034 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2035 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2036 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2037 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2038 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2039 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2040 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2041 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2042 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2043 OPTIMIZATION_MISMATCH.
2044 * tree-inline.c (expand_call_inline): Emit errors during
2045 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2047 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2050 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2051 mode attribute in insn output.
2053 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2055 * output.h (output_constant): Delete.
2056 * varasm.c (output_constant): Make private.
2058 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2060 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2062 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2065 * tree.h (fold_builtin_strcat): New prototype.
2066 * builtins.c (fold_builtin_strcat): No longer static. Add len
2067 argument, if non-NULL, don't call c_strlen. Optimize
2068 directly into __builtin_memcpy instead of __builtin_strcpy.
2069 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2070 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2072 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2074 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2075 for SImode_address_operand operands, having only a REG argument.
2077 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2079 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2080 loader name using mbig-endian.
2081 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2083 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2085 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2089 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2091 * config/aarch64/aarch64-protos.h
2092 (aarch64_cannot_change_mode_class_ptr): Declare.
2093 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2094 aarch64_cannot_change_mode_class_ptr): New.
2095 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2096 backend hook aarch64_cannot_change_mode_class.
2098 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2100 * common/config/aarch64/aarch64-common.c
2101 (aarch64_handle_option): Don't handle any option order logic here.
2102 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2103 selected_cpu, warn on architecture version mismatch.
2104 (aarch64_override_options): Fix parsing order for option strings.
2106 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2107 Iain Sandoe <iain@codesourcery.com>
2110 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2111 warning. Amend comment to reflect current functionality.
2113 2014-01-20 Richard Biener <rguenther@suse.de>
2116 * builtins.c (fold_builtin_strcat): Remove case better handled
2117 by tree-ssa-strlen.c.
2119 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2121 * config/aarch64/aarch64.opt
2122 (mcpu, march, mtune): Make case-insensitive.
2124 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2127 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2128 if operands[1] is a REG or ZERO_EXTEND of a REG.
2130 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2132 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2134 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2136 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2137 long non-pic millicode calls.
2139 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2141 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2143 2014-01-19 Kito Cheng <kito@0xlab.org>
2145 * builtins.c (expand_movstr): Check movstr expand done or fail.
2147 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2148 H.J. Lu <hongjiu.lu@intel.com>
2151 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2152 to DImode for zero-extended addresses.
2154 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2156 PR rtl-optimization/57763
2157 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2158 on the new indirect jump_insn and increment LABEL_NUSES (label).
2160 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2164 * config.gcc (x86_archs): New variable.
2165 (x86_64_archs): Likewise.
2166 (x86_cpus): Likewise.
2167 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2168 --with-arch/--with-cpu= options.
2169 Support --with-arch=/--with-cpu={nehalem,westmere,
2170 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2172 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2174 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2175 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2177 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2179 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2181 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2184 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2185 clear cpp_get_options (parse_in)->warn_unused_macros for
2186 ix86_target_macros_internal with cpp_define.
2188 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2190 * jump.c (delete_related_insns): Keep (use (insn))s.
2191 * reorg.c (redundant_insn): Check for barriers too.
2193 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2195 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2197 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2199 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2200 call to $$dyncall when TARGET_LONG_CALLS is true.
2202 2014-01-17 Jeff Law <law@redhat.com>
2204 * ree.c (combine_set_extension): Temporarily disable test for
2205 changing number of hard registers.
2207 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2210 * ipa-inline-analysis.c (inline_free_summary):
2211 Do not free summary of aliases.
2213 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2216 * gimplify.c (gimplify_expr): Use create_tmp_var
2217 instead of create_tmp_var_raw. If cond doesn't have
2218 integral type, don't add the IFN_ANNOTATE builtin at all.
2220 2014-01-17 Martin Jambor <mjambor@suse.cz>
2223 * ipa-cp.c (prev_edge_clone): New variable.
2224 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2225 Also resize prev_edge_clone vector.
2226 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2227 (ipcp_edge_removal_hook): New function.
2228 (ipcp_driver): Register ipcp_edge_removal_hook.
2230 2014-01-17 Andrew Pinski <apinski@cavium.com>
2231 Steve Ellcey <sellcey@mips.com>
2234 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2237 2014-01-17 Jeff Law <law@redhat.com>
2240 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2241 so that pass_ccp runs first.
2243 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2245 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2246 (ix86_adjust_cost): Use !TARGET_XXX.
