1 2014-02-14 Kyle McMartin <kyle@redhat.com>
4 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
6 2014-02-14 Richard Biener <rguenther@suse.de>
8 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
9 (get_frame_arg): Drop the assert with langhook types_compatible_p.
10 Do not strip INDIRECT_REFs.
12 2014-02-14 Richard Biener <rguenther@suse.de>
15 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
16 DECL_FUNCTION_SPECIFIC_TARGET.
17 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
18 * tree-streamer-out.c (pack_ts_target_option): Remove.
19 (streamer_pack_tree_bitfields): Do not stream
21 (write_ts_function_decl_tree_pointers): Do not stream
22 DECL_FUNCTION_SPECIFIC_TARGET.
23 * tree-streamer-in.c (unpack_ts_target_option): Remove.
24 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
25 (lto_input_ts_function_decl_tree_pointers): Do not stream
26 DECL_FUNCTION_SPECIFIC_TARGET.
28 2014-02-14 Jakub Jelinek <jakub@redhat.com>
30 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
32 (get_initial_def_for_induction, vectorizable_induction): Ignore
33 debug stmts when looking for exit_phi.
34 (vectorizable_live_operation): Fix up condition.
36 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
38 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
39 nreverse() because it changes the content of original tree list.
41 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
43 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
44 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
46 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
48 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
51 2014-02-13 Jakub Jelinek <jakub@redhat.com>
54 * dwarf2out.c (gen_subprogram_die): Don't call
55 add_calling_convention_attribute if subr_die is old_die.
57 2014-02-13 Sharad Singhai <singhai@google.com>
59 * doc/optinfo.texi: Fix order of nodes.
61 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
63 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
64 operands[2], not operands[3].
66 2014-02-13 Richard Biener <rguenther@suse.de>
69 * doc/install.texi (ISL): Update recommended version to 0.12.2,
70 mention the possibility of an in-tree build.
71 (CLooG): Update recommended version to 0.18.1, mention the
72 possibility of an in-tree build and clarify that the ISL
73 bundled with CLooG does not work.
75 2014-02-13 Jakub Jelinek <jakub@redhat.com>
78 * expr.c (compress_float_constant): If x is a hard register,
79 extend into a pseudo and then move to x.
81 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
83 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
84 caused by bad second argument to warning_at() with -mhotpatch and
85 nested functions (e.g. with gfortran).
87 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
89 * opts.c (option_name): Remove "enabled by default" rider.
91 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
93 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
95 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
96 Uros Bizjak <ubizjak@gmail.com>
99 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
100 * configure: Regenerated.
102 2014-02-12 Richard Biener <rguenther@suse.de>
104 * vec.c (vec_prefix::calculate_allocation): Move as
105 inline variant to vec.h.
106 (vec_prefix::calculate_allocation_1): New out-of-line version.
107 * vec.h (vec_prefix::calculate_allocation_1): Declare.
108 (vec_prefix::m_has_auto_buf): Rename to ...
109 (vec_prefix::m_using_auto_storage): ... this.
110 (vec_prefix::calculate_allocation): Inline the easy cases
111 and dispatch to calculate_allocation_1 which doesn't need the
113 (va_heap::reserve): Use gcc_checking_assert.
114 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
115 m_using_auto_storage.
116 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
118 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
119 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
120 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
122 2014-02-12 Richard Biener <rguenther@suse.de>
124 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
125 when we found a dependence.
127 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
129 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
131 (maybe_fold_stmt): ... into this new function.
132 * omp-low.c (lower_omp): Update comment.
134 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
137 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
140 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
142 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
143 identifiers in comments.
144 (cortexa53_extra_costs): Likewise.
145 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
146 (cortexa7_extra_costs): Likewise.
147 (cortexa12_extra_costs): Likewise.
148 (cortexa15_extra_costs): Likewise.
149 (v7m_extra_costs): Likewise.
151 2014-02-12 Richard Biener <rguenther@suse.de>
154 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
155 of posix_memalign being successful.
156 (lower_stmt): Restrict lowering of posix_memalign to when
157 -ftree-bit-ccp is enabled.
159 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
161 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
163 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
165 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
167 PR rtl-optimization/60116
168 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
169 other_insn once the combination has been validated.
171 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
174 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
176 * ipa-devirt.c: Include demangle.h
177 (odr_violation_reported): New static variable.
178 (add_type_duplicate): Update odr_violations.
179 (maybe_record_node): Add completep parameter; update it.
180 (record_target_from_binfo): Add COMPLETEP parameter;
182 (possible_polymorphic_call_targets_1): Likewise.
183 (struct polymorphic_call_target_d): Add nonconstruction_targets;
184 rename FINAL to COMPLETE.
185 (record_targets_from_bases): Sanity check we found the binfo;
186 fix COMPLETEP updating.
187 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
188 parameter, fix computing of COMPLETEP.
189 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
190 at LTO time do demangling.
191 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
192 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
194 (gimple_get_virt_method_for_binfo): Likewise.
195 * gimple-fold.h (gimple_get_virt_method_for_binfo,
196 gimple_get_virt_method_for_vtable): Update prototypes.
198 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
201 * genautomata.c (add_presence_absence): Fix typo with
202 {final_}presence_list.
204 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
207 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
208 for VSX/Altivec vectors that land in GPR registers.
210 2014-02-11 Richard Henderson <rth@redhat.com>
211 Jakub Jelinek <jakub@redhat.com>
214 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
215 around drhs if type conversion to lacc->type is not useless.
217 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
219 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
221 (cortex-a57.cortex-a53): Likewise.
222 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
224 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
226 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
229 2014-02-11 Renlin Li <Renlin.Li@arm.com>
231 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
232 add_options_for_arm_vfp3.
234 2014-02-11 Jeff Law <law@redhat.com>
237 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
238 object with an undesirable mode.
240 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
243 * config/i386/sol2-9.h: New file.
244 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
245 *-*-solaris2.9*): Use it.
247 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
249 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
250 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
252 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
254 * config/microblaze/microblaze.c: Extend mcpu version format
256 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
258 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
260 2014-02-10 Richard Henderson <rth@redhat.com>
263 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
264 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
265 ms-abi vs -mno-accumulate-outgoing-args.
266 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
267 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
270 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
273 * cfgexpand.c (expand_asm_operands): Attach source location to
274 ASM_INPUT rtx objects.
275 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
277 2014-02-10 Nick Clifton <nickc@redhat.com>
279 * config/mn10300/mn10300.c (popcount): New function.
280 (mn10300_expand_prologue): Include saved registers in stack usage
283 2014-02-10 Jeff Law <law@redhat.com>
286 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
287 when changing the SET_DEST of a prior insn to avoid an input reload.
289 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
291 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
292 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
293 -mcall-openbsd, or -mcall-linux.
294 (CC1_ENDIAN_BIG_SPEC): Remove.
295 (CC1_ENDIAN_LITTLE_SPEC): Remove.
296 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
297 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
298 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
299 and %cc1_endian_default.
300 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
302 2014-02-10 Richard Biener <rguenther@suse.de>
304 PR tree-optimization/60115
305 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
306 MEM_REF handling. Properly verify that the accesses are not
307 out of the objects bound.
309 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
311 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
314 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
316 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
317 proper constants and fix formatting.
318 (possible_polymorphic_call_targets): Fix formatting.
320 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
321 Ilya Tocar <ilya.tocar@intel.com>
323 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
324 (_mm512_loadu_epi32): Renamed into...
325 (_mm512_loadu_si512): This.
326 (_mm512_storeu_epi32): Renamed into...
327 (_mm512_storeu_si512): This.
328 (_mm512_maskz_ceil_ps): Removed.
329 (_mm512_maskz_ceil_pd): Ditto.
330 (_mm512_maskz_floor_ps): Ditto.
331 (_mm512_maskz_floor_pd): Ditto.
332 (_mm512_floor_round_ps): Ditto.
333 (_mm512_floor_round_pd): Ditto.
334 (_mm512_ceil_round_ps): Ditto.
335 (_mm512_ceil_round_pd): Ditto.
336 (_mm512_mask_floor_round_ps): Ditto.
337 (_mm512_mask_floor_round_pd): Ditto.
338 (_mm512_mask_ceil_round_ps): Ditto.
339 (_mm512_mask_ceil_round_pd): Ditto.
340 (_mm512_maskz_floor_round_ps): Ditto.
341 (_mm512_maskz_floor_round_pd): Ditto.
342 (_mm512_maskz_ceil_round_ps): Ditto.
343 (_mm512_maskz_ceil_round_pd): Ditto.
344 (_mm512_expand_pd): Ditto.
345 (_mm512_expand_ps): Ditto.
346 * config/i386/i386.c (ix86_builtins): Remove
347 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
349 * config/i386/predicates.md (const1256_operand): New.
350 (const_1_to_2_operand): Ditto.
351 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
352 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
353 (*avx512pf_gatherpf<mode>sf): Ditto.
354 (avx512pf_gatherpf<mode>df): Ditto.
355 (*avx512pf_gatherpf<mode>df_mask): Ditto.
356 (*avx512pf_gatherpf<mode>df): Ditto.
357 (avx512pf_scatterpf<mode>sf): Ditto.
358 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
359 (*avx512pf_scatterpf<mode>sf): Ditto.
360 (avx512pf_scatterpf<mode>df): Ditto.
361 (*avx512pf_scatterpf<mode>df_mask): Ditto.
362 (*avx512pf_scatterpf<mode>df): Ditto.
363 (avx512f_expand<mode>): Removed.
364 (<shift_insn><mode>3<mask_name>): Change predicate type.
366 2014-02-08 Jakub Jelinek <jakub@redhat.com>
368 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
369 not at the end of datarefs vector use ordered_remove to avoid
370 reordering datarefs vector.
373 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
374 mark local addressable non-static vars as GOVD_PRIVATE
375 instead of GOVD_LOCAL.
376 * omp-low.c (lower_omp_for): Move gimple_bind_vars
377 and BLOCK_VARS of gimple_bind_block to new_stmt rather
381 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
382 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
383 assume_aligned or alloc_align attributes.
384 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
385 arguments. Handle also assume_aligned and alloc_align attributes.
386 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
387 calls to functions with assume_aligned or alloc_align attributes.
388 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
390 2014-02-08 Terry Guo <terry.guo@arm.com>
392 * doc/invoke.texi: Document ARM -march=armv7e-m.
394 2014-02-08 Jakub Jelinek <jakub@redhat.com>
396 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
397 flag on __cilkrts_rethrow builtin.
400 * ipa-cp.c (determine_versionability): Fail at -O0
401 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
402 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
405 2014-02-04 Jakub Jelinek <jakub@redhat.com>
408 * tree-inline.c (copy_forbidden): Fail for
409 __attribute__((optimize (0))) functions.
411 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
413 * varpool.c: Include pointer-set.h.
414 (varpool_remove_unreferenced_decls): Variables in other partitions
415 will not be output; be however careful to not lose information
418 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
420 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
421 lookup in the vtable constructor.
423 2014-02-07 Jeff Law <law@redhat.com>
426 * config/m68k/m68k.md (ashldi_extsi): Turn into a
427 define_insn_and_split.
429 * ipa-inline.c (inline_small_functions): Fix typos.
431 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
433 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
434 (s390_can_use_return_insn): Declare.
435 * config/s390/s390.h (EPILOGUE_USES): Define.
436 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
438 (s390_chunkify_start): Handle return JUMP_LABELs.
