1 2014-02-14 Jeff Law <law@redhat.com>
3 PR rtl-optimization/60131
4 * ree.c (get_extended_src_reg): New function.
5 (combine_reaching_defs): Use it rather than assuming location
7 (find_and_remove_re): Verify first operand of extension is
8 a REG before adding the insns to the copy list.
10 2014-02-14 Roland McGrath <mcgrathr@google.com>
12 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
13 * configure: Regenerated.
14 * config.in: Regenerated.
15 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
18 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
19 Richard Earnshaw <rearnsha@arm.com>
21 PR rtl-optimization/59535
22 * lra-constraints.c (process_alt_operands): Encourage alternative
23 when unassigned pseudo class is superset of the alternative class.
24 (inherit_reload_reg): Don't inherit when optimizing for code size.
25 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
26 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
27 modes not less than 4 for Thumb1.
29 2014-02-14 Kyle McMartin <kyle@redhat.com>
32 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
34 2014-02-14 Richard Biener <rguenther@suse.de>
36 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
37 (get_frame_arg): Drop the assert with langhook types_compatible_p.
38 Do not strip INDIRECT_REFs.
40 2014-02-14 Richard Biener <rguenther@suse.de>
43 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
44 DECL_FUNCTION_SPECIFIC_TARGET.
45 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
46 * tree-streamer-out.c (pack_ts_target_option): Remove.
47 (streamer_pack_tree_bitfields): Do not stream
49 (write_ts_function_decl_tree_pointers): Do not stream
50 DECL_FUNCTION_SPECIFIC_TARGET.
51 * tree-streamer-in.c (unpack_ts_target_option): Remove.
52 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
53 (lto_input_ts_function_decl_tree_pointers): Do not stream
54 DECL_FUNCTION_SPECIFIC_TARGET.
56 2014-02-14 Jakub Jelinek <jakub@redhat.com>
58 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
60 (get_initial_def_for_induction, vectorizable_induction): Ignore
61 debug stmts when looking for exit_phi.
62 (vectorizable_live_operation): Fix up condition.
64 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
66 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
67 nreverse() because it changes the content of original tree list.
69 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
71 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
72 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
74 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
76 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
79 2014-02-13 Jakub Jelinek <jakub@redhat.com>
82 * dwarf2out.c (gen_subprogram_die): Don't call
83 add_calling_convention_attribute if subr_die is old_die.
85 2014-02-13 Sharad Singhai <singhai@google.com>
87 * doc/optinfo.texi: Fix order of nodes.
89 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
91 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
92 operands[2], not operands[3].
94 2014-02-13 Richard Biener <rguenther@suse.de>
97 * doc/install.texi (ISL): Update recommended version to 0.12.2,
98 mention the possibility of an in-tree build.
99 (CLooG): Update recommended version to 0.18.1, mention the
100 possibility of an in-tree build and clarify that the ISL
101 bundled with CLooG does not work.
103 2014-02-13 Jakub Jelinek <jakub@redhat.com>
106 * expr.c (compress_float_constant): If x is a hard register,
107 extend into a pseudo and then move to x.
109 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
111 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
112 caused by bad second argument to warning_at() with -mhotpatch and
113 nested functions (e.g. with gfortran).
115 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
117 * opts.c (option_name): Remove "enabled by default" rider.
119 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
121 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
123 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
124 Uros Bizjak <ubizjak@gmail.com>
127 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
128 * configure: Regenerated.
130 2014-02-12 Richard Biener <rguenther@suse.de>
132 * vec.c (vec_prefix::calculate_allocation): Move as
133 inline variant to vec.h.
134 (vec_prefix::calculate_allocation_1): New out-of-line version.
135 * vec.h (vec_prefix::calculate_allocation_1): Declare.
136 (vec_prefix::m_has_auto_buf): Rename to ...
137 (vec_prefix::m_using_auto_storage): ... this.
138 (vec_prefix::calculate_allocation): Inline the easy cases
139 and dispatch to calculate_allocation_1 which doesn't need the
141 (va_heap::reserve): Use gcc_checking_assert.
142 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
143 m_using_auto_storage.
144 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
146 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
147 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
148 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
150 2014-02-12 Richard Biener <rguenther@suse.de>
152 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
153 when we found a dependence.
155 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
157 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
159 (maybe_fold_stmt): ... into this new function.
160 * omp-low.c (lower_omp): Update comment.
162 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
165 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
168 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
170 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
171 identifiers in comments.
172 (cortexa53_extra_costs): Likewise.
173 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
174 (cortexa7_extra_costs): Likewise.
175 (cortexa12_extra_costs): Likewise.
176 (cortexa15_extra_costs): Likewise.
177 (v7m_extra_costs): Likewise.
179 2014-02-12 Richard Biener <rguenther@suse.de>
182 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
183 of posix_memalign being successful.
184 (lower_stmt): Restrict lowering of posix_memalign to when
185 -ftree-bit-ccp is enabled.
187 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
189 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
191 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
193 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
195 PR rtl-optimization/60116
196 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
197 other_insn once the combination has been validated.
199 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
202 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
204 * ipa-devirt.c: Include demangle.h
205 (odr_violation_reported): New static variable.
206 (add_type_duplicate): Update odr_violations.
207 (maybe_record_node): Add completep parameter; update it.
208 (record_target_from_binfo): Add COMPLETEP parameter;
210 (possible_polymorphic_call_targets_1): Likewise.
211 (struct polymorphic_call_target_d): Add nonconstruction_targets;
212 rename FINAL to COMPLETE.
213 (record_targets_from_bases): Sanity check we found the binfo;
214 fix COMPLETEP updating.
215 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
216 parameter, fix computing of COMPLETEP.
217 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
218 at LTO time do demangling.
219 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
220 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
222 (gimple_get_virt_method_for_binfo): Likewise.
223 * gimple-fold.h (gimple_get_virt_method_for_binfo,
224 gimple_get_virt_method_for_vtable): Update prototypes.
226 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
229 * genautomata.c (add_presence_absence): Fix typo with
230 {final_}presence_list.
232 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
235 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
236 for VSX/Altivec vectors that land in GPR registers.
238 2014-02-11 Richard Henderson <rth@redhat.com>
239 Jakub Jelinek <jakub@redhat.com>
242 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
243 around drhs if type conversion to lacc->type is not useless.
245 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
247 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
249 (cortex-a57.cortex-a53): Likewise.
250 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
252 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
254 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
257 2014-02-11 Renlin Li <Renlin.Li@arm.com>
259 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
260 add_options_for_arm_vfp3.
262 2014-02-11 Jeff Law <law@redhat.com>
265 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
266 object with an undesirable mode.
268 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
271 * config/i386/sol2-9.h: New file.
272 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
273 *-*-solaris2.9*): Use it.
275 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
277 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
278 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
280 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
282 * config/microblaze/microblaze.c: Extend mcpu version format
284 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
286 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
288 2014-02-10 Richard Henderson <rth@redhat.com>
291 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
292 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
293 ms-abi vs -mno-accumulate-outgoing-args.
294 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
295 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
298 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
301 * cfgexpand.c (expand_asm_operands): Attach source location to
302 ASM_INPUT rtx objects.
303 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
305 2014-02-10 Nick Clifton <nickc@redhat.com>
307 * config/mn10300/mn10300.c (popcount): New function.
308 (mn10300_expand_prologue): Include saved registers in stack usage
311 2014-02-10 Jeff Law <law@redhat.com>
314 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
315 when changing the SET_DEST of a prior insn to avoid an input reload.
317 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
319 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
320 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
321 -mcall-openbsd, or -mcall-linux.
322 (CC1_ENDIAN_BIG_SPEC): Remove.
323 (CC1_ENDIAN_LITTLE_SPEC): Remove.
324 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
325 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
326 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
327 and %cc1_endian_default.
328 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
330 2014-02-10 Richard Biener <rguenther@suse.de>
332 PR tree-optimization/60115
333 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
334 MEM_REF handling. Properly verify that the accesses are not
335 out of the objects bound.
337 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
339 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
342 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
344 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
345 proper constants and fix formatting.
346 (possible_polymorphic_call_targets): Fix formatting.
348 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
349 Ilya Tocar <ilya.tocar@intel.com>
351 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
352 (_mm512_loadu_epi32): Renamed into...
353 (_mm512_loadu_si512): This.
354 (_mm512_storeu_epi32): Renamed into...
355 (_mm512_storeu_si512): This.
356 (_mm512_maskz_ceil_ps): Removed.
357 (_mm512_maskz_ceil_pd): Ditto.
358 (_mm512_maskz_floor_ps): Ditto.
359 (_mm512_maskz_floor_pd): Ditto.
360 (_mm512_floor_round_ps): Ditto.
361 (_mm512_floor_round_pd): Ditto.
362 (_mm512_ceil_round_ps): Ditto.
363 (_mm512_ceil_round_pd): Ditto.
364 (_mm512_mask_floor_round_ps): Ditto.
365 (_mm512_mask_floor_round_pd): Ditto.
366 (_mm512_mask_ceil_round_ps): Ditto.
367 (_mm512_mask_ceil_round_pd): Ditto.
368 (_mm512_maskz_floor_round_ps): Ditto.
369 (_mm512_maskz_floor_round_pd): Ditto.
370 (_mm512_maskz_ceil_round_ps): Ditto.
371 (_mm512_maskz_ceil_round_pd): Ditto.
372 (_mm512_expand_pd): Ditto.
373 (_mm512_expand_ps): Ditto.
374 * config/i386/i386.c (ix86_builtins): Remove
375 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
377 * config/i386/predicates.md (const1256_operand): New.
378 (const_1_to_2_operand): Ditto.
379 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
380 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
381 (*avx512pf_gatherpf<mode>sf): Ditto.
382 (avx512pf_gatherpf<mode>df): Ditto.
383 (*avx512pf_gatherpf<mode>df_mask): Ditto.
384 (*avx512pf_gatherpf<mode>df): Ditto.
385 (avx512pf_scatterpf<mode>sf): Ditto.
386 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
387 (*avx512pf_scatterpf<mode>sf): Ditto.
388 (avx512pf_scatterpf<mode>df): Ditto.
389 (*avx512pf_scatterpf<mode>df_mask): Ditto.
390 (*avx512pf_scatterpf<mode>df): Ditto.
391 (avx512f_expand<mode>): Removed.
392 (<shift_insn><mode>3<mask_name>): Change predicate type.
394 2014-02-08 Jakub Jelinek <jakub@redhat.com>
396 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
397 not at the end of datarefs vector use ordered_remove to avoid
398 reordering datarefs vector.
401 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
402 mark local addressable non-static vars as GOVD_PRIVATE
403 instead of GOVD_LOCAL.
404 * omp-low.c (lower_omp_for): Move gimple_bind_vars
405 and BLOCK_VARS of gimple_bind_block to new_stmt rather
409 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
410 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
411 assume_aligned or alloc_align attributes.
412 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
413 arguments. Handle also assume_aligned and alloc_align attributes.
414 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
415 calls to functions with assume_aligned or alloc_align attributes.
416 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
418 2014-02-08 Terry Guo <terry.guo@arm.com>
420 * doc/invoke.texi: Document ARM -march=armv7e-m.
422 2014-02-08 Jakub Jelinek <jakub@redhat.com>
424 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
425 flag on __cilkrts_rethrow builtin.
428 * ipa-cp.c (determine_versionability): Fail at -O0
429 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
430 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
433 2014-02-04 Jakub Jelinek <jakub@redhat.com>
436 * tree-inline.c (copy_forbidden): Fail for
437 __attribute__((optimize (0))) functions.