2247 (do_reorder_for_imul): Likewise.
2248 (swap_top_of_ready_list): Likewise.
2249 (ix86_sched_reorder): Likewise.
2251 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2253 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2254 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2255 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2256 (intel_memset): New. Duplicate slm_memset.
2257 (intel_cost): New. Duplicate slm_cost.
2258 (m_INTEL): New macro.
2259 (processor_target_table): Add "intel".
2260 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2261 with PROCESSOR_INTEL for "intel".
2262 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2263 PROCESSOR_SILVERMONT.
2264 (ix86_issue_rate): Likewise.
2265 (ix86_adjust_cost): Likewise.
2266 (ia32_multipass_dfa_lookahead): Likewise.
2267 (swap_top_of_ready_list): Likewise.
2268 (ix86_sched_reorder): Likewise.
2269 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2270 instead of TARGET_OPT_AGU.
2271 * config/i386/i386.h (TARGET_INTEL): New.
2272 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2273 (processor_type): Add PROCESSOR_INTEL.
2274 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2275 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2277 2014-01-17 Marek Polacek <polacek@redhat.com>
2280 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2283 2014-01-17 Richard Biener <rguenther@suse.de>
2285 PR tree-optimization/46590
2286 * opts.c (default_options_table): Add entries for
2287 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2288 all enabled at -O1 but not for -Og.
2289 * common.opt (fbranch-count-reg): Remove Init(1).
2290 (fmove-loop-invariants): Likewise.
2291 (ftree-pta): Likewise.
2293 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2295 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2296 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2297 decls to at least the GCC 4.8 used alignments.
2300 * tree-nested.c (convert_nonlocal_reference_stmt,
2301 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2302 of GIMPLE_BIND stmts, adjust associated decls.
2304 2014-01-17 Richard Biener <rguenther@suse.de>
2306 PR tree-optimization/46590
2307 * vec.h (vec<>::bseach): New member function implementing
2308 binary search according to C89 bsearch.
2309 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2310 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2311 bitmap pointer again. Make accesses_in_loop a flat array.
2312 (mem_ref_obstack): New global.
2313 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2314 (mark_ref_stored): Likewise.
2315 (ref_indep_loop_p_2): Likewise.
2316 (set_ref_stored_in_loop): New helper function.
2317 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2318 (memref_free): Adjust.
2319 (record_mem_ref_loc): Simplify.
2320 (gather_mem_refs_stmt): Adjust.
2321 (sort_locs_in_loop_postorder_cmp): New function.
2322 (analyze_memory_references): Sort accesses_in_loop after
2323 loop postorder number.
2324 (find_ref_loc_in_loop_cmp): New function.
2325 (for_all_locs_in_loop): Find relevant cluster of locs in
2326 accesses_in_loop and iterate without recursion.
2327 (execute_sm): Avoid uninit warning.
2328 (struct ref_always_accessed): Simplify.
2329 (ref_always_accessed::operator ()): Likewise.
2330 (ref_always_accessed_p): Likewise.
2331 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2332 loop postorder numbers here.
2333 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2336 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2339 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2340 on decls for which assemble_alias has been called.
2342 2014-01-17 Nick Clifton <nickc@redhat.com>
2344 * config/msp430/msp430.opt: (mcpu): New option.
2345 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2346 (msp430_option_override): Parse target_cpu. If the MCU name
2347 matches a generic string, clear target_mcu.
2348 (msp430_attr): Allow numeric interrupt values up to 63.
2349 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2350 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2352 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2354 * config/msp430/msp430.md (popm): Use %J rather than %I.
2355 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2356 (addhi_cy_i): Use immediate_operand for operand 2.
2357 * doc/invoke.texi: Document -mcpu option.
2359 2014-01-17 Richard Biener <rguenther@suse.de>
2361 PR rtl-optimization/38518
2362 * df.h (df_analyze_loop): Declare.
2363 * df-core.c: Include cfgloop.h.
2364 (df_analyze_1): Split out main part of df_analyze.
2365 (df_analyze): Adjust.
2366 (loop_inverted_post_order_compute): New function.
2367 (loop_post_order_compute): Likewise.
2368 (df_analyze_loop): New function avoiding whole-function
2370 * loop-invariant.c (find_defs): Use df_analyze_loop.
2371 (find_invariants): Adjust.
2372 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2374 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2376 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2377 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2379 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2381 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2382 traversal when removing references.