439 (s390_early_mach): Emit a main_pool instruction on the entry edge.
440 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
441 (s390_can_use_return_insn): New functions.
442 (s390_fix_long_loop_prediction): Handle conditional returns.
443 (TARGET_SET_UP_BY_PROLOGUE): Define.
444 * config/s390/s390.md (ANY_RETURN): New code iterator.
445 (*creturn, *csimple_return, return, simple_return): New patterns.
447 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
449 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
450 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
451 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
452 REG_CFA_RESTORE list when deciding not to restore a register.
454 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
456 * config/s390/s390.c: Include tree-pass.h and context.h.
457 (s390_early_mach): New function, split out from...
458 (s390_emit_prologue): ...here.
459 (pass_data_s390_early_mach): New pass structure.
460 (pass_s390_early_mach): New class.
461 (s390_option_override): Create and register early_mach pass.
464 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
466 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
467 to match for the exit block.
469 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
471 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
472 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
473 Reject misaligned operands.
475 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
477 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
479 2014-02-07 Richard Biener <rguenther@suse.de>
482 * gimple-low.c (lower_builtin_posix_memalign): New function.
483 (lower_stmt): Call it to lower posix_memalign in a way
484 to make alignment info accessible.
486 2014-02-07 Jakub Jelinek <jakub@redhat.com>
489 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
490 __builtin_setjmp_receiver.
492 2014-02-07 Richard Biener <rguenther@suse.de>
495 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
496 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
497 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
498 Handle BUILT_IN_POSIX_MEMALIGN.
499 (find_func_clobbers): Likewise.
500 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
501 (call_may_clobber_ref_p_1): Likewise.
503 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
506 * ipa-devirt.c (record_target_from_binfo): Remove overactive
509 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
512 * lto-cgraph.c (lto_output_node): Use
513 symtab_get_symbol_partitioning_class.
514 (lto_output_varpool_node): likewise.
515 (symtab_get_symbol_partitioning_class): Move here from
517 * cgraph.h (symbol_partitioning_class): Likewise.
518 (symtab_get_symbol_partitioning_class): Declare.
520 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
522 * ggc.h (ggc_internal_cleared_alloc): New macro.
523 * vec.h (vec_safe_copy): Handle memory stats.
524 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
525 * target-globals.c (save_target_globals): Likewise.
527 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
530 * expr.c (emit_move_resolve_push): Export; be bit more selective
531 on when to clear alias set.
532 * expr.h (emit_move_resolve_push): Declare.
533 * function.h (struct function): Add tail_call_marked.
534 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
535 * config/i386/i386-protos.h (ix86_expand_push): Remove.
536 * config/i386/i386.md (TImode move expander): De not call
538 (FP push expanders): Preserve memory attributes.
539 * config/i386/sse.md (push<mode>1): Remove.
540 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
541 (ix86_expand_push): Remove.
542 * config/i386/mmx.md (push<mode>1): Remove.
544 2014-02-06 Jakub Jelinek <jakub@redhat.com>
546 PR rtl-optimization/60030
547 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
548 lopart with paradoxical subreg before shifting it up by hprec.
550 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
552 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
553 Remove extra newline at end of file.
554 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
555 (arm_issue_rate): Handle cortexa57.
556 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
557 (cortex-a57.cortex-a53): Likewise.
559 2014-02-06 Jakub Jelinek <jakub@redhat.com>
562 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
563 don't record in REG_FRAME_RELATED_EXPR registers not set in that
565 (arm_expand_prologue): Adjust all callers.
566 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
567 info, registers also at the lowest numbered registers side. Use
568 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
572 * var-tracking.c (adjust_mems): Before adding a SET to
573 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
575 2014-02-06 Alan Modra <amodra@gmail.com>
578 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
579 change SDmode to DDmode when lra_in_progress.
581 2014-02-06 Jakub Jelinek <jakub@redhat.com>
584 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
585 free_data_ref on the dr first, and before goto again also set dr
586 to the next dr. For simd_lane_access, free old datarefs[i] before
587 overwriting it. For get_vectype_for_scalar_type failure, don't
588 free_data_ref if simd_lane_access.
590 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
593 * tree.h (opts_for_fn): New inline function.
594 (opt_for_fn): Define.
595 * config/i386/i386.c (ix86_function_regparm): Use
596 opt_for_fn (decl, optimize) instead of optimize.
598 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
600 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
601 for SYMBOL_REF in large memory model.
603 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
605 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
607 (cortex-a57): Likewise.
608 (cortex-a57.cortex-a53): Likewise.
610 2014-02-06 Yury Gribov <y.gribov@samsung.com>
611 Kugan Vivekanandarajah <kuganv@linaro.org>
613 * config/arm/arm.c (arm_vector_alignment_reachable): Check
615 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
617 2014-02-06 Richard Biener <rguenther@suse.de>
619 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
620 set_loop_copy and initialize_original_copy_tables.
622 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
624 * config/aarch64/aarch64-simd.md
625 (aarch64_ashr_simddi): Change QI to SI.
627 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
628 Jakub Jelinek <jakub@redhat.com>
631 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
634 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
636 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
637 CODE_FOR_altivec_vpku[hw]um to
638 CODE_FOR_altivec_vpku[hw]um_direct.
639 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
640 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
641 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
642 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
644 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
646 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
647 generation for -maltivec=be.
648 (altivec_vsumsws): Simplify redundant test.
650 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
652 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
653 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
654 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
655 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
657 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
659 (altivec_vpks<VI_char>ss): Likewise.
660 (altivec_vpks<VI_char>us): Likewise.
661 (altivec_vpku<VI_char>us): Likewise.
662 (altivec_vpku<VI_char>um): Likewise.
663 (altivec_vpku<VI_char>um_direct): New (copy of
664 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
666 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
667 target is little endian and -maltivec=be is not specified.
668 (*altivec_vupkhs<VU_char>_direct): New (copy of
669 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
670 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
671 target is little endian and -maltivec=be is not specified.
672 (*altivec_vupkls<VU_char>_direct): New (copy of
673 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
674 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
675 little endian and -maltivec=be is not specified.
676 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
677 little endian and -maltivec=be is not specified.
679 2014-02-05 Richard Henderson <rth@redhat.com>
682 * combine-stack-adj.c: Revert r206943.
683 * sched-int.h (struct deps_desc): Add last_args_size.
684 * sched-deps.c (init_deps): Initialize it.
685 (sched_analyze_insn): Add OUTPUT dependencies between insns that
686 contain REG_ARGS_SIZE notes.
688 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
690 * lto-cgraph.c (asm_nodes_output): Make global.
691 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
692 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
693 (driver_handle_option): Handle OPT_fwpa.
695 2014-02-05 Jakub Jelinek <jakub@redhat.com>
698 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
699 a comment typo and formatting issue. If odr_hash hasn't been
700 created, return vNULL and set *completep to false.
703 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
704 bb with no successors.
706 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
709 * doc/invoke.texi (-march): Clarify documentation for ARM.
713 2014-02-05 Richard Biener <rguenther@suse.de>
715 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
716 when not vectorizing because of too many alias checks.
717 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
718 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
720 2014-02-05 Nick Clifton <nickc@redhat.com>
722 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
723 accept extended registers in any mode when compiling for the MN10300.
725 2014-02-05 Yury Gribov <y.gribov@samsung.com>
727 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
728 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
729 sanitization attributes.
730 (can_inline_edge_p): Likewise.
731 (sanitize_attrs_match_for_inline_p): New function.
733 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
735 * ipa-prop.c (detect_type_change): Shor circuit testing of
736 type changes on THIS pointer.
738 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
741 * config/pa/pa.c (legitimize_tls_address): Return original address
742 if not passed a SYMBOL_REF rtx.
743 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
745 (pa_emit_move_sequence): Simplify TLS source operands.
746 (pa_legitimate_constant_p): Reject all TLS constants.
747 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
748 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
750 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
752 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
753 groups when we know they are controlled by LTO.
754 * varasm.c (default_binds_local_p_1): If object is in other partition,
755 it will be resolved locally.
757 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
759 * config/host-linux.c (linux_gt_pch_use_address): Don't
760 use SSIZE_MAX because it is not always defined.
762 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
765 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
766 threshold for pseudo splitting.
767 (update_ebb_live_info): Process call argument hard registers and
768 hard registers from insn definition too.
769 (max_small_class_regs_num): New constant.
770 (inherit_in_ebb): Update live hard regs through EBBs. Update
771 reloads_num only for small register classes. Don't split for
774 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
777 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
780 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
782 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
785 2014-02-04 Marek Polacek <polacek@redhat.com>
787 * gdbinit.in (pel): Define.
789 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
791 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
794 2014-02-04 Richard Biener <rguenther@suse.de>
797 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
798 in function context local.
799 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
800 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
801 similar to LTO_imported_decl_ref.
803 2014-02-04 Jakub Jelinek <jakub@redhat.com>
805 PR tree-optimization/60002
806 * cgraphclones.c (build_function_decl_skip_args): Clear
809 PR tree-optimization/60023
810 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
811 false to gsi_replace.
812 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
813 has been in some EH region and vec_stmt could throw, add
814 vec_stmt into the same EH region.
815 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
816 has no lhs, ignore it.
817 * internal-fn.c (expand_MASK_LOAD): Likewise.
820 * tree-inline.c (copy_forbidden): Fail for
821 __attribute__((optimize (0))) functions.
824 * omp-low.c (simd_clone_struct_copy): If from->inbranch
825 is set, copy one less argument.
826 (expand_simd_clones): Don't subtract clone_info->inbranch
827 from simd_clone_struct_alloc argument.
829 PR rtl-optimization/57915
830 * recog.c (simplify_while_replacing): If all unary/binary/relational
831 operation arguments are constant, attempt to simplify those.
834 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
835 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
837 2014-02-04 Richard Biener <rguenther@suse.de>
839 PR tree-optimization/60012
840 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
841 TBAA disambiguation to all DDRs.
843 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
846 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
847 (LINK_SPEC): Use it for -shared, -shared-libgcc.
849 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
852 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
854 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
856 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
857 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
859 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
862 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
863 to figure out targets of polymorphic calls with known decl.
864 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
865 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
866 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
867 (get_polymorphic_call_info): ... here.
868 (get_polymorphic_call_info_from_invariant): New function.
870 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
872 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
873 lookup via vtable pointer; check for type consistency
874 and turn inconsitent facts into UNREACHABLE.
875 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
876 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
877 type inconsistent querries; return UNREACHABLE instead.
879 2014-02-03 Richard Henderson <rth@twiddle.net>
882 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
883 already processed this node.
884 (normalize_one_pred_1): Pass along mark_set.
885 (normalize_one_pred): Create and destroy a pointer_set_t.
886 (normalize_one_pred_chain): Likewise.
888 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
890 PR gcov-profile/58602
891 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
893 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
896 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
897 -fno-devirtualize; try to devirtualize by the knowledge of
898 virtual table pointer given by aggregate propagation.
899 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
900 (ipa_print_node_jump_functions): Dump also offset that
901 is relevant for polymorphic calls.
902 (determine_known_aggregate_parts): Add arg_type parameter; use it
903 instead of determining the type from pointer type.
904 (ipa_compute_jump_functions_for_edge): Update call of
905 determine_known_aggregate_parts.
906 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
907 (gimple_get_virt_method_for_binfo): ... here; simplify using
908 vtable_pointer_value_to_vtable.
