1 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2 Ilya Tocar <ilya.tocar@intel.com>
4 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
5 arguments order in builtin.
6 (_mm512_permutexvar_epi64): Ditto.
7 (_mm512_mask_permutexvar_epi64): Ditto
8 (_mm512_maskz_permutexvar_epi32): Ditto
9 (_mm512_permutexvar_epi32): Ditto
10 (_mm512_mask_permutexvar_epi32): Ditto
12 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian
16 (p8_vmrgow): Likewise.
18 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
23 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
27 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
28 into 64-bit and 32-bit moves. On 64-bit moves, add support for
29 using direct move instructions on ISA 2.07. Also adjust
30 instruction length for 64-bit.
31 (mov<mode>_64bit, TFmode/TDmode): Likewise.
32 (mov<mode>_32bit, TFmode/TDmode): Likewise.
34 2014-02-15 Alan Modra <amodra@gmail.com>
38 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
39 find_replacement on parts of insn rtl that might be reloaded.
41 2014-02-15 Richard Biener <rguenther@suse.de>
43 PR tree-optimization/60183
44 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating
46 (tree_ssa_phiprop): Calculate and free post-dominators.
48 2014-02-14 Jeff Law <law@redhat.com>
50 PR rtl-optimization/60131
51 * ree.c (get_extended_src_reg): New function.
52 (combine_reaching_defs): Use it rather than assuming location
54 (find_and_remove_re): Verify first operand of extension is
55 a REG before adding the insns to the copy list.
57 2014-02-14 Roland McGrath <mcgrathr@google.com>
59 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
60 * configure: Regenerated.
61 * config.in: Regenerated.
62 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
65 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
66 Richard Earnshaw <rearnsha@arm.com>
68 PR rtl-optimization/59535
69 * lra-constraints.c (process_alt_operands): Encourage alternative
70 when unassigned pseudo class is superset of the alternative class.
71 (inherit_reload_reg): Don't inherit when optimizing for code size.
72 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
73 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
74 modes not less than 4 for Thumb1.
76 2014-02-14 Kyle McMartin <kyle@redhat.com>
79 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
81 2014-02-14 Richard Biener <rguenther@suse.de>
83 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
84 (get_frame_arg): Drop the assert with langhook types_compatible_p.
85 Do not strip INDIRECT_REFs.
87 2014-02-14 Richard Biener <rguenther@suse.de>
90 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
91 DECL_FUNCTION_SPECIFIC_TARGET.
92 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
93 * tree-streamer-out.c (pack_ts_target_option): Remove.
94 (streamer_pack_tree_bitfields): Do not stream
96 (write_ts_function_decl_tree_pointers): Do not stream
97 DECL_FUNCTION_SPECIFIC_TARGET.
98 * tree-streamer-in.c (unpack_ts_target_option): Remove.
99 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
100 (lto_input_ts_function_decl_tree_pointers): Do not stream
101 DECL_FUNCTION_SPECIFIC_TARGET.
103 2014-02-14 Jakub Jelinek <jakub@redhat.com>
105 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
107 (get_initial_def_for_induction, vectorizable_induction): Ignore
108 debug stmts when looking for exit_phi.
109 (vectorizable_live_operation): Fix up condition.
111 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
113 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
114 nreverse() because it changes the content of original tree list.
116 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
118 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
119 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
121 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
123 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
124 GNU coding standards.
126 2014-02-13 Jakub Jelinek <jakub@redhat.com>
129 * dwarf2out.c (gen_subprogram_die): Don't call
130 add_calling_convention_attribute if subr_die is old_die.
132 2014-02-13 Sharad Singhai <singhai@google.com>
134 * doc/optinfo.texi: Fix order of nodes.
136 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
138 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
139 operands[2], not operands[3].
141 2014-02-13 Richard Biener <rguenther@suse.de>
144 * doc/install.texi (ISL): Update recommended version to 0.12.2,
145 mention the possibility of an in-tree build.
146 (CLooG): Update recommended version to 0.18.1, mention the
147 possibility of an in-tree build and clarify that the ISL
148 bundled with CLooG does not work.
150 2014-02-13 Jakub Jelinek <jakub@redhat.com>
153 * expr.c (compress_float_constant): If x is a hard register,
154 extend into a pseudo and then move to x.
156 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
158 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
159 caused by bad second argument to warning_at() with -mhotpatch and
160 nested functions (e.g. with gfortran).
162 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
164 * opts.c (option_name): Remove "enabled by default" rider.
166 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
168 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
170 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
171 Uros Bizjak <ubizjak@gmail.com>
174 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
175 * configure: Regenerated.
177 2014-02-12 Richard Biener <rguenther@suse.de>
179 * vec.c (vec_prefix::calculate_allocation): Move as
180 inline variant to vec.h.
181 (vec_prefix::calculate_allocation_1): New out-of-line version.
182 * vec.h (vec_prefix::calculate_allocation_1): Declare.
183 (vec_prefix::m_has_auto_buf): Rename to ...
184 (vec_prefix::m_using_auto_storage): ... this.
185 (vec_prefix::calculate_allocation): Inline the easy cases
186 and dispatch to calculate_allocation_1 which doesn't need the
188 (va_heap::reserve): Use gcc_checking_assert.
189 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
190 m_using_auto_storage.
191 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
193 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
194 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
195 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
197 2014-02-12 Richard Biener <rguenther@suse.de>
199 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
200 when we found a dependence.
202 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
204 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
206 (maybe_fold_stmt): ... into this new function.
207 * omp-low.c (lower_omp): Update comment.
209 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
212 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
215 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
217 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
218 identifiers in comments.
219 (cortexa53_extra_costs): Likewise.
220 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
221 (cortexa7_extra_costs): Likewise.
222 (cortexa12_extra_costs): Likewise.
223 (cortexa15_extra_costs): Likewise.
224 (v7m_extra_costs): Likewise.
226 2014-02-12 Richard Biener <rguenther@suse.de>
229 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
230 of posix_memalign being successful.
231 (lower_stmt): Restrict lowering of posix_memalign to when
232 -ftree-bit-ccp is enabled.
234 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
236 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
238 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
240 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
242 PR rtl-optimization/60116
243 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
244 other_insn once the combination has been validated.
246 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
249 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
251 * ipa-devirt.c: Include demangle.h
252 (odr_violation_reported): New static variable.
253 (add_type_duplicate): Update odr_violations.
254 (maybe_record_node): Add completep parameter; update it.
255 (record_target_from_binfo): Add COMPLETEP parameter;
257 (possible_polymorphic_call_targets_1): Likewise.
258 (struct polymorphic_call_target_d): Add nonconstruction_targets;
259 rename FINAL to COMPLETE.
260 (record_targets_from_bases): Sanity check we found the binfo;
261 fix COMPLETEP updating.
262 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
263 parameter, fix computing of COMPLETEP.
264 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
265 at LTO time do demangling.
266 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
267 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
269 (gimple_get_virt_method_for_binfo): Likewise.
270 * gimple-fold.h (gimple_get_virt_method_for_binfo,
271 gimple_get_virt_method_for_vtable): Update prototypes.
273 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
276 * genautomata.c (add_presence_absence): Fix typo with
277 {final_}presence_list.
279 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
282 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
283 for VSX/Altivec vectors that land in GPR registers.
285 2014-02-11 Richard Henderson <rth@redhat.com>
286 Jakub Jelinek <jakub@redhat.com>
289 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
290 around drhs if type conversion to lacc->type is not useless.
292 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
294 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
296 (cortex-a57.cortex-a53): Likewise.
297 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
299 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
301 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
304 2014-02-11 Renlin Li <Renlin.Li@arm.com>
306 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
307 add_options_for_arm_vfp3.
309 2014-02-11 Jeff Law <law@redhat.com>
312 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
313 object with an undesirable mode.
315 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
318 * config/i386/sol2-9.h: New file.
319 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
320 *-*-solaris2.9*): Use it.
322 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
324 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
325 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
327 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
329 * config/microblaze/microblaze.c: Extend mcpu version format
331 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
333 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
335 2014-02-10 Richard Henderson <rth@redhat.com>
338 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
339 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
340 ms-abi vs -mno-accumulate-outgoing-args.
341 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
342 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
345 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
348 * cfgexpand.c (expand_asm_operands): Attach source location to
349 ASM_INPUT rtx objects.
350 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
352 2014-02-10 Nick Clifton <nickc@redhat.com>
354 * config/mn10300/mn10300.c (popcount): New function.
355 (mn10300_expand_prologue): Include saved registers in stack usage
358 2014-02-10 Jeff Law <law@redhat.com>
361 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
362 when changing the SET_DEST of a prior insn to avoid an input reload.
364 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
366 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
367 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
368 -mcall-openbsd, or -mcall-linux.
369 (CC1_ENDIAN_BIG_SPEC): Remove.
370 (CC1_ENDIAN_LITTLE_SPEC): Remove.
371 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
372 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
373 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
374 and %cc1_endian_default.
375 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
377 2014-02-10 Richard Biener <rguenther@suse.de>
379 PR tree-optimization/60115
380 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
381 MEM_REF handling. Properly verify that the accesses are not
382 out of the objects bound.
384 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
386 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
389 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
391 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
392 proper constants and fix formatting.
393 (possible_polymorphic_call_targets): Fix formatting.
395 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
396 Ilya Tocar <ilya.tocar@intel.com>
398 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
399 (_mm512_loadu_epi32): Renamed into...
400 (_mm512_loadu_si512): This.
401 (_mm512_storeu_epi32): Renamed into...
402 (_mm512_storeu_si512): This.
403 (_mm512_maskz_ceil_ps): Removed.
404 (_mm512_maskz_ceil_pd): Ditto.
405 (_mm512_maskz_floor_ps): Ditto.
406 (_mm512_maskz_floor_pd): Ditto.
407 (_mm512_floor_round_ps): Ditto.
408 (_mm512_floor_round_pd): Ditto.
409 (_mm512_ceil_round_ps): Ditto.
410 (_mm512_ceil_round_pd): Ditto.
411 (_mm512_mask_floor_round_ps): Ditto.
412 (_mm512_mask_floor_round_pd): Ditto.
413 (_mm512_mask_ceil_round_ps): Ditto.
414 (_mm512_mask_ceil_round_pd): Ditto.
415 (_mm512_maskz_floor_round_ps): Ditto.
416 (_mm512_maskz_floor_round_pd): Ditto.
417 (_mm512_maskz_ceil_round_ps): Ditto.
418 (_mm512_maskz_ceil_round_pd): Ditto.
419 (_mm512_expand_pd): Ditto.
420 (_mm512_expand_ps): Ditto.
421 * config/i386/i386.c (ix86_builtins): Remove
422 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
424 * config/i386/predicates.md (const1256_operand): New.
425 (const_1_to_2_operand): Ditto.
426 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
427 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
428 (*avx512pf_gatherpf<mode>sf): Ditto.
429 (avx512pf_gatherpf<mode>df): Ditto.
430 (*avx512pf_gatherpf<mode>df_mask): Ditto.
431 (*avx512pf_gatherpf<mode>df): Ditto.
432 (avx512pf_scatterpf<mode>sf): Ditto.
433 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
434 (*avx512pf_scatterpf<mode>sf): Ditto.
435 (avx512pf_scatterpf<mode>df): Ditto.
436 (*avx512pf_scatterpf<mode>df_mask): Ditto.
437 (*avx512pf_scatterpf<mode>df): Ditto.
438 (avx512f_expand<mode>): Removed.
439 (<shift_insn><mode>3<mask_name>): Change predicate type.
