1 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2 Uros Bizjak <ubizjak@gmail.com>
5 * config/i386/driver-i386.c (host_detect_local_cpu): If
6 YMM state is not saved by the OS, also clear has_f16c. Move
7 CPUID 0x80000001 handling before YMM state saving checking.
9 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
11 PR rtl-optimization/58960
12 * haifa-sched.c (alloc_global_sched_pressure_data): New,
14 (sched_init): ... here.
15 (free_global_sched_pressure_data): New, factored out from ...
16 (sched_finish): ... here.
17 * sched-int.h (free_global_sched_pressure_data): Declare.
18 * sched-rgn.c (nr_regions_initial): New static global.
19 (haifa_find_rgns): Initialize it.
20 (schedule_region): Disable sched-pressure for the newly
23 2014-02-17 Richard Biener <rguenther@suse.de>
25 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
26 release SSA defs of pattern stmts.
28 2014-02-17 Richard Biener <rguenther@suse.de>
30 * tree-inline.c (expand_call_inline): Release the virtual
31 operand defined by the call we are about to inline.
33 2014-02-17 Richard Biener <rguenther@suse.de>
35 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
37 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
38 Ilya Tocar <ilya.tocar@intel.com>
40 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
41 arguments order in builtin.
42 (_mm512_permutexvar_epi64): Ditto.
43 (_mm512_mask_permutexvar_epi64): Ditto
44 (_mm512_maskz_permutexvar_epi32): Ditto
45 (_mm512_permutexvar_epi32): Ditto
46 (_mm512_mask_permutexvar_epi32): Ditto
48 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
50 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian
52 (p8_vmrgow): Likewise.
54 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
56 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
59 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
62 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
63 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
64 into 64-bit and 32-bit moves. On 64-bit moves, add support for
65 using direct move instructions on ISA 2.07. Also adjust
66 instruction length for 64-bit.
67 (mov<mode>_64bit, TFmode/TDmode): Likewise.
68 (mov<mode>_32bit, TFmode/TDmode): Likewise.
70 2014-02-15 Alan Modra <amodra@gmail.com>
74 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
75 find_replacement on parts of insn rtl that might be reloaded.
77 2014-02-15 Richard Biener <rguenther@suse.de>
79 PR tree-optimization/60183
80 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating
82 (tree_ssa_phiprop): Calculate and free post-dominators.
84 2014-02-14 Jeff Law <law@redhat.com>
86 PR rtl-optimization/60131
87 * ree.c (get_extended_src_reg): New function.
88 (combine_reaching_defs): Use it rather than assuming location
90 (find_and_remove_re): Verify first operand of extension is
91 a REG before adding the insns to the copy list.
93 2014-02-14 Roland McGrath <mcgrathr@google.com>
95 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
96 * configure: Regenerated.
97 * config.in: Regenerated.
98 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
101 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
102 Richard Earnshaw <rearnsha@arm.com>
104 PR rtl-optimization/59535
105 * lra-constraints.c (process_alt_operands): Encourage alternative
106 when unassigned pseudo class is superset of the alternative class.
107 (inherit_reload_reg): Don't inherit when optimizing for code size.
108 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
109 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
110 modes not less than 4 for Thumb1.
112 2014-02-14 Kyle McMartin <kyle@redhat.com>
115 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
117 2014-02-14 Richard Biener <rguenther@suse.de>
119 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
120 (get_frame_arg): Drop the assert with langhook types_compatible_p.
121 Do not strip INDIRECT_REFs.
123 2014-02-14 Richard Biener <rguenther@suse.de>
126 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
127 DECL_FUNCTION_SPECIFIC_TARGET.
128 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
129 * tree-streamer-out.c (pack_ts_target_option): Remove.
130 (streamer_pack_tree_bitfields): Do not stream
132 (write_ts_function_decl_tree_pointers): Do not stream
133 DECL_FUNCTION_SPECIFIC_TARGET.
134 * tree-streamer-in.c (unpack_ts_target_option): Remove.
135 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
136 (lto_input_ts_function_decl_tree_pointers): Do not stream
137 DECL_FUNCTION_SPECIFIC_TARGET.
139 2014-02-14 Jakub Jelinek <jakub@redhat.com>
141 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
143 (get_initial_def_for_induction, vectorizable_induction): Ignore
144 debug stmts when looking for exit_phi.
145 (vectorizable_live_operation): Fix up condition.
147 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
149 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
150 nreverse() because it changes the content of original tree list.
152 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
154 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
155 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
157 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
159 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
160 GNU coding standards.
162 2014-02-13 Jakub Jelinek <jakub@redhat.com>
165 * dwarf2out.c (gen_subprogram_die): Don't call
166 add_calling_convention_attribute if subr_die is old_die.
168 2014-02-13 Sharad Singhai <singhai@google.com>
170 * doc/optinfo.texi: Fix order of nodes.
172 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
174 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
175 operands[2], not operands[3].
177 2014-02-13 Richard Biener <rguenther@suse.de>
180 * doc/install.texi (ISL): Update recommended version to 0.12.2,
181 mention the possibility of an in-tree build.
182 (CLooG): Update recommended version to 0.18.1, mention the
183 possibility of an in-tree build and clarify that the ISL
184 bundled with CLooG does not work.
186 2014-02-13 Jakub Jelinek <jakub@redhat.com>
189 * expr.c (compress_float_constant): If x is a hard register,
190 extend into a pseudo and then move to x.
192 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
194 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
195 caused by bad second argument to warning_at() with -mhotpatch and
196 nested functions (e.g. with gfortran).
198 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
200 * opts.c (option_name): Remove "enabled by default" rider.
202 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
204 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
206 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
207 Uros Bizjak <ubizjak@gmail.com>
210 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
211 * configure: Regenerated.
213 2014-02-12 Richard Biener <rguenther@suse.de>
215 * vec.c (vec_prefix::calculate_allocation): Move as
216 inline variant to vec.h.
217 (vec_prefix::calculate_allocation_1): New out-of-line version.
218 * vec.h (vec_prefix::calculate_allocation_1): Declare.
219 (vec_prefix::m_has_auto_buf): Rename to ...
220 (vec_prefix::m_using_auto_storage): ... this.
221 (vec_prefix::calculate_allocation): Inline the easy cases
222 and dispatch to calculate_allocation_1 which doesn't need the
224 (va_heap::reserve): Use gcc_checking_assert.
225 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
226 m_using_auto_storage.
227 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
229 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
230 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
231 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
233 2014-02-12 Richard Biener <rguenther@suse.de>
235 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
236 when we found a dependence.
238 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
240 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
242 (maybe_fold_stmt): ... into this new function.
243 * omp-low.c (lower_omp): Update comment.
245 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
248 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
251 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
253 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
254 identifiers in comments.
255 (cortexa53_extra_costs): Likewise.
256 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
257 (cortexa7_extra_costs): Likewise.
258 (cortexa12_extra_costs): Likewise.
259 (cortexa15_extra_costs): Likewise.
260 (v7m_extra_costs): Likewise.
262 2014-02-12 Richard Biener <rguenther@suse.de>
265 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
266 of posix_memalign being successful.
267 (lower_stmt): Restrict lowering of posix_memalign to when
268 -ftree-bit-ccp is enabled.
270 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
272 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
274 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
276 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
278 PR rtl-optimization/60116
279 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
280 other_insn once the combination has been validated.
282 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
285 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
287 * ipa-devirt.c: Include demangle.h
288 (odr_violation_reported): New static variable.
289 (add_type_duplicate): Update odr_violations.
290 (maybe_record_node): Add completep parameter; update it.
291 (record_target_from_binfo): Add COMPLETEP parameter;
293 (possible_polymorphic_call_targets_1): Likewise.
294 (struct polymorphic_call_target_d): Add nonconstruction_targets;
295 rename FINAL to COMPLETE.
296 (record_targets_from_bases): Sanity check we found the binfo;
297 fix COMPLETEP updating.
298 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
299 parameter, fix computing of COMPLETEP.
300 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
301 at LTO time do demangling.
302 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
303 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
305 (gimple_get_virt_method_for_binfo): Likewise.
306 * gimple-fold.h (gimple_get_virt_method_for_binfo,
307 gimple_get_virt_method_for_vtable): Update prototypes.
309 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
312 * genautomata.c (add_presence_absence): Fix typo with
313 {final_}presence_list.
315 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
318 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
319 for VSX/Altivec vectors that land in GPR registers.
321 2014-02-11 Richard Henderson <rth@redhat.com>
322 Jakub Jelinek <jakub@redhat.com>
325 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
326 around drhs if type conversion to lacc->type is not useless.
328 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
330 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
332 (cortex-a57.cortex-a53): Likewise.
333 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
335 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
337 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
340 2014-02-11 Renlin Li <Renlin.Li@arm.com>
342 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
343 add_options_for_arm_vfp3.
345 2014-02-11 Jeff Law <law@redhat.com>
348 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
349 object with an undesirable mode.
351 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
354 * config/i386/sol2-9.h: New file.
355 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
356 *-*-solaris2.9*): Use it.
358 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
360 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
361 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
363 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
365 * config/microblaze/microblaze.c: Extend mcpu version format
367 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
369 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
371 2014-02-10 Richard Henderson <rth@redhat.com>
374 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
375 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
376 ms-abi vs -mno-accumulate-outgoing-args.
377 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
378 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
381 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
384 * cfgexpand.c (expand_asm_operands): Attach source location to
385 ASM_INPUT rtx objects.
386 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
388 2014-02-10 Nick Clifton <nickc@redhat.com>
390 * config/mn10300/mn10300.c (popcount): New function.
391 (mn10300_expand_prologue): Include saved registers in stack usage
394 2014-02-10 Jeff Law <law@redhat.com>
397 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
398 when changing the SET_DEST of a prior insn to avoid an input reload.
400 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
402 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
403 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
404 -mcall-openbsd, or -mcall-linux.
405 (CC1_ENDIAN_BIG_SPEC): Remove.
406 (CC1_ENDIAN_LITTLE_SPEC): Remove.
407 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
408 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
409 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
410 and %cc1_endian_default.
411 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
413 2014-02-10 Richard Biener <rguenther@suse.de>
415 PR tree-optimization/60115
416 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
417 MEM_REF handling. Properly verify that the accesses are not
418 out of the objects bound.
420 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
422 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
425 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
427 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
428 proper constants and fix formatting.
429 (possible_polymorphic_call_targets): Fix formatting.
431 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
432 Ilya Tocar <ilya.tocar@intel.com>
434 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
435 (_mm512_loadu_epi32): Renamed into...
436 (_mm512_loadu_si512): This.
437 (_mm512_storeu_epi32): Renamed into...
438 (_mm512_storeu_si512): This.
439 (_mm512_maskz_ceil_ps): Removed.
440 (_mm512_maskz_ceil_pd): Ditto.
441 (_mm512_maskz_floor_ps): Ditto.
442 (_mm512_maskz_floor_pd): Ditto.
443 (_mm512_floor_round_ps): Ditto.
444 (_mm512_floor_round_pd): Ditto.
445 (_mm512_ceil_round_ps): Ditto.
446 (_mm512_ceil_round_pd): Ditto.
447 (_mm512_mask_floor_round_ps): Ditto.
448 (_mm512_mask_floor_round_pd): Ditto.
449 (_mm512_mask_ceil_round_ps): Ditto.
450 (_mm512_mask_ceil_round_pd): Ditto.
451 (_mm512_maskz_floor_round_ps): Ditto.
452 (_mm512_maskz_floor_round_pd): Ditto.
453 (_mm512_maskz_ceil_round_ps): Ditto.
454 (_mm512_maskz_ceil_round_pd): Ditto.
455 (_mm512_expand_pd): Ditto.
456 (_mm512_expand_ps): Ditto.