909 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
910 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
911 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
912 (vtable_pointer_value_to_vtable): Break out from ...; handle also
914 (vtable_pointer_value_to_binfo): ... here.
915 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
917 2014-02-03 Teresa Johnson <tejohnson@google.com>
919 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
920 redef of outer loop index variable.
922 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
926 * doc/extend.texi (Function Attributes): Typo.
928 2014-02-03 Cong Hou <congh@google.com>
930 PR tree-optimization/60000
931 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
932 if the vectorized statement is a store. A store statement can only
933 appear at the end of pattern statements.
935 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
937 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
938 (ix86_option_override_internal): Default long double to 64-bit for
939 32-bit Bionic and to 128-bit for 64-bit Bionic.
941 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
942 TARGET_LONG_DOUBLE_128 is true.
943 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
945 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
946 (mlong-double-64): Negate -mlong-double-128.
947 (mlong-double-128): New option.
949 * config/i386/i386-c.c (ix86_target_macros): Define
950 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
952 * doc/invoke.texi: Document -mlong-double-128.
954 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
956 PR rtl-optimization/60024
957 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
959 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
961 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
963 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
965 PR rtl-optimization/57662
966 * sel-sched.c (code_motion_path_driver): Do not mark already not
967 existing blocks in the visiting bitmap.
969 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
971 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
972 on the insn being emitted.
974 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
975 Will Deacon <will.deacon@arm.com>
977 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
979 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
981 * config/arm/arm-tables.opt: Regenerate.
983 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
985 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
986 for vector types other than V16QImode.
987 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
988 define_expand, and call altivec_expand_vec_perm_le when producing
989 code with little endian element order.
990 (*altivec_vperm_<mode>_internal): New insn having previous
991 behavior of altivec_vperm_<mode>.
992 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
993 altivec_expand_vec_perm_le when producing code with little endian
995 (*altivec_vperm_<mode>_uns_internal): New insn having previous
996 behavior of altivec_vperm_<mode>_uns.
998 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1000 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1001 (altivec_vsumsws): Add handling for -maltivec=be with a little
1003 (altivec_vsumsws_direct): New.
1004 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1005 gen_altivec_vsumsws.
1007 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1009 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1010 vtable_pointer_value_to_binfo): New functions.
1011 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1012 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1014 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1016 * config/nios2/nios2.md (load_got_register): Initialize GOT
1017 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1018 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1020 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1022 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1023 preserverd by passthrough, do not propagate the type.
1025 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1027 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1028 (mips_atomic_assign_expand_fenv): New function.
1029 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1031 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1033 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1034 (__builtin_mips_set_fcsr): Likewise.
1035 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1036 MIPS_USI_FTYPE_VOID.
1037 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1038 (mips16_expand_set_fcsr): Likewise.
1039 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1040 (mips16_set_fcsr_stub): Likewise.
1041 (mips16_get_fcsr_one_only_stub): New class.
1042 (mips16_set_fcsr_one_only_stub): Likewise.
1043 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1044 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1045 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1046 (hard_float): New availability predicate.
1047 (mips_builtins): Add get_fcsr and set_fcsr.
1048 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1049 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1050 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1051 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1052 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1055 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1057 * config/mips/mips.c (mips_one_only_stub): New class.
1058 (mips_need_mips16_rdhwr_p): Replace with...
1059 (mips16_rdhwr_stub): ...this new variable.
1060 (mips16_stub_call_address): New function.
1061 (mips16_rdhwr_one_only_stub): New class.
1062 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1063 (mips_output_mips16_rdhwr): Delete.
1064 (mips_finish_stub): New function.
1065 (mips_code_end): Use it to handle rdhwr stubs.
1067 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1070 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1071 when calculating size of integer atomic types.
1073 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1075 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1077 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1079 PR tree-optimization/60003
1080 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1081 * profile.c (branch_prob): Use gimple_call_builtin_p
1082 to check for BUILT_IN_SETJMP_RECEIVER.
1083 * tree-inline.c (copy_bb): Call notice_special_calls.
1085 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1088 * lra-constraints.c (process_alt_operands): Update reload_sum only
1091 2014-01-31 Richard Henderson <rth@redhat.com>
1094 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1095 until after else_eh is processed.
1097 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1099 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1100 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1101 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1102 in smmintrin.h, remove them.
1103 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1104 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1105 * config/i386/i386.md (ROUND_SAE): Fix value.
1106 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1107 (const48_operand): New.
1108 * config/i386/subst.md (round), (round_expand): Use
1109 const_4_or_8_to_11_operand.
1110 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1112 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1114 * config/i386/constraints.md (Yk): Swap meaning with k.
1115 * config/i386/i386.md (movhi_internal): Change Yk to k.
1116 (movqi_internal): Ditto.
1117 (*k<logic><mode>): Ditto.
1120 (kandn<mode>): Ditto.
1121 (*<code>hi_1): Ditto.
1122 (*<code>qi_1): Ditto.
1123 (kxnor<mode>): Ditto.
1124 (kortestzhi): Ditto.
1125 (kortestchi): Ditto.
1127 (*one_cmplhi2_1): Ditto.
1128 (*one_cmplqi2_1): Ditto.
1129 * config/i386/sse.md (): Change k to Yk.
1130 (avx512f_load<mode>_mask): Ditto.
1131 (avx512f_blendm<mode>): Ditto.
1132 (avx512f_store<mode>_mask): Ditto.
1133 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1134 (avx512f_storedqu<mode>_mask): Ditto.
1135 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1137 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1138 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1139 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1140 (avx512f_maskcmp<mode>3): Ditto.
1141 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1142 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1143 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1144 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1145 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1146 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1147 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1148 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1149 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1150 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1151 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1152 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1153 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1154 (vec_extract_lo_<mode>_maskm): Ditto.
1155 (vec_extract_hi_<mode>_maskm): Ditto.
1156 (avx512f_vternlog<mode>_mask): Ditto.
1157 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1158 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1159 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1160 (avx512f_<code>v8div16qi2_mask): Ditto.
1161 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1162 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1163 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1164 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1165 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1166 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1167 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1168 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1169 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1170 (avx512cd_maskb_vec_dupv8di): Ditto.
1171 (avx512cd_maskw_vec_dupv16si): Ditto.
1172 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1173 (avx512f_vpermi2var<mode>3_mask): Ditto.
1174 (avx512f_vpermi2var<mode>3_mask): Ditto.
1175 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1176 (*avx512f_gathersi<mode>): Ditto.
1177 (*avx512f_gathersi<mode>_2): Ditto.
1178 (*avx512f_gatherdi<mode>): Ditto.
1179 (*avx512f_gatherdi<mode>_2): Ditto.
1180 (*avx512f_scattersi<mode>): Ditto.
1181 (*avx512f_scatterdi<mode>): Ditto.
1182 (avx512f_compress<mode>_mask): Ditto.
1183 (avx512f_compressstore<mode>_mask): Ditto.
1184 (avx512f_expand<mode>_mask): Ditto.
1185 * config/i386/subst.md (mask): Change k to Yk.
1186 (mask_scalar_merge): Ditto.
1189 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1191 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1193 2014-01-31 Richard Biener <rguenther@suse.de>
1196 * builtins.c (fold_builtin_memory_op): Make sure to not
1197 use a floating-point mode or a boolean or enumeral type for
1200 2014-01-30 DJ Delorie <dj@redhat.com>
1202 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1203 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1204 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1205 whenever main() has an epilogue.
1207 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1209 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1210 unused variable "field".
1211 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1212 (vsx_mergeh_<mode>): Likewise.
1213 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1214 (altivec_vmrghh): Likewise.
1215 (altivec_vmrghw): Likewise.
1216 (altivec_vmrglb): Likewise.
1217 (altivec_vmrglh): Likewise.
1218 (altivec_vmrglw): Likewise.
1219 (altivec_vspltb): Add missing uses.
1220 (altivec_vsplth): Likewise.
1221 (altivec_vspltw): Likewise.
1222 (altivec_vspltsf): Likewise.
1224 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1227 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1228 frame related instructions.
1230 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1232 PR rtl-optimization/59959
1233 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1234 any reload of register whose subreg is invalid.
1236 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1238 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1239 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1240 Add missing return type - void.
1242 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1244 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1245 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1246 remove element index adjustment for endian (now handled in vsx.md
1248 (altivec_expand_vec_perm_const): Use
1249 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1250 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1251 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1252 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1253 define_expand and a new define_insn *altivec_vspltb_internal;
1254 adjust for -maltivec=be on a little endian target.
1255 (altivec_vspltb_direct): New.
1256 (altivec_vsplth): Divide into a define_expand and a new
1257 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1258 little endian target.
1259 (altivec_vsplth_direct): New.
1260 (altivec_vspltw): Divide into a define_expand and a new
1261 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1262 little endian target.
1263 (altivec_vspltw_direct): New.
1264 (altivec_vspltsf): Divide into a define_expand and a new
1265 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1266 a little endian target.
1268 2014-01-30 Richard Biener <rguenther@suse.de>
1270 PR tree-optimization/59993
1271 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1272 can propagate form the earlier stmt and avoid the transform
1273 when the intermediate result is needed.
1275 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1277 * README.Portability: Fix typo.
1279 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1281 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1282 comparison_operator with ordered_comparison_operator.
1284 2014-01-30 Nick Clifton <nickc@redhat.com>
1286 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1287 Rename to mn10300_store_multiple_regs.
1288 * config/mn10300/mn10300.c: Likewise.
1289 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1290 store_multiple_regs.
1291 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1292 Call mn10300_store_multiple_regs.
1294 2014-01-30 Nick Clifton <nickc@redhat.com>
1295 DJ Delorie <dj@redhat.com>
1297 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1298 %fp 2 to keep registers after it properly word-aligned.
1299 (rl78_alloc_physical_registers_umul): Handle the case where both
1300 input operands are the same.
1302 2014-01-30 Richard Biener <rguenther@suse.de>
1304 PR tree-optimization/59903
1305 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1308 2014-01-30 Jason Merrill <jason@redhat.com>
1311 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1314 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1316 2014-01-30 Richard Biener <rguenther@suse.de>
1318 PR tree-optimization/59951
1319 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1321 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1324 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1325 SFmode to DFmode case.
1327 2014-01-29 DJ Delorie <dj@redhat.com>
1329 * config/msp430/msp430.opt (-minrt): New.
1330 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1332 (ENDFILE_SPEC): Likewise.
1334 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1336 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1337 (estimate_function_body_sizes): Use it.
1339 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1342 * dwarf2out.c (is_cxx_auto): New.
1343 (is_base_type): Use it.
1344 (gen_type_die_with_usage): Likewise.
1346 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1348 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1349 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1350 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1351 -maltivec=be with LE targets.
1352 (vsx_mergeh_<mode>): Likewise.
1353 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1354 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1355 (altivec_vmrghb): Replace with define_expand and new
1356 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1357 (altivec_vmrghb_direct): New define_insn.
1358 (altivec_vmrghh): Replace with define_expand and new
1359 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1360 (altivec_vmrghh_direct): New define_insn.
1361 (altivec_vmrghw): Replace with define_expand and new
1362 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1363 (altivec_vmrghw_direct): New define_insn.
1364 (*altivec_vmrghsf): Adjust for endianness.
1365 (altivec_vmrglb): Replace with define_expand and new
1366 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1367 (altivec_vmrglb_direct): New define_insn.