441 2014-02-08 Jakub Jelinek <jakub@redhat.com>
443 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
444 not at the end of datarefs vector use ordered_remove to avoid
445 reordering datarefs vector.
448 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
449 mark local addressable non-static vars as GOVD_PRIVATE
450 instead of GOVD_LOCAL.
451 * omp-low.c (lower_omp_for): Move gimple_bind_vars
452 and BLOCK_VARS of gimple_bind_block to new_stmt rather
456 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
457 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
458 assume_aligned or alloc_align attributes.
459 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
460 arguments. Handle also assume_aligned and alloc_align attributes.
461 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
462 calls to functions with assume_aligned or alloc_align attributes.
463 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
465 2014-02-08 Terry Guo <terry.guo@arm.com>
467 * doc/invoke.texi: Document ARM -march=armv7e-m.
469 2014-02-08 Jakub Jelinek <jakub@redhat.com>
471 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
472 flag on __cilkrts_rethrow builtin.
475 * ipa-cp.c (determine_versionability): Fail at -O0
476 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
477 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
480 2014-02-04 Jakub Jelinek <jakub@redhat.com>
483 * tree-inline.c (copy_forbidden): Fail for
484 __attribute__((optimize (0))) functions.
486 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
488 * varpool.c: Include pointer-set.h.
489 (varpool_remove_unreferenced_decls): Variables in other partitions
490 will not be output; be however careful to not lose information
493 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
495 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
496 lookup in the vtable constructor.
498 2014-02-07 Jeff Law <law@redhat.com>
501 * config/m68k/m68k.md (ashldi_extsi): Turn into a
502 define_insn_and_split.
504 * ipa-inline.c (inline_small_functions): Fix typos.
506 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
508 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
509 (s390_can_use_return_insn): Declare.
510 * config/s390/s390.h (EPILOGUE_USES): Define.
511 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
513 (s390_chunkify_start): Handle return JUMP_LABELs.
514 (s390_early_mach): Emit a main_pool instruction on the entry edge.
515 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
516 (s390_can_use_return_insn): New functions.
517 (s390_fix_long_loop_prediction): Handle conditional returns.
518 (TARGET_SET_UP_BY_PROLOGUE): Define.
519 * config/s390/s390.md (ANY_RETURN): New code iterator.
520 (*creturn, *csimple_return, return, simple_return): New patterns.
522 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
524 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
525 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
526 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
527 REG_CFA_RESTORE list when deciding not to restore a register.
529 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
531 * config/s390/s390.c: Include tree-pass.h and context.h.
532 (s390_early_mach): New function, split out from...
533 (s390_emit_prologue): ...here.
534 (pass_data_s390_early_mach): New pass structure.
535 (pass_s390_early_mach): New class.
536 (s390_option_override): Create and register early_mach pass.
539 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
541 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
542 to match for the exit block.
544 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
546 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
547 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
548 Reject misaligned operands.
550 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
552 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
554 2014-02-07 Richard Biener <rguenther@suse.de>
557 * gimple-low.c (lower_builtin_posix_memalign): New function.
558 (lower_stmt): Call it to lower posix_memalign in a way
559 to make alignment info accessible.
561 2014-02-07 Jakub Jelinek <jakub@redhat.com>
564 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
565 __builtin_setjmp_receiver.
567 2014-02-07 Richard Biener <rguenther@suse.de>
570 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
571 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
572 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
573 Handle BUILT_IN_POSIX_MEMALIGN.
574 (find_func_clobbers): Likewise.
575 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
576 (call_may_clobber_ref_p_1): Likewise.
578 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
581 * ipa-devirt.c (record_target_from_binfo): Remove overactive
584 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
587 * lto-cgraph.c (lto_output_node): Use
588 symtab_get_symbol_partitioning_class.
589 (lto_output_varpool_node): likewise.
590 (symtab_get_symbol_partitioning_class): Move here from
592 * cgraph.h (symbol_partitioning_class): Likewise.
593 (symtab_get_symbol_partitioning_class): Declare.
595 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
597 * ggc.h (ggc_internal_cleared_alloc): New macro.
598 * vec.h (vec_safe_copy): Handle memory stats.
599 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
600 * target-globals.c (save_target_globals): Likewise.
602 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
605 * expr.c (emit_move_resolve_push): Export; be bit more selective
606 on when to clear alias set.
607 * expr.h (emit_move_resolve_push): Declare.
608 * function.h (struct function): Add tail_call_marked.
609 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
610 * config/i386/i386-protos.h (ix86_expand_push): Remove.
611 * config/i386/i386.md (TImode move expander): De not call
613 (FP push expanders): Preserve memory attributes.
614 * config/i386/sse.md (push<mode>1): Remove.
615 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
616 (ix86_expand_push): Remove.
617 * config/i386/mmx.md (push<mode>1): Remove.
619 2014-02-06 Jakub Jelinek <jakub@redhat.com>
621 PR rtl-optimization/60030
622 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
623 lopart with paradoxical subreg before shifting it up by hprec.
625 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
627 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
628 Remove extra newline at end of file.
629 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
630 (arm_issue_rate): Handle cortexa57.
631 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
632 (cortex-a57.cortex-a53): Likewise.
634 2014-02-06 Jakub Jelinek <jakub@redhat.com>
637 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
638 don't record in REG_FRAME_RELATED_EXPR registers not set in that
640 (arm_expand_prologue): Adjust all callers.
641 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
642 info, registers also at the lowest numbered registers side. Use
643 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
647 * var-tracking.c (adjust_mems): Before adding a SET to
648 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
650 2014-02-06 Alan Modra <amodra@gmail.com>
653 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
654 change SDmode to DDmode when lra_in_progress.
656 2014-02-06 Jakub Jelinek <jakub@redhat.com>
659 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
660 free_data_ref on the dr first, and before goto again also set dr
661 to the next dr. For simd_lane_access, free old datarefs[i] before
662 overwriting it. For get_vectype_for_scalar_type failure, don't
663 free_data_ref if simd_lane_access.
665 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
668 * tree.h (opts_for_fn): New inline function.
669 (opt_for_fn): Define.
670 * config/i386/i386.c (ix86_function_regparm): Use
671 opt_for_fn (decl, optimize) instead of optimize.
673 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
675 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
676 for SYMBOL_REF in large memory model.
678 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
680 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
682 (cortex-a57): Likewise.
683 (cortex-a57.cortex-a53): Likewise.
685 2014-02-06 Yury Gribov <y.gribov@samsung.com>
686 Kugan Vivekanandarajah <kuganv@linaro.org>
688 * config/arm/arm.c (arm_vector_alignment_reachable): Check
690 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
692 2014-02-06 Richard Biener <rguenther@suse.de>
694 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
695 set_loop_copy and initialize_original_copy_tables.
697 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
699 * config/aarch64/aarch64-simd.md
700 (aarch64_ashr_simddi): Change QI to SI.
702 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
703 Jakub Jelinek <jakub@redhat.com>
706 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
709 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
711 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
712 CODE_FOR_altivec_vpku[hw]um to
713 CODE_FOR_altivec_vpku[hw]um_direct.
714 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
715 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
716 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
717 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
719 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
721 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
722 generation for -maltivec=be.
723 (altivec_vsumsws): Simplify redundant test.
725 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
727 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
728 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
729 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
730 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
732 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
734 (altivec_vpks<VI_char>ss): Likewise.
735 (altivec_vpks<VI_char>us): Likewise.
736 (altivec_vpku<VI_char>us): Likewise.
737 (altivec_vpku<VI_char>um): Likewise.
738 (altivec_vpku<VI_char>um_direct): New (copy of
739 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
741 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
742 target is little endian and -maltivec=be is not specified.
743 (*altivec_vupkhs<VU_char>_direct): New (copy of
744 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
745 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
746 target is little endian and -maltivec=be is not specified.
747 (*altivec_vupkls<VU_char>_direct): New (copy of
748 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
749 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
750 little endian and -maltivec=be is not specified.
751 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
752 little endian and -maltivec=be is not specified.
754 2014-02-05 Richard Henderson <rth@redhat.com>
757 * combine-stack-adj.c: Revert r206943.
758 * sched-int.h (struct deps_desc): Add last_args_size.
759 * sched-deps.c (init_deps): Initialize it.
760 (sched_analyze_insn): Add OUTPUT dependencies between insns that
761 contain REG_ARGS_SIZE notes.
763 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
765 * lto-cgraph.c (asm_nodes_output): Make global.
766 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
767 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
768 (driver_handle_option): Handle OPT_fwpa.
770 2014-02-05 Jakub Jelinek <jakub@redhat.com>
773 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
774 a comment typo and formatting issue. If odr_hash hasn't been
775 created, return vNULL and set *completep to false.
778 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
779 bb with no successors.
781 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
784 * doc/invoke.texi (-march): Clarify documentation for ARM.
788 2014-02-05 Richard Biener <rguenther@suse.de>
790 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
791 when not vectorizing because of too many alias checks.
792 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
793 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
795 2014-02-05 Nick Clifton <nickc@redhat.com>
797 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
798 accept extended registers in any mode when compiling for the MN10300.
800 2014-02-05 Yury Gribov <y.gribov@samsung.com>
802 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
803 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
804 sanitization attributes.
805 (can_inline_edge_p): Likewise.
806 (sanitize_attrs_match_for_inline_p): New function.
808 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
810 * ipa-prop.c (detect_type_change): Shor circuit testing of
811 type changes on THIS pointer.
813 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
816 * config/pa/pa.c (legitimize_tls_address): Return original address
817 if not passed a SYMBOL_REF rtx.
818 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
820 (pa_emit_move_sequence): Simplify TLS source operands.
821 (pa_legitimate_constant_p): Reject all TLS constants.
822 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
823 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
825 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
827 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
828 groups when we know they are controlled by LTO.
829 * varasm.c (default_binds_local_p_1): If object is in other partition,
830 it will be resolved locally.
832 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
834 * config/host-linux.c (linux_gt_pch_use_address): Don't
835 use SSIZE_MAX because it is not always defined.
837 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
840 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
841 threshold for pseudo splitting.
842 (update_ebb_live_info): Process call argument hard registers and
843 hard registers from insn definition too.
844 (max_small_class_regs_num): New constant.
845 (inherit_in_ebb): Update live hard regs through EBBs. Update
846 reloads_num only for small register classes. Don't split for
849 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
852 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
855 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
857 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
860 2014-02-04 Marek Polacek <polacek@redhat.com>
862 * gdbinit.in (pel): Define.
864 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
866 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
869 2014-02-04 Richard Biener <rguenther@suse.de>
872 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
873 in function context local.
874 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
875 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
876 similar to LTO_imported_decl_ref.
878 2014-02-04 Jakub Jelinek <jakub@redhat.com>
880 PR tree-optimization/60002
881 * cgraphclones.c (build_function_decl_skip_args): Clear
884 PR tree-optimization/60023
885 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
886 false to gsi_replace.
887 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
888 has been in some EH region and vec_stmt could throw, add
889 vec_stmt into the same EH region.
890 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
891 has no lhs, ignore it.
892 * internal-fn.c (expand_MASK_LOAD): Likewise.
895 * tree-inline.c (copy_forbidden): Fail for
896 __attribute__((optimize (0))) functions.
899 * omp-low.c (simd_clone_struct_copy): If from->inbranch
900 is set, copy one less argument.
901 (expand_simd_clones): Don't subtract clone_info->inbranch
902 from simd_clone_struct_alloc argument.
904 PR rtl-optimization/57915
905 * recog.c (simplify_while_replacing): If all unary/binary/relational
906 operation arguments are constant, attempt to simplify those.