457 * config/i386/i386.c (ix86_builtins): Remove
458 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
460 * config/i386/predicates.md (const1256_operand): New.
461 (const_1_to_2_operand): Ditto.
462 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
463 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
464 (*avx512pf_gatherpf<mode>sf): Ditto.
465 (avx512pf_gatherpf<mode>df): Ditto.
466 (*avx512pf_gatherpf<mode>df_mask): Ditto.
467 (*avx512pf_gatherpf<mode>df): Ditto.
468 (avx512pf_scatterpf<mode>sf): Ditto.
469 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
470 (*avx512pf_scatterpf<mode>sf): Ditto.
471 (avx512pf_scatterpf<mode>df): Ditto.
472 (*avx512pf_scatterpf<mode>df_mask): Ditto.
473 (*avx512pf_scatterpf<mode>df): Ditto.
474 (avx512f_expand<mode>): Removed.
475 (<shift_insn><mode>3<mask_name>): Change predicate type.
477 2014-02-08 Jakub Jelinek <jakub@redhat.com>
479 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
480 not at the end of datarefs vector use ordered_remove to avoid
481 reordering datarefs vector.
484 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
485 mark local addressable non-static vars as GOVD_PRIVATE
486 instead of GOVD_LOCAL.
487 * omp-low.c (lower_omp_for): Move gimple_bind_vars
488 and BLOCK_VARS of gimple_bind_block to new_stmt rather
492 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
493 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
494 assume_aligned or alloc_align attributes.
495 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
496 arguments. Handle also assume_aligned and alloc_align attributes.
497 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
498 calls to functions with assume_aligned or alloc_align attributes.
499 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
501 2014-02-08 Terry Guo <terry.guo@arm.com>
503 * doc/invoke.texi: Document ARM -march=armv7e-m.
505 2014-02-08 Jakub Jelinek <jakub@redhat.com>
507 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
508 flag on __cilkrts_rethrow builtin.
511 * ipa-cp.c (determine_versionability): Fail at -O0
512 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
513 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
516 2014-02-04 Jakub Jelinek <jakub@redhat.com>
519 * tree-inline.c (copy_forbidden): Fail for
520 __attribute__((optimize (0))) functions.
522 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
524 * varpool.c: Include pointer-set.h.
525 (varpool_remove_unreferenced_decls): Variables in other partitions
526 will not be output; be however careful to not lose information
529 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
531 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
532 lookup in the vtable constructor.
534 2014-02-07 Jeff Law <law@redhat.com>
537 * config/m68k/m68k.md (ashldi_extsi): Turn into a
538 define_insn_and_split.
540 * ipa-inline.c (inline_small_functions): Fix typos.
542 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
544 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
545 (s390_can_use_return_insn): Declare.
546 * config/s390/s390.h (EPILOGUE_USES): Define.
547 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
549 (s390_chunkify_start): Handle return JUMP_LABELs.
550 (s390_early_mach): Emit a main_pool instruction on the entry edge.
551 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
552 (s390_can_use_return_insn): New functions.
553 (s390_fix_long_loop_prediction): Handle conditional returns.
554 (TARGET_SET_UP_BY_PROLOGUE): Define.
555 * config/s390/s390.md (ANY_RETURN): New code iterator.
556 (*creturn, *csimple_return, return, simple_return): New patterns.
558 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
560 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
561 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
562 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
563 REG_CFA_RESTORE list when deciding not to restore a register.
565 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
567 * config/s390/s390.c: Include tree-pass.h and context.h.
568 (s390_early_mach): New function, split out from...
569 (s390_emit_prologue): ...here.
570 (pass_data_s390_early_mach): New pass structure.
571 (pass_s390_early_mach): New class.
572 (s390_option_override): Create and register early_mach pass.
575 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
577 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
578 to match for the exit block.
580 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
582 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
583 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
584 Reject misaligned operands.
586 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
588 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
590 2014-02-07 Richard Biener <rguenther@suse.de>
593 * gimple-low.c (lower_builtin_posix_memalign): New function.
594 (lower_stmt): Call it to lower posix_memalign in a way
595 to make alignment info accessible.
597 2014-02-07 Jakub Jelinek <jakub@redhat.com>
600 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
601 __builtin_setjmp_receiver.
603 2014-02-07 Richard Biener <rguenther@suse.de>
606 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
607 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
608 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
609 Handle BUILT_IN_POSIX_MEMALIGN.
610 (find_func_clobbers): Likewise.
611 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
612 (call_may_clobber_ref_p_1): Likewise.
614 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
617 * ipa-devirt.c (record_target_from_binfo): Remove overactive
620 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
623 * lto-cgraph.c (lto_output_node): Use
624 symtab_get_symbol_partitioning_class.
625 (lto_output_varpool_node): likewise.
626 (symtab_get_symbol_partitioning_class): Move here from
628 * cgraph.h (symbol_partitioning_class): Likewise.
629 (symtab_get_symbol_partitioning_class): Declare.
631 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
633 * ggc.h (ggc_internal_cleared_alloc): New macro.
634 * vec.h (vec_safe_copy): Handle memory stats.
635 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
636 * target-globals.c (save_target_globals): Likewise.
638 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
641 * expr.c (emit_move_resolve_push): Export; be bit more selective
642 on when to clear alias set.
643 * expr.h (emit_move_resolve_push): Declare.
644 * function.h (struct function): Add tail_call_marked.
645 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
646 * config/i386/i386-protos.h (ix86_expand_push): Remove.
647 * config/i386/i386.md (TImode move expander): De not call
649 (FP push expanders): Preserve memory attributes.
650 * config/i386/sse.md (push<mode>1): Remove.
651 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
652 (ix86_expand_push): Remove.
653 * config/i386/mmx.md (push<mode>1): Remove.
655 2014-02-06 Jakub Jelinek <jakub@redhat.com>
657 PR rtl-optimization/60030
658 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
659 lopart with paradoxical subreg before shifting it up by hprec.
661 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
663 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
664 Remove extra newline at end of file.
665 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
666 (arm_issue_rate): Handle cortexa57.
667 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
668 (cortex-a57.cortex-a53): Likewise.
670 2014-02-06 Jakub Jelinek <jakub@redhat.com>
673 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
674 don't record in REG_FRAME_RELATED_EXPR registers not set in that
676 (arm_expand_prologue): Adjust all callers.
677 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
678 info, registers also at the lowest numbered registers side. Use
679 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
683 * var-tracking.c (adjust_mems): Before adding a SET to
684 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
686 2014-02-06 Alan Modra <amodra@gmail.com>
689 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
690 change SDmode to DDmode when lra_in_progress.
692 2014-02-06 Jakub Jelinek <jakub@redhat.com>
695 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
696 free_data_ref on the dr first, and before goto again also set dr
697 to the next dr. For simd_lane_access, free old datarefs[i] before
698 overwriting it. For get_vectype_for_scalar_type failure, don't
699 free_data_ref if simd_lane_access.
701 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
704 * tree.h (opts_for_fn): New inline function.
705 (opt_for_fn): Define.
706 * config/i386/i386.c (ix86_function_regparm): Use
707 opt_for_fn (decl, optimize) instead of optimize.
709 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
711 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
712 for SYMBOL_REF in large memory model.
714 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
716 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
718 (cortex-a57): Likewise.
719 (cortex-a57.cortex-a53): Likewise.
721 2014-02-06 Yury Gribov <y.gribov@samsung.com>
722 Kugan Vivekanandarajah <kuganv@linaro.org>
724 * config/arm/arm.c (arm_vector_alignment_reachable): Check
726 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
728 2014-02-06 Richard Biener <rguenther@suse.de>
730 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
731 set_loop_copy and initialize_original_copy_tables.
733 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
735 * config/aarch64/aarch64-simd.md
736 (aarch64_ashr_simddi): Change QI to SI.
738 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
739 Jakub Jelinek <jakub@redhat.com>
742 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
745 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
747 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
748 CODE_FOR_altivec_vpku[hw]um to
749 CODE_FOR_altivec_vpku[hw]um_direct.
750 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
751 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
752 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
753 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
755 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
757 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
758 generation for -maltivec=be.
759 (altivec_vsumsws): Simplify redundant test.
761 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
763 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
764 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
765 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
766 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
768 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
770 (altivec_vpks<VI_char>ss): Likewise.
771 (altivec_vpks<VI_char>us): Likewise.
772 (altivec_vpku<VI_char>us): Likewise.
773 (altivec_vpku<VI_char>um): Likewise.
774 (altivec_vpku<VI_char>um_direct): New (copy of
775 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
777 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
778 target is little endian and -maltivec=be is not specified.
779 (*altivec_vupkhs<VU_char>_direct): New (copy of
780 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
781 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
782 target is little endian and -maltivec=be is not specified.
783 (*altivec_vupkls<VU_char>_direct): New (copy of
784 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
785 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
786 little endian and -maltivec=be is not specified.
787 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
788 little endian and -maltivec=be is not specified.
790 2014-02-05 Richard Henderson <rth@redhat.com>
793 * combine-stack-adj.c: Revert r206943.
794 * sched-int.h (struct deps_desc): Add last_args_size.
795 * sched-deps.c (init_deps): Initialize it.
796 (sched_analyze_insn): Add OUTPUT dependencies between insns that
797 contain REG_ARGS_SIZE notes.
799 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
801 * lto-cgraph.c (asm_nodes_output): Make global.
802 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
803 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
804 (driver_handle_option): Handle OPT_fwpa.
806 2014-02-05 Jakub Jelinek <jakub@redhat.com>
809 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
810 a comment typo and formatting issue. If odr_hash hasn't been
811 created, return vNULL and set *completep to false.
814 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
815 bb with no successors.
817 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
820 * doc/invoke.texi (-march): Clarify documentation for ARM.
824 2014-02-05 Richard Biener <rguenther@suse.de>
826 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
827 when not vectorizing because of too many alias checks.
828 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
829 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
831 2014-02-05 Nick Clifton <nickc@redhat.com>
833 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
834 accept extended registers in any mode when compiling for the MN10300.
836 2014-02-05 Yury Gribov <y.gribov@samsung.com>
838 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
839 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
840 sanitization attributes.
841 (can_inline_edge_p): Likewise.
842 (sanitize_attrs_match_for_inline_p): New function.
844 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
846 * ipa-prop.c (detect_type_change): Shor circuit testing of
847 type changes on THIS pointer.
849 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
852 * config/pa/pa.c (legitimize_tls_address): Return original address
853 if not passed a SYMBOL_REF rtx.
854 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
856 (pa_emit_move_sequence): Simplify TLS source operands.
857 (pa_legitimate_constant_p): Reject all TLS constants.
858 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
859 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
861 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
863 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
864 groups when we know they are controlled by LTO.
865 * varasm.c (default_binds_local_p_1): If object is in other partition,
866 it will be resolved locally.
868 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
870 * config/host-linux.c (linux_gt_pch_use_address): Don't
871 use SSIZE_MAX because it is not always defined.
873 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
876 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
877 threshold for pseudo splitting.
878 (update_ebb_live_info): Process call argument hard registers and
879 hard registers from insn definition too.
880 (max_small_class_regs_num): New constant.
881 (inherit_in_ebb): Update live hard regs through EBBs. Update
882 reloads_num only for small register classes. Don't split for
885 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
888 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
891 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
893 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
896 2014-02-04 Marek Polacek <polacek@redhat.com>
898 * gdbinit.in (pel): Define.
900 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
902 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
905 2014-02-04 Richard Biener <rguenther@suse.de>
908 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
909 in function context local.
910 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
911 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
912 similar to LTO_imported_decl_ref.
914 2014-02-04 Jakub Jelinek <jakub@redhat.com>
916 PR tree-optimization/60002
917 * cgraphclones.c (build_function_decl_skip_args): Clear
920 PR tree-optimization/60023
921 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
922 false to gsi_replace.