2384 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2387 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2389 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2392 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2393 pushing a reload for an autoinc when we had previously reloaded an
2394 inner part of the address.
2396 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2398 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2400 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2401 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2402 when not giving up or versioning for alias only because of
2404 (vect_analyze_data_ref_dependences): Set to true.
2405 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2407 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2408 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2412 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2415 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2416 operand 0 predicate for gathers, use a new pseudo as subtarget.
2418 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2421 * lra-constraints.c (process_alt_operands): Add printing debug info.
2422 Check absence of input/output reloads for matched operands too.
2424 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2426 PR rtl-optimization/59835
2427 * ira.c (ira_init_register_move_cost): Increase cost for
2430 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2432 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2434 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2437 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2438 non-register objects. Use gen_(high/low)part more consistently.
2441 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2444 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2445 endian support, remove tests for WORDS_BIG_ENDIAN.
2446 (p8_mfvsrd_3_<mode>): Likewise.
2447 (reload_gpr_from_vsx<mode>): Likewise.
2448 (reload_gpr_from_vsxsf): Likewise.
2449 (p8_mfvsrd_4_disf): Likewise.
2451 2014-01-16 Richard Biener <rguenther@suse.de>
2453 PR rtl-optimization/46590
2454 * lcm.c (compute_antinout_edge): Use postorder iteration.
2455 (compute_laterin): Use inverted postorder iteration.
2457 2014-01-16 Nick Clifton <nickc@redhat.com>
2460 * varasm.c (output_constant): Return the number of bytes actually
2462 (output_constructor_array_range): Update the field size with the
2463 number of bytes emitted by output_constant.
2464 (output_constructor_regular_field): Likewise. Also do not
2465 complain if the total number of bytes emitted is now greater
2466 than the expected fieldpos.
2467 * output.h (output_constant): Update prototype and descriptive comment.
2469 2014-01-16 Marek Polacek <polacek@redhat.com>
2472 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2473 it is error_mark_node.
2475 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2477 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2478 VALID_AVX256_REG_OR_OI_MODE.
2480 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2482 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2483 current procedure should be profiled.
2485 2014-01-15 Andrew Pinski <apinski@cavium.com>
2487 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2488 of moving from/to the STACK_REG register class.
2490 2014-01-15 Richard Henderson <rth@redhat.com>
2493 * reginfo.c (global_regs_decl): Globalize.
2494 * rtl.h (global_regs_decl): Declare.
2495 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2496 reserved via global_regs.
2498 2014-01-15 Teresa Johnson <tejohnson@google.com>
2500 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2502 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2504 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2505 and vmulosh rather than call gen_vec_widen_smult_*.
2506 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2507 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2508 (vec_widen_smult_even_v16qi): Likewise.
2509 (vec_widen_umult_even_v8hi): Likewise.
2510 (vec_widen_smult_even_v8hi): Likewise.
2511 (vec_widen_umult_odd_v16qi): Likewise.
2512 (vec_widen_smult_odd_v16qi): Likewise.
2513 (vec_widen_umult_odd_v8hi): Likewise.
2514 (vec_widen_smult_odd_v8hi): Likewise.
2515 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2516 vmuloub rather than call gen_vec_widen_umult_*.
2517 (vec_widen_umult_lo_v16qi): Likewise.
2518 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2519 vmulosb rather than call gen_vec_widen_smult_*.
2520 (vec_widen_smult_lo_v16qi): Likewise.
2521 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2522 rather than call gen_vec_widen_umult_*.
2523 (vec_widen_umult_lo_v8hi): Likewise.
2524 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2525 rather than call gen_vec_widen_smult_*.
2526 (vec_widen_smult_lo_v8hi): Likewise.
2528 2014-01-15 Jeff Law <law@redhat.com>
2530 PR tree-optimization/59747
2531 * ree.c (find_and_remove_re): Properly handle case where a second
2532 eliminated extension requires widening a copy created for elimination
2533 of a prior extension.
2534 (combine_set_extension): Ensure that the number of hard regs needed
2535 for a destination register does not change when we widen it.
2537 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2539 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2540 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2541 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2542 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2543 (avr-*-rtems*): Likewise.
2544 (bfin*-rtems*): Likewise.
2545 (moxie-*-rtems*): Likewise.
2546 (h8300-*-rtems*): Likewise.
2547 (i[34567]86-*-rtems*): Likewise.
2548 (lm32-*-rtems*): Likewise.
2549 (m32r-*-rtems*): Likewise.
