1 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
3 PR rtl-optimization/58960
4 * haifa-sched.c (alloc_global_sched_pressure_data): New,
6 (sched_init): ... here.
7 (free_global_sched_pressure_data): New, factored out from ...
8 (sched_finish): ... here.
9 * sched-int.h (free_global_sched_pressure_data): Declare.
10 * sched-rgn.c (nr_regions_initial): New static global.
11 (haifa_find_rgns): Initialize it.
12 (schedule_region): Disable sched-pressure for the newly
15 2014-02-17 Richard Biener <rguenther@suse.de>
17 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
18 release SSA defs of pattern stmts.
20 2014-02-17 Richard Biener <rguenther@suse.de>
22 * tree-inline.c (expand_call_inline): Release the virtual
23 operand defined by the call we are about to inline.
25 2014-02-17 Richard Biener <rguenther@suse.de>
27 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
29 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
30 Ilya Tocar <ilya.tocar@intel.com>
32 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
33 arguments order in builtin.
34 (_mm512_permutexvar_epi64): Ditto.
35 (_mm512_mask_permutexvar_epi64): Ditto
36 (_mm512_maskz_permutexvar_epi32): Ditto
37 (_mm512_permutexvar_epi32): Ditto
38 (_mm512_mask_permutexvar_epi32): Ditto
40 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
42 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian
44 (p8_vmrgow): Likewise.
46 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
48 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
51 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
54 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
55 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
56 into 64-bit and 32-bit moves. On 64-bit moves, add support for
57 using direct move instructions on ISA 2.07. Also adjust
58 instruction length for 64-bit.
59 (mov<mode>_64bit, TFmode/TDmode): Likewise.
60 (mov<mode>_32bit, TFmode/TDmode): Likewise.
62 2014-02-15 Alan Modra <amodra@gmail.com>
66 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
67 find_replacement on parts of insn rtl that might be reloaded.
69 2014-02-15 Richard Biener <rguenther@suse.de>
71 PR tree-optimization/60183
72 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating
74 (tree_ssa_phiprop): Calculate and free post-dominators.
76 2014-02-14 Jeff Law <law@redhat.com>
78 PR rtl-optimization/60131
79 * ree.c (get_extended_src_reg): New function.
80 (combine_reaching_defs): Use it rather than assuming location
82 (find_and_remove_re): Verify first operand of extension is
83 a REG before adding the insns to the copy list.
85 2014-02-14 Roland McGrath <mcgrathr@google.com>
87 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
88 * configure: Regenerated.
89 * config.in: Regenerated.
90 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
93 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
94 Richard Earnshaw <rearnsha@arm.com>
96 PR rtl-optimization/59535
97 * lra-constraints.c (process_alt_operands): Encourage alternative
98 when unassigned pseudo class is superset of the alternative class.
99 (inherit_reload_reg): Don't inherit when optimizing for code size.
100 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
101 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
102 modes not less than 4 for Thumb1.
104 2014-02-14 Kyle McMartin <kyle@redhat.com>
107 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
109 2014-02-14 Richard Biener <rguenther@suse.de>
111 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
112 (get_frame_arg): Drop the assert with langhook types_compatible_p.
113 Do not strip INDIRECT_REFs.
115 2014-02-14 Richard Biener <rguenther@suse.de>
118 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
119 DECL_FUNCTION_SPECIFIC_TARGET.
120 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
121 * tree-streamer-out.c (pack_ts_target_option): Remove.
122 (streamer_pack_tree_bitfields): Do not stream
124 (write_ts_function_decl_tree_pointers): Do not stream
125 DECL_FUNCTION_SPECIFIC_TARGET.
126 * tree-streamer-in.c (unpack_ts_target_option): Remove.
127 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
128 (lto_input_ts_function_decl_tree_pointers): Do not stream
129 DECL_FUNCTION_SPECIFIC_TARGET.
131 2014-02-14 Jakub Jelinek <jakub@redhat.com>
133 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
135 (get_initial_def_for_induction, vectorizable_induction): Ignore
136 debug stmts when looking for exit_phi.
137 (vectorizable_live_operation): Fix up condition.
139 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
141 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
142 nreverse() because it changes the content of original tree list.
144 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
146 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
147 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
149 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
151 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
152 GNU coding standards.
154 2014-02-13 Jakub Jelinek <jakub@redhat.com>
157 * dwarf2out.c (gen_subprogram_die): Don't call
158 add_calling_convention_attribute if subr_die is old_die.
160 2014-02-13 Sharad Singhai <singhai@google.com>
162 * doc/optinfo.texi: Fix order of nodes.
164 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
166 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
167 operands[2], not operands[3].
169 2014-02-13 Richard Biener <rguenther@suse.de>
172 * doc/install.texi (ISL): Update recommended version to 0.12.2,
173 mention the possibility of an in-tree build.
174 (CLooG): Update recommended version to 0.18.1, mention the
175 possibility of an in-tree build and clarify that the ISL
176 bundled with CLooG does not work.
178 2014-02-13 Jakub Jelinek <jakub@redhat.com>
181 * expr.c (compress_float_constant): If x is a hard register,
182 extend into a pseudo and then move to x.
184 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
186 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
187 caused by bad second argument to warning_at() with -mhotpatch and
188 nested functions (e.g. with gfortran).
190 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
192 * opts.c (option_name): Remove "enabled by default" rider.
194 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
196 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
198 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
199 Uros Bizjak <ubizjak@gmail.com>
202 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
203 * configure: Regenerated.
205 2014-02-12 Richard Biener <rguenther@suse.de>
207 * vec.c (vec_prefix::calculate_allocation): Move as
208 inline variant to vec.h.
209 (vec_prefix::calculate_allocation_1): New out-of-line version.
210 * vec.h (vec_prefix::calculate_allocation_1): Declare.
211 (vec_prefix::m_has_auto_buf): Rename to ...
212 (vec_prefix::m_using_auto_storage): ... this.
213 (vec_prefix::calculate_allocation): Inline the easy cases
214 and dispatch to calculate_allocation_1 which doesn't need the
216 (va_heap::reserve): Use gcc_checking_assert.
217 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
218 m_using_auto_storage.
219 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
221 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
222 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
223 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
225 2014-02-12 Richard Biener <rguenther@suse.de>
227 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
228 when we found a dependence.
230 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
232 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
234 (maybe_fold_stmt): ... into this new function.
235 * omp-low.c (lower_omp): Update comment.
237 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
240 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
243 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
245 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
246 identifiers in comments.
247 (cortexa53_extra_costs): Likewise.
248 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
249 (cortexa7_extra_costs): Likewise.
250 (cortexa12_extra_costs): Likewise.
251 (cortexa15_extra_costs): Likewise.
252 (v7m_extra_costs): Likewise.
254 2014-02-12 Richard Biener <rguenther@suse.de>
257 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
258 of posix_memalign being successful.
259 (lower_stmt): Restrict lowering of posix_memalign to when
260 -ftree-bit-ccp is enabled.
262 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
264 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
266 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
268 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
270 PR rtl-optimization/60116
271 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
272 other_insn once the combination has been validated.
274 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
277 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
279 * ipa-devirt.c: Include demangle.h
280 (odr_violation_reported): New static variable.
281 (add_type_duplicate): Update odr_violations.
282 (maybe_record_node): Add completep parameter; update it.
283 (record_target_from_binfo): Add COMPLETEP parameter;
285 (possible_polymorphic_call_targets_1): Likewise.
286 (struct polymorphic_call_target_d): Add nonconstruction_targets;
287 rename FINAL to COMPLETE.
288 (record_targets_from_bases): Sanity check we found the binfo;
289 fix COMPLETEP updating.
290 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
291 parameter, fix computing of COMPLETEP.
292 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
293 at LTO time do demangling.
294 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
295 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
297 (gimple_get_virt_method_for_binfo): Likewise.
298 * gimple-fold.h (gimple_get_virt_method_for_binfo,
299 gimple_get_virt_method_for_vtable): Update prototypes.
301 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
304 * genautomata.c (add_presence_absence): Fix typo with
305 {final_}presence_list.
307 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
310 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
311 for VSX/Altivec vectors that land in GPR registers.
313 2014-02-11 Richard Henderson <rth@redhat.com>
314 Jakub Jelinek <jakub@redhat.com>
317 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
318 around drhs if type conversion to lacc->type is not useless.
320 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
322 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
324 (cortex-a57.cortex-a53): Likewise.
325 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
327 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
329 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
332 2014-02-11 Renlin Li <Renlin.Li@arm.com>
334 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
335 add_options_for_arm_vfp3.
337 2014-02-11 Jeff Law <law@redhat.com>
340 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
341 object with an undesirable mode.
343 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
346 * config/i386/sol2-9.h: New file.
347 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
348 *-*-solaris2.9*): Use it.
350 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
352 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
353 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
355 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
357 * config/microblaze/microblaze.c: Extend mcpu version format
359 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
361 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
363 2014-02-10 Richard Henderson <rth@redhat.com>
366 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
367 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
368 ms-abi vs -mno-accumulate-outgoing-args.
369 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
370 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
373 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
376 * cfgexpand.c (expand_asm_operands): Attach source location to
377 ASM_INPUT rtx objects.
378 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
380 2014-02-10 Nick Clifton <nickc@redhat.com>
382 * config/mn10300/mn10300.c (popcount): New function.
383 (mn10300_expand_prologue): Include saved registers in stack usage
386 2014-02-10 Jeff Law <law@redhat.com>
389 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
390 when changing the SET_DEST of a prior insn to avoid an input reload.
392 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
394 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
395 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
396 -mcall-openbsd, or -mcall-linux.
397 (CC1_ENDIAN_BIG_SPEC): Remove.
398 (CC1_ENDIAN_LITTLE_SPEC): Remove.
399 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
400 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
401 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
402 and %cc1_endian_default.
403 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
405 2014-02-10 Richard Biener <rguenther@suse.de>
407 PR tree-optimization/60115
408 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
409 MEM_REF handling. Properly verify that the accesses are not
410 out of the objects bound.
412 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
414 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
417 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
419 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
420 proper constants and fix formatting.
421 (possible_polymorphic_call_targets): Fix formatting.
423 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
424 Ilya Tocar <ilya.tocar@intel.com>
426 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
427 (_mm512_loadu_epi32): Renamed into...
428 (_mm512_loadu_si512): This.
429 (_mm512_storeu_epi32): Renamed into...
430 (_mm512_storeu_si512): This.
431 (_mm512_maskz_ceil_ps): Removed.
432 (_mm512_maskz_ceil_pd): Ditto.
433 (_mm512_maskz_floor_ps): Ditto.
434 (_mm512_maskz_floor_pd): Ditto.
435 (_mm512_floor_round_ps): Ditto.
436 (_mm512_floor_round_pd): Ditto.
437 (_mm512_ceil_round_ps): Ditto.
438 (_mm512_ceil_round_pd): Ditto.
439 (_mm512_mask_floor_round_ps): Ditto.
440 (_mm512_mask_floor_round_pd): Ditto.
441 (_mm512_mask_ceil_round_ps): Ditto.
442 (_mm512_mask_ceil_round_pd): Ditto.
443 (_mm512_maskz_floor_round_ps): Ditto.
444 (_mm512_maskz_floor_round_pd): Ditto.
445 (_mm512_maskz_ceil_round_ps): Ditto.
446 (_mm512_maskz_ceil_round_pd): Ditto.
447 (_mm512_expand_pd): Ditto.
448 (_mm512_expand_ps): Ditto.
449 * config/i386/i386.c (ix86_builtins): Remove
450 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
452 * config/i386/predicates.md (const1256_operand): New.
453 (const_1_to_2_operand): Ditto.
454 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
455 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
456 (*avx512pf_gatherpf<mode>sf): Ditto.