909 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
910 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
912 2014-02-04 Richard Biener <rguenther@suse.de>
914 PR tree-optimization/60012
915 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
916 TBAA disambiguation to all DDRs.
918 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
921 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
922 (LINK_SPEC): Use it for -shared, -shared-libgcc.
924 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
927 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
929 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
931 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
932 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
934 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
937 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
938 to figure out targets of polymorphic calls with known decl.
939 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
940 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
941 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
942 (get_polymorphic_call_info): ... here.
943 (get_polymorphic_call_info_from_invariant): New function.
945 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
947 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
948 lookup via vtable pointer; check for type consistency
949 and turn inconsitent facts into UNREACHABLE.
950 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
951 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
952 type inconsistent querries; return UNREACHABLE instead.
954 2014-02-03 Richard Henderson <rth@twiddle.net>
957 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
958 already processed this node.
959 (normalize_one_pred_1): Pass along mark_set.
960 (normalize_one_pred): Create and destroy a pointer_set_t.
961 (normalize_one_pred_chain): Likewise.
963 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
965 PR gcov-profile/58602
966 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
968 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
971 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
972 -fno-devirtualize; try to devirtualize by the knowledge of
973 virtual table pointer given by aggregate propagation.
974 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
975 (ipa_print_node_jump_functions): Dump also offset that
976 is relevant for polymorphic calls.
977 (determine_known_aggregate_parts): Add arg_type parameter; use it
978 instead of determining the type from pointer type.
979 (ipa_compute_jump_functions_for_edge): Update call of
980 determine_known_aggregate_parts.
981 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
982 (gimple_get_virt_method_for_binfo): ... here; simplify using
983 vtable_pointer_value_to_vtable.
984 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
985 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
986 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
987 (vtable_pointer_value_to_vtable): Break out from ...; handle also
989 (vtable_pointer_value_to_binfo): ... here.
990 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
992 2014-02-03 Teresa Johnson <tejohnson@google.com>
994 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
995 redef of outer loop index variable.
997 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1001 * doc/extend.texi (Function Attributes): Typo.
1003 2014-02-03 Cong Hou <congh@google.com>
1005 PR tree-optimization/60000
1006 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1007 if the vectorized statement is a store. A store statement can only
1008 appear at the end of pattern statements.
1010 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1012 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1013 (ix86_option_override_internal): Default long double to 64-bit for
1014 32-bit Bionic and to 128-bit for 64-bit Bionic.
1016 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1017 TARGET_LONG_DOUBLE_128 is true.
1018 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1020 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1021 (mlong-double-64): Negate -mlong-double-128.
1022 (mlong-double-128): New option.
1024 * config/i386/i386-c.c (ix86_target_macros): Define
1025 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1027 * doc/invoke.texi: Document -mlong-double-128.
1029 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1031 PR rtl-optimization/60024
1032 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1034 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1036 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1038 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1040 PR rtl-optimization/57662
1041 * sel-sched.c (code_motion_path_driver): Do not mark already not
1042 existing blocks in the visiting bitmap.
1044 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1046 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1047 on the insn being emitted.
1049 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1050 Will Deacon <will.deacon@arm.com>
1052 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1054 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1056 * config/arm/arm-tables.opt: Regenerate.
1058 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1060 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1061 for vector types other than V16QImode.
1062 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1063 define_expand, and call altivec_expand_vec_perm_le when producing
1064 code with little endian element order.
1065 (*altivec_vperm_<mode>_internal): New insn having previous
1066 behavior of altivec_vperm_<mode>.
1067 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1068 altivec_expand_vec_perm_le when producing code with little endian
1070 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1071 behavior of altivec_vperm_<mode>_uns.
1073 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1075 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1076 (altivec_vsumsws): Add handling for -maltivec=be with a little
1078 (altivec_vsumsws_direct): New.
1079 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1080 gen_altivec_vsumsws.
1082 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1084 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1085 vtable_pointer_value_to_binfo): New functions.
1086 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1087 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1089 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1091 * config/nios2/nios2.md (load_got_register): Initialize GOT
1092 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1093 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1095 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1097 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1098 preserverd by passthrough, do not propagate the type.
1100 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1102 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1103 (mips_atomic_assign_expand_fenv): New function.
1104 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1106 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1108 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1109 (__builtin_mips_set_fcsr): Likewise.
1110 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1111 MIPS_USI_FTYPE_VOID.
1112 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1113 (mips16_expand_set_fcsr): Likewise.
1114 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1115 (mips16_set_fcsr_stub): Likewise.
1116 (mips16_get_fcsr_one_only_stub): New class.
1117 (mips16_set_fcsr_one_only_stub): Likewise.
1118 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1119 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1120 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1121 (hard_float): New availability predicate.
1122 (mips_builtins): Add get_fcsr and set_fcsr.
1123 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1124 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1125 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1126 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1127 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1130 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1132 * config/mips/mips.c (mips_one_only_stub): New class.
1133 (mips_need_mips16_rdhwr_p): Replace with...
1134 (mips16_rdhwr_stub): ...this new variable.
1135 (mips16_stub_call_address): New function.
1136 (mips16_rdhwr_one_only_stub): New class.
1137 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1138 (mips_output_mips16_rdhwr): Delete.
1139 (mips_finish_stub): New function.
1140 (mips_code_end): Use it to handle rdhwr stubs.
1142 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1145 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1146 when calculating size of integer atomic types.
1148 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1150 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1152 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1154 PR tree-optimization/60003
1155 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1156 * profile.c (branch_prob): Use gimple_call_builtin_p
1157 to check for BUILT_IN_SETJMP_RECEIVER.
1158 * tree-inline.c (copy_bb): Call notice_special_calls.
1160 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1163 * lra-constraints.c (process_alt_operands): Update reload_sum only
1166 2014-01-31 Richard Henderson <rth@redhat.com>
1169 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1170 until after else_eh is processed.
1172 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1174 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1175 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1176 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1177 in smmintrin.h, remove them.
1178 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1179 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1180 * config/i386/i386.md (ROUND_SAE): Fix value.
1181 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1182 (const48_operand): New.
1183 * config/i386/subst.md (round), (round_expand): Use
1184 const_4_or_8_to_11_operand.
1185 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1187 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1189 * config/i386/constraints.md (Yk): Swap meaning with k.
1190 * config/i386/i386.md (movhi_internal): Change Yk to k.
1191 (movqi_internal): Ditto.
1192 (*k<logic><mode>): Ditto.
1195 (kandn<mode>): Ditto.
1196 (*<code>hi_1): Ditto.
1197 (*<code>qi_1): Ditto.
1198 (kxnor<mode>): Ditto.
1199 (kortestzhi): Ditto.
1200 (kortestchi): Ditto.
1202 (*one_cmplhi2_1): Ditto.
1203 (*one_cmplqi2_1): Ditto.
1204 * config/i386/sse.md (): Change k to Yk.
1205 (avx512f_load<mode>_mask): Ditto.
1206 (avx512f_blendm<mode>): Ditto.
1207 (avx512f_store<mode>_mask): Ditto.
1208 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1209 (avx512f_storedqu<mode>_mask): Ditto.
1210 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1212 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1213 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1214 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1215 (avx512f_maskcmp<mode>3): Ditto.
1216 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1217 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1218 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1219 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1220 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1221 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1222 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1223 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1224 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1225 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1226 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1227 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1228 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1229 (vec_extract_lo_<mode>_maskm): Ditto.
1230 (vec_extract_hi_<mode>_maskm): Ditto.
1231 (avx512f_vternlog<mode>_mask): Ditto.
1232 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1233 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1234 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1235 (avx512f_<code>v8div16qi2_mask): Ditto.
1236 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1237 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1238 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1239 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1240 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1241 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1242 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1243 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1244 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1245 (avx512cd_maskb_vec_dupv8di): Ditto.
1246 (avx512cd_maskw_vec_dupv16si): Ditto.
1247 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1248 (avx512f_vpermi2var<mode>3_mask): Ditto.
1249 (avx512f_vpermi2var<mode>3_mask): Ditto.
1250 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1251 (*avx512f_gathersi<mode>): Ditto.
1252 (*avx512f_gathersi<mode>_2): Ditto.
1253 (*avx512f_gatherdi<mode>): Ditto.
1254 (*avx512f_gatherdi<mode>_2): Ditto.
1255 (*avx512f_scattersi<mode>): Ditto.
1256 (*avx512f_scatterdi<mode>): Ditto.
1257 (avx512f_compress<mode>_mask): Ditto.
1258 (avx512f_compressstore<mode>_mask): Ditto.
1259 (avx512f_expand<mode>_mask): Ditto.
1260 * config/i386/subst.md (mask): Change k to Yk.
1261 (mask_scalar_merge): Ditto.
1264 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1266 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1268 2014-01-31 Richard Biener <rguenther@suse.de>
1271 * builtins.c (fold_builtin_memory_op): Make sure to not
1272 use a floating-point mode or a boolean or enumeral type for
1275 2014-01-30 DJ Delorie <dj@redhat.com>
1277 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1278 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1279 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1280 whenever main() has an epilogue.
1282 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1284 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1285 unused variable "field".
1286 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1287 (vsx_mergeh_<mode>): Likewise.
1288 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1289 (altivec_vmrghh): Likewise.
1290 (altivec_vmrghw): Likewise.
1291 (altivec_vmrglb): Likewise.
1292 (altivec_vmrglh): Likewise.
1293 (altivec_vmrglw): Likewise.
1294 (altivec_vspltb): Add missing uses.
1295 (altivec_vsplth): Likewise.
1296 (altivec_vspltw): Likewise.
1297 (altivec_vspltsf): Likewise.
1299 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1302 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1303 frame related instructions.
1305 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1307 PR rtl-optimization/59959
1308 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1309 any reload of register whose subreg is invalid.
1311 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1313 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1314 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1315 Add missing return type - void.
1317 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1319 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1320 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1321 remove element index adjustment for endian (now handled in vsx.md
1323 (altivec_expand_vec_perm_const): Use
1324 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1325 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1326 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1327 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1328 define_expand and a new define_insn *altivec_vspltb_internal;
1329 adjust for -maltivec=be on a little endian target.
1330 (altivec_vspltb_direct): New.
1331 (altivec_vsplth): Divide into a define_expand and a new
1332 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1333 little endian target.
1334 (altivec_vsplth_direct): New.
1335 (altivec_vspltw): Divide into a define_expand and a new
1336 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1337 little endian target.
1338 (altivec_vspltw_direct): New.
1339 (altivec_vspltsf): Divide into a define_expand and a new
1340 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1341 a little endian target.
1343 2014-01-30 Richard Biener <rguenther@suse.de>
1345 PR tree-optimization/59993
1346 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1347 can propagate form the earlier stmt and avoid the transform
1348 when the intermediate result is needed.
1350 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1352 * README.Portability: Fix typo.
1354 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1356 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1357 comparison_operator with ordered_comparison_operator.
1359 2014-01-30 Nick Clifton <nickc@redhat.com>
1361 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1362 Rename to mn10300_store_multiple_regs.
1363 * config/mn10300/mn10300.c: Likewise.
1364 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1365 store_multiple_regs.
1366 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1367 Call mn10300_store_multiple_regs.
1369 2014-01-30 Nick Clifton <nickc@redhat.com>
1370 DJ Delorie <dj@redhat.com>
1372 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1373 %fp 2 to keep registers after it properly word-aligned.
1374 (rl78_alloc_physical_registers_umul): Handle the case where both
1375 input operands are the same.
1377 2014-01-30 Richard Biener <rguenther@suse.de>
1379 PR tree-optimization/59903
1380 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1383 2014-01-30 Jason Merrill <jason@redhat.com>
1386 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1389 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1391 2014-01-30 Richard Biener <rguenther@suse.de>
1393 PR tree-optimization/59951
1394 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1396 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1399 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1400 SFmode to DFmode case.
