1 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
6 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
8 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
9 identifiers in comments.
10 (cortexa53_extra_costs): Likewise.
11 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in
13 (cortexa7_extra_costs): Likewise.
14 (cortexa12_extra_costs): Likewise.
15 (cortexa15_extra_costs): Likewise.
16 (v7m_extra_costs): Likewise.
18 2014-02-12 Richard Biener <rguenther@suse.de>
21 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
22 of posix_memalign being successful.
23 (lower_stmt): Restrict lowering of posix_memalign to when
24 -ftree-bit-ccp is enabled.
26 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
28 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
30 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
32 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
34 PR rtl-optimization/60116
35 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
36 other_insn once the combination has been validated.
38 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
41 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
43 * ipa-devirt.c: Include demangle.h
44 (odr_violation_reported): New static variable.
45 (add_type_duplicate): Update odr_violations.
46 (maybe_record_node): Add completep parameter; update it.
47 (record_target_from_binfo): Add COMPLETEP parameter;
49 (possible_polymorphic_call_targets_1): Likewise.
50 (struct polymorphic_call_target_d): Add nonconstruction_targets;
51 rename FINAL to COMPLETE.
52 (record_targets_from_bases): Sanity check we found the binfo;
53 fix COMPLETEP updating.
54 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
55 parameter, fix computing of COMPLETEP.
56 (dump_possible_polymorphic_call_targets): Imrove readability of dump; at
57 LTO time do demangling.
58 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
59 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
61 (gimple_get_virt_method_for_binfo): Likewise.
62 * gimple-fold.h (gimple_get_virt_method_for_binfo,
63 gimple_get_virt_method_for_vtable): Update prototypes.
65 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
68 * genautomata.c (add_presence_absence): Fix typo with
69 {final_}presence_list.
71 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
74 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
75 for VSX/Altivec vectors that land in GPR registers.
77 2014-02-11 Richard Henderson <rth@redhat.com>
78 Jakub Jelinek <jakub@redhat.com>
81 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
82 around drhs if type conversion to lacc->type is not useless.
84 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
86 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
88 (cortex-a57.cortex-a53): Likewise.
89 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
91 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
93 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
96 2014-02-11 Renlin Li <Renlin.Li@arm.com>
98 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
99 add_options_for_arm_vfp3.
101 2014-02-11 Jeff Law <law@redhat.com>
104 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
105 object with an undesirable mode.
107 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
110 * config/i386/sol2-9.h: New file.
111 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
112 *-*-solaris2.9*): Use it.
114 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
116 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
117 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
119 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
121 * config/microblaze/microblaze.c: Extend mcpu version format
123 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
125 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
127 2014-02-10 Richard Henderson <rth@redhat.com>
130 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
131 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
132 ms-abi vs -mno-accumulate-outgoing-args.
133 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
134 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
137 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
140 * cfgexpand.c (expand_asm_operands): Attach source location to
141 ASM_INPUT rtx objects.
142 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
144 2014-02-10 Nick Clifton <nickc@redhat.com>
146 * config/mn10300/mn10300.c (popcount): New function.
147 (mn10300_expand_prologue): Include saved registers in stack usage
150 2014-02-10 Jeff Law <law@redhat.com>
153 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
154 when changing the SET_DEST of a prior insn to avoid an input
157 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
159 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
160 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
161 -mcall-openbsd, or -mcall-linux.
162 (CC1_ENDIAN_BIG_SPEC): Remove.
163 (CC1_ENDIAN_LITTLE_SPEC): Remove.
164 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
165 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
166 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
167 and %cc1_endian_default.
168 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
170 2014-02-10 Richard Biener <rguenther@suse.de>
172 PR tree-optimization/60115
173 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
174 MEM_REF handling. Properly verify that the accesses are not
175 out of the objects bound.
177 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
179 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
182 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
184 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
185 proper constants and fix formatting.
186 (possible_polymorphic_call_targets): Fix formatting.
188 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
189 Ilya Tocar <ilya.tocar@intel.com>
191 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
192 (_mm512_loadu_epi32): Renamed into...
193 (_mm512_loadu_si512): This.
194 (_mm512_storeu_epi32): Renamed into...
195 (_mm512_storeu_si512): This.
196 (_mm512_maskz_ceil_ps): Removed.
197 (_mm512_maskz_ceil_pd): Ditto.
198 (_mm512_maskz_floor_ps): Ditto.
199 (_mm512_maskz_floor_pd): Ditto.
200 (_mm512_floor_round_ps): Ditto.
201 (_mm512_floor_round_pd): Ditto.
202 (_mm512_ceil_round_ps): Ditto.
203 (_mm512_ceil_round_pd): Ditto.
204 (_mm512_mask_floor_round_ps): Ditto.
205 (_mm512_mask_floor_round_pd): Ditto.
206 (_mm512_mask_ceil_round_ps): Ditto.
207 (_mm512_mask_ceil_round_pd): Ditto.
208 (_mm512_maskz_floor_round_ps): Ditto.
209 (_mm512_maskz_floor_round_pd): Ditto.
210 (_mm512_maskz_ceil_round_ps): Ditto.
211 (_mm512_maskz_ceil_round_pd): Ditto.
212 (_mm512_expand_pd): Ditto.
213 (_mm512_expand_ps): Ditto.
214 * config/i386/i386.c (ix86_builtins): Remove
215 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
217 * config/i386/predicates.md (const1256_operand): New.
218 (const_1_to_2_operand): Ditto.
219 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
220 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
221 (*avx512pf_gatherpf<mode>sf): Ditto.
222 (avx512pf_gatherpf<mode>df): Ditto.
223 (*avx512pf_gatherpf<mode>df_mask): Ditto.
224 (*avx512pf_gatherpf<mode>df): Ditto.
225 (avx512pf_scatterpf<mode>sf): Ditto.
226 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
227 (*avx512pf_scatterpf<mode>sf): Ditto.
228 (avx512pf_scatterpf<mode>df): Ditto.
229 (*avx512pf_scatterpf<mode>df_mask): Ditto.
230 (*avx512pf_scatterpf<mode>df): Ditto.
231 (avx512f_expand<mode>): Removed.
232 (<shift_insn><mode>3<mask_name>): Change predicate type.
234 2014-02-08 Jakub Jelinek <jakub@redhat.com>
236 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
237 not at the end of datarefs vector use ordered_remove to avoid
238 reordering datarefs vector.
241 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
242 mark local addressable non-static vars as GOVD_PRIVATE
243 instead of GOVD_LOCAL.
244 * omp-low.c (lower_omp_for): Move gimple_bind_vars
245 and BLOCK_VARS of gimple_bind_block to new_stmt rather
249 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
250 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
251 assume_aligned or alloc_align attributes.
252 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
253 arguments. Handle also assume_aligned and alloc_align attributes.
254 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
255 Handle calls to functions with assume_aligned or alloc_align
257 * doc/extend.texi: Document assume_aligned and alloc_align
260 2014-02-08 Terry Guo <terry.guo@arm.com>
262 * doc/invoke.texi: Document ARM -march=armv7e-m.
264 2014-02-08 Jakub Jelinek <jakub@redhat.com>
266 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
267 flag on __cilkrts_rethrow builtin.
270 * ipa-cp.c (determine_versionability): Fail at -O0
271 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
272 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
275 2014-02-04 Jakub Jelinek <jakub@redhat.com>
278 * tree-inline.c (copy_forbidden): Fail for
279 __attribute__((optimize (0))) functions.
281 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
283 * varpool.c: Include pointer-set.h.
284 (varpool_remove_unreferenced_decls): Variables in other partitions
285 will not be output; be however careful to not lose information
288 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
290 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
291 lookup in the vtable constructor.
293 2014-02-07 Jeff Law <law@redhat.com>
296 * config/m68k/m68k.md (ashldi_extsi): Turn into a
297 define_insn_and_split.
299 * ipa-inline.c (inline_small_functions): Fix typos.
301 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
303 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
304 (s390_can_use_return_insn): Declare.
305 * config/s390/s390.h (EPILOGUE_USES): Define.
306 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
308 (s390_chunkify_start): Handle return JUMP_LABELs.
309 (s390_early_mach): Emit a main_pool instruction on the entry edge.
310 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
311 (s390_can_use_return_insn): New functions.
312 (s390_fix_long_loop_prediction): Handle conditional returns.
313 (TARGET_SET_UP_BY_PROLOGUE): Define.
314 * config/s390/s390.md (ANY_RETURN): New code iterator.
315 (*creturn, *csimple_return, return, simple_return): New patterns.
317 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
319 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
320 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
321 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
322 REG_CFA_RESTORE list when deciding not to restore a register.
324 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
326 * config/s390/s390.c: Include tree-pass.h and context.h.
327 (s390_early_mach): New function, split out from...
328 (s390_emit_prologue): ...here.
329 (pass_data_s390_early_mach): New pass structure.
330 (pass_s390_early_mach): New class.
331 (s390_option_override): Create and register early_mach pass.
334 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
336 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
337 to match for the exit block.
339 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
341 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
342 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
343 Reject misaligned operands.
345 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
347 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
350 2014-02-07 Richard Biener <rguenther@suse.de>
353 * gimple-low.c (lower_builtin_posix_memalign): New function.
354 (lower_stmt): Call it to lower posix_memalign in a way
355 to make alignment info accessible.
357 2014-02-07 Jakub Jelinek <jakub@redhat.com>
360 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
361 __builtin_setjmp_receiver.
363 2014-02-07 Richard Biener <rguenther@suse.de>
366 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
367 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
368 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
369 Handle BUILT_IN_POSIX_MEMALIGN.
370 (find_func_clobbers): Likewise.
371 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
372 (call_may_clobber_ref_p_1): Likewise.
374 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
377 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
379 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
382 * lto-cgraph.c (lto_output_node): Use
383 symtab_get_symbol_partitioning_class.
384 (lto_output_varpool_node): likewise.
385 (symtab_get_symbol_partitioning_class): Move here from
387 * cgraph.h (symbol_partitioning_class): Likewise.
388 (symtab_get_symbol_partitioning_class): Declare.
390 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
392 * ggc.h (ggc_internal_cleared_alloc): New macro.
393 * vec.h (vec_safe_copy): Handle memory stats.
394 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
395 * target-globals.c (save_target_globals): Likewise.
397 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
400 * expr.c (emit_move_resolve_push): Export; be bit more selective
401 on when to clear alias set.
402 * expr.h (emit_move_resolve_push): Declare.
403 * function.h (struct function): Add tail_call_marked.
404 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
405 * config/i386/i386-protos.h (ix86_expand_push): Remove.
406 * config/i386/i386.md (TImode move expander): De not call
408 (FP push expanders): Preserve memory attributes.
409 * config/i386/sse.md (push<mode>1): Remove.
410 * config/i386/i386.c (ix86_expand_vector_move): Handle push
412 (ix86_expand_push): Remove.
413 * config/i386/mmx.md (push<mode>1): Remove.