457 (avx512pf_gatherpf<mode>df): Ditto.
458 (*avx512pf_gatherpf<mode>df_mask): Ditto.
459 (*avx512pf_gatherpf<mode>df): Ditto.
460 (avx512pf_scatterpf<mode>sf): Ditto.
461 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
462 (*avx512pf_scatterpf<mode>sf): Ditto.
463 (avx512pf_scatterpf<mode>df): Ditto.
464 (*avx512pf_scatterpf<mode>df_mask): Ditto.
465 (*avx512pf_scatterpf<mode>df): Ditto.
466 (avx512f_expand<mode>): Removed.
467 (<shift_insn><mode>3<mask_name>): Change predicate type.
469 2014-02-08 Jakub Jelinek <jakub@redhat.com>
471 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
472 not at the end of datarefs vector use ordered_remove to avoid
473 reordering datarefs vector.
476 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
477 mark local addressable non-static vars as GOVD_PRIVATE
478 instead of GOVD_LOCAL.
479 * omp-low.c (lower_omp_for): Move gimple_bind_vars
480 and BLOCK_VARS of gimple_bind_block to new_stmt rather
484 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
485 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
486 assume_aligned or alloc_align attributes.
487 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
488 arguments. Handle also assume_aligned and alloc_align attributes.
489 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
490 calls to functions with assume_aligned or alloc_align attributes.
491 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
493 2014-02-08 Terry Guo <terry.guo@arm.com>
495 * doc/invoke.texi: Document ARM -march=armv7e-m.
497 2014-02-08 Jakub Jelinek <jakub@redhat.com>
499 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
500 flag on __cilkrts_rethrow builtin.
503 * ipa-cp.c (determine_versionability): Fail at -O0
504 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
505 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
508 2014-02-04 Jakub Jelinek <jakub@redhat.com>
511 * tree-inline.c (copy_forbidden): Fail for
512 __attribute__((optimize (0))) functions.
514 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
516 * varpool.c: Include pointer-set.h.
517 (varpool_remove_unreferenced_decls): Variables in other partitions
518 will not be output; be however careful to not lose information
521 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
523 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
524 lookup in the vtable constructor.
526 2014-02-07 Jeff Law <law@redhat.com>
529 * config/m68k/m68k.md (ashldi_extsi): Turn into a
530 define_insn_and_split.
532 * ipa-inline.c (inline_small_functions): Fix typos.
534 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
536 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
537 (s390_can_use_return_insn): Declare.
538 * config/s390/s390.h (EPILOGUE_USES): Define.
539 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
541 (s390_chunkify_start): Handle return JUMP_LABELs.
542 (s390_early_mach): Emit a main_pool instruction on the entry edge.
543 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
544 (s390_can_use_return_insn): New functions.
545 (s390_fix_long_loop_prediction): Handle conditional returns.
546 (TARGET_SET_UP_BY_PROLOGUE): Define.
547 * config/s390/s390.md (ANY_RETURN): New code iterator.
548 (*creturn, *csimple_return, return, simple_return): New patterns.
550 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
552 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
553 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
554 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
555 REG_CFA_RESTORE list when deciding not to restore a register.
557 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
559 * config/s390/s390.c: Include tree-pass.h and context.h.
560 (s390_early_mach): New function, split out from...
561 (s390_emit_prologue): ...here.
562 (pass_data_s390_early_mach): New pass structure.
563 (pass_s390_early_mach): New class.
564 (s390_option_override): Create and register early_mach pass.
567 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
569 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
570 to match for the exit block.
572 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
574 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
575 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
576 Reject misaligned operands.
578 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
580 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
582 2014-02-07 Richard Biener <rguenther@suse.de>
585 * gimple-low.c (lower_builtin_posix_memalign): New function.
586 (lower_stmt): Call it to lower posix_memalign in a way
587 to make alignment info accessible.
589 2014-02-07 Jakub Jelinek <jakub@redhat.com>
592 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
593 __builtin_setjmp_receiver.
595 2014-02-07 Richard Biener <rguenther@suse.de>
598 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
599 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
600 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
601 Handle BUILT_IN_POSIX_MEMALIGN.
602 (find_func_clobbers): Likewise.
603 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
604 (call_may_clobber_ref_p_1): Likewise.
606 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
609 * ipa-devirt.c (record_target_from_binfo): Remove overactive
612 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
615 * lto-cgraph.c (lto_output_node): Use
616 symtab_get_symbol_partitioning_class.
617 (lto_output_varpool_node): likewise.
618 (symtab_get_symbol_partitioning_class): Move here from
620 * cgraph.h (symbol_partitioning_class): Likewise.
621 (symtab_get_symbol_partitioning_class): Declare.
623 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
625 * ggc.h (ggc_internal_cleared_alloc): New macro.
626 * vec.h (vec_safe_copy): Handle memory stats.
627 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
628 * target-globals.c (save_target_globals): Likewise.
630 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
633 * expr.c (emit_move_resolve_push): Export; be bit more selective
634 on when to clear alias set.
635 * expr.h (emit_move_resolve_push): Declare.
636 * function.h (struct function): Add tail_call_marked.
637 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
638 * config/i386/i386-protos.h (ix86_expand_push): Remove.
639 * config/i386/i386.md (TImode move expander): De not call
641 (FP push expanders): Preserve memory attributes.
642 * config/i386/sse.md (push<mode>1): Remove.
643 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
644 (ix86_expand_push): Remove.
645 * config/i386/mmx.md (push<mode>1): Remove.
647 2014-02-06 Jakub Jelinek <jakub@redhat.com>
649 PR rtl-optimization/60030
650 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
651 lopart with paradoxical subreg before shifting it up by hprec.
653 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
655 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
656 Remove extra newline at end of file.
657 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
658 (arm_issue_rate): Handle cortexa57.
659 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
660 (cortex-a57.cortex-a53): Likewise.
662 2014-02-06 Jakub Jelinek <jakub@redhat.com>
665 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
666 don't record in REG_FRAME_RELATED_EXPR registers not set in that
668 (arm_expand_prologue): Adjust all callers.
669 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
670 info, registers also at the lowest numbered registers side. Use
671 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
675 * var-tracking.c (adjust_mems): Before adding a SET to
676 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
678 2014-02-06 Alan Modra <amodra@gmail.com>
681 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
682 change SDmode to DDmode when lra_in_progress.
684 2014-02-06 Jakub Jelinek <jakub@redhat.com>
687 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
688 free_data_ref on the dr first, and before goto again also set dr
689 to the next dr. For simd_lane_access, free old datarefs[i] before
690 overwriting it. For get_vectype_for_scalar_type failure, don't
691 free_data_ref if simd_lane_access.
693 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
696 * tree.h (opts_for_fn): New inline function.
697 (opt_for_fn): Define.
698 * config/i386/i386.c (ix86_function_regparm): Use
699 opt_for_fn (decl, optimize) instead of optimize.
701 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
703 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
704 for SYMBOL_REF in large memory model.
706 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
708 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
710 (cortex-a57): Likewise.
711 (cortex-a57.cortex-a53): Likewise.
713 2014-02-06 Yury Gribov <y.gribov@samsung.com>
714 Kugan Vivekanandarajah <kuganv@linaro.org>
716 * config/arm/arm.c (arm_vector_alignment_reachable): Check
718 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
720 2014-02-06 Richard Biener <rguenther@suse.de>
722 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
723 set_loop_copy and initialize_original_copy_tables.
725 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
727 * config/aarch64/aarch64-simd.md
728 (aarch64_ashr_simddi): Change QI to SI.
730 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
731 Jakub Jelinek <jakub@redhat.com>
734 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
737 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
739 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
740 CODE_FOR_altivec_vpku[hw]um to
741 CODE_FOR_altivec_vpku[hw]um_direct.
742 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
743 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
744 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
745 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
747 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
749 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
750 generation for -maltivec=be.
751 (altivec_vsumsws): Simplify redundant test.
753 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
755 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
756 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
757 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
758 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
760 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
762 (altivec_vpks<VI_char>ss): Likewise.
763 (altivec_vpks<VI_char>us): Likewise.
764 (altivec_vpku<VI_char>us): Likewise.
765 (altivec_vpku<VI_char>um): Likewise.
766 (altivec_vpku<VI_char>um_direct): New (copy of
767 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
769 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
770 target is little endian and -maltivec=be is not specified.
771 (*altivec_vupkhs<VU_char>_direct): New (copy of
772 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
773 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
774 target is little endian and -maltivec=be is not specified.
775 (*altivec_vupkls<VU_char>_direct): New (copy of
776 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
777 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
778 little endian and -maltivec=be is not specified.
779 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
780 little endian and -maltivec=be is not specified.
782 2014-02-05 Richard Henderson <rth@redhat.com>
785 * combine-stack-adj.c: Revert r206943.
786 * sched-int.h (struct deps_desc): Add last_args_size.
787 * sched-deps.c (init_deps): Initialize it.
788 (sched_analyze_insn): Add OUTPUT dependencies between insns that
789 contain REG_ARGS_SIZE notes.
791 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
793 * lto-cgraph.c (asm_nodes_output): Make global.
794 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
795 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
796 (driver_handle_option): Handle OPT_fwpa.
798 2014-02-05 Jakub Jelinek <jakub@redhat.com>
801 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
802 a comment typo and formatting issue. If odr_hash hasn't been
803 created, return vNULL and set *completep to false.
806 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
807 bb with no successors.
809 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
812 * doc/invoke.texi (-march): Clarify documentation for ARM.
816 2014-02-05 Richard Biener <rguenther@suse.de>
818 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
819 when not vectorizing because of too many alias checks.
820 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
821 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
823 2014-02-05 Nick Clifton <nickc@redhat.com>
825 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
826 accept extended registers in any mode when compiling for the MN10300.
828 2014-02-05 Yury Gribov <y.gribov@samsung.com>
830 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
831 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
832 sanitization attributes.
833 (can_inline_edge_p): Likewise.
834 (sanitize_attrs_match_for_inline_p): New function.
836 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
838 * ipa-prop.c (detect_type_change): Shor circuit testing of
839 type changes on THIS pointer.
841 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
844 * config/pa/pa.c (legitimize_tls_address): Return original address
845 if not passed a SYMBOL_REF rtx.
846 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
848 (pa_emit_move_sequence): Simplify TLS source operands.
849 (pa_legitimate_constant_p): Reject all TLS constants.
850 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
851 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
853 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
855 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
856 groups when we know they are controlled by LTO.
857 * varasm.c (default_binds_local_p_1): If object is in other partition,
858 it will be resolved locally.
860 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
862 * config/host-linux.c (linux_gt_pch_use_address): Don't
863 use SSIZE_MAX because it is not always defined.
865 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
868 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
869 threshold for pseudo splitting.
870 (update_ebb_live_info): Process call argument hard registers and
871 hard registers from insn definition too.
872 (max_small_class_regs_num): New constant.
873 (inherit_in_ebb): Update live hard regs through EBBs. Update
874 reloads_num only for small register classes. Don't split for
877 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
880 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
883 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
885 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
888 2014-02-04 Marek Polacek <polacek@redhat.com>
890 * gdbinit.in (pel): Define.
892 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
894 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
897 2014-02-04 Richard Biener <rguenther@suse.de>
900 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
901 in function context local.
902 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
903 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
904 similar to LTO_imported_decl_ref.
906 2014-02-04 Jakub Jelinek <jakub@redhat.com>
908 PR tree-optimization/60002
909 * cgraphclones.c (build_function_decl_skip_args): Clear
912 PR tree-optimization/60023
913 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
914 false to gsi_replace.
915 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
916 has been in some EH region and vec_stmt could throw, add
917 vec_stmt into the same EH region.
918 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
919 has no lhs, ignore it.