923 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
924 has been in some EH region and vec_stmt could throw, add
925 vec_stmt into the same EH region.
926 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
927 has no lhs, ignore it.
928 * internal-fn.c (expand_MASK_LOAD): Likewise.
931 * tree-inline.c (copy_forbidden): Fail for
932 __attribute__((optimize (0))) functions.
935 * omp-low.c (simd_clone_struct_copy): If from->inbranch
936 is set, copy one less argument.
937 (expand_simd_clones): Don't subtract clone_info->inbranch
938 from simd_clone_struct_alloc argument.
940 PR rtl-optimization/57915
941 * recog.c (simplify_while_replacing): If all unary/binary/relational
942 operation arguments are constant, attempt to simplify those.
945 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
946 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
948 2014-02-04 Richard Biener <rguenther@suse.de>
950 PR tree-optimization/60012
951 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
952 TBAA disambiguation to all DDRs.
954 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
957 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
958 (LINK_SPEC): Use it for -shared, -shared-libgcc.
960 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
963 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
965 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
967 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
968 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
970 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
973 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
974 to figure out targets of polymorphic calls with known decl.
975 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
976 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
977 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
978 (get_polymorphic_call_info): ... here.
979 (get_polymorphic_call_info_from_invariant): New function.
981 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
983 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
984 lookup via vtable pointer; check for type consistency
985 and turn inconsitent facts into UNREACHABLE.
986 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
987 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
988 type inconsistent querries; return UNREACHABLE instead.
990 2014-02-03 Richard Henderson <rth@twiddle.net>
993 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
994 already processed this node.
995 (normalize_one_pred_1): Pass along mark_set.
996 (normalize_one_pred): Create and destroy a pointer_set_t.
997 (normalize_one_pred_chain): Likewise.
999 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1001 PR gcov-profile/58602
1002 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1004 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1007 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1008 -fno-devirtualize; try to devirtualize by the knowledge of
1009 virtual table pointer given by aggregate propagation.
1010 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1011 (ipa_print_node_jump_functions): Dump also offset that
1012 is relevant for polymorphic calls.
1013 (determine_known_aggregate_parts): Add arg_type parameter; use it
1014 instead of determining the type from pointer type.
1015 (ipa_compute_jump_functions_for_edge): Update call of
1016 determine_known_aggregate_parts.
1017 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1018 (gimple_get_virt_method_for_binfo): ... here; simplify using
1019 vtable_pointer_value_to_vtable.
1020 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1021 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1022 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1023 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1025 (vtable_pointer_value_to_binfo): ... here.
1026 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1028 2014-02-03 Teresa Johnson <tejohnson@google.com>
1030 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1031 redef of outer loop index variable.
1033 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1037 * doc/extend.texi (Function Attributes): Typo.
1039 2014-02-03 Cong Hou <congh@google.com>
1041 PR tree-optimization/60000
1042 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1043 if the vectorized statement is a store. A store statement can only
1044 appear at the end of pattern statements.
1046 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1048 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1049 (ix86_option_override_internal): Default long double to 64-bit for
1050 32-bit Bionic and to 128-bit for 64-bit Bionic.
1052 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1053 TARGET_LONG_DOUBLE_128 is true.
1054 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1056 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1057 (mlong-double-64): Negate -mlong-double-128.
1058 (mlong-double-128): New option.
1060 * config/i386/i386-c.c (ix86_target_macros): Define
1061 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1063 * doc/invoke.texi: Document -mlong-double-128.
1065 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1067 PR rtl-optimization/60024
1068 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1070 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1072 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1074 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1076 PR rtl-optimization/57662
1077 * sel-sched.c (code_motion_path_driver): Do not mark already not
1078 existing blocks in the visiting bitmap.
1080 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1082 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1083 on the insn being emitted.
1085 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1086 Will Deacon <will.deacon@arm.com>
1088 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1090 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1092 * config/arm/arm-tables.opt: Regenerate.
1094 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1096 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1097 for vector types other than V16QImode.
1098 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1099 define_expand, and call altivec_expand_vec_perm_le when producing
1100 code with little endian element order.
1101 (*altivec_vperm_<mode>_internal): New insn having previous
1102 behavior of altivec_vperm_<mode>.
1103 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1104 altivec_expand_vec_perm_le when producing code with little endian
1106 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1107 behavior of altivec_vperm_<mode>_uns.
1109 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1111 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1112 (altivec_vsumsws): Add handling for -maltivec=be with a little
1114 (altivec_vsumsws_direct): New.
1115 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1116 gen_altivec_vsumsws.
1118 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1120 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1121 vtable_pointer_value_to_binfo): New functions.
1122 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1123 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1125 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1127 * config/nios2/nios2.md (load_got_register): Initialize GOT
1128 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1129 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1131 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1133 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1134 preserverd by passthrough, do not propagate the type.
1136 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1138 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1139 (mips_atomic_assign_expand_fenv): New function.
1140 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1142 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1144 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1145 (__builtin_mips_set_fcsr): Likewise.
1146 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1147 MIPS_USI_FTYPE_VOID.
1148 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1149 (mips16_expand_set_fcsr): Likewise.
1150 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1151 (mips16_set_fcsr_stub): Likewise.
1152 (mips16_get_fcsr_one_only_stub): New class.
1153 (mips16_set_fcsr_one_only_stub): Likewise.
1154 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1155 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1156 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1157 (hard_float): New availability predicate.
1158 (mips_builtins): Add get_fcsr and set_fcsr.
1159 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1160 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1161 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1162 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1163 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1166 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1168 * config/mips/mips.c (mips_one_only_stub): New class.
1169 (mips_need_mips16_rdhwr_p): Replace with...
1170 (mips16_rdhwr_stub): ...this new variable.
1171 (mips16_stub_call_address): New function.
1172 (mips16_rdhwr_one_only_stub): New class.
1173 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1174 (mips_output_mips16_rdhwr): Delete.
1175 (mips_finish_stub): New function.
1176 (mips_code_end): Use it to handle rdhwr stubs.
1178 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1181 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1182 when calculating size of integer atomic types.
1184 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1186 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1188 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1190 PR tree-optimization/60003
1191 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1192 * profile.c (branch_prob): Use gimple_call_builtin_p
1193 to check for BUILT_IN_SETJMP_RECEIVER.
1194 * tree-inline.c (copy_bb): Call notice_special_calls.
1196 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1199 * lra-constraints.c (process_alt_operands): Update reload_sum only
1202 2014-01-31 Richard Henderson <rth@redhat.com>
1205 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1206 until after else_eh is processed.
1208 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1210 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1211 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1212 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1213 in smmintrin.h, remove them.
1214 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1215 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1216 * config/i386/i386.md (ROUND_SAE): Fix value.
1217 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1218 (const48_operand): New.
1219 * config/i386/subst.md (round), (round_expand): Use
1220 const_4_or_8_to_11_operand.
1221 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1223 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1225 * config/i386/constraints.md (Yk): Swap meaning with k.
1226 * config/i386/i386.md (movhi_internal): Change Yk to k.
1227 (movqi_internal): Ditto.
1228 (*k<logic><mode>): Ditto.
1231 (kandn<mode>): Ditto.
1232 (*<code>hi_1): Ditto.
1233 (*<code>qi_1): Ditto.
1234 (kxnor<mode>): Ditto.
1235 (kortestzhi): Ditto.
1236 (kortestchi): Ditto.
1238 (*one_cmplhi2_1): Ditto.
1239 (*one_cmplqi2_1): Ditto.
1240 * config/i386/sse.md (): Change k to Yk.
1241 (avx512f_load<mode>_mask): Ditto.
1242 (avx512f_blendm<mode>): Ditto.
1243 (avx512f_store<mode>_mask): Ditto.
1244 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1245 (avx512f_storedqu<mode>_mask): Ditto.
1246 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1248 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1249 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1250 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1251 (avx512f_maskcmp<mode>3): Ditto.
1252 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1253 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1254 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1255 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1256 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1257 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1258 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1259 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1260 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1261 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1262 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1263 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1264 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1265 (vec_extract_lo_<mode>_maskm): Ditto.
1266 (vec_extract_hi_<mode>_maskm): Ditto.
1267 (avx512f_vternlog<mode>_mask): Ditto.
1268 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1269 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1270 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1271 (avx512f_<code>v8div16qi2_mask): Ditto.
1272 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1273 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1274 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1275 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1276 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1277 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1278 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1279 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1280 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1281 (avx512cd_maskb_vec_dupv8di): Ditto.
1282 (avx512cd_maskw_vec_dupv16si): Ditto.
1283 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1284 (avx512f_vpermi2var<mode>3_mask): Ditto.
1285 (avx512f_vpermi2var<mode>3_mask): Ditto.
1286 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1287 (*avx512f_gathersi<mode>): Ditto.
1288 (*avx512f_gathersi<mode>_2): Ditto.
1289 (*avx512f_gatherdi<mode>): Ditto.
1290 (*avx512f_gatherdi<mode>_2): Ditto.
1291 (*avx512f_scattersi<mode>): Ditto.
1292 (*avx512f_scatterdi<mode>): Ditto.
1293 (avx512f_compress<mode>_mask): Ditto.
1294 (avx512f_compressstore<mode>_mask): Ditto.
1295 (avx512f_expand<mode>_mask): Ditto.
1296 * config/i386/subst.md (mask): Change k to Yk.
1297 (mask_scalar_merge): Ditto.
1300 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1302 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1304 2014-01-31 Richard Biener <rguenther@suse.de>
1307 * builtins.c (fold_builtin_memory_op): Make sure to not
1308 use a floating-point mode or a boolean or enumeral type for
1311 2014-01-30 DJ Delorie <dj@redhat.com>
1313 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1314 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1315 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1316 whenever main() has an epilogue.
1318 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1320 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1321 unused variable "field".
1322 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1323 (vsx_mergeh_<mode>): Likewise.
1324 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1325 (altivec_vmrghh): Likewise.
1326 (altivec_vmrghw): Likewise.
1327 (altivec_vmrglb): Likewise.
1328 (altivec_vmrglh): Likewise.
1329 (altivec_vmrglw): Likewise.
1330 (altivec_vspltb): Add missing uses.
1331 (altivec_vsplth): Likewise.
1332 (altivec_vspltw): Likewise.
1333 (altivec_vspltsf): Likewise.
1335 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1338 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1339 frame related instructions.
1341 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1343 PR rtl-optimization/59959
1344 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1345 any reload of register whose subreg is invalid.
1347 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1349 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1350 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1351 Add missing return type - void.
1353 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1355 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1356 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1357 remove element index adjustment for endian (now handled in vsx.md
1359 (altivec_expand_vec_perm_const): Use
1360 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1361 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1362 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1363 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1364 define_expand and a new define_insn *altivec_vspltb_internal;
1365 adjust for -maltivec=be on a little endian target.
1366 (altivec_vspltb_direct): New.
1367 (altivec_vsplth): Divide into a define_expand and a new
1368 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1369 little endian target.
1370 (altivec_vsplth_direct): New.
1371 (altivec_vspltw): Divide into a define_expand and a new
1372 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1373 little endian target.
1374 (altivec_vspltw_direct): New.
1375 (altivec_vspltsf): Divide into a define_expand and a new
1376 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1377 a little endian target.
1379 2014-01-30 Richard Biener <rguenther@suse.de>
1381 PR tree-optimization/59993
1382 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1383 can propagate form the earlier stmt and avoid the transform
1384 when the intermediate result is needed.
1386 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1388 * README.Portability: Fix typo.
1390 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1392 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1393 comparison_operator with ordered_comparison_operator.
1395 2014-01-30 Nick Clifton <nickc@redhat.com>
1397 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1398 Rename to mn10300_store_multiple_regs.
