1 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3 PR rtl-optimization/60116
4 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
5 other_insn once the combination has been validated.
7 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
10 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
12 * ipa-devirt.c: Include demangle.h
13 (odr_violation_reported): New static variable.
14 (add_type_duplicate): Update odr_violations.
15 (maybe_record_node): Add completep parameter; update it.
16 (record_target_from_binfo): Add COMPLETEP parameter;
18 (possible_polymorphic_call_targets_1): Likewise.
19 (struct polymorphic_call_target_d): Add nonconstruction_targets;
20 rename FINAL to COMPLETE.
21 (record_targets_from_bases): Sanity check we found the binfo;
22 fix COMPLETEP updating.
23 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
24 parameter, fix computing of COMPLETEP.
25 (dump_possible_polymorphic_call_targets): Imrove readability of dump; at
26 LTO time do demangling.
27 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
28 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
30 (gimple_get_virt_method_for_binfo): Likewise.
31 * gimple-fold.h (gimple_get_virt_method_for_binfo,
32 gimple_get_virt_method_for_vtable): Update prototypes.
34 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
37 * genautomata.c (add_presence_absence): Fix typo with
38 {final_}presence_list.
40 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
43 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
44 for VSX/Altivec vectors that land in GPR registers.
46 2014-02-11 Richard Henderson <rth@redhat.com>
47 Jakub Jelinek <jakub@redhat.com>
50 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
51 around drhs if type conversion to lacc->type is not useless.
53 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
55 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
57 (cortex-a57.cortex-a53): Likewise.
58 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
60 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
62 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
65 2014-02-11 Renlin Li <Renlin.Li@arm.com>
67 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
68 add_options_for_arm_vfp3.
70 2014-02-11 Jeff Law <law@redhat.com>
73 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
74 object with an undesirable mode.
76 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
79 * config/i386/sol2-9.h: New file.
80 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
81 *-*-solaris2.9*): Use it.
83 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
85 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
86 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
88 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
90 * config/microblaze/microblaze.c: Extend mcpu version format
92 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
94 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
96 2014-02-10 Richard Henderson <rth@redhat.com>
99 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
100 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
101 ms-abi vs -mno-accumulate-outgoing-args.
102 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
103 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
106 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
109 * cfgexpand.c (expand_asm_operands): Attach source location to
110 ASM_INPUT rtx objects.
111 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
113 2014-02-10 Nick Clifton <nickc@redhat.com>
115 * config/mn10300/mn10300.c (popcount): New function.
116 (mn10300_expand_prologue): Include saved registers in stack usage
119 2014-02-10 Jeff Law <law@redhat.com>
122 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
123 when changing the SET_DEST of a prior insn to avoid an input
126 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
128 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
129 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
130 -mcall-openbsd, or -mcall-linux.
131 (CC1_ENDIAN_BIG_SPEC): Remove.
132 (CC1_ENDIAN_LITTLE_SPEC): Remove.
133 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
134 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
135 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
136 and %cc1_endian_default.
137 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
139 2014-02-10 Richard Biener <rguenther@suse.de>
141 PR tree-optimization/60115
142 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
143 MEM_REF handling. Properly verify that the accesses are not
144 out of the objects bound.
146 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
148 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
151 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
153 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
154 proper constants and fix formatting.
155 (possible_polymorphic_call_targets): Fix formatting.
157 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
158 Ilya Tocar <ilya.tocar@intel.com>
160 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
161 (_mm512_loadu_epi32): Renamed into...
162 (_mm512_loadu_si512): This.
163 (_mm512_storeu_epi32): Renamed into...
164 (_mm512_storeu_si512): This.
165 (_mm512_maskz_ceil_ps): Removed.
166 (_mm512_maskz_ceil_pd): Ditto.
167 (_mm512_maskz_floor_ps): Ditto.
168 (_mm512_maskz_floor_pd): Ditto.
169 (_mm512_floor_round_ps): Ditto.
170 (_mm512_floor_round_pd): Ditto.
171 (_mm512_ceil_round_ps): Ditto.
172 (_mm512_ceil_round_pd): Ditto.
173 (_mm512_mask_floor_round_ps): Ditto.
174 (_mm512_mask_floor_round_pd): Ditto.
175 (_mm512_mask_ceil_round_ps): Ditto.
176 (_mm512_mask_ceil_round_pd): Ditto.
177 (_mm512_maskz_floor_round_ps): Ditto.
178 (_mm512_maskz_floor_round_pd): Ditto.
179 (_mm512_maskz_ceil_round_ps): Ditto.
180 (_mm512_maskz_ceil_round_pd): Ditto.
181 (_mm512_expand_pd): Ditto.
182 (_mm512_expand_ps): Ditto.
183 * config/i386/i386.c (ix86_builtins): Remove
184 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
186 * config/i386/predicates.md (const1256_operand): New.
187 (const_1_to_2_operand): Ditto.
188 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
189 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
190 (*avx512pf_gatherpf<mode>sf): Ditto.
191 (avx512pf_gatherpf<mode>df): Ditto.
192 (*avx512pf_gatherpf<mode>df_mask): Ditto.
193 (*avx512pf_gatherpf<mode>df): Ditto.
194 (avx512pf_scatterpf<mode>sf): Ditto.
195 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
196 (*avx512pf_scatterpf<mode>sf): Ditto.
197 (avx512pf_scatterpf<mode>df): Ditto.
198 (*avx512pf_scatterpf<mode>df_mask): Ditto.
199 (*avx512pf_scatterpf<mode>df): Ditto.
200 (avx512f_expand<mode>): Removed.
201 (<shift_insn><mode>3<mask_name>): Change predicate type.
203 2014-02-08 Jakub Jelinek <jakub@redhat.com>
205 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
206 not at the end of datarefs vector use ordered_remove to avoid
207 reordering datarefs vector.
210 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
211 mark local addressable non-static vars as GOVD_PRIVATE
212 instead of GOVD_LOCAL.
213 * omp-low.c (lower_omp_for): Move gimple_bind_vars
214 and BLOCK_VARS of gimple_bind_block to new_stmt rather
218 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
219 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
220 assume_aligned or alloc_align attributes.
221 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
222 arguments. Handle also assume_aligned and alloc_align attributes.
223 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
224 Handle calls to functions with assume_aligned or alloc_align
226 * doc/extend.texi: Document assume_aligned and alloc_align
229 2014-02-08 Terry Guo <terry.guo@arm.com>
231 * doc/invoke.texi: Document ARM -march=armv7e-m.
233 2014-02-08 Jakub Jelinek <jakub@redhat.com>
235 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
236 flag on __cilkrts_rethrow builtin.
239 * ipa-cp.c (determine_versionability): Fail at -O0
240 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
241 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
244 2014-02-04 Jakub Jelinek <jakub@redhat.com>
247 * tree-inline.c (copy_forbidden): Fail for
248 __attribute__((optimize (0))) functions.
250 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
252 * varpool.c: Include pointer-set.h.
253 (varpool_remove_unreferenced_decls): Variables in other partitions
254 will not be output; be however careful to not lose information
257 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
259 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
260 lookup in the vtable constructor.
262 2014-02-07 Jeff Law <law@redhat.com>
265 * config/m68k/m68k.md (ashldi_extsi): Turn into a
266 define_insn_and_split.
268 * ipa-inline.c (inline_small_functions): Fix typos.
270 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
272 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
273 (s390_can_use_return_insn): Declare.
274 * config/s390/s390.h (EPILOGUE_USES): Define.
275 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
277 (s390_chunkify_start): Handle return JUMP_LABELs.
278 (s390_early_mach): Emit a main_pool instruction on the entry edge.
279 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
280 (s390_can_use_return_insn): New functions.
281 (s390_fix_long_loop_prediction): Handle conditional returns.
282 (TARGET_SET_UP_BY_PROLOGUE): Define.
283 * config/s390/s390.md (ANY_RETURN): New code iterator.
284 (*creturn, *csimple_return, return, simple_return): New patterns.
286 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
288 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
289 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
290 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
291 REG_CFA_RESTORE list when deciding not to restore a register.
293 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
295 * config/s390/s390.c: Include tree-pass.h and context.h.
296 (s390_early_mach): New function, split out from...
297 (s390_emit_prologue): ...here.
298 (pass_data_s390_early_mach): New pass structure.
299 (pass_s390_early_mach): New class.
300 (s390_option_override): Create and register early_mach pass.
303 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
305 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
306 to match for the exit block.
308 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
310 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
311 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
312 Reject misaligned operands.
314 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
316 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
319 2014-02-07 Richard Biener <rguenther@suse.de>
322 * gimple-low.c (lower_builtin_posix_memalign): New function.
323 (lower_stmt): Call it to lower posix_memalign in a way
324 to make alignment info accessible.
326 2014-02-07 Jakub Jelinek <jakub@redhat.com>
329 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
330 __builtin_setjmp_receiver.
332 2014-02-07 Richard Biener <rguenther@suse.de>
335 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
336 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
337 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
338 Handle BUILT_IN_POSIX_MEMALIGN.
339 (find_func_clobbers): Likewise.
340 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
341 (call_may_clobber_ref_p_1): Likewise.
343 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
346 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
348 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
351 * lto-cgraph.c (lto_output_node): Use
352 symtab_get_symbol_partitioning_class.
353 (lto_output_varpool_node): likewise.
354 (symtab_get_symbol_partitioning_class): Move here from
356 * cgraph.h (symbol_partitioning_class): Likewise.
357 (symtab_get_symbol_partitioning_class): Declare.
359 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
361 * ggc.h (ggc_internal_cleared_alloc): New macro.
362 * vec.h (vec_safe_copy): Handle memory stats.
363 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
364 * target-globals.c (save_target_globals): Likewise.
366 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
369 * expr.c (emit_move_resolve_push): Export; be bit more selective
370 on when to clear alias set.
371 * expr.h (emit_move_resolve_push): Declare.
372 * function.h (struct function): Add tail_call_marked.
373 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
374 * config/i386/i386-protos.h (ix86_expand_push): Remove.
375 * config/i386/i386.md (TImode move expander): De not call
377 (FP push expanders): Preserve memory attributes.
378 * config/i386/sse.md (push<mode>1): Remove.
379 * config/i386/i386.c (ix86_expand_vector_move): Handle push
381 (ix86_expand_push): Remove.
382 * config/i386/mmx.md (push<mode>1): Remove.
384 2014-02-06 Jakub Jelinek <jakub@redhat.com>
386 PR rtl-optimization/60030
387 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
388 lopart with paradoxical subreg before shifting it up by hprec.
390 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
392 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
393 Remove extra newline at end of file.
394 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
395 (arm_issue_rate): Handle cortexa57.
396 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
397 (cortex-a57.cortex-a53): Likewise.
399 2014-02-06 Jakub Jelinek <jakub@redhat.com>
402 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
403 don't record in REG_FRAME_RELATED_EXPR registers not set in that
405 (arm_expand_prologue): Adjust all callers.