439 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
441 * varpool.c: Include pointer-set.h.
442 (varpool_remove_unreferenced_decls): Variables in other partitions
443 will not be output; be however careful to not lose information
446 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
448 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
449 lookup in the vtable constructor.
451 2014-02-07 Jeff Law <law@redhat.com>
454 * config/m68k/m68k.md (ashldi_extsi): Turn into a
455 define_insn_and_split.
457 * ipa-inline.c (inline_small_functions): Fix typos.
459 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
461 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
462 (s390_can_use_return_insn): Declare.
463 * config/s390/s390.h (EPILOGUE_USES): Define.
464 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
466 (s390_chunkify_start): Handle return JUMP_LABELs.
467 (s390_early_mach): Emit a main_pool instruction on the entry edge.
468 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
469 (s390_can_use_return_insn): New functions.
470 (s390_fix_long_loop_prediction): Handle conditional returns.
471 (TARGET_SET_UP_BY_PROLOGUE): Define.
472 * config/s390/s390.md (ANY_RETURN): New code iterator.
473 (*creturn, *csimple_return, return, simple_return): New patterns.
475 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
477 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
478 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
479 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
480 REG_CFA_RESTORE list when deciding not to restore a register.
482 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
484 * config/s390/s390.c: Include tree-pass.h and context.h.
485 (s390_early_mach): New function, split out from...
486 (s390_emit_prologue): ...here.
487 (pass_data_s390_early_mach): New pass structure.
488 (pass_s390_early_mach): New class.
489 (s390_option_override): Create and register early_mach pass.
492 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
494 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
495 to match for the exit block.
497 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
499 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
500 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
501 Reject misaligned operands.
503 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
505 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
507 2014-02-07 Richard Biener <rguenther@suse.de>
510 * gimple-low.c (lower_builtin_posix_memalign): New function.
511 (lower_stmt): Call it to lower posix_memalign in a way
512 to make alignment info accessible.
514 2014-02-07 Jakub Jelinek <jakub@redhat.com>
517 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
518 __builtin_setjmp_receiver.
520 2014-02-07 Richard Biener <rguenther@suse.de>
523 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
524 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
525 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
526 Handle BUILT_IN_POSIX_MEMALIGN.
527 (find_func_clobbers): Likewise.
528 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
529 (call_may_clobber_ref_p_1): Likewise.
531 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
534 * ipa-devirt.c (record_target_from_binfo): Remove overactive
537 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
540 * lto-cgraph.c (lto_output_node): Use
541 symtab_get_symbol_partitioning_class.
542 (lto_output_varpool_node): likewise.
543 (symtab_get_symbol_partitioning_class): Move here from
545 * cgraph.h (symbol_partitioning_class): Likewise.
546 (symtab_get_symbol_partitioning_class): Declare.
548 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
550 * ggc.h (ggc_internal_cleared_alloc): New macro.
551 * vec.h (vec_safe_copy): Handle memory stats.
552 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
553 * target-globals.c (save_target_globals): Likewise.
555 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
558 * expr.c (emit_move_resolve_push): Export; be bit more selective
559 on when to clear alias set.
560 * expr.h (emit_move_resolve_push): Declare.
561 * function.h (struct function): Add tail_call_marked.
562 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
563 * config/i386/i386-protos.h (ix86_expand_push): Remove.
564 * config/i386/i386.md (TImode move expander): De not call
566 (FP push expanders): Preserve memory attributes.
567 * config/i386/sse.md (push<mode>1): Remove.
568 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
569 (ix86_expand_push): Remove.
570 * config/i386/mmx.md (push<mode>1): Remove.
572 2014-02-06 Jakub Jelinek <jakub@redhat.com>
574 PR rtl-optimization/60030
575 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
576 lopart with paradoxical subreg before shifting it up by hprec.
578 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
580 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
581 Remove extra newline at end of file.
582 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
583 (arm_issue_rate): Handle cortexa57.
584 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
585 (cortex-a57.cortex-a53): Likewise.
587 2014-02-06 Jakub Jelinek <jakub@redhat.com>
590 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
591 don't record in REG_FRAME_RELATED_EXPR registers not set in that
593 (arm_expand_prologue): Adjust all callers.
594 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
595 info, registers also at the lowest numbered registers side. Use
596 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
600 * var-tracking.c (adjust_mems): Before adding a SET to
601 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
603 2014-02-06 Alan Modra <amodra@gmail.com>
606 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
607 change SDmode to DDmode when lra_in_progress.
609 2014-02-06 Jakub Jelinek <jakub@redhat.com>
612 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
613 free_data_ref on the dr first, and before goto again also set dr
614 to the next dr. For simd_lane_access, free old datarefs[i] before
615 overwriting it. For get_vectype_for_scalar_type failure, don't
616 free_data_ref if simd_lane_access.
618 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
621 * tree.h (opts_for_fn): New inline function.
622 (opt_for_fn): Define.
623 * config/i386/i386.c (ix86_function_regparm): Use
624 opt_for_fn (decl, optimize) instead of optimize.
626 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
628 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
629 for SYMBOL_REF in large memory model.
631 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
633 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
635 (cortex-a57): Likewise.
636 (cortex-a57.cortex-a53): Likewise.
638 2014-02-06 Yury Gribov <y.gribov@samsung.com>
639 Kugan Vivekanandarajah <kuganv@linaro.org>
641 * config/arm/arm.c (arm_vector_alignment_reachable): Check
643 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
645 2014-02-06 Richard Biener <rguenther@suse.de>
647 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
648 set_loop_copy and initialize_original_copy_tables.
650 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
652 * config/aarch64/aarch64-simd.md
653 (aarch64_ashr_simddi): Change QI to SI.
655 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
656 Jakub Jelinek <jakub@redhat.com>
659 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
662 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
664 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
665 CODE_FOR_altivec_vpku[hw]um to
666 CODE_FOR_altivec_vpku[hw]um_direct.
667 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
668 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
669 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
670 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
672 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
674 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
675 generation for -maltivec=be.
676 (altivec_vsumsws): Simplify redundant test.
678 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
680 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
681 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
682 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
683 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
685 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
687 (altivec_vpks<VI_char>ss): Likewise.
688 (altivec_vpks<VI_char>us): Likewise.
689 (altivec_vpku<VI_char>us): Likewise.
690 (altivec_vpku<VI_char>um): Likewise.
691 (altivec_vpku<VI_char>um_direct): New (copy of
692 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
694 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
695 target is little endian and -maltivec=be is not specified.
696 (*altivec_vupkhs<VU_char>_direct): New (copy of
697 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
698 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
699 target is little endian and -maltivec=be is not specified.
700 (*altivec_vupkls<VU_char>_direct): New (copy of
701 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
702 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
703 little endian and -maltivec=be is not specified.
704 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
705 little endian and -maltivec=be is not specified.
707 2014-02-05 Richard Henderson <rth@redhat.com>
710 * combine-stack-adj.c: Revert r206943.
711 * sched-int.h (struct deps_desc): Add last_args_size.
712 * sched-deps.c (init_deps): Initialize it.
713 (sched_analyze_insn): Add OUTPUT dependencies between insns that
714 contain REG_ARGS_SIZE notes.
716 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
718 * lto-cgraph.c (asm_nodes_output): Make global.
719 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
720 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
721 (driver_handle_option): Handle OPT_fwpa.
723 2014-02-05 Jakub Jelinek <jakub@redhat.com>
726 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
727 a comment typo and formatting issue. If odr_hash hasn't been
728 created, return vNULL and set *completep to false.
731 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
732 bb with no successors.
734 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
737 * doc/invoke.texi (-march): Clarify documentation for ARM.
741 2014-02-05 Richard Biener <rguenther@suse.de>
743 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
744 when not vectorizing because of too many alias checks.
745 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
746 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
748 2014-02-05 Nick Clifton <nickc@redhat.com>
750 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
751 accept extended registers in any mode when compiling for the MN10300.
753 2014-02-05 Yury Gribov <y.gribov@samsung.com>
755 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
756 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
757 sanitization attributes.
758 (can_inline_edge_p): Likewise.
759 (sanitize_attrs_match_for_inline_p): New function.
761 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
763 * ipa-prop.c (detect_type_change): Shor circuit testing of
764 type changes on THIS pointer.
766 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
769 * config/pa/pa.c (legitimize_tls_address): Return original address
770 if not passed a SYMBOL_REF rtx.
771 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
773 (pa_emit_move_sequence): Simplify TLS source operands.
774 (pa_legitimate_constant_p): Reject all TLS constants.
775 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
776 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
778 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
780 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
781 groups when we know they are controlled by LTO.
782 * varasm.c (default_binds_local_p_1): If object is in other partition,
783 it will be resolved locally.
785 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
787 * config/host-linux.c (linux_gt_pch_use_address): Don't
788 use SSIZE_MAX because it is not always defined.
790 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
793 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
794 threshold for pseudo splitting.
795 (update_ebb_live_info): Process call argument hard registers and
796 hard registers from insn definition too.
797 (max_small_class_regs_num): New constant.
798 (inherit_in_ebb): Update live hard regs through EBBs. Update
799 reloads_num only for small register classes. Don't split for
802 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
805 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
808 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
810 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
813 2014-02-04 Marek Polacek <polacek@redhat.com>
815 * gdbinit.in (pel): Define.
817 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
819 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
822 2014-02-04 Richard Biener <rguenther@suse.de>
825 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
826 in function context local.
827 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
828 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
829 similar to LTO_imported_decl_ref.
831 2014-02-04 Jakub Jelinek <jakub@redhat.com>
833 PR tree-optimization/60002
834 * cgraphclones.c (build_function_decl_skip_args): Clear
837 PR tree-optimization/60023
838 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
839 false to gsi_replace.
840 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
841 has been in some EH region and vec_stmt could throw, add
842 vec_stmt into the same EH region.
843 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
844 has no lhs, ignore it.
845 * internal-fn.c (expand_MASK_LOAD): Likewise.
848 * tree-inline.c (copy_forbidden): Fail for
849 __attribute__((optimize (0))) functions.
852 * omp-low.c (simd_clone_struct_copy): If from->inbranch
853 is set, copy one less argument.
854 (expand_simd_clones): Don't subtract clone_info->inbranch
855 from simd_clone_struct_alloc argument.
857 PR rtl-optimization/57915
858 * recog.c (simplify_while_replacing): If all unary/binary/relational
859 operation arguments are constant, attempt to simplify those.
862 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
863 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
865 2014-02-04 Richard Biener <rguenther@suse.de>
867 PR tree-optimization/60012
868 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
869 TBAA disambiguation to all DDRs.
871 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
874 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
875 (LINK_SPEC): Use it for -shared, -shared-libgcc.
877 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
880 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
882 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
884 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
885 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
887 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
890 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
891 to figure out targets of polymorphic calls with known decl.
892 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
893 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
894 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
895 (get_polymorphic_call_info): ... here.
896 (get_polymorphic_call_info_from_invariant): New function.
898 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
900 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
901 lookup via vtable pointer; check for type consistency
902 and turn inconsitent facts into UNREACHABLE.
903 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
904 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
905 type inconsistent querries; return UNREACHABLE instead.
907 2014-02-03 Richard Henderson <rth@twiddle.net>
910 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
911 already processed this node.
912 (normalize_one_pred_1): Pass along mark_set.
913 (normalize_one_pred): Create and destroy a pointer_set_t.