1399 * config/mn10300/mn10300.c: Likewise.
1400 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1401 store_multiple_regs.
1402 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1403 Call mn10300_store_multiple_regs.
1405 2014-01-30 Nick Clifton <nickc@redhat.com>
1406 DJ Delorie <dj@redhat.com>
1408 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1409 %fp 2 to keep registers after it properly word-aligned.
1410 (rl78_alloc_physical_registers_umul): Handle the case where both
1411 input operands are the same.
1413 2014-01-30 Richard Biener <rguenther@suse.de>
1415 PR tree-optimization/59903
1416 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1419 2014-01-30 Jason Merrill <jason@redhat.com>
1422 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1425 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1427 2014-01-30 Richard Biener <rguenther@suse.de>
1429 PR tree-optimization/59951
1430 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1432 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1435 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1436 SFmode to DFmode case.
1438 2014-01-29 DJ Delorie <dj@redhat.com>
1440 * config/msp430/msp430.opt (-minrt): New.
1441 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1443 (ENDFILE_SPEC): Likewise.
1445 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1447 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1448 (estimate_function_body_sizes): Use it.
1450 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1453 * dwarf2out.c (is_cxx_auto): New.
1454 (is_base_type): Use it.
1455 (gen_type_die_with_usage): Likewise.
1457 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1459 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1460 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1461 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1462 -maltivec=be with LE targets.
1463 (vsx_mergeh_<mode>): Likewise.
1464 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1465 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1466 (altivec_vmrghb): Replace with define_expand and new
1467 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1468 (altivec_vmrghb_direct): New define_insn.
1469 (altivec_vmrghh): Replace with define_expand and new
1470 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1471 (altivec_vmrghh_direct): New define_insn.
1472 (altivec_vmrghw): Replace with define_expand and new
1473 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1474 (altivec_vmrghw_direct): New define_insn.
1475 (*altivec_vmrghsf): Adjust for endianness.
1476 (altivec_vmrglb): Replace with define_expand and new
1477 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1478 (altivec_vmrglb_direct): New define_insn.
1479 (altivec_vmrglh): Replace with define_expand and new
1480 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1481 (altivec_vmrglh_direct): New define_insn.
1482 (altivec_vmrglw): Replace with define_expand and new
1483 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1484 (altivec_vmrglw_direct): New define_insn.
1485 (*altivec_vmrglsf): Adjust for endianness.
1486 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1487 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1488 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1489 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1490 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1491 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1492 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1493 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1495 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1497 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1498 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1501 2014-01-29 Richard Biener <rguenther@suse.de>
1503 PR tree-optimization/58742
1504 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1505 associate_pointerplus_align.
1506 (associate_pointerplus_diff): New function.
1507 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1508 and associate_pointerplus_diff.
1510 2014-01-29 Richard Biener <rguenther@suse.de>
1512 * lto-streamer.h (LTO_major_version): Bump to 3.
1513 (LTO_minor_version): Reset to 0.
1515 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1517 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1518 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1519 (arm_file_start): Generate correct asm header for armv7ve.
1520 * config/arm/bpabi.h: Add multilib support for armv7ve.
1521 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1522 and cortex-a15 to armv7ve.
1523 * config/arm/t-aprofile: Add multilib support for armv7ve.
1524 * doc/invoke.texi: Document -march=armv7ve.
1526 2014-01-29 Richard Biener <rguenther@suse.de>
1528 PR tree-optimization/58742
1529 * tree-ssa-forwprop.c (associate_plusminus): Return true
1530 if we changed sth, defer EH cleanup to ...
1531 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1532 (simplify_mult): New function.
1534 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1537 PR tree-optimization/59920
1538 * tree.c (build_common_builtin_nodes): Remove
1539 __builtin_setjmp_dispatcher initialization.
1540 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1541 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1542 instead of gsi_after_labels + manually skipping debug stmts.
1543 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1544 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1545 * tree-inline.c (copy_edges_for_bb): Remove
1546 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1547 argument. Ignore computed_goto_p stmts. Don't call
1548 make_abnormal_goto_edges. If a call might need abnormal edges
1549 for non-local gotos, see if it already has an edge to
1550 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1551 with true argument, don't do anything then, otherwise add
1552 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1553 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1555 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1556 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1557 (lower_stmt): Don't set data->calls_builtin_setjmp.
1558 (lower_builtin_setjmp): Adjust comment.
1559 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1560 * tree-cfg.c (found_computed_goto): Remove.
1561 (factor_computed_gotos): Remove.
1562 (make_goto_expr_edges): Return bool, true for computed gotos.
1563 Don't call make_abnormal_goto_edges.
1564 (build_gimple_cfg): Don't set found_computed_goto, don't call
1565 factor_computed_gotos.
1566 (computed_goto_p): No longer static.
1567 (make_blocks): Don't set found_computed_goto.
1568 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1569 (make_edges): If make_goto_expr_edges returns true, push bb
1570 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1571 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1572 vector. Record mapping between bbs and OpenMP regions if there
1573 are any, adjust make_gimple_omp_edges caller. Call
1574 handle_abnormal_edges.
1575 (make_abnormal_goto_edges): Remove.
1576 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1577 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1578 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1579 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1580 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1581 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1582 filling *region also set *region_idx to (*region)->entry->index.
1585 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1586 For REGs set ORIGINAL_REGNO.
1588 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1590 * doc/md.texi: Mention that a target shouldn't implement
1591 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1594 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1596 PR tree-optimization/59594
1597 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1598 a copy of the datarefs vector rather than the vector itself.
1600 2014-01-28 Jason Merrill <jason@redhat.com>
1603 * dwarf2out.c (auto_die): New static.
1604 (gen_type_die_with_usage): Handle C++1y 'auto'.
1605 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1608 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1611 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1612 (SPEC_X32): Likewise.
1613 (SPEC_64): Likewise.
1614 * config/i386/i386.c (ix86_option_override_internal): Turn off
1615 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1617 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1618 * config/i386/i386.h (TARGET_16BIT): New macro.
1619 (TARGET_16BIT_P): Likewise.
1620 * config/i386/i386.opt: Add m16.
1621 * doc/invoke.texi: Document -m16.
1623 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1625 PR preprocessor/59935
1626 * input.c (location_get_source_line): Bail out on when line number
1627 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1629 2014-01-28 Richard Biener <rguenther@suse.de>
1631 PR tree-optimization/58742
1632 * tree-ssa-forwprop.c (associate_plusminus): Handle
1633 pointer subtraction of the form (T)(P + A) - (T)P.
1635 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1637 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1640 2014-01-28 Richard Biener <rguenther@suse.de>
1643 2014-01-28 Richard Biener <rguenther@suse.de>
1645 PR rtl-optimization/45364
1646 PR rtl-optimization/59890
1647 * var-tracking.c (local_get_addr_clear_given_value): Handle
1648 already cleared slot.
1649 (val_reset): Handle not allocated local_get_addr_cache.
1650 (vt_find_locations): Use post-order on the inverted CFG.
1652 2014-01-28 Richard Biener <rguenther@suse.de>
1654 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1656 2014-01-28 Richard Biener <rguenther@suse.de>
1658 PR rtl-optimization/45364
1659 PR rtl-optimization/59890
1660 * var-tracking.c (local_get_addr_clear_given_value): Handle
1661 already cleared slot.
1662 (val_reset): Handle not allocated local_get_addr_cache.
1663 (vt_find_locations): Use post-order on the inverted CFG.
1665 2014-01-28 Alan Modra <amodra@gmail.com>
1667 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1668 * configure.ac <recursive call for build != host>: Define
1669 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1670 and LD_FOR_BUILD too.
1671 * configure: Regenerate.
1673 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1675 * config/i386/i386.c (get_builtin_code_for_version): Separate
1676 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1677 Broadwell from Haswell.
1679 2014-01-27 Steve Ellcey <sellcey@mips.com>
1681 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1682 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1683 * config/mips/mips.c (mips_option_override): Change setting
1685 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1686 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1687 Change from Mask to Var.
1689 2014-01-27 Jeff Law <law@redhat.com>
1691 * ipa-inline.c (inline_small_functions): Fix typo.
1693 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1695 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1696 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1697 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1698 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1699 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1700 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1701 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1702 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1703 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1704 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1705 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1706 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1707 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1708 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1709 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1710 (_mm512_storeu_epi64): Ditto.
1711 (_mm512_cmpge_epi32_mask): Ditto.
1712 (_mm512_cmpge_epu32_mask): Ditto.
1713 (_mm512_cmpge_epi64_mask): Ditto.
1714 (_mm512_cmpge_epu64_mask): Ditto.
1715 (_mm512_cmple_epi32_mask): Ditto.
1716 (_mm512_cmple_epu32_mask): Ditto.
1717 (_mm512_cmple_epi64_mask): Ditto.
1718 (_mm512_cmple_epu64_mask): Ditto.
1719 (_mm512_cmplt_epi32_mask): Ditto.
1720 (_mm512_cmplt_epu32_mask): Ditto.
1721 (_mm512_cmplt_epi64_mask): Ditto.
1722 (_mm512_cmplt_epu64_mask): Ditto.
1723 (_mm512_cmpneq_epi32_mask): Ditto.
1724 (_mm512_cmpneq_epu32_mask): Ditto.
1725 (_mm512_cmpneq_epi64_mask): Ditto.
1726 (_mm512_cmpneq_epu64_mask): Ditto.
1727 (_mm512_expand_pd): Ditto.
1728 (_mm512_expand_ps): Ditto.
1729 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1730 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1731 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1732 * config/i386/i386.c (ix86_builtins): Add
1733 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1734 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1735 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1736 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1737 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1738 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1739 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1740 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1741 IX86_BUILTIN_PMOVUSQW512_MEM.
1742 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1743 __builtin_ia32_pmovsqd512mem_mask,
1744 __builtin_ia32_pmovqd512mem_mask,
1745 __builtin_ia32_pmovusqw512mem_mask,
1746 __builtin_ia32_pmovsqw512mem_mask,
1747 __builtin_ia32_pmovqw512mem_mask,
1748 __builtin_ia32_pmovusdw512mem_mask,
1749 __builtin_ia32_pmovsdw512mem_mask,
1750 __builtin_ia32_pmovdw512mem_mask,
1751 __builtin_ia32_pmovqb512mem_mask,
1752 __builtin_ia32_pmovusqb512mem_mask,
1753 __builtin_ia32_pmovsqb512mem_mask,
1754 __builtin_ia32_pmovusdb512mem_mask,
1755 __builtin_ia32_pmovsdb512mem_mask,
1756 __builtin_ia32_pmovdb512mem_mask.
1757 (bdesc_args): Add __builtin_ia32_expanddf512,
1758 __builtin_ia32_expandsf512.
1759 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1760 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1761 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1762 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1763 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1764 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1765 (avx512f_<code>v8div16qi2_mask_store): This.
1766 (avx512f_expand<mode>): New.
1768 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1770 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1772 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1773 (_mm512_prefetch_i32scatter_pd): Ditto.
1774 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1775 (_mm512_prefetch_i64scatter_pd): Ditto.
1776 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1777 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1778 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1779 (_mm512_prefetch_i32scatter_ps): Ditto.
1780 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1781 (_mm512_prefetch_i64scatter_ps): Ditto.
1782 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1783 * config/i386/i386-builtin-types.def: Define
1784 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1785 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1786 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1787 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1788 IX86_BUILTIN_SCATTERPFQPD.
1789 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1790 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1791 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1792 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1793 __builtin_ia32_scatterpfqps.
1794 (ix86_expand_builtin): Expand new built-ins.
1795 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1796 fix memory access data type.
1797 (*avx512pf_gatherpf<mode>_mask): Ditto.
1798 (*avx512pf_gatherpf<mode>): Ditto.