406 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
407 info, registers also at the lowest numbered registers side. Use
408 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
412 * var-tracking.c (adjust_mems): Before adding a SET
413 to amd->side_effects, adjust it's SET_SRC using
414 simplify_replace_fn_rtx.
416 2014-02-06 Alan Modra <amodra@gmail.com>
419 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
420 change SDmode to DDmode when lra_in_progress.
422 2014-02-06 Jakub Jelinek <jakub@redhat.com>
425 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
426 free_data_ref on the dr first, and before goto again also set dr
427 to the next dr. For simd_lane_access, free old datarefs[i] before
428 overwriting it. For get_vectype_for_scalar_type failure, don't
429 free_data_ref if simd_lane_access.
431 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
434 * tree.h (opts_for_fn): New inline function.
435 (opt_for_fn): Define.
436 * config/i386/i386.c (ix86_function_regparm): Use
437 opt_for_fn (decl, optimize) instead of optimize.
439 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
441 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
442 for SYMBOL_REF in large memory model.
444 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
446 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
448 (cortex-a57): Likewise.
449 (cortex-a57.cortex-a53): Likewise.
451 2014-02-06 Yury Gribov <y.gribov@samsung.com>
452 Kugan Vivekanandarajah <kuganv@linaro.org>
454 * config/arm/arm.c (arm_vector_alignment_reachable): Check
456 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
458 2014-02-06 Richard Biener <rguenther@suse.de>
460 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
461 set_loop_copy and initialize_original_copy_tables.
463 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
465 * config/aarch64/aarch64-simd.md
466 (aarch64_ashr_simddi): Change QI to SI.
468 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
469 Jakub Jelinek <jakub@redhat.com>
472 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
475 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
477 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
478 CODE_FOR_altivec_vpku[hw]um to
479 CODE_FOR_altivec_vpku[hw]um_direct.
480 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
481 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
482 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
483 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
485 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
487 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
488 generation for -maltivec=be.
489 (altivec_vsumsws): Simplify redundant test.
491 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
493 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
494 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
495 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
496 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
498 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
500 (altivec_vpks<VI_char>ss): Likewise.
501 (altivec_vpks<VI_char>us): Likewise.
502 (altivec_vpku<VI_char>us): Likewise.
503 (altivec_vpku<VI_char>um): Likewise.
504 (altivec_vpku<VI_char>um_direct): New (copy of
505 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
507 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
508 target is little endian and -maltivec=be is not specified.
509 (*altivec_vupkhs<VU_char>_direct): New (copy of
510 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
512 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
513 target is little endian and -maltivec=be is not specified.
514 (*altivec_vupkls<VU_char>_direct): New (copy of
515 altivec_vupkls<VU_char> that always emits vupkls*, for internal
517 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
518 little endian and -maltivec=be is not specified.
519 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
520 little endian and -maltivec=be is not specified.
522 2014-02-05 Richard Henderson <rth@redhat.com>
525 * combine-stack-adj.c: Revert r206943.
526 * sched-int.h (struct deps_desc): Add last_args_size.
527 * sched-deps.c (init_deps): Initialize it.
528 (sched_analyze_insn): Add OUTPUT dependencies between insns that
529 contain REG_ARGS_SIZE notes.
531 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
533 * lto-cgraph.c (asm_nodes_output): Make global.
534 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
535 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
537 (driver_handle_option): Handle OPT_fwpa.
539 2014-02-05 Jakub Jelinek <jakub@redhat.com>
542 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
543 a comment typo and formatting issue. If odr_hash hasn't been
544 created, return vNULL and set *completep to false.
547 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
548 bb with no successors.
550 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
553 * doc/invoke.texi (-march): Clarify documentation for ARM.
557 2014-02-05 Richard Biener <rguenther@suse.de>
559 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
560 when not vectorizing because of too many alias checks.
561 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
562 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
564 2014-02-05 Nick Clifton <nickc@redhat.com>
566 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
567 accept extended registers in any mode when compiling for the
570 2014-02-05 Yury Gribov <y.gribov@samsung.com>
572 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
573 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
574 sanitization attributes.
575 (can_inline_edge_p): Likewise.
576 (sanitize_attrs_match_for_inline_p): New function.
578 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
580 * ipa-prop.c (detect_type_change): Shor circuit testing of
581 type changes on THIS pointer.
583 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
586 * config/pa/pa.c (legitimize_tls_address): Return original address
587 if not passed a SYMBOL_REF rtx.
588 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
590 (pa_emit_move_sequence): Simplify TLS source operands.
591 (pa_legitimate_constant_p): Reject all TLS constants.
592 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
593 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
595 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
597 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
598 groups when we know they are controlled by LTO.
599 * varasm.c (default_binds_local_p_1): If object is in other partition,
600 it will be resolved locally.
602 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
604 * config/host-linux.c (linux_gt_pch_use_address): Don't
605 use SSIZE_MAX because it is not always defined.
607 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
610 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
611 threshold for pseudo splitting.
612 (update_ebb_live_info): Process call argument hard registers and
613 hard registers from insn definition too.
614 (max_small_class_regs_num): New constant.
615 (inherit_in_ebb): Update live hard regs through EBBs. Update
616 reloads_num only for small register classes. Don't split for
619 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
622 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
625 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
627 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
629 2014-02-04 Marek Polacek <polacek@redhat.com>
631 * gdbinit.in (pel): Define.
633 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
635 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
638 2014-02-04 Richard Biener <rguenther@suse.de>
641 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
642 in function context local.
643 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
644 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
645 similar to LTO_imported_decl_ref.
647 2014-02-04 Jakub Jelinek <jakub@redhat.com>
649 PR tree-optimization/60002
650 * cgraphclones.c (build_function_decl_skip_args): Clear
653 PR tree-optimization/60023
654 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
655 false to gsi_replace.
656 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
657 has been in some EH region and vec_stmt could throw, add
658 vec_stmt into the same EH region.
659 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
660 has no lhs, ignore it.
661 * internal-fn.c (expand_MASK_LOAD): Likewise.
664 * tree-inline.c (copy_forbidden): Fail for
665 __attribute__((optimize (0))) functions.
668 * omp-low.c (simd_clone_struct_copy): If from->inbranch
669 is set, copy one less argument.
670 (expand_simd_clones): Don't subtract clone_info->inbranch
671 from simd_clone_struct_alloc argument.
673 PR rtl-optimization/57915
674 * recog.c (simplify_while_replacing): If all unary/binary/relational
675 operation arguments are constant, attempt to simplify those.
678 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
679 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
681 2014-02-04 Richard Biener <rguenther@suse.de>
683 PR tree-optimization/60012
684 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
685 TBAA disambiguation to all DDRs.
687 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
690 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
691 (LINK_SPEC): Use it for -shared, -shared-libgcc.
693 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
696 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
698 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
700 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
701 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
703 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
706 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
707 to figure out targets of polymorphic calls with known decl.
708 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
709 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
710 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
711 (get_polymorphic_call_info): ... here.
712 (get_polymorphic_call_info_from_invariant): New function.
714 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
716 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
717 lookup via vtable pointer; check for type consistency
718 and turn inconsitent facts into UNREACHABLE.
719 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
720 * gimple-fold.c (gimple_get_virt_method_for_vtable):
721 Do not ICE on type inconsistent querries; return UNREACHABLE
724 2014-02-03 Richard Henderson <rth@twiddle.net>
727 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
728 already processed this node.
729 (normalize_one_pred_1): Pass along mark_set.
730 (normalize_one_pred): Create and destroy a pointer_set_t.
731 (normalize_one_pred_chain): Likewise.
733 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
735 PR gcov-profile/58602
736 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
738 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
741 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
742 Try to devirtualize by the knowledge of virtual table pointer given by
743 aggregate propagation.
744 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
745 ipa_print_node_jump_functions): Dump also offset that
746 is relevant for polymorphic calls.
747 (determine_known_aggregate_parts): Add arg_type parameter; use it
748 instead of determining the type from pointer type.
749 (ipa_compute_jump_functions_for_edge): Update call of
750 determine_known_aggregate_parts.
751 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
752 (gimple_get_virt_method_for_binfo): ... here; simplify using
753 vtable_pointer_value_to_vtable.
754 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
755 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
756 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
757 (vtable_pointer_value_to_vtable): Break out from ...; handle also
759 (vtable_pointer_value_to_binfo): ... here.
760 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
762 2014-02-03 Teresa Johnson <tejohnson@google.com>
764 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
765 redef of outer loop index variable.
767 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
771 * doc/extend.texi (Function Attributes): Typo.
773 2014-02-03 Cong Hou <congh@google.com>
775 PR tree-optimization/60000
776 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
777 if the vectorized statement is a store. A store statement can only
778 appear at the end of pattern statements.
780 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
782 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
783 (ix86_option_override_internal): Default long double to 64-bit for
784 32-bit Bionic and to 128-bit for 64-bit Bionic.
786 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
787 TARGET_LONG_DOUBLE_128 is true.
788 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
790 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
791 (mlong-double-64): Negate -mlong-double-128.
792 (mlong-double-128): New option.
794 * config/i386/i386-c.c (ix86_target_macros): Define
795 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
797 * doc/invoke.texi: Document -mlong-double-128.
799 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
801 PR rtl-optimization/60024
802 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
804 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
806 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
808 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
810 PR rtl-optimization/57662
811 * sel-sched.c (code_motion_path_driver): Do not mark already not
812 existing blocks in the visiting bitmap.
814 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
816 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
817 on the insn being emitted.
819 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
820 Will Deacon <will.deacon@arm.com>
822 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
824 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
826 * config/arm/arm-tables.opt: Regenerate.
828 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
830 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
831 for vector types other than V16QImode.
832 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
833 define_expand, and call altivec_expand_vec_perm_le when producing
834 code with little endian element order.
835 (*altivec_vperm_<mode>_internal): New insn having previous
836 behavior of altivec_vperm_<mode>.
837 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
838 altivec_expand_vec_perm_le when producing code with little endian
840 (*altivec_vperm_<mode>_uns_internal): New insn having previous
841 behavior of altivec_vperm_<mode>_uns.
843 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
845 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
846 (altivec_vsumsws): Add handling for -maltivec=be with a little
848 (altivec_vsumsws_direct): New.
849 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
852 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
854 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
855 vtable_pointer_value_to_binfo): New functions.
856 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
857 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
859 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
861 * config/nios2/nios2.md (load_got_register): Initialize GOT
862 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
863 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
865 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
867 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
868 preserverd by passthrough, do not propagate the type.
870 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
872 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
873 (mips_atomic_assign_expand_fenv): New function.
874 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
876 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
878 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
879 (__builtin_mips_set_fcsr): Likewise.
880 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
882 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
883 (mips16_expand_set_fcsr): Likewise.
884 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
885 (mips16_set_fcsr_stub): Likewise.
886 (mips16_get_fcsr_one_only_stub): New class.