1368 (altivec_vmrglh): Replace with define_expand and new
1369 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1370 (altivec_vmrglh_direct): New define_insn.
1371 (altivec_vmrglw): Replace with define_expand and new
1372 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1373 (altivec_vmrglw_direct): New define_insn.
1374 (*altivec_vmrglsf): Adjust for endianness.
1375 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1376 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1377 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1378 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1379 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1380 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1381 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1382 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1384 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1386 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1387 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1390 2014-01-29 Richard Biener <rguenther@suse.de>
1392 PR tree-optimization/58742
1393 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1394 associate_pointerplus_align.
1395 (associate_pointerplus_diff): New function.
1396 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1397 and associate_pointerplus_diff.
1399 2014-01-29 Richard Biener <rguenther@suse.de>
1401 * lto-streamer.h (LTO_major_version): Bump to 3.
1402 (LTO_minor_version): Reset to 0.
1404 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1406 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1407 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1408 (arm_file_start): Generate correct asm header for armv7ve.
1409 * config/arm/bpabi.h: Add multilib support for armv7ve.
1410 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1411 and cortex-a15 to armv7ve.
1412 * config/arm/t-aprofile: Add multilib support for armv7ve.
1413 * doc/invoke.texi: Document -march=armv7ve.
1415 2014-01-29 Richard Biener <rguenther@suse.de>
1417 PR tree-optimization/58742
1418 * tree-ssa-forwprop.c (associate_plusminus): Return true
1419 if we changed sth, defer EH cleanup to ...
1420 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1421 (simplify_mult): New function.
1423 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1426 PR tree-optimization/59920
1427 * tree.c (build_common_builtin_nodes): Remove
1428 __builtin_setjmp_dispatcher initialization.
1429 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1430 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1431 instead of gsi_after_labels + manually skipping debug stmts.
1432 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1433 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1434 * tree-inline.c (copy_edges_for_bb): Remove
1435 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1436 argument. Ignore computed_goto_p stmts. Don't call
1437 make_abnormal_goto_edges. If a call might need abnormal edges
1438 for non-local gotos, see if it already has an edge to
1439 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1440 with true argument, don't do anything then, otherwise add
1441 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1442 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1444 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1445 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1446 (lower_stmt): Don't set data->calls_builtin_setjmp.
1447 (lower_builtin_setjmp): Adjust comment.
1448 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1449 * tree-cfg.c (found_computed_goto): Remove.
1450 (factor_computed_gotos): Remove.
1451 (make_goto_expr_edges): Return bool, true for computed gotos.
1452 Don't call make_abnormal_goto_edges.
1453 (build_gimple_cfg): Don't set found_computed_goto, don't call
1454 factor_computed_gotos.
1455 (computed_goto_p): No longer static.
1456 (make_blocks): Don't set found_computed_goto.
1457 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1458 (make_edges): If make_goto_expr_edges returns true, push bb
1459 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1460 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1461 vector. Record mapping between bbs and OpenMP regions if there
1462 are any, adjust make_gimple_omp_edges caller. Call
1463 handle_abnormal_edges.
1464 (make_abnormal_goto_edges): Remove.
1465 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1466 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1467 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1468 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1469 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1470 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1471 filling *region also set *region_idx to (*region)->entry->index.
1474 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1475 For REGs set ORIGINAL_REGNO.
1477 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1479 * doc/md.texi: Mention that a target shouldn't implement
1480 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1483 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1485 PR tree-optimization/59594
1486 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1487 a copy of the datarefs vector rather than the vector itself.
1489 2014-01-28 Jason Merrill <jason@redhat.com>
1492 * dwarf2out.c (auto_die): New static.
1493 (gen_type_die_with_usage): Handle C++1y 'auto'.
1494 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1497 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1500 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1501 (SPEC_X32): Likewise.
1502 (SPEC_64): Likewise.
1503 * config/i386/i386.c (ix86_option_override_internal): Turn off
1504 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1506 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1507 * config/i386/i386.h (TARGET_16BIT): New macro.
1508 (TARGET_16BIT_P): Likewise.
1509 * config/i386/i386.opt: Add m16.
1510 * doc/invoke.texi: Document -m16.
1512 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1514 PR preprocessor/59935
1515 * input.c (location_get_source_line): Bail out on when line number
1516 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1518 2014-01-28 Richard Biener <rguenther@suse.de>
1520 PR tree-optimization/58742
1521 * tree-ssa-forwprop.c (associate_plusminus): Handle
1522 pointer subtraction of the form (T)(P + A) - (T)P.
1524 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1526 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1529 2014-01-28 Richard Biener <rguenther@suse.de>
1532 2014-01-28 Richard Biener <rguenther@suse.de>
1534 PR rtl-optimization/45364
1535 PR rtl-optimization/59890
1536 * var-tracking.c (local_get_addr_clear_given_value): Handle
1537 already cleared slot.
1538 (val_reset): Handle not allocated local_get_addr_cache.
1539 (vt_find_locations): Use post-order on the inverted CFG.
1541 2014-01-28 Richard Biener <rguenther@suse.de>
1543 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1545 2014-01-28 Richard Biener <rguenther@suse.de>
1547 PR rtl-optimization/45364
1548 PR rtl-optimization/59890
1549 * var-tracking.c (local_get_addr_clear_given_value): Handle
1550 already cleared slot.
1551 (val_reset): Handle not allocated local_get_addr_cache.
1552 (vt_find_locations): Use post-order on the inverted CFG.
1554 2014-01-28 Alan Modra <amodra@gmail.com>
1556 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1557 * configure.ac <recursive call for build != host>: Define
1558 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1559 and LD_FOR_BUILD too.
1560 * configure: Regenerate.
1562 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1564 * config/i386/i386.c (get_builtin_code_for_version): Separate
1565 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1566 Broadwell from Haswell.
1568 2014-01-27 Steve Ellcey <sellcey@mips.com>
1570 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1571 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1572 * config/mips/mips.c (mips_option_override): Change setting
1574 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1575 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1576 Change from Mask to Var.
1578 2014-01-27 Jeff Law <law@redhat.com>
1580 * ipa-inline.c (inline_small_functions): Fix typo.
1582 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1584 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1585 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1586 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1587 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1588 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1589 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1590 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1591 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1592 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1593 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1594 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1595 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1596 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1597 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1598 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1599 (_mm512_storeu_epi64): Ditto.
1600 (_mm512_cmpge_epi32_mask): Ditto.
1601 (_mm512_cmpge_epu32_mask): Ditto.
1602 (_mm512_cmpge_epi64_mask): Ditto.
1603 (_mm512_cmpge_epu64_mask): Ditto.
1604 (_mm512_cmple_epi32_mask): Ditto.
1605 (_mm512_cmple_epu32_mask): Ditto.
1606 (_mm512_cmple_epi64_mask): Ditto.
1607 (_mm512_cmple_epu64_mask): Ditto.
1608 (_mm512_cmplt_epi32_mask): Ditto.
1609 (_mm512_cmplt_epu32_mask): Ditto.
1610 (_mm512_cmplt_epi64_mask): Ditto.
1611 (_mm512_cmplt_epu64_mask): Ditto.
1612 (_mm512_cmpneq_epi32_mask): Ditto.
1613 (_mm512_cmpneq_epu32_mask): Ditto.
1614 (_mm512_cmpneq_epi64_mask): Ditto.
1615 (_mm512_cmpneq_epu64_mask): Ditto.
1616 (_mm512_expand_pd): Ditto.
1617 (_mm512_expand_ps): Ditto.
1618 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1619 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1620 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1621 * config/i386/i386.c (ix86_builtins): Add
1622 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1623 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1624 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1625 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1626 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1627 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1628 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1629 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1630 IX86_BUILTIN_PMOVUSQW512_MEM.
1631 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1632 __builtin_ia32_pmovsqd512mem_mask,
1633 __builtin_ia32_pmovqd512mem_mask,
1634 __builtin_ia32_pmovusqw512mem_mask,
1635 __builtin_ia32_pmovsqw512mem_mask,
1636 __builtin_ia32_pmovqw512mem_mask,
1637 __builtin_ia32_pmovusdw512mem_mask,
1638 __builtin_ia32_pmovsdw512mem_mask,
1639 __builtin_ia32_pmovdw512mem_mask,
1640 __builtin_ia32_pmovqb512mem_mask,
1641 __builtin_ia32_pmovusqb512mem_mask,
1642 __builtin_ia32_pmovsqb512mem_mask,
1643 __builtin_ia32_pmovusdb512mem_mask,
1644 __builtin_ia32_pmovsdb512mem_mask,
1645 __builtin_ia32_pmovdb512mem_mask.
1646 (bdesc_args): Add __builtin_ia32_expanddf512,
1647 __builtin_ia32_expandsf512.
1648 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1649 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1650 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1651 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1652 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1653 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1654 (avx512f_<code>v8div16qi2_mask_store): This.
1655 (avx512f_expand<mode>): New.
1657 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1659 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1661 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1662 (_mm512_prefetch_i32scatter_pd): Ditto.
1663 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1664 (_mm512_prefetch_i64scatter_pd): Ditto.
1665 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1666 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1667 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1668 (_mm512_prefetch_i32scatter_ps): Ditto.
1669 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1670 (_mm512_prefetch_i64scatter_ps): Ditto.
1671 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1672 * config/i386/i386-builtin-types.def: Define
1673 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1674 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1675 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1676 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1677 IX86_BUILTIN_SCATTERPFQPD.
1678 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1679 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1680 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1681 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1682 __builtin_ia32_scatterpfqps.
1683 (ix86_expand_builtin): Expand new built-ins.
1684 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1685 fix memory access data type.
1686 (*avx512pf_gatherpf<mode>_mask): Ditto.
1687 (*avx512pf_gatherpf<mode>): Ditto.
1688 (avx512pf_scatterpf<mode>): Ditto.
1689 (*avx512pf_scatterpf<mode>_mask): Ditto.
1690 (*avx512pf_scatterpf<mode>): Ditto.
1691 (GATHER_SCATTER_SF_MEM_MODE): New.
1692 (avx512pf_gatherpf<mode>df): Ditto.
1693 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1694 (*avx512pf_scatterpf<mode>df): Ditto.
1696 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1699 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1700 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1703 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1705 * common/config/arm/arm-common.c
1706 (arm_rewrite_mcpu): Handle multiple names.
1708 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1710 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1712 * gimple-builder.h (create_gimple_tmp): Delete.
1714 2014-01-27 Christian Bruel <christian.bruel@st.com>
1716 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1719 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1721 * config/pa/pa.md (call): Generate indirect long calls to non-local
1722 functions when outputing 32-bit code.
1723 (call_value): Likewise except for special call to buggy powf function.
1725 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1726 portable runtime and PIC indirect calls.
1727 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1728 and PIC call sequences. Use ldo instead of blr to set return register
1729 in PIC call sequence.
1731 2014-01-25 Walter Lee <walt@tilera.com>
1733 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1734 avoid clobbering a live register.
1736 2014-01-25 Walter Lee <walt@tilera.com>
1738 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1739 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1740 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1741 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1743 2014-01-25 Walter Lee <walt@tilera.com>
1745 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1746 arguments on even registers.
1747 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1749 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1750 (BIGGEST_ALIGNMENT): Ditto.
1751 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1753 2014-01-25 Walter Lee <walt@tilera.com>
1755 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1756 insns before bundling.