1402 2014-01-29 DJ Delorie <dj@redhat.com>
1404 * config/msp430/msp430.opt (-minrt): New.
1405 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1407 (ENDFILE_SPEC): Likewise.
1409 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1411 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1412 (estimate_function_body_sizes): Use it.
1414 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1417 * dwarf2out.c (is_cxx_auto): New.
1418 (is_base_type): Use it.
1419 (gen_type_die_with_usage): Likewise.
1421 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1423 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1424 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1425 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1426 -maltivec=be with LE targets.
1427 (vsx_mergeh_<mode>): Likewise.
1428 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1429 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1430 (altivec_vmrghb): Replace with define_expand and new
1431 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1432 (altivec_vmrghb_direct): New define_insn.
1433 (altivec_vmrghh): Replace with define_expand and new
1434 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1435 (altivec_vmrghh_direct): New define_insn.
1436 (altivec_vmrghw): Replace with define_expand and new
1437 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1438 (altivec_vmrghw_direct): New define_insn.
1439 (*altivec_vmrghsf): Adjust for endianness.
1440 (altivec_vmrglb): Replace with define_expand and new
1441 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1442 (altivec_vmrglb_direct): New define_insn.
1443 (altivec_vmrglh): Replace with define_expand and new
1444 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1445 (altivec_vmrglh_direct): New define_insn.
1446 (altivec_vmrglw): Replace with define_expand and new
1447 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1448 (altivec_vmrglw_direct): New define_insn.
1449 (*altivec_vmrglsf): Adjust for endianness.
1450 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1451 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1452 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1453 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1454 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1455 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1456 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1457 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1459 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1461 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1462 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1465 2014-01-29 Richard Biener <rguenther@suse.de>
1467 PR tree-optimization/58742
1468 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1469 associate_pointerplus_align.
1470 (associate_pointerplus_diff): New function.
1471 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1472 and associate_pointerplus_diff.
1474 2014-01-29 Richard Biener <rguenther@suse.de>
1476 * lto-streamer.h (LTO_major_version): Bump to 3.
1477 (LTO_minor_version): Reset to 0.
1479 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1481 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1482 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1483 (arm_file_start): Generate correct asm header for armv7ve.
1484 * config/arm/bpabi.h: Add multilib support for armv7ve.
1485 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1486 and cortex-a15 to armv7ve.
1487 * config/arm/t-aprofile: Add multilib support for armv7ve.
1488 * doc/invoke.texi: Document -march=armv7ve.
1490 2014-01-29 Richard Biener <rguenther@suse.de>
1492 PR tree-optimization/58742
1493 * tree-ssa-forwprop.c (associate_plusminus): Return true
1494 if we changed sth, defer EH cleanup to ...
1495 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1496 (simplify_mult): New function.
1498 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1501 PR tree-optimization/59920
1502 * tree.c (build_common_builtin_nodes): Remove
1503 __builtin_setjmp_dispatcher initialization.
1504 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1505 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1506 instead of gsi_after_labels + manually skipping debug stmts.
1507 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1508 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1509 * tree-inline.c (copy_edges_for_bb): Remove
1510 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1511 argument. Ignore computed_goto_p stmts. Don't call
1512 make_abnormal_goto_edges. If a call might need abnormal edges
1513 for non-local gotos, see if it already has an edge to
1514 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1515 with true argument, don't do anything then, otherwise add
1516 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1517 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1519 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1520 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1521 (lower_stmt): Don't set data->calls_builtin_setjmp.
1522 (lower_builtin_setjmp): Adjust comment.
1523 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1524 * tree-cfg.c (found_computed_goto): Remove.
1525 (factor_computed_gotos): Remove.
1526 (make_goto_expr_edges): Return bool, true for computed gotos.
1527 Don't call make_abnormal_goto_edges.
1528 (build_gimple_cfg): Don't set found_computed_goto, don't call
1529 factor_computed_gotos.
1530 (computed_goto_p): No longer static.
1531 (make_blocks): Don't set found_computed_goto.
1532 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1533 (make_edges): If make_goto_expr_edges returns true, push bb
1534 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1535 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1536 vector. Record mapping between bbs and OpenMP regions if there
1537 are any, adjust make_gimple_omp_edges caller. Call
1538 handle_abnormal_edges.
1539 (make_abnormal_goto_edges): Remove.
1540 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1541 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1542 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1543 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1544 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1545 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1546 filling *region also set *region_idx to (*region)->entry->index.
1549 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1550 For REGs set ORIGINAL_REGNO.
1552 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1554 * doc/md.texi: Mention that a target shouldn't implement
1555 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1558 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1560 PR tree-optimization/59594
1561 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1562 a copy of the datarefs vector rather than the vector itself.
1564 2014-01-28 Jason Merrill <jason@redhat.com>
1567 * dwarf2out.c (auto_die): New static.
1568 (gen_type_die_with_usage): Handle C++1y 'auto'.
1569 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1572 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1575 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1576 (SPEC_X32): Likewise.
1577 (SPEC_64): Likewise.
1578 * config/i386/i386.c (ix86_option_override_internal): Turn off
1579 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1581 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1582 * config/i386/i386.h (TARGET_16BIT): New macro.
1583 (TARGET_16BIT_P): Likewise.
1584 * config/i386/i386.opt: Add m16.
1585 * doc/invoke.texi: Document -m16.
1587 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1589 PR preprocessor/59935
1590 * input.c (location_get_source_line): Bail out on when line number
1591 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1593 2014-01-28 Richard Biener <rguenther@suse.de>
1595 PR tree-optimization/58742
1596 * tree-ssa-forwprop.c (associate_plusminus): Handle
1597 pointer subtraction of the form (T)(P + A) - (T)P.
1599 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1601 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1604 2014-01-28 Richard Biener <rguenther@suse.de>
1607 2014-01-28 Richard Biener <rguenther@suse.de>
1609 PR rtl-optimization/45364
1610 PR rtl-optimization/59890
1611 * var-tracking.c (local_get_addr_clear_given_value): Handle
1612 already cleared slot.
1613 (val_reset): Handle not allocated local_get_addr_cache.
1614 (vt_find_locations): Use post-order on the inverted CFG.
1616 2014-01-28 Richard Biener <rguenther@suse.de>
1618 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1620 2014-01-28 Richard Biener <rguenther@suse.de>
1622 PR rtl-optimization/45364
1623 PR rtl-optimization/59890
1624 * var-tracking.c (local_get_addr_clear_given_value): Handle
1625 already cleared slot.
1626 (val_reset): Handle not allocated local_get_addr_cache.
1627 (vt_find_locations): Use post-order on the inverted CFG.
1629 2014-01-28 Alan Modra <amodra@gmail.com>
1631 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1632 * configure.ac <recursive call for build != host>: Define
1633 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1634 and LD_FOR_BUILD too.
1635 * configure: Regenerate.
1637 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1639 * config/i386/i386.c (get_builtin_code_for_version): Separate
1640 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1641 Broadwell from Haswell.
1643 2014-01-27 Steve Ellcey <sellcey@mips.com>
1645 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1646 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1647 * config/mips/mips.c (mips_option_override): Change setting
1649 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1650 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1651 Change from Mask to Var.
1653 2014-01-27 Jeff Law <law@redhat.com>
1655 * ipa-inline.c (inline_small_functions): Fix typo.
1657 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1659 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1660 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1661 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1662 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1663 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1664 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1665 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1666 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1667 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1668 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1669 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1670 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1671 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1672 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1673 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1674 (_mm512_storeu_epi64): Ditto.
1675 (_mm512_cmpge_epi32_mask): Ditto.
1676 (_mm512_cmpge_epu32_mask): Ditto.
1677 (_mm512_cmpge_epi64_mask): Ditto.
1678 (_mm512_cmpge_epu64_mask): Ditto.
1679 (_mm512_cmple_epi32_mask): Ditto.
1680 (_mm512_cmple_epu32_mask): Ditto.
1681 (_mm512_cmple_epi64_mask): Ditto.
1682 (_mm512_cmple_epu64_mask): Ditto.
1683 (_mm512_cmplt_epi32_mask): Ditto.
1684 (_mm512_cmplt_epu32_mask): Ditto.
1685 (_mm512_cmplt_epi64_mask): Ditto.
1686 (_mm512_cmplt_epu64_mask): Ditto.
1687 (_mm512_cmpneq_epi32_mask): Ditto.
1688 (_mm512_cmpneq_epu32_mask): Ditto.
1689 (_mm512_cmpneq_epi64_mask): Ditto.
1690 (_mm512_cmpneq_epu64_mask): Ditto.
1691 (_mm512_expand_pd): Ditto.
1692 (_mm512_expand_ps): Ditto.
1693 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1694 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1695 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1696 * config/i386/i386.c (ix86_builtins): Add
1697 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1698 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1699 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1700 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1701 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1702 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1703 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1704 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1705 IX86_BUILTIN_PMOVUSQW512_MEM.
1706 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1707 __builtin_ia32_pmovsqd512mem_mask,
1708 __builtin_ia32_pmovqd512mem_mask,
1709 __builtin_ia32_pmovusqw512mem_mask,
1710 __builtin_ia32_pmovsqw512mem_mask,
1711 __builtin_ia32_pmovqw512mem_mask,
1712 __builtin_ia32_pmovusdw512mem_mask,
1713 __builtin_ia32_pmovsdw512mem_mask,
1714 __builtin_ia32_pmovdw512mem_mask,
1715 __builtin_ia32_pmovqb512mem_mask,
1716 __builtin_ia32_pmovusqb512mem_mask,
1717 __builtin_ia32_pmovsqb512mem_mask,
1718 __builtin_ia32_pmovusdb512mem_mask,
1719 __builtin_ia32_pmovsdb512mem_mask,
1720 __builtin_ia32_pmovdb512mem_mask.
1721 (bdesc_args): Add __builtin_ia32_expanddf512,
1722 __builtin_ia32_expandsf512.
1723 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1724 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1725 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1726 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1727 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1728 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1729 (avx512f_<code>v8div16qi2_mask_store): This.
1730 (avx512f_expand<mode>): New.
1732 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1734 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1736 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1737 (_mm512_prefetch_i32scatter_pd): Ditto.
1738 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1739 (_mm512_prefetch_i64scatter_pd): Ditto.
1740 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1741 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1742 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1743 (_mm512_prefetch_i32scatter_ps): Ditto.
1744 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1745 (_mm512_prefetch_i64scatter_ps): Ditto.
1746 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1747 * config/i386/i386-builtin-types.def: Define
1748 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1749 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1750 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1751 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1752 IX86_BUILTIN_SCATTERPFQPD.
1753 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1754 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1755 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1756 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1757 __builtin_ia32_scatterpfqps.
1758 (ix86_expand_builtin): Expand new built-ins.
1759 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1760 fix memory access data type.
1761 (*avx512pf_gatherpf<mode>_mask): Ditto.
1762 (*avx512pf_gatherpf<mode>): Ditto.
1763 (avx512pf_scatterpf<mode>): Ditto.
1764 (*avx512pf_scatterpf<mode>_mask): Ditto.
1765 (*avx512pf_scatterpf<mode>): Ditto.
1766 (GATHER_SCATTER_SF_MEM_MODE): New.
1767 (avx512pf_gatherpf<mode>df): Ditto.
1768 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1769 (*avx512pf_scatterpf<mode>df): Ditto.
1771 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1774 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1775 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1778 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1780 * common/config/arm/arm-common.c
1781 (arm_rewrite_mcpu): Handle multiple names.