914 (normalize_one_pred_chain): Likewise.
916 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
918 PR gcov-profile/58602
919 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
921 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
924 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
925 -fno-devirtualize; try to devirtualize by the knowledge of
926 virtual table pointer given by aggregate propagation.
927 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
928 (ipa_print_node_jump_functions): Dump also offset that
929 is relevant for polymorphic calls.
930 (determine_known_aggregate_parts): Add arg_type parameter; use it
931 instead of determining the type from pointer type.
932 (ipa_compute_jump_functions_for_edge): Update call of
933 determine_known_aggregate_parts.
934 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
935 (gimple_get_virt_method_for_binfo): ... here; simplify using
936 vtable_pointer_value_to_vtable.
937 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
938 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
939 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
940 (vtable_pointer_value_to_vtable): Break out from ...; handle also
942 (vtable_pointer_value_to_binfo): ... here.
943 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
945 2014-02-03 Teresa Johnson <tejohnson@google.com>
947 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
948 redef of outer loop index variable.
950 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
954 * doc/extend.texi (Function Attributes): Typo.
956 2014-02-03 Cong Hou <congh@google.com>
958 PR tree-optimization/60000
959 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
960 if the vectorized statement is a store. A store statement can only
961 appear at the end of pattern statements.
963 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
965 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
966 (ix86_option_override_internal): Default long double to 64-bit for
967 32-bit Bionic and to 128-bit for 64-bit Bionic.
969 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
970 TARGET_LONG_DOUBLE_128 is true.
971 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
973 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
974 (mlong-double-64): Negate -mlong-double-128.
975 (mlong-double-128): New option.
977 * config/i386/i386-c.c (ix86_target_macros): Define
978 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
980 * doc/invoke.texi: Document -mlong-double-128.
982 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
984 PR rtl-optimization/60024
985 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
987 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
989 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
991 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
993 PR rtl-optimization/57662
994 * sel-sched.c (code_motion_path_driver): Do not mark already not
995 existing blocks in the visiting bitmap.
997 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
999 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1000 on the insn being emitted.
1002 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1003 Will Deacon <will.deacon@arm.com>
1005 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1007 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1009 * config/arm/arm-tables.opt: Regenerate.
1011 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1013 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1014 for vector types other than V16QImode.
1015 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1016 define_expand, and call altivec_expand_vec_perm_le when producing
1017 code with little endian element order.
1018 (*altivec_vperm_<mode>_internal): New insn having previous
1019 behavior of altivec_vperm_<mode>.
1020 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1021 altivec_expand_vec_perm_le when producing code with little endian
1023 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1024 behavior of altivec_vperm_<mode>_uns.
1026 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1028 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1029 (altivec_vsumsws): Add handling for -maltivec=be with a little
1031 (altivec_vsumsws_direct): New.
1032 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1033 gen_altivec_vsumsws.
1035 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1037 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1038 vtable_pointer_value_to_binfo): New functions.
1039 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1040 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1042 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1044 * config/nios2/nios2.md (load_got_register): Initialize GOT
1045 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1046 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1048 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1050 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1051 preserverd by passthrough, do not propagate the type.
1053 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1055 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1056 (mips_atomic_assign_expand_fenv): New function.
1057 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1059 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1061 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1062 (__builtin_mips_set_fcsr): Likewise.
1063 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1064 MIPS_USI_FTYPE_VOID.
1065 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1066 (mips16_expand_set_fcsr): Likewise.
1067 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1068 (mips16_set_fcsr_stub): Likewise.
1069 (mips16_get_fcsr_one_only_stub): New class.
1070 (mips16_set_fcsr_one_only_stub): Likewise.
1071 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1072 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1073 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1074 (hard_float): New availability predicate.
1075 (mips_builtins): Add get_fcsr and set_fcsr.
1076 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1077 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1078 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1079 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1080 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1083 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1085 * config/mips/mips.c (mips_one_only_stub): New class.
1086 (mips_need_mips16_rdhwr_p): Replace with...
1087 (mips16_rdhwr_stub): ...this new variable.
1088 (mips16_stub_call_address): New function.
1089 (mips16_rdhwr_one_only_stub): New class.
1090 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1091 (mips_output_mips16_rdhwr): Delete.
1092 (mips_finish_stub): New function.
1093 (mips_code_end): Use it to handle rdhwr stubs.
1095 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1098 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1099 when calculating size of integer atomic types.
1101 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1103 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1105 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1107 PR tree-optimization/60003
1108 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1109 * profile.c (branch_prob): Use gimple_call_builtin_p
1110 to check for BUILT_IN_SETJMP_RECEIVER.
1111 * tree-inline.c (copy_bb): Call notice_special_calls.
1113 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1116 * lra-constraints.c (process_alt_operands): Update reload_sum only
1119 2014-01-31 Richard Henderson <rth@redhat.com>
1122 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1123 until after else_eh is processed.
1125 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1127 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1128 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1129 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1130 in smmintrin.h, remove them.
1131 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1132 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1133 * config/i386/i386.md (ROUND_SAE): Fix value.
1134 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1135 (const48_operand): New.
1136 * config/i386/subst.md (round), (round_expand): Use
1137 const_4_or_8_to_11_operand.
1138 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1140 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1142 * config/i386/constraints.md (Yk): Swap meaning with k.
1143 * config/i386/i386.md (movhi_internal): Change Yk to k.
1144 (movqi_internal): Ditto.
1145 (*k<logic><mode>): Ditto.
1148 (kandn<mode>): Ditto.
1149 (*<code>hi_1): Ditto.
1150 (*<code>qi_1): Ditto.
1151 (kxnor<mode>): Ditto.
1152 (kortestzhi): Ditto.
1153 (kortestchi): Ditto.
1155 (*one_cmplhi2_1): Ditto.
1156 (*one_cmplqi2_1): Ditto.
1157 * config/i386/sse.md (): Change k to Yk.
1158 (avx512f_load<mode>_mask): Ditto.
1159 (avx512f_blendm<mode>): Ditto.
1160 (avx512f_store<mode>_mask): Ditto.
1161 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1162 (avx512f_storedqu<mode>_mask): Ditto.
1163 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1165 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1166 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1167 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1168 (avx512f_maskcmp<mode>3): Ditto.
1169 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1170 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1171 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1172 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1173 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1174 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1175 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1176 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1177 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1178 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1179 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1180 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1181 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1182 (vec_extract_lo_<mode>_maskm): Ditto.
1183 (vec_extract_hi_<mode>_maskm): Ditto.
1184 (avx512f_vternlog<mode>_mask): Ditto.
1185 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1186 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1187 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1188 (avx512f_<code>v8div16qi2_mask): Ditto.
1189 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1190 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1191 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1192 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1193 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1194 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1195 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1196 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1197 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1198 (avx512cd_maskb_vec_dupv8di): Ditto.
1199 (avx512cd_maskw_vec_dupv16si): Ditto.
1200 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1201 (avx512f_vpermi2var<mode>3_mask): Ditto.
1202 (avx512f_vpermi2var<mode>3_mask): Ditto.
1203 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1204 (*avx512f_gathersi<mode>): Ditto.
1205 (*avx512f_gathersi<mode>_2): Ditto.
1206 (*avx512f_gatherdi<mode>): Ditto.
1207 (*avx512f_gatherdi<mode>_2): Ditto.
1208 (*avx512f_scattersi<mode>): Ditto.
1209 (*avx512f_scatterdi<mode>): Ditto.
1210 (avx512f_compress<mode>_mask): Ditto.
1211 (avx512f_compressstore<mode>_mask): Ditto.
1212 (avx512f_expand<mode>_mask): Ditto.
1213 * config/i386/subst.md (mask): Change k to Yk.
1214 (mask_scalar_merge): Ditto.
1217 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1219 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1221 2014-01-31 Richard Biener <rguenther@suse.de>
1224 * builtins.c (fold_builtin_memory_op): Make sure to not
1225 use a floating-point mode or a boolean or enumeral type for
1228 2014-01-30 DJ Delorie <dj@redhat.com>
1230 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1231 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1232 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1233 whenever main() has an epilogue.
1235 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1237 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1238 unused variable "field".
1239 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1240 (vsx_mergeh_<mode>): Likewise.
1241 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1242 (altivec_vmrghh): Likewise.
1243 (altivec_vmrghw): Likewise.
1244 (altivec_vmrglb): Likewise.
1245 (altivec_vmrglh): Likewise.
1246 (altivec_vmrglw): Likewise.
1247 (altivec_vspltb): Add missing uses.
1248 (altivec_vsplth): Likewise.
1249 (altivec_vspltw): Likewise.
1250 (altivec_vspltsf): Likewise.
1252 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1255 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1256 frame related instructions.
1258 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1260 PR rtl-optimization/59959
1261 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1262 any reload of register whose subreg is invalid.
1264 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1266 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1267 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1268 Add missing return type - void.
1270 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1272 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1273 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1274 remove element index adjustment for endian (now handled in vsx.md
1276 (altivec_expand_vec_perm_const): Use
1277 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1278 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1279 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1280 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1281 define_expand and a new define_insn *altivec_vspltb_internal;
1282 adjust for -maltivec=be on a little endian target.
1283 (altivec_vspltb_direct): New.
1284 (altivec_vsplth): Divide into a define_expand and a new
1285 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1286 little endian target.
1287 (altivec_vsplth_direct): New.
1288 (altivec_vspltw): Divide into a define_expand and a new
1289 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1290 little endian target.
1291 (altivec_vspltw_direct): New.
1292 (altivec_vspltsf): Divide into a define_expand and a new
1293 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1294 a little endian target.
1296 2014-01-30 Richard Biener <rguenther@suse.de>
1298 PR tree-optimization/59993
1299 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1300 can propagate form the earlier stmt and avoid the transform
1301 when the intermediate result is needed.
1303 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1305 * README.Portability: Fix typo.
1307 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1309 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1310 comparison_operator with ordered_comparison_operator.
1312 2014-01-30 Nick Clifton <nickc@redhat.com>
1314 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1315 Rename to mn10300_store_multiple_regs.
1316 * config/mn10300/mn10300.c: Likewise.
1317 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1318 store_multiple_regs.
1319 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1320 Call mn10300_store_multiple_regs.
1322 2014-01-30 Nick Clifton <nickc@redhat.com>
1323 DJ Delorie <dj@redhat.com>
1325 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1326 %fp 2 to keep registers after it properly word-aligned.
1327 (rl78_alloc_physical_registers_umul): Handle the case where both
1328 input operands are the same.
1330 2014-01-30 Richard Biener <rguenther@suse.de>
1332 PR tree-optimization/59903
1333 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1336 2014-01-30 Jason Merrill <jason@redhat.com>
1339 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1342 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1344 2014-01-30 Richard Biener <rguenther@suse.de>
1346 PR tree-optimization/59951
1347 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1349 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1352 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1353 SFmode to DFmode case.
1355 2014-01-29 DJ Delorie <dj@redhat.com>
1357 * config/msp430/msp430.opt (-minrt): New.
1358 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1360 (ENDFILE_SPEC): Likewise.
1362 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1364 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1365 (estimate_function_body_sizes): Use it.
1367 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1370 * dwarf2out.c (is_cxx_auto): New.
1371 (is_base_type): Use it.
1372 (gen_type_die_with_usage): Likewise.
1374 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1376 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1377 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1378 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1379 -maltivec=be with LE targets.
1380 (vsx_mergeh_<mode>): Likewise.