1757 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1759 2014-01-25 Walter Lee <walt@tilera.com>
1761 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1762 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1763 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1765 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1767 * config/mips/constraints.md (kl): Delete.
1768 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1769 define expands, using...
1770 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1771 instructions for MIPS16.
1772 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1773 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1775 2014-01-25 Walter Lee <walt@tilera.com>
1777 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1781 2014-01-25 Walter Lee <walt@tilera.com>
1783 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1784 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1786 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1788 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1791 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1793 * print-rtl.c (in_call_function_usage): New var.
1794 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1795 EXPR_LIST mode as mode and not as reg note name.
1798 * cfgloopmanip.c (copy_loop_info): If
1799 loop->warned_aggressive_loop_optimizations, make sure
1800 the flag is set in target loop too.
1802 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1804 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1806 * builtins.def: Likewise.
1807 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1808 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1809 * ira.c (ira_setup_eliminable_regset): Likewise.
1810 * omp-low.c (gate_expand_omp): Likewise.
1811 (execute_lower_omp): Likewise.
1812 (diagnose_sb_0): Likewise.
1813 (gate_diagnose_omp_blocks): Likewise.
1814 (simd_clone_clauses_extract): Likewise.
1817 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1819 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1820 correction for little endian...
1821 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1824 2014-01-24 Jeff Law <law@redhat.com>
1826 PR tree-optimization/59919
1827 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1828 for non-returning calls.
1830 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1832 * common/config/aarch64/aarch64-common.c
1833 (aarch64_rewrite_mcpu): Handle multiple names.
1834 * config/aarch64/aarch64.h
1835 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1837 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1839 * input.c (add_file_to_cache_tab): Handle the case where fopen
1842 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1845 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1846 from push operand if code of push isn't PRE_DEC.
1848 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1851 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1852 -mquad-memory-atomic. Update -mquad-memory documentation to say
1853 it is only used for non-atomic loads/stores.
1855 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1856 -mquad-memory or -mquad-memory-atomic switches.
1858 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1859 -mquad-memory-atomic to ISA 2.07 support.
1861 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1862 to separate support of normal quad word memory operations (ldq, stq)
1863 from the atomic quad word memory operations.
1865 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1866 support to separate non-atomic quad word operations from atomic
1867 quad word operations. Disable non-atomic quad word operations in
1868 little endian mode so that we don't have to swap words after the
1869 load and before the store.
1870 (quad_load_store_p): Add comment about atomic quad word support.
1871 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1872 options printed with -mdebug=reg.
1874 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1875 -mquad-memory-atomic as the test for whether we have quad word
1876 atomic instructions.
1877 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1878 or -mp8-vector are used, allow byte/half-word atomic operations.
1880 * config/rs6000/sync.md (load_lockedti): Insure that the address
1881 is a proper indexed or indirect address for the lqarx instruction.
1882 On little endian systems, swap the hi/lo registers after the lqarx
1884 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1885 insure the address is valid for the lqarx instruction.
1886 (store_conditionalti): Insure that the address is a proper indexed
1887 or indirect address for the stqcrx. instruction. On little endian
1888 systems, swap the hi/lo registers before doing the stqcrx.
1890 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1891 insure the address is valid for the stqcrx. instruction.
1893 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1894 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1895 type of quad memory support is available.
1897 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1900 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1901 there is a danger of looping.
1903 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1905 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1906 force flag_ira_loop_pressure if set via command line.
1908 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1910 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1911 (ashr_simd): New builtin handling DI mode.
1912 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1913 (aarch64_sshr_simddi): New match pattern.
1914 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1915 (vshrd_n_s64): Likewise.
1916 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1918 2014-01-23 Nick Clifton <nickc@redhat.com>
1920 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1921 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1922 favour of mcu specific scripts.
1923 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1926 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1927 Alex Velenko <Alex.Velenko@arm.com>
1929 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1930 (vaddv_s16): Likewise.
1931 (vaddv_s32): Likewise.
1932 (vaddv_u8): Likewise.
1933 (vaddv_u16): Likewise.
1934 (vaddv_u32): Likewise.
1935 (vaddvq_s8): Likewise.
1936 (vaddvq_s16): Likewise.
1937 (vaddvq_s32): Likewise.
1938 (vaddvq_s64): Likewise.
1939 (vaddvq_u8): Likewise.
1940 (vaddvq_u16): Likewise.
1941 (vaddvq_u32): Likewise.
1942 (vaddvq_u64): Likewise.
1943 (vaddv_f32): Likewise.
1944 (vaddvq_f32): Likewise.
1945 (vaddvq_f64): Likewise.
1946 (vmaxv_f32): Likewise.
1947 (vmaxv_s8): Likewise.
1948 (vmaxv_s16): Likewise.
1949 (vmaxv_s32): Likewise.
1950 (vmaxv_u8): Likewise.
1951 (vmaxv_u16): Likewise.
1952 (vmaxv_u32): Likewise.
1953 (vmaxvq_f32): Likewise.
1954 (vmaxvq_f64): Likewise.
1955 (vmaxvq_s8): Likewise.
1956 (vmaxvq_s16): Likewise.
1957 (vmaxvq_s32): Likewise.
1958 (vmaxvq_u8): Likewise.
1959 (vmaxvq_u16): Likewise.
1960 (vmaxvq_u32): Likewise.
1961 (vmaxnmv_f32): Likewise.
1962 (vmaxnmvq_f32): Likewise.
1963 (vmaxnmvq_f64): Likewise.
1964 (vminv_f32): Likewise.
1965 (vminv_s8): Likewise.
1966 (vminv_s16): Likewise.
1967 (vminv_s32): Likewise.
1968 (vminv_u8): Likewise.
1969 (vminv_u16): Likewise.
1970 (vminv_u32): Likewise.
1971 (vminvq_f32): Likewise.
1972 (vminvq_f64): Likewise.
1973 (vminvq_s8): Likewise.
1974 (vminvq_s16): Likewise.
1975 (vminvq_s32): Likewise.
1976 (vminvq_u8): Likewise.
1977 (vminvq_u16): Likewise.
1978 (vminvq_u32): Likewise.
1979 (vminnmv_f32): Likewise.
1980 (vminnmvq_f32): Likewise.
1981 (vminnmvq_f64): Likewise.
1983 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1985 * config/aarch64/aarch64-simd.md
1986 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1987 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1988 (*aarch64_mul3_elt<mode>): Likewise.
1989 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1990 (*aarch64_mul3_elt_to_64v2df): Likewise.
1991 (*aarch64_mla_elt<mode>): Likewise.
1992 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1993 (*aarch64_mls_elt<mode>): Likewise.
1994 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1995 (*aarch64_fma4_elt<mode>): Likewise.
1996 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1997 (*aarch64_fma4_elt_to_64v2df): Likewise.
1998 (*aarch64_fnma4_elt<mode>): Likewise.
1999 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2000 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2001 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2002 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2003 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2004 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2005 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2006 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2007 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2009 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2011 * config/aarch64/aarch64-simd.md
2012 (aarch64_be_checked_get_lane<mode>): New define_expand.
2013 * config/aarch64/aarch64-simd-builtins.def
2014 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2015 New builtin definition.
2016 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2017 Use new safe be builtin.
2019 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2021 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2023 (aarch64_be_st1<mode>): Likewise.
2024 (aarch_ld1<VALL:mode>): Define_expand modified.
2025 (aarch_st1<VALL:mode>): Likewise.
2026 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2027 (UNSPEC_ST1): Likewise.
2029 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2031 * config/microblaze/microblaze.md: Add trap insn and attribute
2033 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2035 PR preprocessor/58580
2036 * input.h (location_get_source_line): Take an additional line_size
2038 (void diagnostics_file_cache_fini): Declare new function.
2039 * input.c (struct fcache): New type.
2040 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2041 New static constants.
2042 (diagnostic_file_cache_init, total_lines_num)
2043 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2044 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2045 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2046 (get_next_line, read_next_line, goto_next_line, read_line_num):
2047 New static function definitions.
2048 (diagnostic_file_cache_fini): New function.
2049 (location_get_source_line): Take an additional output line_len
2050 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2052 * diagnostic.c (diagnostic_finish): Call
2053 diagnostic_file_cache_fini.
2054 (adjust_line): Take an additional input parameter for the length
2055 of the line, rather than calculating it with strlen.
2056 (diagnostic_show_locus): Adjust the use of
2057 location_get_source_line and adjust_line with respect to their new
2058 signature. While displaying a line now, do not stop at the first
2059 null byte. Rather, display the zero byte as a space and keep
2060 going until we reach the size of the line.
2061 * Makefile.in: Add vec.o to OBJS-libcommon
2063 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2064 Ilya Tocar <ilya.tocar@intel.com>
2066 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2067 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2068 (__builtin_ia32_kmov16): Ditto.
2069 * config/i386/i386.md (UNSPEC_KMOV): New.
2072 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2074 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2075 (_mm512_storeu_si512): Ditto.
2077 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2080 * rtl.h (get_referenced_operands): Declare.
2081 * recog.c (get_referenced_operands): New function.
2082 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2083 operands have been referenced when recording LO_SUM references.
2085 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2087 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2089 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2091 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2092 Enable for generic and recent AMD targets.
2094 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2096 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2097 ARG_SIZE note when adjustment was eliminated.
2099 2014-01-22 Jeff Law <law@redhat.com>
2101 PR tree-optimization/59597
2102 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2103 in file. Accept new argument REGISTERING and use it to modify
2104 dump output appropriately.
2105 (register_jump_thread): Corresponding changes.
2106 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2107 thread paths involving joiner blocks. Add code to dump cancelled
2108 jump threading paths.
2110 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2112 PR rtl-optimization/59477
2113 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2114 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2116 2014-01-22 Tom Tromey <tromey@redhat.com>
2118 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2120 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2122 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2124 PR rtl-optimization/59896
2125 * lra-constraints.c (process_alt_operands): Check unused note for
2126 matched operands of insn with no output reloads.
2128 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2130 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2131 (mips_move_from_gpr_cost): Likewise.
2133 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2135 PR rtl-optimization/59858
2136 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2137 ira_class_hard_regs_num.
2138 (process_alt_operands): Increase reject for dying matched operand.
2140 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2143 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2144 smaller than size, perform several stores or loads and stores
2145 at dst + count - size to store or copy all of size bytes, rather
2146 than just last modesize bytes.
2148 2014-01-20 DJ Delorie <dj@redhat.com>
2150 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2151 that CLOBBERs are REGs before propogating their values.
2153 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2156 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2157 (cgraph_inline_failed_type): New function.
2158 * cgraph.h (DEFCIFCODE): Add type.
2159 (cgraph_inline_failed_type_t): New enum.
2160 (cgraph_inline_failed_type): New prototype.
2161 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2162 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2163 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2164 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2165 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2166 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2167 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2168 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2169 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2170 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2171 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2172 OPTIMIZATION_MISMATCH.
2173 * tree-inline.c (expand_call_inline): Emit errors during
2174 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2176 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2179 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2180 mode attribute in insn output.
2182 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2184 * output.h (output_constant): Delete.
2185 * varasm.c (output_constant): Make private.
2187 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2189 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2191 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2194 * tree.h (fold_builtin_strcat): New prototype.
2195 * builtins.c (fold_builtin_strcat): No longer static. Add len
2196 argument, if non-NULL, don't call c_strlen. Optimize
2197 directly into __builtin_memcpy instead of __builtin_strcpy.