887 (mips16_set_fcsr_one_only_stub): Likewise.
888 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
889 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
890 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
891 (hard_float): New availability predicate.
892 (mips_builtins): Add get_fcsr and set_fcsr.
893 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
894 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
895 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
896 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
897 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
900 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
902 * config/mips/mips.c (mips_one_only_stub): New class.
903 (mips_need_mips16_rdhwr_p): Replace with...
904 (mips16_rdhwr_stub): ...this new variable.
905 (mips16_stub_call_address): New function.
906 (mips16_rdhwr_one_only_stub): New class.
907 (mips_expand_thread_pointer): Use mips16_stub_call_address.
908 (mips_output_mips16_rdhwr): Delete.
909 (mips_finish_stub): New function.
910 (mips_code_end): Use it to handle rdhwr stubs.
912 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
915 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
916 when calculating size of integer atomic types.
918 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
920 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
922 2014-02-01 Jakub Jelinek <jakub@redhat.com>
924 PR tree-optimization/60003
925 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
926 * profile.c (branch_prob): Use gimple_call_builtin_p
927 to check for BUILT_IN_SETJMP_RECEIVER.
928 * tree-inline.c (copy_bb): Call notice_special_calls.
930 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
933 * lra-constraints.c (process_alt_operands): Update reload_sum only
936 2014-01-31 Richard Henderson <rth@redhat.com>
939 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
940 until after else_eh is processed.
942 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
944 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
945 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
946 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
947 in smmintrin.h, remove them.
948 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
949 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
950 * config/i386/i386.md (ROUND_SAE): Fix value.
951 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
952 (const48_operand): New.
953 * config/i386/subst.md (round), (round_expand): Use
954 const_4_or_8_to_11_operand.
955 (round_saeonly), (round_saeonly_expand): Use const48_operand.
957 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
959 * config/i386/constraints.md (Yk): Swap meaning with k.
960 * config/i386/i386.md (movhi_internal): Change Yk to k.
961 (movqi_internal): Ditto.
962 (*k<logic><mode>): Ditto.
965 (kandn<mode>): Ditto.
966 (*<code>hi_1): Ditto.
967 (*<code>qi_1): Ditto.
968 (kxnor<mode>): Ditto.
972 (*one_cmplhi2_1): Ditto.
973 (*one_cmplqi2_1): Ditto.
974 * config/i386/sse.md (): Change k to Yk.
975 (avx512f_load<mode>_mask): Ditto.
976 (avx512f_blendm<mode>): Ditto.
977 (avx512f_store<mode>_mask): Ditto.
978 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
979 (avx512f_storedqu<mode>_mask): Ditto.
980 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
982 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
983 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
984 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
985 (avx512f_maskcmp<mode>3): Ditto.
986 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
987 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
988 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
989 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
990 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
991 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
992 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
993 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
994 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
995 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
996 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
997 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
998 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
999 (vec_extract_lo_<mode>_maskm): Ditto.
1000 (vec_extract_hi_<mode>_maskm): Ditto.
1001 (avx512f_vternlog<mode>_mask): Ditto.
1002 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1003 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1004 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1005 (avx512f_<code>v8div16qi2_mask): Ditto.
1006 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1007 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1008 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1009 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1010 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1011 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1012 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1013 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1014 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1015 (avx512cd_maskb_vec_dupv8di): Ditto.
1016 (avx512cd_maskw_vec_dupv16si): Ditto.
1017 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1018 (avx512f_vpermi2var<mode>3_mask): Ditto.
1019 (avx512f_vpermi2var<mode>3_mask): Ditto.
1020 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1021 (*avx512f_gathersi<mode>): Ditto.
1022 (*avx512f_gathersi<mode>_2): Ditto.
1023 (*avx512f_gatherdi<mode>): Ditto.
1024 (*avx512f_gatherdi<mode>_2): Ditto.
1025 (*avx512f_scattersi<mode>): Ditto.
1026 (*avx512f_scatterdi<mode>): Ditto.
1027 (avx512f_compress<mode>_mask): Ditto.
1028 (avx512f_compressstore<mode>_mask): Ditto.
1029 (avx512f_expand<mode>_mask): Ditto.
1030 * config/i386/subst.md (mask): Change k to Yk.
1031 (mask_scalar_merge): Ditto.
1034 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1036 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1038 2014-01-31 Richard Biener <rguenther@suse.de>
1041 * builtins.c (fold_builtin_memory_op): Make sure to not
1042 use a floating-point mode or a boolean or enumeral type for
1045 2014-01-30 DJ Delorie <dj@redhat.com>
1047 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1048 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1049 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1050 whenever main() has an epilogue.
1052 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1054 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1055 unused variable "field".
1056 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1057 (vsx_mergeh_<mode>): Likewise.
1058 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1059 (altivec_vmrghh): Likewise.
1060 (altivec_vmrghw): Likewise.
1061 (altivec_vmrglb): Likewise.
1062 (altivec_vmrglh): Likewise.
1063 (altivec_vmrglw): Likewise.
1064 (altivec_vspltb): Add missing uses.
1065 (altivec_vsplth): Likewise.
1066 (altivec_vspltw): Likewise.
1067 (altivec_vspltsf): Likewise.
1069 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1072 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1073 frame related instructions.
1075 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1077 PR rtl-optimization/59959
1078 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1079 any reload of register whose subreg is invalid.
1081 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1083 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1084 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1085 Add missing return type - void.
1087 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1089 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1090 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1091 remove element index adjustment for endian (now handled in vsx.md
1093 (altivec_expand_vec_perm_const): Use
1094 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1095 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1096 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1097 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1098 define_expand and a new define_insn *altivec_vspltb_internal;
1099 adjust for -maltivec=be on a little endian target.
1100 (altivec_vspltb_direct): New.
1101 (altivec_vsplth): Divide into a define_expand and a new
1102 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1103 little endian target.
1104 (altivec_vsplth_direct): New.
1105 (altivec_vspltw): Divide into a define_expand and a new
1106 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1107 little endian target.
1108 (altivec_vspltw_direct): New.
1109 (altivec_vspltsf): Divide into a define_expand and a new
1110 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1111 a little endian target.
1113 2014-01-30 Richard Biener <rguenther@suse.de>
1115 PR tree-optimization/59993
1116 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1117 can propagate form the earlier stmt and avoid the transform
1118 when the intermediate result is needed.
1120 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1122 * README.Portability: Fix typo.
1124 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1126 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1127 comparison_operator with ordered_comparison_operator.
1129 2014-01-30 Nick Clifton <nickc@redhat.com>
1131 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1132 Rename to mn10300_store_multiple_regs.
1133 * config/mn10300/mn10300.c: Likewise.
1134 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1135 store_multiple_regs.
1136 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1137 Call mn10300_store_multiple_regs.
1139 2014-01-30 Nick Clifton <nickc@redhat.com>
1140 DJ Delorie <dj@redhat.com>
1142 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1143 %fp 2 to keep registers after it properly word-aligned.
1144 (rl78_alloc_physical_registers_umul): Handle the case where both
1145 input operands are the same.
1147 2014-01-30 Richard Biener <rguenther@suse.de>
1149 PR tree-optimization/59903
1150 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1153 2014-01-30 Jason Merrill <jason@redhat.com>
1156 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1159 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1161 2014-01-30 Richard Biener <rguenther@suse.de>
1163 PR tree-optimization/59951
1164 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1166 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1169 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1170 SFmode to DFmode case.
1172 2014-01-29 DJ Delorie <dj@redhat.com>
1174 * config/msp430/msp430.opt (-minrt): New.
1175 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1177 (ENDFILE_SPEC): Likewise.
1179 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1181 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1182 (estimate_function_body_sizes): Use it.
1184 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1187 * dwarf2out.c (is_cxx_auto): New.
1188 (is_base_type): Use it.
1189 (gen_type_die_with_usage): Likewise.
1191 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1193 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1194 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1195 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1196 -maltivec=be with LE targets.
1197 (vsx_mergeh_<mode>): Likewise.
1198 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1199 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1200 (altivec_vmrghb): Replace with define_expand and new
1201 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1202 (altivec_vmrghb_direct): New define_insn.
1203 (altivec_vmrghh): Replace with define_expand and new
1204 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1205 (altivec_vmrghh_direct): New define_insn.
1206 (altivec_vmrghw): Replace with define_expand and new
1207 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1208 (altivec_vmrghw_direct): New define_insn.
1209 (*altivec_vmrghsf): Adjust for endianness.
1210 (altivec_vmrglb): Replace with define_expand and new
1211 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1212 (altivec_vmrglb_direct): New define_insn.
1213 (altivec_vmrglh): Replace with define_expand and new
1214 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1215 (altivec_vmrglh_direct): New define_insn.
1216 (altivec_vmrglw): Replace with define_expand and new
1217 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1218 (altivec_vmrglw_direct): New define_insn.
1219 (*altivec_vmrglsf): Adjust for endianness.
1220 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1221 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1222 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1223 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1224 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1225 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1226 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1227 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1229 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1231 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1232 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1235 2014-01-29 Richard Biener <rguenther@suse.de>
1237 PR tree-optimization/58742
1238 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1239 associate_pointerplus_align.
1240 (associate_pointerplus_diff): New function.
1241 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1242 and associate_pointerplus_diff.
1244 2014-01-29 Richard Biener <rguenther@suse.de>
1246 * lto-streamer.h (LTO_major_version): Bump to 3.
1247 (LTO_minor_version): Reset to 0.
1249 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1251 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1252 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1253 (arm_file_start): Generate correct asm header for armv7ve.
1254 * config/arm/bpabi.h: Add multilib support for armv7ve.
1255 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1256 and cortex-a15 to armv7ve.
1257 * config/arm/t-aprofile: Add multilib support for armv7ve.
1258 * doc/invoke.texi: Document -march=armv7ve.
1260 2014-01-29 Richard Biener <rguenther@suse.de>
1262 PR tree-optimization/58742
1263 * tree-ssa-forwprop.c (associate_plusminus): Return true
1264 if we changed sth, defer EH cleanup to ...
1265 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1266 (simplify_mult): New function.
1268 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1271 PR tree-optimization/59920
1272 * tree.c (build_common_builtin_nodes): Remove
1273 __builtin_setjmp_dispatcher initialization.
1274 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1275 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1276 instead of gsi_after_labels + manually skipping debug stmts.
1277 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1278 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1279 * tree-inline.c (copy_edges_for_bb): Remove
1280 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1281 argument. Ignore computed_goto_p stmts. Don't call
1282 make_abnormal_goto_edges. If a call might need abnormal edges
1283 for non-local gotos, see if it already has an edge to
1284 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1285 with true argument, don't do anything then, otherwise add
1286 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1287 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1289 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1290 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1291 (lower_stmt): Don't set data->calls_builtin_setjmp.
1292 (lower_builtin_setjmp): Adjust comment.
1293 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1294 * tree-cfg.c (found_computed_goto): Remove.