1381 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1382 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1383 (altivec_vmrghb): Replace with define_expand and new
1384 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1385 (altivec_vmrghb_direct): New define_insn.
1386 (altivec_vmrghh): Replace with define_expand and new
1387 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1388 (altivec_vmrghh_direct): New define_insn.
1389 (altivec_vmrghw): Replace with define_expand and new
1390 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1391 (altivec_vmrghw_direct): New define_insn.
1392 (*altivec_vmrghsf): Adjust for endianness.
1393 (altivec_vmrglb): Replace with define_expand and new
1394 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1395 (altivec_vmrglb_direct): New define_insn.
1396 (altivec_vmrglh): Replace with define_expand and new
1397 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1398 (altivec_vmrglh_direct): New define_insn.
1399 (altivec_vmrglw): Replace with define_expand and new
1400 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1401 (altivec_vmrglw_direct): New define_insn.
1402 (*altivec_vmrglsf): Adjust for endianness.
1403 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1404 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1405 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1406 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1407 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1408 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1409 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1410 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1412 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1414 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1415 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1418 2014-01-29 Richard Biener <rguenther@suse.de>
1420 PR tree-optimization/58742
1421 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1422 associate_pointerplus_align.
1423 (associate_pointerplus_diff): New function.
1424 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1425 and associate_pointerplus_diff.
1427 2014-01-29 Richard Biener <rguenther@suse.de>
1429 * lto-streamer.h (LTO_major_version): Bump to 3.
1430 (LTO_minor_version): Reset to 0.
1432 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1434 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1435 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1436 (arm_file_start): Generate correct asm header for armv7ve.
1437 * config/arm/bpabi.h: Add multilib support for armv7ve.
1438 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1439 and cortex-a15 to armv7ve.
1440 * config/arm/t-aprofile: Add multilib support for armv7ve.
1441 * doc/invoke.texi: Document -march=armv7ve.
1443 2014-01-29 Richard Biener <rguenther@suse.de>
1445 PR tree-optimization/58742
1446 * tree-ssa-forwprop.c (associate_plusminus): Return true
1447 if we changed sth, defer EH cleanup to ...
1448 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1449 (simplify_mult): New function.
1451 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1454 PR tree-optimization/59920
1455 * tree.c (build_common_builtin_nodes): Remove
1456 __builtin_setjmp_dispatcher initialization.
1457 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1458 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1459 instead of gsi_after_labels + manually skipping debug stmts.
1460 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1461 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1462 * tree-inline.c (copy_edges_for_bb): Remove
1463 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1464 argument. Ignore computed_goto_p stmts. Don't call
1465 make_abnormal_goto_edges. If a call might need abnormal edges
1466 for non-local gotos, see if it already has an edge to
1467 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1468 with true argument, don't do anything then, otherwise add
1469 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1470 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1472 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1473 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1474 (lower_stmt): Don't set data->calls_builtin_setjmp.
1475 (lower_builtin_setjmp): Adjust comment.
1476 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1477 * tree-cfg.c (found_computed_goto): Remove.
1478 (factor_computed_gotos): Remove.
1479 (make_goto_expr_edges): Return bool, true for computed gotos.
1480 Don't call make_abnormal_goto_edges.
1481 (build_gimple_cfg): Don't set found_computed_goto, don't call
1482 factor_computed_gotos.
1483 (computed_goto_p): No longer static.
1484 (make_blocks): Don't set found_computed_goto.
1485 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1486 (make_edges): If make_goto_expr_edges returns true, push bb
1487 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1488 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1489 vector. Record mapping between bbs and OpenMP regions if there
1490 are any, adjust make_gimple_omp_edges caller. Call
1491 handle_abnormal_edges.
1492 (make_abnormal_goto_edges): Remove.
1493 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1494 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1495 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1496 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1497 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1498 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1499 filling *region also set *region_idx to (*region)->entry->index.
1502 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1503 For REGs set ORIGINAL_REGNO.
1505 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1507 * doc/md.texi: Mention that a target shouldn't implement
1508 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1511 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1513 PR tree-optimization/59594
1514 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1515 a copy of the datarefs vector rather than the vector itself.
1517 2014-01-28 Jason Merrill <jason@redhat.com>
1520 * dwarf2out.c (auto_die): New static.
1521 (gen_type_die_with_usage): Handle C++1y 'auto'.
1522 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1525 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1528 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1529 (SPEC_X32): Likewise.
1530 (SPEC_64): Likewise.
1531 * config/i386/i386.c (ix86_option_override_internal): Turn off
1532 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1534 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1535 * config/i386/i386.h (TARGET_16BIT): New macro.
1536 (TARGET_16BIT_P): Likewise.
1537 * config/i386/i386.opt: Add m16.
1538 * doc/invoke.texi: Document -m16.
1540 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1542 PR preprocessor/59935
1543 * input.c (location_get_source_line): Bail out on when line number
1544 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1546 2014-01-28 Richard Biener <rguenther@suse.de>
1548 PR tree-optimization/58742
1549 * tree-ssa-forwprop.c (associate_plusminus): Handle
1550 pointer subtraction of the form (T)(P + A) - (T)P.
1552 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1554 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1557 2014-01-28 Richard Biener <rguenther@suse.de>
1560 2014-01-28 Richard Biener <rguenther@suse.de>
1562 PR rtl-optimization/45364
1563 PR rtl-optimization/59890
1564 * var-tracking.c (local_get_addr_clear_given_value): Handle
1565 already cleared slot.
1566 (val_reset): Handle not allocated local_get_addr_cache.
1567 (vt_find_locations): Use post-order on the inverted CFG.
1569 2014-01-28 Richard Biener <rguenther@suse.de>
1571 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1573 2014-01-28 Richard Biener <rguenther@suse.de>
1575 PR rtl-optimization/45364
1576 PR rtl-optimization/59890
1577 * var-tracking.c (local_get_addr_clear_given_value): Handle
1578 already cleared slot.
1579 (val_reset): Handle not allocated local_get_addr_cache.
1580 (vt_find_locations): Use post-order on the inverted CFG.
1582 2014-01-28 Alan Modra <amodra@gmail.com>
1584 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1585 * configure.ac <recursive call for build != host>: Define
1586 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1587 and LD_FOR_BUILD too.
1588 * configure: Regenerate.
1590 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1592 * config/i386/i386.c (get_builtin_code_for_version): Separate
1593 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1594 Broadwell from Haswell.
1596 2014-01-27 Steve Ellcey <sellcey@mips.com>
1598 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1599 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1600 * config/mips/mips.c (mips_option_override): Change setting
1602 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1603 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1604 Change from Mask to Var.
1606 2014-01-27 Jeff Law <law@redhat.com>
1608 * ipa-inline.c (inline_small_functions): Fix typo.
1610 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1612 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1613 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1614 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1615 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1616 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1617 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1618 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1619 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1620 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1621 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1622 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1623 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1624 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1625 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1626 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1627 (_mm512_storeu_epi64): Ditto.
1628 (_mm512_cmpge_epi32_mask): Ditto.
1629 (_mm512_cmpge_epu32_mask): Ditto.
1630 (_mm512_cmpge_epi64_mask): Ditto.
1631 (_mm512_cmpge_epu64_mask): Ditto.
1632 (_mm512_cmple_epi32_mask): Ditto.
1633 (_mm512_cmple_epu32_mask): Ditto.
1634 (_mm512_cmple_epi64_mask): Ditto.
1635 (_mm512_cmple_epu64_mask): Ditto.
1636 (_mm512_cmplt_epi32_mask): Ditto.
1637 (_mm512_cmplt_epu32_mask): Ditto.
1638 (_mm512_cmplt_epi64_mask): Ditto.
1639 (_mm512_cmplt_epu64_mask): Ditto.
1640 (_mm512_cmpneq_epi32_mask): Ditto.
1641 (_mm512_cmpneq_epu32_mask): Ditto.
1642 (_mm512_cmpneq_epi64_mask): Ditto.
1643 (_mm512_cmpneq_epu64_mask): Ditto.
1644 (_mm512_expand_pd): Ditto.
1645 (_mm512_expand_ps): Ditto.
1646 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1647 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1648 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1649 * config/i386/i386.c (ix86_builtins): Add
1650 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1651 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1652 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1653 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1654 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1655 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1656 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1657 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1658 IX86_BUILTIN_PMOVUSQW512_MEM.
1659 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1660 __builtin_ia32_pmovsqd512mem_mask,
1661 __builtin_ia32_pmovqd512mem_mask,
1662 __builtin_ia32_pmovusqw512mem_mask,
1663 __builtin_ia32_pmovsqw512mem_mask,
1664 __builtin_ia32_pmovqw512mem_mask,
1665 __builtin_ia32_pmovusdw512mem_mask,
1666 __builtin_ia32_pmovsdw512mem_mask,
1667 __builtin_ia32_pmovdw512mem_mask,
1668 __builtin_ia32_pmovqb512mem_mask,
1669 __builtin_ia32_pmovusqb512mem_mask,
1670 __builtin_ia32_pmovsqb512mem_mask,
1671 __builtin_ia32_pmovusdb512mem_mask,
1672 __builtin_ia32_pmovsdb512mem_mask,
1673 __builtin_ia32_pmovdb512mem_mask.
1674 (bdesc_args): Add __builtin_ia32_expanddf512,
1675 __builtin_ia32_expandsf512.
1676 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1677 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1678 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1679 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1680 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1681 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1682 (avx512f_<code>v8div16qi2_mask_store): This.
1683 (avx512f_expand<mode>): New.
1685 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1687 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1689 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1690 (_mm512_prefetch_i32scatter_pd): Ditto.
1691 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1692 (_mm512_prefetch_i64scatter_pd): Ditto.
1693 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1694 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1695 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1696 (_mm512_prefetch_i32scatter_ps): Ditto.
1697 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1698 (_mm512_prefetch_i64scatter_ps): Ditto.
1699 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1700 * config/i386/i386-builtin-types.def: Define
1701 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1702 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1703 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1704 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1705 IX86_BUILTIN_SCATTERPFQPD.
1706 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1707 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1708 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1709 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1710 __builtin_ia32_scatterpfqps.
1711 (ix86_expand_builtin): Expand new built-ins.
1712 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1713 fix memory access data type.
1714 (*avx512pf_gatherpf<mode>_mask): Ditto.
1715 (*avx512pf_gatherpf<mode>): Ditto.
1716 (avx512pf_scatterpf<mode>): Ditto.
1717 (*avx512pf_scatterpf<mode>_mask): Ditto.
1718 (*avx512pf_scatterpf<mode>): Ditto.
1719 (GATHER_SCATTER_SF_MEM_MODE): New.
1720 (avx512pf_gatherpf<mode>df): Ditto.
1721 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1722 (*avx512pf_scatterpf<mode>df): Ditto.
1724 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1727 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1728 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1731 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1733 * common/config/arm/arm-common.c
1734 (arm_rewrite_mcpu): Handle multiple names.
1736 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1738 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1740 * gimple-builder.h (create_gimple_tmp): Delete.
1742 2014-01-27 Christian Bruel <christian.bruel@st.com>
1744 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1747 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1749 * config/pa/pa.md (call): Generate indirect long calls to non-local
1750 functions when outputing 32-bit code.
1751 (call_value): Likewise except for special call to buggy powf function.
1753 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1754 portable runtime and PIC indirect calls.
1755 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1756 and PIC call sequences. Use ldo instead of blr to set return register
1757 in PIC call sequence.
1759 2014-01-25 Walter Lee <walt@tilera.com>
1761 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1762 avoid clobbering a live register.