2198 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2199 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2201 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2203 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2204 for SImode_address_operand operands, having only a REG argument.
2206 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2208 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2209 loader name using mbig-endian.
2210 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2212 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2214 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2218 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2220 * config/aarch64/aarch64-protos.h
2221 (aarch64_cannot_change_mode_class_ptr): Declare.
2222 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2223 aarch64_cannot_change_mode_class_ptr): New.
2224 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2225 backend hook aarch64_cannot_change_mode_class.
2227 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2229 * common/config/aarch64/aarch64-common.c
2230 (aarch64_handle_option): Don't handle any option order logic here.
2231 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2232 selected_cpu, warn on architecture version mismatch.
2233 (aarch64_override_options): Fix parsing order for option strings.
2235 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2236 Iain Sandoe <iain@codesourcery.com>
2239 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2240 warning. Amend comment to reflect current functionality.
2242 2014-01-20 Richard Biener <rguenther@suse.de>
2245 * builtins.c (fold_builtin_strcat): Remove case better handled
2246 by tree-ssa-strlen.c.
2248 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2250 * config/aarch64/aarch64.opt
2251 (mcpu, march, mtune): Make case-insensitive.
2253 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2256 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2257 if operands[1] is a REG or ZERO_EXTEND of a REG.
2259 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2261 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2263 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2265 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2266 long non-pic millicode calls.
2268 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2270 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2272 2014-01-19 Kito Cheng <kito@0xlab.org>
2274 * builtins.c (expand_movstr): Check movstr expand done or fail.
2276 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2277 H.J. Lu <hongjiu.lu@intel.com>
2280 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2281 to DImode for zero-extended addresses.
2283 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2285 PR rtl-optimization/57763
2286 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2287 on the new indirect jump_insn and increment LABEL_NUSES (label).
2289 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2293 * config.gcc (x86_archs): New variable.
2294 (x86_64_archs): Likewise.
2295 (x86_cpus): Likewise.
2296 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2297 --with-arch/--with-cpu= options.
2298 Support --with-arch=/--with-cpu={nehalem,westmere,
2299 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2301 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2303 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2304 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2306 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2308 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2310 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2313 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2314 clear cpp_get_options (parse_in)->warn_unused_macros for
2315 ix86_target_macros_internal with cpp_define.
2317 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2319 * jump.c (delete_related_insns): Keep (use (insn))s.
2320 * reorg.c (redundant_insn): Check for barriers too.
2322 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2324 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2326 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2328 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2329 call to $$dyncall when TARGET_LONG_CALLS is true.
2331 2014-01-17 Jeff Law <law@redhat.com>
2333 * ree.c (combine_set_extension): Temporarily disable test for
2334 changing number of hard registers.
2336 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2339 * ipa-inline-analysis.c (inline_free_summary):
2340 Do not free summary of aliases.
2342 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2345 * gimplify.c (gimplify_expr): Use create_tmp_var
2346 instead of create_tmp_var_raw. If cond doesn't have
2347 integral type, don't add the IFN_ANNOTATE builtin at all.
2349 2014-01-17 Martin Jambor <mjambor@suse.cz>
2352 * ipa-cp.c (prev_edge_clone): New variable.
2353 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2354 Also resize prev_edge_clone vector.
2355 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2356 (ipcp_edge_removal_hook): New function.
2357 (ipcp_driver): Register ipcp_edge_removal_hook.
2359 2014-01-17 Andrew Pinski <apinski@cavium.com>
2360 Steve Ellcey <sellcey@mips.com>
2363 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2366 2014-01-17 Jeff Law <law@redhat.com>
2369 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2370 so that pass_ccp runs first.
2372 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2374 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2375 (ix86_adjust_cost): Use !TARGET_XXX.
2376 (do_reorder_for_imul): Likewise.
2377 (swap_top_of_ready_list): Likewise.
2378 (ix86_sched_reorder): Likewise.
2380 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2382 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2383 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2384 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2385 (intel_memset): New. Duplicate slm_memset.
2386 (intel_cost): New. Duplicate slm_cost.
2387 (m_INTEL): New macro.
2388 (processor_target_table): Add "intel".
2389 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2390 with PROCESSOR_INTEL for "intel".
2391 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2392 PROCESSOR_SILVERMONT.
2393 (ix86_issue_rate): Likewise.
2394 (ix86_adjust_cost): Likewise.
2395 (ia32_multipass_dfa_lookahead): Likewise.
2396 (swap_top_of_ready_list): Likewise.
2397 (ix86_sched_reorder): Likewise.
2398 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2399 instead of TARGET_OPT_AGU.
2400 * config/i386/i386.h (TARGET_INTEL): New.
2401 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2402 (processor_type): Add PROCESSOR_INTEL.
2403 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2404 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2406 2014-01-17 Marek Polacek <polacek@redhat.com>
2409 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2412 2014-01-17 Richard Biener <rguenther@suse.de>
2414 PR tree-optimization/46590
2415 * opts.c (default_options_table): Add entries for
2416 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2417 all enabled at -O1 but not for -Og.
2418 * common.opt (fbranch-count-reg): Remove Init(1).
2419 (fmove-loop-invariants): Likewise.
2420 (ftree-pta): Likewise.
2422 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2424 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2425 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2426 decls to at least the GCC 4.8 used alignments.
2429 * tree-nested.c (convert_nonlocal_reference_stmt,
2430 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2431 of GIMPLE_BIND stmts, adjust associated decls.
2433 2014-01-17 Richard Biener <rguenther@suse.de>
2435 PR tree-optimization/46590
2436 * vec.h (vec<>::bseach): New member function implementing
2437 binary search according to C89 bsearch.
2438 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2439 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2440 bitmap pointer again. Make accesses_in_loop a flat array.
2441 (mem_ref_obstack): New global.
2442 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2443 (mark_ref_stored): Likewise.
2444 (ref_indep_loop_p_2): Likewise.
2445 (set_ref_stored_in_loop): New helper function.
2446 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2447 (memref_free): Adjust.
2448 (record_mem_ref_loc): Simplify.
2449 (gather_mem_refs_stmt): Adjust.
2450 (sort_locs_in_loop_postorder_cmp): New function.
2451 (analyze_memory_references): Sort accesses_in_loop after
2452 loop postorder number.
2453 (find_ref_loc_in_loop_cmp): New function.
2454 (for_all_locs_in_loop): Find relevant cluster of locs in
2455 accesses_in_loop and iterate without recursion.
2456 (execute_sm): Avoid uninit warning.
2457 (struct ref_always_accessed): Simplify.
2458 (ref_always_accessed::operator ()): Likewise.
2459 (ref_always_accessed_p): Likewise.
2460 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2461 loop postorder numbers here.
2462 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2465 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2468 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2469 on decls for which assemble_alias has been called.
2471 2014-01-17 Nick Clifton <nickc@redhat.com>
2473 * config/msp430/msp430.opt: (mcpu): New option.
2474 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2475 (msp430_option_override): Parse target_cpu. If the MCU name
2476 matches a generic string, clear target_mcu.
2477 (msp430_attr): Allow numeric interrupt values up to 63.
2478 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2479 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2481 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2483 * config/msp430/msp430.md (popm): Use %J rather than %I.
2484 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2485 (addhi_cy_i): Use immediate_operand for operand 2.
2486 * doc/invoke.texi: Document -mcpu option.
2488 2014-01-17 Richard Biener <rguenther@suse.de>
2490 PR rtl-optimization/38518
2491 * df.h (df_analyze_loop): Declare.
2492 * df-core.c: Include cfgloop.h.
2493 (df_analyze_1): Split out main part of df_analyze.
2494 (df_analyze): Adjust.
2495 (loop_inverted_post_order_compute): New function.
2496 (loop_post_order_compute): Likewise.
2497 (df_analyze_loop): New function avoiding whole-function
2499 * loop-invariant.c (find_defs): Use df_analyze_loop.
2500 (find_invariants): Adjust.
2501 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2503 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2505 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2506 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2508 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2510 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2511 traversal when removing references.
2513 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2516 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2518 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2521 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2522 pushing a reload for an autoinc when we had previously reloaded an
2523 inner part of the address.
2525 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2527 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2529 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2530 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2531 when not giving up or versioning for alias only because of
2533 (vect_analyze_data_ref_dependences): Set to true.
2534 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2536 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2537 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2541 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2544 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2545 operand 0 predicate for gathers, use a new pseudo as subtarget.
2547 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2550 * lra-constraints.c (process_alt_operands): Add printing debug info.
2551 Check absence of input/output reloads for matched operands too.
2553 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2555 PR rtl-optimization/59835
2556 * ira.c (ira_init_register_move_cost): Increase cost for
2559 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2561 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2563 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2566 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2567 non-register objects. Use gen_(high/low)part more consistently.
2570 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2573 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2574 endian support, remove tests for WORDS_BIG_ENDIAN.
2575 (p8_mfvsrd_3_<mode>): Likewise.
2576 (reload_gpr_from_vsx<mode>): Likewise.
2577 (reload_gpr_from_vsxsf): Likewise.
2578 (p8_mfvsrd_4_disf): Likewise.
2580 2014-01-16 Richard Biener <rguenther@suse.de>
2582 PR rtl-optimization/46590
2583 * lcm.c (compute_antinout_edge): Use postorder iteration.
2584 (compute_laterin): Use inverted postorder iteration.
2586 2014-01-16 Nick Clifton <nickc@redhat.com>
2589 * varasm.c (output_constant): Return the number of bytes actually
2591 (output_constructor_array_range): Update the field size with the
2592 number of bytes emitted by output_constant.
2593 (output_constructor_regular_field): Likewise. Also do not
2594 complain if the total number of bytes emitted is now greater
2595 than the expected fieldpos.
2596 * output.h (output_constant): Update prototype and descriptive comment.
2598 2014-01-16 Marek Polacek <polacek@redhat.com>
2601 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2602 it is error_mark_node.
2604 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2606 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2607 VALID_AVX256_REG_OR_OI_MODE.
2609 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2611 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2612 current procedure should be profiled.
2614 2014-01-15 Andrew Pinski <apinski@cavium.com>
2616 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2617 of moving from/to the STACK_REG register class.
2619 2014-01-15 Richard Henderson <rth@redhat.com>
2622 * reginfo.c (global_regs_decl): Globalize.
2623 * rtl.h (global_regs_decl): Declare.
2624 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2625 reserved via global_regs.
2627 2014-01-15 Teresa Johnson <tejohnson@google.com>
2629 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2631 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2633 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2634 and vmulosh rather than call gen_vec_widen_smult_*.
2635 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2636 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2637 (vec_widen_smult_even_v16qi): Likewise.
2638 (vec_widen_umult_even_v8hi): Likewise.
2639 (vec_widen_smult_even_v8hi): Likewise.
2640 (vec_widen_umult_odd_v16qi): Likewise.
2641 (vec_widen_smult_odd_v16qi): Likewise.
2642 (vec_widen_umult_odd_v8hi): Likewise.
2643 (vec_widen_smult_odd_v8hi): Likewise.
2644 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2645 vmuloub rather than call gen_vec_widen_umult_*.
2646 (vec_widen_umult_lo_v16qi): Likewise.
2647 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2648 vmulosb rather than call gen_vec_widen_smult_*.
2649 (vec_widen_smult_lo_v16qi): Likewise.