1783 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1785 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1787 * gimple-builder.h (create_gimple_tmp): Delete.
1789 2014-01-27 Christian Bruel <christian.bruel@st.com>
1791 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1794 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1796 * config/pa/pa.md (call): Generate indirect long calls to non-local
1797 functions when outputing 32-bit code.
1798 (call_value): Likewise except for special call to buggy powf function.
1800 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1801 portable runtime and PIC indirect calls.
1802 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1803 and PIC call sequences. Use ldo instead of blr to set return register
1804 in PIC call sequence.
1806 2014-01-25 Walter Lee <walt@tilera.com>
1808 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1809 avoid clobbering a live register.
1811 2014-01-25 Walter Lee <walt@tilera.com>
1813 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1814 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1815 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1816 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1818 2014-01-25 Walter Lee <walt@tilera.com>
1820 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1821 arguments on even registers.
1822 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1824 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1825 (BIGGEST_ALIGNMENT): Ditto.
1826 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1828 2014-01-25 Walter Lee <walt@tilera.com>
1830 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1831 insns before bundling.
1832 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1834 2014-01-25 Walter Lee <walt@tilera.com>
1836 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1837 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1838 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1840 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1842 * config/mips/constraints.md (kl): Delete.
1843 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1844 define expands, using...
1845 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1846 instructions for MIPS16.
1847 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1848 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1850 2014-01-25 Walter Lee <walt@tilera.com>
1852 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1856 2014-01-25 Walter Lee <walt@tilera.com>
1858 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1859 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1861 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1863 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1866 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1868 * print-rtl.c (in_call_function_usage): New var.
1869 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1870 EXPR_LIST mode as mode and not as reg note name.
1873 * cfgloopmanip.c (copy_loop_info): If
1874 loop->warned_aggressive_loop_optimizations, make sure
1875 the flag is set in target loop too.
1877 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1879 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1881 * builtins.def: Likewise.
1882 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1883 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1884 * ira.c (ira_setup_eliminable_regset): Likewise.
1885 * omp-low.c (gate_expand_omp): Likewise.
1886 (execute_lower_omp): Likewise.
1887 (diagnose_sb_0): Likewise.
1888 (gate_diagnose_omp_blocks): Likewise.
1889 (simd_clone_clauses_extract): Likewise.
1892 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1894 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1895 correction for little endian...
1896 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1899 2014-01-24 Jeff Law <law@redhat.com>
1901 PR tree-optimization/59919
1902 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1903 for non-returning calls.
1905 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1907 * common/config/aarch64/aarch64-common.c
1908 (aarch64_rewrite_mcpu): Handle multiple names.
1909 * config/aarch64/aarch64.h
1910 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1912 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1914 * input.c (add_file_to_cache_tab): Handle the case where fopen
1917 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1920 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1921 from push operand if code of push isn't PRE_DEC.
1923 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1926 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1927 -mquad-memory-atomic. Update -mquad-memory documentation to say
1928 it is only used for non-atomic loads/stores.
1930 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1931 -mquad-memory or -mquad-memory-atomic switches.
1933 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1934 -mquad-memory-atomic to ISA 2.07 support.
1936 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1937 to separate support of normal quad word memory operations (ldq, stq)
1938 from the atomic quad word memory operations.
1940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1941 support to separate non-atomic quad word operations from atomic
1942 quad word operations. Disable non-atomic quad word operations in
1943 little endian mode so that we don't have to swap words after the
1944 load and before the store.
1945 (quad_load_store_p): Add comment about atomic quad word support.
1946 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1947 options printed with -mdebug=reg.
1949 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1950 -mquad-memory-atomic as the test for whether we have quad word
1951 atomic instructions.
1952 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1953 or -mp8-vector are used, allow byte/half-word atomic operations.
1955 * config/rs6000/sync.md (load_lockedti): Insure that the address
1956 is a proper indexed or indirect address for the lqarx instruction.
1957 On little endian systems, swap the hi/lo registers after the lqarx
1959 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1960 insure the address is valid for the lqarx instruction.
1961 (store_conditionalti): Insure that the address is a proper indexed
1962 or indirect address for the stqcrx. instruction. On little endian
1963 systems, swap the hi/lo registers before doing the stqcrx.
1965 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1966 insure the address is valid for the stqcrx. instruction.
1968 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1969 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1970 type of quad memory support is available.
1972 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1975 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1976 there is a danger of looping.
1978 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1980 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1981 force flag_ira_loop_pressure if set via command line.
1983 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1985 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1986 (ashr_simd): New builtin handling DI mode.
1987 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1988 (aarch64_sshr_simddi): New match pattern.
1989 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1990 (vshrd_n_s64): Likewise.
1991 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1993 2014-01-23 Nick Clifton <nickc@redhat.com>
1995 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1996 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1997 favour of mcu specific scripts.
1998 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2001 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2002 Alex Velenko <Alex.Velenko@arm.com>
2004 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2005 (vaddv_s16): Likewise.
2006 (vaddv_s32): Likewise.
2007 (vaddv_u8): Likewise.
2008 (vaddv_u16): Likewise.
2009 (vaddv_u32): Likewise.
2010 (vaddvq_s8): Likewise.
2011 (vaddvq_s16): Likewise.
2012 (vaddvq_s32): Likewise.
2013 (vaddvq_s64): Likewise.
2014 (vaddvq_u8): Likewise.
2015 (vaddvq_u16): Likewise.
2016 (vaddvq_u32): Likewise.
2017 (vaddvq_u64): Likewise.
2018 (vaddv_f32): Likewise.
2019 (vaddvq_f32): Likewise.
2020 (vaddvq_f64): Likewise.
2021 (vmaxv_f32): Likewise.
2022 (vmaxv_s8): Likewise.
2023 (vmaxv_s16): Likewise.
2024 (vmaxv_s32): Likewise.
2025 (vmaxv_u8): Likewise.
2026 (vmaxv_u16): Likewise.
2027 (vmaxv_u32): Likewise.
2028 (vmaxvq_f32): Likewise.
2029 (vmaxvq_f64): Likewise.
2030 (vmaxvq_s8): Likewise.
2031 (vmaxvq_s16): Likewise.
2032 (vmaxvq_s32): Likewise.
2033 (vmaxvq_u8): Likewise.
2034 (vmaxvq_u16): Likewise.
2035 (vmaxvq_u32): Likewise.
2036 (vmaxnmv_f32): Likewise.
2037 (vmaxnmvq_f32): Likewise.
2038 (vmaxnmvq_f64): Likewise.
2039 (vminv_f32): Likewise.
2040 (vminv_s8): Likewise.
2041 (vminv_s16): Likewise.
2042 (vminv_s32): Likewise.
2043 (vminv_u8): Likewise.
2044 (vminv_u16): Likewise.
2045 (vminv_u32): Likewise.
2046 (vminvq_f32): Likewise.
2047 (vminvq_f64): Likewise.
2048 (vminvq_s8): Likewise.
2049 (vminvq_s16): Likewise.
2050 (vminvq_s32): Likewise.
2051 (vminvq_u8): Likewise.
2052 (vminvq_u16): Likewise.
2053 (vminvq_u32): Likewise.
2054 (vminnmv_f32): Likewise.
2055 (vminnmvq_f32): Likewise.
2056 (vminnmvq_f64): Likewise.
2058 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2060 * config/aarch64/aarch64-simd.md
2061 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2062 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2063 (*aarch64_mul3_elt<mode>): Likewise.
2064 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2065 (*aarch64_mul3_elt_to_64v2df): Likewise.
2066 (*aarch64_mla_elt<mode>): Likewise.
2067 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2068 (*aarch64_mls_elt<mode>): Likewise.
2069 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2070 (*aarch64_fma4_elt<mode>): Likewise.
2071 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2072 (*aarch64_fma4_elt_to_64v2df): Likewise.
2073 (*aarch64_fnma4_elt<mode>): Likewise.
2074 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2075 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2076 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2077 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2078 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2079 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2080 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2081 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2082 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2084 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2086 * config/aarch64/aarch64-simd.md
2087 (aarch64_be_checked_get_lane<mode>): New define_expand.
2088 * config/aarch64/aarch64-simd-builtins.def
2089 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2090 New builtin definition.
2091 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2092 Use new safe be builtin.
2094 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2096 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2098 (aarch64_be_st1<mode>): Likewise.
2099 (aarch_ld1<VALL:mode>): Define_expand modified.
2100 (aarch_st1<VALL:mode>): Likewise.
2101 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2102 (UNSPEC_ST1): Likewise.
2104 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2106 * config/microblaze/microblaze.md: Add trap insn and attribute
2108 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2110 PR preprocessor/58580
2111 * input.h (location_get_source_line): Take an additional line_size
2113 (void diagnostics_file_cache_fini): Declare new function.
2114 * input.c (struct fcache): New type.
2115 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2116 New static constants.
2117 (diagnostic_file_cache_init, total_lines_num)
2118 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2119 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2120 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2121 (get_next_line, read_next_line, goto_next_line, read_line_num):
2122 New static function definitions.
2123 (diagnostic_file_cache_fini): New function.
2124 (location_get_source_line): Take an additional output line_len
2125 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2127 * diagnostic.c (diagnostic_finish): Call
2128 diagnostic_file_cache_fini.
2129 (adjust_line): Take an additional input parameter for the length
2130 of the line, rather than calculating it with strlen.
2131 (diagnostic_show_locus): Adjust the use of
2132 location_get_source_line and adjust_line with respect to their new
2133 signature. While displaying a line now, do not stop at the first
2134 null byte. Rather, display the zero byte as a space and keep
2135 going until we reach the size of the line.
2136 * Makefile.in: Add vec.o to OBJS-libcommon
2138 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2139 Ilya Tocar <ilya.tocar@intel.com>
2141 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2142 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2143 (__builtin_ia32_kmov16): Ditto.
2144 * config/i386/i386.md (UNSPEC_KMOV): New.
2147 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2149 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2150 (_mm512_storeu_si512): Ditto.
2152 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2155 * rtl.h (get_referenced_operands): Declare.
2156 * recog.c (get_referenced_operands): New function.
2157 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2158 operands have been referenced when recording LO_SUM references.
2160 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2162 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2164 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2166 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2167 Enable for generic and recent AMD targets.
2169 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2171 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2172 ARG_SIZE note when adjustment was eliminated.
2174 2014-01-22 Jeff Law <law@redhat.com>
2176 PR tree-optimization/59597
2177 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2178 in file. Accept new argument REGISTERING and use it to modify
2179 dump output appropriately.
2180 (register_jump_thread): Corresponding changes.
2181 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2182 thread paths involving joiner blocks. Add code to dump cancelled
2183 jump threading paths.
2185 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2187 PR rtl-optimization/59477
2188 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2189 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2191 2014-01-22 Tom Tromey <tromey@redhat.com>
2193 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2195 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2197 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2199 PR rtl-optimization/59896
2200 * lra-constraints.c (process_alt_operands): Check unused note for
2201 matched operands of insn with no output reloads.
2203 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2205 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2206 (mips_move_from_gpr_cost): Likewise.
2208 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2210 PR rtl-optimization/59858
2211 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2212 ira_class_hard_regs_num.
2213 (process_alt_operands): Increase reject for dying matched operand.
2215 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2218 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2219 smaller than size, perform several stores or loads and stores
2220 at dst + count - size to store or copy all of size bytes, rather
2221 than just last modesize bytes.
2223 2014-01-20 DJ Delorie <dj@redhat.com>
2225 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2226 that CLOBBERs are REGs before propogating their values.
2228 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2231 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2232 (cgraph_inline_failed_type): New function.
2233 * cgraph.h (DEFCIFCODE): Add type.