1764 2014-01-25 Walter Lee <walt@tilera.com>
1766 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1767 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1768 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1769 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1771 2014-01-25 Walter Lee <walt@tilera.com>
1773 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1774 arguments on even registers.
1775 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1777 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1778 (BIGGEST_ALIGNMENT): Ditto.
1779 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1781 2014-01-25 Walter Lee <walt@tilera.com>
1783 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1784 insns before bundling.
1785 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1787 2014-01-25 Walter Lee <walt@tilera.com>
1789 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1790 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1791 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1793 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1795 * config/mips/constraints.md (kl): Delete.
1796 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1797 define expands, using...
1798 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1799 instructions for MIPS16.
1800 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1801 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1803 2014-01-25 Walter Lee <walt@tilera.com>
1805 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1809 2014-01-25 Walter Lee <walt@tilera.com>
1811 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1812 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1814 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1816 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1819 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1821 * print-rtl.c (in_call_function_usage): New var.
1822 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1823 EXPR_LIST mode as mode and not as reg note name.
1826 * cfgloopmanip.c (copy_loop_info): If
1827 loop->warned_aggressive_loop_optimizations, make sure
1828 the flag is set in target loop too.
1830 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1832 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1834 * builtins.def: Likewise.
1835 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1836 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1837 * ira.c (ira_setup_eliminable_regset): Likewise.
1838 * omp-low.c (gate_expand_omp): Likewise.
1839 (execute_lower_omp): Likewise.
1840 (diagnose_sb_0): Likewise.
1841 (gate_diagnose_omp_blocks): Likewise.
1842 (simd_clone_clauses_extract): Likewise.
1845 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1847 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1848 correction for little endian...
1849 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1852 2014-01-24 Jeff Law <law@redhat.com>
1854 PR tree-optimization/59919
1855 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1856 for non-returning calls.
1858 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1860 * common/config/aarch64/aarch64-common.c
1861 (aarch64_rewrite_mcpu): Handle multiple names.
1862 * config/aarch64/aarch64.h
1863 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1865 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1867 * input.c (add_file_to_cache_tab): Handle the case where fopen
1870 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1873 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1874 from push operand if code of push isn't PRE_DEC.
1876 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1879 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1880 -mquad-memory-atomic. Update -mquad-memory documentation to say
1881 it is only used for non-atomic loads/stores.
1883 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1884 -mquad-memory or -mquad-memory-atomic switches.
1886 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1887 -mquad-memory-atomic to ISA 2.07 support.
1889 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1890 to separate support of normal quad word memory operations (ldq, stq)
1891 from the atomic quad word memory operations.
1893 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1894 support to separate non-atomic quad word operations from atomic
1895 quad word operations. Disable non-atomic quad word operations in
1896 little endian mode so that we don't have to swap words after the
1897 load and before the store.
1898 (quad_load_store_p): Add comment about atomic quad word support.
1899 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1900 options printed with -mdebug=reg.
1902 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1903 -mquad-memory-atomic as the test for whether we have quad word
1904 atomic instructions.
1905 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1906 or -mp8-vector are used, allow byte/half-word atomic operations.
1908 * config/rs6000/sync.md (load_lockedti): Insure that the address
1909 is a proper indexed or indirect address for the lqarx instruction.
1910 On little endian systems, swap the hi/lo registers after the lqarx
1912 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1913 insure the address is valid for the lqarx instruction.
1914 (store_conditionalti): Insure that the address is a proper indexed
1915 or indirect address for the stqcrx. instruction. On little endian
1916 systems, swap the hi/lo registers before doing the stqcrx.
1918 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1919 insure the address is valid for the stqcrx. instruction.
1921 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1922 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1923 type of quad memory support is available.
1925 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1928 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1929 there is a danger of looping.
1931 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1934 force flag_ira_loop_pressure if set via command line.
1936 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1938 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1939 (ashr_simd): New builtin handling DI mode.
1940 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1941 (aarch64_sshr_simddi): New match pattern.
1942 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1943 (vshrd_n_s64): Likewise.
1944 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1946 2014-01-23 Nick Clifton <nickc@redhat.com>
1948 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1949 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1950 favour of mcu specific scripts.
1951 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1954 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1955 Alex Velenko <Alex.Velenko@arm.com>
1957 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1958 (vaddv_s16): Likewise.
1959 (vaddv_s32): Likewise.
1960 (vaddv_u8): Likewise.
1961 (vaddv_u16): Likewise.
1962 (vaddv_u32): Likewise.
1963 (vaddvq_s8): Likewise.
1964 (vaddvq_s16): Likewise.
1965 (vaddvq_s32): Likewise.
1966 (vaddvq_s64): Likewise.
1967 (vaddvq_u8): Likewise.
1968 (vaddvq_u16): Likewise.
1969 (vaddvq_u32): Likewise.
1970 (vaddvq_u64): Likewise.
1971 (vaddv_f32): Likewise.
1972 (vaddvq_f32): Likewise.
1973 (vaddvq_f64): Likewise.
1974 (vmaxv_f32): Likewise.
1975 (vmaxv_s8): Likewise.
1976 (vmaxv_s16): Likewise.
1977 (vmaxv_s32): Likewise.
1978 (vmaxv_u8): Likewise.
1979 (vmaxv_u16): Likewise.
1980 (vmaxv_u32): Likewise.
1981 (vmaxvq_f32): Likewise.
1982 (vmaxvq_f64): Likewise.
1983 (vmaxvq_s8): Likewise.
1984 (vmaxvq_s16): Likewise.
1985 (vmaxvq_s32): Likewise.
1986 (vmaxvq_u8): Likewise.
1987 (vmaxvq_u16): Likewise.
1988 (vmaxvq_u32): Likewise.
1989 (vmaxnmv_f32): Likewise.
1990 (vmaxnmvq_f32): Likewise.
1991 (vmaxnmvq_f64): Likewise.
1992 (vminv_f32): Likewise.
1993 (vminv_s8): Likewise.
1994 (vminv_s16): Likewise.
1995 (vminv_s32): Likewise.
1996 (vminv_u8): Likewise.
1997 (vminv_u16): Likewise.
1998 (vminv_u32): Likewise.
1999 (vminvq_f32): Likewise.
2000 (vminvq_f64): Likewise.
2001 (vminvq_s8): Likewise.
2002 (vminvq_s16): Likewise.
2003 (vminvq_s32): Likewise.
2004 (vminvq_u8): Likewise.
2005 (vminvq_u16): Likewise.
2006 (vminvq_u32): Likewise.
2007 (vminnmv_f32): Likewise.
2008 (vminnmvq_f32): Likewise.
2009 (vminnmvq_f64): Likewise.
2011 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2013 * config/aarch64/aarch64-simd.md
2014 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2015 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2016 (*aarch64_mul3_elt<mode>): Likewise.
2017 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2018 (*aarch64_mul3_elt_to_64v2df): Likewise.
2019 (*aarch64_mla_elt<mode>): Likewise.
2020 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2021 (*aarch64_mls_elt<mode>): Likewise.
2022 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2023 (*aarch64_fma4_elt<mode>): Likewise.
2024 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2025 (*aarch64_fma4_elt_to_64v2df): Likewise.
2026 (*aarch64_fnma4_elt<mode>): Likewise.
2027 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2028 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2029 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2030 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2031 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2032 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2033 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2034 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2035 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2037 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2039 * config/aarch64/aarch64-simd.md
2040 (aarch64_be_checked_get_lane<mode>): New define_expand.
2041 * config/aarch64/aarch64-simd-builtins.def
2042 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2043 New builtin definition.
2044 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2045 Use new safe be builtin.
2047 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2049 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2051 (aarch64_be_st1<mode>): Likewise.
2052 (aarch_ld1<VALL:mode>): Define_expand modified.
2053 (aarch_st1<VALL:mode>): Likewise.
2054 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2055 (UNSPEC_ST1): Likewise.
2057 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2059 * config/microblaze/microblaze.md: Add trap insn and attribute
2061 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2063 PR preprocessor/58580
2064 * input.h (location_get_source_line): Take an additional line_size
2066 (void diagnostics_file_cache_fini): Declare new function.
2067 * input.c (struct fcache): New type.
2068 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2069 New static constants.
2070 (diagnostic_file_cache_init, total_lines_num)
2071 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2072 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2073 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2074 (get_next_line, read_next_line, goto_next_line, read_line_num):
2075 New static function definitions.
2076 (diagnostic_file_cache_fini): New function.
2077 (location_get_source_line): Take an additional output line_len
2078 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2080 * diagnostic.c (diagnostic_finish): Call
2081 diagnostic_file_cache_fini.
2082 (adjust_line): Take an additional input parameter for the length
2083 of the line, rather than calculating it with strlen.
2084 (diagnostic_show_locus): Adjust the use of
2085 location_get_source_line and adjust_line with respect to their new
2086 signature. While displaying a line now, do not stop at the first
2087 null byte. Rather, display the zero byte as a space and keep
2088 going until we reach the size of the line.
2089 * Makefile.in: Add vec.o to OBJS-libcommon
2091 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2092 Ilya Tocar <ilya.tocar@intel.com>
2094 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2095 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2096 (__builtin_ia32_kmov16): Ditto.
2097 * config/i386/i386.md (UNSPEC_KMOV): New.
2100 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2102 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2103 (_mm512_storeu_si512): Ditto.
2105 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2108 * rtl.h (get_referenced_operands): Declare.
2109 * recog.c (get_referenced_operands): New function.
2110 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2111 operands have been referenced when recording LO_SUM references.
2113 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2115 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2117 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2119 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2120 Enable for generic and recent AMD targets.
2122 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2124 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2125 ARG_SIZE note when adjustment was eliminated.
2127 2014-01-22 Jeff Law <law@redhat.com>
2129 PR tree-optimization/59597
2130 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2131 in file. Accept new argument REGISTERING and use it to modify
2132 dump output appropriately.
2133 (register_jump_thread): Corresponding changes.
2134 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2135 thread paths involving joiner blocks. Add code to dump cancelled
2136 jump threading paths.
2138 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2140 PR rtl-optimization/59477
2141 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2142 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2144 2014-01-22 Tom Tromey <tromey@redhat.com>
2146 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2148 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2150 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2152 PR rtl-optimization/59896
2153 * lra-constraints.c (process_alt_operands): Check unused note for
2154 matched operands of insn with no output reloads.
2156 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2158 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2159 (mips_move_from_gpr_cost): Likewise.
2161 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2163 PR rtl-optimization/59858
2164 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2165 ira_class_hard_regs_num.
2166 (process_alt_operands): Increase reject for dying matched operand.
2168 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2171 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2172 smaller than size, perform several stores or loads and stores
2173 at dst + count - size to store or copy all of size bytes, rather
2174 than just last modesize bytes.
2176 2014-01-20 DJ Delorie <dj@redhat.com>
2178 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2179 that CLOBBERs are REGs before propogating their values.
2181 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2184 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2185 (cgraph_inline_failed_type): New function.
2186 * cgraph.h (DEFCIFCODE): Add type.
2187 (cgraph_inline_failed_type_t): New enum.
2188 (cgraph_inline_failed_type): New prototype.
2189 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2190 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2191 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2192 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2193 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2194 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2195 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2196 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2197 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2198 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2199 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2200 OPTIMIZATION_MISMATCH.
2201 * tree-inline.c (expand_call_inline): Emit errors during
2202 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2204 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2207 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2208 mode attribute in insn output.
2210 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2212 * output.h (output_constant): Delete.