2550 (m68k-*-rtems*): Likewise.
2551 (microblaze*-*-rtems*): Likewise.
2552 (mips*-*-rtems*): Likewise.
2553 (powerpc-*-rtems*): Likewise.
2554 (sh-*-rtems*): Likewise.
2555 (sparc-*-rtems*): Likewise.
2556 (sparc64-*-rtems*): Likewise.
2557 (v850-*-rtems*): Likewise.
2558 (m32c-*-rtems*): Likewise.
2560 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2562 PR rtl-optimization/59511
2563 * ira.c (ira_init_register_move_cost): Use memory costs for some
2564 cases of register move cost calculations.
2565 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2566 instead of BB frequency.
2567 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2568 * lra-assigns.c (find_hard_regno_for): Ditto.
2570 2014-01-15 Richard Biener <rguenther@suse.de>
2572 PR tree-optimization/59822
2573 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2574 (vectorizable_load): Use it to hoist defs of uses of invariant
2575 loads out of the loop.
2577 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2578 Kugan Vivekanandarajah <kuganv@linaro.org>
2581 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2584 2014-01-15 Richard Biener <rguenther@suse.de>
2586 PR rtl-optimization/59802
2587 * lcm.c (compute_available): Use inverted postorder to seed
2588 the initial worklist.
2590 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2593 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2594 ADDR_REGS for invalid symrefs in non-PIC code.
2596 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2599 * builtins.c (determine_block_size): Initialize *probable_max_size
2600 even if len_rtx is CONST_INT.
2602 2014-01-14 Andrew Pinski <apinski@cavium.com>
2604 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2605 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2606 (cortexa53_tunings): Likewise.
2607 (aarch64_sched_issue_rate): New function.
2608 (TARGET_SCHED_ISSUE_RATE): Define.
2610 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2612 * ira-costs.c (find_costs_and_classes): Add missed
2613 ira_init_register_move_cost_if_necessary.
2615 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2618 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2620 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2623 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2624 to indicate if type is used for function return value. Warn ABI
2625 change if the vector mode isn't available for function return value.
2626 (ix86_function_arg_advance): Pass false to type_natural_mode.
2627 (ix86_function_arg): Likewise.
2628 (ix86_gimplify_va_arg): Likewise.
2629 (function_arg_32): Don't warn ABI change.
2630 (ix86_function_value): Pass true to type_natural_mode.
2631 (ix86_return_in_memory): Likewise.
2632 (ix86_struct_value_rtx): Removed.
2633 (TARGET_STRUCT_VALUE_RTX): Likewise.
2635 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2637 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2638 converting a conditional jump into a conditional return.
2640 2014-01-14 Richard Biener <rguenther@suse.de>
2642 PR tree-optimization/58921
2643 PR tree-optimization/59006
2644 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2645 hoisting invariant stmts.
2646 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2647 invariant loads on the preheader edge if possible.
2649 2014-01-14 Joey Ye <joey.ye@arm.com>
2651 * doc/plugin.texi (Building GCC plugins): Update to C++.
2653 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2655 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2656 (_mm_rcp28_round_ss): Ditto.
2657 (_mm_rsqrt28_round_sd): Ditto.
2658 (_mm_rsqrt28_round_ss): Ditto.
2659 (_mm_rcp28_sd): Ditto.
2660 (_mm_rcp28_ss): Ditto.
2661 (_mm_rsqrt28_sd): Ditto.
2662 (_mm_rsqrt28_ss): Ditto.
2663 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2664 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2665 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2666 (IX86_BUILTIN_RCP28SD): Ditto.
2667 (IX86_BUILTIN_RCP28SS): Ditto.
2668 (IX86_BUILTIN_RSQRT28SD): Ditto.
2669 (IX86_BUILTIN_RSQRT28SS): Ditto.
2670 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2671 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2672 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2673 (ix86_expand_special_args_builtin): Expand new FTYPE.
2674 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2675 (srcp14<mode>): Make insn unary.
2676 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2677 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2678 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2679 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2680 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2681 Fix rounding: make it SAE only.
2682 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2684 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2686 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2687 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2688 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2689 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2690 (round_saeonly_mask_scalar_operand4): Ditto.
2691 (round_saeonly_mask_scalar_op3): Ditto.
2692 (round_saeonly_mask_scalar_op4): Ditto.
2694 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2696 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2697 Implement -maltivec=be for vec_insert and vec_extract.