2234 (cgraph_inline_failed_type_t): New enum.
2235 (cgraph_inline_failed_type): New prototype.
2236 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2237 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2238 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2239 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2240 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2241 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2242 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2243 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2244 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2245 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2246 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2247 OPTIMIZATION_MISMATCH.
2248 * tree-inline.c (expand_call_inline): Emit errors during
2249 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2251 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2254 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2255 mode attribute in insn output.
2257 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2259 * output.h (output_constant): Delete.
2260 * varasm.c (output_constant): Make private.
2262 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2264 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2266 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2269 * tree.h (fold_builtin_strcat): New prototype.
2270 * builtins.c (fold_builtin_strcat): No longer static. Add len
2271 argument, if non-NULL, don't call c_strlen. Optimize
2272 directly into __builtin_memcpy instead of __builtin_strcpy.
2273 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2274 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2276 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2278 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2279 for SImode_address_operand operands, having only a REG argument.
2281 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2283 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2284 loader name using mbig-endian.
2285 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2287 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2289 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2293 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2295 * config/aarch64/aarch64-protos.h
2296 (aarch64_cannot_change_mode_class_ptr): Declare.
2297 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2298 aarch64_cannot_change_mode_class_ptr): New.
2299 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2300 backend hook aarch64_cannot_change_mode_class.
2302 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2304 * common/config/aarch64/aarch64-common.c
2305 (aarch64_handle_option): Don't handle any option order logic here.
2306 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2307 selected_cpu, warn on architecture version mismatch.
2308 (aarch64_override_options): Fix parsing order for option strings.
2310 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2311 Iain Sandoe <iain@codesourcery.com>
2314 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2315 warning. Amend comment to reflect current functionality.
2317 2014-01-20 Richard Biener <rguenther@suse.de>
2320 * builtins.c (fold_builtin_strcat): Remove case better handled
2321 by tree-ssa-strlen.c.
2323 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2325 * config/aarch64/aarch64.opt
2326 (mcpu, march, mtune): Make case-insensitive.
2328 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2331 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2332 if operands[1] is a REG or ZERO_EXTEND of a REG.
2334 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2336 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2338 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2340 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2341 long non-pic millicode calls.
2343 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2345 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2347 2014-01-19 Kito Cheng <kito@0xlab.org>
2349 * builtins.c (expand_movstr): Check movstr expand done or fail.
2351 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2352 H.J. Lu <hongjiu.lu@intel.com>
2355 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2356 to DImode for zero-extended addresses.
2358 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2360 PR rtl-optimization/57763
2361 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2362 on the new indirect jump_insn and increment LABEL_NUSES (label).
2364 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2368 * config.gcc (x86_archs): New variable.
2369 (x86_64_archs): Likewise.
2370 (x86_cpus): Likewise.
2371 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2372 --with-arch/--with-cpu= options.
2373 Support --with-arch=/--with-cpu={nehalem,westmere,
2374 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2376 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2378 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2379 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2381 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2383 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2385 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2388 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2389 clear cpp_get_options (parse_in)->warn_unused_macros for
2390 ix86_target_macros_internal with cpp_define.
2392 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2394 * jump.c (delete_related_insns): Keep (use (insn))s.
2395 * reorg.c (redundant_insn): Check for barriers too.
2397 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2399 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2401 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2403 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2404 call to $$dyncall when TARGET_LONG_CALLS is true.
2406 2014-01-17 Jeff Law <law@redhat.com>
2408 * ree.c (combine_set_extension): Temporarily disable test for
2409 changing number of hard registers.
2411 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2414 * ipa-inline-analysis.c (inline_free_summary):
2415 Do not free summary of aliases.
2417 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2420 * gimplify.c (gimplify_expr): Use create_tmp_var
2421 instead of create_tmp_var_raw. If cond doesn't have
2422 integral type, don't add the IFN_ANNOTATE builtin at all.
2424 2014-01-17 Martin Jambor <mjambor@suse.cz>
2427 * ipa-cp.c (prev_edge_clone): New variable.
2428 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2429 Also resize prev_edge_clone vector.
2430 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2431 (ipcp_edge_removal_hook): New function.
2432 (ipcp_driver): Register ipcp_edge_removal_hook.
2434 2014-01-17 Andrew Pinski <apinski@cavium.com>
2435 Steve Ellcey <sellcey@mips.com>
2438 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2441 2014-01-17 Jeff Law <law@redhat.com>
2444 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2445 so that pass_ccp runs first.
2447 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2449 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2450 (ix86_adjust_cost): Use !TARGET_XXX.
2451 (do_reorder_for_imul): Likewise.
2452 (swap_top_of_ready_list): Likewise.
2453 (ix86_sched_reorder): Likewise.
2455 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2457 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2458 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2459 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2460 (intel_memset): New. Duplicate slm_memset.
2461 (intel_cost): New. Duplicate slm_cost.
2462 (m_INTEL): New macro.
2463 (processor_target_table): Add "intel".
2464 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2465 with PROCESSOR_INTEL for "intel".
2466 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2467 PROCESSOR_SILVERMONT.
2468 (ix86_issue_rate): Likewise.
2469 (ix86_adjust_cost): Likewise.
2470 (ia32_multipass_dfa_lookahead): Likewise.
2471 (swap_top_of_ready_list): Likewise.
2472 (ix86_sched_reorder): Likewise.
2473 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2474 instead of TARGET_OPT_AGU.
2475 * config/i386/i386.h (TARGET_INTEL): New.
2476 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2477 (processor_type): Add PROCESSOR_INTEL.
2478 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2479 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2481 2014-01-17 Marek Polacek <polacek@redhat.com>
2484 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2487 2014-01-17 Richard Biener <rguenther@suse.de>
2489 PR tree-optimization/46590
2490 * opts.c (default_options_table): Add entries for
2491 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2492 all enabled at -O1 but not for -Og.
2493 * common.opt (fbranch-count-reg): Remove Init(1).
2494 (fmove-loop-invariants): Likewise.
2495 (ftree-pta): Likewise.
2497 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2499 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2500 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2501 decls to at least the GCC 4.8 used alignments.
2504 * tree-nested.c (convert_nonlocal_reference_stmt,
2505 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2506 of GIMPLE_BIND stmts, adjust associated decls.
2508 2014-01-17 Richard Biener <rguenther@suse.de>
2510 PR tree-optimization/46590
2511 * vec.h (vec<>::bseach): New member function implementing
2512 binary search according to C89 bsearch.
2513 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2514 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2515 bitmap pointer again. Make accesses_in_loop a flat array.
2516 (mem_ref_obstack): New global.
2517 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2518 (mark_ref_stored): Likewise.
2519 (ref_indep_loop_p_2): Likewise.
2520 (set_ref_stored_in_loop): New helper function.
2521 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2522 (memref_free): Adjust.
2523 (record_mem_ref_loc): Simplify.
2524 (gather_mem_refs_stmt): Adjust.
2525 (sort_locs_in_loop_postorder_cmp): New function.
2526 (analyze_memory_references): Sort accesses_in_loop after
2527 loop postorder number.
2528 (find_ref_loc_in_loop_cmp): New function.
2529 (for_all_locs_in_loop): Find relevant cluster of locs in
2530 accesses_in_loop and iterate without recursion.
2531 (execute_sm): Avoid uninit warning.
2532 (struct ref_always_accessed): Simplify.
2533 (ref_always_accessed::operator ()): Likewise.
2534 (ref_always_accessed_p): Likewise.
2535 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2536 loop postorder numbers here.
2537 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2540 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2543 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2544 on decls for which assemble_alias has been called.
2546 2014-01-17 Nick Clifton <nickc@redhat.com>
2548 * config/msp430/msp430.opt: (mcpu): New option.
2549 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2550 (msp430_option_override): Parse target_cpu. If the MCU name
2551 matches a generic string, clear target_mcu.
2552 (msp430_attr): Allow numeric interrupt values up to 63.
2553 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2554 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2556 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2558 * config/msp430/msp430.md (popm): Use %J rather than %I.
2559 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2560 (addhi_cy_i): Use immediate_operand for operand 2.
2561 * doc/invoke.texi: Document -mcpu option.
2563 2014-01-17 Richard Biener <rguenther@suse.de>
2565 PR rtl-optimization/38518
2566 * df.h (df_analyze_loop): Declare.
2567 * df-core.c: Include cfgloop.h.
2568 (df_analyze_1): Split out main part of df_analyze.
2569 (df_analyze): Adjust.
2570 (loop_inverted_post_order_compute): New function.
2571 (loop_post_order_compute): Likewise.
2572 (df_analyze_loop): New function avoiding whole-function
2574 * loop-invariant.c (find_defs): Use df_analyze_loop.
2575 (find_invariants): Adjust.
2576 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2578 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2580 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2581 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2583 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2585 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2586 traversal when removing references.
2588 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2591 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2593 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2596 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2597 pushing a reload for an autoinc when we had previously reloaded an
2598 inner part of the address.
2600 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2602 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2604 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2605 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2606 when not giving up or versioning for alias only because of
2608 (vect_analyze_data_ref_dependences): Set to true.
2609 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2611 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2612 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2616 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2619 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2620 operand 0 predicate for gathers, use a new pseudo as subtarget.
2622 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2625 * lra-constraints.c (process_alt_operands): Add printing debug info.
2626 Check absence of input/output reloads for matched operands too.
2628 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2630 PR rtl-optimization/59835
2631 * ira.c (ira_init_register_move_cost): Increase cost for
2634 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2636 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2638 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2641 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2642 non-register objects. Use gen_(high/low)part more consistently.
2645 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2648 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2649 endian support, remove tests for WORDS_BIG_ENDIAN.
2650 (p8_mfvsrd_3_<mode>): Likewise.
2651 (reload_gpr_from_vsx<mode>): Likewise.
2652 (reload_gpr_from_vsxsf): Likewise.
2653 (p8_mfvsrd_4_disf): Likewise.
2655 2014-01-16 Richard Biener <rguenther@suse.de>
2657 PR rtl-optimization/46590
2658 * lcm.c (compute_antinout_edge): Use postorder iteration.
2659 (compute_laterin): Use inverted postorder iteration.
2661 2014-01-16 Nick Clifton <nickc@redhat.com>
2664 * varasm.c (output_constant): Return the number of bytes actually
2666 (output_constructor_array_range): Update the field size with the
2667 number of bytes emitted by output_constant.
2668 (output_constructor_regular_field): Likewise. Also do not
2669 complain if the total number of bytes emitted is now greater
2670 than the expected fieldpos.
2671 * output.h (output_constant): Update prototype and descriptive comment.
2673 2014-01-16 Marek Polacek <polacek@redhat.com>
2676 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2677 it is error_mark_node.
2679 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2681 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2682 VALID_AVX256_REG_OR_OI_MODE.
2684 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2686 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2687 current procedure should be profiled.
2689 2014-01-15 Andrew Pinski <apinski@cavium.com>
2691 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2692 of moving from/to the STACK_REG register class.
2694 2014-01-15 Richard Henderson <rth@redhat.com>
2697 * reginfo.c (global_regs_decl): Globalize.
2698 * rtl.h (global_regs_decl): Declare.
2699 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2700 reserved via global_regs.
2702 2014-01-15 Teresa Johnson <tejohnson@google.com>
2704 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2706 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2708 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2709 and vmulosh rather than call gen_vec_widen_smult_*.
2710 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2711 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2712 (vec_widen_smult_even_v16qi): Likewise.
2713 (vec_widen_umult_even_v8hi): Likewise.
2714 (vec_widen_smult_even_v8hi): Likewise.
2715 (vec_widen_umult_odd_v16qi): Likewise.