1295 (factor_computed_gotos): Remove.
1296 (make_goto_expr_edges): Return bool, true for computed gotos.
1297 Don't call make_abnormal_goto_edges.
1298 (build_gimple_cfg): Don't set found_computed_goto, don't call
1299 factor_computed_gotos.
1300 (computed_goto_p): No longer static.
1301 (make_blocks): Don't set found_computed_goto.
1302 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1303 (make_edges): If make_goto_expr_edges returns true, push bb
1304 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1305 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1306 vector. Record mapping between bbs and OpenMP regions if there
1307 are any, adjust make_gimple_omp_edges caller. Call
1308 handle_abnormal_edges.
1309 (make_abnormal_goto_edges): Remove.
1310 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1311 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1312 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1313 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1314 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1315 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1316 filling *region also set *region_idx to (*region)->entry->index.
1319 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1320 For REGs set ORIGINAL_REGNO.
1322 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1324 * doc/md.texi: Mention that a target shouldn't implement
1325 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1328 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1330 PR tree-optimization/59594
1331 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1332 a copy of the datarefs vector rather than the vector itself.
1334 2014-01-28 Jason Merrill <jason@redhat.com>
1337 * dwarf2out.c (auto_die): New static.
1338 (gen_type_die_with_usage): Handle C++1y 'auto'.
1339 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1342 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1345 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1346 (SPEC_X32): Likewise.
1347 (SPEC_64): Likewise.
1348 * config/i386/i386.c (ix86_option_override_internal): Turn off
1349 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1351 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1352 * config/i386/i386.h (TARGET_16BIT): New macro.
1353 (TARGET_16BIT_P): Likewise.
1354 * config/i386/i386.opt: Add m16.
1355 * doc/invoke.texi: Document -m16.
1357 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1359 PR preprocessor/59935
1360 * input.c (location_get_source_line): Bail out on when line number
1361 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1363 2014-01-28 Richard Biener <rguenther@suse.de>
1365 PR tree-optimization/58742
1366 * tree-ssa-forwprop.c (associate_plusminus): Handle
1367 pointer subtraction of the form (T)(P + A) - (T)P.
1369 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1371 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1374 2014-01-28 Richard Biener <rguenther@suse.de>
1377 2014-01-28 Richard Biener <rguenther@suse.de>
1379 PR rtl-optimization/45364
1380 PR rtl-optimization/59890
1381 * var-tracking.c (local_get_addr_clear_given_value): Handle
1382 already cleared slot.
1383 (val_reset): Handle not allocated local_get_addr_cache.
1384 (vt_find_locations): Use post-order on the inverted CFG.
1386 2014-01-28 Richard Biener <rguenther@suse.de>
1388 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1390 2014-01-28 Richard Biener <rguenther@suse.de>
1392 PR rtl-optimization/45364
1393 PR rtl-optimization/59890
1394 * var-tracking.c (local_get_addr_clear_given_value): Handle
1395 already cleared slot.
1396 (val_reset): Handle not allocated local_get_addr_cache.
1397 (vt_find_locations): Use post-order on the inverted CFG.
1399 2014-01-28 Alan Modra <amodra@gmail.com>
1401 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1402 * configure.ac <recursive call for build != host>: Define
1403 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1404 and LD_FOR_BUILD too.
1405 * configure: Regenerate.
1407 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1409 * config/i386/i386.c (get_builtin_code_for_version): Separate
1410 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1411 Broadwell from Haswell.
1413 2014-01-27 Steve Ellcey <sellcey@mips.com>
1415 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1416 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1417 * config/mips/mips.c (mips_option_override): Change setting
1419 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1420 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1421 Change from Mask to Var.
1423 2014-01-27 Jeff Law <law@redhat.com>
1425 * ipa-inline.c (inline_small_functions): Fix typo.
1427 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1429 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1430 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1431 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1432 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1433 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1434 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1435 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1436 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1437 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1438 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1439 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1440 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1441 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1442 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1443 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1444 (_mm512_storeu_epi64): Ditto.
1445 (_mm512_cmpge_epi32_mask): Ditto.
1446 (_mm512_cmpge_epu32_mask): Ditto.
1447 (_mm512_cmpge_epi64_mask): Ditto.
1448 (_mm512_cmpge_epu64_mask): Ditto.
1449 (_mm512_cmple_epi32_mask): Ditto.
1450 (_mm512_cmple_epu32_mask): Ditto.
1451 (_mm512_cmple_epi64_mask): Ditto.
1452 (_mm512_cmple_epu64_mask): Ditto.
1453 (_mm512_cmplt_epi32_mask): Ditto.
1454 (_mm512_cmplt_epu32_mask): Ditto.
1455 (_mm512_cmplt_epi64_mask): Ditto.
1456 (_mm512_cmplt_epu64_mask): Ditto.
1457 (_mm512_cmpneq_epi32_mask): Ditto.
1458 (_mm512_cmpneq_epu32_mask): Ditto.
1459 (_mm512_cmpneq_epi64_mask): Ditto.
1460 (_mm512_cmpneq_epu64_mask): Ditto.
1461 (_mm512_expand_pd): Ditto.
1462 (_mm512_expand_ps): Ditto.
1463 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1464 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1465 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1466 * config/i386/i386.c (ix86_builtins): Add
1467 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1468 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1469 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1470 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1471 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1472 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1473 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1474 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1475 IX86_BUILTIN_PMOVUSQW512_MEM.
1476 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1477 __builtin_ia32_pmovsqd512mem_mask,
1478 __builtin_ia32_pmovqd512mem_mask,
1479 __builtin_ia32_pmovusqw512mem_mask,
1480 __builtin_ia32_pmovsqw512mem_mask,
1481 __builtin_ia32_pmovqw512mem_mask,
1482 __builtin_ia32_pmovusdw512mem_mask,
1483 __builtin_ia32_pmovsdw512mem_mask,
1484 __builtin_ia32_pmovdw512mem_mask,
1485 __builtin_ia32_pmovqb512mem_mask,
1486 __builtin_ia32_pmovusqb512mem_mask,
1487 __builtin_ia32_pmovsqb512mem_mask,
1488 __builtin_ia32_pmovusdb512mem_mask,
1489 __builtin_ia32_pmovsdb512mem_mask,
1490 __builtin_ia32_pmovdb512mem_mask.
1491 (bdesc_args): Add __builtin_ia32_expanddf512,
1492 __builtin_ia32_expandsf512.
1493 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1494 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1495 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1496 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1497 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1498 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1499 (avx512f_<code>v8div16qi2_mask_store): This.
1500 (avx512f_expand<mode>): New.
1502 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1504 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1506 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1507 (_mm512_prefetch_i32scatter_pd): Ditto.
1508 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1509 (_mm512_prefetch_i64scatter_pd): Ditto.
1510 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1511 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1512 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1513 (_mm512_prefetch_i32scatter_ps): Ditto.
1514 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1515 (_mm512_prefetch_i64scatter_ps): Ditto.
1516 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1517 * config/i386/i386-builtin-types.def: Define
1518 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1519 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1520 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1521 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1522 IX86_BUILTIN_SCATTERPFQPD.
1523 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1524 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1525 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1526 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1527 __builtin_ia32_scatterpfqps.
1528 (ix86_expand_builtin): Expand new built-ins.
1529 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1530 fix memory access data type.
1531 (*avx512pf_gatherpf<mode>_mask): Ditto.
1532 (*avx512pf_gatherpf<mode>): Ditto.
1533 (avx512pf_scatterpf<mode>): Ditto.
1534 (*avx512pf_scatterpf<mode>_mask): Ditto.
1535 (*avx512pf_scatterpf<mode>): Ditto.
1536 (GATHER_SCATTER_SF_MEM_MODE): New.
1537 (avx512pf_gatherpf<mode>df): Ditto.
1538 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1539 (*avx512pf_scatterpf<mode>df): Ditto.
1541 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1544 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1545 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1548 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1550 * common/config/arm/arm-common.c
1551 (arm_rewrite_mcpu): Handle multiple names.
1553 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1555 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1557 * gimple-builder.h (create_gimple_tmp): Delete.
1559 2014-01-27 Christian Bruel <christian.bruel@st.com>
1561 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1564 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1566 * config/pa/pa.md (call): Generate indirect long calls to non-local
1567 functions when outputing 32-bit code.
1568 (call_value): Likewise except for special call to buggy powf function.
1570 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1571 portable runtime and PIC indirect calls.
1572 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1573 and PIC call sequences. Use ldo instead of blr to set return register
1574 in PIC call sequence.
1576 2014-01-25 Walter Lee <walt@tilera.com>
1578 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1579 avoid clobbering a live register.
1581 2014-01-25 Walter Lee <walt@tilera.com>
1583 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1584 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1585 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1586 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1588 2014-01-25 Walter Lee <walt@tilera.com>
1590 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1591 arguments on even registers.
1592 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1594 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1595 (BIGGEST_ALIGNMENT): Ditto.
1596 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1598 2014-01-25 Walter Lee <walt@tilera.com>
1600 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1601 insns before bundling.
1602 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1604 2014-01-25 Walter Lee <walt@tilera.com>
1606 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1607 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1608 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1610 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1612 * config/mips/constraints.md (kl): Delete.
1613 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1614 define expands, using...
1615 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1616 instructions for MIPS16.
1617 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1618 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1620 2014-01-25 Walter Lee <walt@tilera.com>
1622 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1626 2014-01-25 Walter Lee <walt@tilera.com>
1628 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1629 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1631 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1633 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1636 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1638 * print-rtl.c (in_call_function_usage): New var.
1639 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1640 EXPR_LIST mode as mode and not as reg note name.
1643 * cfgloopmanip.c (copy_loop_info): If
1644 loop->warned_aggressive_loop_optimizations, make sure
1645 the flag is set in target loop too.
1647 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1649 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1651 * builtins.def: Likewise.
1652 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1653 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1654 * ira.c (ira_setup_eliminable_regset): Likewise.
1655 * omp-low.c (gate_expand_omp): Likewise.
1656 (execute_lower_omp): Likewise.
1657 (diagnose_sb_0): Likewise.
1658 (gate_diagnose_omp_blocks): Likewise.
1659 (simd_clone_clauses_extract): Likewise.
1662 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1664 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1665 correction for little endian...
1666 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1669 2014-01-24 Jeff Law <law@redhat.com>
1671 PR tree-optimization/59919
1672 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1673 for non-returning calls.
1675 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1677 * common/config/aarch64/aarch64-common.c
1678 (aarch64_rewrite_mcpu): Handle multiple names.
1679 * config/aarch64/aarch64.h
1680 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1682 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1684 * input.c (add_file_to_cache_tab): Handle the case where fopen
1687 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1690 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1691 from push operand if code of push isn't PRE_DEC.
1693 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1696 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1697 -mquad-memory-atomic. Update -mquad-memory documentation to say
1698 it is only used for non-atomic loads/stores.