2699 2014-01-10 DJ Delorie <dj@redhat.com>
2701 * config/msp430/msp430.md (call_internal): Don't allow memory
2702 references with SP as the base register.
2703 (call_value_internal): Likewise.
2704 * config/msp430/constraints.md (Yc): New. For memory references
2705 that don't use SP as a base register.
2707 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2708 "an integer without a # prefix"
2709 * config/msp430/msp430.md (epilogue_helper): Use it.
2711 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2714 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2715 AVX512F gather builtins.
2716 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2717 on gather decls with INTEGER_TYPE masktype.
2718 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2719 directly into the builtin rather than hoisting it before loop.
2721 PR tree-optimization/59387
2722 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2723 (scev_const_prop): If folded_casts and type has undefined overflow,
2724 use force_gimple_operand instead of force_gimple_operand_gsi and
2725 for each added stmt if it is assign with
2726 arith_code_with_undefined_signed_overflow, call
2727 rewrite_to_defined_overflow.
2728 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2729 gimple-fold.h instead.
2730 (arith_code_with_undefined_signed_overflow,
2731 rewrite_to_defined_overflow): Moved to ...
2732 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2733 rewrite_to_defined_overflow): ... here. No longer static.
2734 Include gimplify-me.h.
2735 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2736 rewrite_to_defined_overflow): New prototypes.
2738 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2740 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2742 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2744 * builtins.c (get_object_alignment_2): Minor tweak.
2745 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2747 2014-01-13 Christian Bruel <christian.bruel@st.com>
2749 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2750 optimized non constant lengths.
2752 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2755 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2756 load as __atomic_load_N if possible.
2758 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2760 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2762 (rs6000_expand_builtin): Adjust call.
2764 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2767 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2768 * config/rs6000/rs6000.c: Include target-globals.h.
2769 (rs6000_set_current_function): Instead of doing target_reinit
2770 unconditionally, use save_target_globals_default_opts and
2771 restore_target_globals.
2773 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2775 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2776 (rs6000_expand_builtin): Handle mffs and mtfsf.
2777 (rs6000_init_builtins): Define mffs and mtfsf.
2778 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2779 (rs6000_mffs): New pattern.
2780 (rs6000_mtfsf): New pattern.
2782 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2784 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2785 Start narrowing with START. Apply candidate-use pair
2786 and check overall cost in narrowing.
2787 (iv_ca_prune): Pass new argument.
2789 2014-01-10 Jeff Law <law@redhat.com>
2792 * ree.c (combine_reaching_defs): Ensure the defining statement
2793 occurs before the extension when optimizing extensions with
2794 different source and destination hard registers.
2796 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2799 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2800 vtables into the type inheritance graph.
2802 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2804 PR rtl-optimization/59754
2805 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2806 modes in the REGNO != REGNO case.
2808 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2810 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2812 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2814 PR tree-optimization/59745
2815 * tree-predcom.c (tree_predictive_commoning_loop): Call
2816 free_affine_expand_cache if giving up because components is NULL.
2818 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2819 GC in payload of target_globals struct instead of allocating them on
2820 the heap and the larger structs separately using GC.
2821 * target-globals.h (struct target_globals): Make regs, hard_regs,
2822 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2823 of GTY((skip)) and change type to void *.
2824 (reset_target_globals): Cast loads from those fields to corresponding
2827 2014-01-10 Steve Ellcey <sellcey@mips.com>
2830 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2831 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2832 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2834 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2837 * aarch64-modes.def (CC_Zmode): New flags mode.
2838 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2839 represents an equality.
2840 (aarch64_get_condition_code): Handle CC_Zmode.
2841 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2843 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2845 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2846 extraction in good case.
2848 2014-01-10 Richard Biener <rguenther@suse.de>
2850 PR tree-optimization/59374
2851 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2852 checking after SLP discovery. Mark stmts not participating
2853 in any SLP instance properly.
2855 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2857 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2858 when handling a SET rtx.
2860 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2862 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2863 (cortex-a57): Likewise.
2864 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2866 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2868 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2869 non-iwmmxt builtins.
2871 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2875 * ipa-devirt.c record_target_from_binfo): Take as argument
2876 stack of binfos and lookup matching one for virtual inheritance.
2877 (possible_polymorphic_call_targets_1): Update.
2879 2014-01-10 Huacai Chen <chenhc@lemote.com>
2881 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2882 kernel strings for Loongson-2E/2F/3A.
2884 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2887 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2888 is_gimple_call before calling gimple_call_internal_p.
