1 2014-03-05 Jakub Jelinek <jakub@redhat.com>
4 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
5 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
8 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
9 (TM_H): Add x86-tune.def.
11 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13 * config/aarch64/aarch64.c (generic_tunings):
14 Use cortexa57_extra_costs.
16 2014-03-05 Jakub Jelinek <jakub@redhat.com>
19 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
20 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
21 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
24 2014-03-04 Heiher <r@hev.cc>
26 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
27 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
29 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
31 * config/i386/predicates.md (const2356_operand): Change to ...
32 (const2367_operand): ... this.
33 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
35 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
36 (*avx512pf_scatterpf<mode>sf): Ditto.
37 (avx512pf_scatterpf<mode>df): Ditto.
38 (*avx512pf_scatterpf<mode>df_mask): Ditto.
39 (*avx512pf_scatterpf<mode>df): Ditto.
40 * config/i386/i386.c (ix86_expand_builtin): Update
41 incorrect hint operand error message.
43 2014-03-04 Richard Biener <rguenther@suse.de>
45 * lto-section-in.c (lto_get_section_data): Fix const cast.
47 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
49 * tree-streamer.c (record_common_node): Assert we don't record
50 nodes with type double.
51 (preload_common_node): Skip type double, complex double and
52 double pointer since it is now frontend dependent due to
55 2014-03-04 Richard Biener <rguenther@suse.de>
58 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
59 (lto_input_toplevel_asms): Likewise.
60 * lto-section-in.c (lto_get_section_data): Instead do it here
63 2014-03-04 Richard Biener <rguenther@suse.de>
65 PR tree-optimization/60382
66 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
67 dead PHIs a reduction.
69 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
71 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
73 (_mm_prefetch): Move out of GCC target("sse") pragma.
74 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
75 GCC target("prfchw") pragma.
76 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
78 * config/i386/i386.c (ix86_option_override_internal): Enable
79 -mprfchw with -mprefetchwt1.
81 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
83 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
86 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
88 * opts.h (CL_PCH_IGNORE): Define.
89 * targhooks.c (option_affects_pch_p):
90 Return false for options that have CL_PCH_IGNORE set.
91 * opt-functions.awk: Process PchIgnore.
92 * doc/options.texi: Document PchIgnore.
94 * config/arc/arc.opt (misize): Add PchIgnore property.
96 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
98 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
99 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
100 constraint on constants to permit them being loaded into
101 GENERAL_REGS or BASE_REGS.
103 2014-03-03 Nick Clifton <nickc@redhat.com>
105 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
106 anti-cacnonical alternatives.
107 (negandhi3_real): New pattern.
108 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
110 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
112 * config/avr/avr-mcus.def: Remove atxmega16x1.
113 * config/avr/avr-tables.opt: Regenerate.
114 * config/avr/t-multilib: Regenerate.
115 * doc/avr-mmcu.texi: Regenerate.
117 2014-03-03 Tobias Grosser <tobias@grosser.es>
118 Mircea Namolaru <mircea.namolaru@inria.fr>
120 PR tree-optimization/58028
121 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
124 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
126 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
127 not handled by recognizers.
129 2014-03-03 Jakub Jelinek <jakub@redhat.com>
132 * function.c (expand_function_end): Don't emit
133 clobber_return_register sequence if clobber_after is a BARRIER.
134 * cfgexpand.c (construct_exit_block): Append instructions before
135 return_label to prev_bb.
137 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
139 * config/rs6000/constraints.md: Document reserved use of "wc".
141 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
144 * ipa.c (function_and_variable_visibility): When dissolving comdat
145 group, also set all symbols to local.
147 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
152 2013-12-14 Jan Hubicka <jh@suse.cz>
154 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
156 2014-03-02 Jon Beniston <jon@beniston.com>
162 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
163 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
164 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
165 (simple_return, *simple_return): New patterns
166 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
167 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
169 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
171 * dwarf2out.c (gen_subprogram_die): Tidy.
173 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
176 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
177 (*mov_t_msb_neg_negc): ... this new insn.
179 2014-02-28 Jason Merrill <jason@redhat.com>
182 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
185 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
188 * dwarf2out.c (decltype_auto_die): New static.
189 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
190 (gen_type_die_with_usage): Handle 'decltype(auto)'.
191 (is_cxx_auto): Likewise.
193 2014-02-28 Ian Bolton <ian.bolton@arm.com>
195 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
196 we are not using general regs only.
198 2014-02-28 Richard Biener <rguenther@suse.de>
201 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
202 previous fix and only allow to remove trivial pre-headers
203 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
204 (remove_forwarder_block): Properly update the latch of a loop.
206 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
209 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
210 (cselib_preserved_hash_table): New.
211 (preserve_constants_and_equivs): Move preserved vals to it.
212 (cselib_find_slot): Look it up first.
213 (cselib_init): Initialize it.
214 (cselib_finish): Release it.
215 (dump_cselib_table): Dump it.
217 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
220 * cselib.c (remove_useless_values): Skip to avoid quadratic
221 behavior if the condition moved from...
222 (cselib_process_insn): ... here holds.
224 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
227 * var-tracking.c (vt_initialize): Apply the same condition to
228 preserve the CFA base value.
230 2014-02-28 Joey Ye <joey.ye@arm.com>
233 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
234 if reload in progress or completed.
236 2014-02-28 Tobias Burnus <burnus@net-b.de>
239 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
242 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
244 * doc/tm.texi.in (Condition Code Status): Update documention for
245 relative locations of cc0-setter and cc0-user.
247 2014-02-27 Jeff Law <law@redhat.com>
249 PR rtl-optimization/52714
250 * combine.c (try_combine): When splitting an unrecognized PARALLEL
251 into two independent simple sets, if I3 is a jump, ensure the
252 pattern we place into I3 is a (set (pc) ...).
254 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
255 Jeff Law <law@redhat.com>
257 PR rtl-optimization/49847
258 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
259 are in different blocks.
260 * doc/tm.texi (Condition Code Status): Update documention for
261 relative locations of cc0-setter and cc0-user.
263 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
266 * lra.c (lra_emit_add): Check SUBREG too.
268 2014-02-27 Andreas Schwab <schwab@suse.de>
270 * config/m68k/m68k.c (m68k_option_override): Disable
271 -flive-range-shrinkage for classic m68k.
272 (m68k_override_options_after_change): Likewise.
274 2014-02-27 Marek Polacek <polacek@redhat.com>
277 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
278 -Wmaybe-uninitialized.
280 2014-02-27 Alan Modra <amodra@gmail.com>
283 * reload1.c (emit_input_reload_insns): When reload_override_in,
284 set old to rl->in_reg when rl->in_reg is a subreg.
286 2014-02-26 Richard Biener <rguenther@suse.de>
289 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
291 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
293 * common/config/i386/predicates.md (const1256_operand): Remove.
294 (const2356_operand): New.
295 (const_1_to_2_operand): Remove.
296 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
297 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
298 (*avx512pf_gatherpf<mode>sf): Ditto.
299 (avx512pf_gatherpf<mode>df): Ditto.
300 (*avx512pf_gatherpf<mode>df_mask): Ditto.
301 (*avx512pf_gatherpf<mode>df): Ditto.
302 (avx512pf_scatterpf<mode>sf): Ditto.
303 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
304 (*avx512pf_scatterpf<mode>sf): Ditto.
305 (avx512pf_scatterpf<mode>df): Ditto.
306 (*avx512pf_scatterpf<mode>df_mask): Ditto.
307 (*avx512pf_scatterpf<mode>df): Ditto.
308 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
310 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
312 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
313 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
314 (_mm512_mask_testn_epi64_mask): Move to ...
315 * config/i386/avx512cdintrin.h: Here.
316 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
317 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
318 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
319 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
320 TARGET_AVX512F from TARGET_AVX512CD.
322 2014-02-26 Richard Biener <rguenther@suse.de>
325 * ipa.c (walk_polymorphic_call_targets): Properly guard
326 call to inline_update_overall_summary.
328 2014-02-26 Bin Cheng <bin.cheng@arm.com>
331 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
332 and latches only if requested. Fix latch if it is removed.
333 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
334 LOOPS_HAVE_PREHEADERS.
336 2014-02-25 Andrew Pinski <apinski@cavium.com>
338 * builtins.c (expand_builtin_thread_pointer): Create a new target
339 when the target is NULL.
341 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
343 PR rtl-optimization/60317
344 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
345 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
346 * lra-assigns.c: Include params.h.
347 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
348 other reload pseudos considerations.
350 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
352 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
353 to use canonical form for nor<mode>3.
355 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
358 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
361 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
363 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
364 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
365 (ix86_handle_option): Handle OPT_mprefetchwt1.
366 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
367 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
369 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
370 OPTION_MASK_ISA_PREFETCHWT1.
371 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
372 (PTA_PREFETCHWT1): New.
373 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
374 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
375 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
376 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
377 (*prefetch_avx512pf_<mode>_: Change into ...
378 (*prefetch_prefetchwt1_<mode>: This.
379 * config/i386/i386.opt (mprefetchwt1): New.
380 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
381 (_mm_prefetch): Handle intent to write.
382 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
384 2014-02-25 Richard Biener <rguenther@suse.de>
387 * emit-rtl.c (mem_attrs_htab): Remove.
388 (mem_attrs_htab_hash): Likewise.
389 (mem_attrs_htab_eq): Likewise.
390 (set_mem_attrs): Always allocate new mem-attrs when something changed.
391 (init_emit_once): Do not allocate mem_attrs_htab.
393 2014-02-25 Richard Biener <rguenther@suse.de>
396 * lto-opts.c (lto_write_options): Output non-explicit conservative
397 -fwrapv, -fno-trapv and -fno-strict-overflow.
398 * lto-wrapper.c (merge_and_complain): Handle merging those options.
399 (run_gcc): And pass them through.
401 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
403 * sel-sched.c (calculate_new_fences): New parameter ptime.
404 Calculate it as a maximum over all fence cycles.
405 (sel_sched_region_2): Adjust the call to calculate_new_fences.
406 Print the final schedule timing when sched_verbose.
408 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
410 PR rtl-optimization/60292
411 * sel-sched.c (fill_vec_av_set): Do not reset target availability
412 bit fot the fence instruction.
414 2014-02-24 Alangi Derick <alangiderick@gmail.com>
416 * calls.h: Fix typo in comment.
418 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
420 * config/pa/pa.c (pa_output_move_double): Don't valididate when
421 adjusting offsetable addresses.
423 2014-02-24 Guozhi Wei <carrot@google.com>
425 * sparseset.h (sparseset_pop): Fix the wrong index.
427 2014-02-24 Walter Lee <walt@tilera.com>
429 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
430 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
432 * common/config/tilegx/tilegx-common.c
433 (TARGET_DEFAULT_TARGET_FLAGS): Define.
434 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
436 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
437 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
438 (tilegx_gimplify_va_arg_expr): Handle big endian.
439 (tilegx_expand_unaligned_load): Ditto.
440 (tilegx_expand_unaligned_store): Ditto.
441 (TARGET_RETURN_IN_MSB): New.
442 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
443 (TARGET_ENDIAN_DEFAULT): New.
444 (TARGET_BIG_ENDIAN): Handle big endian.
445 (BYTES_BIG_ENDIAN): Ditto.
446 (WORDS_BIG_ENDIAN): Ditto.
447 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
450 * config/tilegx/tilegx.md (extv): Handle big endian.
453 (insn_st<n>_add<bitsuffix>): Ditto.
454 (insn_stnt<n>): Ditto.
455 (insn_stnt<n>_add<bitsuffix>):Ditto.
456 (vec_interleave_highv8qi): Handle big endian.
457 (vec_interleave_highv8qi_be): New.
458 (vec_interleave_highv8qi_le): New.
459 (insn_v1int_h): Handle big endian.
460 (vec_interleave_lowv8qi): Handle big endian.
461 (vec_interleave_lowv8qi_be): New.
462 (vec_interleave_lowv8qi_le): New.
463 (insn_v1int_l): Handle big endian.
464 (vec_interleave_highv4hi): Handle big endian.
465 (vec_interleave_highv4hi_be): New.
466 (vec_interleave_highv4hi_le): New.
467 (insn_v2int_h): Handle big endian.
468 (vec_interleave_lowv4hi): Handle big endian.
469 (vec_interleave_lowv4hi_be): New.
470 (vec_interleave_lowv4hi_le): New.
471 (insn_v2int_l): Handle big endian.
472 (vec_interleave_highv2si): Handle big endian.
473 (vec_interleave_highv2si_be): New.
474 (vec_interleave_highv2si_le): New.
475 (insn_v4int_h): Handle big endian.
476 (vec_interleave_lowv2si): Handle big endian.
477 (vec_interleave_lowv2si_be): New.
478 (vec_interleave_lowv2si_le): New.
479 (insn_v4int_l): Handle big endian.
480 * config/tilegx/tilegx.opt (mbig-endian): New option.
481 (mlittle-endian): New option.
482 * doc/install.texi: Document tilegxbe-linux.
483 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
485 2014-02-24 Martin Jambor <mjambor@suse.cz>
488 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
489 there are no parameter descriptors.
491 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
493 PR rtl-optimization/60268
494 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
495 initialization to ...
496 (sched_rgn_init): ... here.
497 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
499 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
501 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
504 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
506 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
509 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
511 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
512 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
514 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
516 * config/microblaze/predicates.md: Add cmp_op predicate.
517 * config/microblaze/microblaze.md: Add branch_compare instruction
518 which uses cmp_op predicate and emits cmp insn before branch.
519 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
520 to microblaze_expand_conditional_branch and consolidate logic.
521 (microblaze_expand_conditional_branch): emit branch_compare
522 insn instead of handling cmp op separate from branch insn.
524 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
526 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
529 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
531 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
532 define_insn with define_expand and new define_insn
533 *altivec_lve<VI_char>x_internal.
534 (altivec_stve<VI_char>x): Replace define_insn with define_expand
535 and new define_insn *altivec_stve<VI_char>x_internal.
536 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
538 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
540 (altivec_expand_stvex_be): New function.
542 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
544 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
545 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
546 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
547 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
549 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
552 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
553 instead of emit_move_insn.
555 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
557 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
559 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
562 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
564 * config/rs6000/altivec.md (altivec_lvxl): Rename as
565 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
566 (altivec_lvxl_<mode>): New define_expand incorporating
567 -maltivec=be semantics where needed.