2213 * varasm.c (output_constant): Make private.
2215 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2217 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2219 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2222 * tree.h (fold_builtin_strcat): New prototype.
2223 * builtins.c (fold_builtin_strcat): No longer static. Add len
2224 argument, if non-NULL, don't call c_strlen. Optimize
2225 directly into __builtin_memcpy instead of __builtin_strcpy.
2226 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2227 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2229 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2231 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2232 for SImode_address_operand operands, having only a REG argument.
2234 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2236 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2237 loader name using mbig-endian.
2238 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2240 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2242 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2246 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2248 * config/aarch64/aarch64-protos.h
2249 (aarch64_cannot_change_mode_class_ptr): Declare.
2250 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2251 aarch64_cannot_change_mode_class_ptr): New.
2252 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2253 backend hook aarch64_cannot_change_mode_class.
2255 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2257 * common/config/aarch64/aarch64-common.c
2258 (aarch64_handle_option): Don't handle any option order logic here.
2259 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2260 selected_cpu, warn on architecture version mismatch.
2261 (aarch64_override_options): Fix parsing order for option strings.
2263 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2264 Iain Sandoe <iain@codesourcery.com>
2267 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2268 warning. Amend comment to reflect current functionality.
2270 2014-01-20 Richard Biener <rguenther@suse.de>
2273 * builtins.c (fold_builtin_strcat): Remove case better handled
2274 by tree-ssa-strlen.c.
2276 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2278 * config/aarch64/aarch64.opt
2279 (mcpu, march, mtune): Make case-insensitive.
2281 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2284 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2285 if operands[1] is a REG or ZERO_EXTEND of a REG.
2287 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2289 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2291 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2293 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2294 long non-pic millicode calls.
2296 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2298 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2300 2014-01-19 Kito Cheng <kito@0xlab.org>
2302 * builtins.c (expand_movstr): Check movstr expand done or fail.
2304 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2305 H.J. Lu <hongjiu.lu@intel.com>
2308 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2309 to DImode for zero-extended addresses.
2311 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2313 PR rtl-optimization/57763
2314 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2315 on the new indirect jump_insn and increment LABEL_NUSES (label).
2317 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2321 * config.gcc (x86_archs): New variable.
2322 (x86_64_archs): Likewise.
2323 (x86_cpus): Likewise.
2324 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2325 --with-arch/--with-cpu= options.
2326 Support --with-arch=/--with-cpu={nehalem,westmere,
2327 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2329 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2331 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2332 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2334 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2336 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2338 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2341 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2342 clear cpp_get_options (parse_in)->warn_unused_macros for
2343 ix86_target_macros_internal with cpp_define.
2345 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2347 * jump.c (delete_related_insns): Keep (use (insn))s.
2348 * reorg.c (redundant_insn): Check for barriers too.
2350 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2352 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2354 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2356 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2357 call to $$dyncall when TARGET_LONG_CALLS is true.
2359 2014-01-17 Jeff Law <law@redhat.com>
2361 * ree.c (combine_set_extension): Temporarily disable test for
2362 changing number of hard registers.
2364 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2367 * ipa-inline-analysis.c (inline_free_summary):
2368 Do not free summary of aliases.
2370 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2373 * gimplify.c (gimplify_expr): Use create_tmp_var
2374 instead of create_tmp_var_raw. If cond doesn't have
2375 integral type, don't add the IFN_ANNOTATE builtin at all.
2377 2014-01-17 Martin Jambor <mjambor@suse.cz>
2380 * ipa-cp.c (prev_edge_clone): New variable.
2381 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2382 Also resize prev_edge_clone vector.
2383 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2384 (ipcp_edge_removal_hook): New function.
2385 (ipcp_driver): Register ipcp_edge_removal_hook.
2387 2014-01-17 Andrew Pinski <apinski@cavium.com>
2388 Steve Ellcey <sellcey@mips.com>
2391 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2394 2014-01-17 Jeff Law <law@redhat.com>
2397 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2398 so that pass_ccp runs first.
2400 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2402 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2403 (ix86_adjust_cost): Use !TARGET_XXX.
2404 (do_reorder_for_imul): Likewise.
2405 (swap_top_of_ready_list): Likewise.
2406 (ix86_sched_reorder): Likewise.
2408 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2410 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2411 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2412 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2413 (intel_memset): New. Duplicate slm_memset.
2414 (intel_cost): New. Duplicate slm_cost.
2415 (m_INTEL): New macro.
2416 (processor_target_table): Add "intel".
2417 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2418 with PROCESSOR_INTEL for "intel".
2419 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2420 PROCESSOR_SILVERMONT.
2421 (ix86_issue_rate): Likewise.
2422 (ix86_adjust_cost): Likewise.
2423 (ia32_multipass_dfa_lookahead): Likewise.
2424 (swap_top_of_ready_list): Likewise.
2425 (ix86_sched_reorder): Likewise.
2426 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2427 instead of TARGET_OPT_AGU.
2428 * config/i386/i386.h (TARGET_INTEL): New.
2429 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2430 (processor_type): Add PROCESSOR_INTEL.
2431 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2432 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2434 2014-01-17 Marek Polacek <polacek@redhat.com>
2437 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2440 2014-01-17 Richard Biener <rguenther@suse.de>
2442 PR tree-optimization/46590
2443 * opts.c (default_options_table): Add entries for
2444 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2445 all enabled at -O1 but not for -Og.
2446 * common.opt (fbranch-count-reg): Remove Init(1).
2447 (fmove-loop-invariants): Likewise.
2448 (ftree-pta): Likewise.
2450 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2452 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2453 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2454 decls to at least the GCC 4.8 used alignments.
2457 * tree-nested.c (convert_nonlocal_reference_stmt,
2458 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2459 of GIMPLE_BIND stmts, adjust associated decls.
2461 2014-01-17 Richard Biener <rguenther@suse.de>
2463 PR tree-optimization/46590
2464 * vec.h (vec<>::bseach): New member function implementing
2465 binary search according to C89 bsearch.
2466 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2467 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2468 bitmap pointer again. Make accesses_in_loop a flat array.
2469 (mem_ref_obstack): New global.
2470 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2471 (mark_ref_stored): Likewise.
2472 (ref_indep_loop_p_2): Likewise.
2473 (set_ref_stored_in_loop): New helper function.
2474 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2475 (memref_free): Adjust.
2476 (record_mem_ref_loc): Simplify.
2477 (gather_mem_refs_stmt): Adjust.
2478 (sort_locs_in_loop_postorder_cmp): New function.
2479 (analyze_memory_references): Sort accesses_in_loop after
2480 loop postorder number.
2481 (find_ref_loc_in_loop_cmp): New function.
2482 (for_all_locs_in_loop): Find relevant cluster of locs in
2483 accesses_in_loop and iterate without recursion.
2484 (execute_sm): Avoid uninit warning.
2485 (struct ref_always_accessed): Simplify.
2486 (ref_always_accessed::operator ()): Likewise.
2487 (ref_always_accessed_p): Likewise.
2488 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2489 loop postorder numbers here.
2490 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2493 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2496 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2497 on decls for which assemble_alias has been called.
2499 2014-01-17 Nick Clifton <nickc@redhat.com>
2501 * config/msp430/msp430.opt: (mcpu): New option.
2502 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2503 (msp430_option_override): Parse target_cpu. If the MCU name
2504 matches a generic string, clear target_mcu.
2505 (msp430_attr): Allow numeric interrupt values up to 63.
2506 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2507 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2509 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2511 * config/msp430/msp430.md (popm): Use %J rather than %I.
2512 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2513 (addhi_cy_i): Use immediate_operand for operand 2.
2514 * doc/invoke.texi: Document -mcpu option.
2516 2014-01-17 Richard Biener <rguenther@suse.de>
2518 PR rtl-optimization/38518
2519 * df.h (df_analyze_loop): Declare.
2520 * df-core.c: Include cfgloop.h.
2521 (df_analyze_1): Split out main part of df_analyze.
2522 (df_analyze): Adjust.
2523 (loop_inverted_post_order_compute): New function.
2524 (loop_post_order_compute): Likewise.
2525 (df_analyze_loop): New function avoiding whole-function
2527 * loop-invariant.c (find_defs): Use df_analyze_loop.
2528 (find_invariants): Adjust.
2529 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2531 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2533 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2534 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2536 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2538 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2539 traversal when removing references.
2541 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2544 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2546 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2549 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2550 pushing a reload for an autoinc when we had previously reloaded an
2551 inner part of the address.
2553 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2555 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2557 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2558 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2559 when not giving up or versioning for alias only because of
2561 (vect_analyze_data_ref_dependences): Set to true.
2562 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2564 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2565 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2569 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2572 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2573 operand 0 predicate for gathers, use a new pseudo as subtarget.
2575 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2578 * lra-constraints.c (process_alt_operands): Add printing debug info.
2579 Check absence of input/output reloads for matched operands too.
2581 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2583 PR rtl-optimization/59835
2584 * ira.c (ira_init_register_move_cost): Increase cost for
2587 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2589 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2591 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2594 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2595 non-register objects. Use gen_(high/low)part more consistently.
2598 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2601 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2602 endian support, remove tests for WORDS_BIG_ENDIAN.
2603 (p8_mfvsrd_3_<mode>): Likewise.
2604 (reload_gpr_from_vsx<mode>): Likewise.
2605 (reload_gpr_from_vsxsf): Likewise.
2606 (p8_mfvsrd_4_disf): Likewise.
2608 2014-01-16 Richard Biener <rguenther@suse.de>
2610 PR rtl-optimization/46590
2611 * lcm.c (compute_antinout_edge): Use postorder iteration.
2612 (compute_laterin): Use inverted postorder iteration.
2614 2014-01-16 Nick Clifton <nickc@redhat.com>
2617 * varasm.c (output_constant): Return the number of bytes actually
2619 (output_constructor_array_range): Update the field size with the
2620 number of bytes emitted by output_constant.
2621 (output_constructor_regular_field): Likewise. Also do not
2622 complain if the total number of bytes emitted is now greater
2623 than the expected fieldpos.
2624 * output.h (output_constant): Update prototype and descriptive comment.
2626 2014-01-16 Marek Polacek <polacek@redhat.com>
2629 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2630 it is error_mark_node.
2632 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2634 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2635 VALID_AVX256_REG_OR_OI_MODE.
2637 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2639 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2640 current procedure should be profiled.
2642 2014-01-15 Andrew Pinski <apinski@cavium.com>
2644 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2645 of moving from/to the STACK_REG register class.
2647 2014-01-15 Richard Henderson <rth@redhat.com>
2650 * reginfo.c (global_regs_decl): Globalize.
2651 * rtl.h (global_regs_decl): Declare.
2652 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2653 reserved via global_regs.
2655 2014-01-15 Teresa Johnson <tejohnson@google.com>
2657 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2659 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2661 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2662 and vmulosh rather than call gen_vec_widen_smult_*.
2663 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2664 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2665 (vec_widen_smult_even_v16qi): Likewise.
2666 (vec_widen_umult_even_v8hi): Likewise.
2667 (vec_widen_smult_even_v8hi): Likewise.
2668 (vec_widen_umult_odd_v16qi): Likewise.
2669 (vec_widen_smult_odd_v16qi): Likewise.
2670 (vec_widen_umult_odd_v8hi): Likewise.
2671 (vec_widen_smult_odd_v8hi): Likewise.
2672 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2673 vmuloub rather than call gen_vec_widen_umult_*.