2650 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2651 rather than call gen_vec_widen_umult_*.
2652 (vec_widen_umult_lo_v8hi): Likewise.
2653 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2654 rather than call gen_vec_widen_smult_*.
2655 (vec_widen_smult_lo_v8hi): Likewise.
2657 2014-01-15 Jeff Law <law@redhat.com>
2659 PR tree-optimization/59747
2660 * ree.c (find_and_remove_re): Properly handle case where a second
2661 eliminated extension requires widening a copy created for elimination
2662 of a prior extension.
2663 (combine_set_extension): Ensure that the number of hard regs needed
2664 for a destination register does not change when we widen it.
2666 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2668 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2669 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2670 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2671 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2672 (avr-*-rtems*): Likewise.
2673 (bfin*-rtems*): Likewise.
2674 (moxie-*-rtems*): Likewise.
2675 (h8300-*-rtems*): Likewise.
2676 (i[34567]86-*-rtems*): Likewise.
2677 (lm32-*-rtems*): Likewise.
2678 (m32r-*-rtems*): Likewise.
2679 (m68k-*-rtems*): Likewise.
2680 (microblaze*-*-rtems*): Likewise.
2681 (mips*-*-rtems*): Likewise.
2682 (powerpc-*-rtems*): Likewise.
2683 (sh-*-rtems*): Likewise.
2684 (sparc-*-rtems*): Likewise.
2685 (sparc64-*-rtems*): Likewise.
2686 (v850-*-rtems*): Likewise.
2687 (m32c-*-rtems*): Likewise.
2689 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2691 PR rtl-optimization/59511
2692 * ira.c (ira_init_register_move_cost): Use memory costs for some
2693 cases of register move cost calculations.
2694 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2695 instead of BB frequency.
2696 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2697 * lra-assigns.c (find_hard_regno_for): Ditto.
2699 2014-01-15 Richard Biener <rguenther@suse.de>
2701 PR tree-optimization/59822
2702 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2703 (vectorizable_load): Use it to hoist defs of uses of invariant
2704 loads out of the loop.
2706 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2707 Kugan Vivekanandarajah <kuganv@linaro.org>
2710 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2713 2014-01-15 Richard Biener <rguenther@suse.de>
2715 PR rtl-optimization/59802
2716 * lcm.c (compute_available): Use inverted postorder to seed
2717 the initial worklist.
2719 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2722 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2723 ADDR_REGS for invalid symrefs in non-PIC code.
2725 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2728 * builtins.c (determine_block_size): Initialize *probable_max_size
2729 even if len_rtx is CONST_INT.
2731 2014-01-14 Andrew Pinski <apinski@cavium.com>
2733 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2734 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2735 (cortexa53_tunings): Likewise.
2736 (aarch64_sched_issue_rate): New function.
2737 (TARGET_SCHED_ISSUE_RATE): Define.
2739 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2741 * ira-costs.c (find_costs_and_classes): Add missed
2742 ira_init_register_move_cost_if_necessary.
2744 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2747 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2749 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2752 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2753 to indicate if type is used for function return value. Warn ABI
2754 change if the vector mode isn't available for function return value.
2755 (ix86_function_arg_advance): Pass false to type_natural_mode.
2756 (ix86_function_arg): Likewise.
2757 (ix86_gimplify_va_arg): Likewise.
2758 (function_arg_32): Don't warn ABI change.
2759 (ix86_function_value): Pass true to type_natural_mode.
2760 (ix86_return_in_memory): Likewise.
2761 (ix86_struct_value_rtx): Removed.
2762 (TARGET_STRUCT_VALUE_RTX): Likewise.
2764 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2766 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2767 converting a conditional jump into a conditional return.
2769 2014-01-14 Richard Biener <rguenther@suse.de>
2771 PR tree-optimization/58921
2772 PR tree-optimization/59006
2773 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2774 hoisting invariant stmts.
2775 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2776 invariant loads on the preheader edge if possible.
2778 2014-01-14 Joey Ye <joey.ye@arm.com>
2780 * doc/plugin.texi (Building GCC plugins): Update to C++.
2782 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2784 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2785 (_mm_rcp28_round_ss): Ditto.
2786 (_mm_rsqrt28_round_sd): Ditto.
2787 (_mm_rsqrt28_round_ss): Ditto.
2788 (_mm_rcp28_sd): Ditto.
2789 (_mm_rcp28_ss): Ditto.
2790 (_mm_rsqrt28_sd): Ditto.
2791 (_mm_rsqrt28_ss): Ditto.
2792 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2793 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2794 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2795 (IX86_BUILTIN_RCP28SD): Ditto.
2796 (IX86_BUILTIN_RCP28SS): Ditto.
2797 (IX86_BUILTIN_RSQRT28SD): Ditto.
2798 (IX86_BUILTIN_RSQRT28SS): Ditto.
2799 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2800 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2801 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2802 (ix86_expand_special_args_builtin): Expand new FTYPE.
2803 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2804 (srcp14<mode>): Make insn unary.
2805 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2806 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2807 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2808 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2809 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2810 Fix rounding: make it SAE only.
2811 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2813 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2815 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2816 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2817 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2818 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2819 (round_saeonly_mask_scalar_operand4): Ditto.
2820 (round_saeonly_mask_scalar_op3): Ditto.
2821 (round_saeonly_mask_scalar_op4): Ditto.
2823 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2825 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2826 Implement -maltivec=be for vec_insert and vec_extract.
2828 2014-01-10 DJ Delorie <dj@redhat.com>
2830 * config/msp430/msp430.md (call_internal): Don't allow memory
2831 references with SP as the base register.
2832 (call_value_internal): Likewise.
2833 * config/msp430/constraints.md (Yc): New. For memory references
2834 that don't use SP as a base register.
2836 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2837 "an integer without a # prefix"
2838 * config/msp430/msp430.md (epilogue_helper): Use it.
2840 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2843 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2844 AVX512F gather builtins.
2845 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2846 on gather decls with INTEGER_TYPE masktype.
2847 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2848 directly into the builtin rather than hoisting it before loop.
2850 PR tree-optimization/59387
2851 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2852 (scev_const_prop): If folded_casts and type has undefined overflow,
2853 use force_gimple_operand instead of force_gimple_operand_gsi and
2854 for each added stmt if it is assign with
2855 arith_code_with_undefined_signed_overflow, call
2856 rewrite_to_defined_overflow.
2857 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2858 gimple-fold.h instead.
2859 (arith_code_with_undefined_signed_overflow,
2860 rewrite_to_defined_overflow): Moved to ...
2861 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2862 rewrite_to_defined_overflow): ... here. No longer static.
2863 Include gimplify-me.h.
2864 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2865 rewrite_to_defined_overflow): New prototypes.
2867 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2869 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2871 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2873 * builtins.c (get_object_alignment_2): Minor tweak.
2874 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2876 2014-01-13 Christian Bruel <christian.bruel@st.com>
2878 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2879 optimized non constant lengths.
2881 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2884 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2885 load as __atomic_load_N if possible.
2887 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2889 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2891 (rs6000_expand_builtin): Adjust call.
2893 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2896 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2897 * config/rs6000/rs6000.c: Include target-globals.h.
2898 (rs6000_set_current_function): Instead of doing target_reinit
2899 unconditionally, use save_target_globals_default_opts and
2900 restore_target_globals.
2902 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2904 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2905 (rs6000_expand_builtin): Handle mffs and mtfsf.
2906 (rs6000_init_builtins): Define mffs and mtfsf.
2907 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2908 (rs6000_mffs): New pattern.
2909 (rs6000_mtfsf): New pattern.
2911 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2913 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2914 Start narrowing with START. Apply candidate-use pair
2915 and check overall cost in narrowing.
2916 (iv_ca_prune): Pass new argument.
2918 2014-01-10 Jeff Law <law@redhat.com>
2921 * ree.c (combine_reaching_defs): Ensure the defining statement
2922 occurs before the extension when optimizing extensions with
2923 different source and destination hard registers.
2925 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2928 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2929 vtables into the type inheritance graph.
2931 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2933 PR rtl-optimization/59754
2934 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2935 modes in the REGNO != REGNO case.
2937 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2939 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2941 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2943 PR tree-optimization/59745
2944 * tree-predcom.c (tree_predictive_commoning_loop): Call
2945 free_affine_expand_cache if giving up because components is NULL.
2947 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2948 GC in payload of target_globals struct instead of allocating them on
2949 the heap and the larger structs separately using GC.
2950 * target-globals.h (struct target_globals): Make regs, hard_regs,
2951 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2952 of GTY((skip)) and change type to void *.
2953 (reset_target_globals): Cast loads from those fields to corresponding
2956 2014-01-10 Steve Ellcey <sellcey@mips.com>
2959 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2960 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2961 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2963 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2966 * aarch64-modes.def (CC_Zmode): New flags mode.
2967 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2968 represents an equality.
2969 (aarch64_get_condition_code): Handle CC_Zmode.
2970 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2972 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2974 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2975 extraction in good case.
2977 2014-01-10 Richard Biener <rguenther@suse.de>
2979 PR tree-optimization/59374
2980 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2981 checking after SLP discovery. Mark stmts not participating
2982 in any SLP instance properly.
2984 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2986 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2987 when handling a SET rtx.
2989 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2991 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2992 (cortex-a57): Likewise.
2993 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2995 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2997 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2998 non-iwmmxt builtins.
3000 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3004 * ipa-devirt.c record_target_from_binfo): Take as argument
3005 stack of binfos and lookup matching one for virtual inheritance.
3006 (possible_polymorphic_call_targets_1): Update.
3008 2014-01-10 Huacai Chen <chenhc@lemote.com>
3010 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3011 kernel strings for Loongson-2E/2F/3A.
3013 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3016 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3017 is_gimple_call before calling gimple_call_internal_p.
3019 2014-01-09 Steve Ellcey <sellcey@mips.com>
3021 * Makefile.in (TREE_FLOW_H): Remove.
3022 (TREE_SSA_H): Add file names from tree-flow.h.
3023 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3024 * tree.h: Remove tree-flow.h reference.
3025 * hash-table.h: Remove tree-flow.h reference.
3026 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3027 reference with tree-ssa-loop.h.
3029 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3031 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3032 default element-order behavior for -maltivec.
3033 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3035 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3036 when targeting big endian, at least for now.
3037 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3039 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3042 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3043 var satisfies use_register_for_decl, just take into account type
3044 alignment, rather than decl alignment.
3046 PR tree-optimization/59622
3047 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3048 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3049 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3050 Don't devirtualize for inplace at all. For targets.length () == 1,
3051 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3053 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3055 * config/i386/i386.md (cpu): Remove the unused btver1.
3057 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3059 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3061 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3064 * tree-core.h (struct target_globals): New forward declaration.
3065 (struct tree_target_option): Add globals field.
3066 * tree.h (TREE_TARGET_GLOBALS): Define.
3067 (prepare_target_option_nodes_for_pch): New prototype.
3068 * target-globals.h (struct target_globals): Define even if
3070 * tree.c (prepare_target_option_node_for_pch,
3071 prepare_target_option_nodes_for_pch): New functions.
3072 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3073 * config/i386/i386.c: Include target-globals.h.
3074 (ix86_set_current_function): Instead of doing target_reinit
3075 unconditionally, use save_target_globals_default_opts and
3076 restore_target_globals.