568 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
569 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
570 semantics where needed.
571 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
572 (altivec_stvx_<mode>): New define_expand incorporating
573 -maltivec=be semantics where needed.
574 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
575 VM2 iterator instead of V4SI.
576 (altivec_stvxl_<mode>): New define_expand incorporating
577 -maltivec=be semantics where needed.
578 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
579 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
580 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
581 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
582 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
583 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
584 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
585 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
586 ALTIVEC_BUILTIN_STVXL.
587 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
588 (altivec_expand_stvx_be): Likewise.
589 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
590 (altivec_expand_lvx_be): Likewise.
591 (altivec_expand_stvx_be): Likewise.
592 (altivec_expand_builtin): Add cases for
593 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
594 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
595 (altivec_init_builtins): Add definitions for
596 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
597 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
599 2014-02-21 Catherine Moore <clm@codesourcery.com>
601 * doc/invoke.texi (mvirt, mno-virt): Document.
602 * config/mips/mips.opt (mvirt): New option.
603 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
605 2014-02-21 Richard Biener <rguenther@suse.de>
607 PR tree-optimization/60276
608 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
609 (STMT_VINFO_MIN_NEG_DIST): New macro.
610 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
611 STMT_VINFO_MIN_NEG_DIST.
612 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
613 made for negative dependence distances still hold.
615 2014-02-21 Richard Biener <rguenther@suse.de>
618 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
619 DECL_INITIAL for globals not in the current function context.
621 2014-02-21 Jakub Jelinek <jakub@redhat.com>
623 PR tree-optimization/56490
624 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
625 * tree-ssa-uninit.c: Include params.h.
626 (compute_control_dep_chain): Add num_calls argument, return false
627 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
628 num_calls to recursive call.
629 (find_predicates): Change dep_chain into normal array,
630 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
631 variable and adjust compute_control_dep_chain caller.
632 (find_def_preds): Likewise.
634 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
636 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
637 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
639 2014-02-21 Nick Clifton <nickc@redhat.com>
641 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
643 (popqi1): Add mode to pre_dec.
646 2014-02-21 Jakub Jelinek <jakub@redhat.com>
648 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
649 mode for mask of V8SFmode permutation.
651 2014-02-20 Richard Henderson <rth@redhat.com>
654 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
655 a new pseudo for OLDVAL.
657 2014-02-20 Jakub Jelinek <jakub@redhat.com>
660 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
661 gen_reg_rtx if d->testing_p.
662 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
663 if d->testing_p and we will certainly return true.
664 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
667 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
669 * emit-rtl.c (gen_reg_rtx): Assert that
670 crtl->emit.regno_pointer_align_length is non-zero.
672 2014-02-20 Richard Henderson <rth@redhat.com>
675 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
676 on failure the store back into EXPECT.
678 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
679 Sandra Loosemore <sandra@codesourcery.com>
681 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
682 * config/nios2/nios2.c (nios2_function_profiler): Add
683 -fPIC (flag_pic == 2) support.
684 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
685 (nios2_large_offset_p): New function.
686 (nios2_unspec_reloc_p): Move up position, update to use
687 nios2_large_offset_p.
688 (nios2_unspec_address): Remove function.
689 (nios2_unspec_offset): New function.
690 (nios2_large_got_address): New function.
691 (nios2_got_address): Add large offset support.
692 (nios2_legitimize_tls_address): Update usage of removed and new
694 (nios2_symbol_binds_local_p): New function.
695 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
696 (nios2_legitimize_address): Update to use nios2_large_offset_p.
697 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
698 (nios2_print_operand): Merge H/L processing, add hiadj/lo
699 processing for (const (unspec ...)).
700 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
702 2014-02-20 Richard Biener <rguenther@suse.de>
704 * tree-cfg.c (replace_uses_by): Mark altered BBs before
705 doing the substitution.
706 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
708 2014-02-20 Martin Jambor <mjambor@suse.cz>
711 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
712 info when checking whether lattices are bottom.
714 2014-02-20 Richard Biener <rguenther@suse.de>
717 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
720 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
723 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
724 parameter specifying the scaling.
725 (inline_call): Update.
726 (want_inline_recursively): Guard division by zero.
727 (recursive_inlining): Update.
728 * ipa-inline.h (clone_inlined_nodes): Update.
730 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
733 * config/i386/i386.c (classify_argument): Pass structures of size
734 64 bytes or less in register.
736 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
737 Kirill Yukhin <kirill.yukhin@intel.com>
739 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
740 (_mm_rcp28_round_ss): Ditto.
741 (_mm_rsqrt28_round_sd): Ditto.
742 (_mm_rsqrt28_round_ss): Ditto.
743 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
744 (_mm_rcp14_round_ss): Ditto.
745 (_mm_rsqrt14_round_sd): Ditto.
746 (_mm_rsqrt14_round_ss): Ditto.
747 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
748 the first input operand, get rid of match_dup.
749 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
751 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
753 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
754 operand as the first input operand, set type attribute.
755 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
757 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
758 operand as the first input operand, set type attribute.
760 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
762 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
765 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
768 * config/i386/i386.c (construct_container): Remove TFmode check
769 for X86_64_INTEGER_CLASS.
771 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
774 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
775 only when -Wpsabi is enabled.
777 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
780 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
781 passing arrays in registers are the same as for structs, so remove the
782 special case for them.
784 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
786 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
787 destination type, extract only the valid bits if the source type is not
788 integral and has a different mode.
790 2014-02-19 Richard Biener <rguenther@suse.de>
793 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
796 2014-02-19 Richard Biener <rguenther@suse.de>
799 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
800 (ipa_modify_call_arguments): Emit an argument load explicitely and
801 preserve virtual SSA form there and for the replacement call.
802 Do not update SSA form nor free dominance info.
804 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
806 * ipa.c (function_and_variable_visibility): Also clear WEAK
807 flag when disolving COMDAT_GROUP.
809 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
811 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
812 * ipa-prop.c (ipa_set_jf_known_type): Return early when
814 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
815 do more sanity checks.
816 (detect_type_change): Return true when giving up early.
817 (compute_complex_assign_jump_func): Fix type parameter of
819 (compute_complex_ancestor_jump_func): Likewise.
820 (update_jump_functions_after_inlining): Fix updating of
822 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
824 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
826 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
827 inline clones when edge disappears.
829 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
832 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
833 Split 64-bit moves into 2 patterns. Do not allow the use of
834 direct move for TDmode in little endian, since the decimal value
835 has little endian bytes within a word, but the 64-bit pieces are
836 ordered in a big endian fashion, and normal subreg's of TDmode are
838 (mov<mode>_64bit_dm): Likewise.
839 (movtd_64bit_nodm): Likewise.
841 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
843 PR tree-optimization/60174
844 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
845 statement of an SSA_NAME that occurs in an abnormal PHI node.
847 2014-02-18 Jakub Jelinek <jakub@redhat.com>
850 * final.c (SEEN_BB): Remove.
851 (SEEN_NOTE, SEEN_EMITTED): Renumber.
852 (final_scan_insn): Don't force_source_line on second
853 NOTE_INSN_BASIC_BLOCK.
855 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
858 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
859 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
860 (type_natural_mode): Warn ABI change when %zmm register is not
861 available for AVX512F vector value passing.
863 2014-02-18 Kai Tietz <ktietz@redhat.com>
866 * config/i386/i386.c (ix86_expand_prologue): Use value in
867 rax register as displacement when restoring %r10 or %rax.
868 Fix wrong offset when restoring both registers.
870 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
872 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
873 assertion with conditional return.
875 2014-02-18 Jakub Jelinek <jakub@redhat.com>
876 Uros Bizjak <ubizjak@gmail.com>
879 * config/i386/driver-i386.c (host_detect_local_cpu): If
880 YMM state is not saved by the OS, also clear has_f16c. Move
881 CPUID 0x80000001 handling before YMM state saving checking.
883 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
885 PR rtl-optimization/58960
886 * haifa-sched.c (alloc_global_sched_pressure_data): New,
887 factored out from ...
888 (sched_init): ... here.
889 (free_global_sched_pressure_data): New, factored out from ...
890 (sched_finish): ... here.
891 * sched-int.h (free_global_sched_pressure_data): Declare.
892 * sched-rgn.c (nr_regions_initial): New static global.
893 (haifa_find_rgns): Initialize it.
894 (schedule_region): Disable sched-pressure for the newly
897 2014-02-17 Richard Biener <rguenther@suse.de>
899 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
900 release SSA defs of pattern stmts.
902 2014-02-17 Richard Biener <rguenther@suse.de>
904 * tree-inline.c (expand_call_inline): Release the virtual
905 operand defined by the call we are about to inline.
907 2014-02-17 Richard Biener <rguenther@suse.de>
909 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
911 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
912 Ilya Tocar <ilya.tocar@intel.com>
914 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
915 arguments order in builtin.
916 (_mm512_permutexvar_epi64): Ditto.
917 (_mm512_mask_permutexvar_epi64): Ditto
918 (_mm512_maskz_permutexvar_epi32): Ditto
919 (_mm512_permutexvar_epi32): Ditto
920 (_mm512_mask_permutexvar_epi32): Ditto
922 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
924 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
925 (p8_vmrgow): Likewise.
927 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
929 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
932 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
935 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
936 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
937 into 64-bit and 32-bit moves. On 64-bit moves, add support for
938 using direct move instructions on ISA 2.07. Also adjust
939 instruction length for 64-bit.
940 (mov<mode>_64bit, TFmode/TDmode): Likewise.
941 (mov<mode>_32bit, TFmode/TDmode): Likewise.
943 2014-02-15 Alan Modra <amodra@gmail.com>
947 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
948 find_replacement on parts of insn rtl that might be reloaded.
950 2014-02-15 Richard Biener <rguenther@suse.de>
952 PR tree-optimization/60183
953 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
954 (tree_ssa_phiprop): Calculate and free post-dominators.
956 2014-02-14 Jeff Law <law@redhat.com>
958 PR rtl-optimization/60131
959 * ree.c (get_extended_src_reg): New function.
960 (combine_reaching_defs): Use it rather than assuming location of REG.
961 (find_and_remove_re): Verify first operand of extension is
962 a REG before adding the insns to the copy list.
964 2014-02-14 Roland McGrath <mcgrathr@google.com>
966 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
967 * configure: Regenerated.
968 * config.in: Regenerated.
969 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
970 instead of ASM_SHORT.
972 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
973 Richard Earnshaw <rearnsha@arm.com>
975 PR rtl-optimization/59535
976 * lra-constraints.c (process_alt_operands): Encourage alternative
977 when unassigned pseudo class is superset of the alternative class.
978 (inherit_reload_reg): Don't inherit when optimizing for code size.
979 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
980 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
981 modes not less than 4 for Thumb1.
983 2014-02-14 Kyle McMartin <kyle@redhat.com>
986 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
988 2014-02-14 Richard Biener <rguenther@suse.de>
990 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
991 (get_frame_arg): Drop the assert with langhook types_compatible_p.
992 Do not strip INDIRECT_REFs.
994 2014-02-14 Richard Biener <rguenther@suse.de>
997 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
998 DECL_FUNCTION_SPECIFIC_TARGET.
999 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1000 * tree-streamer-out.c (pack_ts_target_option): Remove.
1001 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1002 (write_ts_function_decl_tree_pointers): Do not stream
1003 DECL_FUNCTION_SPECIFIC_TARGET.
1004 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1005 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1006 (lto_input_ts_function_decl_tree_pointers): Do not stream
1007 DECL_FUNCTION_SPECIFIC_TARGET.
1009 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1011 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1012 (get_initial_def_for_induction, vectorizable_induction): Ignore
1013 debug stmts when looking for exit_phi.
1014 (vectorizable_live_operation): Fix up condition.
1016 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1018 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1019 nreverse() because it changes the content of original tree list.
1021 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1023 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1024 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1026 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1028 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1029 GNU coding standards.
1031 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1034 * dwarf2out.c (gen_subprogram_die): Don't call
1035 add_calling_convention_attribute if subr_die is old_die.
1037 2014-02-13 Sharad Singhai <singhai@google.com>
1039 * doc/optinfo.texi: Fix order of nodes.
1041 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1043 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1044 operands[2], not operands[3].
1046 2014-02-13 Richard Biener <rguenther@suse.de>
1049 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1050 mention the possibility of an in-tree build.
1051 (CLooG): Update recommended version to 0.18.1, mention the
1052 possibility of an in-tree build and clarify that the ISL
1053 bundled with CLooG does not work.
1055 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1058 * expr.c (compress_float_constant): If x is a hard register,
1059 extend into a pseudo and then move to x.
1061 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1063 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1064 caused by bad second argument to warning_at() with -mhotpatch and
1065 nested functions (e.g. with gfortran).
1067 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1069 * opts.c (option_name): Remove "enabled by default" rider.
1071 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1073 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1075 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1076 Uros Bizjak <ubizjak@gmail.com>
1079 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1080 * configure: Regenerated.
1082 2014-02-12 Richard Biener <rguenther@suse.de>
1084 * vec.c (vec_prefix::calculate_allocation): Move as
1085 inline variant to vec.h.
1086 (vec_prefix::calculate_allocation_1): New out-of-line version.
1087 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1088 (vec_prefix::m_has_auto_buf): Rename to ...
1089 (vec_prefix::m_using_auto_storage): ... this.
1090 (vec_prefix::calculate_allocation): Inline the easy cases
1091 and dispatch to calculate_allocation_1 which doesn't need the
1093 (va_heap::reserve): Use gcc_checking_assert.
1094 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1095 m_using_auto_storage.
1096 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1098 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1099 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1100 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1102 2014-02-12 Richard Biener <rguenther@suse.de>
1104 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1105 when we found a dependence.
1107 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1109 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1111 (maybe_fold_stmt): ... into this new function.
1112 * omp-low.c (lower_omp): Update comment.
1114 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1117 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1120 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1122 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1123 identifiers in comments.
1124 (cortexa53_extra_costs): Likewise.
1125 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1126 (cortexa7_extra_costs): Likewise.
1127 (cortexa12_extra_costs): Likewise.
1128 (cortexa15_extra_costs): Likewise.
1129 (v7m_extra_costs): Likewise.