920 * internal-fn.c (expand_MASK_LOAD): Likewise.
923 * tree-inline.c (copy_forbidden): Fail for
924 __attribute__((optimize (0))) functions.
927 * omp-low.c (simd_clone_struct_copy): If from->inbranch
928 is set, copy one less argument.
929 (expand_simd_clones): Don't subtract clone_info->inbranch
930 from simd_clone_struct_alloc argument.
932 PR rtl-optimization/57915
933 * recog.c (simplify_while_replacing): If all unary/binary/relational
934 operation arguments are constant, attempt to simplify those.
937 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
938 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
940 2014-02-04 Richard Biener <rguenther@suse.de>
942 PR tree-optimization/60012
943 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
944 TBAA disambiguation to all DDRs.
946 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
949 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
950 (LINK_SPEC): Use it for -shared, -shared-libgcc.
952 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
955 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
957 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
959 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
960 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
962 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
965 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
966 to figure out targets of polymorphic calls with known decl.
967 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
968 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
969 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
970 (get_polymorphic_call_info): ... here.
971 (get_polymorphic_call_info_from_invariant): New function.
973 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
975 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
976 lookup via vtable pointer; check for type consistency
977 and turn inconsitent facts into UNREACHABLE.
978 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
979 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
980 type inconsistent querries; return UNREACHABLE instead.
982 2014-02-03 Richard Henderson <rth@twiddle.net>
985 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
986 already processed this node.
987 (normalize_one_pred_1): Pass along mark_set.
988 (normalize_one_pred): Create and destroy a pointer_set_t.
989 (normalize_one_pred_chain): Likewise.
991 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
993 PR gcov-profile/58602
994 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
996 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
999 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1000 -fno-devirtualize; try to devirtualize by the knowledge of
1001 virtual table pointer given by aggregate propagation.
1002 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1003 (ipa_print_node_jump_functions): Dump also offset that
1004 is relevant for polymorphic calls.
1005 (determine_known_aggregate_parts): Add arg_type parameter; use it
1006 instead of determining the type from pointer type.
1007 (ipa_compute_jump_functions_for_edge): Update call of
1008 determine_known_aggregate_parts.
1009 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1010 (gimple_get_virt_method_for_binfo): ... here; simplify using
1011 vtable_pointer_value_to_vtable.
1012 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1013 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1014 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1015 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1017 (vtable_pointer_value_to_binfo): ... here.
1018 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1020 2014-02-03 Teresa Johnson <tejohnson@google.com>
1022 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1023 redef of outer loop index variable.
1025 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1029 * doc/extend.texi (Function Attributes): Typo.
1031 2014-02-03 Cong Hou <congh@google.com>
1033 PR tree-optimization/60000
1034 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1035 if the vectorized statement is a store. A store statement can only
1036 appear at the end of pattern statements.
1038 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1040 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1041 (ix86_option_override_internal): Default long double to 64-bit for
1042 32-bit Bionic and to 128-bit for 64-bit Bionic.
1044 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1045 TARGET_LONG_DOUBLE_128 is true.
1046 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1048 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1049 (mlong-double-64): Negate -mlong-double-128.
1050 (mlong-double-128): New option.
1052 * config/i386/i386-c.c (ix86_target_macros): Define
1053 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1055 * doc/invoke.texi: Document -mlong-double-128.
1057 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1059 PR rtl-optimization/60024
1060 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1062 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1064 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1066 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1068 PR rtl-optimization/57662
1069 * sel-sched.c (code_motion_path_driver): Do not mark already not
1070 existing blocks in the visiting bitmap.
1072 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1074 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1075 on the insn being emitted.
1077 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1078 Will Deacon <will.deacon@arm.com>
1080 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1082 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1084 * config/arm/arm-tables.opt: Regenerate.
1086 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1088 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1089 for vector types other than V16QImode.
1090 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1091 define_expand, and call altivec_expand_vec_perm_le when producing
1092 code with little endian element order.
1093 (*altivec_vperm_<mode>_internal): New insn having previous
1094 behavior of altivec_vperm_<mode>.
1095 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1096 altivec_expand_vec_perm_le when producing code with little endian
1098 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1099 behavior of altivec_vperm_<mode>_uns.
1101 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1103 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1104 (altivec_vsumsws): Add handling for -maltivec=be with a little
1106 (altivec_vsumsws_direct): New.
1107 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1108 gen_altivec_vsumsws.
1110 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1112 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1113 vtable_pointer_value_to_binfo): New functions.
1114 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1115 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1117 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1119 * config/nios2/nios2.md (load_got_register): Initialize GOT
1120 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1121 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1123 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1125 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1126 preserverd by passthrough, do not propagate the type.
1128 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1130 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1131 (mips_atomic_assign_expand_fenv): New function.
1132 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1134 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1136 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1137 (__builtin_mips_set_fcsr): Likewise.
1138 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1139 MIPS_USI_FTYPE_VOID.
1140 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1141 (mips16_expand_set_fcsr): Likewise.
1142 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1143 (mips16_set_fcsr_stub): Likewise.
1144 (mips16_get_fcsr_one_only_stub): New class.
1145 (mips16_set_fcsr_one_only_stub): Likewise.
1146 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1147 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1148 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1149 (hard_float): New availability predicate.
1150 (mips_builtins): Add get_fcsr and set_fcsr.
1151 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1152 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1153 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1154 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1155 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1158 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1160 * config/mips/mips.c (mips_one_only_stub): New class.
1161 (mips_need_mips16_rdhwr_p): Replace with...
1162 (mips16_rdhwr_stub): ...this new variable.
1163 (mips16_stub_call_address): New function.
1164 (mips16_rdhwr_one_only_stub): New class.
1165 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1166 (mips_output_mips16_rdhwr): Delete.
1167 (mips_finish_stub): New function.
1168 (mips_code_end): Use it to handle rdhwr stubs.
1170 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1173 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1174 when calculating size of integer atomic types.
1176 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1178 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1180 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1182 PR tree-optimization/60003
1183 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1184 * profile.c (branch_prob): Use gimple_call_builtin_p
1185 to check for BUILT_IN_SETJMP_RECEIVER.
1186 * tree-inline.c (copy_bb): Call notice_special_calls.
1188 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1191 * lra-constraints.c (process_alt_operands): Update reload_sum only
1194 2014-01-31 Richard Henderson <rth@redhat.com>
1197 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1198 until after else_eh is processed.
1200 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1202 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1203 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1204 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1205 in smmintrin.h, remove them.
1206 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1207 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1208 * config/i386/i386.md (ROUND_SAE): Fix value.
1209 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1210 (const48_operand): New.
1211 * config/i386/subst.md (round), (round_expand): Use
1212 const_4_or_8_to_11_operand.
1213 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1215 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1217 * config/i386/constraints.md (Yk): Swap meaning with k.
1218 * config/i386/i386.md (movhi_internal): Change Yk to k.
1219 (movqi_internal): Ditto.
1220 (*k<logic><mode>): Ditto.
1223 (kandn<mode>): Ditto.
1224 (*<code>hi_1): Ditto.
1225 (*<code>qi_1): Ditto.
1226 (kxnor<mode>): Ditto.
1227 (kortestzhi): Ditto.
1228 (kortestchi): Ditto.
1230 (*one_cmplhi2_1): Ditto.
1231 (*one_cmplqi2_1): Ditto.
1232 * config/i386/sse.md (): Change k to Yk.
1233 (avx512f_load<mode>_mask): Ditto.
1234 (avx512f_blendm<mode>): Ditto.
1235 (avx512f_store<mode>_mask): Ditto.
1236 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1237 (avx512f_storedqu<mode>_mask): Ditto.
1238 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1240 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1241 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1242 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1243 (avx512f_maskcmp<mode>3): Ditto.
1244 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1245 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1246 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1247 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1248 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1249 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1250 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1251 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1252 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1253 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1254 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1255 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1256 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1257 (vec_extract_lo_<mode>_maskm): Ditto.
1258 (vec_extract_hi_<mode>_maskm): Ditto.
1259 (avx512f_vternlog<mode>_mask): Ditto.
1260 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1261 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1262 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1263 (avx512f_<code>v8div16qi2_mask): Ditto.
1264 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1265 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1266 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1267 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1268 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1269 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1270 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1271 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1272 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1273 (avx512cd_maskb_vec_dupv8di): Ditto.
1274 (avx512cd_maskw_vec_dupv16si): Ditto.
1275 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1276 (avx512f_vpermi2var<mode>3_mask): Ditto.
1277 (avx512f_vpermi2var<mode>3_mask): Ditto.
1278 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1279 (*avx512f_gathersi<mode>): Ditto.
1280 (*avx512f_gathersi<mode>_2): Ditto.
1281 (*avx512f_gatherdi<mode>): Ditto.
1282 (*avx512f_gatherdi<mode>_2): Ditto.
1283 (*avx512f_scattersi<mode>): Ditto.
1284 (*avx512f_scatterdi<mode>): Ditto.
1285 (avx512f_compress<mode>_mask): Ditto.
1286 (avx512f_compressstore<mode>_mask): Ditto.
1287 (avx512f_expand<mode>_mask): Ditto.
1288 * config/i386/subst.md (mask): Change k to Yk.
1289 (mask_scalar_merge): Ditto.
1292 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1294 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1296 2014-01-31 Richard Biener <rguenther@suse.de>
1299 * builtins.c (fold_builtin_memory_op): Make sure to not
1300 use a floating-point mode or a boolean or enumeral type for
1303 2014-01-30 DJ Delorie <dj@redhat.com>
1305 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1306 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1307 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1308 whenever main() has an epilogue.
1310 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1312 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1313 unused variable "field".
1314 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1315 (vsx_mergeh_<mode>): Likewise.
1316 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1317 (altivec_vmrghh): Likewise.
1318 (altivec_vmrghw): Likewise.
1319 (altivec_vmrglb): Likewise.
1320 (altivec_vmrglh): Likewise.
1321 (altivec_vmrglw): Likewise.
1322 (altivec_vspltb): Add missing uses.
1323 (altivec_vsplth): Likewise.
1324 (altivec_vspltw): Likewise.
1325 (altivec_vspltsf): Likewise.
1327 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1330 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1331 frame related instructions.
1333 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1335 PR rtl-optimization/59959
1336 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1337 any reload of register whose subreg is invalid.
1339 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1341 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1342 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1343 Add missing return type - void.
1345 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1347 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1348 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1349 remove element index adjustment for endian (now handled in vsx.md
1351 (altivec_expand_vec_perm_const): Use
1352 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1353 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1354 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1355 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1356 define_expand and a new define_insn *altivec_vspltb_internal;
1357 adjust for -maltivec=be on a little endian target.
1358 (altivec_vspltb_direct): New.
1359 (altivec_vsplth): Divide into a define_expand and a new
1360 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1361 little endian target.
1362 (altivec_vsplth_direct): New.
1363 (altivec_vspltw): Divide into a define_expand and a new
1364 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1365 little endian target.
1366 (altivec_vspltw_direct): New.
1367 (altivec_vspltsf): Divide into a define_expand and a new
1368 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1369 a little endian target.
1371 2014-01-30 Richard Biener <rguenther@suse.de>
1373 PR tree-optimization/59993
1374 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1375 can propagate form the earlier stmt and avoid the transform
1376 when the intermediate result is needed.
1378 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1380 * README.Portability: Fix typo.
1382 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1384 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1385 comparison_operator with ordered_comparison_operator.
1387 2014-01-30 Nick Clifton <nickc@redhat.com>
1389 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1390 Rename to mn10300_store_multiple_regs.
1391 * config/mn10300/mn10300.c: Likewise.
1392 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1393 store_multiple_regs.
1394 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1395 Call mn10300_store_multiple_regs.