415 2014-02-06 Jakub Jelinek <jakub@redhat.com>
417 PR rtl-optimization/60030
418 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
419 lopart with paradoxical subreg before shifting it up by hprec.
421 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
423 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
424 Remove extra newline at end of file.
425 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
426 (arm_issue_rate): Handle cortexa57.
427 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
428 (cortex-a57.cortex-a53): Likewise.
430 2014-02-06 Jakub Jelinek <jakub@redhat.com>
433 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
434 don't record in REG_FRAME_RELATED_EXPR registers not set in that
436 (arm_expand_prologue): Adjust all callers.
437 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
438 info, registers also at the lowest numbered registers side. Use
439 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
443 * var-tracking.c (adjust_mems): Before adding a SET
444 to amd->side_effects, adjust it's SET_SRC using
445 simplify_replace_fn_rtx.
447 2014-02-06 Alan Modra <amodra@gmail.com>
450 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
451 change SDmode to DDmode when lra_in_progress.
453 2014-02-06 Jakub Jelinek <jakub@redhat.com>
456 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
457 free_data_ref on the dr first, and before goto again also set dr
458 to the next dr. For simd_lane_access, free old datarefs[i] before
459 overwriting it. For get_vectype_for_scalar_type failure, don't
460 free_data_ref if simd_lane_access.
462 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
465 * tree.h (opts_for_fn): New inline function.
466 (opt_for_fn): Define.
467 * config/i386/i386.c (ix86_function_regparm): Use
468 opt_for_fn (decl, optimize) instead of optimize.
470 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
472 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
473 for SYMBOL_REF in large memory model.
475 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
477 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
479 (cortex-a57): Likewise.
480 (cortex-a57.cortex-a53): Likewise.
482 2014-02-06 Yury Gribov <y.gribov@samsung.com>
483 Kugan Vivekanandarajah <kuganv@linaro.org>
485 * config/arm/arm.c (arm_vector_alignment_reachable): Check
487 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
489 2014-02-06 Richard Biener <rguenther@suse.de>
491 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
492 set_loop_copy and initialize_original_copy_tables.
494 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
496 * config/aarch64/aarch64-simd.md
497 (aarch64_ashr_simddi): Change QI to SI.
499 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
500 Jakub Jelinek <jakub@redhat.com>
503 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
506 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
508 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
509 CODE_FOR_altivec_vpku[hw]um to
510 CODE_FOR_altivec_vpku[hw]um_direct.
511 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
512 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
513 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
514 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
516 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
518 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
519 generation for -maltivec=be.
520 (altivec_vsumsws): Simplify redundant test.
522 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
524 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
525 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
526 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
527 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
529 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
531 (altivec_vpks<VI_char>ss): Likewise.
532 (altivec_vpks<VI_char>us): Likewise.
533 (altivec_vpku<VI_char>us): Likewise.
534 (altivec_vpku<VI_char>um): Likewise.
535 (altivec_vpku<VI_char>um_direct): New (copy of
536 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
538 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
539 target is little endian and -maltivec=be is not specified.
540 (*altivec_vupkhs<VU_char>_direct): New (copy of
541 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
543 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
544 target is little endian and -maltivec=be is not specified.
545 (*altivec_vupkls<VU_char>_direct): New (copy of
546 altivec_vupkls<VU_char> that always emits vupkls*, for internal
548 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
549 little endian and -maltivec=be is not specified.
550 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
551 little endian and -maltivec=be is not specified.
553 2014-02-05 Richard Henderson <rth@redhat.com>
556 * combine-stack-adj.c: Revert r206943.
557 * sched-int.h (struct deps_desc): Add last_args_size.
558 * sched-deps.c (init_deps): Initialize it.
559 (sched_analyze_insn): Add OUTPUT dependencies between insns that
560 contain REG_ARGS_SIZE notes.
562 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
564 * lto-cgraph.c (asm_nodes_output): Make global.
565 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
566 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
568 (driver_handle_option): Handle OPT_fwpa.
570 2014-02-05 Jakub Jelinek <jakub@redhat.com>
573 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
574 a comment typo and formatting issue. If odr_hash hasn't been
575 created, return vNULL and set *completep to false.
578 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
579 bb with no successors.
581 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
584 * doc/invoke.texi (-march): Clarify documentation for ARM.
588 2014-02-05 Richard Biener <rguenther@suse.de>
590 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
591 when not vectorizing because of too many alias checks.
592 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
593 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
595 2014-02-05 Nick Clifton <nickc@redhat.com>
597 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
598 accept extended registers in any mode when compiling for the
601 2014-02-05 Yury Gribov <y.gribov@samsung.com>
603 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
604 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
605 sanitization attributes.
606 (can_inline_edge_p): Likewise.
607 (sanitize_attrs_match_for_inline_p): New function.
609 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
611 * ipa-prop.c (detect_type_change): Shor circuit testing of
612 type changes on THIS pointer.
614 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
617 * config/pa/pa.c (legitimize_tls_address): Return original address
618 if not passed a SYMBOL_REF rtx.
619 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
621 (pa_emit_move_sequence): Simplify TLS source operands.
622 (pa_legitimate_constant_p): Reject all TLS constants.
623 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
624 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
626 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
628 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
629 groups when we know they are controlled by LTO.
630 * varasm.c (default_binds_local_p_1): If object is in other partition,
631 it will be resolved locally.
633 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
635 * config/host-linux.c (linux_gt_pch_use_address): Don't
636 use SSIZE_MAX because it is not always defined.
638 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
641 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
642 threshold for pseudo splitting.
643 (update_ebb_live_info): Process call argument hard registers and
644 hard registers from insn definition too.
645 (max_small_class_regs_num): New constant.
646 (inherit_in_ebb): Update live hard regs through EBBs. Update
647 reloads_num only for small register classes. Don't split for
650 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
653 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
656 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
658 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
660 2014-02-04 Marek Polacek <polacek@redhat.com>
662 * gdbinit.in (pel): Define.
664 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
666 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
669 2014-02-04 Richard Biener <rguenther@suse.de>
672 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
673 in function context local.
674 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
675 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
676 similar to LTO_imported_decl_ref.
678 2014-02-04 Jakub Jelinek <jakub@redhat.com>
680 PR tree-optimization/60002
681 * cgraphclones.c (build_function_decl_skip_args): Clear
684 PR tree-optimization/60023
685 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
686 false to gsi_replace.
687 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
688 has been in some EH region and vec_stmt could throw, add
689 vec_stmt into the same EH region.
690 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
691 has no lhs, ignore it.
692 * internal-fn.c (expand_MASK_LOAD): Likewise.
695 * tree-inline.c (copy_forbidden): Fail for
696 __attribute__((optimize (0))) functions.
699 * omp-low.c (simd_clone_struct_copy): If from->inbranch
700 is set, copy one less argument.
701 (expand_simd_clones): Don't subtract clone_info->inbranch
702 from simd_clone_struct_alloc argument.
704 PR rtl-optimization/57915
705 * recog.c (simplify_while_replacing): If all unary/binary/relational
706 operation arguments are constant, attempt to simplify those.
709 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
710 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
712 2014-02-04 Richard Biener <rguenther@suse.de>
714 PR tree-optimization/60012
715 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
716 TBAA disambiguation to all DDRs.
718 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
721 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
722 (LINK_SPEC): Use it for -shared, -shared-libgcc.
724 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
727 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
729 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
731 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
732 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
734 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
737 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
738 to figure out targets of polymorphic calls with known decl.
739 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
740 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
741 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
742 (get_polymorphic_call_info): ... here.
743 (get_polymorphic_call_info_from_invariant): New function.
745 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
747 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
748 lookup via vtable pointer; check for type consistency
749 and turn inconsitent facts into UNREACHABLE.
750 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
751 * gimple-fold.c (gimple_get_virt_method_for_vtable):
752 Do not ICE on type inconsistent querries; return UNREACHABLE
755 2014-02-03 Richard Henderson <rth@twiddle.net>
758 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
759 already processed this node.
760 (normalize_one_pred_1): Pass along mark_set.
761 (normalize_one_pred): Create and destroy a pointer_set_t.
762 (normalize_one_pred_chain): Likewise.
764 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
766 PR gcov-profile/58602
767 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
769 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
772 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
773 Try to devirtualize by the knowledge of virtual table pointer given by
774 aggregate propagation.
775 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
776 ipa_print_node_jump_functions): Dump also offset that
777 is relevant for polymorphic calls.
778 (determine_known_aggregate_parts): Add arg_type parameter; use it
779 instead of determining the type from pointer type.
780 (ipa_compute_jump_functions_for_edge): Update call of
781 determine_known_aggregate_parts.
782 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
783 (gimple_get_virt_method_for_binfo): ... here; simplify using
784 vtable_pointer_value_to_vtable.
785 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
786 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
787 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
788 (vtable_pointer_value_to_vtable): Break out from ...; handle also
790 (vtable_pointer_value_to_binfo): ... here.
791 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
793 2014-02-03 Teresa Johnson <tejohnson@google.com>
795 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
796 redef of outer loop index variable.
798 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
802 * doc/extend.texi (Function Attributes): Typo.
804 2014-02-03 Cong Hou <congh@google.com>
806 PR tree-optimization/60000
807 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
808 if the vectorized statement is a store. A store statement can only
809 appear at the end of pattern statements.
811 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
813 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
814 (ix86_option_override_internal): Default long double to 64-bit for
815 32-bit Bionic and to 128-bit for 64-bit Bionic.
817 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
818 TARGET_LONG_DOUBLE_128 is true.
819 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
821 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
822 (mlong-double-64): Negate -mlong-double-128.
823 (mlong-double-128): New option.
825 * config/i386/i386-c.c (ix86_target_macros): Define
826 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
828 * doc/invoke.texi: Document -mlong-double-128.
830 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
832 PR rtl-optimization/60024
833 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
835 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
837 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
839 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
841 PR rtl-optimization/57662
842 * sel-sched.c (code_motion_path_driver): Do not mark already not
843 existing blocks in the visiting bitmap.
845 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
847 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
848 on the insn being emitted.
850 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
851 Will Deacon <will.deacon@arm.com>
853 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
855 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
857 * config/arm/arm-tables.opt: Regenerate.
859 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
861 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
862 for vector types other than V16QImode.
863 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
864 define_expand, and call altivec_expand_vec_perm_le when producing
865 code with little endian element order.
866 (*altivec_vperm_<mode>_internal): New insn having previous
867 behavior of altivec_vperm_<mode>.
868 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
869 altivec_expand_vec_perm_le when producing code with little endian
871 (*altivec_vperm_<mode>_uns_internal): New insn having previous
872 behavior of altivec_vperm_<mode>_uns.
874 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
876 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
877 (altivec_vsumsws): Add handling for -maltivec=be with a little
879 (altivec_vsumsws_direct): New.
880 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
883 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
885 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
886 vtable_pointer_value_to_binfo): New functions.
887 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
888 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
890 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
892 * config/nios2/nios2.md (load_got_register): Initialize GOT
893 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
894 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
896 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
898 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
899 preserverd by passthrough, do not propagate the type.