1131 2014-02-12 Richard Biener <rguenther@suse.de>
1134 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1135 of posix_memalign being successful.
1136 (lower_stmt): Restrict lowering of posix_memalign to when
1137 -ftree-bit-ccp is enabled.
1139 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1141 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1143 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1145 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1147 PR rtl-optimization/60116
1148 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1149 other_insn once the combination has been validated.
1151 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1154 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1156 * ipa-devirt.c: Include demangle.h
1157 (odr_violation_reported): New static variable.
1158 (add_type_duplicate): Update odr_violations.
1159 (maybe_record_node): Add completep parameter; update it.
1160 (record_target_from_binfo): Add COMPLETEP parameter;
1161 update it as needed.
1162 (possible_polymorphic_call_targets_1): Likewise.
1163 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1164 rename FINAL to COMPLETE.
1165 (record_targets_from_bases): Sanity check we found the binfo;
1166 fix COMPLETEP updating.
1167 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1168 parameter, fix computing of COMPLETEP.
1169 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1170 at LTO time do demangling.
1171 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1172 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1174 (gimple_get_virt_method_for_binfo): Likewise.
1175 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1176 gimple_get_virt_method_for_vtable): Update prototypes.
1178 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1181 * genautomata.c (add_presence_absence): Fix typo with
1182 {final_}presence_list.
1184 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1187 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1188 for VSX/Altivec vectors that land in GPR registers.
1190 2014-02-11 Richard Henderson <rth@redhat.com>
1191 Jakub Jelinek <jakub@redhat.com>
1194 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1195 around drhs if type conversion to lacc->type is not useless.
1197 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1199 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1201 (cortex-a57.cortex-a53): Likewise.
1202 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1204 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1206 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1209 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1211 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1212 add_options_for_arm_vfp3.
1214 2014-02-11 Jeff Law <law@redhat.com>
1217 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1218 object with an undesirable mode.
1220 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1223 * config/i386/sol2-9.h: New file.
1224 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1225 *-*-solaris2.9*): Use it.
1227 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1229 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1230 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1232 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1234 * config/microblaze/microblaze.c: Extend mcpu version format
1236 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1238 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1240 2014-02-10 Richard Henderson <rth@redhat.com>
1243 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1244 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1245 ms-abi vs -mno-accumulate-outgoing-args.
1246 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1247 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1250 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1253 * cfgexpand.c (expand_asm_operands): Attach source location to
1254 ASM_INPUT rtx objects.
1255 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1257 2014-02-10 Nick Clifton <nickc@redhat.com>
1259 * config/mn10300/mn10300.c (popcount): New function.
1260 (mn10300_expand_prologue): Include saved registers in stack usage
1263 2014-02-10 Jeff Law <law@redhat.com>
1266 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1267 when changing the SET_DEST of a prior insn to avoid an input reload.
1269 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1271 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1272 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1273 -mcall-openbsd, or -mcall-linux.
1274 (CC1_ENDIAN_BIG_SPEC): Remove.
1275 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1276 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1277 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1278 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1279 and %cc1_endian_default.
1280 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1282 2014-02-10 Richard Biener <rguenther@suse.de>
1284 PR tree-optimization/60115
1285 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1286 MEM_REF handling. Properly verify that the accesses are not
1287 out of the objects bound.
1289 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1291 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1294 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1296 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1297 proper constants and fix formatting.
1298 (possible_polymorphic_call_targets): Fix formatting.
1300 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1301 Ilya Tocar <ilya.tocar@intel.com>
1303 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1304 (_mm512_loadu_epi32): Renamed into...
1305 (_mm512_loadu_si512): This.
1306 (_mm512_storeu_epi32): Renamed into...
1307 (_mm512_storeu_si512): This.
1308 (_mm512_maskz_ceil_ps): Removed.
1309 (_mm512_maskz_ceil_pd): Ditto.
1310 (_mm512_maskz_floor_ps): Ditto.
1311 (_mm512_maskz_floor_pd): Ditto.
1312 (_mm512_floor_round_ps): Ditto.
1313 (_mm512_floor_round_pd): Ditto.
1314 (_mm512_ceil_round_ps): Ditto.
1315 (_mm512_ceil_round_pd): Ditto.
1316 (_mm512_mask_floor_round_ps): Ditto.
1317 (_mm512_mask_floor_round_pd): Ditto.
1318 (_mm512_mask_ceil_round_ps): Ditto.
1319 (_mm512_mask_ceil_round_pd): Ditto.
1320 (_mm512_maskz_floor_round_ps): Ditto.
1321 (_mm512_maskz_floor_round_pd): Ditto.
1322 (_mm512_maskz_ceil_round_ps): Ditto.
1323 (_mm512_maskz_ceil_round_pd): Ditto.
1324 (_mm512_expand_pd): Ditto.
1325 (_mm512_expand_ps): Ditto.
1326 * config/i386/i386.c (ix86_builtins): Remove
1327 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1328 (bdesc_args): Ditto.
1329 * config/i386/predicates.md (const1256_operand): New.
1330 (const_1_to_2_operand): Ditto.
1331 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1332 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1333 (*avx512pf_gatherpf<mode>sf): Ditto.
1334 (avx512pf_gatherpf<mode>df): Ditto.
1335 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1336 (*avx512pf_gatherpf<mode>df): Ditto.
1337 (avx512pf_scatterpf<mode>sf): Ditto.
1338 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1339 (*avx512pf_scatterpf<mode>sf): Ditto.
1340 (avx512pf_scatterpf<mode>df): Ditto.
1341 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1342 (*avx512pf_scatterpf<mode>df): Ditto.
1343 (avx512f_expand<mode>): Removed.
1344 (<shift_insn><mode>3<mask_name>): Change predicate type.
1346 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1348 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1349 not at the end of datarefs vector use ordered_remove to avoid
1350 reordering datarefs vector.
1353 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1354 mark local addressable non-static vars as GOVD_PRIVATE
1355 instead of GOVD_LOCAL.
1356 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1357 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1361 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1362 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1363 assume_aligned or alloc_align attributes.
1364 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1365 arguments. Handle also assume_aligned and alloc_align attributes.
1366 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1367 calls to functions with assume_aligned or alloc_align attributes.
1368 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1370 2014-02-08 Terry Guo <terry.guo@arm.com>
1372 * doc/invoke.texi: Document ARM -march=armv7e-m.
1374 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1376 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1377 flag on __cilkrts_rethrow builtin.
1380 * ipa-cp.c (determine_versionability): Fail at -O0
1381 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1382 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1385 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1388 * tree-inline.c (copy_forbidden): Fail for
1389 __attribute__((optimize (0))) functions.
1391 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1393 * varpool.c: Include pointer-set.h.
1394 (varpool_remove_unreferenced_decls): Variables in other partitions
1395 will not be output; be however careful to not lose information
1398 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1400 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1401 lookup in the vtable constructor.
1403 2014-02-07 Jeff Law <law@redhat.com>
1406 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1407 define_insn_and_split.
1409 * ipa-inline.c (inline_small_functions): Fix typos.
1411 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1413 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1414 (s390_can_use_return_insn): Declare.
1415 * config/s390/s390.h (EPILOGUE_USES): Define.
1416 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1418 (s390_chunkify_start): Handle return JUMP_LABELs.
1419 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1420 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1421 (s390_can_use_return_insn): New functions.
1422 (s390_fix_long_loop_prediction): Handle conditional returns.
1423 (TARGET_SET_UP_BY_PROLOGUE): Define.
1424 * config/s390/s390.md (ANY_RETURN): New code iterator.
1425 (*creturn, *csimple_return, return, simple_return): New patterns.
1427 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1429 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1430 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1431 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1432 REG_CFA_RESTORE list when deciding not to restore a register.
1434 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1436 * config/s390/s390.c: Include tree-pass.h and context.h.
1437 (s390_early_mach): New function, split out from...
1438 (s390_emit_prologue): ...here.
1439 (pass_data_s390_early_mach): New pass structure.
1440 (pass_s390_early_mach): New class.
1441 (s390_option_override): Create and register early_mach pass.
1442 Move to end of file.
1444 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1446 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1447 to match for the exit block.
1449 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1451 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1452 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1453 Reject misaligned operands.
1455 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1457 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1459 2014-02-07 Richard Biener <rguenther@suse.de>
1462 * gimple-low.c (lower_builtin_posix_memalign): New function.
1463 (lower_stmt): Call it to lower posix_memalign in a way
1464 to make alignment info accessible.
1466 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1469 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1470 __builtin_setjmp_receiver.
1472 2014-02-07 Richard Biener <rguenther@suse.de>
1475 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1476 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1477 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1478 Handle BUILT_IN_POSIX_MEMALIGN.
1479 (find_func_clobbers): Likewise.
1480 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1481 (call_may_clobber_ref_p_1): Likewise.
1483 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1486 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1489 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1492 * lto-cgraph.c (lto_output_node): Use
1493 symtab_get_symbol_partitioning_class.
1494 (lto_output_varpool_node): likewise.
1495 (symtab_get_symbol_partitioning_class): Move here from
1497 * cgraph.h (symbol_partitioning_class): Likewise.
1498 (symtab_get_symbol_partitioning_class): Declare.
1500 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1502 * ggc.h (ggc_internal_cleared_alloc): New macro.
1503 * vec.h (vec_safe_copy): Handle memory stats.
1504 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1505 * target-globals.c (save_target_globals): Likewise.
1507 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1510 * expr.c (emit_move_resolve_push): Export; be bit more selective
1511 on when to clear alias set.
1512 * expr.h (emit_move_resolve_push): Declare.
1513 * function.h (struct function): Add tail_call_marked.
1514 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1515 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1516 * config/i386/i386.md (TImode move expander): De not call
1518 (FP push expanders): Preserve memory attributes.
1519 * config/i386/sse.md (push<mode>1): Remove.
1520 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1521 (ix86_expand_push): Remove.
1522 * config/i386/mmx.md (push<mode>1): Remove.
1524 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1526 PR rtl-optimization/60030
1527 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1528 lopart with paradoxical subreg before shifting it up by hprec.
1530 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1532 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1533 Remove extra newline at end of file.
1534 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1535 (arm_issue_rate): Handle cortexa57.
1536 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1537 (cortex-a57.cortex-a53): Likewise.
1539 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1542 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1543 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1545 (arm_expand_prologue): Adjust all callers.
1546 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1547 info, registers also at the lowest numbered registers side. Use
1548 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1552 * var-tracking.c (adjust_mems): Before adding a SET to
1553 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1555 2014-02-06 Alan Modra <amodra@gmail.com>
1558 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1559 change SDmode to DDmode when lra_in_progress.
1561 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1564 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1565 free_data_ref on the dr first, and before goto again also set dr
1566 to the next dr. For simd_lane_access, free old datarefs[i] before
1567 overwriting it. For get_vectype_for_scalar_type failure, don't
1568 free_data_ref if simd_lane_access.
1570 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1573 * tree.h (opts_for_fn): New inline function.
1574 (opt_for_fn): Define.
1575 * config/i386/i386.c (ix86_function_regparm): Use
1576 opt_for_fn (decl, optimize) instead of optimize.
1578 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1580 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1581 for SYMBOL_REF in large memory model.
1583 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1585 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1587 (cortex-a57): Likewise.
1588 (cortex-a57.cortex-a53): Likewise.
1590 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1591 Kugan Vivekanandarajah <kuganv@linaro.org>
1593 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1595 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1597 2014-02-06 Richard Biener <rguenther@suse.de>
1599 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1600 set_loop_copy and initialize_original_copy_tables.
1602 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1604 * config/aarch64/aarch64-simd.md
1605 (aarch64_ashr_simddi): Change QI to SI.
1607 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1608 Jakub Jelinek <jakub@redhat.com>
1611 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1614 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1616 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1617 CODE_FOR_altivec_vpku[hw]um to
1618 CODE_FOR_altivec_vpku[hw]um_direct.
1619 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1620 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1621 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1622 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1624 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1626 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1627 generation for -maltivec=be.
1628 (altivec_vsumsws): Simplify redundant test.
1630 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1632 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1633 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1634 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1635 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1636 gen_altivec_vpkuwum.
1637 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1639 (altivec_vpks<VI_char>ss): Likewise.
1640 (altivec_vpks<VI_char>us): Likewise.
1641 (altivec_vpku<VI_char>us): Likewise.
1642 (altivec_vpku<VI_char>um): Likewise.
1643 (altivec_vpku<VI_char>um_direct): New (copy of
1644 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1646 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1647 target is little endian and -maltivec=be is not specified.
1648 (*altivec_vupkhs<VU_char>_direct): New (copy of
1649 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1650 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1651 target is little endian and -maltivec=be is not specified.
1652 (*altivec_vupkls<VU_char>_direct): New (copy of
1653 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1654 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1655 little endian and -maltivec=be is not specified.
1656 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1657 little endian and -maltivec=be is not specified.
1659 2014-02-05 Richard Henderson <rth@redhat.com>
1662 * combine-stack-adj.c: Revert r206943.
1663 * sched-int.h (struct deps_desc): Add last_args_size.
1664 * sched-deps.c (init_deps): Initialize it.
1665 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1666 contain REG_ARGS_SIZE notes.
1668 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1670 * lto-cgraph.c (asm_nodes_output): Make global.
1671 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1672 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1673 (driver_handle_option): Handle OPT_fwpa.
1675 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1678 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1679 a comment typo and formatting issue. If odr_hash hasn't been
1680 created, return vNULL and set *completep to false.
1683 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1684 bb with no successors.
1686 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1689 * doc/invoke.texi (-march): Clarify documentation for ARM.
1693 2014-02-05 Richard Biener <rguenther@suse.de>
1695 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1696 when not vectorizing because of too many alias checks.
1697 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1698 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1700 2014-02-05 Nick Clifton <nickc@redhat.com>
1702 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1703 accept extended registers in any mode when compiling for the MN10300.
1705 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1707 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1708 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1709 sanitization attributes.
1710 (can_inline_edge_p): Likewise.
1711 (sanitize_attrs_match_for_inline_p): New function.
1713 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1715 * ipa-prop.c (detect_type_change): Shor circuit testing of
1716 type changes on THIS pointer.
1718 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1721 * config/pa/pa.c (legitimize_tls_address): Return original address
1722 if not passed a SYMBOL_REF rtx.
1723 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1725 (pa_emit_move_sequence): Simplify TLS source operands.