1799 (avx512pf_scatterpf<mode>): Ditto.
1800 (*avx512pf_scatterpf<mode>_mask): Ditto.
1801 (*avx512pf_scatterpf<mode>): Ditto.
1802 (GATHER_SCATTER_SF_MEM_MODE): New.
1803 (avx512pf_gatherpf<mode>df): Ditto.
1804 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1805 (*avx512pf_scatterpf<mode>df): Ditto.
1807 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1810 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1811 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1814 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1816 * common/config/arm/arm-common.c
1817 (arm_rewrite_mcpu): Handle multiple names.
1819 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1821 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1823 * gimple-builder.h (create_gimple_tmp): Delete.
1825 2014-01-27 Christian Bruel <christian.bruel@st.com>
1827 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1830 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1832 * config/pa/pa.md (call): Generate indirect long calls to non-local
1833 functions when outputing 32-bit code.
1834 (call_value): Likewise except for special call to buggy powf function.
1836 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1837 portable runtime and PIC indirect calls.
1838 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1839 and PIC call sequences. Use ldo instead of blr to set return register
1840 in PIC call sequence.
1842 2014-01-25 Walter Lee <walt@tilera.com>
1844 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1845 avoid clobbering a live register.
1847 2014-01-25 Walter Lee <walt@tilera.com>
1849 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1850 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1851 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1852 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1854 2014-01-25 Walter Lee <walt@tilera.com>
1856 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1857 arguments on even registers.
1858 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1860 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1861 (BIGGEST_ALIGNMENT): Ditto.
1862 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1864 2014-01-25 Walter Lee <walt@tilera.com>
1866 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1867 insns before bundling.
1868 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1870 2014-01-25 Walter Lee <walt@tilera.com>
1872 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1873 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1874 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1876 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1878 * config/mips/constraints.md (kl): Delete.
1879 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1880 define expands, using...
1881 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1882 instructions for MIPS16.
1883 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1884 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1886 2014-01-25 Walter Lee <walt@tilera.com>
1888 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1892 2014-01-25 Walter Lee <walt@tilera.com>
1894 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1895 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1897 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1899 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1902 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1904 * print-rtl.c (in_call_function_usage): New var.
1905 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1906 EXPR_LIST mode as mode and not as reg note name.
1909 * cfgloopmanip.c (copy_loop_info): If
1910 loop->warned_aggressive_loop_optimizations, make sure
1911 the flag is set in target loop too.
1913 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1915 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1917 * builtins.def: Likewise.
1918 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1919 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1920 * ira.c (ira_setup_eliminable_regset): Likewise.
1921 * omp-low.c (gate_expand_omp): Likewise.
1922 (execute_lower_omp): Likewise.
1923 (diagnose_sb_0): Likewise.
1924 (gate_diagnose_omp_blocks): Likewise.
1925 (simd_clone_clauses_extract): Likewise.
1928 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1930 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1931 correction for little endian...
1932 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1935 2014-01-24 Jeff Law <law@redhat.com>
1937 PR tree-optimization/59919
1938 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1939 for non-returning calls.
1941 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1943 * common/config/aarch64/aarch64-common.c
1944 (aarch64_rewrite_mcpu): Handle multiple names.
1945 * config/aarch64/aarch64.h
1946 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1948 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1950 * input.c (add_file_to_cache_tab): Handle the case where fopen
1953 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1956 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1957 from push operand if code of push isn't PRE_DEC.
1959 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1962 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1963 -mquad-memory-atomic. Update -mquad-memory documentation to say
1964 it is only used for non-atomic loads/stores.
1966 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1967 -mquad-memory or -mquad-memory-atomic switches.
1969 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1970 -mquad-memory-atomic to ISA 2.07 support.
1972 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1973 to separate support of normal quad word memory operations (ldq, stq)
1974 from the atomic quad word memory operations.
1976 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1977 support to separate non-atomic quad word operations from atomic
1978 quad word operations. Disable non-atomic quad word operations in
1979 little endian mode so that we don't have to swap words after the
1980 load and before the store.
1981 (quad_load_store_p): Add comment about atomic quad word support.
1982 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1983 options printed with -mdebug=reg.
1985 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1986 -mquad-memory-atomic as the test for whether we have quad word
1987 atomic instructions.
1988 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1989 or -mp8-vector are used, allow byte/half-word atomic operations.
1991 * config/rs6000/sync.md (load_lockedti): Insure that the address
1992 is a proper indexed or indirect address for the lqarx instruction.
1993 On little endian systems, swap the hi/lo registers after the lqarx
1995 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1996 insure the address is valid for the lqarx instruction.
1997 (store_conditionalti): Insure that the address is a proper indexed
1998 or indirect address for the stqcrx. instruction. On little endian
1999 systems, swap the hi/lo registers before doing the stqcrx.
2001 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2002 insure the address is valid for the stqcrx. instruction.
2004 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2005 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2006 type of quad memory support is available.
2008 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2011 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2012 there is a danger of looping.
2014 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2017 force flag_ira_loop_pressure if set via command line.
2019 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2021 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2022 (ashr_simd): New builtin handling DI mode.
2023 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2024 (aarch64_sshr_simddi): New match pattern.
2025 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2026 (vshrd_n_s64): Likewise.
2027 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2029 2014-01-23 Nick Clifton <nickc@redhat.com>
2031 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2032 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2033 favour of mcu specific scripts.
2034 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2037 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2038 Alex Velenko <Alex.Velenko@arm.com>
2040 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2041 (vaddv_s16): Likewise.
2042 (vaddv_s32): Likewise.
2043 (vaddv_u8): Likewise.
2044 (vaddv_u16): Likewise.
2045 (vaddv_u32): Likewise.
2046 (vaddvq_s8): Likewise.
2047 (vaddvq_s16): Likewise.
2048 (vaddvq_s32): Likewise.
2049 (vaddvq_s64): Likewise.
2050 (vaddvq_u8): Likewise.
2051 (vaddvq_u16): Likewise.
2052 (vaddvq_u32): Likewise.
2053 (vaddvq_u64): Likewise.
2054 (vaddv_f32): Likewise.
2055 (vaddvq_f32): Likewise.
2056 (vaddvq_f64): Likewise.
2057 (vmaxv_f32): Likewise.
2058 (vmaxv_s8): Likewise.
2059 (vmaxv_s16): Likewise.
2060 (vmaxv_s32): Likewise.
2061 (vmaxv_u8): Likewise.
2062 (vmaxv_u16): Likewise.
2063 (vmaxv_u32): Likewise.
2064 (vmaxvq_f32): Likewise.
2065 (vmaxvq_f64): Likewise.
2066 (vmaxvq_s8): Likewise.
2067 (vmaxvq_s16): Likewise.
2068 (vmaxvq_s32): Likewise.
2069 (vmaxvq_u8): Likewise.
2070 (vmaxvq_u16): Likewise.
2071 (vmaxvq_u32): Likewise.
2072 (vmaxnmv_f32): Likewise.
2073 (vmaxnmvq_f32): Likewise.
2074 (vmaxnmvq_f64): Likewise.
2075 (vminv_f32): Likewise.
2076 (vminv_s8): Likewise.
2077 (vminv_s16): Likewise.
2078 (vminv_s32): Likewise.
2079 (vminv_u8): Likewise.
2080 (vminv_u16): Likewise.
2081 (vminv_u32): Likewise.
2082 (vminvq_f32): Likewise.
2083 (vminvq_f64): Likewise.
2084 (vminvq_s8): Likewise.
2085 (vminvq_s16): Likewise.
2086 (vminvq_s32): Likewise.
2087 (vminvq_u8): Likewise.
2088 (vminvq_u16): Likewise.
2089 (vminvq_u32): Likewise.
2090 (vminnmv_f32): Likewise.
2091 (vminnmvq_f32): Likewise.
2092 (vminnmvq_f64): Likewise.
2094 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2096 * config/aarch64/aarch64-simd.md
2097 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2098 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2099 (*aarch64_mul3_elt<mode>): Likewise.
2100 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2101 (*aarch64_mul3_elt_to_64v2df): Likewise.
2102 (*aarch64_mla_elt<mode>): Likewise.
2103 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2104 (*aarch64_mls_elt<mode>): Likewise.
2105 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2106 (*aarch64_fma4_elt<mode>): Likewise.
2107 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2108 (*aarch64_fma4_elt_to_64v2df): Likewise.
2109 (*aarch64_fnma4_elt<mode>): Likewise.
2110 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2111 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2112 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2113 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2114 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2115 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2116 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2117 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2118 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2120 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2122 * config/aarch64/aarch64-simd.md
2123 (aarch64_be_checked_get_lane<mode>): New define_expand.
2124 * config/aarch64/aarch64-simd-builtins.def
2125 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2126 New builtin definition.
2127 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2128 Use new safe be builtin.
2130 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2132 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2134 (aarch64_be_st1<mode>): Likewise.
2135 (aarch_ld1<VALL:mode>): Define_expand modified.
2136 (aarch_st1<VALL:mode>): Likewise.
2137 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2138 (UNSPEC_ST1): Likewise.
2140 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2142 * config/microblaze/microblaze.md: Add trap insn and attribute
2144 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2146 PR preprocessor/58580
2147 * input.h (location_get_source_line): Take an additional line_size
2149 (void diagnostics_file_cache_fini): Declare new function.
2150 * input.c (struct fcache): New type.
2151 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2152 New static constants.
2153 (diagnostic_file_cache_init, total_lines_num)
2154 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2155 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2156 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2157 (get_next_line, read_next_line, goto_next_line, read_line_num):
2158 New static function definitions.
2159 (diagnostic_file_cache_fini): New function.
2160 (location_get_source_line): Take an additional output line_len
2161 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2163 * diagnostic.c (diagnostic_finish): Call
2164 diagnostic_file_cache_fini.
2165 (adjust_line): Take an additional input parameter for the length
2166 of the line, rather than calculating it with strlen.
2167 (diagnostic_show_locus): Adjust the use of
2168 location_get_source_line and adjust_line with respect to their new
2169 signature. While displaying a line now, do not stop at the first
2170 null byte. Rather, display the zero byte as a space and keep
2171 going until we reach the size of the line.
2172 * Makefile.in: Add vec.o to OBJS-libcommon
2174 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2175 Ilya Tocar <ilya.tocar@intel.com>
2177 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2178 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2179 (__builtin_ia32_kmov16): Ditto.
2180 * config/i386/i386.md (UNSPEC_KMOV): New.
2183 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2185 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2186 (_mm512_storeu_si512): Ditto.
2188 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2191 * rtl.h (get_referenced_operands): Declare.
2192 * recog.c (get_referenced_operands): New function.
2193 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2194 operands have been referenced when recording LO_SUM references.
2196 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2198 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2200 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2202 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2203 Enable for generic and recent AMD targets.
2205 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2207 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2208 ARG_SIZE note when adjustment was eliminated.
2210 2014-01-22 Jeff Law <law@redhat.com>
2212 PR tree-optimization/59597
2213 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2214 in file. Accept new argument REGISTERING and use it to modify
2215 dump output appropriately.
2216 (register_jump_thread): Corresponding changes.
2217 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2218 thread paths involving joiner blocks. Add code to dump cancelled
2219 jump threading paths.
2221 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2223 PR rtl-optimization/59477
2224 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2225 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2227 2014-01-22 Tom Tromey <tromey@redhat.com>
2229 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2231 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2233 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2235 PR rtl-optimization/59896
2236 * lra-constraints.c (process_alt_operands): Check unused note for
2237 matched operands of insn with no output reloads.
2239 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2241 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2242 (mips_move_from_gpr_cost): Likewise.
2244 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2246 PR rtl-optimization/59858
2247 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2248 ira_class_hard_regs_num.
2249 (process_alt_operands): Increase reject for dying matched operand.