2890 2014-01-09 Steve Ellcey <sellcey@mips.com>
2892 * Makefile.in (TREE_FLOW_H): Remove.
2893 (TREE_SSA_H): Add file names from tree-flow.h.
2894 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2895 * tree.h: Remove tree-flow.h reference.
2896 * hash-table.h: Remove tree-flow.h reference.
2897 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2898 reference with tree-ssa-loop.h.
2900 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2902 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2903 default element-order behavior for -maltivec.
2904 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2905 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2906 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2907 when targeting big endian, at least for now.
2908 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2910 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2913 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2914 var satisfies use_register_for_decl, just take into account type
2915 alignment, rather than decl alignment.
2917 PR tree-optimization/59622
2918 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2919 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2920 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2921 Don't devirtualize for inplace at all. For targets.length () == 1,
2922 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2924 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2926 * config/i386/i386.md (cpu): Remove the unused btver1.
2928 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2930 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2932 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2935 * tree-core.h (struct target_globals): New forward declaration.
2936 (struct tree_target_option): Add globals field.
2937 * tree.h (TREE_TARGET_GLOBALS): Define.
2938 (prepare_target_option_nodes_for_pch): New prototype.
2939 * target-globals.h (struct target_globals): Define even if
2941 * tree.c (prepare_target_option_node_for_pch,
2942 prepare_target_option_nodes_for_pch): New functions.
2943 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2944 * config/i386/i386.c: Include target-globals.h.
2945 (ix86_set_current_function): Instead of doing target_reinit
2946 unconditionally, use save_target_globals_default_opts and
2947 restore_target_globals.
2949 2014-01-09 Richard Biener <rguenther@suse.de>
2951 PR tree-optimization/59715
2952 * tree-cfg.h (split_critical_edges): Declare.
2953 * tree-cfg.c (split_critical_edges): Export.
2954 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2956 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2958 * cfgexpand.c (expand_stack_vars): Optionally disable
2959 asan stack protection.
2960 (expand_used_vars): Likewise.
2961 (partition_stack_vars): Likewise.
2962 * asan.c (asan_emit_stack_protection): Optionally disable
2963 after return stack usage.
2964 (instrument_derefs): Optionally disable memory access instrumentation.
2965 (instrument_builtin_call): Likewise.
2966 (instrument_strlen_call): Likewise.
2967 (asan_protect_global): Optionally disable global variables protection.
2968 * doc/invoke.texi: Added doc for new options.
2969 * params.def: Added new options.
2970 * params.h: Likewise.
2972 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2974 PR rtl-optimization/59724
2975 * ifcvt.c (cond_exec_process_if_block): Don't call
2976 flow_find_head_matching_sequence with 0 longest_match.
2977 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2978 non-active insns if !stop_after.
2979 (try_head_merge_bb): Revert 2014-01-07 changes.
2981 2014-01-08 Jeff Law <law@redhat.com>
2983 * ree.c (get_sub_rtx): New function, extracted from...
2984 (merge_def_and_ext): Here.
2985 (combine_reaching_defs): Use get_sub_rtx.
2987 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2989 * cgraph.h (varpool_variable_node): Do not choke on null node.
2991 2014-01-08 Catherine Moore <clm@codesourcery.com>
2993 * config/mips/mips.md (simple_return): Attempt to use JRC
2995 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2997 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2999 PR rtl-optimization/59137
3000 * reorg.c (steal_delay_list_from_target): Call update_block for
3002 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3004 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3006 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3007 two duplicate entries.
3009 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3012 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3014 * config/mips/mips.c (mips_truncated_op_cost): New function.
3015 (mips_rtx_costs): Adjust test for BADDU.
3016 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3018 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3020 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3021 (*baddu_si): ...this new pattern.
3023 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3026 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3028 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3031 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3033 (expand_expr, expand_normal): Adjust.
3034 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3035 inner_reference_p. Use inner_reference_p to expand inner references.
3036 (store_expr): Adjust.
3037 * cfgexpand.c (expand_call_stmt): Adjust.
3039 2014-01-08 Rong Xu <xur@google.com>
3041 * gcov-io.c (gcov_var): Move from gcov-io.h.
3042 (gcov_position): Ditto.
3043 (gcov_is_error): Ditto.
3044 (gcov_rewrite): Ditto.
3045 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3046 only part to libgcc/libgcov.h.