2716 (vec_widen_smult_odd_v16qi): Likewise.
2717 (vec_widen_umult_odd_v8hi): Likewise.
2718 (vec_widen_smult_odd_v8hi): Likewise.
2719 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2720 vmuloub rather than call gen_vec_widen_umult_*.
2721 (vec_widen_umult_lo_v16qi): Likewise.
2722 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2723 vmulosb rather than call gen_vec_widen_smult_*.
2724 (vec_widen_smult_lo_v16qi): Likewise.
2725 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2726 rather than call gen_vec_widen_umult_*.
2727 (vec_widen_umult_lo_v8hi): Likewise.
2728 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2729 rather than call gen_vec_widen_smult_*.
2730 (vec_widen_smult_lo_v8hi): Likewise.
2732 2014-01-15 Jeff Law <law@redhat.com>
2734 PR tree-optimization/59747
2735 * ree.c (find_and_remove_re): Properly handle case where a second
2736 eliminated extension requires widening a copy created for elimination
2737 of a prior extension.
2738 (combine_set_extension): Ensure that the number of hard regs needed
2739 for a destination register does not change when we widen it.
2741 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2743 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2744 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2745 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2746 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2747 (avr-*-rtems*): Likewise.
2748 (bfin*-rtems*): Likewise.
2749 (moxie-*-rtems*): Likewise.
2750 (h8300-*-rtems*): Likewise.
2751 (i[34567]86-*-rtems*): Likewise.
2752 (lm32-*-rtems*): Likewise.
2753 (m32r-*-rtems*): Likewise.
2754 (m68k-*-rtems*): Likewise.
2755 (microblaze*-*-rtems*): Likewise.
2756 (mips*-*-rtems*): Likewise.
2757 (powerpc-*-rtems*): Likewise.
2758 (sh-*-rtems*): Likewise.
2759 (sparc-*-rtems*): Likewise.
2760 (sparc64-*-rtems*): Likewise.
2761 (v850-*-rtems*): Likewise.
2762 (m32c-*-rtems*): Likewise.
2764 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2766 PR rtl-optimization/59511
2767 * ira.c (ira_init_register_move_cost): Use memory costs for some
2768 cases of register move cost calculations.
2769 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2770 instead of BB frequency.
2771 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2772 * lra-assigns.c (find_hard_regno_for): Ditto.
2774 2014-01-15 Richard Biener <rguenther@suse.de>
2776 PR tree-optimization/59822
2777 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2778 (vectorizable_load): Use it to hoist defs of uses of invariant
2779 loads out of the loop.
2781 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2782 Kugan Vivekanandarajah <kuganv@linaro.org>
2785 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2788 2014-01-15 Richard Biener <rguenther@suse.de>
2790 PR rtl-optimization/59802
2791 * lcm.c (compute_available): Use inverted postorder to seed
2792 the initial worklist.
2794 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2797 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2798 ADDR_REGS for invalid symrefs in non-PIC code.
2800 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2803 * builtins.c (determine_block_size): Initialize *probable_max_size
2804 even if len_rtx is CONST_INT.
2806 2014-01-14 Andrew Pinski <apinski@cavium.com>
2808 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2809 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2810 (cortexa53_tunings): Likewise.
2811 (aarch64_sched_issue_rate): New function.
2812 (TARGET_SCHED_ISSUE_RATE): Define.
2814 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2816 * ira-costs.c (find_costs_and_classes): Add missed
2817 ira_init_register_move_cost_if_necessary.
2819 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2822 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2824 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2827 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2828 to indicate if type is used for function return value. Warn ABI
2829 change if the vector mode isn't available for function return value.
2830 (ix86_function_arg_advance): Pass false to type_natural_mode.
2831 (ix86_function_arg): Likewise.
2832 (ix86_gimplify_va_arg): Likewise.
2833 (function_arg_32): Don't warn ABI change.
2834 (ix86_function_value): Pass true to type_natural_mode.
2835 (ix86_return_in_memory): Likewise.
2836 (ix86_struct_value_rtx): Removed.
2837 (TARGET_STRUCT_VALUE_RTX): Likewise.
2839 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2841 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2842 converting a conditional jump into a conditional return.
2844 2014-01-14 Richard Biener <rguenther@suse.de>
2846 PR tree-optimization/58921
2847 PR tree-optimization/59006
2848 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2849 hoisting invariant stmts.
2850 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2851 invariant loads on the preheader edge if possible.
2853 2014-01-14 Joey Ye <joey.ye@arm.com>
2855 * doc/plugin.texi (Building GCC plugins): Update to C++.
2857 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2859 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2860 (_mm_rcp28_round_ss): Ditto.
2861 (_mm_rsqrt28_round_sd): Ditto.
2862 (_mm_rsqrt28_round_ss): Ditto.
2863 (_mm_rcp28_sd): Ditto.
2864 (_mm_rcp28_ss): Ditto.
2865 (_mm_rsqrt28_sd): Ditto.
2866 (_mm_rsqrt28_ss): Ditto.
2867 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2868 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2869 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2870 (IX86_BUILTIN_RCP28SD): Ditto.
2871 (IX86_BUILTIN_RCP28SS): Ditto.
2872 (IX86_BUILTIN_RSQRT28SD): Ditto.
2873 (IX86_BUILTIN_RSQRT28SS): Ditto.
2874 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2875 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2876 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2877 (ix86_expand_special_args_builtin): Expand new FTYPE.
2878 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2879 (srcp14<mode>): Make insn unary.
2880 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2881 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2882 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2883 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2884 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2885 Fix rounding: make it SAE only.
2886 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2888 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2890 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2891 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2892 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2893 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2894 (round_saeonly_mask_scalar_operand4): Ditto.
2895 (round_saeonly_mask_scalar_op3): Ditto.
2896 (round_saeonly_mask_scalar_op4): Ditto.
2898 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2900 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2901 Implement -maltivec=be for vec_insert and vec_extract.
2903 2014-01-10 DJ Delorie <dj@redhat.com>
2905 * config/msp430/msp430.md (call_internal): Don't allow memory
2906 references with SP as the base register.
2907 (call_value_internal): Likewise.
2908 * config/msp430/constraints.md (Yc): New. For memory references
2909 that don't use SP as a base register.
2911 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2912 "an integer without a # prefix"
2913 * config/msp430/msp430.md (epilogue_helper): Use it.
2915 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2918 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2919 AVX512F gather builtins.
2920 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2921 on gather decls with INTEGER_TYPE masktype.
2922 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2923 directly into the builtin rather than hoisting it before loop.
2925 PR tree-optimization/59387
2926 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2927 (scev_const_prop): If folded_casts and type has undefined overflow,
2928 use force_gimple_operand instead of force_gimple_operand_gsi and
2929 for each added stmt if it is assign with
2930 arith_code_with_undefined_signed_overflow, call
2931 rewrite_to_defined_overflow.
2932 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2933 gimple-fold.h instead.
2934 (arith_code_with_undefined_signed_overflow,
2935 rewrite_to_defined_overflow): Moved to ...
2936 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2937 rewrite_to_defined_overflow): ... here. No longer static.
2938 Include gimplify-me.h.
2939 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2940 rewrite_to_defined_overflow): New prototypes.
2942 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2944 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2946 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2948 * builtins.c (get_object_alignment_2): Minor tweak.
2949 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2951 2014-01-13 Christian Bruel <christian.bruel@st.com>
2953 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2954 optimized non constant lengths.
2956 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2959 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2960 load as __atomic_load_N if possible.
2962 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2964 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2966 (rs6000_expand_builtin): Adjust call.
2968 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2971 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2972 * config/rs6000/rs6000.c: Include target-globals.h.
2973 (rs6000_set_current_function): Instead of doing target_reinit
2974 unconditionally, use save_target_globals_default_opts and
2975 restore_target_globals.
2977 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2979 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2980 (rs6000_expand_builtin): Handle mffs and mtfsf.
2981 (rs6000_init_builtins): Define mffs and mtfsf.
2982 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2983 (rs6000_mffs): New pattern.
2984 (rs6000_mtfsf): New pattern.
2986 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2988 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2989 Start narrowing with START. Apply candidate-use pair
2990 and check overall cost in narrowing.
2991 (iv_ca_prune): Pass new argument.
2993 2014-01-10 Jeff Law <law@redhat.com>
2996 * ree.c (combine_reaching_defs): Ensure the defining statement
2997 occurs before the extension when optimizing extensions with
2998 different source and destination hard registers.
3000 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3003 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3004 vtables into the type inheritance graph.
3006 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3008 PR rtl-optimization/59754
3009 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3010 modes in the REGNO != REGNO case.
3012 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3014 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3016 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3018 PR tree-optimization/59745
3019 * tree-predcom.c (tree_predictive_commoning_loop): Call
3020 free_affine_expand_cache if giving up because components is NULL.
3022 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3023 GC in payload of target_globals struct instead of allocating them on
3024 the heap and the larger structs separately using GC.
3025 * target-globals.h (struct target_globals): Make regs, hard_regs,
3026 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3027 of GTY((skip)) and change type to void *.
3028 (reset_target_globals): Cast loads from those fields to corresponding
3031 2014-01-10 Steve Ellcey <sellcey@mips.com>
3034 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3035 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3036 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3038 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3041 * aarch64-modes.def (CC_Zmode): New flags mode.
3042 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3043 represents an equality.
3044 (aarch64_get_condition_code): Handle CC_Zmode.
3045 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3047 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3049 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3050 extraction in good case.
3052 2014-01-10 Richard Biener <rguenther@suse.de>
3054 PR tree-optimization/59374
3055 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3056 checking after SLP discovery. Mark stmts not participating
3057 in any SLP instance properly.
3059 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3061 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3062 when handling a SET rtx.
3064 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3066 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3067 (cortex-a57): Likewise.
3068 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3070 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3072 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3073 non-iwmmxt builtins.
3075 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3079 * ipa-devirt.c record_target_from_binfo): Take as argument
3080 stack of binfos and lookup matching one for virtual inheritance.
3081 (possible_polymorphic_call_targets_1): Update.
3083 2014-01-10 Huacai Chen <chenhc@lemote.com>
3085 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3086 kernel strings for Loongson-2E/2F/3A.
3088 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3091 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3092 is_gimple_call before calling gimple_call_internal_p.
3094 2014-01-09 Steve Ellcey <sellcey@mips.com>
3096 * Makefile.in (TREE_FLOW_H): Remove.
3097 (TREE_SSA_H): Add file names from tree-flow.h.
3098 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3099 * tree.h: Remove tree-flow.h reference.
3100 * hash-table.h: Remove tree-flow.h reference.
3101 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3102 reference with tree-ssa-loop.h.
3104 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3106 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3107 default element-order behavior for -maltivec.
3108 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3109 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3110 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3111 when targeting big endian, at least for now.
3112 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3114 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3117 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3118 var satisfies use_register_for_decl, just take into account type
3119 alignment, rather than decl alignment.
3121 PR tree-optimization/59622
3122 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3123 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3124 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3125 Don't devirtualize for inplace at all. For targets.length () == 1,
3126 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3128 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3130 * config/i386/i386.md (cpu): Remove the unused btver1.
3132 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3134 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3136 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3139 * tree-core.h (struct target_globals): New forward declaration.
3140 (struct tree_target_option): Add globals field.
3141 * tree.h (TREE_TARGET_GLOBALS): Define.
3142 (prepare_target_option_nodes_for_pch): New prototype.
3143 * target-globals.h (struct target_globals): Define even if
3145 * tree.c (prepare_target_option_node_for_pch,
3146 prepare_target_option_nodes_for_pch): New functions.
3147 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3148 * config/i386/i386.c: Include target-globals.h.