901 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
903 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
904 (mips_atomic_assign_expand_fenv): New function.
905 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
907 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
909 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
910 (__builtin_mips_set_fcsr): Likewise.
911 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
913 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
914 (mips16_expand_set_fcsr): Likewise.
915 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
916 (mips16_set_fcsr_stub): Likewise.
917 (mips16_get_fcsr_one_only_stub): New class.
918 (mips16_set_fcsr_one_only_stub): Likewise.
919 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
920 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
921 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
922 (hard_float): New availability predicate.
923 (mips_builtins): Add get_fcsr and set_fcsr.
924 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
925 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
926 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
927 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
928 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
931 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
933 * config/mips/mips.c (mips_one_only_stub): New class.
934 (mips_need_mips16_rdhwr_p): Replace with...
935 (mips16_rdhwr_stub): ...this new variable.
936 (mips16_stub_call_address): New function.
937 (mips16_rdhwr_one_only_stub): New class.
938 (mips_expand_thread_pointer): Use mips16_stub_call_address.
939 (mips_output_mips16_rdhwr): Delete.
940 (mips_finish_stub): New function.
941 (mips_code_end): Use it to handle rdhwr stubs.
943 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
946 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
947 when calculating size of integer atomic types.
949 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
951 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
953 2014-02-01 Jakub Jelinek <jakub@redhat.com>
955 PR tree-optimization/60003
956 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
957 * profile.c (branch_prob): Use gimple_call_builtin_p
958 to check for BUILT_IN_SETJMP_RECEIVER.
959 * tree-inline.c (copy_bb): Call notice_special_calls.
961 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
964 * lra-constraints.c (process_alt_operands): Update reload_sum only
967 2014-01-31 Richard Henderson <rth@redhat.com>
970 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
971 until after else_eh is processed.
973 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
975 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
976 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
977 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
978 in smmintrin.h, remove them.
979 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
980 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
981 * config/i386/i386.md (ROUND_SAE): Fix value.
982 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
983 (const48_operand): New.
984 * config/i386/subst.md (round), (round_expand): Use
985 const_4_or_8_to_11_operand.
986 (round_saeonly), (round_saeonly_expand): Use const48_operand.
988 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
990 * config/i386/constraints.md (Yk): Swap meaning with k.
991 * config/i386/i386.md (movhi_internal): Change Yk to k.
992 (movqi_internal): Ditto.
993 (*k<logic><mode>): Ditto.
996 (kandn<mode>): Ditto.
997 (*<code>hi_1): Ditto.
998 (*<code>qi_1): Ditto.
999 (kxnor<mode>): Ditto.
1000 (kortestzhi): Ditto.
1001 (kortestchi): Ditto.
1003 (*one_cmplhi2_1): Ditto.
1004 (*one_cmplqi2_1): Ditto.
1005 * config/i386/sse.md (): Change k to Yk.
1006 (avx512f_load<mode>_mask): Ditto.
1007 (avx512f_blendm<mode>): Ditto.
1008 (avx512f_store<mode>_mask): Ditto.
1009 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1010 (avx512f_storedqu<mode>_mask): Ditto.
1011 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1013 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1014 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1015 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1016 (avx512f_maskcmp<mode>3): Ditto.
1017 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1018 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1019 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1020 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1021 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1022 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1023 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1024 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1025 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1026 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1027 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1028 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1029 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1030 (vec_extract_lo_<mode>_maskm): Ditto.
1031 (vec_extract_hi_<mode>_maskm): Ditto.
1032 (avx512f_vternlog<mode>_mask): Ditto.
1033 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1034 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1035 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1036 (avx512f_<code>v8div16qi2_mask): Ditto.
1037 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1038 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1039 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1040 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1041 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1042 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1043 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1044 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1045 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1046 (avx512cd_maskb_vec_dupv8di): Ditto.
1047 (avx512cd_maskw_vec_dupv16si): Ditto.
1048 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1049 (avx512f_vpermi2var<mode>3_mask): Ditto.
1050 (avx512f_vpermi2var<mode>3_mask): Ditto.
1051 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1052 (*avx512f_gathersi<mode>): Ditto.
1053 (*avx512f_gathersi<mode>_2): Ditto.
1054 (*avx512f_gatherdi<mode>): Ditto.
1055 (*avx512f_gatherdi<mode>_2): Ditto.
1056 (*avx512f_scattersi<mode>): Ditto.
1057 (*avx512f_scatterdi<mode>): Ditto.
1058 (avx512f_compress<mode>_mask): Ditto.
1059 (avx512f_compressstore<mode>_mask): Ditto.
1060 (avx512f_expand<mode>_mask): Ditto.
1061 * config/i386/subst.md (mask): Change k to Yk.
1062 (mask_scalar_merge): Ditto.
1065 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1067 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1069 2014-01-31 Richard Biener <rguenther@suse.de>
1072 * builtins.c (fold_builtin_memory_op): Make sure to not
1073 use a floating-point mode or a boolean or enumeral type for
1076 2014-01-30 DJ Delorie <dj@redhat.com>
1078 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1079 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1080 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1081 whenever main() has an epilogue.
1083 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1085 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1086 unused variable "field".
1087 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1088 (vsx_mergeh_<mode>): Likewise.
1089 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1090 (altivec_vmrghh): Likewise.
1091 (altivec_vmrghw): Likewise.
1092 (altivec_vmrglb): Likewise.
1093 (altivec_vmrglh): Likewise.
1094 (altivec_vmrglw): Likewise.
1095 (altivec_vspltb): Add missing uses.
1096 (altivec_vsplth): Likewise.
1097 (altivec_vspltw): Likewise.
1098 (altivec_vspltsf): Likewise.
1100 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1103 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1104 frame related instructions.
1106 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1108 PR rtl-optimization/59959
1109 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1110 any reload of register whose subreg is invalid.
1112 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1114 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1115 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1116 Add missing return type - void.
1118 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1120 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1121 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1122 remove element index adjustment for endian (now handled in vsx.md
1124 (altivec_expand_vec_perm_const): Use
1125 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1126 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1127 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1128 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1129 define_expand and a new define_insn *altivec_vspltb_internal;
1130 adjust for -maltivec=be on a little endian target.
1131 (altivec_vspltb_direct): New.
1132 (altivec_vsplth): Divide into a define_expand and a new
1133 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1134 little endian target.
1135 (altivec_vsplth_direct): New.
1136 (altivec_vspltw): Divide into a define_expand and a new
1137 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1138 little endian target.
1139 (altivec_vspltw_direct): New.
1140 (altivec_vspltsf): Divide into a define_expand and a new
1141 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1142 a little endian target.
1144 2014-01-30 Richard Biener <rguenther@suse.de>
1146 PR tree-optimization/59993
1147 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1148 can propagate form the earlier stmt and avoid the transform
1149 when the intermediate result is needed.
1151 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1153 * README.Portability: Fix typo.
1155 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1157 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1158 comparison_operator with ordered_comparison_operator.
1160 2014-01-30 Nick Clifton <nickc@redhat.com>
1162 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1163 Rename to mn10300_store_multiple_regs.
1164 * config/mn10300/mn10300.c: Likewise.
1165 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1166 store_multiple_regs.
1167 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1168 Call mn10300_store_multiple_regs.
1170 2014-01-30 Nick Clifton <nickc@redhat.com>
1171 DJ Delorie <dj@redhat.com>
1173 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1174 %fp 2 to keep registers after it properly word-aligned.
1175 (rl78_alloc_physical_registers_umul): Handle the case where both
1176 input operands are the same.
1178 2014-01-30 Richard Biener <rguenther@suse.de>
1180 PR tree-optimization/59903
1181 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1184 2014-01-30 Jason Merrill <jason@redhat.com>
1187 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1190 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1192 2014-01-30 Richard Biener <rguenther@suse.de>
1194 PR tree-optimization/59951
1195 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1197 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1200 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1201 SFmode to DFmode case.
1203 2014-01-29 DJ Delorie <dj@redhat.com>
1205 * config/msp430/msp430.opt (-minrt): New.
1206 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1208 (ENDFILE_SPEC): Likewise.
1210 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1212 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1213 (estimate_function_body_sizes): Use it.
1215 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1218 * dwarf2out.c (is_cxx_auto): New.
1219 (is_base_type): Use it.
1220 (gen_type_die_with_usage): Likewise.
1222 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1224 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1225 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1226 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1227 -maltivec=be with LE targets.
1228 (vsx_mergeh_<mode>): Likewise.
1229 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1230 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1231 (altivec_vmrghb): Replace with define_expand and new
1232 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1233 (altivec_vmrghb_direct): New define_insn.
1234 (altivec_vmrghh): Replace with define_expand and new
1235 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1236 (altivec_vmrghh_direct): New define_insn.
1237 (altivec_vmrghw): Replace with define_expand and new
1238 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1239 (altivec_vmrghw_direct): New define_insn.
1240 (*altivec_vmrghsf): Adjust for endianness.
1241 (altivec_vmrglb): Replace with define_expand and new
1242 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1243 (altivec_vmrglb_direct): New define_insn.
1244 (altivec_vmrglh): Replace with define_expand and new
1245 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1246 (altivec_vmrglh_direct): New define_insn.
1247 (altivec_vmrglw): Replace with define_expand and new
1248 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1249 (altivec_vmrglw_direct): New define_insn.
1250 (*altivec_vmrglsf): Adjust for endianness.
1251 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1252 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1253 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1254 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1255 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1256 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1257 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1258 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1260 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1262 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1263 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1266 2014-01-29 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/58742
1269 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1270 associate_pointerplus_align.
1271 (associate_pointerplus_diff): New function.
1272 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1273 and associate_pointerplus_diff.
1275 2014-01-29 Richard Biener <rguenther@suse.de>
1277 * lto-streamer.h (LTO_major_version): Bump to 3.
1278 (LTO_minor_version): Reset to 0.
1280 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1282 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1283 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1284 (arm_file_start): Generate correct asm header for armv7ve.
1285 * config/arm/bpabi.h: Add multilib support for armv7ve.
1286 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1287 and cortex-a15 to armv7ve.
1288 * config/arm/t-aprofile: Add multilib support for armv7ve.
1289 * doc/invoke.texi: Document -march=armv7ve.
1291 2014-01-29 Richard Biener <rguenther@suse.de>
1293 PR tree-optimization/58742
1294 * tree-ssa-forwprop.c (associate_plusminus): Return true
1295 if we changed sth, defer EH cleanup to ...
1296 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1297 (simplify_mult): New function.
1299 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1302 PR tree-optimization/59920
1303 * tree.c (build_common_builtin_nodes): Remove
1304 __builtin_setjmp_dispatcher initialization.
1305 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1306 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1307 instead of gsi_after_labels + manually skipping debug stmts.