1397 2014-01-30 Nick Clifton <nickc@redhat.com>
1398 DJ Delorie <dj@redhat.com>
1400 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1401 %fp 2 to keep registers after it properly word-aligned.
1402 (rl78_alloc_physical_registers_umul): Handle the case where both
1403 input operands are the same.
1405 2014-01-30 Richard Biener <rguenther@suse.de>
1407 PR tree-optimization/59903
1408 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1411 2014-01-30 Jason Merrill <jason@redhat.com>
1414 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1417 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1419 2014-01-30 Richard Biener <rguenther@suse.de>
1421 PR tree-optimization/59951
1422 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1424 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1427 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1428 SFmode to DFmode case.
1430 2014-01-29 DJ Delorie <dj@redhat.com>
1432 * config/msp430/msp430.opt (-minrt): New.
1433 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1435 (ENDFILE_SPEC): Likewise.
1437 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1439 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1440 (estimate_function_body_sizes): Use it.
1442 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1445 * dwarf2out.c (is_cxx_auto): New.
1446 (is_base_type): Use it.
1447 (gen_type_die_with_usage): Likewise.
1449 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1451 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1452 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1453 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1454 -maltivec=be with LE targets.
1455 (vsx_mergeh_<mode>): Likewise.
1456 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1457 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1458 (altivec_vmrghb): Replace with define_expand and new
1459 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1460 (altivec_vmrghb_direct): New define_insn.
1461 (altivec_vmrghh): Replace with define_expand and new
1462 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1463 (altivec_vmrghh_direct): New define_insn.
1464 (altivec_vmrghw): Replace with define_expand and new
1465 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1466 (altivec_vmrghw_direct): New define_insn.
1467 (*altivec_vmrghsf): Adjust for endianness.
1468 (altivec_vmrglb): Replace with define_expand and new
1469 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1470 (altivec_vmrglb_direct): New define_insn.
1471 (altivec_vmrglh): Replace with define_expand and new
1472 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1473 (altivec_vmrglh_direct): New define_insn.
1474 (altivec_vmrglw): Replace with define_expand and new
1475 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1476 (altivec_vmrglw_direct): New define_insn.
1477 (*altivec_vmrglsf): Adjust for endianness.
1478 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1479 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1480 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1481 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1482 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1483 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1484 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1485 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1487 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1489 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1490 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1493 2014-01-29 Richard Biener <rguenther@suse.de>
1495 PR tree-optimization/58742
1496 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1497 associate_pointerplus_align.
1498 (associate_pointerplus_diff): New function.
1499 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1500 and associate_pointerplus_diff.
1502 2014-01-29 Richard Biener <rguenther@suse.de>
1504 * lto-streamer.h (LTO_major_version): Bump to 3.
1505 (LTO_minor_version): Reset to 0.
1507 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1509 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1510 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1511 (arm_file_start): Generate correct asm header for armv7ve.
1512 * config/arm/bpabi.h: Add multilib support for armv7ve.
1513 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1514 and cortex-a15 to armv7ve.
1515 * config/arm/t-aprofile: Add multilib support for armv7ve.
1516 * doc/invoke.texi: Document -march=armv7ve.
1518 2014-01-29 Richard Biener <rguenther@suse.de>
1520 PR tree-optimization/58742
1521 * tree-ssa-forwprop.c (associate_plusminus): Return true
1522 if we changed sth, defer EH cleanup to ...
1523 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1524 (simplify_mult): New function.
1526 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1529 PR tree-optimization/59920
1530 * tree.c (build_common_builtin_nodes): Remove
1531 __builtin_setjmp_dispatcher initialization.
1532 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1533 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1534 instead of gsi_after_labels + manually skipping debug stmts.
1535 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1536 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1537 * tree-inline.c (copy_edges_for_bb): Remove
1538 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1539 argument. Ignore computed_goto_p stmts. Don't call
1540 make_abnormal_goto_edges. If a call might need abnormal edges
1541 for non-local gotos, see if it already has an edge to
1542 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1543 with true argument, don't do anything then, otherwise add
1544 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1545 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1547 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1548 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1549 (lower_stmt): Don't set data->calls_builtin_setjmp.
1550 (lower_builtin_setjmp): Adjust comment.
1551 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1552 * tree-cfg.c (found_computed_goto): Remove.
1553 (factor_computed_gotos): Remove.
1554 (make_goto_expr_edges): Return bool, true for computed gotos.
1555 Don't call make_abnormal_goto_edges.
1556 (build_gimple_cfg): Don't set found_computed_goto, don't call
1557 factor_computed_gotos.
1558 (computed_goto_p): No longer static.
1559 (make_blocks): Don't set found_computed_goto.
1560 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1561 (make_edges): If make_goto_expr_edges returns true, push bb
1562 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1563 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1564 vector. Record mapping between bbs and OpenMP regions if there
1565 are any, adjust make_gimple_omp_edges caller. Call
1566 handle_abnormal_edges.
1567 (make_abnormal_goto_edges): Remove.
1568 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1569 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1570 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1571 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1572 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1573 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1574 filling *region also set *region_idx to (*region)->entry->index.
1577 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1578 For REGs set ORIGINAL_REGNO.
1580 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1582 * doc/md.texi: Mention that a target shouldn't implement
1583 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1586 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1588 PR tree-optimization/59594
1589 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1590 a copy of the datarefs vector rather than the vector itself.
1592 2014-01-28 Jason Merrill <jason@redhat.com>
1595 * dwarf2out.c (auto_die): New static.
1596 (gen_type_die_with_usage): Handle C++1y 'auto'.
1597 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1600 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1603 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1604 (SPEC_X32): Likewise.
1605 (SPEC_64): Likewise.
1606 * config/i386/i386.c (ix86_option_override_internal): Turn off
1607 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1609 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1610 * config/i386/i386.h (TARGET_16BIT): New macro.
1611 (TARGET_16BIT_P): Likewise.
1612 * config/i386/i386.opt: Add m16.
1613 * doc/invoke.texi: Document -m16.
1615 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1617 PR preprocessor/59935
1618 * input.c (location_get_source_line): Bail out on when line number
1619 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1621 2014-01-28 Richard Biener <rguenther@suse.de>
1623 PR tree-optimization/58742
1624 * tree-ssa-forwprop.c (associate_plusminus): Handle
1625 pointer subtraction of the form (T)(P + A) - (T)P.
1627 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1629 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1632 2014-01-28 Richard Biener <rguenther@suse.de>
1635 2014-01-28 Richard Biener <rguenther@suse.de>
1637 PR rtl-optimization/45364
1638 PR rtl-optimization/59890
1639 * var-tracking.c (local_get_addr_clear_given_value): Handle
1640 already cleared slot.
1641 (val_reset): Handle not allocated local_get_addr_cache.
1642 (vt_find_locations): Use post-order on the inverted CFG.
1644 2014-01-28 Richard Biener <rguenther@suse.de>
1646 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1648 2014-01-28 Richard Biener <rguenther@suse.de>
1650 PR rtl-optimization/45364
1651 PR rtl-optimization/59890
1652 * var-tracking.c (local_get_addr_clear_given_value): Handle
1653 already cleared slot.
1654 (val_reset): Handle not allocated local_get_addr_cache.
1655 (vt_find_locations): Use post-order on the inverted CFG.
1657 2014-01-28 Alan Modra <amodra@gmail.com>
1659 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1660 * configure.ac <recursive call for build != host>: Define
1661 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1662 and LD_FOR_BUILD too.
1663 * configure: Regenerate.
1665 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1667 * config/i386/i386.c (get_builtin_code_for_version): Separate
1668 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1669 Broadwell from Haswell.
1671 2014-01-27 Steve Ellcey <sellcey@mips.com>
1673 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1674 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1675 * config/mips/mips.c (mips_option_override): Change setting
1677 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1678 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1679 Change from Mask to Var.
1681 2014-01-27 Jeff Law <law@redhat.com>
1683 * ipa-inline.c (inline_small_functions): Fix typo.
1685 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1687 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1688 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1689 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1690 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1691 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1692 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1693 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1694 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1695 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1696 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1697 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1698 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1699 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1700 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1701 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1702 (_mm512_storeu_epi64): Ditto.
1703 (_mm512_cmpge_epi32_mask): Ditto.
1704 (_mm512_cmpge_epu32_mask): Ditto.
1705 (_mm512_cmpge_epi64_mask): Ditto.
1706 (_mm512_cmpge_epu64_mask): Ditto.
1707 (_mm512_cmple_epi32_mask): Ditto.
1708 (_mm512_cmple_epu32_mask): Ditto.
1709 (_mm512_cmple_epi64_mask): Ditto.
1710 (_mm512_cmple_epu64_mask): Ditto.
1711 (_mm512_cmplt_epi32_mask): Ditto.
1712 (_mm512_cmplt_epu32_mask): Ditto.
1713 (_mm512_cmplt_epi64_mask): Ditto.
1714 (_mm512_cmplt_epu64_mask): Ditto.
1715 (_mm512_cmpneq_epi32_mask): Ditto.
1716 (_mm512_cmpneq_epu32_mask): Ditto.
1717 (_mm512_cmpneq_epi64_mask): Ditto.
1718 (_mm512_cmpneq_epu64_mask): Ditto.
1719 (_mm512_expand_pd): Ditto.
1720 (_mm512_expand_ps): Ditto.
1721 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1722 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1723 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1724 * config/i386/i386.c (ix86_builtins): Add
1725 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1726 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1727 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1728 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1729 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1730 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1731 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1732 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1733 IX86_BUILTIN_PMOVUSQW512_MEM.
1734 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1735 __builtin_ia32_pmovsqd512mem_mask,
1736 __builtin_ia32_pmovqd512mem_mask,
1737 __builtin_ia32_pmovusqw512mem_mask,
1738 __builtin_ia32_pmovsqw512mem_mask,
1739 __builtin_ia32_pmovqw512mem_mask,
1740 __builtin_ia32_pmovusdw512mem_mask,
1741 __builtin_ia32_pmovsdw512mem_mask,
1742 __builtin_ia32_pmovdw512mem_mask,
1743 __builtin_ia32_pmovqb512mem_mask,
1744 __builtin_ia32_pmovusqb512mem_mask,
1745 __builtin_ia32_pmovsqb512mem_mask,
1746 __builtin_ia32_pmovusdb512mem_mask,
1747 __builtin_ia32_pmovsdb512mem_mask,
1748 __builtin_ia32_pmovdb512mem_mask.
1749 (bdesc_args): Add __builtin_ia32_expanddf512,
1750 __builtin_ia32_expandsf512.
1751 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1752 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1753 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1754 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1755 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1756 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1757 (avx512f_<code>v8div16qi2_mask_store): This.
1758 (avx512f_expand<mode>): New.
1760 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1762 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1764 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1765 (_mm512_prefetch_i32scatter_pd): Ditto.
1766 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1767 (_mm512_prefetch_i64scatter_pd): Ditto.
1768 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1769 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1770 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1771 (_mm512_prefetch_i32scatter_ps): Ditto.
1772 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1773 (_mm512_prefetch_i64scatter_ps): Ditto.
1774 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1775 * config/i386/i386-builtin-types.def: Define
1776 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1777 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1778 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1779 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1780 IX86_BUILTIN_SCATTERPFQPD.
1781 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1782 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1783 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1784 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1785 __builtin_ia32_scatterpfqps.
1786 (ix86_expand_builtin): Expand new built-ins.
1787 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1788 fix memory access data type.
1789 (*avx512pf_gatherpf<mode>_mask): Ditto.
1790 (*avx512pf_gatherpf<mode>): Ditto.
1791 (avx512pf_scatterpf<mode>): Ditto.
1792 (*avx512pf_scatterpf<mode>_mask): Ditto.