1700 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1701 -mquad-memory or -mquad-memory-atomic switches.
1703 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1704 -mquad-memory-atomic to ISA 2.07 support.
1706 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1707 to separate support of normal quad word memory operations (ldq, stq)
1708 from the atomic quad word memory operations.
1710 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1711 support to separate non-atomic quad word operations from atomic
1712 quad word operations. Disable non-atomic quad word operations in
1713 little endian mode so that we don't have to swap words after the
1714 load and before the store.
1715 (quad_load_store_p): Add comment about atomic quad word support.
1716 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1717 options printed with -mdebug=reg.
1719 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1720 -mquad-memory-atomic as the test for whether we have quad word
1721 atomic instructions.
1722 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1723 or -mp8-vector are used, allow byte/half-word atomic operations.
1725 * config/rs6000/sync.md (load_lockedti): Insure that the address
1726 is a proper indexed or indirect address for the lqarx instruction.
1727 On little endian systems, swap the hi/lo registers after the lqarx
1729 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1730 insure the address is valid for the lqarx instruction.
1731 (store_conditionalti): Insure that the address is a proper indexed
1732 or indirect address for the stqcrx. instruction. On little endian
1733 systems, swap the hi/lo registers before doing the stqcrx.
1735 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1736 insure the address is valid for the stqcrx. instruction.
1738 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1739 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1740 type of quad memory support is available.
1742 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1745 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1746 there is a danger of looping.
1748 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1750 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1751 force flag_ira_loop_pressure if set via command line.
1753 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1755 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1756 (ashr_simd): New builtin handling DI mode.
1757 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1758 (aarch64_sshr_simddi): New match pattern.
1759 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1760 (vshrd_n_s64): Likewise.
1761 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1763 2014-01-23 Nick Clifton <nickc@redhat.com>
1765 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1766 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1767 favour of mcu specific scripts.
1768 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1771 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1772 Alex Velenko <Alex.Velenko@arm.com>
1774 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1775 (vaddv_s16): Likewise.
1776 (vaddv_s32): Likewise.
1777 (vaddv_u8): Likewise.
1778 (vaddv_u16): Likewise.
1779 (vaddv_u32): Likewise.
1780 (vaddvq_s8): Likewise.
1781 (vaddvq_s16): Likewise.
1782 (vaddvq_s32): Likewise.
1783 (vaddvq_s64): Likewise.
1784 (vaddvq_u8): Likewise.
1785 (vaddvq_u16): Likewise.
1786 (vaddvq_u32): Likewise.
1787 (vaddvq_u64): Likewise.
1788 (vaddv_f32): Likewise.
1789 (vaddvq_f32): Likewise.
1790 (vaddvq_f64): Likewise.
1791 (vmaxv_f32): Likewise.
1792 (vmaxv_s8): Likewise.
1793 (vmaxv_s16): Likewise.
1794 (vmaxv_s32): Likewise.
1795 (vmaxv_u8): Likewise.
1796 (vmaxv_u16): Likewise.
1797 (vmaxv_u32): Likewise.
1798 (vmaxvq_f32): Likewise.
1799 (vmaxvq_f64): Likewise.
1800 (vmaxvq_s8): Likewise.
1801 (vmaxvq_s16): Likewise.
1802 (vmaxvq_s32): Likewise.
1803 (vmaxvq_u8): Likewise.
1804 (vmaxvq_u16): Likewise.
1805 (vmaxvq_u32): Likewise.
1806 (vmaxnmv_f32): Likewise.
1807 (vmaxnmvq_f32): Likewise.
1808 (vmaxnmvq_f64): Likewise.
1809 (vminv_f32): Likewise.
1810 (vminv_s8): Likewise.
1811 (vminv_s16): Likewise.
1812 (vminv_s32): Likewise.
1813 (vminv_u8): Likewise.
1814 (vminv_u16): Likewise.
1815 (vminv_u32): Likewise.
1816 (vminvq_f32): Likewise.
1817 (vminvq_f64): Likewise.
1818 (vminvq_s8): Likewise.
1819 (vminvq_s16): Likewise.
1820 (vminvq_s32): Likewise.
1821 (vminvq_u8): Likewise.
1822 (vminvq_u16): Likewise.
1823 (vminvq_u32): Likewise.
1824 (vminnmv_f32): Likewise.
1825 (vminnmvq_f32): Likewise.
1826 (vminnmvq_f64): Likewise.
1828 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1830 * config/aarch64/aarch64-simd.md
1831 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1832 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1833 (*aarch64_mul3_elt<mode>): Likewise.
1834 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1835 (*aarch64_mul3_elt_to_64v2df): Likewise.
1836 (*aarch64_mla_elt<mode>): Likewise.
1837 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1838 (*aarch64_mls_elt<mode>): Likewise.
1839 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1840 (*aarch64_fma4_elt<mode>): Likewise.
1841 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1842 (*aarch64_fma4_elt_to_64v2df): Likewise.
1843 (*aarch64_fnma4_elt<mode>): Likewise.
1844 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1845 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1846 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1847 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1848 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1849 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1850 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1851 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1852 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1854 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1856 * config/aarch64/aarch64-simd.md
1857 (aarch64_be_checked_get_lane<mode>): New define_expand.
1858 * config/aarch64/aarch64-simd-builtins.def
1859 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1860 New builtin definition.
1861 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1862 Use new safe be builtin.
1864 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1866 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1868 (aarch64_be_st1<mode>): Likewise.
1869 (aarch_ld1<VALL:mode>): Define_expand modified.
1870 (aarch_st1<VALL:mode>): Likewise.
1871 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1872 (UNSPEC_ST1): Likewise.
1874 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1876 * config/microblaze/microblaze.md: Add trap insn and attribute
1878 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1880 PR preprocessor/58580
1881 * input.h (location_get_source_line): Take an additional line_size
1883 (void diagnostics_file_cache_fini): Declare new function.
1884 * input.c (struct fcache): New type.
1885 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1886 New static constants.
1887 (diagnostic_file_cache_init, total_lines_num)
1888 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1889 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1890 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1891 (get_next_line, read_next_line, goto_next_line, read_line_num):
1892 New static function definitions.
1893 (diagnostic_file_cache_fini): New function.
1894 (location_get_source_line): Take an additional output line_len
1895 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1897 * diagnostic.c (diagnostic_finish): Call
1898 diagnostic_file_cache_fini.
1899 (adjust_line): Take an additional input parameter for the length
1900 of the line, rather than calculating it with strlen.
1901 (diagnostic_show_locus): Adjust the use of
1902 location_get_source_line and adjust_line with respect to their new
1903 signature. While displaying a line now, do not stop at the first
1904 null byte. Rather, display the zero byte as a space and keep
1905 going until we reach the size of the line.
1906 * Makefile.in: Add vec.o to OBJS-libcommon
1908 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1909 Ilya Tocar <ilya.tocar@intel.com>
1911 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1912 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1913 (__builtin_ia32_kmov16): Ditto.
1914 * config/i386/i386.md (UNSPEC_KMOV): New.
1917 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1919 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1920 (_mm512_storeu_si512): Ditto.
1922 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1925 * rtl.h (get_referenced_operands): Declare.
1926 * recog.c (get_referenced_operands): New function.
1927 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1928 operands have been referenced when recording LO_SUM references.
1930 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1932 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1934 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1936 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1937 Enable for generic and recent AMD targets.
1939 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1941 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1942 ARG_SIZE note when adjustment was eliminated.
1944 2014-01-22 Jeff Law <law@redhat.com>
1946 PR tree-optimization/59597
1947 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1948 in file. Accept new argument REGISTERING and use it to modify
1949 dump output appropriately.
1950 (register_jump_thread): Corresponding changes.
1951 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1952 thread paths involving joiner blocks. Add code to dump cancelled
1953 jump threading paths.
1955 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1957 PR rtl-optimization/59477
1958 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1959 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1961 2014-01-22 Tom Tromey <tromey@redhat.com>
1963 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1965 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1967 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1969 PR rtl-optimization/59896
1970 * lra-constraints.c (process_alt_operands): Check unused note for
1971 matched operands of insn with no output reloads.
1973 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1975 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1976 (mips_move_from_gpr_cost): Likewise.
1978 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1980 PR rtl-optimization/59858
1981 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1982 ira_class_hard_regs_num.
1983 (process_alt_operands): Increase reject for dying matched operand.
1985 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1988 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1989 smaller than size, perform several stores or loads and stores
1990 at dst + count - size to store or copy all of size bytes, rather
1991 than just last modesize bytes.
1993 2014-01-20 DJ Delorie <dj@redhat.com>
1995 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1996 that CLOBBERs are REGs before propogating their values.
1998 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2001 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2002 (cgraph_inline_failed_type): New function.
2003 * cgraph.h (DEFCIFCODE): Add type.
2004 (cgraph_inline_failed_type_t): New enum.
2005 (cgraph_inline_failed_type): New prototype.
2006 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2007 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2008 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2009 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2010 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2011 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2012 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2013 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2014 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2015 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2016 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2017 OPTIMIZATION_MISMATCH.
2018 * tree-inline.c (expand_call_inline): Emit errors during
2019 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2021 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2024 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2025 mode attribute in insn output.
2027 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2029 * output.h (output_constant): Delete.
2030 * varasm.c (output_constant): Make private.
2032 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2034 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2036 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2039 * tree.h (fold_builtin_strcat): New prototype.
2040 * builtins.c (fold_builtin_strcat): No longer static. Add len
2041 argument, if non-NULL, don't call c_strlen. Optimize
2042 directly into __builtin_memcpy instead of __builtin_strcpy.
2043 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2044 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2046 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2048 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2049 for SImode_address_operand operands, having only a REG argument.
2051 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2053 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2054 loader name using mbig-endian.
2055 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2057 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2059 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2063 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2065 * config/aarch64/aarch64-protos.h
2066 (aarch64_cannot_change_mode_class_ptr): Declare.
2067 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2068 aarch64_cannot_change_mode_class_ptr): New.
2069 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2070 backend hook aarch64_cannot_change_mode_class.
2072 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2074 * common/config/aarch64/aarch64-common.c
2075 (aarch64_handle_option): Don't handle any option order logic here.
2076 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2077 selected_cpu, warn on architecture version mismatch.
2078 (aarch64_override_options): Fix parsing order for option strings.
2080 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2081 Iain Sandoe <iain@codesourcery.com>
2084 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2085 warning. Amend comment to reflect current functionality.
2087 2014-01-20 Richard Biener <rguenther@suse.de>
2090 * builtins.c (fold_builtin_strcat): Remove case better handled
2091 by tree-ssa-strlen.c.
2093 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2095 * config/aarch64/aarch64.opt
2096 (mcpu, march, mtune): Make case-insensitive.
2098 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2101 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2102 if operands[1] is a REG or ZERO_EXTEND of a REG.
2104 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2106 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2108 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2110 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2111 long non-pic millicode calls.