2674 (vec_widen_umult_lo_v16qi): Likewise.
2675 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2676 vmulosb rather than call gen_vec_widen_smult_*.
2677 (vec_widen_smult_lo_v16qi): Likewise.
2678 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2679 rather than call gen_vec_widen_umult_*.
2680 (vec_widen_umult_lo_v8hi): Likewise.
2681 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2682 rather than call gen_vec_widen_smult_*.
2683 (vec_widen_smult_lo_v8hi): Likewise.
2685 2014-01-15 Jeff Law <law@redhat.com>
2687 PR tree-optimization/59747
2688 * ree.c (find_and_remove_re): Properly handle case where a second
2689 eliminated extension requires widening a copy created for elimination
2690 of a prior extension.
2691 (combine_set_extension): Ensure that the number of hard regs needed
2692 for a destination register does not change when we widen it.
2694 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2696 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2697 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2698 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2699 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2700 (avr-*-rtems*): Likewise.
2701 (bfin*-rtems*): Likewise.
2702 (moxie-*-rtems*): Likewise.
2703 (h8300-*-rtems*): Likewise.
2704 (i[34567]86-*-rtems*): Likewise.
2705 (lm32-*-rtems*): Likewise.
2706 (m32r-*-rtems*): Likewise.
2707 (m68k-*-rtems*): Likewise.
2708 (microblaze*-*-rtems*): Likewise.
2709 (mips*-*-rtems*): Likewise.
2710 (powerpc-*-rtems*): Likewise.
2711 (sh-*-rtems*): Likewise.
2712 (sparc-*-rtems*): Likewise.
2713 (sparc64-*-rtems*): Likewise.
2714 (v850-*-rtems*): Likewise.
2715 (m32c-*-rtems*): Likewise.
2717 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2719 PR rtl-optimization/59511
2720 * ira.c (ira_init_register_move_cost): Use memory costs for some
2721 cases of register move cost calculations.
2722 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2723 instead of BB frequency.
2724 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2725 * lra-assigns.c (find_hard_regno_for): Ditto.
2727 2014-01-15 Richard Biener <rguenther@suse.de>
2729 PR tree-optimization/59822
2730 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2731 (vectorizable_load): Use it to hoist defs of uses of invariant
2732 loads out of the loop.
2734 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2735 Kugan Vivekanandarajah <kuganv@linaro.org>
2738 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2741 2014-01-15 Richard Biener <rguenther@suse.de>
2743 PR rtl-optimization/59802
2744 * lcm.c (compute_available): Use inverted postorder to seed
2745 the initial worklist.
2747 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2750 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2751 ADDR_REGS for invalid symrefs in non-PIC code.
2753 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2756 * builtins.c (determine_block_size): Initialize *probable_max_size
2757 even if len_rtx is CONST_INT.
2759 2014-01-14 Andrew Pinski <apinski@cavium.com>
2761 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2762 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2763 (cortexa53_tunings): Likewise.
2764 (aarch64_sched_issue_rate): New function.
2765 (TARGET_SCHED_ISSUE_RATE): Define.
2767 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2769 * ira-costs.c (find_costs_and_classes): Add missed
2770 ira_init_register_move_cost_if_necessary.
2772 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2775 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2777 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2780 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2781 to indicate if type is used for function return value. Warn ABI
2782 change if the vector mode isn't available for function return value.
2783 (ix86_function_arg_advance): Pass false to type_natural_mode.
2784 (ix86_function_arg): Likewise.
2785 (ix86_gimplify_va_arg): Likewise.
2786 (function_arg_32): Don't warn ABI change.
2787 (ix86_function_value): Pass true to type_natural_mode.
2788 (ix86_return_in_memory): Likewise.
2789 (ix86_struct_value_rtx): Removed.
2790 (TARGET_STRUCT_VALUE_RTX): Likewise.
2792 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2794 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2795 converting a conditional jump into a conditional return.
2797 2014-01-14 Richard Biener <rguenther@suse.de>
2799 PR tree-optimization/58921
2800 PR tree-optimization/59006
2801 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2802 hoisting invariant stmts.
2803 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2804 invariant loads on the preheader edge if possible.
2806 2014-01-14 Joey Ye <joey.ye@arm.com>
2808 * doc/plugin.texi (Building GCC plugins): Update to C++.
2810 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2812 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2813 (_mm_rcp28_round_ss): Ditto.
2814 (_mm_rsqrt28_round_sd): Ditto.
2815 (_mm_rsqrt28_round_ss): Ditto.
2816 (_mm_rcp28_sd): Ditto.
2817 (_mm_rcp28_ss): Ditto.
2818 (_mm_rsqrt28_sd): Ditto.
2819 (_mm_rsqrt28_ss): Ditto.
2820 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2821 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2822 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2823 (IX86_BUILTIN_RCP28SD): Ditto.
2824 (IX86_BUILTIN_RCP28SS): Ditto.
2825 (IX86_BUILTIN_RSQRT28SD): Ditto.
2826 (IX86_BUILTIN_RSQRT28SS): Ditto.
2827 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2828 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2829 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2830 (ix86_expand_special_args_builtin): Expand new FTYPE.
2831 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2832 (srcp14<mode>): Make insn unary.
2833 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2834 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2835 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2836 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2837 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2838 Fix rounding: make it SAE only.
2839 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2841 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2843 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2844 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2845 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2846 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2847 (round_saeonly_mask_scalar_operand4): Ditto.
2848 (round_saeonly_mask_scalar_op3): Ditto.
2849 (round_saeonly_mask_scalar_op4): Ditto.
2851 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2853 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2854 Implement -maltivec=be for vec_insert and vec_extract.
2856 2014-01-10 DJ Delorie <dj@redhat.com>
2858 * config/msp430/msp430.md (call_internal): Don't allow memory
2859 references with SP as the base register.
2860 (call_value_internal): Likewise.
2861 * config/msp430/constraints.md (Yc): New. For memory references
2862 that don't use SP as a base register.
2864 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2865 "an integer without a # prefix"
2866 * config/msp430/msp430.md (epilogue_helper): Use it.
2868 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2871 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2872 AVX512F gather builtins.
2873 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2874 on gather decls with INTEGER_TYPE masktype.
2875 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2876 directly into the builtin rather than hoisting it before loop.
2878 PR tree-optimization/59387
2879 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2880 (scev_const_prop): If folded_casts and type has undefined overflow,
2881 use force_gimple_operand instead of force_gimple_operand_gsi and
2882 for each added stmt if it is assign with
2883 arith_code_with_undefined_signed_overflow, call
2884 rewrite_to_defined_overflow.
2885 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2886 gimple-fold.h instead.
2887 (arith_code_with_undefined_signed_overflow,
2888 rewrite_to_defined_overflow): Moved to ...
2889 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2890 rewrite_to_defined_overflow): ... here. No longer static.
2891 Include gimplify-me.h.
2892 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2893 rewrite_to_defined_overflow): New prototypes.
2895 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2897 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2899 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2901 * builtins.c (get_object_alignment_2): Minor tweak.
2902 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2904 2014-01-13 Christian Bruel <christian.bruel@st.com>
2906 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2907 optimized non constant lengths.
2909 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2912 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2913 load as __atomic_load_N if possible.
2915 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2917 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2919 (rs6000_expand_builtin): Adjust call.
2921 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2924 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2925 * config/rs6000/rs6000.c: Include target-globals.h.
2926 (rs6000_set_current_function): Instead of doing target_reinit
2927 unconditionally, use save_target_globals_default_opts and
2928 restore_target_globals.
2930 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2932 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2933 (rs6000_expand_builtin): Handle mffs and mtfsf.
2934 (rs6000_init_builtins): Define mffs and mtfsf.
2935 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2936 (rs6000_mffs): New pattern.
2937 (rs6000_mtfsf): New pattern.
2939 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2941 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2942 Start narrowing with START. Apply candidate-use pair
2943 and check overall cost in narrowing.
2944 (iv_ca_prune): Pass new argument.
2946 2014-01-10 Jeff Law <law@redhat.com>
2949 * ree.c (combine_reaching_defs): Ensure the defining statement
2950 occurs before the extension when optimizing extensions with
2951 different source and destination hard registers.
2953 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2956 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2957 vtables into the type inheritance graph.
2959 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2961 PR rtl-optimization/59754
2962 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2963 modes in the REGNO != REGNO case.
2965 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2967 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2969 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2971 PR tree-optimization/59745
2972 * tree-predcom.c (tree_predictive_commoning_loop): Call
2973 free_affine_expand_cache if giving up because components is NULL.
2975 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2976 GC in payload of target_globals struct instead of allocating them on
2977 the heap and the larger structs separately using GC.
2978 * target-globals.h (struct target_globals): Make regs, hard_regs,
2979 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2980 of GTY((skip)) and change type to void *.
2981 (reset_target_globals): Cast loads from those fields to corresponding
2984 2014-01-10 Steve Ellcey <sellcey@mips.com>
2987 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2988 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2989 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2991 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2994 * aarch64-modes.def (CC_Zmode): New flags mode.
2995 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2996 represents an equality.
2997 (aarch64_get_condition_code): Handle CC_Zmode.
2998 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3000 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3002 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3003 extraction in good case.
3005 2014-01-10 Richard Biener <rguenther@suse.de>
3007 PR tree-optimization/59374
3008 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3009 checking after SLP discovery. Mark stmts not participating
3010 in any SLP instance properly.
3012 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3014 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3015 when handling a SET rtx.
3017 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3019 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3020 (cortex-a57): Likewise.
3021 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3023 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3025 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3026 non-iwmmxt builtins.
3028 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3032 * ipa-devirt.c record_target_from_binfo): Take as argument
3033 stack of binfos and lookup matching one for virtual inheritance.
3034 (possible_polymorphic_call_targets_1): Update.
3036 2014-01-10 Huacai Chen <chenhc@lemote.com>
3038 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3039 kernel strings for Loongson-2E/2F/3A.
3041 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3044 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3045 is_gimple_call before calling gimple_call_internal_p.
3047 2014-01-09 Steve Ellcey <sellcey@mips.com>
3049 * Makefile.in (TREE_FLOW_H): Remove.
3050 (TREE_SSA_H): Add file names from tree-flow.h.
3051 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3052 * tree.h: Remove tree-flow.h reference.
3053 * hash-table.h: Remove tree-flow.h reference.
3054 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3055 reference with tree-ssa-loop.h.
3057 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3059 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3060 default element-order behavior for -maltivec.
3061 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3062 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3063 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3064 when targeting big endian, at least for now.
3065 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3067 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3070 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3071 var satisfies use_register_for_decl, just take into account type
3072 alignment, rather than decl alignment.
3074 PR tree-optimization/59622
3075 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3076 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3077 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3078 Don't devirtualize for inplace at all. For targets.length () == 1,
3079 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3081 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3083 * config/i386/i386.md (cpu): Remove the unused btver1.
3085 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3087 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3089 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3092 * tree-core.h (struct target_globals): New forward declaration.
3093 (struct tree_target_option): Add globals field.
3094 * tree.h (TREE_TARGET_GLOBALS): Define.
3095 (prepare_target_option_nodes_for_pch): New prototype.
3096 * target-globals.h (struct target_globals): Define even if
3098 * tree.c (prepare_target_option_node_for_pch,
3099 prepare_target_option_nodes_for_pch): New functions.
3100 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3101 * config/i386/i386.c: Include target-globals.h.
3102 (ix86_set_current_function): Instead of doing target_reinit
3103 unconditionally, use save_target_globals_default_opts and
3104 restore_target_globals.