1726 (pa_legitimate_constant_p): Reject all TLS constants.
1727 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1728 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1730 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1732 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1733 groups when we know they are controlled by LTO.
1734 * varasm.c (default_binds_local_p_1): If object is in other partition,
1735 it will be resolved locally.
1737 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1739 * config/host-linux.c (linux_gt_pch_use_address): Don't
1740 use SSIZE_MAX because it is not always defined.
1742 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1745 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1746 threshold for pseudo splitting.
1747 (update_ebb_live_info): Process call argument hard registers and
1748 hard registers from insn definition too.
1749 (max_small_class_regs_num): New constant.
1750 (inherit_in_ebb): Update live hard regs through EBBs. Update
1751 reloads_num only for small register classes. Don't split for
1754 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1757 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1760 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1762 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1765 2014-02-04 Marek Polacek <polacek@redhat.com>
1767 * gdbinit.in (pel): Define.
1769 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1771 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1774 2014-02-04 Richard Biener <rguenther@suse.de>
1777 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1778 in function context local.
1779 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1780 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1781 similar to LTO_imported_decl_ref.
1783 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1785 PR tree-optimization/60002
1786 * cgraphclones.c (build_function_decl_skip_args): Clear
1789 PR tree-optimization/60023
1790 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1791 false to gsi_replace.
1792 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1793 has been in some EH region and vec_stmt could throw, add
1794 vec_stmt into the same EH region.
1795 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1796 has no lhs, ignore it.
1797 * internal-fn.c (expand_MASK_LOAD): Likewise.
1800 * tree-inline.c (copy_forbidden): Fail for
1801 __attribute__((optimize (0))) functions.
1804 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1805 is set, copy one less argument.
1806 (expand_simd_clones): Don't subtract clone_info->inbranch
1807 from simd_clone_struct_alloc argument.
1809 PR rtl-optimization/57915
1810 * recog.c (simplify_while_replacing): If all unary/binary/relational
1811 operation arguments are constant, attempt to simplify those.
1814 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1815 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1817 2014-02-04 Richard Biener <rguenther@suse.de>
1819 PR tree-optimization/60012
1820 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1821 TBAA disambiguation to all DDRs.
1823 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1826 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1827 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1829 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1832 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1834 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1836 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1837 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1839 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1842 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1843 to figure out targets of polymorphic calls with known decl.
1844 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1845 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1846 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1847 (get_polymorphic_call_info): ... here.
1848 (get_polymorphic_call_info_from_invariant): New function.
1850 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1852 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1853 lookup via vtable pointer; check for type consistency
1854 and turn inconsitent facts into UNREACHABLE.
1855 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1856 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1857 type inconsistent querries; return UNREACHABLE instead.
1859 2014-02-03 Richard Henderson <rth@twiddle.net>
1862 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1863 already processed this node.
1864 (normalize_one_pred_1): Pass along mark_set.
1865 (normalize_one_pred): Create and destroy a pointer_set_t.
1866 (normalize_one_pred_chain): Likewise.
1868 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1870 PR gcov-profile/58602
1871 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1873 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1876 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1877 -fno-devirtualize; try to devirtualize by the knowledge of
1878 virtual table pointer given by aggregate propagation.
1879 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1880 (ipa_print_node_jump_functions): Dump also offset that
1881 is relevant for polymorphic calls.
1882 (determine_known_aggregate_parts): Add arg_type parameter; use it
1883 instead of determining the type from pointer type.
1884 (ipa_compute_jump_functions_for_edge): Update call of
1885 determine_known_aggregate_parts.
1886 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1887 (gimple_get_virt_method_for_binfo): ... here; simplify using
1888 vtable_pointer_value_to_vtable.
1889 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1890 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1891 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1892 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1894 (vtable_pointer_value_to_binfo): ... here.
1895 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1897 2014-02-03 Teresa Johnson <tejohnson@google.com>
1899 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1900 redef of outer loop index variable.
1902 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1906 * doc/extend.texi (Function Attributes): Typo.
1908 2014-02-03 Cong Hou <congh@google.com>
1910 PR tree-optimization/60000
1911 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1912 if the vectorized statement is a store. A store statement can only
1913 appear at the end of pattern statements.
1915 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1917 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1918 (ix86_option_override_internal): Default long double to 64-bit for
1919 32-bit Bionic and to 128-bit for 64-bit Bionic.
1921 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1922 TARGET_LONG_DOUBLE_128 is true.
1923 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1925 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1926 (mlong-double-64): Negate -mlong-double-128.
1927 (mlong-double-128): New option.
1929 * config/i386/i386-c.c (ix86_target_macros): Define
1930 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1932 * doc/invoke.texi: Document -mlong-double-128.
1934 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1936 PR rtl-optimization/60024
1937 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1939 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1941 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1943 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1945 PR rtl-optimization/57662
1946 * sel-sched.c (code_motion_path_driver): Do not mark already not
1947 existing blocks in the visiting bitmap.
1949 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1951 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1952 on the insn being emitted.
1954 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1955 Will Deacon <will.deacon@arm.com>
1957 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1959 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1961 * config/arm/arm-tables.opt: Regenerate.
1963 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1965 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1966 for vector types other than V16QImode.
1967 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1968 define_expand, and call altivec_expand_vec_perm_le when producing
1969 code with little endian element order.
1970 (*altivec_vperm_<mode>_internal): New insn having previous
1971 behavior of altivec_vperm_<mode>.
1972 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1973 altivec_expand_vec_perm_le when producing code with little endian
1975 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1976 behavior of altivec_vperm_<mode>_uns.
1978 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1980 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1981 (altivec_vsumsws): Add handling for -maltivec=be with a little
1983 (altivec_vsumsws_direct): New.
1984 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1985 gen_altivec_vsumsws.
1987 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1989 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1990 vtable_pointer_value_to_binfo): New functions.
1991 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1992 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1994 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1996 * config/nios2/nios2.md (load_got_register): Initialize GOT
1997 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1998 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2000 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2002 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2003 preserverd by passthrough, do not propagate the type.
2005 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2007 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2008 (mips_atomic_assign_expand_fenv): New function.
2009 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2011 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2013 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2014 (__builtin_mips_set_fcsr): Likewise.
2015 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2016 MIPS_USI_FTYPE_VOID.
2017 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2018 (mips16_expand_set_fcsr): Likewise.
2019 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2020 (mips16_set_fcsr_stub): Likewise.
2021 (mips16_get_fcsr_one_only_stub): New class.
2022 (mips16_set_fcsr_one_only_stub): Likewise.
2023 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2024 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2025 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2026 (hard_float): New availability predicate.
2027 (mips_builtins): Add get_fcsr and set_fcsr.
2028 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2029 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2030 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2031 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2032 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2035 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2037 * config/mips/mips.c (mips_one_only_stub): New class.
2038 (mips_need_mips16_rdhwr_p): Replace with...
2039 (mips16_rdhwr_stub): ...this new variable.
2040 (mips16_stub_call_address): New function.
2041 (mips16_rdhwr_one_only_stub): New class.
2042 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2043 (mips_output_mips16_rdhwr): Delete.
2044 (mips_finish_stub): New function.
2045 (mips_code_end): Use it to handle rdhwr stubs.
2047 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2050 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2051 when calculating size of integer atomic types.
2053 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2055 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2057 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2059 PR tree-optimization/60003
2060 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2061 * profile.c (branch_prob): Use gimple_call_builtin_p
2062 to check for BUILT_IN_SETJMP_RECEIVER.
2063 * tree-inline.c (copy_bb): Call notice_special_calls.
2065 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2068 * lra-constraints.c (process_alt_operands): Update reload_sum only
2071 2014-01-31 Richard Henderson <rth@redhat.com>
2074 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2075 until after else_eh is processed.
2077 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2079 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2080 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2081 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2082 in smmintrin.h, remove them.
2083 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2084 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2085 * config/i386/i386.md (ROUND_SAE): Fix value.
2086 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2087 (const48_operand): New.
2088 * config/i386/subst.md (round), (round_expand): Use
2089 const_4_or_8_to_11_operand.
2090 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2092 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2094 * config/i386/constraints.md (Yk): Swap meaning with k.
2095 * config/i386/i386.md (movhi_internal): Change Yk to k.
2096 (movqi_internal): Ditto.
2097 (*k<logic><mode>): Ditto.
2100 (kandn<mode>): Ditto.
2101 (*<code>hi_1): Ditto.
2102 (*<code>qi_1): Ditto.
2103 (kxnor<mode>): Ditto.
2104 (kortestzhi): Ditto.
2105 (kortestchi): Ditto.
2107 (*one_cmplhi2_1): Ditto.
2108 (*one_cmplqi2_1): Ditto.
2109 * config/i386/sse.md (): Change k to Yk.
2110 (avx512f_load<mode>_mask): Ditto.
2111 (avx512f_blendm<mode>): Ditto.
2112 (avx512f_store<mode>_mask): Ditto.
2113 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2114 (avx512f_storedqu<mode>_mask): Ditto.
2115 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2117 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2118 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2119 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2120 (avx512f_maskcmp<mode>3): Ditto.
2121 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2122 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2123 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2124 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2125 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2126 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2127 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2128 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2129 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2130 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2131 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2132 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2133 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2134 (vec_extract_lo_<mode>_maskm): Ditto.
2135 (vec_extract_hi_<mode>_maskm): Ditto.
2136 (avx512f_vternlog<mode>_mask): Ditto.
2137 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2138 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2139 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2140 (avx512f_<code>v8div16qi2_mask): Ditto.
2141 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2142 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2143 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2144 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2145 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2146 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2147 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2148 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2149 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2150 (avx512cd_maskb_vec_dupv8di): Ditto.
2151 (avx512cd_maskw_vec_dupv16si): Ditto.
2152 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2153 (avx512f_vpermi2var<mode>3_mask): Ditto.
2154 (avx512f_vpermi2var<mode>3_mask): Ditto.
2155 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2156 (*avx512f_gathersi<mode>): Ditto.
2157 (*avx512f_gathersi<mode>_2): Ditto.
2158 (*avx512f_gatherdi<mode>): Ditto.
2159 (*avx512f_gatherdi<mode>_2): Ditto.
2160 (*avx512f_scattersi<mode>): Ditto.
2161 (*avx512f_scatterdi<mode>): Ditto.
2162 (avx512f_compress<mode>_mask): Ditto.
2163 (avx512f_compressstore<mode>_mask): Ditto.
2164 (avx512f_expand<mode>_mask): Ditto.
2165 * config/i386/subst.md (mask): Change k to Yk.
2166 (mask_scalar_merge): Ditto.
2169 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2171 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2173 2014-01-31 Richard Biener <rguenther@suse.de>
2176 * builtins.c (fold_builtin_memory_op): Make sure to not
2177 use a floating-point mode or a boolean or enumeral type for
2180 2014-01-30 DJ Delorie <dj@redhat.com>
2182 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2183 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2184 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2185 whenever main() has an epilogue.
2187 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2189 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2190 unused variable "field".
2191 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2192 (vsx_mergeh_<mode>): Likewise.
2193 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2194 (altivec_vmrghh): Likewise.
2195 (altivec_vmrghw): Likewise.
2196 (altivec_vmrglb): Likewise.
2197 (altivec_vmrglh): Likewise.
2198 (altivec_vmrglw): Likewise.
2199 (altivec_vspltb): Add missing uses.
2200 (altivec_vsplth): Likewise.
2201 (altivec_vspltw): Likewise.
2202 (altivec_vspltsf): Likewise.
2204 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2207 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2208 frame related instructions.
2210 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2212 PR rtl-optimization/59959
2213 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2214 any reload of register whose subreg is invalid.
2216 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2218 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2219 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2220 Add missing return type - void.
2222 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2224 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2225 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2226 remove element index adjustment for endian (now handled in vsx.md
2228 (altivec_expand_vec_perm_const): Use
2229 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2230 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2231 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2232 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2233 define_expand and a new define_insn *altivec_vspltb_internal;
2234 adjust for -maltivec=be on a little endian target.
2235 (altivec_vspltb_direct): New.
2236 (altivec_vsplth): Divide into a define_expand and a new
2237 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2238 little endian target.
2239 (altivec_vsplth_direct): New.
2240 (altivec_vspltw): Divide into a define_expand and a new
2241 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2242 little endian target.
2243 (altivec_vspltw_direct): New.
2244 (altivec_vspltsf): Divide into a define_expand and a new
2245 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2246 a little endian target.
2248 2014-01-30 Richard Biener <rguenther@suse.de>
2250 PR tree-optimization/59993
2251 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2252 can propagate form the earlier stmt and avoid the transform
2253 when the intermediate result is needed.
2255 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2257 * README.Portability: Fix typo.
2259 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2261 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2262 comparison_operator with ordered_comparison_operator.
2264 2014-01-30 Nick Clifton <nickc@redhat.com>
2266 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2267 Rename to mn10300_store_multiple_regs.
2268 * config/mn10300/mn10300.c: Likewise.
2269 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2270 store_multiple_regs.
2271 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2272 Call mn10300_store_multiple_regs.
2274 2014-01-30 Nick Clifton <nickc@redhat.com>
2275 DJ Delorie <dj@redhat.com>
2277 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2278 %fp 2 to keep registers after it properly word-aligned.
2279 (rl78_alloc_physical_registers_umul): Handle the case where both
2280 input operands are the same.
2282 2014-01-30 Richard Biener <rguenther@suse.de>
2284 PR tree-optimization/59903
2285 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2288 2014-01-30 Jason Merrill <jason@redhat.com>
2291 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2294 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2296 2014-01-30 Richard Biener <rguenther@suse.de>
2298 PR tree-optimization/59951
2299 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2301 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2304 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2305 SFmode to DFmode case.
2307 2014-01-29 DJ Delorie <dj@redhat.com>
2309 * config/msp430/msp430.opt (-minrt): New.
2310 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2312 (ENDFILE_SPEC): Likewise.
2314 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2316 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2317 (estimate_function_body_sizes): Use it.
2319 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2322 * dwarf2out.c (is_cxx_auto): New.
2323 (is_base_type): Use it.
2324 (gen_type_die_with_usage): Likewise.
2326 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2328 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2329 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2330 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2331 -maltivec=be with LE targets.
2332 (vsx_mergeh_<mode>): Likewise.