2113 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2115 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2117 2014-01-19 Kito Cheng <kito@0xlab.org>
2119 * builtins.c (expand_movstr): Check movstr expand done or fail.
2121 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2122 H.J. Lu <hongjiu.lu@intel.com>
2125 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2126 to DImode for zero-extended addresses.
2128 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2130 PR rtl-optimization/57763
2131 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2132 on the new indirect jump_insn and increment LABEL_NUSES (label).
2134 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2138 * config.gcc (x86_archs): New variable.
2139 (x86_64_archs): Likewise.
2140 (x86_cpus): Likewise.
2141 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2142 --with-arch/--with-cpu= options.
2143 Support --with-arch=/--with-cpu={nehalem,westmere,
2144 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2146 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2148 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2149 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2151 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2153 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2155 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2158 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2159 clear cpp_get_options (parse_in)->warn_unused_macros for
2160 ix86_target_macros_internal with cpp_define.
2162 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2164 * jump.c (delete_related_insns): Keep (use (insn))s.
2165 * reorg.c (redundant_insn): Check for barriers too.
2167 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2169 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2171 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2173 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2174 call to $$dyncall when TARGET_LONG_CALLS is true.
2176 2014-01-17 Jeff Law <law@redhat.com>
2178 * ree.c (combine_set_extension): Temporarily disable test for
2179 changing number of hard registers.
2181 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2184 * ipa-inline-analysis.c (inline_free_summary):
2185 Do not free summary of aliases.
2187 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2190 * gimplify.c (gimplify_expr): Use create_tmp_var
2191 instead of create_tmp_var_raw. If cond doesn't have
2192 integral type, don't add the IFN_ANNOTATE builtin at all.
2194 2014-01-17 Martin Jambor <mjambor@suse.cz>
2197 * ipa-cp.c (prev_edge_clone): New variable.
2198 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2199 Also resize prev_edge_clone vector.
2200 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2201 (ipcp_edge_removal_hook): New function.
2202 (ipcp_driver): Register ipcp_edge_removal_hook.
2204 2014-01-17 Andrew Pinski <apinski@cavium.com>
2205 Steve Ellcey <sellcey@mips.com>
2208 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2211 2014-01-17 Jeff Law <law@redhat.com>
2214 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2215 so that pass_ccp runs first.
2217 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2219 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2220 (ix86_adjust_cost): Use !TARGET_XXX.
2221 (do_reorder_for_imul): Likewise.
2222 (swap_top_of_ready_list): Likewise.
2223 (ix86_sched_reorder): Likewise.
2225 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2227 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2228 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2229 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2230 (intel_memset): New. Duplicate slm_memset.
2231 (intel_cost): New. Duplicate slm_cost.
2232 (m_INTEL): New macro.
2233 (processor_target_table): Add "intel".
2234 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2235 with PROCESSOR_INTEL for "intel".
2236 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2237 PROCESSOR_SILVERMONT.
2238 (ix86_issue_rate): Likewise.
2239 (ix86_adjust_cost): Likewise.
2240 (ia32_multipass_dfa_lookahead): Likewise.
2241 (swap_top_of_ready_list): Likewise.
2242 (ix86_sched_reorder): Likewise.
2243 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2244 instead of TARGET_OPT_AGU.
2245 * config/i386/i386.h (TARGET_INTEL): New.
2246 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2247 (processor_type): Add PROCESSOR_INTEL.
2248 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2249 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2251 2014-01-17 Marek Polacek <polacek@redhat.com>
2254 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2257 2014-01-17 Richard Biener <rguenther@suse.de>
2259 PR tree-optimization/46590
2260 * opts.c (default_options_table): Add entries for
2261 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2262 all enabled at -O1 but not for -Og.
2263 * common.opt (fbranch-count-reg): Remove Init(1).
2264 (fmove-loop-invariants): Likewise.
2265 (ftree-pta): Likewise.
2267 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2269 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2270 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2271 decls to at least the GCC 4.8 used alignments.
2274 * tree-nested.c (convert_nonlocal_reference_stmt,
2275 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2276 of GIMPLE_BIND stmts, adjust associated decls.
2278 2014-01-17 Richard Biener <rguenther@suse.de>
2280 PR tree-optimization/46590
2281 * vec.h (vec<>::bseach): New member function implementing
2282 binary search according to C89 bsearch.
2283 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2284 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2285 bitmap pointer again. Make accesses_in_loop a flat array.
2286 (mem_ref_obstack): New global.
2287 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2288 (mark_ref_stored): Likewise.
2289 (ref_indep_loop_p_2): Likewise.
2290 (set_ref_stored_in_loop): New helper function.
2291 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2292 (memref_free): Adjust.
2293 (record_mem_ref_loc): Simplify.
2294 (gather_mem_refs_stmt): Adjust.
2295 (sort_locs_in_loop_postorder_cmp): New function.
2296 (analyze_memory_references): Sort accesses_in_loop after
2297 loop postorder number.
2298 (find_ref_loc_in_loop_cmp): New function.
2299 (for_all_locs_in_loop): Find relevant cluster of locs in
2300 accesses_in_loop and iterate without recursion.
2301 (execute_sm): Avoid uninit warning.
2302 (struct ref_always_accessed): Simplify.
2303 (ref_always_accessed::operator ()): Likewise.
2304 (ref_always_accessed_p): Likewise.
2305 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2306 loop postorder numbers here.
2307 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2310 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2313 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2314 on decls for which assemble_alias has been called.
2316 2014-01-17 Nick Clifton <nickc@redhat.com>
2318 * config/msp430/msp430.opt: (mcpu): New option.
2319 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2320 (msp430_option_override): Parse target_cpu. If the MCU name
2321 matches a generic string, clear target_mcu.
2322 (msp430_attr): Allow numeric interrupt values up to 63.
2323 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2324 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2326 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2328 * config/msp430/msp430.md (popm): Use %J rather than %I.
2329 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2330 (addhi_cy_i): Use immediate_operand for operand 2.
2331 * doc/invoke.texi: Document -mcpu option.
2333 2014-01-17 Richard Biener <rguenther@suse.de>
2335 PR rtl-optimization/38518
2336 * df.h (df_analyze_loop): Declare.
2337 * df-core.c: Include cfgloop.h.
2338 (df_analyze_1): Split out main part of df_analyze.
2339 (df_analyze): Adjust.
2340 (loop_inverted_post_order_compute): New function.
2341 (loop_post_order_compute): Likewise.
2342 (df_analyze_loop): New function avoiding whole-function
2344 * loop-invariant.c (find_defs): Use df_analyze_loop.
2345 (find_invariants): Adjust.
2346 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2348 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2350 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2351 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2353 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2355 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2356 traversal when removing references.
2358 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2361 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2363 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2366 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2367 pushing a reload for an autoinc when we had previously reloaded an
2368 inner part of the address.
2370 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2372 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2374 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2375 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2376 when not giving up or versioning for alias only because of
2378 (vect_analyze_data_ref_dependences): Set to true.
2379 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2381 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2382 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2386 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2389 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2390 operand 0 predicate for gathers, use a new pseudo as subtarget.
2392 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2395 * lra-constraints.c (process_alt_operands): Add printing debug info.
2396 Check absence of input/output reloads for matched operands too.
2398 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2400 PR rtl-optimization/59835
2401 * ira.c (ira_init_register_move_cost): Increase cost for
2404 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2406 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2408 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2411 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2412 non-register objects. Use gen_(high/low)part more consistently.
2415 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2418 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2419 endian support, remove tests for WORDS_BIG_ENDIAN.
2420 (p8_mfvsrd_3_<mode>): Likewise.
2421 (reload_gpr_from_vsx<mode>): Likewise.
2422 (reload_gpr_from_vsxsf): Likewise.
2423 (p8_mfvsrd_4_disf): Likewise.
2425 2014-01-16 Richard Biener <rguenther@suse.de>
2427 PR rtl-optimization/46590
2428 * lcm.c (compute_antinout_edge): Use postorder iteration.
2429 (compute_laterin): Use inverted postorder iteration.
2431 2014-01-16 Nick Clifton <nickc@redhat.com>
2434 * varasm.c (output_constant): Return the number of bytes actually
2436 (output_constructor_array_range): Update the field size with the
2437 number of bytes emitted by output_constant.
2438 (output_constructor_regular_field): Likewise. Also do not
2439 complain if the total number of bytes emitted is now greater
2440 than the expected fieldpos.
2441 * output.h (output_constant): Update prototype and descriptive comment.
2443 2014-01-16 Marek Polacek <polacek@redhat.com>
2446 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2447 it is error_mark_node.
2449 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2451 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2452 VALID_AVX256_REG_OR_OI_MODE.
2454 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2456 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2457 current procedure should be profiled.
2459 2014-01-15 Andrew Pinski <apinski@cavium.com>
2461 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2462 of moving from/to the STACK_REG register class.
2464 2014-01-15 Richard Henderson <rth@redhat.com>
2467 * reginfo.c (global_regs_decl): Globalize.
2468 * rtl.h (global_regs_decl): Declare.
2469 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2470 reserved via global_regs.
2472 2014-01-15 Teresa Johnson <tejohnson@google.com>
2474 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2476 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2478 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2479 and vmulosh rather than call gen_vec_widen_smult_*.
2480 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2481 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2482 (vec_widen_smult_even_v16qi): Likewise.
2483 (vec_widen_umult_even_v8hi): Likewise.
2484 (vec_widen_smult_even_v8hi): Likewise.
2485 (vec_widen_umult_odd_v16qi): Likewise.
2486 (vec_widen_smult_odd_v16qi): Likewise.
2487 (vec_widen_umult_odd_v8hi): Likewise.
2488 (vec_widen_smult_odd_v8hi): Likewise.
2489 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2490 vmuloub rather than call gen_vec_widen_umult_*.
2491 (vec_widen_umult_lo_v16qi): Likewise.
2492 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2493 vmulosb rather than call gen_vec_widen_smult_*.
2494 (vec_widen_smult_lo_v16qi): Likewise.
2495 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2496 rather than call gen_vec_widen_umult_*.
2497 (vec_widen_umult_lo_v8hi): Likewise.
2498 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2499 rather than call gen_vec_widen_smult_*.
2500 (vec_widen_smult_lo_v8hi): Likewise.
2502 2014-01-15 Jeff Law <law@redhat.com>
2504 PR tree-optimization/59747
2505 * ree.c (find_and_remove_re): Properly handle case where a second
2506 eliminated extension requires widening a copy created for elimination
2507 of a prior extension.
2508 (combine_set_extension): Ensure that the number of hard regs needed
2509 for a destination register does not change when we widen it.
2511 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2513 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2514 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2515 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2516 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2517 (avr-*-rtems*): Likewise.
2518 (bfin*-rtems*): Likewise.
2519 (moxie-*-rtems*): Likewise.
2520 (h8300-*-rtems*): Likewise.
2521 (i[34567]86-*-rtems*): Likewise.
2522 (lm32-*-rtems*): Likewise.