2251 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2254 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2255 smaller than size, perform several stores or loads and stores
2256 at dst + count - size to store or copy all of size bytes, rather
2257 than just last modesize bytes.
2259 2014-01-20 DJ Delorie <dj@redhat.com>
2261 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2262 that CLOBBERs are REGs before propogating their values.
2264 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2267 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2268 (cgraph_inline_failed_type): New function.
2269 * cgraph.h (DEFCIFCODE): Add type.
2270 (cgraph_inline_failed_type_t): New enum.
2271 (cgraph_inline_failed_type): New prototype.
2272 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2273 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2274 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2275 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2276 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2277 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2278 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2279 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2280 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2281 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2282 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2283 OPTIMIZATION_MISMATCH.
2284 * tree-inline.c (expand_call_inline): Emit errors during
2285 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2287 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2290 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2291 mode attribute in insn output.
2293 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2295 * output.h (output_constant): Delete.
2296 * varasm.c (output_constant): Make private.
2298 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2300 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2302 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2305 * tree.h (fold_builtin_strcat): New prototype.
2306 * builtins.c (fold_builtin_strcat): No longer static. Add len
2307 argument, if non-NULL, don't call c_strlen. Optimize
2308 directly into __builtin_memcpy instead of __builtin_strcpy.
2309 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2310 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2312 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2314 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2315 for SImode_address_operand operands, having only a REG argument.
2317 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2319 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2320 loader name using mbig-endian.
2321 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2323 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2325 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2329 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2331 * config/aarch64/aarch64-protos.h
2332 (aarch64_cannot_change_mode_class_ptr): Declare.
2333 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2334 aarch64_cannot_change_mode_class_ptr): New.
2335 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2336 backend hook aarch64_cannot_change_mode_class.
2338 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2340 * common/config/aarch64/aarch64-common.c
2341 (aarch64_handle_option): Don't handle any option order logic here.
2342 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2343 selected_cpu, warn on architecture version mismatch.
2344 (aarch64_override_options): Fix parsing order for option strings.
2346 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2347 Iain Sandoe <iain@codesourcery.com>
2350 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2351 warning. Amend comment to reflect current functionality.
2353 2014-01-20 Richard Biener <rguenther@suse.de>
2356 * builtins.c (fold_builtin_strcat): Remove case better handled
2357 by tree-ssa-strlen.c.
2359 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2361 * config/aarch64/aarch64.opt
2362 (mcpu, march, mtune): Make case-insensitive.
2364 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2367 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2368 if operands[1] is a REG or ZERO_EXTEND of a REG.
2370 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2372 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2374 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2376 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2377 long non-pic millicode calls.
2379 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2381 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2383 2014-01-19 Kito Cheng <kito@0xlab.org>
2385 * builtins.c (expand_movstr): Check movstr expand done or fail.
2387 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2388 H.J. Lu <hongjiu.lu@intel.com>
2391 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2392 to DImode for zero-extended addresses.
2394 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2396 PR rtl-optimization/57763
2397 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2398 on the new indirect jump_insn and increment LABEL_NUSES (label).
2400 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2404 * config.gcc (x86_archs): New variable.
2405 (x86_64_archs): Likewise.
2406 (x86_cpus): Likewise.
2407 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2408 --with-arch/--with-cpu= options.
2409 Support --with-arch=/--with-cpu={nehalem,westmere,
2410 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2412 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2414 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2415 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2417 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2419 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2421 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2424 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2425 clear cpp_get_options (parse_in)->warn_unused_macros for
2426 ix86_target_macros_internal with cpp_define.
2428 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2430 * jump.c (delete_related_insns): Keep (use (insn))s.
2431 * reorg.c (redundant_insn): Check for barriers too.
2433 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2435 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2437 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2439 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2440 call to $$dyncall when TARGET_LONG_CALLS is true.
2442 2014-01-17 Jeff Law <law@redhat.com>
2444 * ree.c (combine_set_extension): Temporarily disable test for
2445 changing number of hard registers.
2447 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2450 * ipa-inline-analysis.c (inline_free_summary):
2451 Do not free summary of aliases.
2453 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2456 * gimplify.c (gimplify_expr): Use create_tmp_var
2457 instead of create_tmp_var_raw. If cond doesn't have
2458 integral type, don't add the IFN_ANNOTATE builtin at all.
2460 2014-01-17 Martin Jambor <mjambor@suse.cz>
2463 * ipa-cp.c (prev_edge_clone): New variable.
2464 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2465 Also resize prev_edge_clone vector.
2466 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2467 (ipcp_edge_removal_hook): New function.
2468 (ipcp_driver): Register ipcp_edge_removal_hook.
2470 2014-01-17 Andrew Pinski <apinski@cavium.com>
2471 Steve Ellcey <sellcey@mips.com>
2474 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2477 2014-01-17 Jeff Law <law@redhat.com>
2480 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2481 so that pass_ccp runs first.
2483 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2485 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2486 (ix86_adjust_cost): Use !TARGET_XXX.
2487 (do_reorder_for_imul): Likewise.
2488 (swap_top_of_ready_list): Likewise.
2489 (ix86_sched_reorder): Likewise.
2491 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2493 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2494 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2495 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2496 (intel_memset): New. Duplicate slm_memset.
2497 (intel_cost): New. Duplicate slm_cost.
2498 (m_INTEL): New macro.
2499 (processor_target_table): Add "intel".
2500 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2501 with PROCESSOR_INTEL for "intel".
2502 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2503 PROCESSOR_SILVERMONT.
2504 (ix86_issue_rate): Likewise.
2505 (ix86_adjust_cost): Likewise.
2506 (ia32_multipass_dfa_lookahead): Likewise.
2507 (swap_top_of_ready_list): Likewise.
2508 (ix86_sched_reorder): Likewise.
2509 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2510 instead of TARGET_OPT_AGU.
2511 * config/i386/i386.h (TARGET_INTEL): New.
2512 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2513 (processor_type): Add PROCESSOR_INTEL.
2514 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2515 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2517 2014-01-17 Marek Polacek <polacek@redhat.com>
2520 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2523 2014-01-17 Richard Biener <rguenther@suse.de>
2525 PR tree-optimization/46590
2526 * opts.c (default_options_table): Add entries for
2527 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2528 all enabled at -O1 but not for -Og.
2529 * common.opt (fbranch-count-reg): Remove Init(1).
2530 (fmove-loop-invariants): Likewise.
2531 (ftree-pta): Likewise.
2533 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2535 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2536 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2537 decls to at least the GCC 4.8 used alignments.
2540 * tree-nested.c (convert_nonlocal_reference_stmt,
2541 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2542 of GIMPLE_BIND stmts, adjust associated decls.
2544 2014-01-17 Richard Biener <rguenther@suse.de>
2546 PR tree-optimization/46590
2547 * vec.h (vec<>::bseach): New member function implementing
2548 binary search according to C89 bsearch.
2549 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2550 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2551 bitmap pointer again. Make accesses_in_loop a flat array.
2552 (mem_ref_obstack): New global.
2553 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2554 (mark_ref_stored): Likewise.
2555 (ref_indep_loop_p_2): Likewise.
2556 (set_ref_stored_in_loop): New helper function.
2557 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2558 (memref_free): Adjust.
2559 (record_mem_ref_loc): Simplify.
2560 (gather_mem_refs_stmt): Adjust.
2561 (sort_locs_in_loop_postorder_cmp): New function.
2562 (analyze_memory_references): Sort accesses_in_loop after
2563 loop postorder number.
2564 (find_ref_loc_in_loop_cmp): New function.
2565 (for_all_locs_in_loop): Find relevant cluster of locs in
2566 accesses_in_loop and iterate without recursion.
2567 (execute_sm): Avoid uninit warning.
2568 (struct ref_always_accessed): Simplify.
2569 (ref_always_accessed::operator ()): Likewise.
2570 (ref_always_accessed_p): Likewise.
2571 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2572 loop postorder numbers here.
2573 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2576 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2579 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2580 on decls for which assemble_alias has been called.
2582 2014-01-17 Nick Clifton <nickc@redhat.com>
2584 * config/msp430/msp430.opt: (mcpu): New option.
2585 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2586 (msp430_option_override): Parse target_cpu. If the MCU name
2587 matches a generic string, clear target_mcu.
2588 (msp430_attr): Allow numeric interrupt values up to 63.
2589 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2590 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2592 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2594 * config/msp430/msp430.md (popm): Use %J rather than %I.
2595 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2596 (addhi_cy_i): Use immediate_operand for operand 2.
2597 * doc/invoke.texi: Document -mcpu option.
2599 2014-01-17 Richard Biener <rguenther@suse.de>
2601 PR rtl-optimization/38518
2602 * df.h (df_analyze_loop): Declare.
2603 * df-core.c: Include cfgloop.h.
2604 (df_analyze_1): Split out main part of df_analyze.
2605 (df_analyze): Adjust.
2606 (loop_inverted_post_order_compute): New function.
2607 (loop_post_order_compute): Likewise.
2608 (df_analyze_loop): New function avoiding whole-function
2610 * loop-invariant.c (find_defs): Use df_analyze_loop.
2611 (find_invariants): Adjust.
2612 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2614 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2616 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2617 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2619 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2621 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2622 traversal when removing references.
2624 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2627 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2629 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2632 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2633 pushing a reload for an autoinc when we had previously reloaded an
2634 inner part of the address.
2636 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2638 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2640 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2641 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2642 when not giving up or versioning for alias only because of
2644 (vect_analyze_data_ref_dependences): Set to true.
2645 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2647 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2648 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2652 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2655 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2656 operand 0 predicate for gathers, use a new pseudo as subtarget.
2658 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2661 * lra-constraints.c (process_alt_operands): Add printing debug info.
2662 Check absence of input/output reloads for matched operands too.
2664 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2666 PR rtl-optimization/59835
2667 * ira.c (ira_init_register_move_cost): Increase cost for
2670 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2672 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2674 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2677 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2678 non-register objects. Use gen_(high/low)part more consistently.
2681 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2684 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2685 endian support, remove tests for WORDS_BIG_ENDIAN.
2686 (p8_mfvsrd_3_<mode>): Likewise.
2687 (reload_gpr_from_vsx<mode>): Likewise.
2688 (reload_gpr_from_vsxsf): Likewise.
2689 (p8_mfvsrd_4_disf): Likewise.
2691 2014-01-16 Richard Biener <rguenther@suse.de>
2693 PR rtl-optimization/46590
2694 * lcm.c (compute_antinout_edge): Use postorder iteration.
2695 (compute_laterin): Use inverted postorder iteration.
2697 2014-01-16 Nick Clifton <nickc@redhat.com>
2700 * varasm.c (output_constant): Return the number of bytes actually
2702 (output_constructor_array_range): Update the field size with the
2703 number of bytes emitted by output_constant.
2704 (output_constructor_regular_field): Likewise. Also do not
2705 complain if the total number of bytes emitted is now greater
2706 than the expected fieldpos.
2707 * output.h (output_constant): Update prototype and descriptive comment.
2709 2014-01-16 Marek Polacek <polacek@redhat.com>
2712 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2713 it is error_mark_node.
2715 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2717 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2718 VALID_AVX256_REG_OR_OI_MODE.
2720 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2722 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2723 current procedure should be profiled.
2725 2014-01-15 Andrew Pinski <apinski@cavium.com>
2727 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2728 of moving from/to the STACK_REG register class.
2730 2014-01-15 Richard Henderson <rth@redhat.com>
2733 * reginfo.c (global_regs_decl): Globalize.
2734 * rtl.h (global_regs_decl): Declare.
2735 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2736 reserved via global_regs.
2738 2014-01-15 Teresa Johnson <tejohnson@google.com>
2740 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2742 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2744 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2745 and vmulosh rather than call gen_vec_widen_smult_*.