1793 (*avx512pf_scatterpf<mode>): Ditto.
1794 (GATHER_SCATTER_SF_MEM_MODE): New.
1795 (avx512pf_gatherpf<mode>df): Ditto.
1796 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1797 (*avx512pf_scatterpf<mode>df): Ditto.
1799 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1802 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1803 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1806 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1808 * common/config/arm/arm-common.c
1809 (arm_rewrite_mcpu): Handle multiple names.
1811 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1813 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1815 * gimple-builder.h (create_gimple_tmp): Delete.
1817 2014-01-27 Christian Bruel <christian.bruel@st.com>
1819 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1822 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1824 * config/pa/pa.md (call): Generate indirect long calls to non-local
1825 functions when outputing 32-bit code.
1826 (call_value): Likewise except for special call to buggy powf function.
1828 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1829 portable runtime and PIC indirect calls.
1830 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1831 and PIC call sequences. Use ldo instead of blr to set return register
1832 in PIC call sequence.
1834 2014-01-25 Walter Lee <walt@tilera.com>
1836 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1837 avoid clobbering a live register.
1839 2014-01-25 Walter Lee <walt@tilera.com>
1841 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1842 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1843 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1844 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1846 2014-01-25 Walter Lee <walt@tilera.com>
1848 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1849 arguments on even registers.
1850 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1852 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1853 (BIGGEST_ALIGNMENT): Ditto.
1854 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1856 2014-01-25 Walter Lee <walt@tilera.com>
1858 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1859 insns before bundling.
1860 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1862 2014-01-25 Walter Lee <walt@tilera.com>
1864 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1865 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1866 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1868 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1870 * config/mips/constraints.md (kl): Delete.
1871 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1872 define expands, using...
1873 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1874 instructions for MIPS16.
1875 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1876 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1878 2014-01-25 Walter Lee <walt@tilera.com>
1880 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1884 2014-01-25 Walter Lee <walt@tilera.com>
1886 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1887 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1889 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1891 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1894 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1896 * print-rtl.c (in_call_function_usage): New var.
1897 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1898 EXPR_LIST mode as mode and not as reg note name.
1901 * cfgloopmanip.c (copy_loop_info): If
1902 loop->warned_aggressive_loop_optimizations, make sure
1903 the flag is set in target loop too.
1905 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1907 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1909 * builtins.def: Likewise.
1910 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1911 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1912 * ira.c (ira_setup_eliminable_regset): Likewise.
1913 * omp-low.c (gate_expand_omp): Likewise.
1914 (execute_lower_omp): Likewise.
1915 (diagnose_sb_0): Likewise.
1916 (gate_diagnose_omp_blocks): Likewise.
1917 (simd_clone_clauses_extract): Likewise.
1920 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1922 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1923 correction for little endian...
1924 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1927 2014-01-24 Jeff Law <law@redhat.com>
1929 PR tree-optimization/59919
1930 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1931 for non-returning calls.
1933 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1935 * common/config/aarch64/aarch64-common.c
1936 (aarch64_rewrite_mcpu): Handle multiple names.
1937 * config/aarch64/aarch64.h
1938 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1940 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1942 * input.c (add_file_to_cache_tab): Handle the case where fopen
1945 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1948 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1949 from push operand if code of push isn't PRE_DEC.
1951 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1954 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1955 -mquad-memory-atomic. Update -mquad-memory documentation to say
1956 it is only used for non-atomic loads/stores.
1958 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1959 -mquad-memory or -mquad-memory-atomic switches.
1961 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1962 -mquad-memory-atomic to ISA 2.07 support.
1964 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1965 to separate support of normal quad word memory operations (ldq, stq)
1966 from the atomic quad word memory operations.
1968 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1969 support to separate non-atomic quad word operations from atomic
1970 quad word operations. Disable non-atomic quad word operations in
1971 little endian mode so that we don't have to swap words after the
1972 load and before the store.
1973 (quad_load_store_p): Add comment about atomic quad word support.
1974 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1975 options printed with -mdebug=reg.
1977 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1978 -mquad-memory-atomic as the test for whether we have quad word
1979 atomic instructions.
1980 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1981 or -mp8-vector are used, allow byte/half-word atomic operations.
1983 * config/rs6000/sync.md (load_lockedti): Insure that the address
1984 is a proper indexed or indirect address for the lqarx instruction.
1985 On little endian systems, swap the hi/lo registers after the lqarx
1987 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1988 insure the address is valid for the lqarx instruction.
1989 (store_conditionalti): Insure that the address is a proper indexed
1990 or indirect address for the stqcrx. instruction. On little endian
1991 systems, swap the hi/lo registers before doing the stqcrx.
1993 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1994 insure the address is valid for the stqcrx. instruction.
1996 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1997 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1998 type of quad memory support is available.
2000 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2003 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2004 there is a danger of looping.
2006 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2008 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2009 force flag_ira_loop_pressure if set via command line.
2011 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2013 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2014 (ashr_simd): New builtin handling DI mode.
2015 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2016 (aarch64_sshr_simddi): New match pattern.
2017 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2018 (vshrd_n_s64): Likewise.
2019 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2021 2014-01-23 Nick Clifton <nickc@redhat.com>
2023 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2024 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2025 favour of mcu specific scripts.
2026 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2029 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2030 Alex Velenko <Alex.Velenko@arm.com>
2032 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2033 (vaddv_s16): Likewise.
2034 (vaddv_s32): Likewise.
2035 (vaddv_u8): Likewise.
2036 (vaddv_u16): Likewise.
2037 (vaddv_u32): Likewise.
2038 (vaddvq_s8): Likewise.
2039 (vaddvq_s16): Likewise.
2040 (vaddvq_s32): Likewise.
2041 (vaddvq_s64): Likewise.
2042 (vaddvq_u8): Likewise.
2043 (vaddvq_u16): Likewise.
2044 (vaddvq_u32): Likewise.
2045 (vaddvq_u64): Likewise.
2046 (vaddv_f32): Likewise.
2047 (vaddvq_f32): Likewise.
2048 (vaddvq_f64): Likewise.
2049 (vmaxv_f32): Likewise.
2050 (vmaxv_s8): Likewise.
2051 (vmaxv_s16): Likewise.
2052 (vmaxv_s32): Likewise.
2053 (vmaxv_u8): Likewise.
2054 (vmaxv_u16): Likewise.
2055 (vmaxv_u32): Likewise.
2056 (vmaxvq_f32): Likewise.
2057 (vmaxvq_f64): Likewise.
2058 (vmaxvq_s8): Likewise.
2059 (vmaxvq_s16): Likewise.
2060 (vmaxvq_s32): Likewise.
2061 (vmaxvq_u8): Likewise.
2062 (vmaxvq_u16): Likewise.
2063 (vmaxvq_u32): Likewise.
2064 (vmaxnmv_f32): Likewise.
2065 (vmaxnmvq_f32): Likewise.
2066 (vmaxnmvq_f64): Likewise.
2067 (vminv_f32): Likewise.
2068 (vminv_s8): Likewise.
2069 (vminv_s16): Likewise.
2070 (vminv_s32): Likewise.
2071 (vminv_u8): Likewise.
2072 (vminv_u16): Likewise.
2073 (vminv_u32): Likewise.
2074 (vminvq_f32): Likewise.
2075 (vminvq_f64): Likewise.
2076 (vminvq_s8): Likewise.
2077 (vminvq_s16): Likewise.
2078 (vminvq_s32): Likewise.
2079 (vminvq_u8): Likewise.
2080 (vminvq_u16): Likewise.
2081 (vminvq_u32): Likewise.
2082 (vminnmv_f32): Likewise.
2083 (vminnmvq_f32): Likewise.
2084 (vminnmvq_f64): Likewise.
2086 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2088 * config/aarch64/aarch64-simd.md
2089 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2090 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2091 (*aarch64_mul3_elt<mode>): Likewise.
2092 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2093 (*aarch64_mul3_elt_to_64v2df): Likewise.
2094 (*aarch64_mla_elt<mode>): Likewise.
2095 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2096 (*aarch64_mls_elt<mode>): Likewise.
2097 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2098 (*aarch64_fma4_elt<mode>): Likewise.
2099 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2100 (*aarch64_fma4_elt_to_64v2df): Likewise.
2101 (*aarch64_fnma4_elt<mode>): Likewise.
2102 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2103 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2104 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2105 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2106 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2107 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2108 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2109 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2110 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2112 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2114 * config/aarch64/aarch64-simd.md
2115 (aarch64_be_checked_get_lane<mode>): New define_expand.
2116 * config/aarch64/aarch64-simd-builtins.def
2117 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2118 New builtin definition.
2119 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2120 Use new safe be builtin.
2122 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2124 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2126 (aarch64_be_st1<mode>): Likewise.
2127 (aarch_ld1<VALL:mode>): Define_expand modified.
2128 (aarch_st1<VALL:mode>): Likewise.
2129 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2130 (UNSPEC_ST1): Likewise.
2132 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2134 * config/microblaze/microblaze.md: Add trap insn and attribute
2136 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2138 PR preprocessor/58580
2139 * input.h (location_get_source_line): Take an additional line_size
2141 (void diagnostics_file_cache_fini): Declare new function.
2142 * input.c (struct fcache): New type.
2143 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2144 New static constants.
2145 (diagnostic_file_cache_init, total_lines_num)
2146 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2147 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2148 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2149 (get_next_line, read_next_line, goto_next_line, read_line_num):
2150 New static function definitions.
2151 (diagnostic_file_cache_fini): New function.
2152 (location_get_source_line): Take an additional output line_len
2153 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2155 * diagnostic.c (diagnostic_finish): Call
2156 diagnostic_file_cache_fini.
2157 (adjust_line): Take an additional input parameter for the length
2158 of the line, rather than calculating it with strlen.
2159 (diagnostic_show_locus): Adjust the use of
2160 location_get_source_line and adjust_line with respect to their new
2161 signature. While displaying a line now, do not stop at the first
2162 null byte. Rather, display the zero byte as a space and keep
2163 going until we reach the size of the line.
2164 * Makefile.in: Add vec.o to OBJS-libcommon
2166 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2167 Ilya Tocar <ilya.tocar@intel.com>
2169 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2170 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2171 (__builtin_ia32_kmov16): Ditto.
2172 * config/i386/i386.md (UNSPEC_KMOV): New.
2175 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2177 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2178 (_mm512_storeu_si512): Ditto.
2180 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2183 * rtl.h (get_referenced_operands): Declare.
2184 * recog.c (get_referenced_operands): New function.
2185 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2186 operands have been referenced when recording LO_SUM references.
2188 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2190 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2192 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2194 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2195 Enable for generic and recent AMD targets.
2197 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2199 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2200 ARG_SIZE note when adjustment was eliminated.
2202 2014-01-22 Jeff Law <law@redhat.com>
2204 PR tree-optimization/59597
2205 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2206 in file. Accept new argument REGISTERING and use it to modify
2207 dump output appropriately.
2208 (register_jump_thread): Corresponding changes.
2209 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2210 thread paths involving joiner blocks. Add code to dump cancelled
2211 jump threading paths.
2213 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2215 PR rtl-optimization/59477
2216 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2217 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2219 2014-01-22 Tom Tromey <tromey@redhat.com>
2221 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2223 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2225 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2227 PR rtl-optimization/59896
2228 * lra-constraints.c (process_alt_operands): Check unused note for
2229 matched operands of insn with no output reloads.
2231 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2233 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2234 (mips_move_from_gpr_cost): Likewise.
2236 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2238 PR rtl-optimization/59858
2239 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2240 ira_class_hard_regs_num.
2241 (process_alt_operands): Increase reject for dying matched operand.
2243 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2246 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2247 smaller than size, perform several stores or loads and stores
2248 at dst + count - size to store or copy all of size bytes, rather
2249 than just last modesize bytes.