2333 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2334 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2335 (altivec_vmrghb): Replace with define_expand and new
2336 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2337 (altivec_vmrghb_direct): New define_insn.
2338 (altivec_vmrghh): Replace with define_expand and new
2339 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2340 (altivec_vmrghh_direct): New define_insn.
2341 (altivec_vmrghw): Replace with define_expand and new
2342 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2343 (altivec_vmrghw_direct): New define_insn.
2344 (*altivec_vmrghsf): Adjust for endianness.
2345 (altivec_vmrglb): Replace with define_expand and new
2346 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2347 (altivec_vmrglb_direct): New define_insn.
2348 (altivec_vmrglh): Replace with define_expand and new
2349 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2350 (altivec_vmrglh_direct): New define_insn.
2351 (altivec_vmrglw): Replace with define_expand and new
2352 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2353 (altivec_vmrglw_direct): New define_insn.
2354 (*altivec_vmrglsf): Adjust for endianness.
2355 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2356 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2357 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2358 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2359 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2360 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2361 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2362 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2364 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2366 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2367 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2370 2014-01-29 Richard Biener <rguenther@suse.de>
2372 PR tree-optimization/58742
2373 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2374 associate_pointerplus_align.
2375 (associate_pointerplus_diff): New function.
2376 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2377 and associate_pointerplus_diff.
2379 2014-01-29 Richard Biener <rguenther@suse.de>
2381 * lto-streamer.h (LTO_major_version): Bump to 3.
2382 (LTO_minor_version): Reset to 0.
2384 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2386 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2387 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2388 (arm_file_start): Generate correct asm header for armv7ve.
2389 * config/arm/bpabi.h: Add multilib support for armv7ve.
2390 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2391 and cortex-a15 to armv7ve.
2392 * config/arm/t-aprofile: Add multilib support for armv7ve.
2393 * doc/invoke.texi: Document -march=armv7ve.
2395 2014-01-29 Richard Biener <rguenther@suse.de>
2397 PR tree-optimization/58742
2398 * tree-ssa-forwprop.c (associate_plusminus): Return true
2399 if we changed sth, defer EH cleanup to ...
2400 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2401 (simplify_mult): New function.
2403 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2406 PR tree-optimization/59920
2407 * tree.c (build_common_builtin_nodes): Remove
2408 __builtin_setjmp_dispatcher initialization.
2409 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2410 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2411 instead of gsi_after_labels + manually skipping debug stmts.
2412 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2413 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2414 * tree-inline.c (copy_edges_for_bb): Remove
2415 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2416 argument. Ignore computed_goto_p stmts. Don't call
2417 make_abnormal_goto_edges. If a call might need abnormal edges
2418 for non-local gotos, see if it already has an edge to
2419 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2420 with true argument, don't do anything then, otherwise add
2421 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2422 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2424 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2425 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2426 (lower_stmt): Don't set data->calls_builtin_setjmp.
2427 (lower_builtin_setjmp): Adjust comment.
2428 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2429 * tree-cfg.c (found_computed_goto): Remove.
2430 (factor_computed_gotos): Remove.
2431 (make_goto_expr_edges): Return bool, true for computed gotos.
2432 Don't call make_abnormal_goto_edges.
2433 (build_gimple_cfg): Don't set found_computed_goto, don't call
2434 factor_computed_gotos.
2435 (computed_goto_p): No longer static.
2436 (make_blocks): Don't set found_computed_goto.
2437 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2438 (make_edges): If make_goto_expr_edges returns true, push bb
2439 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2440 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2441 vector. Record mapping between bbs and OpenMP regions if there
2442 are any, adjust make_gimple_omp_edges caller. Call
2443 handle_abnormal_edges.
2444 (make_abnormal_goto_edges): Remove.
2445 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2446 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2447 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2448 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2449 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2450 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2451 filling *region also set *region_idx to (*region)->entry->index.
2454 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2455 For REGs set ORIGINAL_REGNO.
2457 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2459 * doc/md.texi: Mention that a target shouldn't implement
2460 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2463 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2465 PR tree-optimization/59594
2466 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2467 a copy of the datarefs vector rather than the vector itself.
2469 2014-01-28 Jason Merrill <jason@redhat.com>
2472 * dwarf2out.c (auto_die): New static.
2473 (gen_type_die_with_usage): Handle C++1y 'auto'.
2474 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2477 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2480 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2481 (SPEC_X32): Likewise.
2482 (SPEC_64): Likewise.
2483 * config/i386/i386.c (ix86_option_override_internal): Turn off
2484 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2486 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2487 * config/i386/i386.h (TARGET_16BIT): New macro.
2488 (TARGET_16BIT_P): Likewise.
2489 * config/i386/i386.opt: Add m16.
2490 * doc/invoke.texi: Document -m16.
2492 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2494 PR preprocessor/59935
2495 * input.c (location_get_source_line): Bail out on when line number
2496 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2498 2014-01-28 Richard Biener <rguenther@suse.de>
2500 PR tree-optimization/58742
2501 * tree-ssa-forwprop.c (associate_plusminus): Handle
2502 pointer subtraction of the form (T)(P + A) - (T)P.
2504 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2506 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2509 2014-01-28 Richard Biener <rguenther@suse.de>
2512 2014-01-28 Richard Biener <rguenther@suse.de>
2514 PR rtl-optimization/45364
2515 PR rtl-optimization/59890
2516 * var-tracking.c (local_get_addr_clear_given_value): Handle
2517 already cleared slot.
2518 (val_reset): Handle not allocated local_get_addr_cache.
2519 (vt_find_locations): Use post-order on the inverted CFG.
2521 2014-01-28 Richard Biener <rguenther@suse.de>
2523 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2525 2014-01-28 Richard Biener <rguenther@suse.de>
2527 PR rtl-optimization/45364
2528 PR rtl-optimization/59890
2529 * var-tracking.c (local_get_addr_clear_given_value): Handle
2530 already cleared slot.
2531 (val_reset): Handle not allocated local_get_addr_cache.
2532 (vt_find_locations): Use post-order on the inverted CFG.
2534 2014-01-28 Alan Modra <amodra@gmail.com>
2536 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2537 * configure.ac <recursive call for build != host>: Define
2538 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2539 and LD_FOR_BUILD too.
2540 * configure: Regenerate.
2542 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2544 * config/i386/i386.c (get_builtin_code_for_version): Separate
2545 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2546 Broadwell from Haswell.
2548 2014-01-27 Steve Ellcey <sellcey@mips.com>
2550 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2551 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2552 * config/mips/mips.c (mips_option_override): Change setting
2554 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2555 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2556 Change from Mask to Var.
2558 2014-01-27 Jeff Law <law@redhat.com>
2560 * ipa-inline.c (inline_small_functions): Fix typo.
2562 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2564 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2565 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2566 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2567 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2568 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2569 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2570 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2571 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2572 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2573 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2574 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2575 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2576 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2577 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2578 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2579 (_mm512_storeu_epi64): Ditto.
2580 (_mm512_cmpge_epi32_mask): Ditto.
2581 (_mm512_cmpge_epu32_mask): Ditto.
2582 (_mm512_cmpge_epi64_mask): Ditto.
2583 (_mm512_cmpge_epu64_mask): Ditto.
2584 (_mm512_cmple_epi32_mask): Ditto.
2585 (_mm512_cmple_epu32_mask): Ditto.
2586 (_mm512_cmple_epi64_mask): Ditto.
2587 (_mm512_cmple_epu64_mask): Ditto.
2588 (_mm512_cmplt_epi32_mask): Ditto.
2589 (_mm512_cmplt_epu32_mask): Ditto.
2590 (_mm512_cmplt_epi64_mask): Ditto.
2591 (_mm512_cmplt_epu64_mask): Ditto.
2592 (_mm512_cmpneq_epi32_mask): Ditto.
2593 (_mm512_cmpneq_epu32_mask): Ditto.
2594 (_mm512_cmpneq_epi64_mask): Ditto.
2595 (_mm512_cmpneq_epu64_mask): Ditto.
2596 (_mm512_expand_pd): Ditto.
2597 (_mm512_expand_ps): Ditto.
2598 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2599 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2600 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2601 * config/i386/i386.c (ix86_builtins): Add
2602 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2603 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2604 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2605 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2606 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2607 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2608 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2609 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2610 IX86_BUILTIN_PMOVUSQW512_MEM.
2611 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2612 __builtin_ia32_pmovsqd512mem_mask,
2613 __builtin_ia32_pmovqd512mem_mask,
2614 __builtin_ia32_pmovusqw512mem_mask,
2615 __builtin_ia32_pmovsqw512mem_mask,
2616 __builtin_ia32_pmovqw512mem_mask,
2617 __builtin_ia32_pmovusdw512mem_mask,
2618 __builtin_ia32_pmovsdw512mem_mask,
2619 __builtin_ia32_pmovdw512mem_mask,
2620 __builtin_ia32_pmovqb512mem_mask,
2621 __builtin_ia32_pmovusqb512mem_mask,
2622 __builtin_ia32_pmovsqb512mem_mask,
2623 __builtin_ia32_pmovusdb512mem_mask,
2624 __builtin_ia32_pmovsdb512mem_mask,
2625 __builtin_ia32_pmovdb512mem_mask.
2626 (bdesc_args): Add __builtin_ia32_expanddf512,
2627 __builtin_ia32_expandsf512.
2628 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2629 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2630 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2631 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2632 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2633 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2634 (avx512f_<code>v8div16qi2_mask_store): This.
2635 (avx512f_expand<mode>): New.
2637 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2639 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2641 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2642 (_mm512_prefetch_i32scatter_pd): Ditto.
2643 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2644 (_mm512_prefetch_i64scatter_pd): Ditto.
2645 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2646 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2647 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2648 (_mm512_prefetch_i32scatter_ps): Ditto.
2649 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2650 (_mm512_prefetch_i64scatter_ps): Ditto.
2651 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2652 * config/i386/i386-builtin-types.def: Define
2653 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2654 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2655 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2656 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2657 IX86_BUILTIN_SCATTERPFQPD.
2658 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2659 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2660 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2661 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2662 __builtin_ia32_scatterpfqps.
2663 (ix86_expand_builtin): Expand new built-ins.
2664 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2665 fix memory access data type.
2666 (*avx512pf_gatherpf<mode>_mask): Ditto.
2667 (*avx512pf_gatherpf<mode>): Ditto.
2668 (avx512pf_scatterpf<mode>): Ditto.
2669 (*avx512pf_scatterpf<mode>_mask): Ditto.
2670 (*avx512pf_scatterpf<mode>): Ditto.
2671 (GATHER_SCATTER_SF_MEM_MODE): New.
2672 (avx512pf_gatherpf<mode>df): Ditto.
2673 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2674 (*avx512pf_scatterpf<mode>df): Ditto.
2676 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2679 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2680 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2683 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2685 * common/config/arm/arm-common.c
2686 (arm_rewrite_mcpu): Handle multiple names.
2688 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2690 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2692 * gimple-builder.h (create_gimple_tmp): Delete.
2694 2014-01-27 Christian Bruel <christian.bruel@st.com>
2696 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2699 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2701 * config/pa/pa.md (call): Generate indirect long calls to non-local
2702 functions when outputing 32-bit code.
2703 (call_value): Likewise except for special call to buggy powf function.
2705 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2706 portable runtime and PIC indirect calls.
2707 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2708 and PIC call sequences. Use ldo instead of blr to set return register
2709 in PIC call sequence.
2711 2014-01-25 Walter Lee <walt@tilera.com>
2713 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2714 avoid clobbering a live register.
2716 2014-01-25 Walter Lee <walt@tilera.com>
2718 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2719 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2720 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2721 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2723 2014-01-25 Walter Lee <walt@tilera.com>
2725 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2726 arguments on even registers.
2727 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2729 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2730 (BIGGEST_ALIGNMENT): Ditto.
2731 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2733 2014-01-25 Walter Lee <walt@tilera.com>
2735 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2736 insns before bundling.
2737 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2739 2014-01-25 Walter Lee <walt@tilera.com>
2741 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2742 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2743 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2745 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2747 * config/mips/constraints.md (kl): Delete.
2748 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2749 define expands, using...
2750 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2751 instructions for MIPS16.
2752 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2753 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2755 2014-01-25 Walter Lee <walt@tilera.com>
2757 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2761 2014-01-25 Walter Lee <walt@tilera.com>
2763 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2764 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2766 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2768 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2771 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2773 * print-rtl.c (in_call_function_usage): New var.
2774 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2775 EXPR_LIST mode as mode and not as reg note name.
2778 * cfgloopmanip.c (copy_loop_info): If
2779 loop->warned_aggressive_loop_optimizations, make sure
2780 the flag is set in target loop too.
2782 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2784 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2786 * builtins.def: Likewise.
2787 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2788 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2789 * ira.c (ira_setup_eliminable_regset): Likewise.
2790 * omp-low.c (gate_expand_omp): Likewise.
2791 (execute_lower_omp): Likewise.
2792 (diagnose_sb_0): Likewise.
2793 (gate_diagnose_omp_blocks): Likewise.
2794 (simd_clone_clauses_extract): Likewise.
2797 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2799 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2800 correction for little endian...
2801 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2804 2014-01-24 Jeff Law <law@redhat.com>
2806 PR tree-optimization/59919
2807 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2808 for non-returning calls.
2810 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2812 * common/config/aarch64/aarch64-common.c
2813 (aarch64_rewrite_mcpu): Handle multiple names.
2814 * config/aarch64/aarch64.h
2815 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2817 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2819 * input.c (add_file_to_cache_tab): Handle the case where fopen
2822 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2825 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2826 from push operand if code of push isn't PRE_DEC.
2828 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2831 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2832 -mquad-memory-atomic. Update -mquad-memory documentation to say
2833 it is only used for non-atomic loads/stores.
2835 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2836 -mquad-memory or -mquad-memory-atomic switches.
2838 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2839 -mquad-memory-atomic to ISA 2.07 support.
2841 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2842 to separate support of normal quad word memory operations (ldq, stq)
2843 from the atomic quad word memory operations.
2845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2846 support to separate non-atomic quad word operations from atomic
2847 quad word operations. Disable non-atomic quad word operations in
2848 little endian mode so that we don't have to swap words after the
2849 load and before the store.
2850 (quad_load_store_p): Add comment about atomic quad word support.