2523 (m32r-*-rtems*): Likewise.
2524 (m68k-*-rtems*): Likewise.
2525 (microblaze*-*-rtems*): Likewise.
2526 (mips*-*-rtems*): Likewise.
2527 (powerpc-*-rtems*): Likewise.
2528 (sh-*-rtems*): Likewise.
2529 (sparc-*-rtems*): Likewise.
2530 (sparc64-*-rtems*): Likewise.
2531 (v850-*-rtems*): Likewise.
2532 (m32c-*-rtems*): Likewise.
2534 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2536 PR rtl-optimization/59511
2537 * ira.c (ira_init_register_move_cost): Use memory costs for some
2538 cases of register move cost calculations.
2539 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2540 instead of BB frequency.
2541 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2542 * lra-assigns.c (find_hard_regno_for): Ditto.
2544 2014-01-15 Richard Biener <rguenther@suse.de>
2546 PR tree-optimization/59822
2547 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2548 (vectorizable_load): Use it to hoist defs of uses of invariant
2549 loads out of the loop.
2551 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2552 Kugan Vivekanandarajah <kuganv@linaro.org>
2555 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2558 2014-01-15 Richard Biener <rguenther@suse.de>
2560 PR rtl-optimization/59802
2561 * lcm.c (compute_available): Use inverted postorder to seed
2562 the initial worklist.
2564 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2567 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2568 ADDR_REGS for invalid symrefs in non-PIC code.
2570 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2573 * builtins.c (determine_block_size): Initialize *probable_max_size
2574 even if len_rtx is CONST_INT.
2576 2014-01-14 Andrew Pinski <apinski@cavium.com>
2578 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2579 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2580 (cortexa53_tunings): Likewise.
2581 (aarch64_sched_issue_rate): New function.
2582 (TARGET_SCHED_ISSUE_RATE): Define.
2584 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2586 * ira-costs.c (find_costs_and_classes): Add missed
2587 ira_init_register_move_cost_if_necessary.
2589 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2592 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2594 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2597 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2598 to indicate if type is used for function return value. Warn ABI
2599 change if the vector mode isn't available for function return value.
2600 (ix86_function_arg_advance): Pass false to type_natural_mode.
2601 (ix86_function_arg): Likewise.
2602 (ix86_gimplify_va_arg): Likewise.
2603 (function_arg_32): Don't warn ABI change.
2604 (ix86_function_value): Pass true to type_natural_mode.
2605 (ix86_return_in_memory): Likewise.
2606 (ix86_struct_value_rtx): Removed.
2607 (TARGET_STRUCT_VALUE_RTX): Likewise.
2609 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2611 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2612 converting a conditional jump into a conditional return.
2614 2014-01-14 Richard Biener <rguenther@suse.de>
2616 PR tree-optimization/58921
2617 PR tree-optimization/59006
2618 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2619 hoisting invariant stmts.
2620 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2621 invariant loads on the preheader edge if possible.
2623 2014-01-14 Joey Ye <joey.ye@arm.com>
2625 * doc/plugin.texi (Building GCC plugins): Update to C++.
2627 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2629 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2630 (_mm_rcp28_round_ss): Ditto.
2631 (_mm_rsqrt28_round_sd): Ditto.
2632 (_mm_rsqrt28_round_ss): Ditto.
2633 (_mm_rcp28_sd): Ditto.
2634 (_mm_rcp28_ss): Ditto.
2635 (_mm_rsqrt28_sd): Ditto.
2636 (_mm_rsqrt28_ss): Ditto.
2637 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2638 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2639 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2640 (IX86_BUILTIN_RCP28SD): Ditto.
2641 (IX86_BUILTIN_RCP28SS): Ditto.
2642 (IX86_BUILTIN_RSQRT28SD): Ditto.
2643 (IX86_BUILTIN_RSQRT28SS): Ditto.
2644 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2645 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2646 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2647 (ix86_expand_special_args_builtin): Expand new FTYPE.
2648 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2649 (srcp14<mode>): Make insn unary.
2650 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2651 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2652 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2653 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2654 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2655 Fix rounding: make it SAE only.
2656 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2658 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2660 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2661 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2662 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2663 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2664 (round_saeonly_mask_scalar_operand4): Ditto.
2665 (round_saeonly_mask_scalar_op3): Ditto.
2666 (round_saeonly_mask_scalar_op4): Ditto.
2668 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2670 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2671 Implement -maltivec=be for vec_insert and vec_extract.
2673 2014-01-10 DJ Delorie <dj@redhat.com>
2675 * config/msp430/msp430.md (call_internal): Don't allow memory
2676 references with SP as the base register.
2677 (call_value_internal): Likewise.
2678 * config/msp430/constraints.md (Yc): New. For memory references
2679 that don't use SP as a base register.
2681 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2682 "an integer without a # prefix"
2683 * config/msp430/msp430.md (epilogue_helper): Use it.
2685 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2688 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2689 AVX512F gather builtins.
2690 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2691 on gather decls with INTEGER_TYPE masktype.
2692 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2693 directly into the builtin rather than hoisting it before loop.
2695 PR tree-optimization/59387
2696 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2697 (scev_const_prop): If folded_casts and type has undefined overflow,
2698 use force_gimple_operand instead of force_gimple_operand_gsi and
2699 for each added stmt if it is assign with
2700 arith_code_with_undefined_signed_overflow, call
2701 rewrite_to_defined_overflow.
2702 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2703 gimple-fold.h instead.
2704 (arith_code_with_undefined_signed_overflow,
2705 rewrite_to_defined_overflow): Moved to ...
2706 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2707 rewrite_to_defined_overflow): ... here. No longer static.
2708 Include gimplify-me.h.
2709 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2710 rewrite_to_defined_overflow): New prototypes.
2712 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2714 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2716 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2718 * builtins.c (get_object_alignment_2): Minor tweak.
2719 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2721 2014-01-13 Christian Bruel <christian.bruel@st.com>
2723 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2724 optimized non constant lengths.
2726 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2729 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2730 load as __atomic_load_N if possible.
2732 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2734 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2736 (rs6000_expand_builtin): Adjust call.
2738 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2741 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2742 * config/rs6000/rs6000.c: Include target-globals.h.
2743 (rs6000_set_current_function): Instead of doing target_reinit
2744 unconditionally, use save_target_globals_default_opts and
2745 restore_target_globals.
2747 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2749 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2750 (rs6000_expand_builtin): Handle mffs and mtfsf.
2751 (rs6000_init_builtins): Define mffs and mtfsf.
2752 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2753 (rs6000_mffs): New pattern.
2754 (rs6000_mtfsf): New pattern.
2756 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2758 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2759 Start narrowing with START. Apply candidate-use pair
2760 and check overall cost in narrowing.
2761 (iv_ca_prune): Pass new argument.
2763 2014-01-10 Jeff Law <law@redhat.com>
2766 * ree.c (combine_reaching_defs): Ensure the defining statement
2767 occurs before the extension when optimizing extensions with
2768 different source and destination hard registers.
2770 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2773 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2774 vtables into the type inheritance graph.
2776 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2778 PR rtl-optimization/59754
2779 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2780 modes in the REGNO != REGNO case.
2782 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2784 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2786 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2788 PR tree-optimization/59745
2789 * tree-predcom.c (tree_predictive_commoning_loop): Call
2790 free_affine_expand_cache if giving up because components is NULL.
2792 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2793 GC in payload of target_globals struct instead of allocating them on
2794 the heap and the larger structs separately using GC.
2795 * target-globals.h (struct target_globals): Make regs, hard_regs,
2796 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2797 of GTY((skip)) and change type to void *.
2798 (reset_target_globals): Cast loads from those fields to corresponding
2801 2014-01-10 Steve Ellcey <sellcey@mips.com>
2804 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2805 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2806 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2808 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2811 * aarch64-modes.def (CC_Zmode): New flags mode.
2812 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2813 represents an equality.
2814 (aarch64_get_condition_code): Handle CC_Zmode.
2815 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2817 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2819 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2820 extraction in good case.
2822 2014-01-10 Richard Biener <rguenther@suse.de>
2824 PR tree-optimization/59374
2825 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2826 checking after SLP discovery. Mark stmts not participating
2827 in any SLP instance properly.
2829 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2831 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2832 when handling a SET rtx.
2834 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2836 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2837 (cortex-a57): Likewise.
2838 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2840 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2842 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2843 non-iwmmxt builtins.
2845 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2849 * ipa-devirt.c record_target_from_binfo): Take as argument
2850 stack of binfos and lookup matching one for virtual inheritance.
2851 (possible_polymorphic_call_targets_1): Update.
2853 2014-01-10 Huacai Chen <chenhc@lemote.com>
2855 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2856 kernel strings for Loongson-2E/2F/3A.
2858 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2861 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2862 is_gimple_call before calling gimple_call_internal_p.
2864 2014-01-09 Steve Ellcey <sellcey@mips.com>
2866 * Makefile.in (TREE_FLOW_H): Remove.
2867 (TREE_SSA_H): Add file names from tree-flow.h.
2868 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2869 * tree.h: Remove tree-flow.h reference.
2870 * hash-table.h: Remove tree-flow.h reference.
2871 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2872 reference with tree-ssa-loop.h.
2874 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2876 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2877 default element-order behavior for -maltivec.
2878 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2879 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2880 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2881 when targeting big endian, at least for now.
2882 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2884 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2887 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2888 var satisfies use_register_for_decl, just take into account type
2889 alignment, rather than decl alignment.
2891 PR tree-optimization/59622
2892 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2893 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2894 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2895 Don't devirtualize for inplace at all. For targets.length () == 1,
2896 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2898 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2900 * config/i386/i386.md (cpu): Remove the unused btver1.
2902 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2904 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2906 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2909 * tree-core.h (struct target_globals): New forward declaration.
2910 (struct tree_target_option): Add globals field.
2911 * tree.h (TREE_TARGET_GLOBALS): Define.
2912 (prepare_target_option_nodes_for_pch): New prototype.
2913 * target-globals.h (struct target_globals): Define even if
2915 * tree.c (prepare_target_option_node_for_pch,
2916 prepare_target_option_nodes_for_pch): New functions.
2917 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2918 * config/i386/i386.c: Include target-globals.h.
2919 (ix86_set_current_function): Instead of doing target_reinit
2920 unconditionally, use save_target_globals_default_opts and
2921 restore_target_globals.
2923 2014-01-09 Richard Biener <rguenther@suse.de>
2925 PR tree-optimization/59715
2926 * tree-cfg.h (split_critical_edges): Declare.
2927 * tree-cfg.c (split_critical_edges): Export.
2928 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2930 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2932 * cfgexpand.c (expand_stack_vars): Optionally disable
2933 asan stack protection.
2934 (expand_used_vars): Likewise.
2935 (partition_stack_vars): Likewise.
2936 * asan.c (asan_emit_stack_protection): Optionally disable
2937 after return stack usage.