2251 2014-01-20 DJ Delorie <dj@redhat.com>
2253 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2254 that CLOBBERs are REGs before propogating their values.
2256 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2259 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2260 (cgraph_inline_failed_type): New function.
2261 * cgraph.h (DEFCIFCODE): Add type.
2262 (cgraph_inline_failed_type_t): New enum.
2263 (cgraph_inline_failed_type): New prototype.
2264 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2265 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2266 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2267 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2268 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2269 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2270 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2271 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2272 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2273 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2274 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2275 OPTIMIZATION_MISMATCH.
2276 * tree-inline.c (expand_call_inline): Emit errors during
2277 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2279 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2282 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2283 mode attribute in insn output.
2285 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2287 * output.h (output_constant): Delete.
2288 * varasm.c (output_constant): Make private.
2290 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2292 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2294 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2297 * tree.h (fold_builtin_strcat): New prototype.
2298 * builtins.c (fold_builtin_strcat): No longer static. Add len
2299 argument, if non-NULL, don't call c_strlen. Optimize
2300 directly into __builtin_memcpy instead of __builtin_strcpy.
2301 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2302 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2304 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2306 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2307 for SImode_address_operand operands, having only a REG argument.
2309 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2311 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2312 loader name using mbig-endian.
2313 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2315 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2317 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2321 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2323 * config/aarch64/aarch64-protos.h
2324 (aarch64_cannot_change_mode_class_ptr): Declare.
2325 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2326 aarch64_cannot_change_mode_class_ptr): New.
2327 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2328 backend hook aarch64_cannot_change_mode_class.
2330 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2332 * common/config/aarch64/aarch64-common.c
2333 (aarch64_handle_option): Don't handle any option order logic here.
2334 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2335 selected_cpu, warn on architecture version mismatch.
2336 (aarch64_override_options): Fix parsing order for option strings.
2338 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2339 Iain Sandoe <iain@codesourcery.com>
2342 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2343 warning. Amend comment to reflect current functionality.
2345 2014-01-20 Richard Biener <rguenther@suse.de>
2348 * builtins.c (fold_builtin_strcat): Remove case better handled
2349 by tree-ssa-strlen.c.
2351 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2353 * config/aarch64/aarch64.opt
2354 (mcpu, march, mtune): Make case-insensitive.
2356 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2359 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2360 if operands[1] is a REG or ZERO_EXTEND of a REG.
2362 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2364 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2366 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2368 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2369 long non-pic millicode calls.
2371 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2373 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2375 2014-01-19 Kito Cheng <kito@0xlab.org>
2377 * builtins.c (expand_movstr): Check movstr expand done or fail.
2379 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2380 H.J. Lu <hongjiu.lu@intel.com>
2383 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2384 to DImode for zero-extended addresses.
2386 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2388 PR rtl-optimization/57763
2389 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2390 on the new indirect jump_insn and increment LABEL_NUSES (label).
2392 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2396 * config.gcc (x86_archs): New variable.
2397 (x86_64_archs): Likewise.
2398 (x86_cpus): Likewise.
2399 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2400 --with-arch/--with-cpu= options.
2401 Support --with-arch=/--with-cpu={nehalem,westmere,
2402 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2404 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2406 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2407 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2409 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2411 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2413 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2416 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2417 clear cpp_get_options (parse_in)->warn_unused_macros for
2418 ix86_target_macros_internal with cpp_define.
2420 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2422 * jump.c (delete_related_insns): Keep (use (insn))s.
2423 * reorg.c (redundant_insn): Check for barriers too.
2425 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2427 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2429 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2431 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2432 call to $$dyncall when TARGET_LONG_CALLS is true.
2434 2014-01-17 Jeff Law <law@redhat.com>
2436 * ree.c (combine_set_extension): Temporarily disable test for
2437 changing number of hard registers.
2439 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2442 * ipa-inline-analysis.c (inline_free_summary):
2443 Do not free summary of aliases.
2445 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2448 * gimplify.c (gimplify_expr): Use create_tmp_var
2449 instead of create_tmp_var_raw. If cond doesn't have
2450 integral type, don't add the IFN_ANNOTATE builtin at all.
2452 2014-01-17 Martin Jambor <mjambor@suse.cz>
2455 * ipa-cp.c (prev_edge_clone): New variable.
2456 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2457 Also resize prev_edge_clone vector.
2458 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2459 (ipcp_edge_removal_hook): New function.
2460 (ipcp_driver): Register ipcp_edge_removal_hook.
2462 2014-01-17 Andrew Pinski <apinski@cavium.com>
2463 Steve Ellcey <sellcey@mips.com>
2466 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2469 2014-01-17 Jeff Law <law@redhat.com>
2472 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2473 so that pass_ccp runs first.
2475 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2477 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2478 (ix86_adjust_cost): Use !TARGET_XXX.
2479 (do_reorder_for_imul): Likewise.
2480 (swap_top_of_ready_list): Likewise.
2481 (ix86_sched_reorder): Likewise.
2483 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2485 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2486 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2487 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2488 (intel_memset): New. Duplicate slm_memset.
2489 (intel_cost): New. Duplicate slm_cost.
2490 (m_INTEL): New macro.
2491 (processor_target_table): Add "intel".
2492 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2493 with PROCESSOR_INTEL for "intel".
2494 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2495 PROCESSOR_SILVERMONT.
2496 (ix86_issue_rate): Likewise.
2497 (ix86_adjust_cost): Likewise.
2498 (ia32_multipass_dfa_lookahead): Likewise.
2499 (swap_top_of_ready_list): Likewise.
2500 (ix86_sched_reorder): Likewise.
2501 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2502 instead of TARGET_OPT_AGU.
2503 * config/i386/i386.h (TARGET_INTEL): New.
2504 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2505 (processor_type): Add PROCESSOR_INTEL.
2506 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2507 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2509 2014-01-17 Marek Polacek <polacek@redhat.com>
2512 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2515 2014-01-17 Richard Biener <rguenther@suse.de>
2517 PR tree-optimization/46590
2518 * opts.c (default_options_table): Add entries for
2519 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2520 all enabled at -O1 but not for -Og.
2521 * common.opt (fbranch-count-reg): Remove Init(1).
2522 (fmove-loop-invariants): Likewise.
2523 (ftree-pta): Likewise.
2525 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2527 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2528 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2529 decls to at least the GCC 4.8 used alignments.
2532 * tree-nested.c (convert_nonlocal_reference_stmt,
2533 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2534 of GIMPLE_BIND stmts, adjust associated decls.
2536 2014-01-17 Richard Biener <rguenther@suse.de>
2538 PR tree-optimization/46590
2539 * vec.h (vec<>::bseach): New member function implementing
2540 binary search according to C89 bsearch.
2541 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2542 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2543 bitmap pointer again. Make accesses_in_loop a flat array.
2544 (mem_ref_obstack): New global.
2545 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2546 (mark_ref_stored): Likewise.
2547 (ref_indep_loop_p_2): Likewise.
2548 (set_ref_stored_in_loop): New helper function.
2549 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2550 (memref_free): Adjust.
2551 (record_mem_ref_loc): Simplify.
2552 (gather_mem_refs_stmt): Adjust.
2553 (sort_locs_in_loop_postorder_cmp): New function.
2554 (analyze_memory_references): Sort accesses_in_loop after
2555 loop postorder number.
2556 (find_ref_loc_in_loop_cmp): New function.
2557 (for_all_locs_in_loop): Find relevant cluster of locs in
2558 accesses_in_loop and iterate without recursion.
2559 (execute_sm): Avoid uninit warning.
2560 (struct ref_always_accessed): Simplify.
2561 (ref_always_accessed::operator ()): Likewise.
2562 (ref_always_accessed_p): Likewise.
2563 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2564 loop postorder numbers here.
2565 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2568 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2571 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2572 on decls for which assemble_alias has been called.
2574 2014-01-17 Nick Clifton <nickc@redhat.com>
2576 * config/msp430/msp430.opt: (mcpu): New option.
2577 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2578 (msp430_option_override): Parse target_cpu. If the MCU name
2579 matches a generic string, clear target_mcu.
2580 (msp430_attr): Allow numeric interrupt values up to 63.
2581 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2582 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2584 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2586 * config/msp430/msp430.md (popm): Use %J rather than %I.
2587 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2588 (addhi_cy_i): Use immediate_operand for operand 2.
2589 * doc/invoke.texi: Document -mcpu option.
2591 2014-01-17 Richard Biener <rguenther@suse.de>
2593 PR rtl-optimization/38518
2594 * df.h (df_analyze_loop): Declare.
2595 * df-core.c: Include cfgloop.h.
2596 (df_analyze_1): Split out main part of df_analyze.
2597 (df_analyze): Adjust.
2598 (loop_inverted_post_order_compute): New function.
2599 (loop_post_order_compute): Likewise.
2600 (df_analyze_loop): New function avoiding whole-function
2602 * loop-invariant.c (find_defs): Use df_analyze_loop.
2603 (find_invariants): Adjust.
2604 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2606 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2608 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2609 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2611 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2613 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2614 traversal when removing references.
2616 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2619 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2621 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2624 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2625 pushing a reload for an autoinc when we had previously reloaded an
2626 inner part of the address.
2628 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2630 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2632 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2633 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2634 when not giving up or versioning for alias only because of
2636 (vect_analyze_data_ref_dependences): Set to true.
2637 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2639 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2640 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2644 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2647 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2648 operand 0 predicate for gathers, use a new pseudo as subtarget.
2650 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2653 * lra-constraints.c (process_alt_operands): Add printing debug info.
2654 Check absence of input/output reloads for matched operands too.
2656 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2658 PR rtl-optimization/59835
2659 * ira.c (ira_init_register_move_cost): Increase cost for
2662 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2664 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2666 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2669 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2670 non-register objects. Use gen_(high/low)part more consistently.
2673 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2676 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2677 endian support, remove tests for WORDS_BIG_ENDIAN.
2678 (p8_mfvsrd_3_<mode>): Likewise.
2679 (reload_gpr_from_vsx<mode>): Likewise.
2680 (reload_gpr_from_vsxsf): Likewise.
2681 (p8_mfvsrd_4_disf): Likewise.
2683 2014-01-16 Richard Biener <rguenther@suse.de>
2685 PR rtl-optimization/46590
2686 * lcm.c (compute_antinout_edge): Use postorder iteration.
2687 (compute_laterin): Use inverted postorder iteration.
2689 2014-01-16 Nick Clifton <nickc@redhat.com>
2692 * varasm.c (output_constant): Return the number of bytes actually
2694 (output_constructor_array_range): Update the field size with the
2695 number of bytes emitted by output_constant.
2696 (output_constructor_regular_field): Likewise. Also do not
2697 complain if the total number of bytes emitted is now greater
2698 than the expected fieldpos.
2699 * output.h (output_constant): Update prototype and descriptive comment.
2701 2014-01-16 Marek Polacek <polacek@redhat.com>
2704 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2705 it is error_mark_node.
2707 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2709 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2710 VALID_AVX256_REG_OR_OI_MODE.
2712 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2714 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2715 current procedure should be profiled.
2717 2014-01-15 Andrew Pinski <apinski@cavium.com>
2719 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2720 of moving from/to the STACK_REG register class.
2722 2014-01-15 Richard Henderson <rth@redhat.com>
2725 * reginfo.c (global_regs_decl): Globalize.
2726 * rtl.h (global_regs_decl): Declare.
2727 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2728 reserved via global_regs.
2730 2014-01-15 Teresa Johnson <tejohnson@google.com>
2732 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2734 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2736 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2737 and vmulosh rather than call gen_vec_widen_smult_*.
2738 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2739 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2740 (vec_widen_smult_even_v16qi): Likewise.