3106 2014-01-09 Richard Biener <rguenther@suse.de>
3108 PR tree-optimization/59715
3109 * tree-cfg.h (split_critical_edges): Declare.
3110 * tree-cfg.c (split_critical_edges): Export.
3111 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3113 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3115 * cfgexpand.c (expand_stack_vars): Optionally disable
3116 asan stack protection.
3117 (expand_used_vars): Likewise.
3118 (partition_stack_vars): Likewise.
3119 * asan.c (asan_emit_stack_protection): Optionally disable
3120 after return stack usage.
3121 (instrument_derefs): Optionally disable memory access instrumentation.
3122 (instrument_builtin_call): Likewise.
3123 (instrument_strlen_call): Likewise.
3124 (asan_protect_global): Optionally disable global variables protection.
3125 * doc/invoke.texi: Added doc for new options.
3126 * params.def: Added new options.
3127 * params.h: Likewise.
3129 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3131 PR rtl-optimization/59724
3132 * ifcvt.c (cond_exec_process_if_block): Don't call
3133 flow_find_head_matching_sequence with 0 longest_match.
3134 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3135 non-active insns if !stop_after.
3136 (try_head_merge_bb): Revert 2014-01-07 changes.
3138 2014-01-08 Jeff Law <law@redhat.com>
3140 * ree.c (get_sub_rtx): New function, extracted from...
3141 (merge_def_and_ext): Here.
3142 (combine_reaching_defs): Use get_sub_rtx.
3144 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3146 * cgraph.h (varpool_variable_node): Do not choke on null node.
3148 2014-01-08 Catherine Moore <clm@codesourcery.com>
3150 * config/mips/mips.md (simple_return): Attempt to use JRC
3152 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3154 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3156 PR rtl-optimization/59137
3157 * reorg.c (steal_delay_list_from_target): Call update_block for
3159 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3161 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3163 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3164 two duplicate entries.
3166 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3169 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3171 * config/mips/mips.c (mips_truncated_op_cost): New function.
3172 (mips_rtx_costs): Adjust test for BADDU.
3173 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3175 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3177 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3178 (*baddu_si): ...this new pattern.
3180 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3183 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3185 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3188 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3190 (expand_expr, expand_normal): Adjust.
3191 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3192 inner_reference_p. Use inner_reference_p to expand inner references.
3193 (store_expr): Adjust.
3194 * cfgexpand.c (expand_call_stmt): Adjust.
3196 2014-01-08 Rong Xu <xur@google.com>
3198 * gcov-io.c (gcov_var): Move from gcov-io.h.
3199 (gcov_position): Ditto.
3200 (gcov_is_error): Ditto.
3201 (gcov_rewrite): Ditto.
3202 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3203 only part to libgcc/libgcov.h.
3205 2014-01-08 Marek Polacek <polacek@redhat.com>
3208 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3210 2014-01-08 Marek Polacek <polacek@redhat.com>
3213 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3215 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3217 PR rtl-optimization/59649
3218 * stor-layout.c (get_mode_bounds): For BImode return
3219 0 and STORE_FLAG_VALUE.
3221 2014-01-08 Richard Biener <rguenther@suse.de>
3224 * gimple.h (is_gimple_builtin_call): Remove.
3225 (gimple_builtin_call_types_compatible_p): New.
3226 (gimple_call_builtin_p): New overload.
3227 * gimple.c (is_gimple_builtin_call): Remove.
3228 (validate_call): Rename to ...
3229 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3231 (validate_type): New static function.
3232 (gimple_call_builtin_p): New overload and adjust.
3233 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3234 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3235 (gimple_fold_stmt_to_constant_1): Likewise.
3236 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3238 2014-01-08 Richard Biener <rguenther@suse.de>
3241 * gimplify.c (gimplify_expr): Gimplify register-register type
3242 VIEW_CONVERT_EXPRs to separate stmts.
3244 2014-01-07 Jeff Law <law@redhat.com>
3247 * ree.c (combine_set_extension): Handle case where source
3248 and destination registers in an extension insn are different.
3249 (combine_reaching_defs): Allow source and destination registers
3250 in extension to be different under limited circumstances.
3251 (add_removable_extension): Remove restriction that the
3252 source and destination registers in the extension are the same.
3253 (find_and_remove_re): Emit a copy from the extension's
3254 destination to its source after the defining insn if
3255 the source and destination registers are different.
3258 * ifcvt.c (merge_if_block): If we are merging a block with more than
3259 one successor with a block with no successors, remove any BARRIER
3260 after the second block.
3262 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3264 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3266 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3269 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3270 for 14-bit register offsets when INT14_OK_STRICT is false.
3272 2014-01-07 Roland Stigge <stigge@antcom.de>
3273 Michael Meissner <meissner@linux.vnet.ibm.com>
3276 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3277 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3279 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3281 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3284 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3286 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3287 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3288 rtx is const0_rtx or not.
3290 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3293 * target-globals.c (save_target_globals): Remove this_fn_optab
3295 * toplev.c: Include optabs.h.
3296 (target_reinit): Temporarily restore the global options if another
3297 set of options are in force.
3299 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3301 PR rtl-optimization/58668
3302 * cfgcleanup.c (flow_find_cross_jump): Don't count
3303 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3304 to determine what is counted.
3305 (flow_find_head_matching_sequence): Use active_insn_p to determine
3307 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3309 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3310 determine what is counted.
3312 PR tree-optimization/59643
3313 * tree-predcom.c (split_data_refs_to_components): If one dr is
3314 read and one write, determine_offset fails and the write isn't
3315 in the bad component, just put the read into the bad component.
3317 2014-01-07 Mike Stump <mikestump@comcast.net>
3318 Jakub Jelinek <jakub@redhat.com>
3321 * tree-core.h (struct tree_optimization_option): Change optabs
3322 type from unsigned char * to void *.
3323 * optabs.c (init_tree_optimization_optabs): Adjust
3324 TREE_OPTIMIZATION_OPTABS initialization.
3326 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3329 * config/i386/i386.h (struct machine_function): Add
3330 no_drap_save_restore field.
3331 * config/i386/i386.c (ix86_save_reg): Use
3332 !cfun->machine->no_drap_save_restore instead of
3333 crtl->stack_realign_needed.
3334 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3335 this function clears frame_pointer_needed. Set
3336 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3337 and DRAP reg is needed.
3339 2014-01-06 Marek Polacek <polacek@redhat.com>
3342 * doc/implement-c.texi: Mention that other integer types are
3343 permitted as bit-field types in strictly conforming mode.
3345 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3347 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3350 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3352 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3354 2014-01-06 Martin Jambor <mjambor@suse.cz>
3357 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3359 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3361 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3365 * var-tracking.c (add_stores): Preserve the value of the source even if
3366 we don't record the store.
3368 2014-01-06 Terry Guo <terry.guo@arm.com>
3370 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3372 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3375 * config/darwin.c (darwin_function_section): Adjust return values to
3376 correspond to optimisation changes made in r206070.
3378 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3380 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3381 from prefetch_block tune setting.
3382 (nocona_cost): Correct size of prefetch block to 64.
3384 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3386 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3387 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3388 used to save the static chain register in the computation of the offset
3389 from which the FP registers need to be restored.
3391 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3393 PR tree-optimization/59519
3394 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3395 ICE if get_current_def (current_new_name) is already non-NULL, as long
3396 as it is a phi result of some other phi in *new_exit_bb that has
3399 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3400 or vmovdqu* for misaligned_operand.
3401 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3402 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3403 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3404 aligned_mem for AVX512F masked aligned load and store builtins and for
3407 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3409 PR tree-optimization/59651
3410 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3411 Address range for negative step should be added by TYPE_SIZE_UNIT.
3413 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3415 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3416 overlapping registers also for registers other than the stack pointer.
3418 2014-01-03 Marek Polacek <polacek@redhat.com>
3421 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3424 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3427 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3430 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3431 (push splitter): Use <P:MODE_SIZE> instead of
3432 GET_MODE_SIZE (<P:MODE>mode).
3433 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3434 (mov -1, reg peephole2): Likewise.
3435 * config/i386/sse.md (*mov<mode>_internal,
3436 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3437 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3438 *<code><mode>3, *andnot<mode>3<mask_name>,
3439 <mask_codefor><code><mode>3<mask_name>): Likewise.
3440 * config/i386/subst.md (mask_mode512bit_condition,
3441 sd_mask_mode512bit_condition): Likewise.
3443 2014-01-02 Xinliang David Li <davidxl@google.com>
3445 PR tree-optimization/59303
3446 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3447 (dump_predicates): Better output format.
3448 (pred_equal_p): New function.
3449 (is_neq_relop_p): Ditto.
3450 (is_neq_zero_form_p): Ditto.
3451 (pred_expr_equal_p): Ditto.
3452 (pred_neg_p): Ditto.
3453 (simplify_pred): Ditto.
3454 (simplify_preds_2): Ditto.
3455 (simplify_preds_3): Ditto.
3456 (simplify_preds_4): Ditto.
3457 (simplify_preds): Ditto.
3459 (push_to_worklist): Ditto.
3460 (get_pred_info_from_cmp): Ditto.
3461 (is_degenerated_phi): Ditto.
3462 (normalize_one_pred_1): Ditto.
3463 (normalize_one_pred): Ditto.
3464 (normalize_one_pred_chain): Ditto.
3465 (normalize_preds): Ditto.
3466 (normalize_cond_1): Remove function.
3467 (normalize_cond): Ditto.
3468 (is_gcond_subset_of): Ditto.
3469 (is_subset_of_any): Ditto.
3470 (is_or_set_subset_of): Ditto.
3471 (is_and_set_subset_of): Ditto.
3472 (is_norm_cond_subset_of): Ditto.
3473 (pred_chain_length_cmp): Ditto.
3474 (convert_control_dep_chain_into_preds): Type change.
3475 (find_predicates): Ditto.
3476 (find_def_preds): Ditto.
3477 (destroy_predicates_vecs): Ditto.
3478 (find_matching_predicates_in_rest_chains): Ditto.
3479 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3480 (is_pred_expr_subset): Ditto.
3481 (is_pred_chain_subset_of): Ditto.
3482 (is_included_in): Ditto.
3483 (is_superset_of): Ditto.
3485 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3487 Update copyright years.
3489 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3491 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3492 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3493 config/arc/arc.md, config/arc/arc.opt,
3494 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3495 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3496 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3497 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3498 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3499 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3500 vtable-verify.c, vtable-verify.h: Use the standard form for the
3503 2014-01-02 Tobias Burnus <burnus@net-b.de>
3505 * gcc.c (process_command): Update copyright notice dates.
3506 * gcov-dump.c: Ditto.
3508 * doc/cpp.texi: Bump @copying's copyright year.
3509 * doc/cppinternals.texi: Ditto.
3510 * doc/gcc.texi: Ditto.
3511 * doc/gccint.texi: Ditto.
3512 * doc/gcov.texi: Ditto.
3513 * doc/install.texi: Ditto.
3514 * doc/invoke.texi: Ditto.
3516 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3518 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3520 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3522 * config/i386/sse.md (*mov<mode>_internal): Guard
3523 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3525 PR rtl-optimization/59647
3526 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3527 new_rtx into UNSIGNED_FLOAT rtxes.
3529 Copyright (C) 2014 Free Software Foundation, Inc.
3531 Copying and distribution of this file, with or without modification,
3532 are permitted in any medium without royalty provided the copyright
3533 notice and this notice are preserved.