3149 (ix86_set_current_function): Instead of doing target_reinit
3150 unconditionally, use save_target_globals_default_opts and
3151 restore_target_globals.
3153 2014-01-09 Richard Biener <rguenther@suse.de>
3155 PR tree-optimization/59715
3156 * tree-cfg.h (split_critical_edges): Declare.
3157 * tree-cfg.c (split_critical_edges): Export.
3158 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3160 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3162 * cfgexpand.c (expand_stack_vars): Optionally disable
3163 asan stack protection.
3164 (expand_used_vars): Likewise.
3165 (partition_stack_vars): Likewise.
3166 * asan.c (asan_emit_stack_protection): Optionally disable
3167 after return stack usage.
3168 (instrument_derefs): Optionally disable memory access instrumentation.
3169 (instrument_builtin_call): Likewise.
3170 (instrument_strlen_call): Likewise.
3171 (asan_protect_global): Optionally disable global variables protection.
3172 * doc/invoke.texi: Added doc for new options.
3173 * params.def: Added new options.
3174 * params.h: Likewise.
3176 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3178 PR rtl-optimization/59724
3179 * ifcvt.c (cond_exec_process_if_block): Don't call
3180 flow_find_head_matching_sequence with 0 longest_match.
3181 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3182 non-active insns if !stop_after.
3183 (try_head_merge_bb): Revert 2014-01-07 changes.
3185 2014-01-08 Jeff Law <law@redhat.com>
3187 * ree.c (get_sub_rtx): New function, extracted from...
3188 (merge_def_and_ext): Here.
3189 (combine_reaching_defs): Use get_sub_rtx.
3191 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3193 * cgraph.h (varpool_variable_node): Do not choke on null node.
3195 2014-01-08 Catherine Moore <clm@codesourcery.com>
3197 * config/mips/mips.md (simple_return): Attempt to use JRC
3199 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3201 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3203 PR rtl-optimization/59137
3204 * reorg.c (steal_delay_list_from_target): Call update_block for
3206 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3208 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3210 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3211 two duplicate entries.
3213 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3216 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3218 * config/mips/mips.c (mips_truncated_op_cost): New function.
3219 (mips_rtx_costs): Adjust test for BADDU.
3220 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3222 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3224 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3225 (*baddu_si): ...this new pattern.
3227 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3230 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3232 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3235 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3237 (expand_expr, expand_normal): Adjust.
3238 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3239 inner_reference_p. Use inner_reference_p to expand inner references.
3240 (store_expr): Adjust.
3241 * cfgexpand.c (expand_call_stmt): Adjust.
3243 2014-01-08 Rong Xu <xur@google.com>
3245 * gcov-io.c (gcov_var): Move from gcov-io.h.
3246 (gcov_position): Ditto.
3247 (gcov_is_error): Ditto.
3248 (gcov_rewrite): Ditto.
3249 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3250 only part to libgcc/libgcov.h.
3252 2014-01-08 Marek Polacek <polacek@redhat.com>
3255 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3257 2014-01-08 Marek Polacek <polacek@redhat.com>
3260 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3262 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3264 PR rtl-optimization/59649
3265 * stor-layout.c (get_mode_bounds): For BImode return
3266 0 and STORE_FLAG_VALUE.
3268 2014-01-08 Richard Biener <rguenther@suse.de>
3271 * gimple.h (is_gimple_builtin_call): Remove.
3272 (gimple_builtin_call_types_compatible_p): New.
3273 (gimple_call_builtin_p): New overload.
3274 * gimple.c (is_gimple_builtin_call): Remove.
3275 (validate_call): Rename to ...
3276 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3278 (validate_type): New static function.
3279 (gimple_call_builtin_p): New overload and adjust.
3280 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3281 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3282 (gimple_fold_stmt_to_constant_1): Likewise.
3283 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3285 2014-01-08 Richard Biener <rguenther@suse.de>
3288 * gimplify.c (gimplify_expr): Gimplify register-register type
3289 VIEW_CONVERT_EXPRs to separate stmts.
3291 2014-01-07 Jeff Law <law@redhat.com>
3294 * ree.c (combine_set_extension): Handle case where source
3295 and destination registers in an extension insn are different.
3296 (combine_reaching_defs): Allow source and destination registers
3297 in extension to be different under limited circumstances.
3298 (add_removable_extension): Remove restriction that the
3299 source and destination registers in the extension are the same.
3300 (find_and_remove_re): Emit a copy from the extension's
3301 destination to its source after the defining insn if
3302 the source and destination registers are different.
3305 * ifcvt.c (merge_if_block): If we are merging a block with more than
3306 one successor with a block with no successors, remove any BARRIER
3307 after the second block.
3309 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3311 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3313 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3316 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3317 for 14-bit register offsets when INT14_OK_STRICT is false.
3319 2014-01-07 Roland Stigge <stigge@antcom.de>
3320 Michael Meissner <meissner@linux.vnet.ibm.com>
3323 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3324 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3326 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3328 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3331 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3333 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3334 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3335 rtx is const0_rtx or not.
3337 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3340 * target-globals.c (save_target_globals): Remove this_fn_optab
3342 * toplev.c: Include optabs.h.
3343 (target_reinit): Temporarily restore the global options if another
3344 set of options are in force.
3346 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3348 PR rtl-optimization/58668
3349 * cfgcleanup.c (flow_find_cross_jump): Don't count
3350 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3351 to determine what is counted.
3352 (flow_find_head_matching_sequence): Use active_insn_p to determine
3354 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3356 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3357 determine what is counted.
3359 PR tree-optimization/59643
3360 * tree-predcom.c (split_data_refs_to_components): If one dr is
3361 read and one write, determine_offset fails and the write isn't
3362 in the bad component, just put the read into the bad component.
3364 2014-01-07 Mike Stump <mikestump@comcast.net>
3365 Jakub Jelinek <jakub@redhat.com>
3368 * tree-core.h (struct tree_optimization_option): Change optabs
3369 type from unsigned char * to void *.
3370 * optabs.c (init_tree_optimization_optabs): Adjust
3371 TREE_OPTIMIZATION_OPTABS initialization.
3373 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3376 * config/i386/i386.h (struct machine_function): Add
3377 no_drap_save_restore field.
3378 * config/i386/i386.c (ix86_save_reg): Use
3379 !cfun->machine->no_drap_save_restore instead of
3380 crtl->stack_realign_needed.
3381 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3382 this function clears frame_pointer_needed. Set
3383 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3384 and DRAP reg is needed.
3386 2014-01-06 Marek Polacek <polacek@redhat.com>
3389 * doc/implement-c.texi: Mention that other integer types are
3390 permitted as bit-field types in strictly conforming mode.
3392 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3394 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3397 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3399 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3401 2014-01-06 Martin Jambor <mjambor@suse.cz>
3404 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3406 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3408 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3412 * var-tracking.c (add_stores): Preserve the value of the source even if
3413 we don't record the store.
3415 2014-01-06 Terry Guo <terry.guo@arm.com>
3417 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3419 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3422 * config/darwin.c (darwin_function_section): Adjust return values to
3423 correspond to optimisation changes made in r206070.
3425 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3427 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3428 from prefetch_block tune setting.
3429 (nocona_cost): Correct size of prefetch block to 64.
3431 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3433 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3434 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3435 used to save the static chain register in the computation of the offset
3436 from which the FP registers need to be restored.
3438 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3440 PR tree-optimization/59519
3441 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3442 ICE if get_current_def (current_new_name) is already non-NULL, as long
3443 as it is a phi result of some other phi in *new_exit_bb that has
3446 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3447 or vmovdqu* for misaligned_operand.
3448 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3449 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3450 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3451 aligned_mem for AVX512F masked aligned load and store builtins and for
3454 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3456 PR tree-optimization/59651
3457 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3458 Address range for negative step should be added by TYPE_SIZE_UNIT.
3460 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3462 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3463 overlapping registers also for registers other than the stack pointer.
3465 2014-01-03 Marek Polacek <polacek@redhat.com>
3468 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3471 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3474 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3477 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3478 (push splitter): Use <P:MODE_SIZE> instead of
3479 GET_MODE_SIZE (<P:MODE>mode).
3480 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3481 (mov -1, reg peephole2): Likewise.
3482 * config/i386/sse.md (*mov<mode>_internal,
3483 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3484 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3485 *<code><mode>3, *andnot<mode>3<mask_name>,
3486 <mask_codefor><code><mode>3<mask_name>): Likewise.
3487 * config/i386/subst.md (mask_mode512bit_condition,
3488 sd_mask_mode512bit_condition): Likewise.
3490 2014-01-02 Xinliang David Li <davidxl@google.com>
3492 PR tree-optimization/59303
3493 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3494 (dump_predicates): Better output format.
3495 (pred_equal_p): New function.
3496 (is_neq_relop_p): Ditto.
3497 (is_neq_zero_form_p): Ditto.
3498 (pred_expr_equal_p): Ditto.
3499 (pred_neg_p): Ditto.
3500 (simplify_pred): Ditto.
3501 (simplify_preds_2): Ditto.
3502 (simplify_preds_3): Ditto.
3503 (simplify_preds_4): Ditto.
3504 (simplify_preds): Ditto.
3506 (push_to_worklist): Ditto.
3507 (get_pred_info_from_cmp): Ditto.
3508 (is_degenerated_phi): Ditto.
3509 (normalize_one_pred_1): Ditto.
3510 (normalize_one_pred): Ditto.
3511 (normalize_one_pred_chain): Ditto.
3512 (normalize_preds): Ditto.
3513 (normalize_cond_1): Remove function.
3514 (normalize_cond): Ditto.
3515 (is_gcond_subset_of): Ditto.
3516 (is_subset_of_any): Ditto.
3517 (is_or_set_subset_of): Ditto.
3518 (is_and_set_subset_of): Ditto.
3519 (is_norm_cond_subset_of): Ditto.
3520 (pred_chain_length_cmp): Ditto.
3521 (convert_control_dep_chain_into_preds): Type change.
3522 (find_predicates): Ditto.
3523 (find_def_preds): Ditto.
3524 (destroy_predicates_vecs): Ditto.
3525 (find_matching_predicates_in_rest_chains): Ditto.
3526 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3527 (is_pred_expr_subset): Ditto.
3528 (is_pred_chain_subset_of): Ditto.
3529 (is_included_in): Ditto.
3530 (is_superset_of): Ditto.
3532 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3534 Update copyright years.
3536 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3538 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3539 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3540 config/arc/arc.md, config/arc/arc.opt,
3541 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3542 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3543 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3544 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3545 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3546 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3547 vtable-verify.c, vtable-verify.h: Use the standard form for the
3550 2014-01-02 Tobias Burnus <burnus@net-b.de>
3552 * gcc.c (process_command): Update copyright notice dates.
3553 * gcov-dump.c: Ditto.
3555 * doc/cpp.texi: Bump @copying's copyright year.
3556 * doc/cppinternals.texi: Ditto.
3557 * doc/gcc.texi: Ditto.
3558 * doc/gccint.texi: Ditto.
3559 * doc/gcov.texi: Ditto.
3560 * doc/install.texi: Ditto.
3561 * doc/invoke.texi: Ditto.
3563 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3565 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3567 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3569 * config/i386/sse.md (*mov<mode>_internal): Guard
3570 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3572 PR rtl-optimization/59647
3573 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3574 new_rtx into UNSIGNED_FLOAT rtxes.
3576 Copyright (C) 2014 Free Software Foundation, Inc.
3578 Copying and distribution of this file, with or without modification,
3579 are permitted in any medium without royalty provided the copyright
3580 notice and this notice are preserved.