2741 (vec_widen_umult_even_v8hi): Likewise.
2742 (vec_widen_smult_even_v8hi): Likewise.
2743 (vec_widen_umult_odd_v16qi): Likewise.
2744 (vec_widen_smult_odd_v16qi): Likewise.
2745 (vec_widen_umult_odd_v8hi): Likewise.
2746 (vec_widen_smult_odd_v8hi): Likewise.
2747 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2748 vmuloub rather than call gen_vec_widen_umult_*.
2749 (vec_widen_umult_lo_v16qi): Likewise.
2750 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2751 vmulosb rather than call gen_vec_widen_smult_*.
2752 (vec_widen_smult_lo_v16qi): Likewise.
2753 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2754 rather than call gen_vec_widen_umult_*.
2755 (vec_widen_umult_lo_v8hi): Likewise.
2756 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2757 rather than call gen_vec_widen_smult_*.
2758 (vec_widen_smult_lo_v8hi): Likewise.
2760 2014-01-15 Jeff Law <law@redhat.com>
2762 PR tree-optimization/59747
2763 * ree.c (find_and_remove_re): Properly handle case where a second
2764 eliminated extension requires widening a copy created for elimination
2765 of a prior extension.
2766 (combine_set_extension): Ensure that the number of hard regs needed
2767 for a destination register does not change when we widen it.
2769 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2771 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2772 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2773 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2774 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2775 (avr-*-rtems*): Likewise.
2776 (bfin*-rtems*): Likewise.
2777 (moxie-*-rtems*): Likewise.
2778 (h8300-*-rtems*): Likewise.
2779 (i[34567]86-*-rtems*): Likewise.
2780 (lm32-*-rtems*): Likewise.
2781 (m32r-*-rtems*): Likewise.
2782 (m68k-*-rtems*): Likewise.
2783 (microblaze*-*-rtems*): Likewise.
2784 (mips*-*-rtems*): Likewise.
2785 (powerpc-*-rtems*): Likewise.
2786 (sh-*-rtems*): Likewise.
2787 (sparc-*-rtems*): Likewise.
2788 (sparc64-*-rtems*): Likewise.
2789 (v850-*-rtems*): Likewise.
2790 (m32c-*-rtems*): Likewise.
2792 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2794 PR rtl-optimization/59511
2795 * ira.c (ira_init_register_move_cost): Use memory costs for some
2796 cases of register move cost calculations.
2797 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2798 instead of BB frequency.
2799 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2800 * lra-assigns.c (find_hard_regno_for): Ditto.
2802 2014-01-15 Richard Biener <rguenther@suse.de>
2804 PR tree-optimization/59822
2805 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2806 (vectorizable_load): Use it to hoist defs of uses of invariant
2807 loads out of the loop.
2809 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2810 Kugan Vivekanandarajah <kuganv@linaro.org>
2813 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2816 2014-01-15 Richard Biener <rguenther@suse.de>
2818 PR rtl-optimization/59802
2819 * lcm.c (compute_available): Use inverted postorder to seed
2820 the initial worklist.
2822 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2825 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2826 ADDR_REGS for invalid symrefs in non-PIC code.
2828 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2831 * builtins.c (determine_block_size): Initialize *probable_max_size
2832 even if len_rtx is CONST_INT.
2834 2014-01-14 Andrew Pinski <apinski@cavium.com>
2836 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2837 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2838 (cortexa53_tunings): Likewise.
2839 (aarch64_sched_issue_rate): New function.
2840 (TARGET_SCHED_ISSUE_RATE): Define.
2842 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2844 * ira-costs.c (find_costs_and_classes): Add missed
2845 ira_init_register_move_cost_if_necessary.
2847 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2850 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2852 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2855 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2856 to indicate if type is used for function return value. Warn ABI
2857 change if the vector mode isn't available for function return value.
2858 (ix86_function_arg_advance): Pass false to type_natural_mode.
2859 (ix86_function_arg): Likewise.
2860 (ix86_gimplify_va_arg): Likewise.
2861 (function_arg_32): Don't warn ABI change.
2862 (ix86_function_value): Pass true to type_natural_mode.
2863 (ix86_return_in_memory): Likewise.
2864 (ix86_struct_value_rtx): Removed.
2865 (TARGET_STRUCT_VALUE_RTX): Likewise.
2867 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2869 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2870 converting a conditional jump into a conditional return.
2872 2014-01-14 Richard Biener <rguenther@suse.de>
2874 PR tree-optimization/58921
2875 PR tree-optimization/59006
2876 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2877 hoisting invariant stmts.
2878 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2879 invariant loads on the preheader edge if possible.
2881 2014-01-14 Joey Ye <joey.ye@arm.com>
2883 * doc/plugin.texi (Building GCC plugins): Update to C++.
2885 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2887 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2888 (_mm_rcp28_round_ss): Ditto.
2889 (_mm_rsqrt28_round_sd): Ditto.
2890 (_mm_rsqrt28_round_ss): Ditto.
2891 (_mm_rcp28_sd): Ditto.
2892 (_mm_rcp28_ss): Ditto.
2893 (_mm_rsqrt28_sd): Ditto.
2894 (_mm_rsqrt28_ss): Ditto.
2895 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2896 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2897 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2898 (IX86_BUILTIN_RCP28SD): Ditto.
2899 (IX86_BUILTIN_RCP28SS): Ditto.
2900 (IX86_BUILTIN_RSQRT28SD): Ditto.
2901 (IX86_BUILTIN_RSQRT28SS): Ditto.
2902 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2903 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2904 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2905 (ix86_expand_special_args_builtin): Expand new FTYPE.
2906 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2907 (srcp14<mode>): Make insn unary.
2908 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2909 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2910 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2911 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2912 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2913 Fix rounding: make it SAE only.
2914 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2916 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2918 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2919 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2920 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2921 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2922 (round_saeonly_mask_scalar_operand4): Ditto.
2923 (round_saeonly_mask_scalar_op3): Ditto.
2924 (round_saeonly_mask_scalar_op4): Ditto.
2926 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2928 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2929 Implement -maltivec=be for vec_insert and vec_extract.
2931 2014-01-10 DJ Delorie <dj@redhat.com>
2933 * config/msp430/msp430.md (call_internal): Don't allow memory
2934 references with SP as the base register.
2935 (call_value_internal): Likewise.
2936 * config/msp430/constraints.md (Yc): New. For memory references
2937 that don't use SP as a base register.
2939 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2940 "an integer without a # prefix"
2941 * config/msp430/msp430.md (epilogue_helper): Use it.
2943 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2946 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2947 AVX512F gather builtins.
2948 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2949 on gather decls with INTEGER_TYPE masktype.
2950 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2951 directly into the builtin rather than hoisting it before loop.
2953 PR tree-optimization/59387
2954 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2955 (scev_const_prop): If folded_casts and type has undefined overflow,
2956 use force_gimple_operand instead of force_gimple_operand_gsi and
2957 for each added stmt if it is assign with
2958 arith_code_with_undefined_signed_overflow, call
2959 rewrite_to_defined_overflow.
2960 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2961 gimple-fold.h instead.
2962 (arith_code_with_undefined_signed_overflow,
2963 rewrite_to_defined_overflow): Moved to ...
2964 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2965 rewrite_to_defined_overflow): ... here. No longer static.
2966 Include gimplify-me.h.
2967 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2968 rewrite_to_defined_overflow): New prototypes.
2970 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2972 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2974 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2976 * builtins.c (get_object_alignment_2): Minor tweak.
2977 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2979 2014-01-13 Christian Bruel <christian.bruel@st.com>
2981 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2982 optimized non constant lengths.
2984 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2987 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2988 load as __atomic_load_N if possible.
2990 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2992 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2994 (rs6000_expand_builtin): Adjust call.
2996 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2999 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3000 * config/rs6000/rs6000.c: Include target-globals.h.
3001 (rs6000_set_current_function): Instead of doing target_reinit
3002 unconditionally, use save_target_globals_default_opts and
3003 restore_target_globals.
3005 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3007 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3008 (rs6000_expand_builtin): Handle mffs and mtfsf.
3009 (rs6000_init_builtins): Define mffs and mtfsf.
3010 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3011 (rs6000_mffs): New pattern.
3012 (rs6000_mtfsf): New pattern.
3014 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3016 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3017 Start narrowing with START. Apply candidate-use pair
3018 and check overall cost in narrowing.
3019 (iv_ca_prune): Pass new argument.
3021 2014-01-10 Jeff Law <law@redhat.com>
3024 * ree.c (combine_reaching_defs): Ensure the defining statement
3025 occurs before the extension when optimizing extensions with
3026 different source and destination hard registers.
3028 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3031 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3032 vtables into the type inheritance graph.
3034 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3036 PR rtl-optimization/59754
3037 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3038 modes in the REGNO != REGNO case.
3040 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3042 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3044 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3046 PR tree-optimization/59745
3047 * tree-predcom.c (tree_predictive_commoning_loop): Call
3048 free_affine_expand_cache if giving up because components is NULL.
3050 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3051 GC in payload of target_globals struct instead of allocating them on
3052 the heap and the larger structs separately using GC.
3053 * target-globals.h (struct target_globals): Make regs, hard_regs,
3054 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3055 of GTY((skip)) and change type to void *.
3056 (reset_target_globals): Cast loads from those fields to corresponding
3059 2014-01-10 Steve Ellcey <sellcey@mips.com>
3062 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3063 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3064 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3066 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3069 * aarch64-modes.def (CC_Zmode): New flags mode.
3070 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3071 represents an equality.
3072 (aarch64_get_condition_code): Handle CC_Zmode.
3073 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3075 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3077 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3078 extraction in good case.
3080 2014-01-10 Richard Biener <rguenther@suse.de>
3082 PR tree-optimization/59374
3083 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3084 checking after SLP discovery. Mark stmts not participating
3085 in any SLP instance properly.
3087 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3089 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3090 when handling a SET rtx.
3092 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3094 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3095 (cortex-a57): Likewise.
3096 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3098 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3100 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3101 non-iwmmxt builtins.
3103 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3107 * ipa-devirt.c record_target_from_binfo): Take as argument
3108 stack of binfos and lookup matching one for virtual inheritance.
3109 (possible_polymorphic_call_targets_1): Update.
3111 2014-01-10 Huacai Chen <chenhc@lemote.com>
3113 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3114 kernel strings for Loongson-2E/2F/3A.
3116 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3119 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3120 is_gimple_call before calling gimple_call_internal_p.
3122 2014-01-09 Steve Ellcey <sellcey@mips.com>
3124 * Makefile.in (TREE_FLOW_H): Remove.
3125 (TREE_SSA_H): Add file names from tree-flow.h.
3126 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3127 * tree.h: Remove tree-flow.h reference.
3128 * hash-table.h: Remove tree-flow.h reference.
3129 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3130 reference with tree-ssa-loop.h.
3132 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3134 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3135 default element-order behavior for -maltivec.
3136 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3137 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3138 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3139 when targeting big endian, at least for now.
3140 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3142 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3145 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3146 var satisfies use_register_for_decl, just take into account type
3147 alignment, rather than decl alignment.
3149 PR tree-optimization/59622
3150 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3151 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3152 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3153 Don't devirtualize for inplace at all. For targets.length () == 1,
3154 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3156 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3158 * config/i386/i386.md (cpu): Remove the unused btver1.
3160 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3162 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3164 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3167 * tree-core.h (struct target_globals): New forward declaration.
3168 (struct tree_target_option): Add globals field.
3169 * tree.h (TREE_TARGET_GLOBALS): Define.
3170 (prepare_target_option_nodes_for_pch): New prototype.
3171 * target-globals.h (struct target_globals): Define even if
3173 * tree.c (prepare_target_option_node_for_pch,
3174 prepare_target_option_nodes_for_pch): New functions.