2746 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2747 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2748 (vec_widen_smult_even_v16qi): Likewise.
2749 (vec_widen_umult_even_v8hi): Likewise.
2750 (vec_widen_smult_even_v8hi): Likewise.
2751 (vec_widen_umult_odd_v16qi): Likewise.
2752 (vec_widen_smult_odd_v16qi): Likewise.
2753 (vec_widen_umult_odd_v8hi): Likewise.
2754 (vec_widen_smult_odd_v8hi): Likewise.
2755 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2756 vmuloub rather than call gen_vec_widen_umult_*.
2757 (vec_widen_umult_lo_v16qi): Likewise.
2758 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2759 vmulosb rather than call gen_vec_widen_smult_*.
2760 (vec_widen_smult_lo_v16qi): Likewise.
2761 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2762 rather than call gen_vec_widen_umult_*.
2763 (vec_widen_umult_lo_v8hi): Likewise.
2764 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2765 rather than call gen_vec_widen_smult_*.
2766 (vec_widen_smult_lo_v8hi): Likewise.
2768 2014-01-15 Jeff Law <law@redhat.com>
2770 PR tree-optimization/59747
2771 * ree.c (find_and_remove_re): Properly handle case where a second
2772 eliminated extension requires widening a copy created for elimination
2773 of a prior extension.
2774 (combine_set_extension): Ensure that the number of hard regs needed
2775 for a destination register does not change when we widen it.
2777 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2779 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2780 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2781 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2782 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2783 (avr-*-rtems*): Likewise.
2784 (bfin*-rtems*): Likewise.
2785 (moxie-*-rtems*): Likewise.
2786 (h8300-*-rtems*): Likewise.
2787 (i[34567]86-*-rtems*): Likewise.
2788 (lm32-*-rtems*): Likewise.
2789 (m32r-*-rtems*): Likewise.
2790 (m68k-*-rtems*): Likewise.
2791 (microblaze*-*-rtems*): Likewise.
2792 (mips*-*-rtems*): Likewise.
2793 (powerpc-*-rtems*): Likewise.
2794 (sh-*-rtems*): Likewise.
2795 (sparc-*-rtems*): Likewise.
2796 (sparc64-*-rtems*): Likewise.
2797 (v850-*-rtems*): Likewise.
2798 (m32c-*-rtems*): Likewise.
2800 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2802 PR rtl-optimization/59511
2803 * ira.c (ira_init_register_move_cost): Use memory costs for some
2804 cases of register move cost calculations.
2805 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2806 instead of BB frequency.
2807 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2808 * lra-assigns.c (find_hard_regno_for): Ditto.
2810 2014-01-15 Richard Biener <rguenther@suse.de>
2812 PR tree-optimization/59822
2813 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2814 (vectorizable_load): Use it to hoist defs of uses of invariant
2815 loads out of the loop.
2817 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2818 Kugan Vivekanandarajah <kuganv@linaro.org>
2821 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2824 2014-01-15 Richard Biener <rguenther@suse.de>
2826 PR rtl-optimization/59802
2827 * lcm.c (compute_available): Use inverted postorder to seed
2828 the initial worklist.
2830 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2833 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2834 ADDR_REGS for invalid symrefs in non-PIC code.
2836 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2839 * builtins.c (determine_block_size): Initialize *probable_max_size
2840 even if len_rtx is CONST_INT.
2842 2014-01-14 Andrew Pinski <apinski@cavium.com>
2844 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2845 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2846 (cortexa53_tunings): Likewise.
2847 (aarch64_sched_issue_rate): New function.
2848 (TARGET_SCHED_ISSUE_RATE): Define.
2850 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2852 * ira-costs.c (find_costs_and_classes): Add missed
2853 ira_init_register_move_cost_if_necessary.
2855 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2858 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2860 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2863 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2864 to indicate if type is used for function return value. Warn ABI
2865 change if the vector mode isn't available for function return value.
2866 (ix86_function_arg_advance): Pass false to type_natural_mode.
2867 (ix86_function_arg): Likewise.
2868 (ix86_gimplify_va_arg): Likewise.
2869 (function_arg_32): Don't warn ABI change.
2870 (ix86_function_value): Pass true to type_natural_mode.
2871 (ix86_return_in_memory): Likewise.
2872 (ix86_struct_value_rtx): Removed.
2873 (TARGET_STRUCT_VALUE_RTX): Likewise.
2875 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2877 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2878 converting a conditional jump into a conditional return.
2880 2014-01-14 Richard Biener <rguenther@suse.de>
2882 PR tree-optimization/58921
2883 PR tree-optimization/59006
2884 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2885 hoisting invariant stmts.
2886 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2887 invariant loads on the preheader edge if possible.
2889 2014-01-14 Joey Ye <joey.ye@arm.com>
2891 * doc/plugin.texi (Building GCC plugins): Update to C++.
2893 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2895 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2896 (_mm_rcp28_round_ss): Ditto.
2897 (_mm_rsqrt28_round_sd): Ditto.
2898 (_mm_rsqrt28_round_ss): Ditto.
2899 (_mm_rcp28_sd): Ditto.
2900 (_mm_rcp28_ss): Ditto.
2901 (_mm_rsqrt28_sd): Ditto.
2902 (_mm_rsqrt28_ss): Ditto.
2903 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2904 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2905 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2906 (IX86_BUILTIN_RCP28SD): Ditto.
2907 (IX86_BUILTIN_RCP28SS): Ditto.
2908 (IX86_BUILTIN_RSQRT28SD): Ditto.
2909 (IX86_BUILTIN_RSQRT28SS): Ditto.
2910 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2911 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2912 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2913 (ix86_expand_special_args_builtin): Expand new FTYPE.
2914 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2915 (srcp14<mode>): Make insn unary.
2916 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2917 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2918 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2919 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2920 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2921 Fix rounding: make it SAE only.
2922 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2924 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2926 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2927 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2928 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2929 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2930 (round_saeonly_mask_scalar_operand4): Ditto.
2931 (round_saeonly_mask_scalar_op3): Ditto.
2932 (round_saeonly_mask_scalar_op4): Ditto.
2934 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2936 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2937 Implement -maltivec=be for vec_insert and vec_extract.
2939 2014-01-10 DJ Delorie <dj@redhat.com>
2941 * config/msp430/msp430.md (call_internal): Don't allow memory
2942 references with SP as the base register.
2943 (call_value_internal): Likewise.
2944 * config/msp430/constraints.md (Yc): New. For memory references
2945 that don't use SP as a base register.
2947 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2948 "an integer without a # prefix"
2949 * config/msp430/msp430.md (epilogue_helper): Use it.
2951 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2954 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2955 AVX512F gather builtins.
2956 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2957 on gather decls with INTEGER_TYPE masktype.
2958 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2959 directly into the builtin rather than hoisting it before loop.
2961 PR tree-optimization/59387
2962 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2963 (scev_const_prop): If folded_casts and type has undefined overflow,
2964 use force_gimple_operand instead of force_gimple_operand_gsi and
2965 for each added stmt if it is assign with
2966 arith_code_with_undefined_signed_overflow, call
2967 rewrite_to_defined_overflow.
2968 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2969 gimple-fold.h instead.
2970 (arith_code_with_undefined_signed_overflow,
2971 rewrite_to_defined_overflow): Moved to ...
2972 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2973 rewrite_to_defined_overflow): ... here. No longer static.
2974 Include gimplify-me.h.
2975 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2976 rewrite_to_defined_overflow): New prototypes.
2978 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2980 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2982 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2984 * builtins.c (get_object_alignment_2): Minor tweak.
2985 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2987 2014-01-13 Christian Bruel <christian.bruel@st.com>
2989 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2990 optimized non constant lengths.
2992 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2995 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2996 load as __atomic_load_N if possible.
2998 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3000 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3002 (rs6000_expand_builtin): Adjust call.
3004 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3007 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3008 * config/rs6000/rs6000.c: Include target-globals.h.
3009 (rs6000_set_current_function): Instead of doing target_reinit
3010 unconditionally, use save_target_globals_default_opts and
3011 restore_target_globals.
3013 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3015 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3016 (rs6000_expand_builtin): Handle mffs and mtfsf.
3017 (rs6000_init_builtins): Define mffs and mtfsf.
3018 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3019 (rs6000_mffs): New pattern.
3020 (rs6000_mtfsf): New pattern.
3022 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3024 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3025 Start narrowing with START. Apply candidate-use pair
3026 and check overall cost in narrowing.
3027 (iv_ca_prune): Pass new argument.
3029 2014-01-10 Jeff Law <law@redhat.com>
3032 * ree.c (combine_reaching_defs): Ensure the defining statement
3033 occurs before the extension when optimizing extensions with
3034 different source and destination hard registers.
3036 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3039 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3040 vtables into the type inheritance graph.
3042 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3044 PR rtl-optimization/59754
3045 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3046 modes in the REGNO != REGNO case.
3048 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3050 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3052 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3054 PR tree-optimization/59745
3055 * tree-predcom.c (tree_predictive_commoning_loop): Call
3056 free_affine_expand_cache if giving up because components is NULL.
3058 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3059 GC in payload of target_globals struct instead of allocating them on
3060 the heap and the larger structs separately using GC.
3061 * target-globals.h (struct target_globals): Make regs, hard_regs,
3062 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3063 of GTY((skip)) and change type to void *.
3064 (reset_target_globals): Cast loads from those fields to corresponding
3067 2014-01-10 Steve Ellcey <sellcey@mips.com>
3070 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3071 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3072 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3074 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3077 * aarch64-modes.def (CC_Zmode): New flags mode.
3078 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3079 represents an equality.
3080 (aarch64_get_condition_code): Handle CC_Zmode.
3081 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3083 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3085 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3086 extraction in good case.
3088 2014-01-10 Richard Biener <rguenther@suse.de>
3090 PR tree-optimization/59374
3091 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3092 checking after SLP discovery. Mark stmts not participating
3093 in any SLP instance properly.
3095 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3097 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3098 when handling a SET rtx.
3100 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3102 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3103 (cortex-a57): Likewise.
3104 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3106 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3108 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3109 non-iwmmxt builtins.
3111 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3115 * ipa-devirt.c record_target_from_binfo): Take as argument
3116 stack of binfos and lookup matching one for virtual inheritance.
3117 (possible_polymorphic_call_targets_1): Update.
3119 2014-01-10 Huacai Chen <chenhc@lemote.com>
3121 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3122 kernel strings for Loongson-2E/2F/3A.
3124 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3127 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3128 is_gimple_call before calling gimple_call_internal_p.
3130 2014-01-09 Steve Ellcey <sellcey@mips.com>
3132 * Makefile.in (TREE_FLOW_H): Remove.
3133 (TREE_SSA_H): Add file names from tree-flow.h.
3134 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3135 * tree.h: Remove tree-flow.h reference.
3136 * hash-table.h: Remove tree-flow.h reference.
3137 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3138 reference with tree-ssa-loop.h.
3140 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3142 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3143 default element-order behavior for -maltivec.
3144 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3145 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3146 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3147 when targeting big endian, at least for now.
3148 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3150 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3153 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3154 var satisfies use_register_for_decl, just take into account type
3155 alignment, rather than decl alignment.
3157 PR tree-optimization/59622
3158 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3159 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3160 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3161 Don't devirtualize for inplace at all. For targets.length () == 1,
3162 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3164 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3166 * config/i386/i386.md (cpu): Remove the unused btver1.
3168 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3170 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3172 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3175 * tree-core.h (struct target_globals): New forward declaration.
3176 (struct tree_target_option): Add globals field.
3177 * tree.h (TREE_TARGET_GLOBALS): Define.
3178 (prepare_target_option_nodes_for_pch): New prototype.