2851 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2852 options printed with -mdebug=reg.
2854 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2855 -mquad-memory-atomic as the test for whether we have quad word
2856 atomic instructions.
2857 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2858 or -mp8-vector are used, allow byte/half-word atomic operations.
2860 * config/rs6000/sync.md (load_lockedti): Insure that the address
2861 is a proper indexed or indirect address for the lqarx instruction.
2862 On little endian systems, swap the hi/lo registers after the lqarx
2864 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2865 insure the address is valid for the lqarx instruction.
2866 (store_conditionalti): Insure that the address is a proper indexed
2867 or indirect address for the stqcrx. instruction. On little endian
2868 systems, swap the hi/lo registers before doing the stqcrx.
2870 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2871 insure the address is valid for the stqcrx. instruction.
2873 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2874 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2875 type of quad memory support is available.
2877 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2880 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2881 there is a danger of looping.
2883 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2885 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2886 force flag_ira_loop_pressure if set via command line.
2888 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2890 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2891 (ashr_simd): New builtin handling DI mode.
2892 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2893 (aarch64_sshr_simddi): New match pattern.
2894 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2895 (vshrd_n_s64): Likewise.
2896 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2898 2014-01-23 Nick Clifton <nickc@redhat.com>
2900 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2901 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2902 favour of mcu specific scripts.
2903 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2906 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2907 Alex Velenko <Alex.Velenko@arm.com>
2909 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2910 (vaddv_s16): Likewise.
2911 (vaddv_s32): Likewise.
2912 (vaddv_u8): Likewise.
2913 (vaddv_u16): Likewise.
2914 (vaddv_u32): Likewise.
2915 (vaddvq_s8): Likewise.
2916 (vaddvq_s16): Likewise.
2917 (vaddvq_s32): Likewise.
2918 (vaddvq_s64): Likewise.
2919 (vaddvq_u8): Likewise.
2920 (vaddvq_u16): Likewise.
2921 (vaddvq_u32): Likewise.
2922 (vaddvq_u64): Likewise.
2923 (vaddv_f32): Likewise.
2924 (vaddvq_f32): Likewise.
2925 (vaddvq_f64): Likewise.
2926 (vmaxv_f32): Likewise.
2927 (vmaxv_s8): Likewise.
2928 (vmaxv_s16): Likewise.
2929 (vmaxv_s32): Likewise.
2930 (vmaxv_u8): Likewise.
2931 (vmaxv_u16): Likewise.
2932 (vmaxv_u32): Likewise.
2933 (vmaxvq_f32): Likewise.
2934 (vmaxvq_f64): Likewise.
2935 (vmaxvq_s8): Likewise.
2936 (vmaxvq_s16): Likewise.
2937 (vmaxvq_s32): Likewise.
2938 (vmaxvq_u8): Likewise.
2939 (vmaxvq_u16): Likewise.
2940 (vmaxvq_u32): Likewise.
2941 (vmaxnmv_f32): Likewise.
2942 (vmaxnmvq_f32): Likewise.
2943 (vmaxnmvq_f64): Likewise.
2944 (vminv_f32): Likewise.
2945 (vminv_s8): Likewise.
2946 (vminv_s16): Likewise.
2947 (vminv_s32): Likewise.
2948 (vminv_u8): Likewise.
2949 (vminv_u16): Likewise.
2950 (vminv_u32): Likewise.
2951 (vminvq_f32): Likewise.
2952 (vminvq_f64): Likewise.
2953 (vminvq_s8): Likewise.
2954 (vminvq_s16): Likewise.
2955 (vminvq_s32): Likewise.
2956 (vminvq_u8): Likewise.
2957 (vminvq_u16): Likewise.
2958 (vminvq_u32): Likewise.
2959 (vminnmv_f32): Likewise.
2960 (vminnmvq_f32): Likewise.
2961 (vminnmvq_f64): Likewise.
2963 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2965 * config/aarch64/aarch64-simd.md
2966 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2967 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2968 (*aarch64_mul3_elt<mode>): Likewise.
2969 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2970 (*aarch64_mul3_elt_to_64v2df): Likewise.
2971 (*aarch64_mla_elt<mode>): Likewise.
2972 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2973 (*aarch64_mls_elt<mode>): Likewise.
2974 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2975 (*aarch64_fma4_elt<mode>): Likewise.
2976 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2977 (*aarch64_fma4_elt_to_64v2df): Likewise.
2978 (*aarch64_fnma4_elt<mode>): Likewise.
2979 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2980 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2981 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2982 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2983 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2984 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2985 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2986 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2987 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2989 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2991 * config/aarch64/aarch64-simd.md
2992 (aarch64_be_checked_get_lane<mode>): New define_expand.
2993 * config/aarch64/aarch64-simd-builtins.def
2994 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2995 New builtin definition.
2996 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2997 Use new safe be builtin.
2999 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3001 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3003 (aarch64_be_st1<mode>): Likewise.
3004 (aarch_ld1<VALL:mode>): Define_expand modified.
3005 (aarch_st1<VALL:mode>): Likewise.
3006 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3007 (UNSPEC_ST1): Likewise.
3009 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3011 * config/microblaze/microblaze.md: Add trap insn and attribute
3013 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3015 PR preprocessor/58580
3016 * input.h (location_get_source_line): Take an additional line_size
3018 (void diagnostics_file_cache_fini): Declare new function.
3019 * input.c (struct fcache): New type.
3020 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3021 New static constants.
3022 (diagnostic_file_cache_init, total_lines_num)
3023 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3024 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3025 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3026 (get_next_line, read_next_line, goto_next_line, read_line_num):
3027 New static function definitions.
3028 (diagnostic_file_cache_fini): New function.
3029 (location_get_source_line): Take an additional output line_len
3030 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3032 * diagnostic.c (diagnostic_finish): Call
3033 diagnostic_file_cache_fini.
3034 (adjust_line): Take an additional input parameter for the length
3035 of the line, rather than calculating it with strlen.
3036 (diagnostic_show_locus): Adjust the use of
3037 location_get_source_line and adjust_line with respect to their new
3038 signature. While displaying a line now, do not stop at the first
3039 null byte. Rather, display the zero byte as a space and keep
3040 going until we reach the size of the line.
3041 * Makefile.in: Add vec.o to OBJS-libcommon
3043 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3044 Ilya Tocar <ilya.tocar@intel.com>
3046 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3047 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3048 (__builtin_ia32_kmov16): Ditto.
3049 * config/i386/i386.md (UNSPEC_KMOV): New.
3052 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3054 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3055 (_mm512_storeu_si512): Ditto.
3057 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3060 * rtl.h (get_referenced_operands): Declare.
3061 * recog.c (get_referenced_operands): New function.
3062 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3063 operands have been referenced when recording LO_SUM references.
3065 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3067 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3069 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3071 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3072 Enable for generic and recent AMD targets.
3074 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3076 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3077 ARG_SIZE note when adjustment was eliminated.
3079 2014-01-22 Jeff Law <law@redhat.com>
3081 PR tree-optimization/59597
3082 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3083 in file. Accept new argument REGISTERING and use it to modify
3084 dump output appropriately.
3085 (register_jump_thread): Corresponding changes.
3086 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3087 thread paths involving joiner blocks. Add code to dump cancelled
3088 jump threading paths.
3090 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3092 PR rtl-optimization/59477
3093 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3094 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3096 2014-01-22 Tom Tromey <tromey@redhat.com>
3098 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3100 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3102 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3104 PR rtl-optimization/59896
3105 * lra-constraints.c (process_alt_operands): Check unused note for
3106 matched operands of insn with no output reloads.
3108 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3110 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3111 (mips_move_from_gpr_cost): Likewise.
3113 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3115 PR rtl-optimization/59858
3116 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3117 ira_class_hard_regs_num.
3118 (process_alt_operands): Increase reject for dying matched operand.
3120 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3123 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3124 smaller than size, perform several stores or loads and stores
3125 at dst + count - size to store or copy all of size bytes, rather
3126 than just last modesize bytes.
3128 2014-01-20 DJ Delorie <dj@redhat.com>
3130 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3131 that CLOBBERs are REGs before propogating their values.
3133 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3136 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3137 (cgraph_inline_failed_type): New function.
3138 * cgraph.h (DEFCIFCODE): Add type.
3139 (cgraph_inline_failed_type_t): New enum.
3140 (cgraph_inline_failed_type): New prototype.
3141 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3142 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3143 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3144 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3145 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3146 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3147 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3148 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3149 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3150 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3151 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3152 OPTIMIZATION_MISMATCH.
3153 * tree-inline.c (expand_call_inline): Emit errors during
3154 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3156 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3159 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3160 mode attribute in insn output.
3162 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3164 * output.h (output_constant): Delete.
3165 * varasm.c (output_constant): Make private.
3167 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3169 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3171 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3174 * tree.h (fold_builtin_strcat): New prototype.
3175 * builtins.c (fold_builtin_strcat): No longer static. Add len
3176 argument, if non-NULL, don't call c_strlen. Optimize
3177 directly into __builtin_memcpy instead of __builtin_strcpy.
3178 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3179 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3181 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3183 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3184 for SImode_address_operand operands, having only a REG argument.
3186 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3188 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3189 loader name using mbig-endian.
3190 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3192 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3194 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3198 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3200 * config/aarch64/aarch64-protos.h
3201 (aarch64_cannot_change_mode_class_ptr): Declare.
3202 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3203 aarch64_cannot_change_mode_class_ptr): New.
3204 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3205 backend hook aarch64_cannot_change_mode_class.
3207 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3209 * common/config/aarch64/aarch64-common.c
3210 (aarch64_handle_option): Don't handle any option order logic here.
3211 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3212 selected_cpu, warn on architecture version mismatch.
3213 (aarch64_override_options): Fix parsing order for option strings.
3215 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3216 Iain Sandoe <iain@codesourcery.com>
3219 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3220 warning. Amend comment to reflect current functionality.
3222 2014-01-20 Richard Biener <rguenther@suse.de>
3225 * builtins.c (fold_builtin_strcat): Remove case better handled
3226 by tree-ssa-strlen.c.
3228 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3230 * config/aarch64/aarch64.opt
3231 (mcpu, march, mtune): Make case-insensitive.
3233 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3236 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3237 if operands[1] is a REG or ZERO_EXTEND of a REG.
3239 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3241 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3243 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3245 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3246 long non-pic millicode calls.
3248 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3250 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3252 2014-01-19 Kito Cheng <kito@0xlab.org>
3254 * builtins.c (expand_movstr): Check movstr expand done or fail.
3256 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3257 H.J. Lu <hongjiu.lu@intel.com>
3260 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3261 to DImode for zero-extended addresses.
3263 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3265 PR rtl-optimization/57763
3266 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3267 on the new indirect jump_insn and increment LABEL_NUSES (label).
3269 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3273 * config.gcc (x86_archs): New variable.
3274 (x86_64_archs): Likewise.
3275 (x86_cpus): Likewise.
3276 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3277 --with-arch/--with-cpu= options.
3278 Support --with-arch=/--with-cpu={nehalem,westmere,
3279 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3281 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3283 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3284 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3286 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3288 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3290 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3293 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3294 clear cpp_get_options (parse_in)->warn_unused_macros for
3295 ix86_target_macros_internal with cpp_define.
3297 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3299 * jump.c (delete_related_insns): Keep (use (insn))s.
3300 * reorg.c (redundant_insn): Check for barriers too.
3302 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3304 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3306 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3308 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3309 call to $$dyncall when TARGET_LONG_CALLS is true.
3311 2014-01-17 Jeff Law <law@redhat.com>
3313 * ree.c (combine_set_extension): Temporarily disable test for
3314 changing number of hard registers.
3316 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3319 * ipa-inline-analysis.c (inline_free_summary):
3320 Do not free summary of aliases.
3322 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3325 * gimplify.c (gimplify_expr): Use create_tmp_var
3326 instead of create_tmp_var_raw. If cond doesn't have
3327 integral type, don't add the IFN_ANNOTATE builtin at all.
3329 2014-01-17 Martin Jambor <mjambor@suse.cz>
3332 * ipa-cp.c (prev_edge_clone): New variable.
3333 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3334 Also resize prev_edge_clone vector.
3335 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3336 (ipcp_edge_removal_hook): New function.
3337 (ipcp_driver): Register ipcp_edge_removal_hook.
3339 2014-01-17 Andrew Pinski <apinski@cavium.com>
3340 Steve Ellcey <sellcey@mips.com>
3343 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3346 2014-01-17 Jeff Law <law@redhat.com>
3349 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3350 so that pass_ccp runs first.
3352 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3354 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3355 (ix86_adjust_cost): Use !TARGET_XXX.
3356 (do_reorder_for_imul): Likewise.
3357 (swap_top_of_ready_list): Likewise.
3358 (ix86_sched_reorder): Likewise.
3360 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3362 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3363 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3364 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3365 (intel_memset): New. Duplicate slm_memset.
3366 (intel_cost): New. Duplicate slm_cost.
3367 (m_INTEL): New macro.
3368 (processor_target_table): Add "intel".
3369 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3370 with PROCESSOR_INTEL for "intel".
3371 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3372 PROCESSOR_SILVERMONT.
3373 (ix86_issue_rate): Likewise.
3374 (ix86_adjust_cost): Likewise.
3375 (ia32_multipass_dfa_lookahead): Likewise.
3376 (swap_top_of_ready_list): Likewise.
3377 (ix86_sched_reorder): Likewise.
3378 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3379 instead of TARGET_OPT_AGU.
3380 * config/i386/i386.h (TARGET_INTEL): New.
3381 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3382 (processor_type): Add PROCESSOR_INTEL.
3383 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3384 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3386 2014-01-17 Marek Polacek <polacek@redhat.com>
3389 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3392 2014-01-17 Richard Biener <rguenther@suse.de>
3394 PR tree-optimization/46590
3395 * opts.c (default_options_table): Add entries for
3396 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3397 all enabled at -O1 but not for -Og.
3398 * common.opt (fbranch-count-reg): Remove Init(1).
3399 (fmove-loop-invariants): Likewise.
3400 (ftree-pta): Likewise.
3402 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3404 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3405 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3406 decls to at least the GCC 4.8 used alignments.
3409 * tree-nested.c (convert_nonlocal_reference_stmt,
3410 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3411 of GIMPLE_BIND stmts, adjust associated decls.