1308 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1309 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1310 * tree-inline.c (copy_edges_for_bb): Remove
1311 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1312 argument. Ignore computed_goto_p stmts. Don't call
1313 make_abnormal_goto_edges. If a call might need abnormal edges
1314 for non-local gotos, see if it already has an edge to
1315 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1316 with true argument, don't do anything then, otherwise add
1317 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1318 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1320 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1321 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1322 (lower_stmt): Don't set data->calls_builtin_setjmp.
1323 (lower_builtin_setjmp): Adjust comment.
1324 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1325 * tree-cfg.c (found_computed_goto): Remove.
1326 (factor_computed_gotos): Remove.
1327 (make_goto_expr_edges): Return bool, true for computed gotos.
1328 Don't call make_abnormal_goto_edges.
1329 (build_gimple_cfg): Don't set found_computed_goto, don't call
1330 factor_computed_gotos.
1331 (computed_goto_p): No longer static.
1332 (make_blocks): Don't set found_computed_goto.
1333 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1334 (make_edges): If make_goto_expr_edges returns true, push bb
1335 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1336 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1337 vector. Record mapping between bbs and OpenMP regions if there
1338 are any, adjust make_gimple_omp_edges caller. Call
1339 handle_abnormal_edges.
1340 (make_abnormal_goto_edges): Remove.
1341 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1342 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1343 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1344 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1345 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1346 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1347 filling *region also set *region_idx to (*region)->entry->index.
1350 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1351 For REGs set ORIGINAL_REGNO.
1353 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1355 * doc/md.texi: Mention that a target shouldn't implement
1356 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1359 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1361 PR tree-optimization/59594
1362 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1363 a copy of the datarefs vector rather than the vector itself.
1365 2014-01-28 Jason Merrill <jason@redhat.com>
1368 * dwarf2out.c (auto_die): New static.
1369 (gen_type_die_with_usage): Handle C++1y 'auto'.
1370 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1373 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1376 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1377 (SPEC_X32): Likewise.
1378 (SPEC_64): Likewise.
1379 * config/i386/i386.c (ix86_option_override_internal): Turn off
1380 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1382 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1383 * config/i386/i386.h (TARGET_16BIT): New macro.
1384 (TARGET_16BIT_P): Likewise.
1385 * config/i386/i386.opt: Add m16.
1386 * doc/invoke.texi: Document -m16.
1388 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1390 PR preprocessor/59935
1391 * input.c (location_get_source_line): Bail out on when line number
1392 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1394 2014-01-28 Richard Biener <rguenther@suse.de>
1396 PR tree-optimization/58742
1397 * tree-ssa-forwprop.c (associate_plusminus): Handle
1398 pointer subtraction of the form (T)(P + A) - (T)P.
1400 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1402 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1405 2014-01-28 Richard Biener <rguenther@suse.de>
1408 2014-01-28 Richard Biener <rguenther@suse.de>
1410 PR rtl-optimization/45364
1411 PR rtl-optimization/59890
1412 * var-tracking.c (local_get_addr_clear_given_value): Handle
1413 already cleared slot.
1414 (val_reset): Handle not allocated local_get_addr_cache.
1415 (vt_find_locations): Use post-order on the inverted CFG.
1417 2014-01-28 Richard Biener <rguenther@suse.de>
1419 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1421 2014-01-28 Richard Biener <rguenther@suse.de>
1423 PR rtl-optimization/45364
1424 PR rtl-optimization/59890
1425 * var-tracking.c (local_get_addr_clear_given_value): Handle
1426 already cleared slot.
1427 (val_reset): Handle not allocated local_get_addr_cache.
1428 (vt_find_locations): Use post-order on the inverted CFG.
1430 2014-01-28 Alan Modra <amodra@gmail.com>
1432 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1433 * configure.ac <recursive call for build != host>: Define
1434 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1435 and LD_FOR_BUILD too.
1436 * configure: Regenerate.
1438 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1440 * config/i386/i386.c (get_builtin_code_for_version): Separate
1441 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1442 Broadwell from Haswell.
1444 2014-01-27 Steve Ellcey <sellcey@mips.com>
1446 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1447 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1448 * config/mips/mips.c (mips_option_override): Change setting
1450 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1451 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1452 Change from Mask to Var.
1454 2014-01-27 Jeff Law <law@redhat.com>
1456 * ipa-inline.c (inline_small_functions): Fix typo.
1458 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1460 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1461 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1462 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1463 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1464 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1465 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1466 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1467 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1468 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1469 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1470 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1471 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1472 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1473 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1474 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1475 (_mm512_storeu_epi64): Ditto.
1476 (_mm512_cmpge_epi32_mask): Ditto.
1477 (_mm512_cmpge_epu32_mask): Ditto.
1478 (_mm512_cmpge_epi64_mask): Ditto.
1479 (_mm512_cmpge_epu64_mask): Ditto.
1480 (_mm512_cmple_epi32_mask): Ditto.
1481 (_mm512_cmple_epu32_mask): Ditto.
1482 (_mm512_cmple_epi64_mask): Ditto.
1483 (_mm512_cmple_epu64_mask): Ditto.
1484 (_mm512_cmplt_epi32_mask): Ditto.
1485 (_mm512_cmplt_epu32_mask): Ditto.
1486 (_mm512_cmplt_epi64_mask): Ditto.
1487 (_mm512_cmplt_epu64_mask): Ditto.
1488 (_mm512_cmpneq_epi32_mask): Ditto.
1489 (_mm512_cmpneq_epu32_mask): Ditto.
1490 (_mm512_cmpneq_epi64_mask): Ditto.
1491 (_mm512_cmpneq_epu64_mask): Ditto.
1492 (_mm512_expand_pd): Ditto.
1493 (_mm512_expand_ps): Ditto.
1494 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1495 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1496 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1497 * config/i386/i386.c (ix86_builtins): Add
1498 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1499 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1500 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1501 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1502 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1503 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1504 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1505 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1506 IX86_BUILTIN_PMOVUSQW512_MEM.
1507 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1508 __builtin_ia32_pmovsqd512mem_mask,
1509 __builtin_ia32_pmovqd512mem_mask,
1510 __builtin_ia32_pmovusqw512mem_mask,
1511 __builtin_ia32_pmovsqw512mem_mask,
1512 __builtin_ia32_pmovqw512mem_mask,
1513 __builtin_ia32_pmovusdw512mem_mask,
1514 __builtin_ia32_pmovsdw512mem_mask,
1515 __builtin_ia32_pmovdw512mem_mask,
1516 __builtin_ia32_pmovqb512mem_mask,
1517 __builtin_ia32_pmovusqb512mem_mask,
1518 __builtin_ia32_pmovsqb512mem_mask,
1519 __builtin_ia32_pmovusdb512mem_mask,
1520 __builtin_ia32_pmovsdb512mem_mask,
1521 __builtin_ia32_pmovdb512mem_mask.
1522 (bdesc_args): Add __builtin_ia32_expanddf512,
1523 __builtin_ia32_expandsf512.
1524 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1525 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1526 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1527 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1528 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1529 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1530 (avx512f_<code>v8div16qi2_mask_store): This.
1531 (avx512f_expand<mode>): New.
1533 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1535 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1537 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1538 (_mm512_prefetch_i32scatter_pd): Ditto.
1539 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1540 (_mm512_prefetch_i64scatter_pd): Ditto.
1541 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1542 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1543 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1544 (_mm512_prefetch_i32scatter_ps): Ditto.
1545 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1546 (_mm512_prefetch_i64scatter_ps): Ditto.
1547 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1548 * config/i386/i386-builtin-types.def: Define
1549 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1550 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1551 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1552 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1553 IX86_BUILTIN_SCATTERPFQPD.
1554 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1555 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1556 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1557 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1558 __builtin_ia32_scatterpfqps.
1559 (ix86_expand_builtin): Expand new built-ins.
1560 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1561 fix memory access data type.
1562 (*avx512pf_gatherpf<mode>_mask): Ditto.
1563 (*avx512pf_gatherpf<mode>): Ditto.
1564 (avx512pf_scatterpf<mode>): Ditto.
1565 (*avx512pf_scatterpf<mode>_mask): Ditto.
1566 (*avx512pf_scatterpf<mode>): Ditto.
1567 (GATHER_SCATTER_SF_MEM_MODE): New.
1568 (avx512pf_gatherpf<mode>df): Ditto.
1569 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1570 (*avx512pf_scatterpf<mode>df): Ditto.
1572 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1575 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1576 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1579 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1581 * common/config/arm/arm-common.c
1582 (arm_rewrite_mcpu): Handle multiple names.
1584 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1586 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1588 * gimple-builder.h (create_gimple_tmp): Delete.
1590 2014-01-27 Christian Bruel <christian.bruel@st.com>
1592 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1595 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1597 * config/pa/pa.md (call): Generate indirect long calls to non-local
1598 functions when outputing 32-bit code.
1599 (call_value): Likewise except for special call to buggy powf function.
1601 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1602 portable runtime and PIC indirect calls.
1603 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1604 and PIC call sequences. Use ldo instead of blr to set return register
1605 in PIC call sequence.
1607 2014-01-25 Walter Lee <walt@tilera.com>
1609 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1610 avoid clobbering a live register.
1612 2014-01-25 Walter Lee <walt@tilera.com>
1614 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1615 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1616 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1617 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1619 2014-01-25 Walter Lee <walt@tilera.com>
1621 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1622 arguments on even registers.
1623 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1625 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1626 (BIGGEST_ALIGNMENT): Ditto.
1627 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1629 2014-01-25 Walter Lee <walt@tilera.com>
1631 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1632 insns before bundling.
1633 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1635 2014-01-25 Walter Lee <walt@tilera.com>
1637 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1638 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1639 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1641 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1643 * config/mips/constraints.md (kl): Delete.
1644 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1645 define expands, using...
1646 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1647 instructions for MIPS16.
1648 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1649 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1651 2014-01-25 Walter Lee <walt@tilera.com>
1653 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1657 2014-01-25 Walter Lee <walt@tilera.com>
1659 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1660 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1662 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1664 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1667 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1669 * print-rtl.c (in_call_function_usage): New var.
1670 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1671 EXPR_LIST mode as mode and not as reg note name.
1674 * cfgloopmanip.c (copy_loop_info): If
1675 loop->warned_aggressive_loop_optimizations, make sure
1676 the flag is set in target loop too.
1678 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1680 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1682 * builtins.def: Likewise.
1683 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1684 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1685 * ira.c (ira_setup_eliminable_regset): Likewise.
1686 * omp-low.c (gate_expand_omp): Likewise.
1687 (execute_lower_omp): Likewise.
1688 (diagnose_sb_0): Likewise.
1689 (gate_diagnose_omp_blocks): Likewise.
1690 (simd_clone_clauses_extract): Likewise.
1693 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1695 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1696 correction for little endian...
1697 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1700 2014-01-24 Jeff Law <law@redhat.com>
1702 PR tree-optimization/59919
1703 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1704 for non-returning calls.
1706 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1708 * common/config/aarch64/aarch64-common.c
1709 (aarch64_rewrite_mcpu): Handle multiple names.