3048 2014-01-08 Marek Polacek <polacek@redhat.com>
3051 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3053 2014-01-08 Marek Polacek <polacek@redhat.com>
3056 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3058 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3060 PR rtl-optimization/59649
3061 * stor-layout.c (get_mode_bounds): For BImode return
3062 0 and STORE_FLAG_VALUE.
3064 2014-01-08 Richard Biener <rguenther@suse.de>
3067 * gimple.h (is_gimple_builtin_call): Remove.
3068 (gimple_builtin_call_types_compatible_p): New.
3069 (gimple_call_builtin_p): New overload.
3070 * gimple.c (is_gimple_builtin_call): Remove.
3071 (validate_call): Rename to ...
3072 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3074 (validate_type): New static function.
3075 (gimple_call_builtin_p): New overload and adjust.
3076 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3077 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3078 (gimple_fold_stmt_to_constant_1): Likewise.
3079 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3081 2014-01-08 Richard Biener <rguenther@suse.de>
3084 * gimplify.c (gimplify_expr): Gimplify register-register type
3085 VIEW_CONVERT_EXPRs to separate stmts.
3087 2014-01-07 Jeff Law <law@redhat.com>
3090 * ree.c (combine_set_extension): Handle case where source
3091 and destination registers in an extension insn are different.
3092 (combine_reaching_defs): Allow source and destination registers
3093 in extension to be different under limited circumstances.
3094 (add_removable_extension): Remove restriction that the
3095 source and destination registers in the extension are the same.
3096 (find_and_remove_re): Emit a copy from the extension's
3097 destination to its source after the defining insn if
3098 the source and destination registers are different.
3101 * ifcvt.c (merge_if_block): If we are merging a block with more than
3102 one successor with a block with no successors, remove any BARRIER
3103 after the second block.
3105 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3107 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3109 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3112 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3113 for 14-bit register offsets when INT14_OK_STRICT is false.
3115 2014-01-07 Roland Stigge <stigge@antcom.de>
3116 Michael Meissner <meissner@linux.vnet.ibm.com>
3119 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3120 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3122 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3124 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3127 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3129 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3130 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3131 rtx is const0_rtx or not.
3133 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3136 * target-globals.c (save_target_globals): Remove this_fn_optab
3138 * toplev.c: Include optabs.h.
3139 (target_reinit): Temporarily restore the global options if another
3140 set of options are in force.
3142 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3144 PR rtl-optimization/58668
3145 * cfgcleanup.c (flow_find_cross_jump): Don't count
3146 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3147 to determine what is counted.
3148 (flow_find_head_matching_sequence): Use active_insn_p to determine
3150 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3152 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3153 determine what is counted.
3155 PR tree-optimization/59643
3156 * tree-predcom.c (split_data_refs_to_components): If one dr is
3157 read and one write, determine_offset fails and the write isn't
3158 in the bad component, just put the read into the bad component.
3160 2014-01-07 Mike Stump <mikestump@comcast.net>
3161 Jakub Jelinek <jakub@redhat.com>
3164 * tree-core.h (struct tree_optimization_option): Change optabs
3165 type from unsigned char * to void *.
3166 * optabs.c (init_tree_optimization_optabs): Adjust
3167 TREE_OPTIMIZATION_OPTABS initialization.
3169 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3172 * config/i386/i386.h (struct machine_function): Add
3173 no_drap_save_restore field.
3174 * config/i386/i386.c (ix86_save_reg): Use
3175 !cfun->machine->no_drap_save_restore instead of
3176 crtl->stack_realign_needed.
3177 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3178 this function clears frame_pointer_needed. Set
3179 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3180 and DRAP reg is needed.
3182 2014-01-06 Marek Polacek <polacek@redhat.com>
3185 * doc/implement-c.texi: Mention that other integer types are
3186 permitted as bit-field types in strictly conforming mode.
3188 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3190 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3193 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3195 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3197 2014-01-06 Martin Jambor <mjambor@suse.cz>
3200 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3202 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3204 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3208 * var-tracking.c (add_stores): Preserve the value of the source even if
3209 we don't record the store.
3211 2014-01-06 Terry Guo <terry.guo@arm.com>
3213 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3215 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3218 * config/darwin.c (darwin_function_section): Adjust return values to
3219 correspond to optimisation changes made in r206070.
3221 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3223 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3224 from prefetch_block tune setting.
3225 (nocona_cost): Correct size of prefetch block to 64.
3227 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3229 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3230 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3231 used to save the static chain register in the computation of the offset
3232 from which the FP registers need to be restored.