3175 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3176 * config/i386/i386.c: Include target-globals.h.
3177 (ix86_set_current_function): Instead of doing target_reinit
3178 unconditionally, use save_target_globals_default_opts and
3179 restore_target_globals.
3181 2014-01-09 Richard Biener <rguenther@suse.de>
3183 PR tree-optimization/59715
3184 * tree-cfg.h (split_critical_edges): Declare.
3185 * tree-cfg.c (split_critical_edges): Export.
3186 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3188 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3190 * cfgexpand.c (expand_stack_vars): Optionally disable
3191 asan stack protection.
3192 (expand_used_vars): Likewise.
3193 (partition_stack_vars): Likewise.
3194 * asan.c (asan_emit_stack_protection): Optionally disable
3195 after return stack usage.
3196 (instrument_derefs): Optionally disable memory access instrumentation.
3197 (instrument_builtin_call): Likewise.
3198 (instrument_strlen_call): Likewise.
3199 (asan_protect_global): Optionally disable global variables protection.
3200 * doc/invoke.texi: Added doc for new options.
3201 * params.def: Added new options.
3202 * params.h: Likewise.
3204 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3206 PR rtl-optimization/59724
3207 * ifcvt.c (cond_exec_process_if_block): Don't call
3208 flow_find_head_matching_sequence with 0 longest_match.
3209 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3210 non-active insns if !stop_after.
3211 (try_head_merge_bb): Revert 2014-01-07 changes.
3213 2014-01-08 Jeff Law <law@redhat.com>
3215 * ree.c (get_sub_rtx): New function, extracted from...
3216 (merge_def_and_ext): Here.
3217 (combine_reaching_defs): Use get_sub_rtx.
3219 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3221 * cgraph.h (varpool_variable_node): Do not choke on null node.
3223 2014-01-08 Catherine Moore <clm@codesourcery.com>
3225 * config/mips/mips.md (simple_return): Attempt to use JRC
3227 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3229 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3231 PR rtl-optimization/59137
3232 * reorg.c (steal_delay_list_from_target): Call update_block for
3234 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3236 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3239 two duplicate entries.
3241 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3244 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3246 * config/mips/mips.c (mips_truncated_op_cost): New function.
3247 (mips_rtx_costs): Adjust test for BADDU.
3248 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3250 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3252 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3253 (*baddu_si): ...this new pattern.
3255 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3258 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3260 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3263 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3265 (expand_expr, expand_normal): Adjust.
3266 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3267 inner_reference_p. Use inner_reference_p to expand inner references.
3268 (store_expr): Adjust.
3269 * cfgexpand.c (expand_call_stmt): Adjust.
3271 2014-01-08 Rong Xu <xur@google.com>
3273 * gcov-io.c (gcov_var): Move from gcov-io.h.
3274 (gcov_position): Ditto.
3275 (gcov_is_error): Ditto.
3276 (gcov_rewrite): Ditto.
3277 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3278 only part to libgcc/libgcov.h.
3280 2014-01-08 Marek Polacek <polacek@redhat.com>
3283 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3285 2014-01-08 Marek Polacek <polacek@redhat.com>
3288 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3290 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3292 PR rtl-optimization/59649
3293 * stor-layout.c (get_mode_bounds): For BImode return
3294 0 and STORE_FLAG_VALUE.
3296 2014-01-08 Richard Biener <rguenther@suse.de>
3299 * gimple.h (is_gimple_builtin_call): Remove.
3300 (gimple_builtin_call_types_compatible_p): New.
3301 (gimple_call_builtin_p): New overload.
3302 * gimple.c (is_gimple_builtin_call): Remove.
3303 (validate_call): Rename to ...
3304 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3306 (validate_type): New static function.
3307 (gimple_call_builtin_p): New overload and adjust.
3308 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3309 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3310 (gimple_fold_stmt_to_constant_1): Likewise.
3311 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3313 2014-01-08 Richard Biener <rguenther@suse.de>
3316 * gimplify.c (gimplify_expr): Gimplify register-register type
3317 VIEW_CONVERT_EXPRs to separate stmts.
3319 2014-01-07 Jeff Law <law@redhat.com>
3322 * ree.c (combine_set_extension): Handle case where source
3323 and destination registers in an extension insn are different.
3324 (combine_reaching_defs): Allow source and destination registers
3325 in extension to be different under limited circumstances.
3326 (add_removable_extension): Remove restriction that the
3327 source and destination registers in the extension are the same.
3328 (find_and_remove_re): Emit a copy from the extension's
3329 destination to its source after the defining insn if
3330 the source and destination registers are different.
3333 * ifcvt.c (merge_if_block): If we are merging a block with more than
3334 one successor with a block with no successors, remove any BARRIER
3335 after the second block.
3337 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3339 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3341 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3344 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3345 for 14-bit register offsets when INT14_OK_STRICT is false.
3347 2014-01-07 Roland Stigge <stigge@antcom.de>
3348 Michael Meissner <meissner@linux.vnet.ibm.com>
3351 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3352 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3354 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3356 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3359 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3361 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3362 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3363 rtx is const0_rtx or not.
3365 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3368 * target-globals.c (save_target_globals): Remove this_fn_optab
3370 * toplev.c: Include optabs.h.
3371 (target_reinit): Temporarily restore the global options if another
3372 set of options are in force.
3374 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3376 PR rtl-optimization/58668
3377 * cfgcleanup.c (flow_find_cross_jump): Don't count
3378 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3379 to determine what is counted.
3380 (flow_find_head_matching_sequence): Use active_insn_p to determine
3382 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3384 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3385 determine what is counted.
3387 PR tree-optimization/59643
3388 * tree-predcom.c (split_data_refs_to_components): If one dr is
3389 read and one write, determine_offset fails and the write isn't
3390 in the bad component, just put the read into the bad component.
3392 2014-01-07 Mike Stump <mikestump@comcast.net>
3393 Jakub Jelinek <jakub@redhat.com>
3396 * tree-core.h (struct tree_optimization_option): Change optabs
3397 type from unsigned char * to void *.
3398 * optabs.c (init_tree_optimization_optabs): Adjust
3399 TREE_OPTIMIZATION_OPTABS initialization.
3401 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3404 * config/i386/i386.h (struct machine_function): Add
3405 no_drap_save_restore field.
3406 * config/i386/i386.c (ix86_save_reg): Use
3407 !cfun->machine->no_drap_save_restore instead of
3408 crtl->stack_realign_needed.
3409 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3410 this function clears frame_pointer_needed. Set
3411 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3412 and DRAP reg is needed.
3414 2014-01-06 Marek Polacek <polacek@redhat.com>
3417 * doc/implement-c.texi: Mention that other integer types are
3418 permitted as bit-field types in strictly conforming mode.
3420 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3422 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3425 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3427 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3429 2014-01-06 Martin Jambor <mjambor@suse.cz>
3432 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3434 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3436 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3440 * var-tracking.c (add_stores): Preserve the value of the source even if
3441 we don't record the store.
3443 2014-01-06 Terry Guo <terry.guo@arm.com>
3445 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3447 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3450 * config/darwin.c (darwin_function_section): Adjust return values to
3451 correspond to optimisation changes made in r206070.
3453 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3455 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3456 from prefetch_block tune setting.
3457 (nocona_cost): Correct size of prefetch block to 64.
3459 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3461 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3462 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3463 used to save the static chain register in the computation of the offset
3464 from which the FP registers need to be restored.
3466 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3468 PR tree-optimization/59519
3469 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3470 ICE if get_current_def (current_new_name) is already non-NULL, as long
3471 as it is a phi result of some other phi in *new_exit_bb that has
3474 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3475 or vmovdqu* for misaligned_operand.
3476 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3477 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3478 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3479 aligned_mem for AVX512F masked aligned load and store builtins and for
3482 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3484 PR tree-optimization/59651
3485 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3486 Address range for negative step should be added by TYPE_SIZE_UNIT.
3488 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3490 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3491 overlapping registers also for registers other than the stack pointer.
3493 2014-01-03 Marek Polacek <polacek@redhat.com>
3496 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3499 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3502 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3505 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3506 (push splitter): Use <P:MODE_SIZE> instead of
3507 GET_MODE_SIZE (<P:MODE>mode).
3508 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3509 (mov -1, reg peephole2): Likewise.
3510 * config/i386/sse.md (*mov<mode>_internal,
3511 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3512 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3513 *<code><mode>3, *andnot<mode>3<mask_name>,
3514 <mask_codefor><code><mode>3<mask_name>): Likewise.
3515 * config/i386/subst.md (mask_mode512bit_condition,
3516 sd_mask_mode512bit_condition): Likewise.
3518 2014-01-02 Xinliang David Li <davidxl@google.com>
3520 PR tree-optimization/59303
3521 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3522 (dump_predicates): Better output format.
3523 (pred_equal_p): New function.
3524 (is_neq_relop_p): Ditto.
3525 (is_neq_zero_form_p): Ditto.
3526 (pred_expr_equal_p): Ditto.
3527 (pred_neg_p): Ditto.
3528 (simplify_pred): Ditto.
3529 (simplify_preds_2): Ditto.
3530 (simplify_preds_3): Ditto.
3531 (simplify_preds_4): Ditto.
3532 (simplify_preds): Ditto.
3534 (push_to_worklist): Ditto.
3535 (get_pred_info_from_cmp): Ditto.
3536 (is_degenerated_phi): Ditto.
3537 (normalize_one_pred_1): Ditto.
3538 (normalize_one_pred): Ditto.
3539 (normalize_one_pred_chain): Ditto.
3540 (normalize_preds): Ditto.
3541 (normalize_cond_1): Remove function.
3542 (normalize_cond): Ditto.
3543 (is_gcond_subset_of): Ditto.
3544 (is_subset_of_any): Ditto.
3545 (is_or_set_subset_of): Ditto.
3546 (is_and_set_subset_of): Ditto.
3547 (is_norm_cond_subset_of): Ditto.
3548 (pred_chain_length_cmp): Ditto.
3549 (convert_control_dep_chain_into_preds): Type change.
3550 (find_predicates): Ditto.
3551 (find_def_preds): Ditto.
3552 (destroy_predicates_vecs): Ditto.
3553 (find_matching_predicates_in_rest_chains): Ditto.
3554 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3555 (is_pred_expr_subset): Ditto.
3556 (is_pred_chain_subset_of): Ditto.
3557 (is_included_in): Ditto.
3558 (is_superset_of): Ditto.
3560 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3562 Update copyright years.
3564 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3566 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3567 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3568 config/arc/arc.md, config/arc/arc.opt,
3569 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3570 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3571 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3572 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3573 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3574 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3575 vtable-verify.c, vtable-verify.h: Use the standard form for the
3578 2014-01-02 Tobias Burnus <burnus@net-b.de>
3580 * gcc.c (process_command): Update copyright notice dates.
3581 * gcov-dump.c: Ditto.
3583 * doc/cpp.texi: Bump @copying's copyright year.
3584 * doc/cppinternals.texi: Ditto.
3585 * doc/gcc.texi: Ditto.
3586 * doc/gccint.texi: Ditto.
3587 * doc/gcov.texi: Ditto.
3588 * doc/install.texi: Ditto.
3589 * doc/invoke.texi: Ditto.
3591 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3593 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3595 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3597 * config/i386/sse.md (*mov<mode>_internal): Guard
3598 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3600 PR rtl-optimization/59647
3601 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3602 new_rtx into UNSIGNED_FLOAT rtxes.
3604 Copyright (C) 2014 Free Software Foundation, Inc.
3606 Copying and distribution of this file, with or without modification,
3607 are permitted in any medium without royalty provided the copyright
3608 notice and this notice are preserved.