1710 * config/aarch64/aarch64.h
1711 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1713 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1715 * input.c (add_file_to_cache_tab): Handle the case where fopen
1718 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1721 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1722 from push operand if code of push isn't PRE_DEC.
1724 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1727 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1728 -mquad-memory-atomic. Update -mquad-memory documentation to say
1729 it is only used for non-atomic loads/stores.
1731 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1732 -mquad-memory or -mquad-memory-atomic switches.
1734 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1735 -mquad-memory-atomic to ISA 2.07 support.
1737 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1738 to separate support of normal quad word memory operations (ldq, stq)
1739 from the atomic quad word memory operations.
1741 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1742 support to separate non-atomic quad word operations from atomic
1743 quad word operations. Disable non-atomic quad word operations in
1744 little endian mode so that we don't have to swap words after the
1745 load and before the store.
1746 (quad_load_store_p): Add comment about atomic quad word support.
1747 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1748 options printed with -mdebug=reg.
1750 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1751 -mquad-memory-atomic as the test for whether we have quad word
1752 atomic instructions.
1753 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1754 or -mp8-vector are used, allow byte/half-word atomic operations.
1756 * config/rs6000/sync.md (load_lockedti): Insure that the address
1757 is a proper indexed or indirect address for the lqarx instruction.
1758 On little endian systems, swap the hi/lo registers after the lqarx
1760 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1761 insure the address is valid for the lqarx instruction.
1762 (store_conditionalti): Insure that the address is a proper indexed
1763 or indirect address for the stqcrx. instruction. On little endian
1764 systems, swap the hi/lo registers before doing the stqcrx.
1766 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1767 insure the address is valid for the stqcrx. instruction.
1769 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1770 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1771 type of quad memory support is available.
1773 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1776 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1777 there is a danger of looping.
1779 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1782 force flag_ira_loop_pressure if set via command line.
1784 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1786 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1787 (ashr_simd): New builtin handling DI mode.
1788 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1789 (aarch64_sshr_simddi): New match pattern.
1790 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1791 (vshrd_n_s64): Likewise.
1792 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1794 2014-01-23 Nick Clifton <nickc@redhat.com>
1796 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1797 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1798 favour of mcu specific scripts.
1799 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1802 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1803 Alex Velenko <Alex.Velenko@arm.com>
1805 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1806 (vaddv_s16): Likewise.
1807 (vaddv_s32): Likewise.
1808 (vaddv_u8): Likewise.
1809 (vaddv_u16): Likewise.
1810 (vaddv_u32): Likewise.
1811 (vaddvq_s8): Likewise.
1812 (vaddvq_s16): Likewise.
1813 (vaddvq_s32): Likewise.
1814 (vaddvq_s64): Likewise.
1815 (vaddvq_u8): Likewise.
1816 (vaddvq_u16): Likewise.
1817 (vaddvq_u32): Likewise.
1818 (vaddvq_u64): Likewise.
1819 (vaddv_f32): Likewise.
1820 (vaddvq_f32): Likewise.
1821 (vaddvq_f64): Likewise.
1822 (vmaxv_f32): Likewise.
1823 (vmaxv_s8): Likewise.
1824 (vmaxv_s16): Likewise.
1825 (vmaxv_s32): Likewise.
1826 (vmaxv_u8): Likewise.
1827 (vmaxv_u16): Likewise.
1828 (vmaxv_u32): Likewise.
1829 (vmaxvq_f32): Likewise.
1830 (vmaxvq_f64): Likewise.
1831 (vmaxvq_s8): Likewise.
1832 (vmaxvq_s16): Likewise.
1833 (vmaxvq_s32): Likewise.
1834 (vmaxvq_u8): Likewise.
1835 (vmaxvq_u16): Likewise.
1836 (vmaxvq_u32): Likewise.
1837 (vmaxnmv_f32): Likewise.
1838 (vmaxnmvq_f32): Likewise.
1839 (vmaxnmvq_f64): Likewise.
1840 (vminv_f32): Likewise.
1841 (vminv_s8): Likewise.
1842 (vminv_s16): Likewise.
1843 (vminv_s32): Likewise.
1844 (vminv_u8): Likewise.
1845 (vminv_u16): Likewise.
1846 (vminv_u32): Likewise.
1847 (vminvq_f32): Likewise.
1848 (vminvq_f64): Likewise.
1849 (vminvq_s8): Likewise.
1850 (vminvq_s16): Likewise.
1851 (vminvq_s32): Likewise.
1852 (vminvq_u8): Likewise.
1853 (vminvq_u16): Likewise.
1854 (vminvq_u32): Likewise.
1855 (vminnmv_f32): Likewise.
1856 (vminnmvq_f32): Likewise.
1857 (vminnmvq_f64): Likewise.
1859 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1861 * config/aarch64/aarch64-simd.md
1862 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1863 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1864 (*aarch64_mul3_elt<mode>): Likewise.
1865 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1866 (*aarch64_mul3_elt_to_64v2df): Likewise.
1867 (*aarch64_mla_elt<mode>): Likewise.
1868 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1869 (*aarch64_mls_elt<mode>): Likewise.
1870 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1871 (*aarch64_fma4_elt<mode>): Likewise.
1872 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1873 (*aarch64_fma4_elt_to_64v2df): Likewise.
1874 (*aarch64_fnma4_elt<mode>): Likewise.
1875 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1876 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1877 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1878 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1879 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1880 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1881 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1882 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1883 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1885 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1887 * config/aarch64/aarch64-simd.md
1888 (aarch64_be_checked_get_lane<mode>): New define_expand.
1889 * config/aarch64/aarch64-simd-builtins.def
1890 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1891 New builtin definition.
1892 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1893 Use new safe be builtin.
1895 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1897 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1899 (aarch64_be_st1<mode>): Likewise.
1900 (aarch_ld1<VALL:mode>): Define_expand modified.
1901 (aarch_st1<VALL:mode>): Likewise.
1902 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1903 (UNSPEC_ST1): Likewise.
1905 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1907 * config/microblaze/microblaze.md: Add trap insn and attribute
1909 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1911 PR preprocessor/58580
1912 * input.h (location_get_source_line): Take an additional line_size
1914 (void diagnostics_file_cache_fini): Declare new function.
1915 * input.c (struct fcache): New type.
1916 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1917 New static constants.
1918 (diagnostic_file_cache_init, total_lines_num)
1919 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1920 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1921 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1922 (get_next_line, read_next_line, goto_next_line, read_line_num):
1923 New static function definitions.
1924 (diagnostic_file_cache_fini): New function.
1925 (location_get_source_line): Take an additional output line_len
1926 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1928 * diagnostic.c (diagnostic_finish): Call
1929 diagnostic_file_cache_fini.
1930 (adjust_line): Take an additional input parameter for the length
1931 of the line, rather than calculating it with strlen.
1932 (diagnostic_show_locus): Adjust the use of
1933 location_get_source_line and adjust_line with respect to their new
1934 signature. While displaying a line now, do not stop at the first
1935 null byte. Rather, display the zero byte as a space and keep
1936 going until we reach the size of the line.
1937 * Makefile.in: Add vec.o to OBJS-libcommon
1939 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1940 Ilya Tocar <ilya.tocar@intel.com>
1942 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1943 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1944 (__builtin_ia32_kmov16): Ditto.
1945 * config/i386/i386.md (UNSPEC_KMOV): New.
1948 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1950 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1951 (_mm512_storeu_si512): Ditto.
1953 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1956 * rtl.h (get_referenced_operands): Declare.
1957 * recog.c (get_referenced_operands): New function.
1958 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1959 operands have been referenced when recording LO_SUM references.
1961 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1963 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1965 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1967 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1968 Enable for generic and recent AMD targets.
1970 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1972 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1973 ARG_SIZE note when adjustment was eliminated.
1975 2014-01-22 Jeff Law <law@redhat.com>
1977 PR tree-optimization/59597
1978 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1979 in file. Accept new argument REGISTERING and use it to modify
1980 dump output appropriately.
1981 (register_jump_thread): Corresponding changes.
1982 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1983 thread paths involving joiner blocks. Add code to dump cancelled
1984 jump threading paths.
1986 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1988 PR rtl-optimization/59477
1989 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1990 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1992 2014-01-22 Tom Tromey <tromey@redhat.com>
1994 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1996 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1998 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2000 PR rtl-optimization/59896
2001 * lra-constraints.c (process_alt_operands): Check unused note for
2002 matched operands of insn with no output reloads.
2004 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2006 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2007 (mips_move_from_gpr_cost): Likewise.
2009 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2011 PR rtl-optimization/59858
2012 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2013 ira_class_hard_regs_num.
2014 (process_alt_operands): Increase reject for dying matched operand.
2016 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2019 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2020 smaller than size, perform several stores or loads and stores
2021 at dst + count - size to store or copy all of size bytes, rather
2022 than just last modesize bytes.
2024 2014-01-20 DJ Delorie <dj@redhat.com>
2026 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2027 that CLOBBERs are REGs before propogating their values.
2029 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2032 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2033 (cgraph_inline_failed_type): New function.
2034 * cgraph.h (DEFCIFCODE): Add type.
2035 (cgraph_inline_failed_type_t): New enum.
2036 (cgraph_inline_failed_type): New prototype.
2037 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2038 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2039 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2040 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2041 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2042 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2043 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2044 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2045 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2046 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2047 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2048 OPTIMIZATION_MISMATCH.
2049 * tree-inline.c (expand_call_inline): Emit errors during
2050 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2052 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2055 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2056 mode attribute in insn output.
2058 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2060 * output.h (output_constant): Delete.
2061 * varasm.c (output_constant): Make private.
2063 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2065 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2067 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2070 * tree.h (fold_builtin_strcat): New prototype.
2071 * builtins.c (fold_builtin_strcat): No longer static. Add len
2072 argument, if non-NULL, don't call c_strlen. Optimize
2073 directly into __builtin_memcpy instead of __builtin_strcpy.
2074 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2075 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2077 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2079 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2080 for SImode_address_operand operands, having only a REG argument.
2082 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2084 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2085 loader name using mbig-endian.
2086 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2088 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2090 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2094 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2096 * config/aarch64/aarch64-protos.h
2097 (aarch64_cannot_change_mode_class_ptr): Declare.
2098 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2099 aarch64_cannot_change_mode_class_ptr): New.
2100 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2101 backend hook aarch64_cannot_change_mode_class.
2103 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2105 * common/config/aarch64/aarch64-common.c
2106 (aarch64_handle_option): Don't handle any option order logic here.
2107 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2108 selected_cpu, warn on architecture version mismatch.
2109 (aarch64_override_options): Fix parsing order for option strings.
2111 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2112 Iain Sandoe <iain@codesourcery.com>
2115 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2116 warning. Amend comment to reflect current functionality.
2118 2014-01-20 Richard Biener <rguenther@suse.de>
2121 * builtins.c (fold_builtin_strcat): Remove case better handled
2122 by tree-ssa-strlen.c.
2124 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2126 * config/aarch64/aarch64.opt
2127 (mcpu, march, mtune): Make case-insensitive.