3234 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3236 PR tree-optimization/59519
3237 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3238 ICE if get_current_def (current_new_name) is already non-NULL, as long
3239 as it is a phi result of some other phi in *new_exit_bb that has
3242 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3243 or vmovdqu* for misaligned_operand.
3244 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3245 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3246 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3247 aligned_mem for AVX512F masked aligned load and store builtins and for
3250 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3252 PR tree-optimization/59651
3253 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3254 Address range for negative step should be added by TYPE_SIZE_UNIT.
3256 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3258 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3259 overlapping registers also for registers other than the stack pointer.
3261 2014-01-03 Marek Polacek <polacek@redhat.com>
3264 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3267 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3270 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3273 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3274 (push splitter): Use <P:MODE_SIZE> instead of
3275 GET_MODE_SIZE (<P:MODE>mode).
3276 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3277 (mov -1, reg peephole2): Likewise.
3278 * config/i386/sse.md (*mov<mode>_internal,
3279 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3280 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3281 *<code><mode>3, *andnot<mode>3<mask_name>,
3282 <mask_codefor><code><mode>3<mask_name>): Likewise.
3283 * config/i386/subst.md (mask_mode512bit_condition,
3284 sd_mask_mode512bit_condition): Likewise.
3286 2014-01-02 Xinliang David Li <davidxl@google.com>
3288 PR tree-optimization/59303
3289 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3290 (dump_predicates): Better output format.
3291 (pred_equal_p): New function.
3292 (is_neq_relop_p): Ditto.
3293 (is_neq_zero_form_p): Ditto.
3294 (pred_expr_equal_p): Ditto.
3295 (pred_neg_p): Ditto.
3296 (simplify_pred): Ditto.
3297 (simplify_preds_2): Ditto.
3298 (simplify_preds_3): Ditto.
3299 (simplify_preds_4): Ditto.
3300 (simplify_preds): Ditto.
3302 (push_to_worklist): Ditto.
3303 (get_pred_info_from_cmp): Ditto.
3304 (is_degenerated_phi): Ditto.
3305 (normalize_one_pred_1): Ditto.
3306 (normalize_one_pred): Ditto.
3307 (normalize_one_pred_chain): Ditto.
3308 (normalize_preds): Ditto.
3309 (normalize_cond_1): Remove function.
3310 (normalize_cond): Ditto.
3311 (is_gcond_subset_of): Ditto.
3312 (is_subset_of_any): Ditto.
3313 (is_or_set_subset_of): Ditto.
3314 (is_and_set_subset_of): Ditto.
3315 (is_norm_cond_subset_of): Ditto.
3316 (pred_chain_length_cmp): Ditto.
3317 (convert_control_dep_chain_into_preds): Type change.
3318 (find_predicates): Ditto.
3319 (find_def_preds): Ditto.
3320 (destroy_predicates_vecs): Ditto.
3321 (find_matching_predicates_in_rest_chains): Ditto.
3322 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3323 (is_pred_expr_subset): Ditto.
3324 (is_pred_chain_subset_of): Ditto.
3325 (is_included_in): Ditto.
3326 (is_superset_of): Ditto.
3328 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3330 Update copyright years.
3332 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3334 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3335 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3336 config/arc/arc.md, config/arc/arc.opt,
3337 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3338 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3339 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3340 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3341 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3342 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3343 vtable-verify.c, vtable-verify.h: Use the standard form for the
3346 2014-01-02 Tobias Burnus <burnus@net-b.de>
3348 * gcc.c (process_command): Update copyright notice dates.
3349 * gcov-dump.c: Ditto.
3351 * doc/cpp.texi: Bump @copying's copyright year.
3352 * doc/cppinternals.texi: Ditto.
3353 * doc/gcc.texi: Ditto.
3354 * doc/gccint.texi: Ditto.
3355 * doc/gcov.texi: Ditto.
3356 * doc/install.texi: Ditto.
3357 * doc/invoke.texi: Ditto.
3359 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3361 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3363 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3365 * config/i386/sse.md (*mov<mode>_internal): Guard
3366 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3368 PR rtl-optimization/59647
3369 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3370 new_rtx into UNSIGNED_FLOAT rtxes.
3372 Copyright (C) 2014 Free Software Foundation, Inc.
3374 Copying and distribution of this file, with or without modification,
3375 are permitted in any medium without royalty provided the copyright
3376 notice and this notice are preserved.