3179 * target-globals.h (struct target_globals): Define even if
3181 * tree.c (prepare_target_option_node_for_pch,
3182 prepare_target_option_nodes_for_pch): New functions.
3183 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3184 * config/i386/i386.c: Include target-globals.h.
3185 (ix86_set_current_function): Instead of doing target_reinit
3186 unconditionally, use save_target_globals_default_opts and
3187 restore_target_globals.
3189 2014-01-09 Richard Biener <rguenther@suse.de>
3191 PR tree-optimization/59715
3192 * tree-cfg.h (split_critical_edges): Declare.
3193 * tree-cfg.c (split_critical_edges): Export.
3194 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3196 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3198 * cfgexpand.c (expand_stack_vars): Optionally disable
3199 asan stack protection.
3200 (expand_used_vars): Likewise.
3201 (partition_stack_vars): Likewise.
3202 * asan.c (asan_emit_stack_protection): Optionally disable
3203 after return stack usage.
3204 (instrument_derefs): Optionally disable memory access instrumentation.
3205 (instrument_builtin_call): Likewise.
3206 (instrument_strlen_call): Likewise.
3207 (asan_protect_global): Optionally disable global variables protection.
3208 * doc/invoke.texi: Added doc for new options.
3209 * params.def: Added new options.
3210 * params.h: Likewise.
3212 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3214 PR rtl-optimization/59724
3215 * ifcvt.c (cond_exec_process_if_block): Don't call
3216 flow_find_head_matching_sequence with 0 longest_match.
3217 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3218 non-active insns if !stop_after.
3219 (try_head_merge_bb): Revert 2014-01-07 changes.
3221 2014-01-08 Jeff Law <law@redhat.com>
3223 * ree.c (get_sub_rtx): New function, extracted from...
3224 (merge_def_and_ext): Here.
3225 (combine_reaching_defs): Use get_sub_rtx.
3227 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3229 * cgraph.h (varpool_variable_node): Do not choke on null node.
3231 2014-01-08 Catherine Moore <clm@codesourcery.com>
3233 * config/mips/mips.md (simple_return): Attempt to use JRC
3235 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3237 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3239 PR rtl-optimization/59137
3240 * reorg.c (steal_delay_list_from_target): Call update_block for
3242 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3244 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3246 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3247 two duplicate entries.
3249 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3252 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3254 * config/mips/mips.c (mips_truncated_op_cost): New function.
3255 (mips_rtx_costs): Adjust test for BADDU.
3256 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3258 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3260 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3261 (*baddu_si): ...this new pattern.
3263 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3266 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3268 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3271 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3273 (expand_expr, expand_normal): Adjust.
3274 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3275 inner_reference_p. Use inner_reference_p to expand inner references.
3276 (store_expr): Adjust.
3277 * cfgexpand.c (expand_call_stmt): Adjust.
3279 2014-01-08 Rong Xu <xur@google.com>
3281 * gcov-io.c (gcov_var): Move from gcov-io.h.
3282 (gcov_position): Ditto.
3283 (gcov_is_error): Ditto.
3284 (gcov_rewrite): Ditto.
3285 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3286 only part to libgcc/libgcov.h.
3288 2014-01-08 Marek Polacek <polacek@redhat.com>
3291 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3293 2014-01-08 Marek Polacek <polacek@redhat.com>
3296 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3298 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3300 PR rtl-optimization/59649
3301 * stor-layout.c (get_mode_bounds): For BImode return
3302 0 and STORE_FLAG_VALUE.
3304 2014-01-08 Richard Biener <rguenther@suse.de>
3307 * gimple.h (is_gimple_builtin_call): Remove.
3308 (gimple_builtin_call_types_compatible_p): New.
3309 (gimple_call_builtin_p): New overload.
3310 * gimple.c (is_gimple_builtin_call): Remove.
3311 (validate_call): Rename to ...
3312 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3314 (validate_type): New static function.
3315 (gimple_call_builtin_p): New overload and adjust.
3316 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3317 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3318 (gimple_fold_stmt_to_constant_1): Likewise.
3319 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3321 2014-01-08 Richard Biener <rguenther@suse.de>
3324 * gimplify.c (gimplify_expr): Gimplify register-register type
3325 VIEW_CONVERT_EXPRs to separate stmts.
3327 2014-01-07 Jeff Law <law@redhat.com>
3330 * ree.c (combine_set_extension): Handle case where source
3331 and destination registers in an extension insn are different.
3332 (combine_reaching_defs): Allow source and destination registers
3333 in extension to be different under limited circumstances.
3334 (add_removable_extension): Remove restriction that the
3335 source and destination registers in the extension are the same.
3336 (find_and_remove_re): Emit a copy from the extension's
3337 destination to its source after the defining insn if
3338 the source and destination registers are different.
3341 * ifcvt.c (merge_if_block): If we are merging a block with more than
3342 one successor with a block with no successors, remove any BARRIER
3343 after the second block.
3345 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3347 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3349 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3352 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3353 for 14-bit register offsets when INT14_OK_STRICT is false.
3355 2014-01-07 Roland Stigge <stigge@antcom.de>
3356 Michael Meissner <meissner@linux.vnet.ibm.com>
3359 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3360 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3362 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3364 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3367 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3369 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3370 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3371 rtx is const0_rtx or not.
3373 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3376 * target-globals.c (save_target_globals): Remove this_fn_optab
3378 * toplev.c: Include optabs.h.
3379 (target_reinit): Temporarily restore the global options if another
3380 set of options are in force.
3382 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3384 PR rtl-optimization/58668
3385 * cfgcleanup.c (flow_find_cross_jump): Don't count
3386 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3387 to determine what is counted.
3388 (flow_find_head_matching_sequence): Use active_insn_p to determine
3390 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3392 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3393 determine what is counted.
3395 PR tree-optimization/59643
3396 * tree-predcom.c (split_data_refs_to_components): If one dr is
3397 read and one write, determine_offset fails and the write isn't
3398 in the bad component, just put the read into the bad component.
3400 2014-01-07 Mike Stump <mikestump@comcast.net>
3401 Jakub Jelinek <jakub@redhat.com>
3404 * tree-core.h (struct tree_optimization_option): Change optabs
3405 type from unsigned char * to void *.
3406 * optabs.c (init_tree_optimization_optabs): Adjust
3407 TREE_OPTIMIZATION_OPTABS initialization.
3409 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3412 * config/i386/i386.h (struct machine_function): Add
3413 no_drap_save_restore field.
3414 * config/i386/i386.c (ix86_save_reg): Use
3415 !cfun->machine->no_drap_save_restore instead of
3416 crtl->stack_realign_needed.
3417 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3418 this function clears frame_pointer_needed. Set
3419 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3420 and DRAP reg is needed.
3422 2014-01-06 Marek Polacek <polacek@redhat.com>
3425 * doc/implement-c.texi: Mention that other integer types are
3426 permitted as bit-field types in strictly conforming mode.
3428 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3430 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3433 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3435 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3437 2014-01-06 Martin Jambor <mjambor@suse.cz>
3440 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3442 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3444 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3448 * var-tracking.c (add_stores): Preserve the value of the source even if
3449 we don't record the store.
3451 2014-01-06 Terry Guo <terry.guo@arm.com>
3453 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3455 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3458 * config/darwin.c (darwin_function_section): Adjust return values to
3459 correspond to optimisation changes made in r206070.
3461 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3463 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3464 from prefetch_block tune setting.
3465 (nocona_cost): Correct size of prefetch block to 64.
3467 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3469 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3470 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3471 used to save the static chain register in the computation of the offset
3472 from which the FP registers need to be restored.
3474 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3476 PR tree-optimization/59519
3477 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3478 ICE if get_current_def (current_new_name) is already non-NULL, as long
3479 as it is a phi result of some other phi in *new_exit_bb that has
3482 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3483 or vmovdqu* for misaligned_operand.
3484 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3485 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3486 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3487 aligned_mem for AVX512F masked aligned load and store builtins and for
3490 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3492 PR tree-optimization/59651
3493 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3494 Address range for negative step should be added by TYPE_SIZE_UNIT.
3496 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3498 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3499 overlapping registers also for registers other than the stack pointer.
3501 2014-01-03 Marek Polacek <polacek@redhat.com>
3504 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3507 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3510 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3513 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3514 (push splitter): Use <P:MODE_SIZE> instead of
3515 GET_MODE_SIZE (<P:MODE>mode).
3516 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3517 (mov -1, reg peephole2): Likewise.
3518 * config/i386/sse.md (*mov<mode>_internal,
3519 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3520 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3521 *<code><mode>3, *andnot<mode>3<mask_name>,
3522 <mask_codefor><code><mode>3<mask_name>): Likewise.
3523 * config/i386/subst.md (mask_mode512bit_condition,
3524 sd_mask_mode512bit_condition): Likewise.
3526 2014-01-02 Xinliang David Li <davidxl@google.com>
3528 PR tree-optimization/59303
3529 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3530 (dump_predicates): Better output format.
3531 (pred_equal_p): New function.
3532 (is_neq_relop_p): Ditto.
3533 (is_neq_zero_form_p): Ditto.
3534 (pred_expr_equal_p): Ditto.
3535 (pred_neg_p): Ditto.
3536 (simplify_pred): Ditto.
3537 (simplify_preds_2): Ditto.
3538 (simplify_preds_3): Ditto.
3539 (simplify_preds_4): Ditto.
3540 (simplify_preds): Ditto.
3542 (push_to_worklist): Ditto.
3543 (get_pred_info_from_cmp): Ditto.
3544 (is_degenerated_phi): Ditto.
3545 (normalize_one_pred_1): Ditto.
3546 (normalize_one_pred): Ditto.
3547 (normalize_one_pred_chain): Ditto.
3548 (normalize_preds): Ditto.
3549 (normalize_cond_1): Remove function.
3550 (normalize_cond): Ditto.
3551 (is_gcond_subset_of): Ditto.
3552 (is_subset_of_any): Ditto.
3553 (is_or_set_subset_of): Ditto.
3554 (is_and_set_subset_of): Ditto.
3555 (is_norm_cond_subset_of): Ditto.
3556 (pred_chain_length_cmp): Ditto.
3557 (convert_control_dep_chain_into_preds): Type change.
3558 (find_predicates): Ditto.
3559 (find_def_preds): Ditto.
3560 (destroy_predicates_vecs): Ditto.
3561 (find_matching_predicates_in_rest_chains): Ditto.
3562 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3563 (is_pred_expr_subset): Ditto.
3564 (is_pred_chain_subset_of): Ditto.
3565 (is_included_in): Ditto.
3566 (is_superset_of): Ditto.
3568 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3570 Update copyright years.
3572 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3574 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3575 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3576 config/arc/arc.md, config/arc/arc.opt,
3577 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3578 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3579 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3580 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3581 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3582 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3583 vtable-verify.c, vtable-verify.h: Use the standard form for the
3586 2014-01-02 Tobias Burnus <burnus@net-b.de>
3588 * gcc.c (process_command): Update copyright notice dates.
3589 * gcov-dump.c: Ditto.
3591 * doc/cpp.texi: Bump @copying's copyright year.
3592 * doc/cppinternals.texi: Ditto.
3593 * doc/gcc.texi: Ditto.
3594 * doc/gccint.texi: Ditto.
3595 * doc/gcov.texi: Ditto.
3596 * doc/install.texi: Ditto.
3597 * doc/invoke.texi: Ditto.
3599 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3601 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3603 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3605 * config/i386/sse.md (*mov<mode>_internal): Guard
3606 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3608 PR rtl-optimization/59647
3609 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3610 new_rtx into UNSIGNED_FLOAT rtxes.
3612 Copyright (C) 2014 Free Software Foundation, Inc.
3614 Copying and distribution of this file, with or without modification,
3615 are permitted in any medium without royalty provided the copyright
3616 notice and this notice are preserved.