2129 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2132 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2133 if operands[1] is a REG or ZERO_EXTEND of a REG.
2135 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2137 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2139 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2141 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2142 long non-pic millicode calls.
2144 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2146 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2148 2014-01-19 Kito Cheng <kito@0xlab.org>
2150 * builtins.c (expand_movstr): Check movstr expand done or fail.
2152 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2153 H.J. Lu <hongjiu.lu@intel.com>
2156 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2157 to DImode for zero-extended addresses.
2159 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2161 PR rtl-optimization/57763
2162 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2163 on the new indirect jump_insn and increment LABEL_NUSES (label).
2165 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2169 * config.gcc (x86_archs): New variable.
2170 (x86_64_archs): Likewise.
2171 (x86_cpus): Likewise.
2172 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2173 --with-arch/--with-cpu= options.
2174 Support --with-arch=/--with-cpu={nehalem,westmere,
2175 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2177 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2179 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2180 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2182 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2184 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2186 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2189 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2190 clear cpp_get_options (parse_in)->warn_unused_macros for
2191 ix86_target_macros_internal with cpp_define.
2193 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2195 * jump.c (delete_related_insns): Keep (use (insn))s.
2196 * reorg.c (redundant_insn): Check for barriers too.
2198 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2200 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2202 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2204 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2205 call to $$dyncall when TARGET_LONG_CALLS is true.
2207 2014-01-17 Jeff Law <law@redhat.com>
2209 * ree.c (combine_set_extension): Temporarily disable test for
2210 changing number of hard registers.
2212 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2215 * ipa-inline-analysis.c (inline_free_summary):
2216 Do not free summary of aliases.
2218 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2221 * gimplify.c (gimplify_expr): Use create_tmp_var
2222 instead of create_tmp_var_raw. If cond doesn't have
2223 integral type, don't add the IFN_ANNOTATE builtin at all.
2225 2014-01-17 Martin Jambor <mjambor@suse.cz>
2228 * ipa-cp.c (prev_edge_clone): New variable.
2229 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2230 Also resize prev_edge_clone vector.
2231 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2232 (ipcp_edge_removal_hook): New function.
2233 (ipcp_driver): Register ipcp_edge_removal_hook.
2235 2014-01-17 Andrew Pinski <apinski@cavium.com>
2236 Steve Ellcey <sellcey@mips.com>
2239 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2242 2014-01-17 Jeff Law <law@redhat.com>
2245 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2246 so that pass_ccp runs first.
2248 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2250 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2251 (ix86_adjust_cost): Use !TARGET_XXX.
2252 (do_reorder_for_imul): Likewise.
2253 (swap_top_of_ready_list): Likewise.
2254 (ix86_sched_reorder): Likewise.
2256 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2258 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2259 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2260 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2261 (intel_memset): New. Duplicate slm_memset.
2262 (intel_cost): New. Duplicate slm_cost.
2263 (m_INTEL): New macro.
2264 (processor_target_table): Add "intel".
2265 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2266 with PROCESSOR_INTEL for "intel".
2267 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2268 PROCESSOR_SILVERMONT.
2269 (ix86_issue_rate): Likewise.
2270 (ix86_adjust_cost): Likewise.
2271 (ia32_multipass_dfa_lookahead): Likewise.
2272 (swap_top_of_ready_list): Likewise.
2273 (ix86_sched_reorder): Likewise.
2274 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2275 instead of TARGET_OPT_AGU.
2276 * config/i386/i386.h (TARGET_INTEL): New.
2277 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2278 (processor_type): Add PROCESSOR_INTEL.
2279 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2280 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2282 2014-01-17 Marek Polacek <polacek@redhat.com>
2285 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2288 2014-01-17 Richard Biener <rguenther@suse.de>
2290 PR tree-optimization/46590
2291 * opts.c (default_options_table): Add entries for
2292 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2293 all enabled at -O1 but not for -Og.
2294 * common.opt (fbranch-count-reg): Remove Init(1).
2295 (fmove-loop-invariants): Likewise.
2296 (ftree-pta): Likewise.
2298 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2300 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2301 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2302 decls to at least the GCC 4.8 used alignments.
2305 * tree-nested.c (convert_nonlocal_reference_stmt,
2306 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2307 of GIMPLE_BIND stmts, adjust associated decls.
2309 2014-01-17 Richard Biener <rguenther@suse.de>
2311 PR tree-optimization/46590
2312 * vec.h (vec<>::bseach): New member function implementing
2313 binary search according to C89 bsearch.
2314 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2315 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2316 bitmap pointer again. Make accesses_in_loop a flat array.
2317 (mem_ref_obstack): New global.
2318 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2319 (mark_ref_stored): Likewise.
2320 (ref_indep_loop_p_2): Likewise.
2321 (set_ref_stored_in_loop): New helper function.
2322 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2323 (memref_free): Adjust.
2324 (record_mem_ref_loc): Simplify.
2325 (gather_mem_refs_stmt): Adjust.
2326 (sort_locs_in_loop_postorder_cmp): New function.
2327 (analyze_memory_references): Sort accesses_in_loop after
2328 loop postorder number.
2329 (find_ref_loc_in_loop_cmp): New function.
2330 (for_all_locs_in_loop): Find relevant cluster of locs in
2331 accesses_in_loop and iterate without recursion.
2332 (execute_sm): Avoid uninit warning.
2333 (struct ref_always_accessed): Simplify.
2334 (ref_always_accessed::operator ()): Likewise.
2335 (ref_always_accessed_p): Likewise.
2336 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2337 loop postorder numbers here.
2338 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2341 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2344 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2345 on decls for which assemble_alias has been called.
2347 2014-01-17 Nick Clifton <nickc@redhat.com>
2349 * config/msp430/msp430.opt: (mcpu): New option.
2350 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2351 (msp430_option_override): Parse target_cpu. If the MCU name
2352 matches a generic string, clear target_mcu.
2353 (msp430_attr): Allow numeric interrupt values up to 63.
2354 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2355 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2357 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2359 * config/msp430/msp430.md (popm): Use %J rather than %I.
2360 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2361 (addhi_cy_i): Use immediate_operand for operand 2.
2362 * doc/invoke.texi: Document -mcpu option.
2364 2014-01-17 Richard Biener <rguenther@suse.de>
2366 PR rtl-optimization/38518
2367 * df.h (df_analyze_loop): Declare.
2368 * df-core.c: Include cfgloop.h.
2369 (df_analyze_1): Split out main part of df_analyze.
2370 (df_analyze): Adjust.
2371 (loop_inverted_post_order_compute): New function.
2372 (loop_post_order_compute): Likewise.
2373 (df_analyze_loop): New function avoiding whole-function
2375 * loop-invariant.c (find_defs): Use df_analyze_loop.
2376 (find_invariants): Adjust.
2377 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2379 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2381 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2382 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2384 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2386 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2387 traversal when removing references.
2389 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2392 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2394 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2397 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2398 pushing a reload for an autoinc when we had previously reloaded an
2399 inner part of the address.
2401 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2403 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2405 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2406 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2407 when not giving up or versioning for alias only because of
2409 (vect_analyze_data_ref_dependences): Set to true.
2410 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2412 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2413 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2417 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2420 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2421 operand 0 predicate for gathers, use a new pseudo as subtarget.
2423 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2426 * lra-constraints.c (process_alt_operands): Add printing debug info.
2427 Check absence of input/output reloads for matched operands too.
2429 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2431 PR rtl-optimization/59835
2432 * ira.c (ira_init_register_move_cost): Increase cost for
2435 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2437 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2439 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2442 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2443 non-register objects. Use gen_(high/low)part more consistently.
2446 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2449 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2450 endian support, remove tests for WORDS_BIG_ENDIAN.
2451 (p8_mfvsrd_3_<mode>): Likewise.
2452 (reload_gpr_from_vsx<mode>): Likewise.
2453 (reload_gpr_from_vsxsf): Likewise.
2454 (p8_mfvsrd_4_disf): Likewise.
2456 2014-01-16 Richard Biener <rguenther@suse.de>
2458 PR rtl-optimization/46590
2459 * lcm.c (compute_antinout_edge): Use postorder iteration.
2460 (compute_laterin): Use inverted postorder iteration.
2462 2014-01-16 Nick Clifton <nickc@redhat.com>
2465 * varasm.c (output_constant): Return the number of bytes actually
2467 (output_constructor_array_range): Update the field size with the
2468 number of bytes emitted by output_constant.
2469 (output_constructor_regular_field): Likewise. Also do not
2470 complain if the total number of bytes emitted is now greater
2471 than the expected fieldpos.
2472 * output.h (output_constant): Update prototype and descriptive comment.
2474 2014-01-16 Marek Polacek <polacek@redhat.com>
2477 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2478 it is error_mark_node.
2480 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2482 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2483 VALID_AVX256_REG_OR_OI_MODE.
2485 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2487 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2488 current procedure should be profiled.
2490 2014-01-15 Andrew Pinski <apinski@cavium.com>
2492 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2493 of moving from/to the STACK_REG register class.
2495 2014-01-15 Richard Henderson <rth@redhat.com>
2498 * reginfo.c (global_regs_decl): Globalize.
2499 * rtl.h (global_regs_decl): Declare.
2500 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2501 reserved via global_regs.
2503 2014-01-15 Teresa Johnson <tejohnson@google.com>
2505 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2507 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2509 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2510 and vmulosh rather than call gen_vec_widen_smult_*.
2511 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2512 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2513 (vec_widen_smult_even_v16qi): Likewise.
2514 (vec_widen_umult_even_v8hi): Likewise.
2515 (vec_widen_smult_even_v8hi): Likewise.
2516 (vec_widen_umult_odd_v16qi): Likewise.
2517 (vec_widen_smult_odd_v16qi): Likewise.
2518 (vec_widen_umult_odd_v8hi): Likewise.
2519 (vec_widen_smult_odd_v8hi): Likewise.
2520 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2521 vmuloub rather than call gen_vec_widen_umult_*.
2522 (vec_widen_umult_lo_v16qi): Likewise.
2523 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2524 vmulosb rather than call gen_vec_widen_smult_*.
2525 (vec_widen_smult_lo_v16qi): Likewise.
2526 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2527 rather than call gen_vec_widen_umult_*.
2528 (vec_widen_umult_lo_v8hi): Likewise.
2529 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2530 rather than call gen_vec_widen_smult_*.
2531 (vec_widen_smult_lo_v8hi): Likewise.
2533 2014-01-15 Jeff Law <law@redhat.com>
2535 PR tree-optimization/59747
2536 * ree.c (find_and_remove_re): Properly handle case where a second
2537 eliminated extension requires widening a copy created for elimination
2538 of a prior extension.
2539 (combine_set_extension): Ensure that the number of hard regs needed
2540 for a destination register does not change when we widen it.
2542 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2544 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2545 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2546 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2547 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2548 (avr-*-rtems*): Likewise.
2549 (bfin*-rtems*): Likewise.
2550 (moxie-*-rtems*): Likewise.