3078 2014-01-09 Richard Biener <rguenther@suse.de>
3080 PR tree-optimization/59715
3081 * tree-cfg.h (split_critical_edges): Declare.
3082 * tree-cfg.c (split_critical_edges): Export.
3083 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3085 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3087 * cfgexpand.c (expand_stack_vars): Optionally disable
3088 asan stack protection.
3089 (expand_used_vars): Likewise.
3090 (partition_stack_vars): Likewise.
3091 * asan.c (asan_emit_stack_protection): Optionally disable
3092 after return stack usage.
3093 (instrument_derefs): Optionally disable memory access instrumentation.
3094 (instrument_builtin_call): Likewise.
3095 (instrument_strlen_call): Likewise.
3096 (asan_protect_global): Optionally disable global variables protection.
3097 * doc/invoke.texi: Added doc for new options.
3098 * params.def: Added new options.
3099 * params.h: Likewise.
3101 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3103 PR rtl-optimization/59724
3104 * ifcvt.c (cond_exec_process_if_block): Don't call
3105 flow_find_head_matching_sequence with 0 longest_match.
3106 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3107 non-active insns if !stop_after.
3108 (try_head_merge_bb): Revert 2014-01-07 changes.
3110 2014-01-08 Jeff Law <law@redhat.com>
3112 * ree.c (get_sub_rtx): New function, extracted from...
3113 (merge_def_and_ext): Here.
3114 (combine_reaching_defs): Use get_sub_rtx.
3116 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3118 * cgraph.h (varpool_variable_node): Do not choke on null node.
3120 2014-01-08 Catherine Moore <clm@codesourcery.com>
3122 * config/mips/mips.md (simple_return): Attempt to use JRC
3124 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3126 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3128 PR rtl-optimization/59137
3129 * reorg.c (steal_delay_list_from_target): Call update_block for
3131 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3133 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3135 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3136 two duplicate entries.
3138 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3141 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3143 * config/mips/mips.c (mips_truncated_op_cost): New function.
3144 (mips_rtx_costs): Adjust test for BADDU.
3145 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3147 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3149 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3150 (*baddu_si): ...this new pattern.
3152 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3155 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3157 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3160 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3162 (expand_expr, expand_normal): Adjust.
3163 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3164 inner_reference_p. Use inner_reference_p to expand inner references.
3165 (store_expr): Adjust.
3166 * cfgexpand.c (expand_call_stmt): Adjust.
3168 2014-01-08 Rong Xu <xur@google.com>
3170 * gcov-io.c (gcov_var): Move from gcov-io.h.
3171 (gcov_position): Ditto.
3172 (gcov_is_error): Ditto.
3173 (gcov_rewrite): Ditto.
3174 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3175 only part to libgcc/libgcov.h.
3177 2014-01-08 Marek Polacek <polacek@redhat.com>
3180 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3182 2014-01-08 Marek Polacek <polacek@redhat.com>
3185 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3187 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3189 PR rtl-optimization/59649
3190 * stor-layout.c (get_mode_bounds): For BImode return
3191 0 and STORE_FLAG_VALUE.
3193 2014-01-08 Richard Biener <rguenther@suse.de>
3196 * gimple.h (is_gimple_builtin_call): Remove.
3197 (gimple_builtin_call_types_compatible_p): New.
3198 (gimple_call_builtin_p): New overload.
3199 * gimple.c (is_gimple_builtin_call): Remove.
3200 (validate_call): Rename to ...
3201 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3203 (validate_type): New static function.
3204 (gimple_call_builtin_p): New overload and adjust.
3205 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3206 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3207 (gimple_fold_stmt_to_constant_1): Likewise.
3208 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3210 2014-01-08 Richard Biener <rguenther@suse.de>
3213 * gimplify.c (gimplify_expr): Gimplify register-register type
3214 VIEW_CONVERT_EXPRs to separate stmts.
3216 2014-01-07 Jeff Law <law@redhat.com>
3219 * ree.c (combine_set_extension): Handle case where source
3220 and destination registers in an extension insn are different.
3221 (combine_reaching_defs): Allow source and destination registers
3222 in extension to be different under limited circumstances.
3223 (add_removable_extension): Remove restriction that the
3224 source and destination registers in the extension are the same.
3225 (find_and_remove_re): Emit a copy from the extension's
3226 destination to its source after the defining insn if
3227 the source and destination registers are different.
3230 * ifcvt.c (merge_if_block): If we are merging a block with more than
3231 one successor with a block with no successors, remove any BARRIER
3232 after the second block.
3234 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3236 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3238 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3241 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3242 for 14-bit register offsets when INT14_OK_STRICT is false.
3244 2014-01-07 Roland Stigge <stigge@antcom.de>
3245 Michael Meissner <meissner@linux.vnet.ibm.com>
3248 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3249 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3251 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3253 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3256 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3258 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3259 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3260 rtx is const0_rtx or not.
3262 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3265 * target-globals.c (save_target_globals): Remove this_fn_optab
3267 * toplev.c: Include optabs.h.
3268 (target_reinit): Temporarily restore the global options if another
3269 set of options are in force.
3271 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3273 PR rtl-optimization/58668
3274 * cfgcleanup.c (flow_find_cross_jump): Don't count
3275 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3276 to determine what is counted.
3277 (flow_find_head_matching_sequence): Use active_insn_p to determine
3279 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3281 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3282 determine what is counted.
3284 PR tree-optimization/59643
3285 * tree-predcom.c (split_data_refs_to_components): If one dr is
3286 read and one write, determine_offset fails and the write isn't
3287 in the bad component, just put the read into the bad component.
3289 2014-01-07 Mike Stump <mikestump@comcast.net>
3290 Jakub Jelinek <jakub@redhat.com>
3293 * tree-core.h (struct tree_optimization_option): Change optabs
3294 type from unsigned char * to void *.
3295 * optabs.c (init_tree_optimization_optabs): Adjust
3296 TREE_OPTIMIZATION_OPTABS initialization.
3298 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3301 * config/i386/i386.h (struct machine_function): Add
3302 no_drap_save_restore field.
3303 * config/i386/i386.c (ix86_save_reg): Use
3304 !cfun->machine->no_drap_save_restore instead of
3305 crtl->stack_realign_needed.
3306 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3307 this function clears frame_pointer_needed. Set
3308 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3309 and DRAP reg is needed.
3311 2014-01-06 Marek Polacek <polacek@redhat.com>
3314 * doc/implement-c.texi: Mention that other integer types are
3315 permitted as bit-field types in strictly conforming mode.
3317 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3319 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3322 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3324 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3326 2014-01-06 Martin Jambor <mjambor@suse.cz>
3329 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3331 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3333 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3337 * var-tracking.c (add_stores): Preserve the value of the source even if
3338 we don't record the store.
3340 2014-01-06 Terry Guo <terry.guo@arm.com>
3342 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3344 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3347 * config/darwin.c (darwin_function_section): Adjust return values to
3348 correspond to optimisation changes made in r206070.
3350 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3352 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3353 from prefetch_block tune setting.
3354 (nocona_cost): Correct size of prefetch block to 64.
3356 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3358 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3359 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3360 used to save the static chain register in the computation of the offset
3361 from which the FP registers need to be restored.
3363 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3365 PR tree-optimization/59519
3366 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3367 ICE if get_current_def (current_new_name) is already non-NULL, as long
3368 as it is a phi result of some other phi in *new_exit_bb that has
3371 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3372 or vmovdqu* for misaligned_operand.
3373 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3374 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3375 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3376 aligned_mem for AVX512F masked aligned load and store builtins and for
3379 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3381 PR tree-optimization/59651
3382 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3383 Address range for negative step should be added by TYPE_SIZE_UNIT.
3385 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3387 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3388 overlapping registers also for registers other than the stack pointer.
3390 2014-01-03 Marek Polacek <polacek@redhat.com>
3393 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3396 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3399 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3402 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3403 (push splitter): Use <P:MODE_SIZE> instead of
3404 GET_MODE_SIZE (<P:MODE>mode).
3405 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3406 (mov -1, reg peephole2): Likewise.
3407 * config/i386/sse.md (*mov<mode>_internal,
3408 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3409 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3410 *<code><mode>3, *andnot<mode>3<mask_name>,
3411 <mask_codefor><code><mode>3<mask_name>): Likewise.
3412 * config/i386/subst.md (mask_mode512bit_condition,
3413 sd_mask_mode512bit_condition): Likewise.
3415 2014-01-02 Xinliang David Li <davidxl@google.com>
3417 PR tree-optimization/59303
3418 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3419 (dump_predicates): Better output format.
3420 (pred_equal_p): New function.
3421 (is_neq_relop_p): Ditto.
3422 (is_neq_zero_form_p): Ditto.
3423 (pred_expr_equal_p): Ditto.
3424 (pred_neg_p): Ditto.
3425 (simplify_pred): Ditto.
3426 (simplify_preds_2): Ditto.
3427 (simplify_preds_3): Ditto.
3428 (simplify_preds_4): Ditto.
3429 (simplify_preds): Ditto.
3431 (push_to_worklist): Ditto.
3432 (get_pred_info_from_cmp): Ditto.
3433 (is_degenerated_phi): Ditto.
3434 (normalize_one_pred_1): Ditto.
3435 (normalize_one_pred): Ditto.
3436 (normalize_one_pred_chain): Ditto.
3437 (normalize_preds): Ditto.
3438 (normalize_cond_1): Remove function.
3439 (normalize_cond): Ditto.
3440 (is_gcond_subset_of): Ditto.
3441 (is_subset_of_any): Ditto.
3442 (is_or_set_subset_of): Ditto.
3443 (is_and_set_subset_of): Ditto.
3444 (is_norm_cond_subset_of): Ditto.
3445 (pred_chain_length_cmp): Ditto.
3446 (convert_control_dep_chain_into_preds): Type change.
3447 (find_predicates): Ditto.
3448 (find_def_preds): Ditto.
3449 (destroy_predicates_vecs): Ditto.
3450 (find_matching_predicates_in_rest_chains): Ditto.
3451 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3452 (is_pred_expr_subset): Ditto.
3453 (is_pred_chain_subset_of): Ditto.
3454 (is_included_in): Ditto.
3455 (is_superset_of): Ditto.
3457 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3459 Update copyright years.
3461 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3463 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3464 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3465 config/arc/arc.md, config/arc/arc.opt,
3466 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3467 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3468 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3469 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3470 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3471 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3472 vtable-verify.c, vtable-verify.h: Use the standard form for the
3475 2014-01-02 Tobias Burnus <burnus@net-b.de>
3477 * gcc.c (process_command): Update copyright notice dates.
3478 * gcov-dump.c: Ditto.
3480 * doc/cpp.texi: Bump @copying's copyright year.
3481 * doc/cppinternals.texi: Ditto.
3482 * doc/gcc.texi: Ditto.
3483 * doc/gccint.texi: Ditto.
3484 * doc/gcov.texi: Ditto.
3485 * doc/install.texi: Ditto.
3486 * doc/invoke.texi: Ditto.
3488 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3490 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3492 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3494 * config/i386/sse.md (*mov<mode>_internal): Guard
3495 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3497 PR rtl-optimization/59647
3498 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3499 new_rtx into UNSIGNED_FLOAT rtxes.
3501 Copyright (C) 2014 Free Software Foundation, Inc.
3503 Copying and distribution of this file, with or without modification,
3504 are permitted in any medium without royalty provided the copyright
3505 notice and this notice are preserved.