2938 (instrument_derefs): Optionally disable memory access instrumentation.
2939 (instrument_builtin_call): Likewise.
2940 (instrument_strlen_call): Likewise.
2941 (asan_protect_global): Optionally disable global variables protection.
2942 * doc/invoke.texi: Added doc for new options.
2943 * params.def: Added new options.
2944 * params.h: Likewise.
2946 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2948 PR rtl-optimization/59724
2949 * ifcvt.c (cond_exec_process_if_block): Don't call
2950 flow_find_head_matching_sequence with 0 longest_match.
2951 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2952 non-active insns if !stop_after.
2953 (try_head_merge_bb): Revert 2014-01-07 changes.
2955 2014-01-08 Jeff Law <law@redhat.com>
2957 * ree.c (get_sub_rtx): New function, extracted from...
2958 (merge_def_and_ext): Here.
2959 (combine_reaching_defs): Use get_sub_rtx.
2961 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2963 * cgraph.h (varpool_variable_node): Do not choke on null node.
2965 2014-01-08 Catherine Moore <clm@codesourcery.com>
2967 * config/mips/mips.md (simple_return): Attempt to use JRC
2969 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2971 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2973 PR rtl-optimization/59137
2974 * reorg.c (steal_delay_list_from_target): Call update_block for
2976 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2978 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2980 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2981 two duplicate entries.
2983 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2986 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2988 * config/mips/mips.c (mips_truncated_op_cost): New function.
2989 (mips_rtx_costs): Adjust test for BADDU.
2990 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2992 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2994 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2995 (*baddu_si): ...this new pattern.
2997 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3000 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3002 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3005 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3007 (expand_expr, expand_normal): Adjust.
3008 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3009 inner_reference_p. Use inner_reference_p to expand inner references.
3010 (store_expr): Adjust.
3011 * cfgexpand.c (expand_call_stmt): Adjust.
3013 2014-01-08 Rong Xu <xur@google.com>
3015 * gcov-io.c (gcov_var): Move from gcov-io.h.
3016 (gcov_position): Ditto.
3017 (gcov_is_error): Ditto.
3018 (gcov_rewrite): Ditto.
3019 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3020 only part to libgcc/libgcov.h.
3022 2014-01-08 Marek Polacek <polacek@redhat.com>
3025 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3027 2014-01-08 Marek Polacek <polacek@redhat.com>
3030 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3032 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3034 PR rtl-optimization/59649
3035 * stor-layout.c (get_mode_bounds): For BImode return
3036 0 and STORE_FLAG_VALUE.
3038 2014-01-08 Richard Biener <rguenther@suse.de>
3041 * gimple.h (is_gimple_builtin_call): Remove.
3042 (gimple_builtin_call_types_compatible_p): New.
3043 (gimple_call_builtin_p): New overload.
3044 * gimple.c (is_gimple_builtin_call): Remove.
3045 (validate_call): Rename to ...
3046 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3048 (validate_type): New static function.
3049 (gimple_call_builtin_p): New overload and adjust.
3050 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3051 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3052 (gimple_fold_stmt_to_constant_1): Likewise.
3053 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3055 2014-01-08 Richard Biener <rguenther@suse.de>
3058 * gimplify.c (gimplify_expr): Gimplify register-register type
3059 VIEW_CONVERT_EXPRs to separate stmts.
3061 2014-01-07 Jeff Law <law@redhat.com>
3064 * ree.c (combine_set_extension): Handle case where source
3065 and destination registers in an extension insn are different.
3066 (combine_reaching_defs): Allow source and destination registers
3067 in extension to be different under limited circumstances.
3068 (add_removable_extension): Remove restriction that the
3069 source and destination registers in the extension are the same.
3070 (find_and_remove_re): Emit a copy from the extension's
3071 destination to its source after the defining insn if
3072 the source and destination registers are different.
3075 * ifcvt.c (merge_if_block): If we are merging a block with more than
3076 one successor with a block with no successors, remove any BARRIER
3077 after the second block.
3079 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3081 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3083 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3086 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3087 for 14-bit register offsets when INT14_OK_STRICT is false.
3089 2014-01-07 Roland Stigge <stigge@antcom.de>
3090 Michael Meissner <meissner@linux.vnet.ibm.com>
3093 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3094 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3096 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3098 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3101 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3103 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3104 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3105 rtx is const0_rtx or not.
3107 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3110 * target-globals.c (save_target_globals): Remove this_fn_optab
3112 * toplev.c: Include optabs.h.
3113 (target_reinit): Temporarily restore the global options if another
3114 set of options are in force.
3116 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3118 PR rtl-optimization/58668
3119 * cfgcleanup.c (flow_find_cross_jump): Don't count
3120 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3121 to determine what is counted.
3122 (flow_find_head_matching_sequence): Use active_insn_p to determine
3124 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3126 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3127 determine what is counted.
3129 PR tree-optimization/59643
3130 * tree-predcom.c (split_data_refs_to_components): If one dr is
3131 read and one write, determine_offset fails and the write isn't
3132 in the bad component, just put the read into the bad component.
3134 2014-01-07 Mike Stump <mikestump@comcast.net>
3135 Jakub Jelinek <jakub@redhat.com>
3138 * tree-core.h (struct tree_optimization_option): Change optabs
3139 type from unsigned char * to void *.
3140 * optabs.c (init_tree_optimization_optabs): Adjust
3141 TREE_OPTIMIZATION_OPTABS initialization.
3143 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3146 * config/i386/i386.h (struct machine_function): Add
3147 no_drap_save_restore field.
3148 * config/i386/i386.c (ix86_save_reg): Use
3149 !cfun->machine->no_drap_save_restore instead of
3150 crtl->stack_realign_needed.
3151 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3152 this function clears frame_pointer_needed. Set
3153 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3154 and DRAP reg is needed.
3156 2014-01-06 Marek Polacek <polacek@redhat.com>
3159 * doc/implement-c.texi: Mention that other integer types are
3160 permitted as bit-field types in strictly conforming mode.
3162 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3164 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3167 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3169 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3171 2014-01-06 Martin Jambor <mjambor@suse.cz>
3174 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3176 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3178 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3182 * var-tracking.c (add_stores): Preserve the value of the source even if
3183 we don't record the store.
3185 2014-01-06 Terry Guo <terry.guo@arm.com>
3187 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3189 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3192 * config/darwin.c (darwin_function_section): Adjust return values to
3193 correspond to optimisation changes made in r206070.
3195 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3197 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3198 from prefetch_block tune setting.
3199 (nocona_cost): Correct size of prefetch block to 64.
3201 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3203 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3204 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3205 used to save the static chain register in the computation of the offset
3206 from which the FP registers need to be restored.
3208 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3210 PR tree-optimization/59519
3211 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3212 ICE if get_current_def (current_new_name) is already non-NULL, as long
3213 as it is a phi result of some other phi in *new_exit_bb that has
3216 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3217 or vmovdqu* for misaligned_operand.
3218 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3219 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3220 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3221 aligned_mem for AVX512F masked aligned load and store builtins and for
3224 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3226 PR tree-optimization/59651
3227 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3228 Address range for negative step should be added by TYPE_SIZE_UNIT.
3230 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3232 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3233 overlapping registers also for registers other than the stack pointer.
3235 2014-01-03 Marek Polacek <polacek@redhat.com>
3238 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3241 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3244 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3247 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3248 (push splitter): Use <P:MODE_SIZE> instead of
3249 GET_MODE_SIZE (<P:MODE>mode).
3250 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3251 (mov -1, reg peephole2): Likewise.
3252 * config/i386/sse.md (*mov<mode>_internal,
3253 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3254 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3255 *<code><mode>3, *andnot<mode>3<mask_name>,
3256 <mask_codefor><code><mode>3<mask_name>): Likewise.
3257 * config/i386/subst.md (mask_mode512bit_condition,
3258 sd_mask_mode512bit_condition): Likewise.
3260 2014-01-02 Xinliang David Li <davidxl@google.com>
3262 PR tree-optimization/59303
3263 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3264 (dump_predicates): Better output format.
3265 (pred_equal_p): New function.
3266 (is_neq_relop_p): Ditto.
3267 (is_neq_zero_form_p): Ditto.
3268 (pred_expr_equal_p): Ditto.
3269 (pred_neg_p): Ditto.
3270 (simplify_pred): Ditto.
3271 (simplify_preds_2): Ditto.
3272 (simplify_preds_3): Ditto.
3273 (simplify_preds_4): Ditto.
3274 (simplify_preds): Ditto.
3276 (push_to_worklist): Ditto.
3277 (get_pred_info_from_cmp): Ditto.
3278 (is_degenerated_phi): Ditto.
3279 (normalize_one_pred_1): Ditto.
3280 (normalize_one_pred): Ditto.
3281 (normalize_one_pred_chain): Ditto.
3282 (normalize_preds): Ditto.
3283 (normalize_cond_1): Remove function.
3284 (normalize_cond): Ditto.
3285 (is_gcond_subset_of): Ditto.
3286 (is_subset_of_any): Ditto.
3287 (is_or_set_subset_of): Ditto.
3288 (is_and_set_subset_of): Ditto.
3289 (is_norm_cond_subset_of): Ditto.
3290 (pred_chain_length_cmp): Ditto.
3291 (convert_control_dep_chain_into_preds): Type change.
3292 (find_predicates): Ditto.
3293 (find_def_preds): Ditto.
3294 (destroy_predicates_vecs): Ditto.
3295 (find_matching_predicates_in_rest_chains): Ditto.
3296 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3297 (is_pred_expr_subset): Ditto.
3298 (is_pred_chain_subset_of): Ditto.
3299 (is_included_in): Ditto.
3300 (is_superset_of): Ditto.
3302 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3304 Update copyright years.
3306 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3308 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3309 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3310 config/arc/arc.md, config/arc/arc.opt,
3311 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3312 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3313 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3314 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3315 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3316 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3317 vtable-verify.c, vtable-verify.h: Use the standard form for the
3320 2014-01-02 Tobias Burnus <burnus@net-b.de>
3322 * gcc.c (process_command): Update copyright notice dates.
3323 * gcov-dump.c: Ditto.
3325 * doc/cpp.texi: Bump @copying's copyright year.
3326 * doc/cppinternals.texi: Ditto.
3327 * doc/gcc.texi: Ditto.
3328 * doc/gccint.texi: Ditto.
3329 * doc/gcov.texi: Ditto.
3330 * doc/install.texi: Ditto.
3331 * doc/invoke.texi: Ditto.
3333 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3335 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3337 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3339 * config/i386/sse.md (*mov<mode>_internal): Guard
3340 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3342 PR rtl-optimization/59647
3343 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3344 new_rtx into UNSIGNED_FLOAT rtxes.
3346 Copyright (C) 2014 Free Software Foundation, Inc.
3348 Copying and distribution of this file, with or without modification,
3349 are permitted in any medium without royalty provided the copyright
3350 notice and this notice are preserved.