2551 (h8300-*-rtems*): Likewise.
2552 (i[34567]86-*-rtems*): Likewise.
2553 (lm32-*-rtems*): Likewise.
2554 (m32r-*-rtems*): Likewise.
2555 (m68k-*-rtems*): Likewise.
2556 (microblaze*-*-rtems*): Likewise.
2557 (mips*-*-rtems*): Likewise.
2558 (powerpc-*-rtems*): Likewise.
2559 (sh-*-rtems*): Likewise.
2560 (sparc-*-rtems*): Likewise.
2561 (sparc64-*-rtems*): Likewise.
2562 (v850-*-rtems*): Likewise.
2563 (m32c-*-rtems*): Likewise.
2565 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2567 PR rtl-optimization/59511
2568 * ira.c (ira_init_register_move_cost): Use memory costs for some
2569 cases of register move cost calculations.
2570 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2571 instead of BB frequency.
2572 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2573 * lra-assigns.c (find_hard_regno_for): Ditto.
2575 2014-01-15 Richard Biener <rguenther@suse.de>
2577 PR tree-optimization/59822
2578 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2579 (vectorizable_load): Use it to hoist defs of uses of invariant
2580 loads out of the loop.
2582 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2583 Kugan Vivekanandarajah <kuganv@linaro.org>
2586 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2589 2014-01-15 Richard Biener <rguenther@suse.de>
2591 PR rtl-optimization/59802
2592 * lcm.c (compute_available): Use inverted postorder to seed
2593 the initial worklist.
2595 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2598 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2599 ADDR_REGS for invalid symrefs in non-PIC code.
2601 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2604 * builtins.c (determine_block_size): Initialize *probable_max_size
2605 even if len_rtx is CONST_INT.
2607 2014-01-14 Andrew Pinski <apinski@cavium.com>
2609 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2610 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2611 (cortexa53_tunings): Likewise.
2612 (aarch64_sched_issue_rate): New function.
2613 (TARGET_SCHED_ISSUE_RATE): Define.
2615 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2617 * ira-costs.c (find_costs_and_classes): Add missed
2618 ira_init_register_move_cost_if_necessary.
2620 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2623 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2625 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2628 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2629 to indicate if type is used for function return value. Warn ABI
2630 change if the vector mode isn't available for function return value.
2631 (ix86_function_arg_advance): Pass false to type_natural_mode.
2632 (ix86_function_arg): Likewise.
2633 (ix86_gimplify_va_arg): Likewise.
2634 (function_arg_32): Don't warn ABI change.
2635 (ix86_function_value): Pass true to type_natural_mode.
2636 (ix86_return_in_memory): Likewise.
2637 (ix86_struct_value_rtx): Removed.
2638 (TARGET_STRUCT_VALUE_RTX): Likewise.
2640 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2642 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2643 converting a conditional jump into a conditional return.
2645 2014-01-14 Richard Biener <rguenther@suse.de>
2647 PR tree-optimization/58921
2648 PR tree-optimization/59006
2649 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2650 hoisting invariant stmts.
2651 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2652 invariant loads on the preheader edge if possible.
2654 2014-01-14 Joey Ye <joey.ye@arm.com>
2656 * doc/plugin.texi (Building GCC plugins): Update to C++.
2658 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2660 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2661 (_mm_rcp28_round_ss): Ditto.
2662 (_mm_rsqrt28_round_sd): Ditto.
2663 (_mm_rsqrt28_round_ss): Ditto.
2664 (_mm_rcp28_sd): Ditto.
2665 (_mm_rcp28_ss): Ditto.
2666 (_mm_rsqrt28_sd): Ditto.
2667 (_mm_rsqrt28_ss): Ditto.
2668 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2669 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2670 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2671 (IX86_BUILTIN_RCP28SD): Ditto.
2672 (IX86_BUILTIN_RCP28SS): Ditto.
2673 (IX86_BUILTIN_RSQRT28SD): Ditto.
2674 (IX86_BUILTIN_RSQRT28SS): Ditto.
2675 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2676 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2677 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2678 (ix86_expand_special_args_builtin): Expand new FTYPE.
2679 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2680 (srcp14<mode>): Make insn unary.
2681 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2682 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2683 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2684 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2685 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2686 Fix rounding: make it SAE only.
2687 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2689 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2691 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2692 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2693 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2694 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2695 (round_saeonly_mask_scalar_operand4): Ditto.
2696 (round_saeonly_mask_scalar_op3): Ditto.
2697 (round_saeonly_mask_scalar_op4): Ditto.
2699 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2701 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2702 Implement -maltivec=be for vec_insert and vec_extract.
2704 2014-01-10 DJ Delorie <dj@redhat.com>
2706 * config/msp430/msp430.md (call_internal): Don't allow memory
2707 references with SP as the base register.
2708 (call_value_internal): Likewise.
2709 * config/msp430/constraints.md (Yc): New. For memory references
2710 that don't use SP as a base register.
2712 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2713 "an integer without a # prefix"
2714 * config/msp430/msp430.md (epilogue_helper): Use it.
2716 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2719 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2720 AVX512F gather builtins.
2721 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2722 on gather decls with INTEGER_TYPE masktype.
2723 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2724 directly into the builtin rather than hoisting it before loop.
2726 PR tree-optimization/59387
2727 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2728 (scev_const_prop): If folded_casts and type has undefined overflow,
2729 use force_gimple_operand instead of force_gimple_operand_gsi and
2730 for each added stmt if it is assign with
2731 arith_code_with_undefined_signed_overflow, call
2732 rewrite_to_defined_overflow.
2733 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2734 gimple-fold.h instead.
2735 (arith_code_with_undefined_signed_overflow,
2736 rewrite_to_defined_overflow): Moved to ...
2737 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2738 rewrite_to_defined_overflow): ... here. No longer static.
2739 Include gimplify-me.h.
2740 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2741 rewrite_to_defined_overflow): New prototypes.
2743 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2745 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2747 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2749 * builtins.c (get_object_alignment_2): Minor tweak.
2750 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2752 2014-01-13 Christian Bruel <christian.bruel@st.com>
2754 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2755 optimized non constant lengths.
2757 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2760 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2761 load as __atomic_load_N if possible.
2763 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2765 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2767 (rs6000_expand_builtin): Adjust call.
2769 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2772 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2773 * config/rs6000/rs6000.c: Include target-globals.h.
2774 (rs6000_set_current_function): Instead of doing target_reinit
2775 unconditionally, use save_target_globals_default_opts and
2776 restore_target_globals.
2778 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2780 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2781 (rs6000_expand_builtin): Handle mffs and mtfsf.
2782 (rs6000_init_builtins): Define mffs and mtfsf.
2783 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2784 (rs6000_mffs): New pattern.
2785 (rs6000_mtfsf): New pattern.
2787 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2789 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2790 Start narrowing with START. Apply candidate-use pair
2791 and check overall cost in narrowing.
2792 (iv_ca_prune): Pass new argument.
2794 2014-01-10 Jeff Law <law@redhat.com>
2797 * ree.c (combine_reaching_defs): Ensure the defining statement
2798 occurs before the extension when optimizing extensions with
2799 different source and destination hard registers.
2801 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2804 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2805 vtables into the type inheritance graph.
2807 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2809 PR rtl-optimization/59754
2810 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2811 modes in the REGNO != REGNO case.
2813 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2815 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2817 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2819 PR tree-optimization/59745
2820 * tree-predcom.c (tree_predictive_commoning_loop): Call
2821 free_affine_expand_cache if giving up because components is NULL.
2823 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2824 GC in payload of target_globals struct instead of allocating them on
2825 the heap and the larger structs separately using GC.
2826 * target-globals.h (struct target_globals): Make regs, hard_regs,
2827 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2828 of GTY((skip)) and change type to void *.
2829 (reset_target_globals): Cast loads from those fields to corresponding
2832 2014-01-10 Steve Ellcey <sellcey@mips.com>
2835 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2836 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2837 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2839 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2842 * aarch64-modes.def (CC_Zmode): New flags mode.
2843 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2844 represents an equality.
2845 (aarch64_get_condition_code): Handle CC_Zmode.
2846 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2848 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2850 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2851 extraction in good case.
2853 2014-01-10 Richard Biener <rguenther@suse.de>
2855 PR tree-optimization/59374
2856 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2857 checking after SLP discovery. Mark stmts not participating
2858 in any SLP instance properly.
2860 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2862 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2863 when handling a SET rtx.
2865 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2867 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2868 (cortex-a57): Likewise.
2869 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2871 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2873 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2874 non-iwmmxt builtins.
2876 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2880 * ipa-devirt.c record_target_from_binfo): Take as argument
2881 stack of binfos and lookup matching one for virtual inheritance.
2882 (possible_polymorphic_call_targets_1): Update.
2884 2014-01-10 Huacai Chen <chenhc@lemote.com>
2886 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2887 kernel strings for Loongson-2E/2F/3A.
2889 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2892 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2893 is_gimple_call before calling gimple_call_internal_p.
2895 2014-01-09 Steve Ellcey <sellcey@mips.com>
2897 * Makefile.in (TREE_FLOW_H): Remove.
2898 (TREE_SSA_H): Add file names from tree-flow.h.
2899 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2900 * tree.h: Remove tree-flow.h reference.
2901 * hash-table.h: Remove tree-flow.h reference.
2902 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2903 reference with tree-ssa-loop.h.
2905 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2907 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2908 default element-order behavior for -maltivec.
2909 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2910 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2911 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2912 when targeting big endian, at least for now.
2913 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2915 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2918 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2919 var satisfies use_register_for_decl, just take into account type
2920 alignment, rather than decl alignment.
2922 PR tree-optimization/59622
2923 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2924 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2925 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2926 Don't devirtualize for inplace at all. For targets.length () == 1,
2927 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2929 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2931 * config/i386/i386.md (cpu): Remove the unused btver1.
2933 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2935 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2937 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2940 * tree-core.h (struct target_globals): New forward declaration.
2941 (struct tree_target_option): Add globals field.
2942 * tree.h (TREE_TARGET_GLOBALS): Define.
2943 (prepare_target_option_nodes_for_pch): New prototype.
2944 * target-globals.h (struct target_globals): Define even if
2946 * tree.c (prepare_target_option_node_for_pch,
2947 prepare_target_option_nodes_for_pch): New functions.
2948 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2949 * config/i386/i386.c: Include target-globals.h.
2950 (ix86_set_current_function): Instead of doing target_reinit
2951 unconditionally, use save_target_globals_default_opts and
2952 restore_target_globals.
2954 2014-01-09 Richard Biener <rguenther@suse.de>
2956 PR tree-optimization/59715
2957 * tree-cfg.h (split_critical_edges): Declare.
2958 * tree-cfg.c (split_critical_edges): Export.
2959 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2961 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2963 * cfgexpand.c (expand_stack_vars): Optionally disable
2964 asan stack protection.
2965 (expand_used_vars): Likewise.
2966 (partition_stack_vars): Likewise.