3413 2014-01-17 Richard Biener <rguenther@suse.de>
3415 PR tree-optimization/46590
3416 * vec.h (vec<>::bseach): New member function implementing
3417 binary search according to C89 bsearch.
3418 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3419 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3420 bitmap pointer again. Make accesses_in_loop a flat array.
3421 (mem_ref_obstack): New global.
3422 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3423 (mark_ref_stored): Likewise.
3424 (ref_indep_loop_p_2): Likewise.
3425 (set_ref_stored_in_loop): New helper function.
3426 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3427 (memref_free): Adjust.
3428 (record_mem_ref_loc): Simplify.
3429 (gather_mem_refs_stmt): Adjust.
3430 (sort_locs_in_loop_postorder_cmp): New function.
3431 (analyze_memory_references): Sort accesses_in_loop after
3432 loop postorder number.
3433 (find_ref_loc_in_loop_cmp): New function.
3434 (for_all_locs_in_loop): Find relevant cluster of locs in
3435 accesses_in_loop and iterate without recursion.
3436 (execute_sm): Avoid uninit warning.
3437 (struct ref_always_accessed): Simplify.
3438 (ref_always_accessed::operator ()): Likewise.
3439 (ref_always_accessed_p): Likewise.
3440 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3441 loop postorder numbers here.
3442 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3445 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3448 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3449 on decls for which assemble_alias has been called.
3451 2014-01-17 Nick Clifton <nickc@redhat.com>
3453 * config/msp430/msp430.opt: (mcpu): New option.
3454 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3455 (msp430_option_override): Parse target_cpu. If the MCU name
3456 matches a generic string, clear target_mcu.
3457 (msp430_attr): Allow numeric interrupt values up to 63.
3458 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3459 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3461 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3463 * config/msp430/msp430.md (popm): Use %J rather than %I.
3464 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3465 (addhi_cy_i): Use immediate_operand for operand 2.
3466 * doc/invoke.texi: Document -mcpu option.
3468 2014-01-17 Richard Biener <rguenther@suse.de>
3470 PR rtl-optimization/38518
3471 * df.h (df_analyze_loop): Declare.
3472 * df-core.c: Include cfgloop.h.
3473 (df_analyze_1): Split out main part of df_analyze.
3474 (df_analyze): Adjust.
3475 (loop_inverted_post_order_compute): New function.
3476 (loop_post_order_compute): Likewise.
3477 (df_analyze_loop): New function avoiding whole-function
3479 * loop-invariant.c (find_defs): Use df_analyze_loop.
3480 (find_invariants): Adjust.
3481 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3483 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3485 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3486 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3488 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3490 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3491 traversal when removing references.
3493 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3496 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3498 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3501 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3502 pushing a reload for an autoinc when we had previously reloaded an
3503 inner part of the address.
3505 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3507 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3509 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3510 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3511 when not giving up or versioning for alias only because of
3513 (vect_analyze_data_ref_dependences): Set to true.
3514 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3516 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3517 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3521 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3524 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3525 operand 0 predicate for gathers, use a new pseudo as subtarget.
3527 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3530 * lra-constraints.c (process_alt_operands): Add printing debug info.
3531 Check absence of input/output reloads for matched operands too.
3533 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3535 PR rtl-optimization/59835
3536 * ira.c (ira_init_register_move_cost): Increase cost for
3539 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3541 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3543 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3546 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3547 non-register objects. Use gen_(high/low)part more consistently.
3550 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3553 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3554 endian support, remove tests for WORDS_BIG_ENDIAN.
3555 (p8_mfvsrd_3_<mode>): Likewise.
3556 (reload_gpr_from_vsx<mode>): Likewise.
3557 (reload_gpr_from_vsxsf): Likewise.
3558 (p8_mfvsrd_4_disf): Likewise.
3560 2014-01-16 Richard Biener <rguenther@suse.de>
3562 PR rtl-optimization/46590
3563 * lcm.c (compute_antinout_edge): Use postorder iteration.
3564 (compute_laterin): Use inverted postorder iteration.
3566 2014-01-16 Nick Clifton <nickc@redhat.com>
3569 * varasm.c (output_constant): Return the number of bytes actually
3571 (output_constructor_array_range): Update the field size with the
3572 number of bytes emitted by output_constant.
3573 (output_constructor_regular_field): Likewise. Also do not
3574 complain if the total number of bytes emitted is now greater
3575 than the expected fieldpos.
3576 * output.h (output_constant): Update prototype and descriptive comment.
3578 2014-01-16 Marek Polacek <polacek@redhat.com>
3581 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3582 it is error_mark_node.
3584 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3586 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3587 VALID_AVX256_REG_OR_OI_MODE.
3589 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3591 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3592 current procedure should be profiled.
3594 2014-01-15 Andrew Pinski <apinski@cavium.com>
3596 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3597 of moving from/to the STACK_REG register class.
3599 2014-01-15 Richard Henderson <rth@redhat.com>
3602 * reginfo.c (global_regs_decl): Globalize.
3603 * rtl.h (global_regs_decl): Declare.
3604 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3605 reserved via global_regs.
3607 2014-01-15 Teresa Johnson <tejohnson@google.com>
3609 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3611 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3613 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3614 and vmulosh rather than call gen_vec_widen_smult_*.
3615 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3616 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3617 (vec_widen_smult_even_v16qi): Likewise.
3618 (vec_widen_umult_even_v8hi): Likewise.
3619 (vec_widen_smult_even_v8hi): Likewise.
3620 (vec_widen_umult_odd_v16qi): Likewise.
3621 (vec_widen_smult_odd_v16qi): Likewise.
3622 (vec_widen_umult_odd_v8hi): Likewise.
3623 (vec_widen_smult_odd_v8hi): Likewise.
3624 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3625 vmuloub rather than call gen_vec_widen_umult_*.
3626 (vec_widen_umult_lo_v16qi): Likewise.
3627 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3628 vmulosb rather than call gen_vec_widen_smult_*.
3629 (vec_widen_smult_lo_v16qi): Likewise.
3630 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3631 rather than call gen_vec_widen_umult_*.
3632 (vec_widen_umult_lo_v8hi): Likewise.
3633 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3634 rather than call gen_vec_widen_smult_*.
3635 (vec_widen_smult_lo_v8hi): Likewise.
3637 2014-01-15 Jeff Law <law@redhat.com>
3639 PR tree-optimization/59747
3640 * ree.c (find_and_remove_re): Properly handle case where a second
3641 eliminated extension requires widening a copy created for elimination
3642 of a prior extension.
3643 (combine_set_extension): Ensure that the number of hard regs needed
3644 for a destination register does not change when we widen it.
3646 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3648 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3649 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3650 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3651 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3652 (avr-*-rtems*): Likewise.
3653 (bfin*-rtems*): Likewise.
3654 (moxie-*-rtems*): Likewise.
3655 (h8300-*-rtems*): Likewise.
3656 (i[34567]86-*-rtems*): Likewise.
3657 (lm32-*-rtems*): Likewise.
3658 (m32r-*-rtems*): Likewise.
3659 (m68k-*-rtems*): Likewise.
3660 (microblaze*-*-rtems*): Likewise.
3661 (mips*-*-rtems*): Likewise.
3662 (powerpc-*-rtems*): Likewise.
3663 (sh-*-rtems*): Likewise.
3664 (sparc-*-rtems*): Likewise.
3665 (sparc64-*-rtems*): Likewise.
3666 (v850-*-rtems*): Likewise.
3667 (m32c-*-rtems*): Likewise.
3669 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3671 PR rtl-optimization/59511
3672 * ira.c (ira_init_register_move_cost): Use memory costs for some
3673 cases of register move cost calculations.
3674 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3675 instead of BB frequency.
3676 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3677 * lra-assigns.c (find_hard_regno_for): Ditto.
3679 2014-01-15 Richard Biener <rguenther@suse.de>
3681 PR tree-optimization/59822
3682 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3683 (vectorizable_load): Use it to hoist defs of uses of invariant
3684 loads out of the loop.
3686 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3687 Kugan Vivekanandarajah <kuganv@linaro.org>
3690 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3693 2014-01-15 Richard Biener <rguenther@suse.de>
3695 PR rtl-optimization/59802
3696 * lcm.c (compute_available): Use inverted postorder to seed
3697 the initial worklist.
3699 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3702 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3703 ADDR_REGS for invalid symrefs in non-PIC code.
3705 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3708 * builtins.c (determine_block_size): Initialize *probable_max_size
3709 even if len_rtx is CONST_INT.
3711 2014-01-14 Andrew Pinski <apinski@cavium.com>
3713 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3714 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3715 (cortexa53_tunings): Likewise.
3716 (aarch64_sched_issue_rate): New function.
3717 (TARGET_SCHED_ISSUE_RATE): Define.
3719 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3721 * ira-costs.c (find_costs_and_classes): Add missed
3722 ira_init_register_move_cost_if_necessary.
3724 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3727 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3729 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3732 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3733 to indicate if type is used for function return value. Warn ABI
3734 change if the vector mode isn't available for function return value.
3735 (ix86_function_arg_advance): Pass false to type_natural_mode.
3736 (ix86_function_arg): Likewise.
3737 (ix86_gimplify_va_arg): Likewise.
3738 (function_arg_32): Don't warn ABI change.
3739 (ix86_function_value): Pass true to type_natural_mode.
3740 (ix86_return_in_memory): Likewise.
3741 (ix86_struct_value_rtx): Removed.
3742 (TARGET_STRUCT_VALUE_RTX): Likewise.
3744 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3746 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3747 converting a conditional jump into a conditional return.
3749 2014-01-14 Richard Biener <rguenther@suse.de>
3751 PR tree-optimization/58921
3752 PR tree-optimization/59006
3753 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3754 hoisting invariant stmts.
3755 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3756 invariant loads on the preheader edge if possible.
3758 2014-01-14 Joey Ye <joey.ye@arm.com>
3760 * doc/plugin.texi (Building GCC plugins): Update to C++.
3762 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3764 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3765 (_mm_rcp28_round_ss): Ditto.
3766 (_mm_rsqrt28_round_sd): Ditto.
3767 (_mm_rsqrt28_round_ss): Ditto.
3768 (_mm_rcp28_sd): Ditto.
3769 (_mm_rcp28_ss): Ditto.
3770 (_mm_rsqrt28_sd): Ditto.
3771 (_mm_rsqrt28_ss): Ditto.
3772 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3773 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3774 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3775 (IX86_BUILTIN_RCP28SD): Ditto.
3776 (IX86_BUILTIN_RCP28SS): Ditto.
3777 (IX86_BUILTIN_RSQRT28SD): Ditto.
3778 (IX86_BUILTIN_RSQRT28SS): Ditto.
3779 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3780 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3781 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3782 (ix86_expand_special_args_builtin): Expand new FTYPE.
3783 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3784 (srcp14<mode>): Make insn unary.
3785 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3786 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3787 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3788 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3789 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3790 Fix rounding: make it SAE only.
3791 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3793 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3795 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3796 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3797 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3798 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3799 (round_saeonly_mask_scalar_operand4): Ditto.
3800 (round_saeonly_mask_scalar_op3): Ditto.
3801 (round_saeonly_mask_scalar_op4): Ditto.
3803 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3805 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3806 Implement -maltivec=be for vec_insert and vec_extract.
3808 2014-01-10 DJ Delorie <dj@redhat.com>
3810 * config/msp430/msp430.md (call_internal): Don't allow memory
3811 references with SP as the base register.
3812 (call_value_internal): Likewise.
3813 * config/msp430/constraints.md (Yc): New. For memory references
3814 that don't use SP as a base register.
3816 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3817 "an integer without a # prefix"
3818 * config/msp430/msp430.md (epilogue_helper): Use it.
3820 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3823 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3824 AVX512F gather builtins.
3825 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3826 on gather decls with INTEGER_TYPE masktype.
3827 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3828 directly into the builtin rather than hoisting it before loop.
3830 PR tree-optimization/59387
3831 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3832 (scev_const_prop): If folded_casts and type has undefined overflow,
3833 use force_gimple_operand instead of force_gimple_operand_gsi and
3834 for each added stmt if it is assign with
3835 arith_code_with_undefined_signed_overflow, call
3836 rewrite_to_defined_overflow.
3837 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3838 gimple-fold.h instead.
3839 (arith_code_with_undefined_signed_overflow,
3840 rewrite_to_defined_overflow): Moved to ...
3841 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3842 rewrite_to_defined_overflow): ... here. No longer static.
3843 Include gimplify-me.h.
3844 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3845 rewrite_to_defined_overflow): New prototypes.
3847 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3849 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3851 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3853 * builtins.c (get_object_alignment_2): Minor tweak.
3854 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3856 2014-01-13 Christian Bruel <christian.bruel@st.com>
3858 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3859 optimized non constant lengths.
3861 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3864 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3865 load as __atomic_load_N if possible.
3867 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3869 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3871 (rs6000_expand_builtin): Adjust call.
3873 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3876 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3877 * config/rs6000/rs6000.c: Include target-globals.h.
3878 (rs6000_set_current_function): Instead of doing target_reinit
3879 unconditionally, use save_target_globals_default_opts and
3880 restore_target_globals.
3882 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3884 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3885 (rs6000_expand_builtin): Handle mffs and mtfsf.
3886 (rs6000_init_builtins): Define mffs and mtfsf.
3887 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3888 (rs6000_mffs): New pattern.
3889 (rs6000_mtfsf): New pattern.
3891 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3893 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3894 Start narrowing with START. Apply candidate-use pair
3895 and check overall cost in narrowing.
3896 (iv_ca_prune): Pass new argument.
3898 2014-01-10 Jeff Law <law@redhat.com>
3901 * ree.c (combine_reaching_defs): Ensure the defining statement
3902 occurs before the extension when optimizing extensions with
3903 different source and destination hard registers.
3905 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3908 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3909 vtables into the type inheritance graph.
3911 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3913 PR rtl-optimization/59754
3914 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3915 modes in the REGNO != REGNO case.
3917 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3919 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3921 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3923 PR tree-optimization/59745
3924 * tree-predcom.c (tree_predictive_commoning_loop): Call
3925 free_affine_expand_cache if giving up because components is NULL.
3927 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3928 GC in payload of target_globals struct instead of allocating them on
3929 the heap and the larger structs separately using GC.
3930 * target-globals.h (struct target_globals): Make regs, hard_regs,
3931 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3932 of GTY((skip)) and change type to void *.
3933 (reset_target_globals): Cast loads from those fields to corresponding
3936 2014-01-10 Steve Ellcey <sellcey@mips.com>
3939 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3940 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3941 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3943 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3946 * aarch64-modes.def (CC_Zmode): New flags mode.
3947 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3948 represents an equality.
3949 (aarch64_get_condition_code): Handle CC_Zmode.
3950 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3952 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3954 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3955 extraction in good case.
3957 2014-01-10 Richard Biener <rguenther@suse.de>
3959 PR tree-optimization/59374
3960 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3961 checking after SLP discovery. Mark stmts not participating
3962 in any SLP instance properly.
3964 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3966 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3967 when handling a SET rtx.
3969 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3971 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3972 (cortex-a57): Likewise.
3973 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3975 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3977 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3978 non-iwmmxt builtins.
3980 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3984 * ipa-devirt.c record_target_from_binfo): Take as argument
3985 stack of binfos and lookup matching one for virtual inheritance.
3986 (possible_polymorphic_call_targets_1): Update.
3988 2014-01-10 Huacai Chen <chenhc@lemote.com>
3990 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3991 kernel strings for Loongson-2E/2F/3A.
3993 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3996 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3997 is_gimple_call before calling gimple_call_internal_p.
3999 2014-01-09 Steve Ellcey <sellcey@mips.com>
4001 * Makefile.in (TREE_FLOW_H): Remove.
4002 (TREE_SSA_H): Add file names from tree-flow.h.
4003 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4004 * tree.h: Remove tree-flow.h reference.
4005 * hash-table.h: Remove tree-flow.h reference.
4006 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4007 reference with tree-ssa-loop.h.
4009 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4011 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4012 default element-order behavior for -maltivec.
4013 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4014 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4015 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4016 when targeting big endian, at least for now.
4017 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4019 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4022 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4023 var satisfies use_register_for_decl, just take into account type
4024 alignment, rather than decl alignment.
4026 PR tree-optimization/59622
4027 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4028 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4029 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4030 Don't devirtualize for inplace at all. For targets.length () == 1,
4031 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4033 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4035 * config/i386/i386.md (cpu): Remove the unused btver1.
4037 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4039 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4041 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4044 * tree-core.h (struct target_globals): New forward declaration.
4045 (struct tree_target_option): Add globals field.
4046 * tree.h (TREE_TARGET_GLOBALS): Define.
4047 (prepare_target_option_nodes_for_pch): New prototype.
4048 * target-globals.h (struct target_globals): Define even if
4050 * tree.c (prepare_target_option_node_for_pch,
4051 prepare_target_option_nodes_for_pch): New functions.
4052 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4053 * config/i386/i386.c: Include target-globals.h.
4054 (ix86_set_current_function): Instead of doing target_reinit
4055 unconditionally, use save_target_globals_default_opts and
4056 restore_target_globals.
4058 2014-01-09 Richard Biener <rguenther@suse.de>
4060 PR tree-optimization/59715
4061 * tree-cfg.h (split_critical_edges): Declare.
4062 * tree-cfg.c (split_critical_edges): Export.
4063 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4065 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4067 * cfgexpand.c (expand_stack_vars): Optionally disable
4068 asan stack protection.
4069 (expand_used_vars): Likewise.
4070 (partition_stack_vars): Likewise.
4071 * asan.c (asan_emit_stack_protection): Optionally disable
4072 after return stack usage.
4073 (instrument_derefs): Optionally disable memory access instrumentation.
4074 (instrument_builtin_call): Likewise.
4075 (instrument_strlen_call): Likewise.
4076 (asan_protect_global): Optionally disable global variables protection.
4077 * doc/invoke.texi: Added doc for new options.
4078 * params.def: Added new options.
4079 * params.h: Likewise.
4081 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4083 PR rtl-optimization/59724
4084 * ifcvt.c (cond_exec_process_if_block): Don't call
4085 flow_find_head_matching_sequence with 0 longest_match.
4086 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4087 non-active insns if !stop_after.
4088 (try_head_merge_bb): Revert 2014-01-07 changes.
4090 2014-01-08 Jeff Law <law@redhat.com>
4092 * ree.c (get_sub_rtx): New function, extracted from...
4093 (merge_def_and_ext): Here.
4094 (combine_reaching_defs): Use get_sub_rtx.
4096 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4098 * cgraph.h (varpool_variable_node): Do not choke on null node.
4100 2014-01-08 Catherine Moore <clm@codesourcery.com>
4102 * config/mips/mips.md (simple_return): Attempt to use JRC
4104 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4106 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4108 PR rtl-optimization/59137
4109 * reorg.c (steal_delay_list_from_target): Call update_block for
4111 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4113 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4115 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4116 two duplicate entries.
4118 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4121 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4123 * config/mips/mips.c (mips_truncated_op_cost): New function.
4124 (mips_rtx_costs): Adjust test for BADDU.
4125 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4127 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4129 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4130 (*baddu_si): ...this new pattern.
4132 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4135 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4137 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4140 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4142 (expand_expr, expand_normal): Adjust.
4143 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4144 inner_reference_p. Use inner_reference_p to expand inner references.
4145 (store_expr): Adjust.
4146 * cfgexpand.c (expand_call_stmt): Adjust.
4148 2014-01-08 Rong Xu <xur@google.com>
4150 * gcov-io.c (gcov_var): Move from gcov-io.h.
4151 (gcov_position): Ditto.
4152 (gcov_is_error): Ditto.
4153 (gcov_rewrite): Ditto.
4154 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4155 only part to libgcc/libgcov.h.
4157 2014-01-08 Marek Polacek <polacek@redhat.com>
4160 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4162 2014-01-08 Marek Polacek <polacek@redhat.com>
4165 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4167 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4169 PR rtl-optimization/59649
4170 * stor-layout.c (get_mode_bounds): For BImode return
4171 0 and STORE_FLAG_VALUE.
4173 2014-01-08 Richard Biener <rguenther@suse.de>
4176 * gimple.h (is_gimple_builtin_call): Remove.
4177 (gimple_builtin_call_types_compatible_p): New.
4178 (gimple_call_builtin_p): New overload.
4179 * gimple.c (is_gimple_builtin_call): Remove.
4180 (validate_call): Rename to ...
4181 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4183 (validate_type): New static function.
4184 (gimple_call_builtin_p): New overload and adjust.
4185 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4186 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4187 (gimple_fold_stmt_to_constant_1): Likewise.
4188 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4190 2014-01-08 Richard Biener <rguenther@suse.de>
4193 * gimplify.c (gimplify_expr): Gimplify register-register type
4194 VIEW_CONVERT_EXPRs to separate stmts.
4196 2014-01-07 Jeff Law <law@redhat.com>
4199 * ree.c (combine_set_extension): Handle case where source
4200 and destination registers in an extension insn are different.
4201 (combine_reaching_defs): Allow source and destination registers
4202 in extension to be different under limited circumstances.
4203 (add_removable_extension): Remove restriction that the
4204 source and destination registers in the extension are the same.
4205 (find_and_remove_re): Emit a copy from the extension's
4206 destination to its source after the defining insn if
4207 the source and destination registers are different.
4210 * ifcvt.c (merge_if_block): If we are merging a block with more than
4211 one successor with a block with no successors, remove any BARRIER
4212 after the second block.
4214 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4216 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4218 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4221 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4222 for 14-bit register offsets when INT14_OK_STRICT is false.
4224 2014-01-07 Roland Stigge <stigge@antcom.de>
4225 Michael Meissner <meissner@linux.vnet.ibm.com>
4228 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4229 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4231 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4233 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4236 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4238 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4239 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4240 rtx is const0_rtx or not.
4242 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4245 * target-globals.c (save_target_globals): Remove this_fn_optab
4247 * toplev.c: Include optabs.h.
4248 (target_reinit): Temporarily restore the global options if another
4249 set of options are in force.
4251 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4253 PR rtl-optimization/58668
4254 * cfgcleanup.c (flow_find_cross_jump): Don't count
4255 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4256 to determine what is counted.
4257 (flow_find_head_matching_sequence): Use active_insn_p to determine
4259 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4261 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4262 determine what is counted.
4264 PR tree-optimization/59643
4265 * tree-predcom.c (split_data_refs_to_components): If one dr is
4266 read and one write, determine_offset fails and the write isn't
4267 in the bad component, just put the read into the bad component.
4269 2014-01-07 Mike Stump <mikestump@comcast.net>
4270 Jakub Jelinek <jakub@redhat.com>
4273 * tree-core.h (struct tree_optimization_option): Change optabs
4274 type from unsigned char * to void *.
4275 * optabs.c (init_tree_optimization_optabs): Adjust
4276 TREE_OPTIMIZATION_OPTABS initialization.
4278 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4281 * config/i386/i386.h (struct machine_function): Add
4282 no_drap_save_restore field.
4283 * config/i386/i386.c (ix86_save_reg): Use
4284 !cfun->machine->no_drap_save_restore instead of
4285 crtl->stack_realign_needed.
4286 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4287 this function clears frame_pointer_needed. Set
4288 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4289 and DRAP reg is needed.
4291 2014-01-06 Marek Polacek <polacek@redhat.com>
4294 * doc/implement-c.texi: Mention that other integer types are
4295 permitted as bit-field types in strictly conforming mode.
4297 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4299 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4302 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4304 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4306 2014-01-06 Martin Jambor <mjambor@suse.cz>
4309 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4311 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4313 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4317 * var-tracking.c (add_stores): Preserve the value of the source even if
4318 we don't record the store.
4320 2014-01-06 Terry Guo <terry.guo@arm.com>
4322 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4324 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4327 * config/darwin.c (darwin_function_section): Adjust return values to
4328 correspond to optimisation changes made in r206070.
4330 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4332 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4333 from prefetch_block tune setting.
4334 (nocona_cost): Correct size of prefetch block to 64.
4336 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4338 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4339 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4340 used to save the static chain register in the computation of the offset
4341 from which the FP registers need to be restored.
4343 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4345 PR tree-optimization/59519
4346 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4347 ICE if get_current_def (current_new_name) is already non-NULL, as long
4348 as it is a phi result of some other phi in *new_exit_bb that has
4351 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4352 or vmovdqu* for misaligned_operand.
4353 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4354 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4355 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4356 aligned_mem for AVX512F masked aligned load and store builtins and for
4359 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4361 PR tree-optimization/59651
4362 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4363 Address range for negative step should be added by TYPE_SIZE_UNIT.
4365 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4367 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4368 overlapping registers also for registers other than the stack pointer.
4370 2014-01-03 Marek Polacek <polacek@redhat.com>
4373 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4376 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4379 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4382 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4383 (push splitter): Use <P:MODE_SIZE> instead of
4384 GET_MODE_SIZE (<P:MODE>mode).
4385 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4386 (mov -1, reg peephole2): Likewise.
4387 * config/i386/sse.md (*mov<mode>_internal,
4388 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4389 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4390 *<code><mode>3, *andnot<mode>3<mask_name>,
4391 <mask_codefor><code><mode>3<mask_name>): Likewise.
4392 * config/i386/subst.md (mask_mode512bit_condition,
4393 sd_mask_mode512bit_condition): Likewise.
4395 2014-01-02 Xinliang David Li <davidxl@google.com>
4397 PR tree-optimization/59303
4398 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4399 (dump_predicates): Better output format.
4400 (pred_equal_p): New function.
4401 (is_neq_relop_p): Ditto.
4402 (is_neq_zero_form_p): Ditto.
4403 (pred_expr_equal_p): Ditto.
4404 (pred_neg_p): Ditto.
4405 (simplify_pred): Ditto.
4406 (simplify_preds_2): Ditto.
4407 (simplify_preds_3): Ditto.
4408 (simplify_preds_4): Ditto.
4409 (simplify_preds): Ditto.
4411 (push_to_worklist): Ditto.
4412 (get_pred_info_from_cmp): Ditto.
4413 (is_degenerated_phi): Ditto.
4414 (normalize_one_pred_1): Ditto.
4415 (normalize_one_pred): Ditto.
4416 (normalize_one_pred_chain): Ditto.
4417 (normalize_preds): Ditto.
4418 (normalize_cond_1): Remove function.
4419 (normalize_cond): Ditto.
4420 (is_gcond_subset_of): Ditto.
4421 (is_subset_of_any): Ditto.
4422 (is_or_set_subset_of): Ditto.
4423 (is_and_set_subset_of): Ditto.
4424 (is_norm_cond_subset_of): Ditto.
4425 (pred_chain_length_cmp): Ditto.
4426 (convert_control_dep_chain_into_preds): Type change.
4427 (find_predicates): Ditto.
4428 (find_def_preds): Ditto.
4429 (destroy_predicates_vecs): Ditto.
4430 (find_matching_predicates_in_rest_chains): Ditto.
4431 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4432 (is_pred_expr_subset): Ditto.
4433 (is_pred_chain_subset_of): Ditto.
4434 (is_included_in): Ditto.
4435 (is_superset_of): Ditto.
4437 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4439 Update copyright years.
4441 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4443 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4444 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4445 config/arc/arc.md, config/arc/arc.opt,
4446 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4447 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4448 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4449 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4450 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4451 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4452 vtable-verify.c, vtable-verify.h: Use the standard form for the
4455 2014-01-02 Tobias Burnus <burnus@net-b.de>
4457 * gcc.c (process_command): Update copyright notice dates.
4458 * gcov-dump.c: Ditto.
4460 * doc/cpp.texi: Bump @copying's copyright year.
4461 * doc/cppinternals.texi: Ditto.
4462 * doc/gcc.texi: Ditto.
4463 * doc/gccint.texi: Ditto.
4464 * doc/gcov.texi: Ditto.
4465 * doc/install.texi: Ditto.
4466 * doc/invoke.texi: Ditto.
4468 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4470 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4472 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4474 * config/i386/sse.md (*mov<mode>_internal): Guard
4475 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4477 PR rtl-optimization/59647
4478 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4479 new_rtx into UNSIGNED_FLOAT rtxes.
4481 Copyright (C) 2014 Free Software Foundation, Inc.
4483 Copying and distribution of this file, with or without modification,
4484 are permitted in any medium without royalty provided the copyright
4485 notice and this notice are preserved.