2967 * asan.c (asan_emit_stack_protection): Optionally disable
2968 after return stack usage.
2969 (instrument_derefs): Optionally disable memory access instrumentation.
2970 (instrument_builtin_call): Likewise.
2971 (instrument_strlen_call): Likewise.
2972 (asan_protect_global): Optionally disable global variables protection.
2973 * doc/invoke.texi: Added doc for new options.
2974 * params.def: Added new options.
2975 * params.h: Likewise.
2977 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2979 PR rtl-optimization/59724
2980 * ifcvt.c (cond_exec_process_if_block): Don't call
2981 flow_find_head_matching_sequence with 0 longest_match.
2982 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2983 non-active insns if !stop_after.
2984 (try_head_merge_bb): Revert 2014-01-07 changes.
2986 2014-01-08 Jeff Law <law@redhat.com>
2988 * ree.c (get_sub_rtx): New function, extracted from...
2989 (merge_def_and_ext): Here.
2990 (combine_reaching_defs): Use get_sub_rtx.
2992 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2994 * cgraph.h (varpool_variable_node): Do not choke on null node.
2996 2014-01-08 Catherine Moore <clm@codesourcery.com>
2998 * config/mips/mips.md (simple_return): Attempt to use JRC
3000 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3002 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3004 PR rtl-optimization/59137
3005 * reorg.c (steal_delay_list_from_target): Call update_block for
3007 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3009 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3011 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3012 two duplicate entries.
3014 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3017 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3019 * config/mips/mips.c (mips_truncated_op_cost): New function.
3020 (mips_rtx_costs): Adjust test for BADDU.
3021 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3023 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3025 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3026 (*baddu_si): ...this new pattern.
3028 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3031 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3033 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3036 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3038 (expand_expr, expand_normal): Adjust.
3039 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3040 inner_reference_p. Use inner_reference_p to expand inner references.
3041 (store_expr): Adjust.
3042 * cfgexpand.c (expand_call_stmt): Adjust.
3044 2014-01-08 Rong Xu <xur@google.com>
3046 * gcov-io.c (gcov_var): Move from gcov-io.h.
3047 (gcov_position): Ditto.
3048 (gcov_is_error): Ditto.
3049 (gcov_rewrite): Ditto.
3050 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3051 only part to libgcc/libgcov.h.
3053 2014-01-08 Marek Polacek <polacek@redhat.com>
3056 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3058 2014-01-08 Marek Polacek <polacek@redhat.com>
3061 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3063 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3065 PR rtl-optimization/59649
3066 * stor-layout.c (get_mode_bounds): For BImode return
3067 0 and STORE_FLAG_VALUE.
3069 2014-01-08 Richard Biener <rguenther@suse.de>
3072 * gimple.h (is_gimple_builtin_call): Remove.
3073 (gimple_builtin_call_types_compatible_p): New.
3074 (gimple_call_builtin_p): New overload.
3075 * gimple.c (is_gimple_builtin_call): Remove.
3076 (validate_call): Rename to ...
3077 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3079 (validate_type): New static function.
3080 (gimple_call_builtin_p): New overload and adjust.
3081 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3082 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3083 (gimple_fold_stmt_to_constant_1): Likewise.
3084 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3086 2014-01-08 Richard Biener <rguenther@suse.de>
3089 * gimplify.c (gimplify_expr): Gimplify register-register type
3090 VIEW_CONVERT_EXPRs to separate stmts.
3092 2014-01-07 Jeff Law <law@redhat.com>
3095 * ree.c (combine_set_extension): Handle case where source
3096 and destination registers in an extension insn are different.
3097 (combine_reaching_defs): Allow source and destination registers
3098 in extension to be different under limited circumstances.
3099 (add_removable_extension): Remove restriction that the
3100 source and destination registers in the extension are the same.
3101 (find_and_remove_re): Emit a copy from the extension's
3102 destination to its source after the defining insn if
3103 the source and destination registers are different.
3106 * ifcvt.c (merge_if_block): If we are merging a block with more than
3107 one successor with a block with no successors, remove any BARRIER
3108 after the second block.
3110 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3112 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3114 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3117 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3118 for 14-bit register offsets when INT14_OK_STRICT is false.
3120 2014-01-07 Roland Stigge <stigge@antcom.de>
3121 Michael Meissner <meissner@linux.vnet.ibm.com>
3124 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3125 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3127 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3129 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3132 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3134 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3135 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3136 rtx is const0_rtx or not.
3138 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3141 * target-globals.c (save_target_globals): Remove this_fn_optab
3143 * toplev.c: Include optabs.h.
3144 (target_reinit): Temporarily restore the global options if another
3145 set of options are in force.
3147 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3149 PR rtl-optimization/58668
3150 * cfgcleanup.c (flow_find_cross_jump): Don't count
3151 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3152 to determine what is counted.
3153 (flow_find_head_matching_sequence): Use active_insn_p to determine
3155 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3157 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3158 determine what is counted.
3160 PR tree-optimization/59643
3161 * tree-predcom.c (split_data_refs_to_components): If one dr is
3162 read and one write, determine_offset fails and the write isn't
3163 in the bad component, just put the read into the bad component.
3165 2014-01-07 Mike Stump <mikestump@comcast.net>
3166 Jakub Jelinek <jakub@redhat.com>
3169 * tree-core.h (struct tree_optimization_option): Change optabs
3170 type from unsigned char * to void *.
3171 * optabs.c (init_tree_optimization_optabs): Adjust
3172 TREE_OPTIMIZATION_OPTABS initialization.
3174 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3177 * config/i386/i386.h (struct machine_function): Add
3178 no_drap_save_restore field.
3179 * config/i386/i386.c (ix86_save_reg): Use
3180 !cfun->machine->no_drap_save_restore instead of
3181 crtl->stack_realign_needed.
3182 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3183 this function clears frame_pointer_needed. Set
3184 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3185 and DRAP reg is needed.
3187 2014-01-06 Marek Polacek <polacek@redhat.com>
3190 * doc/implement-c.texi: Mention that other integer types are
3191 permitted as bit-field types in strictly conforming mode.
3193 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3195 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3198 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3200 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3202 2014-01-06 Martin Jambor <mjambor@suse.cz>
3205 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3207 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3209 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3213 * var-tracking.c (add_stores): Preserve the value of the source even if
3214 we don't record the store.
3216 2014-01-06 Terry Guo <terry.guo@arm.com>
3218 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3220 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3223 * config/darwin.c (darwin_function_section): Adjust return values to
3224 correspond to optimisation changes made in r206070.
3226 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3228 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3229 from prefetch_block tune setting.
3230 (nocona_cost): Correct size of prefetch block to 64.
3232 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3234 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3235 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3236 used to save the static chain register in the computation of the offset
3237 from which the FP registers need to be restored.
3239 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3241 PR tree-optimization/59519
3242 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3243 ICE if get_current_def (current_new_name) is already non-NULL, as long
3244 as it is a phi result of some other phi in *new_exit_bb that has
3247 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3248 or vmovdqu* for misaligned_operand.
3249 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3250 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3251 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3252 aligned_mem for AVX512F masked aligned load and store builtins and for
3255 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3257 PR tree-optimization/59651
3258 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3259 Address range for negative step should be added by TYPE_SIZE_UNIT.
3261 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3263 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3264 overlapping registers also for registers other than the stack pointer.
3266 2014-01-03 Marek Polacek <polacek@redhat.com>
3269 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3272 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3275 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3278 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3279 (push splitter): Use <P:MODE_SIZE> instead of
3280 GET_MODE_SIZE (<P:MODE>mode).
3281 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3282 (mov -1, reg peephole2): Likewise.
3283 * config/i386/sse.md (*mov<mode>_internal,
3284 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3285 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3286 *<code><mode>3, *andnot<mode>3<mask_name>,
3287 <mask_codefor><code><mode>3<mask_name>): Likewise.
3288 * config/i386/subst.md (mask_mode512bit_condition,
3289 sd_mask_mode512bit_condition): Likewise.
3291 2014-01-02 Xinliang David Li <davidxl@google.com>
3293 PR tree-optimization/59303
3294 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3295 (dump_predicates): Better output format.
3296 (pred_equal_p): New function.
3297 (is_neq_relop_p): Ditto.
3298 (is_neq_zero_form_p): Ditto.
3299 (pred_expr_equal_p): Ditto.
3300 (pred_neg_p): Ditto.
3301 (simplify_pred): Ditto.
3302 (simplify_preds_2): Ditto.
3303 (simplify_preds_3): Ditto.
3304 (simplify_preds_4): Ditto.
3305 (simplify_preds): Ditto.
3307 (push_to_worklist): Ditto.
3308 (get_pred_info_from_cmp): Ditto.
3309 (is_degenerated_phi): Ditto.
3310 (normalize_one_pred_1): Ditto.
3311 (normalize_one_pred): Ditto.
3312 (normalize_one_pred_chain): Ditto.
3313 (normalize_preds): Ditto.
3314 (normalize_cond_1): Remove function.
3315 (normalize_cond): Ditto.
3316 (is_gcond_subset_of): Ditto.
3317 (is_subset_of_any): Ditto.
3318 (is_or_set_subset_of): Ditto.
3319 (is_and_set_subset_of): Ditto.
3320 (is_norm_cond_subset_of): Ditto.
3321 (pred_chain_length_cmp): Ditto.
3322 (convert_control_dep_chain_into_preds): Type change.
3323 (find_predicates): Ditto.
3324 (find_def_preds): Ditto.
3325 (destroy_predicates_vecs): Ditto.
3326 (find_matching_predicates_in_rest_chains): Ditto.
3327 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3328 (is_pred_expr_subset): Ditto.
3329 (is_pred_chain_subset_of): Ditto.
3330 (is_included_in): Ditto.
3331 (is_superset_of): Ditto.
3333 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3335 Update copyright years.
3337 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3339 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3340 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3341 config/arc/arc.md, config/arc/arc.opt,
3342 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3343 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3344 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3345 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3346 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3347 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3348 vtable-verify.c, vtable-verify.h: Use the standard form for the
3351 2014-01-02 Tobias Burnus <burnus@net-b.de>
3353 * gcc.c (process_command): Update copyright notice dates.
3354 * gcov-dump.c: Ditto.
3356 * doc/cpp.texi: Bump @copying's copyright year.
3357 * doc/cppinternals.texi: Ditto.
3358 * doc/gcc.texi: Ditto.
3359 * doc/gccint.texi: Ditto.
3360 * doc/gcov.texi: Ditto.
3361 * doc/install.texi: Ditto.
3362 * doc/invoke.texi: Ditto.
3364 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3366 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3368 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3370 * config/i386/sse.md (*mov<mode>_internal): Guard
3371 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3373 PR rtl-optimization/59647
3374 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3375 new_rtx into UNSIGNED_FLOAT rtxes.
3377 Copyright (C) 2014 Free Software Foundation, Inc.
3379 Copying and distribution of this file, with or without modification,
3380 are permitted in any medium without royalty provided the copyright
3381 notice and this notice are preserved.